Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

In our manuscript, we describe a role for the nuclear mRNA export factor UAP56 (a helicase) during metamorphic dendrite and presynapse pruning in flies. We characterize a UAP56 ATPase mutant and find that it rescues the pruning defects of a uap56 mutant. We identify the actin severing enzyme Mical as a potentially crucial UAP56 mRNA target during dendrite pruning and show alterations at both the mRNA and protein level. Finally, loss of UAP56 also causes presynapse pruning defects with actin abnormalities. Indeed, the actin disassembly factor cofilin is required for pruning specifically at the presynapse.

We thank the reviewers for their constructive comments, which we tried to address experimentally as much as possible. To summarize briefly, while all reviewers saw the results as interesting (e. g., Reviewer 3's significance assessment: "Understanding how post-transcriptional events are linked to key functions in neurons is important and would be of interest to a broad audience") and generally methodologically strong, they thought that our conclusions regarding the potential specificity of UAP56 for Mical mRNA was not fully covered by the data. To address this criticism, we added more RNAi analyses of other mRNA export factors and rephrased our conclusions towards a more careful interpretation, i. e., we now state that the pruning process is particularly sensitive to loss of UAP56. In addition, reviewer 1 had technical comments regarding some of our protein and mRNA analyses. We added more explanations and an additional control for the MS2/MCP system. Reviewers 2 and 3 wanted to see a deeper characterization of the ATPase mutant provided. We generated an additional UAP56 mutant transgene, improved our analyses of UAP56 localization, and added a biochemical control experiment. We hope that our revisions make our manuscript suitable for publication.

1. Point-by-point description of the revisions

This section is mandatory. *Please insert a point-by-point reply describing the revisions that were already carried out and included in the transferred manuscript. *

• *

Comments by reviewer 1.

Major comments

1.

For Figure 4, the MS2/MCP system is not quantitative. Using this technique, it is impossible to determine how many RNAs are located in each "dot". Each of these dots looks quite large and likely corresponds to some phase-separated RNP complex where multiple RNAs are stored and/or transported. Thus, these data do not support the conclusion that Mical mRNA levels are reduced upon UAP56 knockdown. A good quantitative microscopic assay would be something like smFISH. Additinally, the localization of Mical mRNA dots to dendrites is not convincing as it looks like regions where there are dendritic swellings, the background is generally brighter.

Our response

We indeed found evidence in the literature that mRNPs labeled with the MS2/MCP or similar systems form condensates (Smith et al., JCB 2015). Unfortunately, smFISH is not established for this developmental stage and would likely be difficult due to the presence of the pupal case. To address whether the Mical mRNPs in control and UAP56 KD neurons are comparable, we characterized the MCP dots in the respective neurons in more detail and found that their sizes did not differ significantly between control and UAP56 KD neurons. To facilitate interpretability, we also increased the individual panel sizes and include larger panels that only show the red (MCP::RFP) channel. We think these changes improved the figure. Thanks for the insight.

Changes introduced: Figure 5 (former Fig. 4): Increased panel size for MCP::RFP images, left out GFP marker for better visibility. Added new analysis of MCP::RFP dot size (new Fig. 5 I).

1.

Alternatively, levels of Mical mRNA could be verified by qPCR in the laval brain following pan-neuronal UAP56 knockdown or in FACS-sorted fluorescently labeled da sensory neurons. Protein levels could be analyzed using a similar approach.

Our response

We thank the reviewer for this comment. Unfortunately, these experiments are not doable as neuron-wide UAP56 KD is lethal (see Flybase entry for UAP56). From our own experience, FACS-sorting of c4da neurons would be extremely difficult as the GFP marker fluorescence intensity of UAP56 KD neurons is weak - this would likely result in preferential sorting of subsets of neurons with weaker RNAi effects. In addition, FACS-sorting whole neurons would not discriminate between nuclear and cytoplasmic mRNA.

The established way of measuring protein content in the Drosophila PNS system is immunofluorescence with strong internal controls. In our case, we also measured Mical fluorescence intensity of neighboring c1da neurons that do not express the RNAi and show expression levels as relative intensities compared to these internal controls. This procedure rules out the influence of staining variation between samples and is used by other labs as well.

1.

In Figure 5, the authors state that Mical expression could not be detected at 0 h APF. The data presented in Fig. 5C, D suggest the opposite as there clearly is some expression. Moreover, the data shown in Fig. 5D looks significantly brighter than the Orco dsRNA control and appears to localize to some type of cytoplasmic granule. So the expression of Mical does not look normal.

Our response

We thank the reviewer for this comment. In the original image in Fig. 5 C, the c4da neuron overlaps with the dendrite from a neighboring PNS neuron (likely c2da or c3da). The latter neuron shows strong Mical staining. We agree that this image is confusing and exchanged this image for another one from the same genotype.

Changes introduced: Figure 5 L (former Fig. 5 C): Exchanged panel for image without overlap from other neuron.

1.

Sufficient data are not presented to conclude any specificity in mRNA export pathways. Data is presented for one export protein (UAP56) and one putative target (Mical). To adequately assess this, the authors would need to do RNA-seq in UAP56 mutants.

Our response

We thank the reviewer for this comment. To address this, we tested whether knockdown of three other mRNA export factors (NXF1, THO2, THOC5) causes dendrite pruning defects, which was not the case (new Fig. S1). While these data are consistent with specific mRNA export pathways, we agree that they are not proof. We therefore toned down our interpretation and removed the conclusion about specificity. Instead, we now use the more neutral term "increased sensibility (to loss of UAP56)".

Changes introduced: Added new Figure S1: RNAi analyses of NXF1, THO2 and THOC5 in dendrite pruning. Introduced concluding sentence at the end of first Results paragraph: We conclude that c4da neuron dendrite pruning is particularly sensitive to loss of UAP56. (p. 6)

1.

In summary, better quantitative assays should be used in Figures 4 and 5 in order to conclude the expression levels of either mRNA or protein. In its current form, this study demonstrates the novel finding that UAP56 regulates dendrite and presynaptic pruning, potentially via regulation of the actin cytoskeleton. However, these data do not convincingly demonstrate that UAP56 controls these processes by regulating of Mical expression and defintately not by controlling export from the nucleus.

Our response

We hope that the changes we introduced above help clarify this.

1.

While there are clearly dendrites shown in Fig. 1C', the cell body is not readily identifiable. This makes it difficult to assess attachment and suggests that the neuron may be dying. This should be replaced with an image that shows the soma.

Our response

We thank the reviewer for this comment. Changes introduced: we replaced the picture in the panel with one where the cell body is more clearly visible.

1.

The level of knockdown in the UAS56 RNAi and P element insertion lines should be determined. It would be useful to mention the nature of the RNAi lines (long/short hairpin). Some must be long since Dcr has been co-expressed. Another issue raised by this is the potential for off-target effects. shRNAi lines would be preferable because these effects are minimized.

Our response

We thank the reviewer for this comment. Assessment of knockdown efficiency is a control to make sure the manipulations work the way they are intended to. As mRNA isolation from Drosophila PNS neurons is extremely difficult, RNAi or mutant phenotypes in this system are controlled by performing several independent manipulations of the same gene. In our case, we used two independent RNAi lines (both long hairpins from VDRC/Bloomington and an additional insertion of the VDRC line, see Table S1) as well as a mutant P element in a MARCM experiment, i. e., a total of three independent manipulations that all cause pruning defects, and the VDRC RNAi lines do not have any predicted OFF targets (not known for the Bloomington line). If any of these manipulations would not have matched, we would have generated sgRNA lines for CRISPR to confirm.

Minor comments:

1.

The authors should explain what EB1:GFP is marking when introduced in the text.

Our response

We thank the reviewer for this comment. Changes introduced: we explain the EB1::GFP assay in the panel with one where the cell body is more clearly visible.

1.

The da neuron images throughout the figures could be a bit larger.

Our response

We thank the reviewer for this comment. Changes introduced: we changed the figure organization to be able to use larger panels:

• the pruning analysis of the ATPase mutations (formerly Fig. 2) is now its own figure (Figure 3).

• we increased the panel sizes of the MCP::RFP images (Figure 5 A - I, formerly Fig. 4).

Reviewer #1 (Significance (Required)):

Strengths:

The methodology used to assess dendrite and presynaptic prunings are strong and the phenotypic analysis is conclusive.

Our response

We thank the reviewer for this comment.

Weakness:

The evidence demonstrating that UAP56 regulates the expression of Mical is unconvincing. Similarly, no data is presented to show that there is any specificity in mRNA export pathways. Thus, these major conclusions are not adequately supported by the data.

Our response

We hope the introduced changes address this comment.

__Reviewer #2 (Evidence, reproducibility and clarity (Required)): __

In this paper, the authors describe dendrite pruning defects in c4da neurons in the DEXD box ATPase UAP56 mutant or in neuronal RNAi knockdown. Overexpression UAP56::GFP or UAP56::GFPE194Q without ATPase activity can rescue dendrite pruning defects in UAP56 mutant. They further characterized the mis-localization of UAP56::GFPE194Q and its binding to nuclear export complexes. Both microtubules and the Ubiquitin-proteasome system are intact in UAP56RNAi neurons. However, they suggest a specific effect on MICAL mRNA nuclear export shown by using the MS2-MCP system., resulting in delay of MICAL protein expression in pruned neurons. Furthermore, the authors show that UAP56 is also involved in presynaptic pruning of c4da neuros in VNC and Mica and actin are also required for actin disassembly in presynapses. They propose that UAP56 is required for dendrite and synapse pruning through actin regulation in Drosophila. Following are my comments.

Major comments

1.

The result that UAP56::GFPE194Q rescues the mutant phenotype while the protein is largely mis-localized suggests a novel mechanism or as the authors suggested rescue from combination of residual activities. The latter possibility requires further support, which is important to support the role mRNA export in dendrite and pre-synapse pruning. One approach would be to examine whether other export components like REF1, and NXF1 show similar mutant phenotypes. Alternatively, depleting residual activity like using null mutant alleles or combining more copies of RNAi transgenes could help.

Our response

We thank the reviewer for this comment. We agree that the mislocalization phenotype is interesting and could inform further studies on the mechanism of UAP56. To further investigate this and to exclude that this could represent a gain-of-function due to the introduced mutation, we made and characterized a new additional transgene, UAP56::GFP E194A. This mutant shows largely the same phenotypes as E194Q, with enhanced interactions with Ref1 and partial mislocalization to the cytoplasm. In addition, we tested whether knockdown of THO2, THOC5 or NXF1 causes pruning defects (no).

Changes introduced:

• added new Figure S1: RNAi analyses of NXF1, THO2 and THOC5 in dendrite pruning.

• made and characterized a new transgene UAP56 E194A (new Fig. 2 B, E, E', 3 C, C', E, F).

1.

The localization of UAP56::GFP (and E194Q) should be analyzed in more details. It is not clear whether the images in Fig. 2A and 2B are from confocal single sections or merged multiple sections. The localization to the nuclear periphery of UAP56::GFP is not clear, and the existence of the E194Q derivative in both nucleus and cytosol (or whether there is still some peripheral enrichment) is not clear if the images are stacked.

Our response

We thank the reviewer for this comment. It is correct that the profiles in the old Figure 2 were from single confocal sections from the displayed images. As it was difficult to create good average profiles with data from multiple neurons, we now introduce an alternative quantification based on categories (nuclear versus dispersed) which includes data from several neurons for each genotype, including the new E194A transgene (new Fig 3 G). Upon further inspection, the increase at the nuclear periphery was not always visible and may have been a misinterpretation. We therefore removed this statement.

Changes introduced:

• added new quantitative analysis of UAP56 wt and E/A, E/Q mutant localization (new Fig 3 G).

1.

The Ub-VV-GFP is a new reagent, and its use to detect active proteasomal degradation is by the lack of GFP signals, which could be also due to the lack of expression. The use of Ub-QQ-GFP cannot confirm the expression of Ub-VV-GFP. The proteasomal subunit RPN7 has been shown to be a prominent component in the dendrite pruning pathway (Development 149, dev200536). Immunostaining using RPN7 antibodies to measure the RPN expression level could be a direct way to address the issue whether the proteasomal pathway is affected or not.

Our response

We thank the reviewer for this comment. We agree that it is wise to not only introduce a positive control for the Ub-VV-GFP sensor (the VCP dominant-negative VCP QQ), but also an independent control. As mutants with defects in proteasomal degradation accumulate ubiquitinated proteins (see, e. g., Rumpf et al., Development 2011), we stained controls and UAP56 KD neurons with antibodies against ubiquitin and found that they had similar levels (new Fig. S3).

Changes introduced:

• added new ubiquitin immunofluorescence analysis (new Fig. S3).

1.

Using the MS2/MCP system to detect the export of MICAL mRNA is a nice approach to confirm the UAP56 activity; lack of UAP56 by RNAi knockdown delays the nuclear export of MS2-MICAL mRNA. The rescue experiment by UAS transgenes could not be performed due to the UAS gene dosage, as suggested by the authors. However, this MS2-MICAL system is also a good assay for the requirement of UAP56 ATPase activity (absence in the E194Q mutant) in this process. Could authors use the MARCM (thus reduce the use of UAS-RNAi transgene) for the rescue experiment? Also, the c4da neuronal marker UAS-CD8-GFP used in Fig4 could be replaced by marker gene directly fused to ppk promoter, which can save a copy of UAS transgene. The results from the rescue experiment would test the dependence of ATPase activity in nuclear export of MICAL mRNA.

Our response

We thank the reviewer for this comment. This is a great idea but unfortunately, this experiment was not feasible due to the (rare) constraints of Drosophila genetics. The MARCM system with rescue already occupies all available chromosomes (X: FLPase, 2nd: FRT, GAL80 + mutant, 3rd: GAL4 + rescue construct), and we would have needed to introduce three additional ones (MCP::RFP and two copies of unmarked genomic MICAL-MS2, all on the third chromosome) that would have needed to be introduced by recombination. Any Drosophilist will see that this is an extreme, likely undoable project :-(

1.

The UAP56 is also involved in presynaptic pruning through regulating actin assembly, and the authors suggest that Mical and cofilin are involved in the process. However, direct observation of lifeact::GFP in Mical or cofilin RNAi knockdown is important to support this conclusion.

Our response

We thank the reviewer for this comment. In response, we analyzed the lifeact::GFP patterns of control and cofilin knockdown neurons and found that loss of cofilin also leads to actin accumulation (new Fig. 7 I, J).

Changes introduced:

• new lifeact analysis (new Fig. 7 I, J).

Minor comments:

1.

RNA localization is important for dendrite development in larval stages (Brechbiel JL, Gavis ER. Curr Biol. 20;18(10):745-750). Yet, the role of UAP56 is relatively specific and shown only in later-stage pruning. It would need thorough discussion.

Our response

We thank reviewer 2 for this comment. We added the following paragraph to the discussion: "UAP56 has also been shown to affect cytoplasmic mRNA localization in Drosophila oocytes (Meignin and Davis, 2008), opening up the possibility that nuclear mRNA export and cytoplasmic transport are linked. It remains to be seen whether this also applies to dendritic mRNA transport (Brechbiel and Gavis, 2008)." (p.13)

1.

Could authors elaborate on the possible upstream regulators that might be involved, as described in "alternatively, several cofilin upstream regulators have been described (Rust, 2015) which might also be involved in presynapse pruning and subject to UAP56 regulation" in Discussion?

Our response

We thank reviewer 2 for this comment. In the corresponding paragraph, we cite as example now that cofilin is regulated by Slingshot phosphatases and LIM kinase (p.14).

1.

In Discussion, the role of cofilin in pre- and post-synaptic processes was described. The role of Tsr/Cofilin regulating actin behaviors in dendrite branching has been described in c3da and c4da neurons (Nithianandam and Chien, 2018 and other references) should be included in Discussion.

Our response

We thank reviewer 2 for this comment. In response we tested whether cofilin is required for dendrite pruning and found that this, in contrast to Mical, is not the case (new Fig. S6). We cite the above paper in the corresponding results section (p.12).

Changes introduced:

• new cofilin dendrite pruning analysis (new Fig. S6).

• added cofilin reference in Results.

1.

The authors speculate distinct actin structures have to be disassembled in dendrite and presynapse pruning in Discussion. What are the possible actin structures in both sites could be elaborated.

Our response

We thank reviewer 2 for this comment. In response, we specify in the Discussion: "As Mical is more effective in disassembling bundled F-actin than cofilin (Rajan et al., 2023), it is interesting to speculate that such bundles are more prevalent in dendrites than at presynapses." (p14)

Reviewer #2 (Significance (Required)):

The study initiated a genetic screen for factors involved in a dendrite pruning system and reveals the involvement of nuclear mRNA export is an important event in this process. They further identified the mRNA of the actin disassembly factor MICAL is a candidate substrate in the exporting process. This is consistent with previous finding that MICAL has to be transcribed and translated when pruning is initiated. As the presynapses of the model c4da neuron in this study is also pruned, the dependence on nuclear export and local actin remodeling were also shown. Thus, this study has added another layer of regulation (the nuclear mRNA export) in c4da neuronal pruning, which would be important for the audience interested in neuronal pruning. The study is limited for the confusing result whether ATPase activity of the exporting factor is required.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

Summary: In the manuscript by Frommeyer, Gigengack et al. entitled "The UAP56 mRNA Export Factor is Required for Dendrite and Synapse Pruning via Actin Regulation in Drosophila" the authors surveyed a number of RNA export/processing factors to identify any required for efficient dendrite and/or synapse pruning. They describe a requirement for a general poly(A) RNA export factor, UAP56, which functions as an RNA helicase. They also study links to aspects of actin regulation.

Overall, while the results are interesting and the impact of loss of UAP56 on the pruning is intriguing, some of the data are overinterpreted as presented. The argument that UAP56 may be specific for the MICAL RNA is not sufficiently supported by the data presented. The two stories about poly(A) RNA export/processing and the actin regulation seem to not quite be connected by the data presented. The events are rather distal within the cell, making connecting the nuclear events with RNA to events at the dendrites/synapse challenging.

Our response

We thank reviewer 3 for this comment. To address this, we tested whether knockdown of three other mRNA export factors (NXF1, THO2, THOC5) causes dendrite pruning defects, which was not the case (new Fig. S1). While these data are consistent with specific mRNA export pathways, we agree that they are not proof. We therefore toned down our interpretation and removed the conclusion about specificity. Instead, we now use the more neutral term "increased sensibility (to loss of UAP56)".

We agree that it is a little hard to tie cofilin to UAP56, as we currently have no evidence that cofilin levels are affected by loss of UAP56, even though both seem to affect lifeact::GFP in a similar way (new Fig. 7 I, J). However, a dysregulation of cofilin can also occur through dysregulation of upstream cofilin regulators such as Slingshot and LIM kinase, making such a relationship possible.

Changes introduced:

• added new Figure S1: RNAi analyses of NXF1, THO2 and THOC5 in dendrite pruning.

• introduced concluding sentence at the end of first Results paragraph: "We conclude that c4da neuron dendrite pruning is particularly sensitive to loss of UAP56." (p. 6)

• add new lifeact::GFP analysis of cofilin KD (new Fig. I, J).

• identify potential other targets from the literature in the Discussion (Slingshot phosphatases and LIM kinase, p.14).

There are a number of specific statements that are not supported by references. See, for example, these sentences within the Introduction- "Dysregulation of pruning pathways has been linked to various neurological disorders such as autism spectrum disorders and schizophrenia. The cell biological mechanisms underlying pruning can be studied in Drosophila." The Drosophila sentence is followed by some specific examples that do include references. The authors also provide no reference to support the variant that they create in UAP56 (E194Q) and whether this is a previously characterized fly variant or based on an orthologous protein in a different system. If so, has the surprising mis-localization been reported in another system?

Our response

We thank reviewer 3 for this comment. We added the following references on pruning and disease:

1) Howes, O.D., Onwordi, E.C., 2023. The synaptic hypothesis of schizophrenia version III: a master mechanism. Mol. Psychiatry 28, 1843-1856.

2) Tang, G., et al., 2014. Loss of mTOR-dependent macroautophagy causes autistic-like synaptic pruning deficits. Neuron 83, 1131-43.

To better introduce the E194 mutations, we explain the position of the DECD motif in the Walker B domain, give the corresponding residues in the human and yeast homologues and cite papers demonstrating the importance of this residue for ATPase activity:

3) Saguez, C., et al., 2013. Mutational analysis of the yeast RNA helicase Sub2p reveals conserved domains required for growth, mRNA export, and genomic stability. RNA 19:1363-71.

4) Shen, J., et al., 2007. Biochemical Characterization of the ATPase and Helicase Activity of UAP56, an Essential Pre-mRNA Splicing and mRNA Export Factor. J. Biol. Chem. 282, P22544-22550.

We are not aware of other studies looking at the relationship between the UAP56 ATPase and its localization. Thank you for pointing this out!

Specific Comments:

Specific Comment 1: Figure 1 shows the impact of loss of UAP56 on neuron dendrite pruning. The experiment employs both two distinct dsRNAs and a MARCM clone, providing confidence that there is a defect in pruning upon loss of UAP56. As the authors mention screening against 92 genes that caused splicing defects in S2 cells, inclusion of some examples of these genes that do not show such a defect would enhance the argument for specificity with regard to the role of UAP56. This control would be in addition to the more technical control that is shown, the mCherry dsRNA.

Our response

We thank reviewer 3 for this comment. To address this, we included the full list of screened genes with their phenotypic categorization regarding pruning (103 RNAi lines targeting 64 genes) as Table S1. In addition, we also tested four RNAi lines targeting the nuclear mRNA export factors Nxf1, THO2 and THOC5 which do not cause dendrite pruning defects (Fig. S1).

Changes introduced:

• added RNAi screen results as a list in Table S1.

• added new Figure S1: RNAi analyses of NXF1, THO2 and THOC5 in dendrite pruning.

Specific Comment 2: Later the authors demonstrate a delay in the accumulation of the Mical protein, so if they assayed these pruning events at later times, would the loss of UAP56 cause a delay in these events as well? Such a correlation would enhance the causality argument the authors make for Mical levels and these pruning events.

Our response

We thank reviewer 3 for this comment. Unfortunately, this is somewhat difficult to assess, as shortly after the 18 h APF timepoint, the epidermal cells that form the attachment substrate for c4da neuron dendrites undergo apoptosis. Where assessed (e. g., Wang et al., 2017, Development) 144: 1851–1862), this process, together with the reduced GAL4 activity of our ppk-GAL4 during the pupal stage (our own observations), eventually leads to pruning, but the causality cannot be easily attributed anymore. We therefore use the 18 h APF timepoint essentially as an endpoint assay.

Specific Comment 3: Figure 2 provides data designed to test the requirement for the ATPase/helicase activity of UAP56 for these trimming events. The first observation, which is surprising, is the mislocalization of the variant (E194Q) that the authors generate. The data shown does not seem to indicate how many cells the results shown represent as a single image and trace is shown the UAP56::GFP wildtype control and the E194Q variant.

Our response

We thank reviewer 3 for this comment. It is correct that the traces shown are from single confocal sections. To better display the phenotypic penetrance, we now added a categorical analysis that shows that the UAP56 E194Q mutant is completely mislocalized in the majority of cells assessed (and the newly added E194A mutant in a subset of cells).

Changes introduced:

• added categorical quantification of UAP56 variant localization (new Fig. 2 G).

__Specific Comment 4: __Given the rather surprising finding that the ATPase activity is not required for the function of UAP56 characterized here, the authors do not provide sufficient references or rationale to support the ATPase mutant that they generate. The E194Q likely lies in the Walker B motif and is equivalent to human E218Q, which can prevent proper ATP hydrolysis in the yeast Sub2 protein. There is no reference to support the nature of the variant created here.

Our response

We thank reviewer 3 for this comment. To better introduce the E194 mutations, we explain the position of the DECD motif in the Walker B domain, give the corresponding residues in the human and yeast homologues (Sub2) and cite papers demonstrating the importance of this residue for ATPase activity:

1) Saguez, C., et al., 2013. Mutational analysis of the yeast RNA helicase Sub2p reveals conserved domains required for growth, mRNA export, and genomic stability. RNA 19:1363-71.

2) Shen, J., et al., 2007. Biochemical Characterization of the ATPase and Helicase Activity of UAP56, an Essential Pre-mRNA Splicing and mRNA Export Factor. J. Biol. Chem. 282, P22544-22550.

__Specific Comment 5: __Given the surprising results, the authors could have included additional variants to ensure the change has the biochemical effect that the authors claim. Previous studies have defined missense mutations in the ATP-binding site- K129A (Lysine to Alanine): This mutation, in both yeast Sub2 and human UAP56, targets a conserved lysine residue that is critical for ATP binding. This prevents proper ATP binding and consequently impairs helicase function. There are also missense mutations in the DEAD-box motif, (Asp-Glu-Ala-Asp) involved in ATP binding and hydrolysis. Mutations in this motif, such as D287A in yeast Sub2 (corresponding to D290A in human UAP56), can severely disrupt ATP hydrolysis, impairing helicase activity. In addition, mutations in the Walker A (GXXXXGKT) and Walker B motifs are can impair ATP binding and hydrolysis in DEAD-box helicases. Missense mutations in these motifs, like G137A (in the Walker A motif), can block ATP binding, while E218Q (in the Walker B motif)- which seems to be the basis for the variant employed here- can prevent proper ATP hydrolysis.

Our response

We thank reviewer 3 for this comment. Our cursory survey of the literature suggested that mutations in the Walker B motif are the most specific as they still preserve ATP binding and their effects have not well been characterized overall. In addition, these mutations can create strong dominant-negatives in related helicases (e. g., Rode et al., 2018 Cell Reports, our lab). To better characterize the role of the Walker B motif in UAP56, we generated and characterized an alternative mutant, UAP56 E194A. While the E194A variant does not show the same penetrance of localization phenotypes as E194Q, it also is partially mislocalized, shows stronger binding to Ref1 and also rescues the uap56 mutant phenotypes without an obvious dominant-negative effect, thus confirming our conclusions regarding E194Q.

Changes introduced:

• added biochemical, localization and phenotypic analysis of newly generated UAP56 E194A variant (new Figs. 2 B, 2 E, E', 3 C, C'). categorical quantification of UAP56 variant localization (new Fig. 2 G).

__Specific Comment 6: __The co-IP results shown in Figure 2C would also seem to have multiple potential interpretations beyond what the authors suggest, an inability to disassemble a complex. The change in protein localization with the E194Q variant could impact the interacting proteins. There is no negative control to show that the UAP56-E194Q variant is not just associated with many, many proteins. Another myc-tagged protein that does not interact would be an ideal control.

Our response

We thank reviewer 3 for this comment. To address this comment, we tried to co-IP UAP56 wt or UAP56 E194Q with a THO complex subunit THOC7 (new Fig. S2). The results show that neither UAP56 variant can co-IP THOC7 under our conditions (likely because the UAP56/THO complex intermediate during mRNA export is disassembled in an ATPase-independent manner (Hohmann et al., Nature 2025)).

Changes introduced:

• added co-IP experiment between UAP56 variants and THOC7 (new Fig. S2).

__Specific Comment 7: __With regard to Figure 3, the authors never define EB1::GFP in the text of the Results, so a reader unfamiliar with this system has no idea what they are seeing. Reading the Materials and Methods does not mitigate this concern as there is only a brief reference to a fly line and how the EB1::GFP is visualized by microscopy. This makes interpretation of the data presented in Figure 3A-C very challenging.

Our response

We thank reviewer 3 for pointing this out. We added a description of the EB1::GFP analysis in the corresponding Results section (p.8).

__Specific Comment 8: __The data shown for MICAL MS2 reporter localization in Figure 4 is nice, but is also fully expected on many former studies analyzing loss of UAP56 or UAP56 hypomorphs in different systems. While creating the reporter is admirable, to make the argument that MICAL localization is in some way preferentially impacted by loss of UAP56, the authors would need to examine several other transcripts. As presented, the authors can merely state that UAP56 seems to be required for the efficient export of an mRNA transcript, which is predicted based on dozens of previous studies dating back to the early 2000s.

Our response

Firstly, thank you for commenting on the validity of the experimental approach! The primary purpose of this experiment was to test whether the mechanism of UAP56 during dendrite pruning conforms with what is known about UAP56's cellular role - which it apparently does. We also noted that our statements regarding the specificity of UAP56 for Mical over other transcripts are difficult. While our experiments would be consistent with such a model, they do not prove it. We therefore toned down the corresponding statements (e. g., the concluding sentence at the end of first Results paragraphis now: "We conclude that c4da neuron dendrite pruning is particularly sensitive to loss of UAP56." (p. 6)).

Minor (and really minor) points:

In the second sentence of the Discussion, the word 'developing' seems to be mis-typed "While a general inhibition of mRNA export might be expected to cause broad defects in cellular processes, our data in develoing c4da neurons indicate that loss of UAP56 mainly affects pruning mechanisms related to actin remodeling."

Sentence in the Results (lack of page numbers makes indicating where exactly a bit tricky)- "We therefore reasoned that Mical expression could be more challenging to c4da neurons." This is a complete sentence as presented, yet, if something is 'more something'- the thing must be 'more than' something else. Presumably, the authors mean that the length of the MICAL transcript could make the processing and export of this transcript more challenging than typical fly transcripts (raising the question of the average length of a mature transcript in flies?).

Our response

Thanks for pointing these out. The typo is fixed, page numbers are added. We changed the sentence to: "Because of the large size of its mRNA, we reasoned that MICAL gene expression could be particularly sensitive to loss of export factors such as UAP56." (p.9) We hope this is more precise language-wise.

Reviewer #3 (Significance (Required)):

Understanding how post-transcriptional events are linked to key functions in neurons is important and would be of interest to a broad audience.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary:

In the manuscript by Frommeyer, Gigengack et al. entitled "The UAP56 mRNA Export Factor is Required for Dendrite and Synapse Pruning via Actin Regulation in Drosophila" the authors surveyed a number of RNA export/processing factors to identify any required for efficient dendrite and/or synapse pruning. They describe a requirement for a general poly(A) RNA export factor, UAP56, which functions as an RNA helicase. They also study links to aspects of actin regulation.

Overall, while the results are interesting and the impact of loss of UAP56 on the pruning is intriguing, some of the data are overinterpreted as presented. The argument that UAP56 may be specific for the MICAL RNA is not sufficiently supported by the data presented. The two stories about poly(A) RNA export/processing and the actin regulation seem to not quite be connected by the data presented. The events are rather distal within the cell, making connecting the nuclear events with RNA to events at the dendrites/synapse challenging.

There are a number of specific statements that are not supported by references. See, for example, these sentences within the Introduction- "Dysregulation of pruning pathways has been linked to various neurological disorders such as autism spectrum disorders and schizophrenia. The cell biological mechanisms underlying pruning can be studied in Drosophila." The Drosophila sentence is followed by some specific examples that do include references. The authors also provide no reference to support the variant that they create in UAP56 (E194Q) and whether this is a previously characterized fly variant or based on an orthologous protein in a different system. If so, has the surprising mis-localization been reported in another system?

Specific Comments:

Figure 1 shows the impact of loss of UAP56 on neuron dendrite pruning. The experiment employs both two distinct dsRNAs and a MARCM clone, providing confidence that there is a defect in pruning upon loss of UAP56. As the authors mention screening against 92 genes that caused splicing defects in S2 cells, inclusion of some examples of these genes that do not show such a defect would enhance the argument for specificity with regard to the role of UAP56. This control would be in addition to the more technical control that is shown, the mCherry dsRNA. Later the authors demonstrate a delay in the accumulation of the Mical protein, so if they assayed these pruning events at later times, would the loss of UAP56 cause a delay in these events as well? Such a correlation would enhance the causality argument the authors make for Mical levels and these pruning events.

Figure 2 provides data designed to test the requirement for the ATPase/helicase activity of UAP56 for these trimming events. The first observation, which is surprising, is the mislocalization of the variant (E194Q) that the authors generate. The data shown does not seem to indicate how many cells the results shown represent as a single image and trace is shown the UAP56::GFP wildtype control and the E194Q variant.

Given the rather surprising finding that the ATPase activity is not required for the function of UAP56 characterized here, the authors do not provide sufficient references or rationale to support the ATPase mutant that they generate. The E194Q likely lies in the Walker B motif and is equivalent to human E218Q, which can prevent proper ATP hydrolysis in the yeast Sub2 protein. There is no reference to support the nature of the variant created here.

Given the surprising results, the authors could have included additional variants to ensure the change has the biochemical effect that the authors claim. Previous studies have defined missense mutations in the ATP-binding site- K129A (Lysine to Alanine): This mutation, in both yeast Sub2 and human UAP56, targets a conserved lysine residue that is critical for ATP binding. This prevents proper ATP binding and consequently impairs helicase function. There are also missense mutations in the DEAD-box motif, (Asp-Glu-Ala-Asp) involved in ATP binding and hydrolysis. Mutations in this motif, such as D287A in yeast Sub2 (corresponding to D290A in human UAP56), can severely disrupt ATP hydrolysis, impairing helicase activity. In addition, mutations in the Walker A (GXXXXGKT) and Walker B motifs are can impair ATP binding and hydrolysis in DEAD-box helicases. Missense mutations in these motifs, like G137A (in the Walker A motif), can block ATP binding, while E218Q (in the Walker B motif)- which seems to be the basis for the variant employed here- can prevent proper ATP hydrolysis.

The co-IP results shown in Figure 2C would also seem to have multiple potential interpretations beyond what the authors suggest, an inability to disassemble a complex. The change in protein localization with the E194Q variant could impact the interacting proteins. There is no negative control to show that the UAP56-E194Q variant is not just associated with many, many proteins. Another myc-tagged protein that does not interact would be an ideal control.

With regard to Figure 3, the authors never define EB1::GFP in the text of the Results, so a reader unfamiliar with this system has no idea what they are seeing. Reading the Materials and Methods does not mitigate this concern as there is only a brief reference to a fly line and how the EB1::GFP is visualized by microscopy. This makes interpretation of the data presented in Figure 3A-C very challenging. The data shown for MICAL MS2 reporter localization in Figure 4 is nice, but is also fully expected on many former studies analyzing loss of UAP56 or UAP56 hypomorphs in different systems. While creating the reporter is admirable, to make the argument that MICAL localization is in some way preferentially impacted by loss of UAP56, the authors would need to examine several other transcripts. As presented, the authors can merely state that UAP56 seems to be required for the efficient export of an mRNA transcript, which is predicted based on dozens of previous studies dating back to the early 2000s.

Minor (and really minor) points:

In the second sentence of the Discussion, the word 'developing' seems to be mis-typed "While a general inhibition of mRNA export might be expected to cause broad defects in cellular processes, our data in develoing c4da neurons indicate that loss of UAP56 mainly affects pruning mechanisms related to actin remodeling."

Sentence in the Results (lack of page numbers makes indicating where exactly a bit tricky)- "We therefore reasoned that Mical expression could be more challenging to c4da neurons." This is a complete sentence as presented, yet, if something is 'more something'- the thing must be 'more than' something else. Presumably, the authors mean that the length of the MICAL transcript could make the processing and export of this transcript more challenging than typical fly transcripts (raising the question of the average length of a mature transcript in flies?).

Significance

Understanding how post-transcriptional events are linked to key functions in neurons is important and would be of interest to a broad audience.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

In this paper, the authors describe dendrite pruning defects in c4da neurons in the DEXD box ATPase UAP56 mutant or in neuronal RNAi knockdown. Overexpression UAP56::GFP or UAP56::GFPE194Q without ATPase activity can rescue dendrite pruning defects in UAP56 mutant. They further characterized the mis-localization of UAP56::GFPE194Q and its binding to nuclear export complexes. Both microtubules and the Ubiquitin-proteasome system are intact in UAP56RNAi neurons. However, they suggest a specific effect on MICAL mRNA nuclear export shown by using the MS2-MCP system., resulting in delay of MICAL protein expression in pruned neurons. Furthermore, the authors show that UAP56 is also involved in presynaptic pruning of c4da neuros in VNC and Mica and actin are also required for actin disassembly in presynapses. They propose that UAP56 is required for dendrite and synapse pruning through actin regulation in Drosophila. Following are my comments.

Major comments

1. The result that UAP56::GFPE194Q rescues the mutant phenotype while the protein is largely mis-localized suggests a novel mechanism or as the authors suggested rescue from combination of residual activities. The latter possibility requires further support, which is important to support the role mRNA export in dendrite and pre-synapse pruning. One approach would be to examine whether other export components like REF1, and NXF1 show similar mutant phenotypes. Alternatively, depleting residual activity like using null mutant alleles or combining more copies of RNAi transgenes could help.

2. The localization of UAP56::GFP (and E194Q) should be analyzed in more details. It is not clear whether the images in Fig. 2A and 2B are from confocal single sections or merged multiple sections. The localization to the nuclear periphery of UAP56::GFP is not clear, and the existence of the E194Q derivative in both nucleus and cytosol (or whether there is still some peripheral enrichment) is not clear if the images are stacked.

3. The Ub-VV-GFP is a new reagent, and its use to detect active proteasomal degradation is by the lack of GFP signals, which could be also due to the lack of expression. The use of Ub-QQ-GFP cannot confirm the expression of Ub-VV-GFP. The proteasomal subunit RPN7 has been shown to be a prominent component in the dendrite pruning pathway (Development 149, dev200536). Immunostaining using RPN7 antibodies to measure the RPN expression level could be a direct way to address the issue whether the proteasomal pathway is affected or not.

4. Using the MS2/MCP system to detect the export of MICAL mRNA is a nice approach to confirm the UAP56 activity; lack of UAP56 by RNAi knockdown delays the nuclear export of MS2-MICAL mRNA. The rescue experiment by UAS transgenes could not be performed due to the UAS gene dosage, as suggested by the authors. However, this MS2-MICAL system is also a good assay for the requirement of UAP56 ATPase activity (absence in the E194Q mutant) in this process. Could authors use the MARCM (thus reduce the use of UAS-RNAi transgene) for the rescue experiment? Also, the c4da neuronal marker UAS-CD8-GFP used in Fig4 could be replaced by marker gene directly fused to ppk promoter, which can save a copy of UAS transgene. The results from the rescue experiment would test the dependence of ATPase activity in nuclear export of MICAL mRNA.

5. The UAP56 is also involved in presynaptic pruning through regulating actin assembly, and the authors suggest that Mical and cofilin are involved in the process. However, direct observation of lifeact::GFP in Mical or cofilin RNAi knockdown is important to support this conclusion.

Minor comments

1. RNA localization is important for dendrite development in larval stages (Brechbiel JL, Gavis ER. Curr Biol. 20;18(10):745-750). Yet, the role of UAP56 is relatively specific and shown only in later-stage pruning. It would need thorough discussion.

2. Could authors elaborate on the possible upstream regulators that might be involved, as described in "alternatively, several cofilin upstream regulators have been described (Rust, 2015) which might also be involved in presynapse pruning and subject to UAP56 regulation" in Discussion?

3. In Discussion, the role of cofilin in pre- and post-synaptic processes was described. The role of Tsr/Cofilin regulating actin behaviors in dendrite branching has been described in c3da and c4da neurons (Nithianandam and Chien, 2018 and other references) should be included in Discussion.

4. The authors speculate distinct actin structures have to be disassembled in dendrite and presynapse pruning in Discussion. What are the possible actin structures in both sites could be elaborated.

Significance

The study initiated a genetic screen for factors involved in a dendrite pruning system and reveals the involvement of nuclear mRNA export is an important event in this process. They further identified the mRNA of the actin disassembly factor MICAL is a candidate substrate in the exporting process. This is consistent with previous finding that MICAL has to be transcribed and translated when pruning is initiated. As the presynapses of the model c4da neuron in this study is also pruned, the dependence on nuclear export and local actin remodeling were also shown. Thus, this study has added another layer of regulation (the nuclear mRNA export) in c4da neuronal pruning, which would be important for the audience interested in neuronal pruning. The study is limited for the confusing result whether ATPase activity of the exporting factor is required.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Summary:

This manuscript by Frommeyer et al. explores the role of the helicase and regulator of nuclear export, UAP56, in the control of dendrite and presynaptic pruning in Drosophila larval da sensory neurons. The authors present evidence showing that UAP56 regulates these processes via the actin cytoskeleton and suggest that this is occurs by controlling the expression of the actin severing enzyme, Mical.

Major comments:

The most signficant issue with the manuscript is that some of the major conclusions are not supported by the data. Additional experiment would need to be completed in order support these claims. These (and other) major comments are as follows:

1. For Figure 4, the ms2/MCP system is not quantitative. Using this technique, it is impossible to determine how many RNAs are located in each "dot". Each of these dots looks quite large and likely corresponds to some phase-separated RNP complex where multiple RNAs are stored and/or transported. Thus, these data do not support the conclusion that Mical mRNA levels are reduced upon UAP56 knockdown. A good quantitative microscopic assay would be something like smFISH. Additinally, the localization of Mical mRNA dots to dendrites is not convincing as it looks like regions where there are dendritic swellings, the background is generally brighter.

2. Alternatively, levels of Mical mRNA could be verified by qPCR in the laval brain following pan-neuronal UAP56 knockdown or in FACS-sorted fluorescently labeled da sensory neurons. Protein levels could be analyzed using a similar approach.

3. In Figure 5, the authors state that Mical expression could not be detected at 0 h APF. The data presented in Fig. 5C, D suggest the opposite as there clearly is some expression. Moreover, the data shown in Fig. 5D looks significantly brighter than the Orco dsRNA control and appears to localize to some type of cytoplasmic granule. So the expression of Mical does not look normal.

4. Sufficient data are not presented to conclude any specificity in mRNA export pathways. Data is presented for one export protein (UAP56) and one putative target (Mical). To adequately assess this, the authors would need to do RNA-seq in UAP56 mutants.

5. In summary, better quantitative assays should be used in Figures 4 and 5 in order to conclude the expression levels of either mRNA or protein. In its current form, this study demonstrates the novel finding that UAP56 regulates dendrite and presynaptic pruning, potentially via regulation of the actin cytoskeleton. However, these data do not convincingly demonstrate that UAP56 controls these processes by regulating of Mical expression and defintately not by controlling export from the nucleus.

6. While there are clearly dendrites shown in Fig. 1C', the cell body is not readily identifiable. This makes it difficult to assess attachment and suggests that the neuron may be dying. This should be replaced with an image that shows the soma.

7. The level of knockdown in the UAS56 RNAi and P element insertion lines should be determined. It would be useful to mention the nature of the RNAi lines (long/short hairpin). Some must be long since Dcr has been co-expressed. Another issue raised by this is the potential for off-target effects. shRNAi lines would be preferable because these effects are minimized.

Minor comments:

1. The authors should explain what EB1:GFP is marking when introduced in the text.

2. The da neuron images througout the figures could be a bit larger.

Significance

Strengths:

The methodology used to assess dendrite and presynaptic prunings are strong and the phenotypic analysis is conclusive.

Weakness:

The evidence demonstrating that UAP56 regulates the expression of Mical is unconvincing. Similarly, no data is presented to show that there is any specificity in mRNA export pathways. Thus, these major conclusions are not adequately supported by the data.

Advance:

The findings that UAP56 regulate dendrite and synaptic pruning are novel. As is its specific regulation of the actin cytoskeleton. These findings are restricted to a phenotypic analysis and do not show that it is not simply due to the disruption of general mRNA export.

Audience:

In its current form the manuscript whould be of interest to an audience who specializes in the study of RNA binding proteins in the control of neurodevelopment. This would include scientists who work in invertebrate and vertebrate model systems.

My expertise:

My lab uses Drosophila to study the role of RNA binding proteins in neurodevelopment and neurodegeneration. Currently, we use flies as a model to better understand the molecular pathogenesis of neurodevelopmenal disorders such as FXS and ASD.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

*Reviewer #1 (Evidence, reproducibility and clarity (Required): *

*Using genetics and microscopy approaches, Cabral et al. investigate how fission yeast regulates its length and width in response to osmotic, oxidative, or low glucose stress. Miller et al. have recently found that the cell cycle regulators Cdc25, Cdc13 and Cdr2 integrate information about cell volume, time and cell surface area into the cellular decision when to divide. Cabral now build on this work and test how disruption of these regulators affects cell size adaptation. They find that each stress condition shows a distinct dependence on the individual regulators, suggesting that the complex size control network enables optimized size adaptation for each condition. Overall, the manuscript is clear and the detailed methods ensure that the experiments can be replicated.

Major comments:

1.) It would be much easier to follow the authors' conclusions, if in addition to surface area to volume ratio, length and width, they would also plot cell volume at division in Figs. 1-4.*

AUTHOR RESPONSE: Due to space constraints in the main (and supplemental) figures, we focused on SA:Vol ratio together with cell length and width, which directly define cell geometry in rod-shaped fission yeast. Surface area and volume are derived from these measurements and can be misleading when considered alone, as similar surface area or volume values can arise from distinct combinations of length and width. The SA:Vol ratio therefore serves as a robust integrative metric for capturing coordinated changes in length and width that reshape cell geometry. We would be happy to include individual surface area and volume plots if requested.

2.) To me, it seems that maybe even more than upon osmotic stress, the cdc13-2x strain differs qualitatively from WT in low glucose conditions, where the increased SA-V ratio is almost completely abolished.

AUTHOR RESPONSE: We agree with the reviewer and have revised the manuscript text to point out this difference. The newly added text states: “Under low glucose, cdc13-2x cells also showed a WT-like response, decreasing length and increasing in SA:Vol ratio (Figures 3B-D). However, this SA:Vol increase was reduced compared to WT (1% vs 8.5%; Figures 1D and 3B), suggesting impaired geometric remodeling under glucose limitation.”

3.) It is not entirely clear to me why two copies of Cdc13 would qualitatively affect the responses. Shouldn't the extra copy behave similarly to the endogenous one and therefore only lead to quantitative changes? Maybe the authors can discuss this more clearly or even test a strain in which Cdc13 function is qualitatively disrupted.

AUTHOR RESPONSE: Increased Cdc13 protein concentration in cdc13-2x cells disrupts the typical time-scaling of Cdc13 protein. Consistent with this, cdc13-2x cells enter mitosis at a smaller cell size. We have modified the text to clarify this point. The new text states: “To access the role of the Cdc13 time-sensing pathway, we disrupted Cdc13 protein abundance by creating a cdc13-2x strain carrying an additional copy of cdc13 integrated at an exogenous locus. cdc13-2x cells divided at a smaller size than WT, reflecting accelerated mitotic entry upon disruption of typical time-scaling of Cdc13 protein (Figure S1A).”

4.) I don't see why the authors come to the conclusion that under osmotic stress cells would maximize cell volume. It leads to a decreased cell length, doesn't it?

AUTHOR RESPONSE: WT cells under osmotic stress do decrease in length, but this is accompanied by an increase in cell width. Because width contributes disproportionately to cell volume in rod-shaped cells, this change results in a modest but reproducible reduction in the SA:Vol ratio relative to WT cells in control medium (Figure 1D). We note that the degree of this change under osmotic stress is small (-0.4%), although statistically significant (p * Likewise, in Figure 2B, they interpret tiny changes in the SA/V. By my estimation, the difference between control and osmotic stress is only 2% (1.195/1.17), less that the wild-type case, which appears to be twice that (which is still pretty modest). The small amplitude of these changes is obscured by the fact that the graphs do not have a baseline at zero, which, as a matter of good data-presentation practice, they should.

*

AUTHOR RESPONSE: We appreciate the reviewer’s distinction between statistical and biological significance and agree that this is an important point to clarify. We now note in the revised text that changes in SA:Vol ratio under osmotic stress are numerically small and should not be overinterpreted. Our revised text now states: “Under oxidative and osmotic stress, the SA:Vol ratio decreased, indicating greater cell volume expansion relative to surface area (Figure 1D). However, we note that the reduction in SA:Vol under osmotic stress, while statistically significant, was modest in magnitude (−0.4%).”

Although small in absolute terms, even subtle geometric changes can be biologically meaningful in fission yeast due to the small size of these cells, where minor shifts in length or width translate into measurable differences in membrane area relative to cytoplasmic volume. Importantly, in Figure 2B, the key observation is not the magnitude of the change but its direction: cdc25-degron-DaMP cells exhibit a ~2% increase in SA:Vol ratio under osmotic stress, in contrast to the decrease observed in WT cells under the same condition. This opposite response reflects altered cell geometry and is supported by corresponding changes in cell length and width. We have revised the Results text to emphasize both the modest magnitude and the directional nature of these effects: “Under osmotic stress, cdc25-degron-DaMP cells exhibited a ~2% increase in SA:Vol ratio, opposite to the modest decrease observed in WT cells. This increase arose from increased cell length and reduced width (Figures 2B-D).”

Regarding data presentation, because SA:Vol ratios vary over a narrow numerical range, setting the y-axis minimum to zero would compress the data and obscure all detectable differences. Instead, we have modifed our SA:Vol ratio graphs in Fig. 1-4 to have consistent axis scaling across panels to accurately convey relative changes while maintaining visual clarity. We are happy to provide full data tables and statistical outputs upon request.

* I am also concerned about the use of manual measurement of width at a single point along the cell. This approach is very sensitive to the choice of width point and to non-cylindrical geometries, several of which are evident in the images presented. MATLAB will return the ??? as well as the length from a mask, but even better, one can more accurately calculate the surface area and volume by assuming rotational symmetry of the mask. Given that surface area and volume calculation need to be redone anyway, as discussed below, I encourage the authors to calculate them directly from the mask, instead of using the cylindrical assumption.*

AUTHOR RESPONSE: In initial experiments to calculate surface area and volume of fission yeast cells for prior work (Miller et al., 2023, Current Biology) we found that automated width measurements by MATLAB or ImageJ were inaccurate for a subset of cells leading to noisy cell surface area and volume values. Measuring cell width by hand and assuming that each cell in a given strain had the same cell radius (average of population) for calculation of cell surface area and volume gave more consistent results and recapitulated established conclusions regarding size control mechanisms.

In this previous work and the current study, abnormally skinny or wide regions of a cell were avoided when drawing a line to measure the cell width by hand. For each strain and condition, an average cell width was determined per independent experiment and used for surface area and volume calculations. Additionally, previous analysis demonstrated that this approach yields results consistent with a rotation method derived directly from cell masks, which does not assume a cylindrical cell shape (Facchetti et al., 2019, Current Biology; Miller et al., 2023, Current Biology).

To test the validity of our size measurements and confirm the robustness of our results in this study we compared the surface area and volume of cells by this rotation method. We have added this additional information to our revised methods section and also added SA:Vol ratio graphs generated from the rotation size measurement to our revised Figure S1 E-J. Importantly, both approaches used to measure cell size gave consistent results and supported the same conclusions.*

The authors also need to be more careful about their claims about size-dependent scaling. The concentration of both Cdc13 and Cdc25 scale with size (perhaps indirectly, in the case of Cdc13), but Cdr2 does not. Cdr2 activity has been proposed to scale with size, and its density at cortical nodes has been reported to scale with size, although that claim has been challenged .*

AUTHOR RESPONSE: We have modified text in the Introduction and Results to address this point. Our revised text in the introduction states: “Recent work has shown that Cdk1 activation integrates size- and time-dependent inputs: the Wee1-inhibitory kinase Cdr2 cortical node density scales with cell surface area (Pan et al., 2014; Facchetti et al., 2019); Cdc25 nuclear accumulation scales with cell volume; and cyclin Cdc13 accumulates over time in the nucleus (Miller et al., 2023) (Figure 1B).” Our revised text in the results section states: “Cdr2 functions as a cortical scaffold that regulates Wee1 activity in relation to cell size, with Cdr2 nodal density reported to scale with cell surface area, enforcing a surface area threshold for mitotic entry (Pan et al., 2014; Allard et al., 2018; Facchetti et al., 2019; Sayyad and Pollard, 2022).”*

Even taking the authors approach at face value, there are observations that do not seem to make sense, which led me to realize that the wrong formulae were used to calculate surface area and volume.

In Figure 1E,F, the KCl-treated cells get shorter and wider; surely, that should result in a lower SA/V ratio. However, as noted above, in Figure 1D, they are shown to have a similar ratio. As a sanity check, I eye-balled the numbers off of the figure (control: 14 µm x 3.6 µm and KCl: 11 µm x 3.8 µm) and calculated their surface area and volume using the formula for a capsule (i.e., a cylinder with hemispheric ends).

SA = the surface area of the two hemispheres + the surface are of the cylinder in between = 4*pi*(width/2)^2 + pi*width*(length-width), the length-width term calculates the side length of the capsule (length without the hemispheres) from the full length of the capsule (length including the hemispheres)

V = the volume of the two hemispheres + the volume of the cylinder in between = 4/3*pi*(width/2)^3 + pi*(width/2)^2*(length-width).

I got SA/V ratios of around 2, which are way off from what is presented in Figure 1D, but my calculated ratio goes down in KCl, as expected, but not as reported.

To make sure I was not doing something wrong, I was going to repeat my calculations with the formulae in Table 1, which made me realize both are incorrect. The stated formula for the cell surface area-2*pi*RL-only represents to surface area of the cylindrical side of the cells, not its hemispherical ends. And it is not even the correct formula for the surface area of the side, because that calls for L to be the length of the side (without the hemispherical ends) not the length of the cell (which includes the hemispherical ends). L here is stated to be cell length (which is what is normally measured in the field, and which is consistent with the reported length of control cells in Figure 1E being 14 µm). The formula for the volume of a capsule in the form use in Table 1 (volume of a cylinder of length L - the volume excluded from the hemispherical ends) is pi*R^2*L - (8-(4/3*pi))*R^3.

Given these problems, I think I spent too much time thinking about the rest of the paper, because all of the calculations, and perhaps their interpretations, need to be redone.*

AUTHOR RESPONSE: The surface area and volume equations for a cylinder with hemispherical ends used in our study and listed in our table are correct and widely used in other work with fission yeast cells (Navarro and Nurse, 2012; Pan et al., 2014; Facchetti et al., 2019; BayBay et al., 2020; and Miller et al., 2023). We write our equations with variables for cell length and radius because these are biologically relevant and measured parameters for fission yeast cells. Cell length (L) refers to the total tip-to-tip length of the cell, including the hemispherical ends, and radius (R) refers to half the measured cell width. We have revised the Methods section to clarify this definition and avoid ambiguity (Please see methods section “Cell geometry measurements”)

Additionally, SA or Vol calculations were performed using the length of each individual cell and the average cell radius of the population. We did not use mean cell length of the population for our calculations like the reviewer assumed in their “sanity check” above. Please see methods section “Cell geometry measurements”. We hope that these clarifications and text revisions improve transparency and reproducibility.

* Minor Points:

Strains should be identified by strain number is the text and figure legends.*

AUTHOR RESPONSE: For clarity and readability, we refer to strains by genotype in the main text and figure legends, which we believe is more informative for readers than strain numbers. All strain numbers corresponding to each genotype are provided in Table S1, ensuring traceability and reproducibility without compromising clarity in data presentation.*

In the Introduction, "Most cell control their size" should be "Most eukaryotic cell control their size".*

• *

AUTHOR RESPONSE: The text has been corrected as suggested.*

Reviewer #2 (Significance (Required)):

Nothing to add.*

*Reviewer #3 (Evidence, reproducibility and clarity (Required)):

Summary This manuscript reports that fission yeast cells exhibit distinct cell size and geometry when exposed to osmotic, oxidative, or low-glucose stress. Based on quantitative measurements of cell length and width, the authors propose that different stress conditions trigger specific 'geometric adaptation' patterns, suggesting that cell size homeostasis is flexibly modulated depending on environmental cues. The study provides phenotypic evidence that multiple environmental stresses lead to distinct outcomes in the balance between cell surface area and volume, which the authors interpret as stress-specific modes of size control.

Major comments 1) The authors define the 48-hour time point as the 'long-term response', but no justification is provided for why 48 hours represents a physiologically relevant adaptation phase. It is unclear whether the size-control mode has stabilized by that time, or whether it may continue to change afterward. At minimum, the authors should provide a rationale (e.g., growth recovery dynamics, transcriptional adaptation plateau, or pilot time-course observations) to demonstrate that 48 hours corresponds to the steady-state adaptive phase rather than an arbitrarily selected time point.*

AUTHOR RESPONSE: We thank the reviewer for this important point and agree that the definition of the long-term response should be clarified. We have addressed this with new experiments and revised text. We now incorporate growth curve data and doubling time analyses for all yeast strains grown under control and stress conditions (See new Figure S3). These analyses show that following an initial transient stress-induced cell cycle delay, growth rates stabilize well before 48 hours. Notably, the slowest growth rate observed was in 1M KCl, with a doubling time of ~4 hours across all yeast strains tested. Thus, by 48 hours, cells in this condition have undergone more than 12 generations of growth, while cells in all other conditions with shorter doubling times have undergone even more divisions. So by allowing cells to grow for 48 hours prior to imaging, we are capturing cells that have resumed sustained cell cycle progression following transient stress-induced cell cycle delays. Because cell size control is tightly linked to the cell cycle, we define 48 hours as a physiologically relevant time point where cells have adapted to stress conditions.

Our revised methods now states: “Cultures were incubated at 25°C while shaking at 180 rpm for 48 h prior to imaging. This time point was chosen to ensure that cells had progressed beyond the initial transient stress response and reached a stable, condition-specific growth state, as confirmed by growth curve and doubling time analyses showing stabilization well before 48 h (Figure S3), including in the slowest growing condition (1 M KCl; doubling time ~4 h).”

* 2*）Related to the above comment, the authors propose that different stresses lead to distinct cell size adaptations, yet the rationale for the chosen stress intensities and exposure times is insufficiently described. It remains unclear whether the osmotic, oxidative, and low-glucose conditions used here induce comparable levels of cellular stress. Dose-response and time-course analyses would greatly strengthen the conclusions. Without such analyses, it is difficult to support the interpretation that geometry modulation represents a direct adaptive response.

AUTHOR RESPONSE: * *We selected the specific stress conditions based on previously published work showing that these doses elicit robust responses while preserving overall cell viability and the capacity for recovery. We note that osmotic, oxidative, and low glucose conditions perturb fundamentally different cellular systems (turgor pressure and cell wall mechanics, redox balance, and metabolism etc.) and therefore do not generate directly comparable levels of cellular stress in a quantitative sense. Our goal was not to equalize stress intensity across conditions, but to examine how cells change their geometry in response to distinct classes of stressors.

We have clarified the rationale for specific stress conditions in the revised methods: “These stress intensities were selected based on prior studies demonstrating robust cellular responses while preserving cell viability and the capacity for recovery (Fantes and Nurse, 1977, Shiozaki and Russell, 1995, Degols, et al., 1996; López-Avilés et al., 2008; Sansó et al., 2008; Satioh et al., 2015, Salat-Canela et al., 2021, Bertaux et al., 2023).”

* 3) The authors describe stress-induced size changes as an 'adaptive' response. While this is an appealing hypothesis, the presented data do not demonstrate that the change in cell size itself confers a fitness advantage. Evidence showing that blocking the size change reduces stress survival-or that the altered size improves growth recovery- would be required to support this claim. Without such data, the use of the term 'geometric adaptation' seems overstated.*

AUTHOR RESPONSE: We have revised the text to remove the term “adaptive” and now describe stress-induced size changes in descriptive terms. As discussed further in response to Comment 4, new growth curve and doubling time analyses show that defects in surface area or volume expansion do not uniformly impair growth or survival over the stress exposure examined here, reinforcing the decision to avoid fitness-based language.*

4) The authors conclude that mutants exhibit no major defects in growth or viability during 48-hour stress exposure based on comparable septation index values (Fig. S2). However, septation index alone does not fully capture growth performance or cell-cycle progression and is not sufficient to support claims regarding fitness or robustness of proliferation. If the authors intend to make statements about 'growth', 'viability', or 'cell-cycle progression', additional quantitative measures (e.g., growth curves, doubling time, colony-forming units, or microcolony growth measurements) would be necessary. Alternatively, the claims should be toned down to align with the measurements currently provided.*

AUTHOR RESPONSE: We have addressed this concern with new experiments and revised text. In addition to septation index measurements (now analyzed using chi-square tests of proportions; Figure S2), we performed growth curve experiments and doubling time analyses for all genotypes under control and stress conditions (new Figure S3). These additional data show that growth rates are largely comparable across genotypes in control, oxidative, and low-glucose conditions, with more pronounced genotype-dependent differences emerging under osmotic stress. Defects in surface area or volume expansion did not uniformly correspond to impaired population growth, indicating that geometric remodeling is not strictly required for proliferation over the 48-hour stress exposure examined here. We have refined our conclusion to emphasize that defects in surface area or volume expansion do not uniformly impair growth or survival. See revised Results text under the heading “Defects in surface area or volume expansion do not uniformly compromise growth or survival”.*

5) Related to the above comment, the manuscript does not adequately rule out the possibility that the decreased division size simply results from slower growth or delayed cell-cycle progression rather than a shift in the size-control mechanism. Measurements and normalizations of growth rate are required; without them, the interpretation remains speculative.*

AUTHOR RESPONSE: We agree that changes in growth rate or altered cell cycle timing are important to consider. We have revised our text: “Changes in growth rate or cell cycle progression under stress may influence division size by altering mitotic regulator accumulation. Future studies measuring mitotic regulator dynamics alongside growth rates will be needed to distinguish direct changes in size control mechanisms from growth- or timing-dependent effects.”

* 6) Regarding the phenotypes of wee1-2x cells, it is interesting that they increase the SA:Vol ratio under all stress conditions and show phenotypes distinct from cdr2Δ cells. From these observations, the authors claims that Cdr2 and Wee1 function as a surface-area-sensing module that complements the volume-sensing and time-sensing pathways to maintain geometric homeostasis. To support this interpretation, the authors could consider additional experiments, such as analyzing cdr2Δ + wee1-2x cells under the same stress conditions. Such data would test whether increased Wee1 can rescue or modify the cdr2Δ phenotype, providing functional evidence for the proposed Cdr2-Wee1-Cdk1 regulatory relationship. Measurements of cell length, width, SA:Vol ratio, and, if feasible, Cdk1 activity markers in the strain would greatly strengthen the mechanistic claims.*

AUTHOR RESPONSE: We thank the reviewer for this insightful suggestion. While analysis of a cdr2Δ wee1-2x strain could provide additional mechanistic detail, such experiments address a distinct question beyond the scope of our current study, which focuses on how cell geometry changes under different stress conditions in cells with perturbed surface area-, volume-, or time-sensing pathways. Our conclusions regarding a surface area-sensing role for Cdr2-Wee1 signaling are based on previous studies (Pan et al., 2014; Facchetti et al., 2019; Miller et al., 2023) and the cell geometry phenotypes we observe of cdr2Δ and wee1-2x cells under stress conditions. *

Minor comments 1) The manuscript focuses on adaptation through changes in the surface-to-volume ratio; however, only the ratio is shown. Presenting the underlying values of surface area and volume would clarify which geometric parameter primary contributes to the observed changes.*

AUTHOR RESPONSE: Please see our response to Reviewer 1 major comment 1.*

*2) Statistical analysis for Fig.S2 should be provided.

AUTHOR RESPONSE: We have completed this. See revised Figure S2 and methods.*

3) The paper by Kellog and Levin 2022 is missing from the reference list.*

AUTHOR RESPONSE: Thank you for catching this. This reference has now been added. *

**Referees cross-commenting**

After reading the other reviewer's reports, I recognize that focal points differ, but they appear sequential rather than contradictory.

Reviewer 2 raises concerns regarding the surface area/volume calculations, which-if incorrect-would influence many of the quantitative conclusions. I agree that confirming the validity of these calculations (and recalculating if necessary) should be the top priority before evaluating the biological interpretations.

Reviewer 1 raises more mechanistic biological questions. These are certainly important, but in my view they depend on the robustness of the quantitative analysis highlighted by Reviewer 2.

Therefore, I regard the reports as complementary rather than conflicting. Once the analytical issue pointed out by Reviewer 2 is resolved, the field will be in a better position to assess the significance of the mechanistic points raised by Reviewer 1 (as well as those in my own report).

Reviewer #3 (Significance (Required)):

General assessment One of the major strengths of this manuscript is its quantitative, side-by-side comparison of multiple environmental stresses under a unified experimental and analytical framework. The authors provide well-controlled morphometric measurements, allowing direct comparison of geometry changes that would otherwise be difficult to evaluate across studies. The observation that different stress types generate distinct geometric outcomes is particularly intriguing and has the potential to stimulate new conceptual thinking in the field of size control. However, the strength of the conceptual conclusion is currently limited by several aspects of the experimental design and interpretation. In particular, it remains unclear whether the observed geometry changes represent active adaptive responses rather than non-specific consequences of prolonged or string stress exposure. Demonstrating whether geometry remodeling provides a fitness advantage, clarifying whether the changes reach a steady-state rather than reflecting slow drift over time, or identifying upstream stress pathways that govern the response would substantially strengthen the conceptual advance. Even if additional mechanistic or fitness-related data cannot be added, refining the interpretation so that it remains aligned with the present evidence will enhance the clarity, and impact of the study.

Advance Previous study - including the 2023 publication by the James B. Moseley group - established that fission yeast integrates distinct size-control pathways related to surface area, volume, and time under normal growth conditions. The present manuscript extends this line of work to stressed environments and argues that each stress condition elicits a distinct size-control pattern. To our knowledge, a systematic comparison of cell geometry across multiple stress types in the context of size-control pathways has not been reported, and this represents a potentially valuable conceptual advance. The advance is primarily phenomenological and conceptual rather than mechanistic: the work presents new correlation between stress types and geometry but does not yet elucidate the pathways governing these responses or demonstrate a functional advantage. With additional evidence - or with qualifiers ensuring that claims match the current data - the study could make an important contribution to understanding how cells integrate environmental cues into size-control strategies.

Audience Although the primary audience consists of researchers in the fields of cell growth, cell-cycle control, and stress responses in yeast, the conceptual contribution may interest broader fields such as growth homeostasis, metabolic adaptation, and pathological cell size changes in higher eukaryotes. Beyond yeast biology, the modular view of size regulation proposed here may inspire new investigations in stem cell biology, cancer research, and biotechnology where environmental adaptation and cell size are closely linked.

Expertise: nuclear morphology; cell morphology; cell growth; cell cycle; cytoskeleton*

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary

This manuscript reports that fission yeast cells exhibit distinct cell size and geometry when exposed to osmotic, oxidative, or low-glucose stress. Based on quantitative measurements of cell length and width, the authors propose that different stress conditions trigger specific 'geometric adaptation' patterns, suggesting that cell size homeostasis is flexibly modulated depending on environmental cues. The study provides phenotypic evidence that multiple environmental stresses lead to distinct outcomes in the balance between cell surface area and volume, which the authors interpret as stress-specific modes of size control.

Major comments

1) The authors define the 48-hour time point as the 'long-term response', but no justification is provided for why 48 hours represents a physiologically relevant adaptation phase. It is unclear whether the size-control mode has stabilized by that time, or whether it may continue to change afterward. At minimum, the authors should provide a rationale (e.g., growth recovery dynamics, transcriptional adaptation plateau, or pilot time-course observations) to demonstrate that 48 hours corresponds to the steady-state adaptive phase rather than an arbitrarily selected time point.

2）Related to the above comment, the authors propose that different stresses lead to distinct cell size adaptations, yet the rationale for the chosen stress intensities and exposure times is insufficiently described. It remains unclear whether the osmotic, oxidative, and low-glucose conditions used here induce comparable levels of cellular stress. Dose-response and time-course analyses would greatly strengthen the conclusions. Without such analyses, it is difficult to support the interpretation that geometry modulation represents a direct adaptive response.

3) The authors describe stress-induced size changes as an 'adaptive' response. While this is an appealing hypothesis, the presented data do not demonstrate that the change in cell size itself confers a fitness advantage. Evidence showing that blocking the size change reduces stress survival-or that the altered size improves growth recovery- would be required to support this claim. Without such data, the use of the term 'geometric adaptation' seems overstated.

4) The authors conclude that mutants exhibit no major defects in growth or viability during 48-hour stress exposure based on comparable septation index values (Fig. S2). However, septation index alone does not fully capture growth performance or cell-cycle progression and is not sufficient to support claims regarding fitness or robustness of proliferation. If the authors intend to make statements about 'growth', 'viability', or 'cell-cycle progression', additional quantitative measures (e.g., growth curves, doubling time, colony-forming units, or microcolony growth measurements) would be necessary. Alternatively, the claims should be toned down to align with the measurements currently provided.

5) Related to the above comment, the manuscript does not adequately rule out the possibility that the decreased division size simply results from slower growth or delayed cell-cycle progression rather than a shift in the size-control mechanism. Measurements and normalizations of growth rate are required; without them, the interpretation remains speculative.

6) Regarding the phenotypes of wee1-2x cells, it is interesting that they increase the SA:Vol ratio under all stress conditions and show phenotypes distinct from cdr2Δ cells. From these observations, the authors claims that Cdr2 and Wee1 function as a surface-area-sensing module that complements the volume-sensing and time-sensing pathways to maintain geometric homeostasis. To support this interpretation, the authors could consider additional experiments, such as analyzing cdr2Δ + wee1-2x cells under the same stress conditions. Such data would test whether increased Wee1 can rescue or modify the cdr2Δ phenotype, providing functional evidence for the proposed Cdr2-Wee1-Cdk1 regulatory relationship. Measurements of cell length, width, SA:Vol ratio, and, if feasible, Cdk1 activity markers in the strain would greatly strengthen the mechanistic claims.

Minor comments

1) The manuscript focuses on adaptation through changes in the surface-to-volume ratio; however, only the ratio is shown. Presenting the underlying values of surface area and volume would clarify which geometric parameter primary contributes to the observed changes.

2) Statistical analysis for Fig.S2 should be provided.

3) The paper by Kellog and Levin 2022 is missing from the reference list.

Referees cross-commenting

After reading the other reviewer's reports, I recognize that focal points differ, but they appear sequential rather than contradictory.

Reviewer 2 raises concerns regarding the surface area/volume calculations, which-if incorrect-would influence many of the quantitative conclusions. I agree that confirming the validity of these calculations (and recalculating if necessary) should be the top priority before evaluating the biological interpretations.

Reviewer 1 raises more mechanistic biological questions. These are certainly important, but in my view they depend on the robustness of the quantitative analysis highlighted by Reviewer 2.

Therefore, I regard the reports as complementary rather than conflicting. Once the analytical issue pointed out by Reviewer 2 is resolved, the field will be in a better position to assess the significance of the mechanistic points raised by Reviewer 1 (as well as those in my own report).

Significance

General assessment

One of the major strengths of this manuscript is its quantitative, side-by-side comparison of multiple environmental stresses under a unified experimental and analytical framework. The authors provide well-controlled morphometric measurements, allowing direct comparison of geometry changes that would otherwise be difficult to evaluate across studies. The observation that different stress types generate distinct geometric outcomes is particularly intriguing and has the potential to stimulate new conceptual thinking in the field of size control. However, the strength of the conceptual conclusion is currently limited by several aspects of the experimental design and interpretation. In particular, it remains unclear whether the observed geometry changes represent active adaptive responses rather than non-specific consequences of prolonged or string stress exposure. Demonstrating whether geometry remodeling provides a fitness advantage, clarifying whether the changes reach a steady-state rather than reflecting slow drift over time, or identifying upstream stress pathways that govern the response would substantially strengthen the conceptual advance. Even if additional mechanistic or fitness-related data cannot be added, refining the interpretation so that it remains aligned with the present evidence will enhance the clarity, and impact of the study.

Advance

Previous study - including the 2023 publication by the James B. Moseley group - established that fission yeast integrates distinct size-control pathways related to surface area, volume, and time under normal growth conditions. The present manuscript extends this line of work to stressed environments and argues that each stress condition elicits a distinct size-control pattern. To our knowledge, a systematic comparison of cell geometry across multiple stress types in the context of size-control pathways has not been reported, and this represents a potentially valuable conceptual advance. The advance is primarily phenomenological and conceptual rather than mechanistic: the work presents new correlation between stress types and geometry but does not yet elucidate the pathways governing these responses or demonstrate a functional advantage. With additional evidence - or with qualifiers ensuring that claims match the current data - the study could make an important contribution to understanding how cells integrate environmental cues into size-control strategies.

Audience

Although the primary audience consists of researchers in the fields of cell growth, cell-cycle control, and stress responses in yeast, the conceptual contribution may interest broader fields such as growth homeostasis, metabolic adaptation, and pathological cell size changes in higher eukaryotes. Beyond yeast biology, the modular view of size regulation proposed here may inspire new investigations in stem cell biology, cancer research, and biotechnology where environmental adaptation and cell size are closely linked.

Expertise: nuclear morphology; cell morphology; cell growth; cell cycle; cytoskeleton.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

Cabral et al. present a analysis of the effects of environmental stress of cellular geometry in the fission yeast S. pombe. The stresses they study-oxidative, osmotic and nutritional-have previously been shown to affect cell size in fission yeast. Here, the authors do a more sophisticated analysis, measuring surface area as well as volume (for which length had previously been used as a proxy, assuming fission yeast cells are cylinders of constant width). In addition, they investigate the effect of mutations in three cell-cycle control proteins that have been proposed to regulate cell geometry: Cdc13, Cdc25 and Cdr2. It is an interesting study that could provide insight into cell-size control and environmental-stress response in fission yeast. However, I have serious concerns about the analysis of the data. In fact, as I was writing up my concerns, I noticed that the formulae in Table 1 for surface area and volume are incorrect, so the whole paper appears to require reanalysis.

One general problem is that the authors seem to confuse statistical significance with biological significance. They claim that both oxidative and osmotic stress cause a reduction in SA/V ratio. For oxidative stress, the difference is evident, but the control and KCl-treated cells look to have indistinguishable distributions. Perhaps there is a significant statistical difference between the, but I am skeptical. (I would ask for the data table to try out the stats myself, but given the revelation below that the number will all need to be recalculated, that point is moot). In any case, the difference is certainly not biologically significant.

Likewise, in Figure 2B, they interpret tiny changes in the SA/V. By my estimation, the difference between control and osmotic stress is only 2% (1.195/1.17), less that the wild-type case, which appears to be twice that (which is still pretty modest). The small amplitude of these changes is obscured by the fact that the graphs do not have a baseline at zero, which, as a matter of good data-presentation practice, they should.

I am also concerned about the use of manual measurement of width at a single point along the cell. This approach is very sensitive to the choice of width point and to non-cylindrical geometries, several of which are evident in the images presented. MATLAB will return the ??? as well as the length from a mask, but even better, one can more accurately calculate the surface area and volume by assuming rotational symmetry of the mask. Given that surface area and volume calculation need to be redone anyway, as discussed below, I encourage the authors to calculate them directly from the mask, instead of using the cylindrical assumption.

The authors also need to be more careful about their claims about size-dependent scaling. The concentration of both Cdc13 and Cdc25 scale with size (perhaps indirectly, in the case of Cdc13), but Cdr2 does not. Cdr2 activity has been proposed to scale with size, and its density at cortical nodes has been reported to scale with size, although that claim has been challenged <https://pubmed.ncbi.nlm.nih.gov/36093997>.

Even taking the authors approach at face value, there are observations that do not seem to make sense, which led me to realize that the wrong formulae were used to calculate surface area and volume.

In Figure 1E,F, the KCl-treated cells get shorter and wider; surely, that should result in a lower SA/V ratio. However, as noted above, in Figure 1D, they are shown to have a similar ratio. As a sanity check, I eye-balled the numbers off of the figure (control: 14 µm x 3.6 µm and KCl: 11 µm x 3.8 µm) and calculated their surface area and volume using the formula for a capsule (i.e., a cylinder with hemispheric ends).

SA = the surface area of the two hemispheres + the surface are of the cylinder in between = 4pi(width/2)^2 + piwidth(length-width), the length-width term calculates the side length of the capsule (length without the hemispheres) from the full length of the capsule (length including the hemispheres)

V = the volume of the two hemispheres + the volume of the cylinder in between = 4/3pi(width/2)^3 + pi(width/2)^2(length-width).

I got SA/V ratios of around 2, which are way off from what is presented in Figure 1D, but my calculated ratio goes down in KCl, as expected, but not as reported.

To make sure I was not doing something wrong, I was going to repeat my calculations with the formulae in Table 1, which made me realize both are incorrect. The stated formula for the cell surface area-2piRL-only represents to surface area of the cylindrical side of the cells, not its hemispherical ends. And it is not even the correct formula for the surface area of the side, because that calls for L to be the length of the side (without the hemispherical ends) not the length of the cell (which includes the hemispherical ends). L here is stated to be cell length (which is what is normally measured in the field, and which is consistent with the reported length of control cells in Figure 1E being 14 µm). The formula for the volume of a capsule in the form use in Table 1 (volume of a cylinder of length L - the volume excluded from the hemispherical ends) is piR^2L - (8-(4/3pi))R^3.

Given these problems, I think I spent too much time thinking about the rest of the paper, because all of the calculations, and perhaps their interpretations, need to be redone.

Minor Points:

Strains should be identified by strain number is the text and figure legends.

In the Introduction, "Most cell control their size" should be "Most eukaryotic cell control their size".

Significance

Nothing to add.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Using genetics and microscopy approaches, Cabral et al. investigate how fission yeast regulates its length and width in response to osmotic, oxidative, or low glucose stress. Miller et al. have recently found that the cell cycle regulators Cdc25, Cdc13 and Cdr2 integrate information about cell volume, time and cell surface area into the cellular decision when to divide. Cabral now build on this work and test how disruption of these regulators affects cell size adaptation. They find that each stress condition shows a distinct dependence on the individual regulators, suggesting that the complex size control network enables optimized size adaptation for each condition. Overall, the manuscript is clear and the detailed methods ensure that the experiments can be replicated.

Major comments:

1. It would be much easier to follow the authors' conclusions, if in addition to surface area to volume ratio, length and width, they would also plot cell volume at division in Figs. 1-4.
2. To me, it seems that maybe even more than upon osmotic stress, the cdc13-2x strain differs qualitatively from WT in low glucose conditions, where the increased SA-V ratio is almost completely abolished.
3. It is not entirely clear to me why two copies of Cdc13 would qualitatively affect the responses. Shouldn't the extra copy behave similarly to the endogenous one and therefore only lead to quantitative changes? Maybe the authors can discuss this more clearly or even test a strain in which Cdc13 function is qualitatively disrupted.
4. I don't see why the authors come to the conclusion that under osmotic stress cells would maximize cell volume. It leads to a decreased cell length, doesn't it?

Significance

Fission yeast has long been used as a model for eukaryotic cell size regulation. So far, this research has been mostly focused on steady state size regulation. While it has long been clear that cells across organisms adapt their size in response to environmental changes, little is known about how these external inputs are processed through the size control network. Dissecting how disruption of the various branches of the size control network affects size adaptation is an important step towards a mechanistic understanding of this process. Future studies will have to build on these observations and investigate how each stress mechanistically affects the respective regulator(s). While the details of the molecular players and their contribution to size adaptation are likely specific to fission yeast, the concept of stress type-specific size adaptation that is mediated through different regulators is likely conserved and thus of broader relevance.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

We sincerely appreciate the feedback, attention to detail and timeliness of the referees for our manuscript. Below, we provide a point-by-point response to all comments from the referees, detailing the changes we have already made, and those that are in progress. Referee's comments will appear in bolded text, while our responses will be unbolded. Any text quoted directly from the manuscript will be italicised and contained within "quotation marks". Additionally, we have grouped all comments into four categories (structural changes, minor text changes, experimental changes, figure changes), comments are numbered 1-n in each of these categories. Please note: this response to reviewer's comments included some images that cannot be embedded in this text-only section.

1. General Statements

We appreciate the overall highly positive and enthusiastic comments from all reviewers, who clearly appreciated the technical difficulty of this study, and noted amongst other things that this study represents" a major contribution to the future advancement of oocyst-sporozoite biology" and the development of the segmentation score for oocysts as a "major advance[ment]". We apologise for the omission of line numbers on the document sent to reviewers, we removed these for the bioRxiv submission without considering that this PDF would be transferred across to Review Commons.

We have responded to all reviewers comments through a variety of text changes, experimental inclusions, or direct query response. Significant changes to the manuscript since initial submission are as follows:

1. Refinement of rhoptry biogenesis model: Reviewers requested more detail around the content of the AORs, which we had previously suggested were a vehicle for rhoptry biogenesis as we saw they carried the rhoptry neck protein RON4. To address this, we first attempted to address this using antibodies against rhoptry bulb proteins but were unsuccessful. We then developed a * berghei* line where there rhoptry bulb protein RhopH3 was GFP-tagged. Using this parasite line, we observed that the earliest rhoptry-like structure, which we had previously interpreted as an AOR contained RhopH3. By contrast, RhopH3 was absent from AORs. Reflecting these observations we have renamed this initial structure the 'pre-rhoptry' and suggested a model for rhoptry biogenesis where rhoptry neck cargo are trafficked via the AOR but rhoptry bulb cargo are trafficked by small vesicles that move along the rootlet fibre (previously observed by EM).
2. Measurement of rhoptry neck vs bulb: While not directly suggested by the reviewers, we have also included an analysis that estimates the proportion of the sporozoite rhoptry that represents the rhoptry neck. By contrast to merozoites, which we show are overwhelmingly represented by the rhoptry bulb, the vast majority of the sporozoite rhoptry represents the rhoptry neck.
3. Measurement of subpellicular microtubules: One reviewer asked if we could measure the length of subpellicular microtubules where we had previously observed that they were longer on one side of the sporozoite than the other. We have now provided absolute and relative (% sporozoite length) length measurements for these subpellicular microtubules and also calculated the proportion of the microtubule that is polyglutamylated.
4. More detailed analysis of RON11cKD rhoptries: Multiple comments suggested a more detailed analysis of the rhoptries that were formed/not formed in RON11cKD We have included an updated analysis that shows the relative position of these rhoptries in sporozoites.

2. Point-by-point description of the revisions

Reviewer #1

Minor text changes (Reviewer #1)

1. __Text on page 12 could be condensed to highlight the new data of ron4 staining of the AOR. __

We agree with the reviewer that it is a reasonable suggestion. After obtaining additional data on the contents of the AOR (as described in General Statements #1), this section has been significantly rewritten to highlight these findings. 2.

__Add reference on page 3 after 'disrupted parasites' __

This sentence has been rewritten slightly with some references included and now reads: "Most data on these processes comes from electron microscopy studies 6-8, with relatively few functional reports on gene deleted or disrupted parasites9-11. 3.

__Change 'the basal complex at the leading edge' - this seems counterintuitive __

This change has been made. 4.

__Change 'mechanisms underlying SG are poorly' - what mechanisms? of invasion or infection? __

This was supposed to read "SG invasion" and has now been fixed. 5.

__On page 4: 'handful of proteins' __

This error has been corrected. 6.

__What are the 'three microtubule spindle structures'? __

The three microtubule spindle structures: hemispindle, mitotic spindle, and interpolar spindle are now listed explicitly in the text. 7.

__On page 5: 'little is known' - please describe what is known, also in other stages. At the end of the paper I would like to know what is the key difference to rhoptry function in other stages? __

The following sentence already detailed that we had recently used U-ExM to visualise rhoptry biogenesis in blood-stage parasites, but the following two sentences have been added to provide extra detail on these findings: "In that study, we defined the timing of rhoptry biogenesis showing that it begun prior to cytokinesis and completed approximate coincident with the final round of mitosis. Additionally, we observed that rhoptry duplication and inheritance was coupled with centriolar plaque duplication and nuclear fission." 8.

__change 'rhoptries golgi-derived, made de novo' __

This has been fixed. 9.

__change 'new understand to' __

This change has been made 10.

__'rhoptry malformations' seem to be similar in sporozoites and merozoites. Is that surprising/new? __

We assume this is in reference to mention of "rhoptry malformations" in the abstract. In the RON11 merozoite study (PMID:39292724) the authors noted no gross rhoptry malformations, only that one was not formed/missing. The abstract sentence has been changed to the following to better reflect this nuance: "*We show that stage-specific disruption of RON11 leads to a formation of sporozoites that only contain half the number of rhoptries of controls like in merozoites, however unlike in merozoites the majority of rhoptries appear grossly malformed."

* 11.

__What is known about crossing the basal lamina. Where rhoptries thought to be involved in this process? Or is it proteins on the surface or in other secretory organelles? __

We are unaware of any studies that specifically look at sporozoites crossing the SG basal lamina. A review, although now ~15 years old stated that "No information is available as to how the sporozoites traverse the basal lamina" (PMID:19608457) and we don't know any more information since then. To try and better define our understanding of rhoptry secretion during SG invasion, we have added the following sentence:

"It is currently unclear precisely when during these steps of SG invasion rhoptry proteins are required, but rhoptry secretion is thought to begin before in the haemolymph before SG invasion16." 12.

__On page change/specify: 'wide range of parasite structures' __

The structures observed have been listed: centriolar plaque, rhoptry, apical polar rings, rootlet fibre, basal complex, apicoplast. 13.

__On page 7: is Airyscan2 a particular method or a specific microscope? __

Airyscan2 is a detector setup on Zeiss LSM microscopes, this was already detailed in the materials and methods sections, but figure legends have been clarified to read: "...imaged by an LSM900 microscopy with an Airyscan2 detector". 14.

__how large is RON11? __

RON11 is 112 kDa in * berghei*, as noted in the text. 15.

__There is no causal link between ookinete invasion and oocyst developmental asynchrony __

We have deleted the sentence that implied that ookinete invasion was responsible for oocyst asynchrony. This section now simply states that "Development of each oocyst within a midgut is asynchronous..." 16.

__First sentence of page 24 appears to contradict what is written in results____ I don't understand the first two sentences in the paragraph titled Comparison between Plasmodium spp __

This sentence was worded confusingly, making it appear contradictory when that was not the intention. The sentence has been changed to more clearly support what is written in the discussion and now reads: "Our extensive analysis only found one additional ultrastructural difference between Plasmodium spp."

__On page 25 or before the vast number of electron microscopy studies should be discussed and compared with the authors new data. __

It is not entirely clear which new data should be specifically discussed based on this comment. However, we have added a new paragraph that broadly compares MoTissU-ExM and our findings with other imaging methods previously used on mosquito-stage malaria parasites:

"*Comparison of MoTissU-ExM and other imaging modalities

Prior to the development of MoTissU-ExM, imaging of mosquito-stage malaria parasites in situ had been performed using electron microscopy7,8,11,28, conventional immunofluorescence assays (IFA)10, and live-cell microscopy25. MoTissU-ExM offers significant advantages over electron microscopy techniques, especially volume electron microscopy, in terms of accessibility, throughput, and detection of multiple targets. While we have benchmarked many of our observations against previous electron microscopy studies, the intracellular detail that can be observed by MoTissU-ExM is not as clear as electron microscopy. For example, previous electron microscopy studies have observed Golgi-derived vesicles trafficking along the rootlet fibre8 and distinguished the apical polar rings44; both of which we could not observe using MoTissU-ExM. Compared to conventional IFA, MoTissU-ExM dramatically improves the number and detail of parasite structures/organelles that can be visualised while maintaining the flexibility of target detection. By contrast, it can be difficult or impossible to reliably quantify fluorescence intensity in samples prepared by expansion microscopy, something that is routine for conventional IFA. For studying temporally complex processes, live-cell microscopy is the 'gold-standard' and there are some processes that fundamentally cannot be studied or observed in fixed cells. We attempt to increase the utility of MoTissU-ExM in discerning temporal relationships through the development of the segmentation score but note that this cannot be applied to the majority of oocyst development. Collectively, MoTissU-ExM offers some benefits over these previously applied techniques but does not replace them and instead serves as a novel and complementary tool in studying the cell biology of mosquito-stage malaria parasites.**"

*

__First sentence on page 27: there are many studies on parasite proteins involved in salivary gland invasion that could be mentioned/discussed. __

The sentence in question is "To the best of our knowledge, the ability of sporozoites to cross the basal lamina and accumulate in the SG intercellular space has never previously been reported."

This sentence has now been changed to read as follows: "While numerous studies have characterized proteins whose disruption inhibited SG invasion9,10,15,59-63, to the best of our knowledge the ability of sporozoites to cross the basal lamina and accumulate in the SG intercellular space has never previously been reported ."

__On page 10 I suggest to qualify the statement 'oocyst development has typcially been inferred by'. There seem a few studies that show that size doesn't reflect maturation. __

In our opinion, this statement is already qualified in the following sentence which reads: "Recent studies have shown that while oocysts increase in size initially, their size eventually plateaus (11 days pot infection (dpi) in P. falciparum4)."

__On page 16 the authors state that different rhoptries might have different function. This is an interesting hypothesis/result that could be mentioned in the abstract. __

The abstract already contains the following statement: "...and provide the first evidence that rhoptry pairs are specialised for different invasion events." We see this as an equivalent statement.

Experimental changes (Reviewer #1)

1. On page 19: do the parasites with the RON11 knockout only have the cytoplasmic or only the apical rhoptries?

The answer to this is not completely clear. We have added the following data to Figures 6 and 8 where we quantify the proportion of rhoptries that are either apical or cytoplasmic: In both wildtype parasites and RON11ctrl parasites, oocyst spz rhoptries are roughly 50:50 apical:cytoplasmic (with a small but consistent majority apical), while almost all rhoptries are found at the apical end (>90%) in SG spz. Presumably, after the initial apical rhoptries are 'used up' during SG invasion, the rhoptries that were previously cytoplasmic take their place. In RON11cKD the ratio of apical:cytoplasmic rhoptries is fairly similar to control oocyst spz. In RON11cKD SG spz, the proportion of cytoplasmic rhoptries decreases but not to the same extent as in wildtype or RON11Ctrl. From this, we infer that the two rhoptries that are lost/not made in RON11cKD sporozoites are likely a combination of both the apical and cytoplasmic rhoptries we find in control sporozoites.

__in panel G: Are the dense granules not micronemes? What are the dark lines? Rhoptries?? __

We have labelled all of Figure 1 more clearly to point out that the 'dark lines' are indeed rhoptries. Additionally, we have renamed the 'protein-dense granules' to 'protein-rich granules', as it seems we are suggesting that these structures are dense granules the secretory organelle. At this stage we simply do not know what all of these granules are. The observation that some but not all of these granules contain CSP (Supplementary Figure 2) suggests that they may represent heterogenous structures. It is indeed possible that some are micronemes, however, we think it is unlikely that they are all micronemes for a number of reasons: (1) micronemes are not nearly this protein dense in other Plasmodium lifecycle stages, (2) some of them carry CSP which has not been demonstrated to be micronemal, (3) very few of these granules are present in SG sporozoites, which would be unexpected because microneme secretion is required for hepatocyte invasion.

__Figure 2 seems to add little extra compared to the following figures and could in my view go to the supplement. __

We agree that Figure 2b adds little and so have moved that to Supplementary Figure 2, but think that the relative ease at which it can be distinguished if sporozoites are in the secretory cavity or SG epithelial cell is a key observation because of the difficulty in doing this by conventional IFA.

__On page 8 the authors mention a second layer of CSP but do not further investigate it. It is likely hard to investigate this further but to just let it stand as it is seems unsatisfactory, considering that CSP is the malaria vaccine. What happens if you add anti-CSP antibodies? I would suggest to shorten the opening paragraphs of this paper and to focus on the rhoptries. This could be done be toning down the text on all aspects that are not rhoptries and point to the open question some of the observations such as the CSP layers raise for future studies. __

When writing the manuscript, we were unsure whether to include this data at all as it is a purely incidental finding. We had no intention of investigating CSP specifically, but anti-CSP antibodies were included in most of the salivary gland imaging experiments so we could more easily find sporozoites. Given the tremendous importance of CSP to the field, we figured that these observations were potentially important enough that they should be reported in the literature even though they are not something we have the intention or resources to investigate subsequently. Additionally, after consultation with other microscopists we think there is a reasonable chance that this double-layer effect could be a product of chemical fixation. To account for this, we have qualified the paragraph on CSP with this sentence:

"We cannot determine if there is any functional significance of this second CSP layer and considering that it has not been observed previously it may well represent an artefact of chemical (paraformaldehyde) fixation."

__Maybe include more detail of the differences between species on rhoptry structure into Figure 4. I would encourage to move the Data on rhoptries in Figure S6 to the main text ie to Figure 4. __

We have moved the images of developing rhoptries in * falciparum *(previously Figure S6a and b) into figure 4, which now looks as follows:

Figure S8 (previously S6c) now consists only of the MG spz rhoptry quantification

Manuscript structural changes (Reviewer #1)

1. Abstract: don't focus on technique but on the questions you tried to answer (ie rewrite or delete the 3rd and 4th sentence)

2. 'range of cell biology processes' - I understand the paper that the key discovery concerns rhoptry biogenesis and function, so focus on that, all other aspects appear rather peripheral.

3. 'Much of this study focuses on the secretory organelles': I would suggest to rewrite the intro to focus solely on those, which yield interesting findings.

4. Page 11: I am tempted to suggest the authors start their study with Figure 3 and add panel A from Figure 2 to it. This leads directly to their nice work on rhoptries. Other features reported in Figures 1 and 2 are comparatively less exciting and could be moved to the supplement or reported in a separate study.____ Page 23: I suggest to delete the first sentence and focus on the functional aspects and the discoveries.

5. __Maybe add a conclusion section rather than a future application section, which reads as if you want to promoted the use of ultrastructure expansion microscopy. To my taste the technological advance is a bit overplayed considering the many applications of this techniques over the last years, especially in parasitology, where it seems widely used. In any case, please delete 'extraordinarily' __

Response to Reviewer#1 manuscript structural changes 1-5: This reviewer considers the findings related to rhoptry biology as the most significant aspect of the study and suggests rewriting the manuscript to emphasize these findings specifically. Doing so might make the key findings easier to interpret. However, in our view, this approach could misrepresent how the study originated and what we see as the most important outcomes. We did not develop MoTissU-ExM specifically to investigate rhoptry biology. Instead, this technique was created independently of any particular biological question, and once established, we asked what questions it could answer, using rhoptry biology as a proof of concept. Given the authors' previous work and available resources, we chose to focus on rhoptry biology. Since this was driven by basic research rather than a specific hypothesis, it's important to acknowledge this in the manuscript. While we agree that the findings related to rhoptry biology are valuable, we believe that highlighting the technique's ability to observe organelles, structures, and phenotypes with unprecedented ease and detail is more important than emphasizing the rhoptry findings alone. For these reasons, we have decided not to restructure the manuscript as suggested.

Reviewer #2

Minor text changes (Reviewer #2)

1. __The 'image Z-depth' value indicated in the figures is ambiguous. It is not clear whether this refers to the distance from the coverslip surface or the starting point of the z-stack image acquisition. A precise definition of this parameter would be beneficial. __

In the legend of Figure 1, the image Z-depth has been clarified as "sum distance of Z-slices in max intensity projection". 2.

__Paragraph 3 of the introduction - line 7, "handful or proteins" should be handful of proteins __

This has been corrected. 3.

__Paragraph 5 of the introduction - line 7, "also able to observed" should be observe __

This has been changed. 4.

__In the final paragraph of the introduction - line 1, "leverage this new understand" should be understanding __

This has been fixed. 5.

__The first paragraph of the discussion summary contains an incomplete sentence on line 7, "PbRON11ctrl-infected SGs." __

This has been removed. 6.

__The second paragraph of the discussion - line 10, "until cytokinesis beings" should be begins __

This mistake has been corrected. 7.

__One minor point that author suggest that oocyst diameter is not appropriate for the development of sporozoite develop. This is not so true as oocyst diameter tells between cell division and cell growth so it is important parameter especially where the proliferation with oocyst does not take place but the growth of oocyst takes place. __

We agree that this was not highlighted enough in the text. The final sentence of the results section about this now reads:

"While diameter is a useful readout for oocyst development in the early stages of its growth, this suggests that diameter is a poor readout for oocyst development once sporozoite formation has begun and highlights the usefulness of the segmentation score as an alternative.", and the final sentence of the discussion section about this now reads "Considering that oocyst size does not plateau until cytokinesis begins4, measuring oocyst diameter may represent a useful biological clock specifically when investigating the early stages of oocyst development." 8.

__How is the apical polarity different to merozoite as some conoid genes are present in ookinete and sporozoite but not in merozoite. __

Our hypothesis is that apical polarity is established by the positioning and attachment of the centriolar plaque to the parasite plasma membrane in both forming merozoites and sporozoites. While the apical polar ring proteins are obviously present at the apical end, and have important functions, we think that they themselves are unlikely to regulate polarity establishment directly. Additionally, it seems that the apical polar rings are visible in forming sporozoites far before the comparable stages of merozoite formation. An important note here is that at this point, this is largely inferences based on observational differences and there is relatively little functional data on proteins that regulate polarity establishment at any stage of the Plasmodium 9.

__Therefore, I think that electron microscopy remains essential for the observation of such ultra-fine structures __

We have added a paragraph in the discussion that provides a more clear comparison between MoTissU-ExM and other imaging modalities previously applied on mosquito-stage parasites (see response to Reviewer#1 (Minor text changes) comment #17). 10.

__The author have not mentioned that sometimes the stage oocyst development is also dependent on the age of mosquito and it vary between different mosquito gut even if the blood feed is done on same day. __

In our opinion this can be inferred through the more general statement that "development of each oocyst within a midgut is asynchronous..."

Figure changes (Reviewer #2)

1. __Fig 3B: stage 2 and 6 does not show the DNA cyan, it would-be good show the sate of DNA at that particular stage, especially at stage 2 when APR is visible. And box the segment in the parent picture whose subset is enlarged below it. __

We completely agree with the reviewer that the stage 2 image would benefit from the addition of a DNA stain. Many of the images in Figure 3b were done on samples that did not have a DNA stain and so in these * yoelii samples we did not find examples of all segmentation scores with the DNA stain. Examples of segmentation score 2 and 6 for P. berghei, and 6 for P. falciparum* can be found with DNA stains in Figure S8. 2.

__For clarity, it would be helpful to add indicators for the centriolar plaques in Figure 1b, as their locations are not immediately obvious. __

The CPs in Figure 1a and 1b have been circled on the NHS ester only panel for clarity. +

__Regarding Figure 1c, the authors state that 'the rootlet fiber is visible'. However, such a structure cannot be confirmed from the provided NHS ester image. Can the authors present a clearer image where the rootlet fibre is more distinct? Furthermore, please provide the basis for identifying this structure as a rootlet fiber based on the NHS ester observation alone. __

The image in Figure 1c has been replaced with one that more clearly shows the rootlet fibre.

Based on electron microscopy studies, the rootlet fibre has been defined as a protein dense structure that connects the centriolar plaque to the apical polar rings (PMID: 17908361). Through NHS ester and tubulin staining, we could identify the apical polar rings and centriolar plaque as sites on the apical end of the parasite and nucleus that microtubules are nucleated from. There is a protein dense fibre that connects these two structures. Based on the fact that the protein density of this structure was previously considered sufficient for its identification by electron microscopy, we consider its visualisation by NHS ester staining sufficient for its identification by U-ExM.

__Fig 1B - could the tubulin image in the hemispindle panel be made brighter? __

The tubulin staining in this panel was not saturated, and so this change has been made.

__Fig 4A - the green text in the first image panel is not visible. Also, the cyan text in the 3rd image in Fig 1A is also difficult to see. There's a few places where this is the case __

We have made all microscopy labels legible at least when printed in A4/Letter size.

__Fig 6A - how do the authors know ron11 expression is reduced by 99%? Did they test this themselves or rely on data from the lab that gifted them the construct? Also please provide mention the number of oocyst and sporozoites were observed. __

The way Figure 6a was previously designed and described was an oversight, that wrongly suggested we had quantified a >99% reduction in *ron11 * The 99% reduction has been removed from Figure 6a and the corresponding part of the figure legend has been rewritten to emphasise that this was previously established:

"(a) Schematic showing previously established Ron11Ctrl and Ron11cKD parasite lines where ron11 expression was reduced by >99%9."

As to the second part of the question, we did not independently test either protein or RNA level expression of RON11, but we were gifted the clonal parasite lines established by Prof. Ishino's lab in PMID: 31247198 not just the genetic constructs.

__Fig 6E - are the data point colours the wrong way round on this graph? Just looking at the graph it looks as though the RON11cKD has more rhoptries than the control which does not match what is said in the text. __

Thank you for pointing out this mistake, the colours have now been corrected.

__Fig S8C, PbRON11 ctrl, pie chart shows 89.7 % spz are present in the secretory cavity while the text shows 100 %, 35/35 __

The text saying 100% (35/35) only considered salivary glands that were infected (ie. Uninfected SGs were removed from the count. The two sentences that report this data have been clarified to reflect this better:

"Of *PbRON11ctrl SGs that were infected (35/39), 100% (35/35) contained sporozoites in the secretory cavity (Figure S8c). Conversely of infected PbRON11cKD SGs (59/82), only 24% (14/59) contained sporozoites within the secretory cavity (Figure S9d)."

*

__Fig S9D shows that RON11 ckd contains 17.1% sporozoites in secretory cavity while the text says 24%. __

Please see the response to Reviewer#2 Figure Changes Comment #8 where this was addressed.

Experimental changes (Reviewer #2)

1. __Why do the congruent rhoptries have similar lengths to each other, while the dimorphic rhoptries have different lengths? Is this morphological difference related to the function of these rhoptries? __

We hypothesise that this morphological difference arises because the congruent rhoptries are 'used' during SG invasion, while the dimorphic rhoptries are utilized during hepatocyte invasion. It is not straightforward to test this functionally at this point, as no protein is known to have differential localization between the two. Additionally, RON11 is likely directly involved in both SG and hepatocyte invasion through a secreted portion of the protein (as seen in RBC invasion). Therefore, RON11cKD sporozoites may have combined defects, meaning we cannot assume any defect is solely due to the absence of two rhoptries. Determining this functionally is of high interest to our research groups and remains an area of ongoing study, but it is beyond the scope of this study. 2.

Would it be possible to show whether RON11 localises to the dimorphic rhoptries, the congruent rhoptries, or both, by using expansion microscopy and a parasite line that expresses RON11 tagged with GFP or a peptide tag?

__ __We do not have access to a parasite line that expresses a tagged copy of RON11, or anti-PbRON11 antibodies. Based on previously published localisation data, however, it seems likely that RON11 localises to both sets of rhoptries. Below are excerpts from Figure 1c of PMID: 31247198, where RON11 (in green) seems to have a more basally-extended localisation in midgut (MG) sporozoites than in salivary gland (SG) sporozoites. From this we infer that in the MG sporozoite you're seeing RON11 in both pairs of rhoptries, but only the one remaining pair in the SG sporozoite.

__The knockdown of RON11 disrupts the rhoptry structure, making the dimorphic and congruent rhoptries indistinguishable. Does this suggest that RON11 is important for the formation of both types of rhoptries? I believe that it would be crucial to confirm whether RON11 localises to all rhoptries or is restricted to specific rhoptries for a more precise discussion of RON11's function. __

Based on our analysis, it does indeed seem that RON11 is important for both types of rhoptries as when RON11 isn't expressed sporozoites still have both apical and cytoplasmic rhoptries (ie. Not just one pair is lost; see Reviewer #1 Experimental changes comment #1).

__The authors state that 64% of RON11cKD SG sporozoites contained no rhoptries at all. Does this mean RON11cKD SG sporozoites used up all rhoptries corresponding to the dimorphic and congruent pairs during SG invasion? If so, this contradicts your claims that sporozoites are 'leaving the dimorphic rhoptries for hepatocyte invasion' and that 'rhoptry pairs are specialized for different invasion events'. If that is not the case, does it mean that RON11cKD sporozoites failed to form the rhoptries corresponding to the dimorphic pair? A more detailed discussion would be needed on this point and, as I mentioned above, on the specific role of RON11 in the formation of each rhoptry pair. __

We do not agree that this constitutes a contradiction; instead, more nuance is needed to fully explain the phenotype. As shown in the new graph added in response to Reviewer#1 Figure changes comment #1 in RON11cKD oocyst sporozoites, 64% of all rhoptries are located at the apical end. Our hypothesis is that these rhoptries are used for SG invasion and, therefore, would not be present in RON11cKD SG sporozoites. Consequently, the fact that 64% of RON11cKD sporozoites lack rhoptries is exactly what we would expect. Essentially, we predict three slightly different 'pathways' for RON11cKD sporozoites: If they had 2 apical rhoptries in the oocyst, we predict they would have zero rhoptries in the SG. If they had 2 cytoplasmic rhoptries in the oocyst, we predict they would have two rhoptries in the SG. If they had one apical and one cytoplasmic rhoptry in the oocyst, we predict they would have one rhoptry in the SG. In any case, we expect the apical rhoptries to be 'used up,' which appears to be supported by the data.

__Out of pure curiosity, is it possible to measure the length and number of subpellicular microtubules in the sporozoites observed in this study using expansion microscopy? __

We have performed an analysis of subpellicular microtubules which is now included as Supplementary Figure 2. We could not always distinguish every SPMT from each other and so have not quantified SPMT number. We have, however, quantified their absolute length on both the 'long side' and 'short side', their relative length (as % sporozoite length) and the degree to which they are polyglutamylated.

A description of this analysis is now found in the results section as follows: "*We quantified the length and degree of polyglutamylation of SPMTs on the 'long side' and 'short side' of the sporozoite (Figure S2). 'Short side' SPMTs were on average 33% shorter (mean = 3.6 µm {plus minus}SD 1.0 µm) than 'long side' SPMTs (mean = 5.3 µm {plus minus}SD 1.5 µm) and extended 17.4% less of the total sporozoite length. While 'short side' SPMTs were significantly shorter, a greater proportion of their length (87.9% {plus minus}SD 11.2%) was polyglutamylated compared to 'long side' SPMTs (69.4% {plus minus}SD 13.8%)." *

Supplementary Figure 2: Analysis of sporozoite subpellicular microtubules. Isolated P. yoelii salivary gland sporozoites were prepared by U-ExM and stained with anti-tubulin (microtubules) and anti-PolyE (polyglutamylated SPMTs) antibodies. SPMTs were defined as being on either the 'long side' (nucleus distant from plasma membrane) or 'short side' (nucleus close to plasma membrane) of the sporozoite as depicted in Figure 1f. (a) SPMT length along with (b) SPMT length as a proportion of sporozoite length were both measured. (c) Additionally, the proportion of the SPMT that was polyglutamylated was measured. Analysis comprises 25 SPMTs (11 long side, 14 short side) from 6 SG sporozoites. ** = p The following section has also been added to the methods to describe this analysis: * "Subpellicular microtubule measurement

• To measure subpellicular microtubule length and polyglutamylation maximum intensity projections were made of sporozoites stained with NHS Ester, anti-tubulin and anti-PolyE antibodies, and SYTOX Deep Red. The side where the nucleus was closest to the parasite plasma membrane was defined as the 'short side', while the side where the nucleus was furthest from the parasite plasma membrane was defined as the 'long side'. Subpellicular microtubules were then measured using a spline contour from the apical end of the sporozoite to the basal-most end of the microtubule with fluorescence intensity across the contour plotted (Zeiss ZEN 3.8). Sporozoite length was defined as the distance from the sporozoite apical polar rings to the basal complex, measuring through the centre of the cytoplasm. The percentage of the subpellicular microtubule that was polyglutamylated was determined by assessing when along the subpellicular microtubule contour the anti-PolyE fluorescence intensity last dropped below a pre-defined threshold."

*

__In addition to the previous point, in the text accompanying Figure 7a, the authors claim that "64% of PbRON11cKD SG sporozoites contained no rhoptries at all, while 9% contained 1 rhoptry and 27% contained 2 rhoptries". Could this data be used to infer which rhoptry pair are missing from the RON11cKD oocyst sporozoites? Can it be inferred that the 64% of salivary gland sporozoites that had no rhoptries in fact had 2 congruent rhoptries in the oocyst sporozoite stage and that these have been discharged already? __

Please see the response to Reviewer #2 Experimental Changes Comment #4.

__Is it possible that the dimorphic rhoptries are simply precursors to the congruent rhoptries? Could it be that after the congruent rhoptries are used for SG invasion, new congruent rhoptries are formed from the dimorphic ones and are then used for the next invasion?____ Would it be possible to investigate this by isolating sporozoites some time after they have invaded the SG and performing expansion microscopy? This would allow you to confirm whether the dimorphic rhoptries truly remain in the same form, or if new congruent rhoptries have been formed, or if there have been any other changes to the morphology of the dimorphic rhoptries. __

In theory, it is possible that the dimorphic rhoptries are precursors to the uniform rhoptries, specifically how the larger one of the two in the dimorphic pair might be a precursor. Maybe the smaller one is, but we have no evidence to suggest that this rhoptry lengthens after SG invasion. We are interested in isolating sporozoites from SGs to add a temporal perspective, but currently, this isn't feasible. When sporozoites are isolated from SGs, they are collected at all stages of invasion. Additionally, we don't know how long each step of SG invasion takes, so a time-based method might not be effective either. We are developing an assay to better determine the timing of events during SG invasion with MoTissU-ExM, but this is beyond the scope of this study.

__In the section titled "Presence of PbRON11cKD sporozoites in the SG intercellular space", the authors state that "the majority of PbRON11cKD-infected mosquitoes contained some sporozoites in their SGs, but these sporozoites were rarely inside either the SG epithelial cell or secretory cavity". - this is suggestive of an invasion defect as the authors suggest. Could the authors collect these sporozoites and see if liver hepatocyte infection can be established by the mutant sporozoites? They previously speculate that the two different types of rhoptries (congruent and dimorphic) may be specific to the two invasion events (salivary gland epithelial cell and liver cell infection). __

It has already been shown that RON11cKD sporozoites fail hepatocyte invasion (PMID: 31247198), even when isolated from the haemolymph and so it seems very unlikely that they would be invasive following SG isolation. As mentioned in the discussion, RON11 in merozoites has a 'dual-function' where it is partially secreted during merozoite invasion in addition to its rhoptry biogenesis functions. Assuming this is also the case in sporozoites, using the RON11cKD parasite line we cannot differentiate these two functions and therefore cannot ascribe invasion defects purely to issues with rhoptry biogenesis. In order to answer this question functionally, we would need to identify a protein that only has roles in rhoptry biogenesis and not invasion directly.

Reviewer #3

Minor text changes (Reviewer #3)

1. __Page 3 last paragraph: ...the molecular mechanisms underlying SG (invasion?) are poorly understood. __

This has been corrected 2.

__The term "APR" does not refer to a tubulin structure per se, but rather to the proteinaceous structure to which tubulin anchors. Are there any specific APR markers that can be used in Figure 1C? If not, I recommend avoiding the use of "APR" in this context. __

The text does not state that the APR is a tubulin structure. Given that it is a proteinaceous structure, we visualise the APRs through protein density (NHS Ester). It has been standard for decades to define APRs by protein density using electron microscopy, and it has previously been sufficient in Plasmodium using expansion microscopy (PMIDs: 41542479, 33705377) so it is unclear why it should not be done so in this study. 3.

__I politely disagree with the bold statements ‚ Little is known about cell biology of sporozoite formation.....from electron microscopy studies now decades old' (p.3, 2nd paragraph); ‚To date, only a handful of (instead of ‚or') proteins have been implicated in SG invasion' (p. 4, 1st paragraph). These claims may overlook existing studies; a more thorough review of the literature is recommended. __

This study includes at least 50 references from papers broadly related to sporozoite biology, covering publications from every decade since the 1970s. The most recent review that discusses salivary gland invasion cites 11 proteins involved in SG invasion. We have replaced "handful" with a more precise term, as it is not the best adjective, but it is hardly an exaggeration.

Figure changes (Reviewer #3)

1. __The hypothesis that Plasmodium utilizes two distinct rhoptry pairs for invading the salivary gland and liver cells is intriguing but remains clearly speculative. Are the "cytoplasmic pair" and "docked pair" composed of the same secretory proteins? Are the paired rhoptries identical? How does the parasite determine which pair to use for salivary gland versus liver cell invasion? Is there any experimental evidence showing that the second pair is activated upon successful liver cell invasion? Without such data this hypothesis seems rather premature. __

We are unaware of any direct protein localisation evidence suggesting that the rhoptry pairs may carry different cargo. However, only a few proteins have been localised in a way that would allow us to determine if they are associated with distinct rhoptry pairs, so this possibility cannot be ruled out either. It seems unlikely that the parasite 'selects' a specific pair, as rhoptries are typically always found at the apical end. What appears more plausible is that the "docked pair" forms first and immediately occupies the apical docking site, preventing the cytoplasmic pair from docking there. Regarding any evidence that the second pair is activated during liver cell invasion, it has been well documented over decades that rhoptries are involved in hepatocyte invasion. If the dimorphic rhoptries are the only ones present in the parasite during hepatocyte invasion, then they must be used for this process. 2.

__The quality of the "Roolet fibre" image is not good and resembles background noise from PolyE staining. Additional or alternative images should be provided to convincingly demonstrate that PolyE staining indeed visualizes the Roolet fibre. It is puzzling that the structure is visible with PolyE staining but not with tubulin staining. __

This is a logical misinterpretation based on the image provided in Figure 1c. Our intention was not to imply that PolyE staining enables us to see the rootlet fibre but that PolyE and tubulin allow us to see the APR to which the rootlet fibre is connected. There is some PolyE staining that likely corresponds to the early SPMTs that in 1c appears to run along the rootlet fibre but this is a product of the max-intensity projection. Please see Reviwer#2 Figure Changes Comment #3 for the updated Figure 1c. 3.

__More arrows should be added to Figures 6b and 6c to guide readers and improve clarity. __

We have added arrows to Figure 6b and 6c which point out what we have defined as normal and aberrant rhoptries more clearly. These panels now look like this: 4.

__Figure 2a zoomed image of P. yoelii infected SG is different than the highligted square. __

We agree that the highlighted square and the zoomed area appear different, but this is due to the differing amounts of light captured by the objectives used in these two panels. The entire SG panel was captured with a 5x objective, while the zoomed panel was captured with a 63x objective. Because of this difference, the plane of focus of the zoomed area is hard to distinguish in the whole SG image. The zoomed image is on the 'top' of the SG (closest to the coverslip), while most of the signal you see in the whole SG image comes from the 'middle' of the SG. To demonstrate this more clearly, we have provided the exact region of interest shown in the 63x image alongside a 5x image and an additional 20x image, all of which are clearly superimposable.__

__ 5.

__Figure 3 legend: "P. yoelii infected midguts harvested on day 15" should be corrected. More general, yes, "...development of each oocyst within a single midgut is asynchronous." but it is still required to provide the dissection days. __

We are unsure what the suggested change here is. We do not know what is wrong with the statement about day 15 post infection, that is when these midguts were dissected. __ Experimental Changes (Reviewer #3)__

1. __The proposed role of AOR in rhoptry biogenesis appears highly speculative. It is unclear how the authors conclude that "AORs carry rhoptry cargo" solely based on the presence of RON4 within the structure. Inclusion of additional markers to characterize the content of AOR and rhoptries will be essential to substantiate the hypothesis that this enigmatic structure supports rhoptry biogenesis. __

It is important to note that the hypothesis that AORs, or rhoptry anlagen, carry rhoptry cargo and serve as vehicles of rhoptry biogenesis was proposed long before this study (PMID: 17908361). In that study, it was assumed that structures now called AORs or rhoptry anlagen were developing rhoptries. Although often visualised by EM and presumed to carry rhoptry cargo (PMID: 33600048, 26565797, 25438048), it was only more recently that AORs became the subject of dedicated investigation (PMID: 31805442), where the authors stated that "...AORs could be immature rhoptr[ies]...". Our observation that AORs contain the rhoptry protein RON4, which is not known to localize to any other organelle, we therefore consider sufficient to conclude that AORs carry rhoptry cargo and are thus vehicles for rhoptry biogenesis. 2.

__The study of RON11 appears to be a continuation of previous work by a collaborator in the same group. However, neither this study nor the previous one adequately addresses the evolutionary context or structural characteristics of RON11. Notably, the presence of an EF-hand motif is an important feature, especially considering the critical role of calcium signaling in parasite stage conversion. Given the absence of a clear ortholog, it would be interesting to know whether other Apicomplexan parasites harbor rhoptry proteins with transmembrane domains and EF-hand motifs, and if these proteins might respond similarly to calcium stimulation. Investigating mutations within the EF-hand domain could provide valuable functional insights into RON11. __

We are unsure what suggests that RON11 lacks a clear orthologue. RON11 is conserved across all apicomplexans and is also present in Vitrella brassicaformis (OrthoMCL orthogroup: OG7_0028843). A phylogenetic comparison of RON11 across apicomplexans has previously been performed (PMID: 31247198), and this study provides a structural prediction of PbRON11 with the dual EF-hand domains annotated (Supplementary Figure 9). 3.

__The study cannot directly confirm that membrane fusion occurs between rhoptries and AORs. __

This is already stated verbatim in the results "Our data cannot directly confirm that membrane fusion occurs between rhoptries and AORs..." 4.

__It is unclear what leads to the formation of the aberrant rhoptries observed in RON11cKD sporozoites. Since mosquitoes were not screened for infection prior to salivary gland dissection, The defect reports and revisited of RON11 knockdown does not aid in interpreting rhoptry pair specialization, as there was no consistent trend as to which rhoptry pair was missing in RON11cKD oocyst sporozoites. The notion that RON11cKD parasites likely have ‚combinatorial defects that effect both rhoptry biogenesis and invasion' poses challenges to understand the molecular role(s) of RON11 on biogenesis versus invasion. Of note, RON11 also plays a role in merozoite invasion. __

We are unclear about the comment or suggestion here, as the claims that RON11cKD does not help interpret rhoptry pair specialization, and that these parasites have combined defects, are both directly stated in the manuscript. 5.

__Do all SG PbRON11cKD sporozoites lose their reduced number of rhoptries during SG invasion as in Figure 7a (no rhoptries)? __

Not all RON11cKD SG sporozoites 'use up' their rhoptries during SG invasion. This is quantified in both Figure 7a and the text, which states: "64% of *PbRON11cKD SG sporozoites contained no rhoptries at all, while 9% contained 1 rhoptry and 27% contained 2 rhoptries."

* 6.

Different mosquito species/strains are used for P. yoelii, P. berghei, and P. falciparum. Does it effect oocyst sizes/stages? Is it ok to compare?

__ __We agree that a direct comparison between for example * yoelii and P. berghei *oocyst size would be inappropriate, however Figure 3c and Supplementary Figure 4 are not direct comparisons between two species, but a summation of all oocysts measured in this study to indicate that the trends we observe transcend parasite/mosquito species differences. Our study was not set up with the experimental power to determine if mosquito host species alter oocyst size. 7.

__While I acknowledge that UExM has significantly advanced resolution capabilities in parasite studies, the value of standard microscopy technique should not be overlooked. Particularly, when discussing the function of RON11, relevant IFA and electron microscopy (EM) images should be included to support claims about RON11's role in rhoptry biogenesis. This would complement the UExM data and substantially strengthen the conclusions. Importantly, UExM can sometimes produce unexpected localization patterns due to the denaturation process, which warrants caution. __

The purpose of this study is not to discredit, undermine, or supersede other imaging techniques. It is simply to use U-ExM to answer biological questions that cannot or have not been answered using other techniques. Please refer to Reviewer # 1 Minor text changes comment#17 to see the new paragraph "Comparison of MoTissU-ExM and other imaging modalities" that addresses this

Both conventional IFA and immunoEM have already been performed on RON11 in sporozoites before (PMID: 31247198). When assessing defects caused by RON11 knockdown, conventional IFA isn't especially helpful because it doesn't allow visualization of individual rhoptries. Thin-section TEM also doesn't provide the whole-cell view needed to draw these kinds of conclusions. Volume EM could likely support these observations, but we don't have access to or expertise in this technique, and we believe it is beyond the scope of this study. It's also important to note that for the defect we observe-missing or abnormal rhoptries-the visualization with NHS ester isn't significantly different from what would be seen with EM-based techniques, where rhoptries are easily identified based on their protein density.

The statement that "UExM can sometimes produce unexpected localisation patterns due to the denaturation process..." is partially correct but lacks important nuance in this context. Based on our extensive experience with U-ExM, there are two main reasons why the localisation of a single protein may look different when comparing U-ExM and traditional IFA images. First, denaturation: in conventional IFAs, antibodies need to recognize conformational epitopes to bind to their target, whereas in U-ExM, antibodies must recognize linear epitopes. This doesn't mean the target protein's localisation changes, only that the antibody's ability to recognize it does. Second, antibody complexes seem unable to freely diffuse out of the gel, which can result in highly fluorescent signals not related to the target protein appearing in the image, as we have previously reported (PMID: 36993603). Importantly, neither of these factors applies to our phenotypic analysis of RON11 knockdown. All phenotypes described are based solely on NHS Ester (total protein) staining, so the considerations about changes in the localisation of individual proteins are not relevant.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Overall, the manuscript is well-written and -structured. However, I would like to raise several major points for consideration:

1. While I acknowledge that UExM has significantly advanced resolution capabilities in parasite studies, the value of standard microscopy technique should not be overlooked. Particularly, when discussing the function of RON11, relevant IFA and electron microscopy (EM) images should be included to support claims about RON11's role in rhoptry biogenesis. This would complement the UExM data and substantially strengthen the conclusions. Importantly, UExM can sometimes produce unexpected localization patterns due to the denaturation process, which warrants caution.
2. The proposed role of AOR in rhoptry biogenesis appears highly speculative. It is unclear how the authors conclude that "AORs carry rhoptry cargo" solely based on the presence of RON4 within the structure. Inclusion of additional markers to characterize the content of AOR and rhoptries will be essential to substantiate the hypothesis that this enigmatic structure supports rhoptry biogenesis.
3. The hypothesis that Plasmodium utilizes two distinct rhoptry pairs for invading the salivary gland and liver cells is intriguing but remains clearly speculative. Are the "cytoplasmic pair" and "docked pair" composed of the same secretory proteins? Are the paired rhoptries identical? How does the parasite determine which pair to use for salivary gland versus liver cell invasion? Is there any experimental evidence showing that the second pair is activated upon successful liver cell invasion? Without such data this hypothesis seems rather premature.
4. The study of RON11 appears to be a continuation of previous work by a collaborator in the same group. However, neither this study nor the previous one adequately addresses the evolutionary context or structural characteristics of RON11. Notably, the presence of an EF-hand motif is an important feature, especially considering the critical role of calcium signaling in parasite stage conversion. Given the absence of a clear ortholog, it would be interesting to know whether other Apicomplexan parasites harbor rhoptry proteins with transmembrane domains and EF-hand motifs, and if these proteins might respond similarly to calcium stimulation. Investigating mutations within the EF-hand domain could provide valuable functional insights into RON11.
5. The study cannot directly confirm that membrane fusion occurs between rhoptries and AORs.
6. It is unclear what leads to the formation of the aberrant rhoptries observed in RON11cKD sporozoites. Since mosquitoes were not screened for infection prior to salivary gland dissection, The defect reports and revisited of RON11 knockdown does not aid in interpreting rhoptry pair specialization, as there was no consistent trend as to which rhoptry pair was missing in RON11cKD oocyst sporozoites. The notion that RON11cKD parasites likely have ‚combinatorial defects that effect both rhoptry biogenesis and invasion' poses challenges to understand the molecular role(s) of RON11 on biogenesis versus invasion. Of note, RON11 also plays a role in merozoite invasion. I like the introduction of a segmentation score to Plasmodium oocyst maturation.

Minor comments:

1. The term "APR" does not refer to a tubulin structure per se, but rather to the proteinaceous structure to which tubulin anchors. Are there any specific APR markers that can be used in Figure 1C? If not, I recommend avoiding the use of "APR" in this context.
2. The quality of the "Roolet fibre" image is not good and resembles background noise from PolyE staining. Additional or alternative images should be provided to convincingly demonstrate that PolyE staining indeed visualizes the Roolet fibre. It is puzzling that the structure is visible with PolyE staining but not with tubulin staining.
3. Figure 2a zoomed image of P. yoelii infected SG is different than the highligted square.
4. Figure 3 legend: "P. yoelii infected midguts harvested on day 15" should be corrected. More general, yes, "...development of each oocyst within a single midgut is asynchronous." but it is still required to provide the dissection days.
5. More arrows should be added to Figures 6b and 6c to guide readers and improve clarity.
6. Do all SG PbRON11cKD sporozoites lose their reduced number of rhoptries during SG invasion as in Figure 7a (no rhoptries)?
7. Different mosquito species/strains are used for P. yoelii, P. berghei, and P. falciparum. Does it effect oocyst sizes/stages? Is it ok to compare?
8. I politely disagree with the bold statements ‚ Little is known about cell biology of sporozoite formation.....from electron microscopy studies now decades old' (p.3, 2nd paragraph); ‚To date, only a handful of (instead of ‚or') proteins have been implicated in SG invasion' (p. 4, 1st paragraph). These claims may overlook existing studies; a more thorough review of the literature is recommended.
9. Page 3 last paragraph: ...the molecular mechanisms underlying SG (invasion?) are poorly understood.

Significance

In this study, the authors explore Ultrastructure Expansion Microscopy (U-ExM) in Plasmodium-infected mosquito tissue with the aim to enhance the visualization of parasite ultrastructure. For this purpose, they revisit sporogony, the maturation of sporozoites inside oocysts, and sporozoite invasion of salivary glands, which has been studied both by cell biological methods and experimental genetics over four decades. They focus their analysis on the biogenesis and function of key secretory organelles, termed rhoptries, which are central to parasite invasion and, again, have been studied extensively.

This study is a follow-up of a previous study by the same authors (Ref. 19). In the former study the authors showed that U-ExM allows to visualize subcellular structures in sporozoites, including the nucleus, rhoptries, Golgi, apical polar rings (APR), and basal complex, as well as midgut-associated oocysts with developing sporozoites. Here, the authors claim a new finding by stating that sporozoites possess two distinct rhoptry pairs. Supposedly, only one pair is utilized during salivary gland invasion. The authors suggest specialization of rhoptries for different cell invasion events. The authors also revisit a RON11 knock-down parasite line, which was previously shown to be deficient in salivary gland invasion, host cell attachment, gliding locomotion, and liver invasion (Ref. 14).

I find it difficult to estimate the significance. Obviously, attention will be limited to Plasmodium researchers only, as this study is descriptive and revisits a well-studied aspect of the Plasmodium life cycle in the Anopheles vector.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

The manuscript by Liffner et al have used the modified expansion microscopy as they term Mosquito Tissue Ultrastructure Expansion Microscopy (MoTissU-ExM) to study a cell biology of temporal development of malaria parasite sporozoite biogenesis within mosquito host. They employed three different malaria parasite models Plasmodium yoelii, P.beghei and P falciparum and infected them in mosquito host.

The application of MoTissU-ExM to infected mosquito tissues is a significant technical advance, enabling visualizations previously only achievable with electron microscopy.

The major conclusion and advances are as following

• The establishment of a "segmentation score" as a great tool for staging asynchronous oocyst development.
• The location of Centriolar plaques, rootlet and other structures which are difficult to analyse
• The first detailed timeline for sporozoite rhoptry biogenesis.
• Clear quantification showing that sporozoites possess four rhoptries and utilise two during salivary gland (SG) invasion.
• A characterization of the RON11 knockout phenotype, linking it to defects in rhoptry biogenesis and a specific block in SG epithelial cell invasion. The following points are intended to further strengthen the paper for publication.

Points for Revision

1. For clarity, it would be helpful to add indicators for the centriolar plaques in Figure 1b, as their locations are not immediately obvious.
2. The 'image Z-depth' value indicated in the figures is ambiguous. It is not clear whether this refers to the distance from the coverslip surface or the starting point of the z-stack image acquisition. A precise definition of this parameter would be beneficial.
3. Regarding Figure 1c, the authors state that 'the rootlet fiber is visible'. However, such a structure cannot be confirmed from the provided NHS ester image. Can the authors present a clearer image where the rootlet fibre is more distinct? Furthermore, please provide the basis for identifying this structure as a rootlet fiber based on the NHS ester observation alone.
4. Why do the congruent rhoptries have similar lengths to each other, while the dimorphic rhoptries have different lengths? Is this morphological difference related to the function of these rhoptries?
5. Would it be possible to show whether RON11 localises to the dimorphic rhoptries, the congruent rhoptries, or both, by using expansion microscopy and a parasite line that expresses RON11 tagged with GFP or a peptide tag?
6. The knockdown of RON11 disrupts the rhoptry structure, making the dimorphic and congruent rhoptries indistinguishable. Does this suggest that RON11 is important for the formation of both types of rhoptries? I believe that it would be crucial to confirm whether RON11 localises to all rhoptries or is restricted to specific rhoptries for a more precise discussion of RON11's function.
7. The authors state that 64% of RON11cKD SG sporozoites contained no rhoptries at all. Does this mean RON11cKD SG sporozoites used up all rhoptries corresponding to the dimorphic and congruent pairs during SG invasion? If so, this contradicts your claims that sporozoites are 'leaving the dimorphic rhoptries for hepatocyte invasion' and that 'rhoptry pairs are specialized for different invasion events'. If that is not the case, does it mean that RON11cKD sporozoites failed to form the rhoptries corresponding to the dimorphic pair? A more detailed discussion would be needed on this point and, as I mentioned above, on the specific role of RON11 in the formation of each rhoptry pair.
8. Out of pure curiosity, is it possible to measure the length and number of subpellicular microtubules in the sporozoites observed in this study using expansion microscopy?
9. Is it possible that the dimorphic rhoptries are simply precursors to the congruent rhoptries? Could it be that after the congruent rhoptries are used for SG invasion, new congruent rhoptries are formed from the dimorphic ones and are then used for the next invasion? Would it be possible to investigate this by isolating sporozoites some time after they have invaded the SG and performing expansion microscopy? This would allow you to confirm whether the dimorphic rhoptries truly remain in the same form, or if new congruent rhoptries have been formed, or if there have been any other changes to the morphology of the dimorphic rhoptries.
10. In addition to the previous point, in the text accompanying Figure 7a, the authors claim that "64% of PbRON11cKD SG sporozoites contained no rhoptries at all, while 9% contained 1 rhoptry and 27% contained 2 rhoptries". Could this data be used to infer which rhoptry pair are missing from the RON11cKD oocyst sporozoites? Can it be inferred that the 64% of salivary gland sporozoites that had no rhoptries in fact had 2 congruent rhoptries in the oocyst sporozoite stage and that these have been discharged already?
11. In the section titled "Presence of PbRON11cKD sporozoites in the SG intercellular space", the authors state that "the majority of PbRON11cKD-infected mosquitoes contained some sporozoites in their SGs, but these sporozoites were rarely inside either the SG epithelial cell or secretory cavity". - this is suggestive of an invasion defect as the authors suggest. Could the authors collect these sporozoites and see if liver hepatocyte infection can be established by the mutant sporozoites? They previously speculate that the two different types of rhoptries (congruent and dimorphic) may be specific to the two invasion events (salivary gland epithelial cell and liver cell infection).

There are a few typing errors in the document:

1. Paragraph 3 of the introduction - line 7, "handful or proteins" should be handful of proteins
2. Paragraph 5 of the introduction - line 7, "also able to observed" should be observe
3. In the final paragraph of the introduction - line 1, "leverage this new understand" should be understanding
4. The first paragraph of the discussion summary contains an incomplete sentence on line 7, "PbRON11ctrl-infected SGs."
5. The second paragraph of the discussion - line 10, "until cytokinesis beings" should be begins

Some suggestions for figures

Fig 1B - could the tubulin image in the hemispindle panel be made brighter?

Fig 3B: stage 2 and 6 does not show the DNA cyan, it would-be good show the sate of DNA at that particular stage, especially at stage 2 when APR is visible. And box the segment in the parent picture whose subset is enlarged below it.

Fig 4A - the green text in the first image panel is not visible. Also, the cyan text in the 3rd image in Fig 1A is also difficult to see. There's a few places where this is the case

Fig 6A - how do the authors know ron11 expression is reduced by 99%? Did they test this themselves or rely on data from the lab that gifted them the construct? Also please provide mention the number of oocyst and sporozoites were observed.

Fig 6E - are the data point colours the wrong way round on this graph? Just looking at the graph it looks as though the RON11cKD has more rhoptries than the control which does not match what is said in the text.

Fig S8C, PbRON11 ctrl, pie chart shows 89.7 % spz are present in the secretory cavity while the text shows 100 %, 35/35

Fig S9D shows that RON11 ckd contains 17.1% sporozoites in secretory cavity while the text says 24%.

Some point to discuss

1.One minor point that author suggest that oocyst diameter is not appropriate for the development of sporozoite develop. This is not so true as oocyst diameter tells between cell division and cell growth so it is important parameter especially where the proliferation with oocyst does not take place but the growth of oocyst takes place.<br /> 2. The author have not mentioned that sometimes the stage oocyst development is also dependent on the age of mosquito and it vary between different mosquito gut even if the blood feed is done on same day. 3. How is the apical polarity different to merozoite as some conoid genes are present in ookinete and sporozoite but not in merozoite.

Significance

The following aspects are important:

This is novel and more cell biology approach to study the challenging stage of malaria parasite within mosquito. By using MoTissU-ExM, the authors have enabled the three-dimensional observation of ultrastructures of oocyst-sporozoite development that were previously difficult to observe with conventional electron microscopy alone. This includes the developmental process and entire ultrastructure of oocysts and sporozoites, and even the tissue architecture of the mosquito salivary gland and its epithelia cells.

Advances:

By observing sporozoites formation within the oocyst and the overall ultrastructure of the sporozoite with MoTissU-ExM, the authors have provided detailed descriptions of the complete structure and three-dimensional spatial relationships of the rhoptries, rootlet fibre, nucleus, and other organelles. Furthermore, their detailed localisation analysis of sporozoites within the salivary gland is also a great achievement. Considering that such observations were technically and laboriously very difficult with conventional electron microscopy, enabling these analyses with higher efficiency and relatively lower difficulty represents a major contribution to the future advancement of oocyst-sporozoite biology. The development of the 'segmentation score' for sporozoite formation within the oocyst is another major advance. I think this will enable detailed descriptions of structural changes at each developmental stage and of the molecular mechanisms involved in the development of oocysts-sporozoites This has its advantages if antibodies can be used and somewhat reduces the need for immuno-EM. Secondly, in terms of sporozoite rhoptry biology, the Schrevel et al Parasitology 2007 seems to only focus on oocyst sporozoite rhoptries as they say that the sporozoites have 4 rhoptries. This study on the other hand also looks at salivary gland sporozoites and shows that there are potentially important differences between the two - namely the reduction from 4 rhoptries to two. This also leads to further questions about the different types of rhoptries in oocyst sporozoites and whether they're adapted to invasion of different cell types (sal gland epithelial cells or liver hepatocytes)

Limitation

It would be that expansion microscopy alone still has its limits when it comes to observing ultra-fine structures. For example, visualising the small vesicular structures that Schrevel et al. observed in detail with electron microscopy, or seeing ultra-high resolution details such as the fusion of membrane structures and their interactions with structures like the rootlet fibre and microtubules. Therefore, I think that electron microscopy remains essential for the observation of such ultra-fine structures The real impact of this work is mostly cell biologist working with malaria parasite and more in mosquito stages. But the approaches can be applied to any material from any species where temporal dynamics need to be studied with tissue related structures and where UExM can be applied. I am parasite cell biologist working with parasites stages within mosquito vector host.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

In this paper the authors use ultrastructure expansion microscopy to investigate the mosquito stages of the malaria parasite, specifically the stage called oocyst and the process of sporozoite development. They report a number of observations of which the ones concerning rhoptries are the most interesting. There are four of these organelles in the first form of sporozoites in the oocyst and only two in the mature form in the salivary gland. Using a gene knockout of a protein that was reported to be important for rhoprty formation in merozoites, the parasites invading into human red blood cells, they found that fewer rhoptries are formed also in sporozoites and that these cannot enter into the salivary gland cells any more. The presented data are in my view conclusive and no additional experiments are needed for this work to be published. The described experiments should be readily reproducible and have a high statistical power. The text is mostly clearly written but could be improved to make it more concise and more precise and to avoid overstatements. Some references could be added. It would have helped to have line numbers in the manuscript. My suggestions are as following:

Abstract: don't focus on technique but on the questions you tried to answer (ie rewrite or delete the 3rd and 4th sentence)

Add reference on page 3 after 'disrupted parasites' Change 'the basal compelx at the leading edge' - this seems counterintuitive Change 'mechanisms underlying SG are poorly' - what mechanisms? of invasion or infection? On page 4: 'handful of proteins' 'range of cell biology processes' - I understand the paper that the key discovery concerns rhoptry biogenesis and function, so focus on that, all other aspects appear rather peripheral. what are the 'three microtubule spindle structures'? 'Much of this study focuses on the secretory organelles': I would suggest to rewrite the intro to focus solely on those, which yield interesting findings. On page 5: 'little is known' - please describe what is known, also in other stages. At the end of the paper I would like to know what is the key difference to rhoptry function in other stages? change 'rhoptries golgi-derived, made de novo' change 'new understand to' 'rhoptry malformations' seem to be similar in sporozoites and merozoites. Is that surprising/new? What is known about crossing the basal lamina. Where rhoptries thought to be involved in this process? Or is it proteins on the surface or in other secretory organelles? On page change/specify: 'wide range of parasite structures' On page 7: is Airyscan2 a particular method or a specific microscope? what are the dark lines in panel E? in panel G: Are the dense granules not micronemes? What are the dark lines? Rhoptries?? On page 8 the authors mention a second layer of CSP but do not further investigate it. It is likely hard to investigate this further but to just let it stand as it is seems unsatisfactory, considering that CSP is the malaria vaccine. What happens if you add anti-CSP antibodies? I would suggest to shorten the opening paragraphs of this paper and to focus on the rhoptries. This could be done be toning down the text on all aspects that are not rhoptries and point to the open question some of the observations such as the CSP layers raise for future studies. Figure 2 seems to add little extra compared to the following figures and could in my view go to the supplement. On page 10 I suggest to qualify the statement 'oocyst development has typcially been inferred by'. There seem a few studies that show that size doesn't reflect maturation. Page 11: I am tempted to suggest the authors start their study with Figure 3 and add panel A from Figure 2 to it. This leads directly to their nice work on rhoptries. Other features reported in Figures 1 and 2 are comparatively less exciting and could be moved to the supplement or reported in a separate study. Text on page 12 could be condensed to highlight the new data of ron4 staining of the AOR. Maybe include more detail of the differences between species on rhoptry structure into Figure 4. I would encourage to move the Data on rhoptries in Figure S6 to the main text ie to Figure 4. On page 16 the authors state that different rhoptries might have different function. This is an interesting hyopthesis/result that could be mentioned in the abstract. how large is RON11? On page 19: do the parasites with the RON11 knockout only have the cytoplasmic or only the apical rhoptries? Page 23: I suggest to delete the first sentence and focus on the functional aspects and the discoveries. There is no causal link between ookinete invasion and oocyst developmental asynchrony First sentence of page 24 appears to contradict what is written in results I don't understand the first two sentences in the paragraph titled Comparison between Plasmodium spp On page 25 or before the vast number of electron microscopy studies should be discussed and compared with the authors new data. First sentence on page 27: there are many studies on parasite proteins involved in salivary gland invasion that could be mentioned/discussed. Maybe add a conclusion section rather than a future application section, which reads as if you want to promoted the use of ultrastructure expansion microscopy. To my taste the technological advance is a bit overplayed considering the many applications of this techniques over the last years, especially in parasitology, where it seems widely used. In any case, please delete 'extraordinarily'

Significance

This interesting study investigates the development of malaria parasites in the mosquito using ultrastructure expansion microscopy adapted to mosquito tissue. It provides new and beautiful views of the process of sporozoite formation. The authors discovered that four secretory vesicles called rhoptires are formed in the sporozoites with two pairs being important for distinct functions, one pair functions during invasion of the salivary glands of the mosquito and the other in liver infection, although the latter is not shown but inferred from prior data.

This study will thus be of interest to scientists investigating malaria parasites in the mosquito as well as to scientist working on vesicle secretion and invasion in these parasites.

The authors use a previously generated parasite line that lack a protein to investigate its function in rhoptry biogenesis and find that its absence leads to fewer rhoptries which impacts the capacity of the parasite to enter into salivary gland cells. This is a nice functional addition to an otherwise largely descriptive study, but mimics largely the previously reported results from the blood stages. It is not clear to this reviewer how much the study advances the field over the many previous electron microscopy studies. This could be better elaborated in the text.

Strength of the study: beautiful microscopy, new insights into rhoptry formation and function, new technique to study malaria parasites in the mosquito

Weakness of the study: Some loose ends in the description of spindles and CSP layers, text could be more focussed on the key advancements reported

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1 Reviewer 1 Point 1- The authors describe cortical neuronal counts across several mammalian species, which is quite impressive, but the information on the methods of counting is lacking: how representative are the data used / shown; how many individuals / brains / sections were used for each species considered? Much more detailed description of the quantifications should be provided to judge the validity of this first conclusion.

Response: We sincerely thank the reviewer for this insightful and constructive suggestion. We agree that the methodological description of our comparative histological analysis, which is the fundamental basis of this study, was insufficient in the original manuscript. Following the reviewer’s advice, we have extensively revised the Materials and Methods section entitled “Nissl staining and neuronal cell number count” (Page 32, Line 15).

Reviewer 1 Point 2- The authors use several markers of cortical neuron identity to confirm their neuron number measurements, but from the data shown in Figure 1D,E it seems that only some markers (Satb2) show species-differences while others do not (CTIP2 / Tbr1). How do the authors explain this discrepancy - does this mean that it is mainly Satb2 neurons that are increased in number? But if so how to explain the relative increase in subcortical projections shown in Figure S7?

Response: We appreciate the reviewer’s insightful comments regarding the marker expression patterns. Upon re-evaluating our data in light of your feedback, we agree that the species differences in deep-layer (DL) markers such as Ctip2 and Tbr1 in the adult stage appear relatively modest compared to the robust differences observed in Satb2 and the projection data shown in Figure S8.

To address this point, we have incorporated a comparison between the adult data (Figure 1) and our findings from P7 (Figure S2). As shown in the revised manuscript, the species differences for all markers are significantly more pronounced at P7 than in the adult. Notably, in the lower layers, rats exhibit a significantly higher number of marker-positive cells across all markers, including those newly added in this revision, compared to mice.

We offer the following interpretation regarding these temporal differences:

1. Developmental Relevance: The marker molecules analyzed are well-established regulators of neuronal subtype fate and projection identity during development. Their critical fate-determining functions are primarily exercised during the migration and maturation phases of nascent neurons.
2. Postnatal Expression Shifts: Whether these molecules maintain functional roles in the fully matured adult brain remains less certain. It is plausible that marker expression may diminish in certain neuronal populations during late postnatal development, leading to the attenuated species differences observed in adults. Consequently, we believe the strong correlation between P7 quantitative data and projection fate provides a biologically sound validation of our hypothesis.

While we have kept the discussion in the main text concise to maintain focus for the general reader, we have provided comprehensive data in Figure 1 and Figure S2. This ensures that the necessary evidence is readily available for specialists interested in these developmental dynamics.

Reviewer 1 Point 3- The authors focus their study almost exclusively on somatosensory cortex, but can they comment on other areas (motor, visual for instance)? It would be nice to provide additional comparative data on other areas, at least for some of the parameters examined across mouse and rat. Alternatively the authors should be more explicit in the abstract and description of the study that it is limited to a single area.

Response: We sincerely appreciate the reviewer’s insightful comment. As suggested, we have revised the Abstract to explicitly state that our current analysis is focused on the somatosensory cortex. Furthermore, as demonstrated in Figure 1B, we have added a discussion regarding the possibility that the species differences observed in the primary somatosensory cortex may be a general feature shared across the entire cerebral cortex, as follows: “This DL-biased thickening in rats was evident in the primary somatosensory area, but is consistently observed throughout the rostral-caudal cortical regions. (Page 19, Lines 29-31)“

Reviewer 1 Point 4- The authors provide convincing evidence of increased Wnt signaling pathway in the rat. They should show more explicitly how other classical pathways of neurogenic balance / temporal patterning are expressed in their mouse and rat transcriptome data sets. These would include Notch, FGF, BMP, for which all the data should be available to provide meaningful species comparison.

Response: We sincerely thank the reviewer for this insightful suggestion. Following your advice, we have newly included comparative data on key signaling pathways essential for cortical development—namely Wnt, FGF, NOTCH, mTOR, SHH, and BMP—across different species. These results are now presented in Figure S17. Rat progenitors show comparable patterns to other species for FGF, mTOR, and Notch signaling, but elevated Wnt and BMP expression, especially at early stages. A detailed heatmap of raw Wnt pathway gene expression across species is also included in the same supplementary figure. We believe these additions provide a more comprehensive evolutionary perspective and significantly strengthen our findings.

Reviewer 1 Point 5- The alignment of mouse and rat trajectories is very nicely showing a delay at early-mid-corticogenesis. But there is also heterochronic transcriptome at latest stages (end of 5). How can this be interpreted? Does this mean potentially prolonged astrogliogenesis in the rat cortex?

Response: We sincerely appreciate the reviewer’s insightful comment and the meticulous attention given to our data. Regarding the heterochronic shift observed at Day 5, we agree that this point was not sufficiently addressed in the original manuscript.

We would like to clarify the two primary reasons for this omission, which are inherent to the current study’s design:

1. Resolution of Stage Alignment at Temporal Extremes: In our developmental stage alignment analysis, corresponding stages are defined by pairs showing the highest transcriptomic similarity within the sampled range. By definition, the precision of this alignment tends to decrease at the earliest and latest time points of a dataset. Since the "true" biological equivalent might lie outside our sampling window, we must be cautious in interpreting shifts at these temporal boundaries.
2. Difference in Validation Rigor: Our study prioritized the early stages of deep-layer (DL) neuron production. Consequently, we rigorously defined the onset of neurogenesis in rats (Day 1) using multiple independent methods, including clonal analysis, immunohistochemistry, and gene expression. In contrast, Day 5 was defined simply as five days post-initiation of neurogenesis, without equivalent multi-modal validation. Given that our primary focus is the early phase of neurogenesis, the precision of the transition from late neurogenesis to gliogenesis is relatively lower. For these reasons, we believe that an in-depth discussion of the heterochronic shift at Day 5 might lead to over-interpretation. To reflect this more accurately and avoid misleading the reader, we have revised Figure 6F to de-emphasize the Day 5 shift. In addition, we revised the manuscript as “Importantly, while this analysis identified stage pairs with the highest similarity, the correspondence at the edges of the temporal sampling window is inherently less certain than at the center. Consequently, we focus on the notable reflection point at the center of our dataset. (Page 13, Lines 37-39)”.

We believe these changes more faithfully represent the biological scope of our data while maintaining the scientific integrity of our primary conclusions.

Reviewer 1 Point 6- Figure 7: description implies that module 3 is a subset of module 4, but this is not obvious at all from the panels shown. Please clarify.

Response: We sincerely appreciate the reviewer’s careful reading of our manuscript. As suggested, we have revised Figure 7 to clarify the hierarchical relationship between Module 3 and Module 4, ensuring that their inclusion is now explicitly presented.

Reviewer #2 Reviewer 2 Point 1. The introduction lacks sufficient background and fails to convey the significance of the study. Specifically, why the research was undertaken, what knowledge gap it addresses, and how the findings could be applied. Addressing these questions already in the introduction would enhance the impact of the work and broaden its readership.

Response: We sincerely appreciate the reviewer’s insightful comment on this point. Our study reports evolutionary insights gained through an unconventional approach: a single-cell level comparison between mice and rats. We agree that clarifying the necessity of this specific approach is crucial for the manuscript. Accordingly, we have added the following two points to the Introduction:

1. At the end of the first paragraph, we emphasized the current lack of research on the evolutionary adaptation of cortical circuits, despite the established functional importance of evolutionarily conserved circuits. (Page 3, Lines 7-10); “Paradoxically, despite the importance of these variations, research has predominantly focused on the conserved aspects of cortical architecture. Consequently, the degree of evolutionary plasticity inherent in these circuits and the cell-intrinsic mechanisms driving their modification remain profoundly enigmatic.”)
2. At the end of the third paragraph, we revised and added text (Page3, Lines 26-27; “This lack of comparative insight represents a significant gap in our understanding of how conserved developmental programs give rise to species-specific brain architectures.”).

Reviewer 2 Point 2. In figure 5 the authors conclude that "differences in cell cycle kinetics and indirect neurogenesis are unlikely to be the primary factors driving the species-specific variation in DL neuron production. Instead, the temporal regulation of progenitor neurogenic competence, which determines the duration of the DL production phase, provides a more plausible explanation for the greater number of DL subtypes observed in rats". It is not clear to this reviewer how the authors come to this conclusion. Authors observe a significant proportion of mitotic cells in rat VZ from day 1, and a higher constant proportion of mitotic progenitors in SVZ rats compared to mouse (Figure 5C). This points to an early difference in mitotic progenitors that may also lead to increased IP numbers, and potentially an increased number in DL cells, even before day 1. In addition, the higher abundance of IPs in the G2/S phase (statistically significant in 4 of the 7 time points) (Figure 5F), would suggest that this difference might play a role in the species-specific variation of DL neuron production. The authors should estimate cell cycle length instead of just measuring proportions to conclude something about cell cycle kinetics. They can then model growth curves to predict the effect caused if there were differences in cell cycle length between equivalent cell types across species.

Response: We sincerely thank the reviewer for their careful reading of our manuscript and for pointing out the overstatements in our original descriptions. We agree that a more nuanced interpretation of the data was necessary. In response to these constructive suggestions, we have made the following revisions:

1. Refinement of Descriptions: We have revised the text to more accurately reflect our findings, specifically noting that the increase in RG division on Day 1 and IP proliferation throughout the neurogenic period showed a significant trend. These features are now described more fairly and cautiously in the revised manuscript. (Page 11, Lines 42-46; “Remarkably, while the temporal dynamics of mitotic density were strikingly conserved between the two species, subtle yet discernible species-specific signatures emerged. Specifically, rats exhibited a higher ratio of mitotic cells in the VZ at the onset of neurogenesis, the precise period when DL subtypes are generated in both species. Further assessment of G2/S-phase cells via pulse-EdU labeling (Figure 5D, E) “)
2. Inclusion of Time-lapse Imaging Data: The reviewer is correct that measuring the proportions of M and G2/S phases provides only a limited snapshot of cell cycle dynamics. To gain a more precise insight, we performed primary cultures of neural progenitor cells (NPCs) from Day 1 and conducted live-cell time-lapse imaging. This allowed us to directly quantify the cell cycle duration of mouse and rat NPCs (Figure S9A-C).
3. Comparative Analysis and Mathematical Modeling: Our new data revealed that the cell cycle lengths of the two species are remarkably similar, with no significant differences observed under these culture conditions. Furthermore, to validate the impact of these findings on overall brain development, we developed a mathematical model based on our experimental data. This model predicts the total number of cells produced over the five-day neurogenic period, providing a more robust theoretical framework for our conclusions (Figure S9D). We believe these additions significantly strengthen the manuscript and address the reviewer's concerns regarding the physiological relevance of our observations.

Reviewer 2 Point 3. In Figure 6 the authors focus only on the mouse and rat datasets. Given the availability of datasets from primates that the author used already for Figure 7, it would give the reader a broader prospective if also these datasets would be integrated in the analysis done for Figure 6, particularly it would be interesting to integrate them in the pseudotime alignment of cortical progenitor. How do human and/or macaque early and late neurogenic phase would compare to mouse and rat in this model?

Response: We sincerely appreciate the reviewer’s insightful suggestion. In accordance with this comment, we have now incorporated pseudotime alignments of cortical progenitors between primates (human, macaque) and rodents (mouse, rat), presented as pairwise gene expression distance matrices with dynamic time warping in Figure S13. These heatmaps illustrate temporal compression or stretching in progenitor gene expression progression across species. Notably, macaque progenitors show no definitive deviations from rodents, whereas human progenitors exhibit distinct protraction relative to rats and even more so to mice. These additions provide a more comprehensive cross-species perspective without altering the study's core conclusions.

Reviewer 2 Point 4. In Figures 6C and 6D, the authors distinguish between cycling and non-cycling NECs and RGCs. Could the authors clarify the rationale behind making this distinction? Could the authors comment on how they interpret the impact of cycling versus non-cycling states on species-specific non-uniform scaling? Do they consider the observed non-linear correspondences to be driven by differences in cell cycle activity?

Response: We are grateful to the reviewer for their insightful observation. We agree that our initial classification of neural progenitor cell (NPC) populations based on proliferation marker expression levels followed a convention used in other studies but was, in the context of this work, unnecessary and potentially misleading. To avoid further confusion and focus on the core biological question, we have re-organized the data by pooling these populations into a single group. Regarding the concern about species differences in cell cycle kinetics, we believe there is no significant divergence between mice and rats that could explain the observed developmental patterns in temporal progression of neurogenesis. This is supported by two lines of evidence:

1. Quantitative analysis of pH3-positive cells (Figure 5).
2. New time-lapse imaging data of primary cultured NPCs, which shows no substantial difference in cell cycle length between the two species (Figure S9). These results indicate that the species-specific differences in deep-layer (DL) neuron production are not driven by cell division kinetics. Consequently, we conclude that the non-linear developmental progression of NPCs occurs independently of cell cycle regulation.

Reviewer 2 Point 5. For the non-uniform scaling in Figure 6F, the authors identify critical inflection points and mention that "the largest delay in rat progenitors occurring where Day 1 and Day 3 progenitors overlapped". It would be good if the authors could discuss what they think all the inflection points represents. How much can it be explained by the heterogeneity within progenitors per time point? There is a clear higher spread of histograms at days 3 and 5, and the histogram at day 5 almost overlaps with day 1. I wonder if the same conclusion about non-uniform scaling would be detected if the distance matrix was built separately for specific cell types, for example only looking at NECs or RGCs.

Response: We sincerely appreciate the reviewer’s insightful perspective on this point. In alignment with the suggestions from both this reviewer and Reviewer 1 (Point 5), we have updated the manuscript to discuss all identified inflection points. Specifically, we have clarified why our discussion focuses on the correspondence between Mouse D1 and Rat Day 3.

A recognized limitation of our current analytical approach is that it identifies the closest matching expression profiles within the specific timeframes sampled for each species. For stages at the beginning or end of our sampling window, the "true" corresponding stage in the other species may lie outside our sampled range, which naturally limits the strength of any conclusions regarding those boundary points. Consequently, while we can confidently confirm the correspondence between Mouse Day 1 and Rat Day 3—both of which sit centrally within our sampled window—we have intentionally avoided over-interpreting data near the temporal boundaries.

Regarding the cell types analyzed, this specific analysis was conducted exclusively on NECs and RGs (now shown in Figure 6F). Extensive prior research (Susan McConnell lab, Sally Temple lab, Fumio Matsuzaki lab, Dennis Jabaudon lab, and more) has established that the time-dependent mechanisms governing the fate determination of cortical excitatory neuron subtypes are encoded within RGs. Therefore, we focused our investigation on these lineages and did not include other cell types in this study. We believe this focused approach maintains the highest degree of biological relevance for our conclusions.

Reviewer 2 Point 6. The authors conclude that the elevated and prolonged expression of Wnt-ligand genes in rat RGs extend the DL neurogenic window and contribute to rat-specific expansion of deep cortical layer. In order to validate this finding it would be good for the authors to perform a perturbation experiment and reduce Wnt signalling/ Axin 2 levels in rats or depleted the Lmx1a and Lhx2 double-positive population. Response: __We thank the reviewer for this insightful suggestion. We agree that providing direct experimental evidence is crucial to demonstrating that elevated Wnt signaling in RG progenitors drives the production of DL subtype neurons in rats. To address this, we performed a functional intervention on Day 3, a stage when Wnt signaling (indicated by Axin2 expression) is significantly higher in rats than in mice (__Figure 7C, D). By introducing a dominant-negative form of TCF7L2 (dnTCF7L2) to inhibit Wnt signaling specifically in RG progenitors, we tracked the fate of the resulting neurons (Figure 7I, J). Our results showed a clear reduction in the proportion of DL neurons, accompanied by a reciprocal increase in upper-layer (UL) neurons. These findings demonstrate that maintained high levels of Wnt signaling are essential for the prolonged neurogenic capacity for DL neurons in rats. This new data has been incorporated into Figure 7.

Reviewer 2 Point 7. The authors conclude that Wnt signaling is a rat specific effect since they did not observe any clear temporal change in wnt receptors in gyrencephalic species, and only a subset of RG in rats co-express Lmx1a and Lhx2. However, specific Wntligands and receptors (Wnt5a, Fzd and Lrp6) seem to be upregulated in human as well (Fig 7G), non RG cells could act as wnt ligand inducers in other species, and it has not been demonstrated that Lmx1a and Lhx2 are the source for Wntligand production. I wonder if the authors can completely rule out a role for Wnt in the protracted neurogenesis of other species.

Response: We sincerely appreciate the reviewer’s insightful and broad perspective regarding Wnt signaling dynamics across diverse species. In this study, our primary focus was to elucidate the specific mechanisms underlying the differences between mice and rats. Consequently, we did not initially explore Wnt dynamics in other species or their roles in developmental timing in great depth in the original manuscript. We fully acknowledge that lineage-specific adaptations occur at the individual gene level; for instance, Silver and colleagues have reported that human-specific upregulation of Wnt receptor gene FZD8 modulates neural progenitor behavior (Boyd et al., Current Biology 2008, Liu et al., Nature 2025). However, our comparative analysis of five mammalian species—carefully aligned by developmental stage—reveals a distinct global trend. While individual gene variations exist like human FZD8, the expression levels of multiple Wnt-related genes, particularly ligands, are markedly higher in rats than in the other four species.

Following the reviewer’s insightful suggestion, we examined the potential role of Lmx1a in activating Wnt ligand transcription in rat cortical progenitors by analyzing their expression correlation at the single-cell level. Our analysis revealed that several Wnt ligand genes are co-expressed with Lmx1a with a remarkably strong positive correlation. While we have not yet experimentally demonstrated the direct transcriptional activation of Wnt ligands by Lmx1a in these cells, this robust correlation at single-cell resolution strongly suggests that Lmx1a regulates Wnt ligand expression. These new findings are now included in Figure 7 and Figure S16, and the corresponding results section (Page 15, Lines 42-44) has been revised accordingly.

__Reviewer 2 Point 8 __Minor comments: The RNAscope experiment is currently qualitative. Is it the mRNA copy number per cell equal in both species but more cells are positive in rat, or are there differences in number of mRNA molecules as well? It is not indicated if the RNAscopeprobes are the same for mouse and rat.

Response: We sincerely thank the reviewer for this insightful suggestion. Following the comment, we performed RNAscope analysis for Axin2 in both mice and rats and quantified the results (now included in Figure 7D). The new data successfully validate the species differences initially observed in our scRNAseq analysis: specifically, the period of high-level Axin2 expression is significantly extended in rats compared to mice. These findings provide histological evidence that reinforces our conclusions regarding the distinct temporal dynamics between the two species.

Regarding probe design, the Axin2 RNAscope probes target conserved and corresponding sequences between mouse and rat, with species-specific probes optimized for each organism to ensure maximal specificity and sensitivity. We have updated the Methods section ("Fluorescent in situ hybridization with RNAscope") to include these details.

Reviewer #3

Reviewer 3 Point 1. Satb2 is also widely recognized as a deep layer marker. The authors need to perform analysis and quantification in Figs 1 and 4 with other II/III and IV markers such as Cux1 and Rorb.

Response: We thank the reviewer for their insightful comments regarding the marker specificity. We fully agree that while Satb2 is a robust marker for callosal projection identity, its broad distribution across both deep and upper layers limits its utility as a layer-specific marker. As the reviewer suggested, Cux1 (Layers 2/3) and Rorb (Layer 4) are indeed superior markers for defining laminar identity.

To address this, we have incorporated new immunohistochemical data for these markers in both the quantification of somatosensory cortical neurons (Figure S2) and the birth-dating analysis (Figure 4).

Our new findings are as follows:

1. Layer Quantification (Figure S2): By utilizing Cux1 and Rorb as more specific upper-layer (UL) markers, we confirmed that there are no significant differences in the number of these neurons between mice and rats.
2. Birth-dating Analysis (Figure 4): These markers allowed us to more precisely define the timing of Cux1/Rorb-positive cell generation, revealing subtle but important differences between the two species. While these additions do not alter the fundamental narrative of the original manuscript, they have significantly enhanced the precision and rigor of our analysis. We are grateful to the reviewer for guiding us toward this more robust validation.

Reviewer 3 Point 2. Rats have larger cortices. Therefore, quantification of neurons should also be normalized to cortical thickness in Fig 1E and also represented with individual data points.

Response: We sincerely appreciate the reviewer’s constructive suggestion. We agree that normalizing the number of cortical neurons by thickness provides a more rigorous comparison. Accordingly, we have calculated the neuronal density (cell count per unit thickness) for Tbr1- and Ctip2-positive cells and included these data in Figure S2C. Our analysis confirms that these populations are distributed at a significantly higher density in mice compared to rats.

Furthermore, we have updated the visualization in Figure 1E to display individual data points, ensuring full transparency of the underlying distribution. We believe these revisions, prompted by the reviewer’s insight, have substantially strengthened the clarity and persuasiveness of our manuscript.

Reviewer 3 Point 3. The clonal analysis in Figs 2 and 3 quantifies GFP and RFP and reports these as neurons. However, without using cell-specific markers, it seems the authors cannot exclude that some progeny are also glia derived from a radial glial progeny. I don't expect all experiments to have this but they must have some measures of both populations to address this possibility. This needs to be addressed to build confidence in the conclusion that there is clonal production of neurons.

Related to this, the relationship between position and fate is not always 1 to 1. The data summarized in Fig 2G are based on position and not using subtype markers. They should include assessment of markers as they do in Fig 4.

Response: We sincerely thank the reviewer for this insightful comment. We agree that a clear definition of cell types is essential for the accuracy of clonal analysis.

In this study, we primarily identified neurons based on their distinct morphological characteristics and performed measurements specifically on these cells. To validate this approach, we confirmed that the vast majority of cells identified as neurons were positive for NeuN and cortical excitatory neuron markers, while remaining negative for glial markers such as Olig2 and SOX9. (Notably, at postnatal day 7, most cells in the glial lineage exist as undifferentiated Olig2-positive progenitors). These observations support our conclusion that the cells analyzed based on morphology are indeed cortical excitatory neurons.

As the reviewer rightly pointed out, evaluating cell composition using fate-specific marker expression is the ideal approach. However, our current experimental setup required multiple fluorescence channels for DAPI staining (to assess tissue architecture) and immunostaining for GFP and RFP (to identify labeled clones). Due to these technical constraints regarding available detection channels and host species compatibility, we relied on morphological criteria for the primary analysis.

To address this concern and ensure the reliability of our findings, we performed additional analyses using a subset of samples. By co-staining retrovirally labeled neurons with cell-fate markers, we obtained results consistent with our other data (Figures 1 and 4) regarding laminar position and marker expression. Based on this consistency, we are confident that our classification based on morphology and laminar position does not alter the fundamental conclusions of this study.

Reviewer 3 Point 4. In Fig 5, the authors use PH3 as well as EdU to measure differences in indirect neurogenesis. Using EdU and Tbr2 they report more dividing IPs. However they need to measure this over the total number of Tbr2 cells as it is not normalized to differences in Tbr2 cells between species. Are there total differences in Tbr2+ cells when normalized to DAPI as well? Moreover, little analyses is performed to measure any impact on radial glia. As no striking differences were observed in IPs this leaves the cellular mechanism a bit unclear and begs the impact on radial glia. Measuring PH3+ cells in VZ and SVZ is not cell specific nor does it yield information to support the prolonged neurogenesis.

Response: We sincerely thank the reviewer for this insightful suggestion. We agree that quantifying Tbr2+/EdU+ double-positive cells alone was insufficient to fully capture the IP dynamics. Following the reviewer’s advice, we have now quantified the total population of Tbr2+ cells, normalized to the number of DAPI-stained nuclei. This new analysis reveals that mice and rats exhibit nearly indistinguishable temporal dynamics (Figure S10). When integrated with the original Tbr2+/EdU+ data in Figure 5, these findings suggest that rats maintain a slightly higher IP pool throughout the neurogenic period. This implies that the increased neuronal production in rats is not restricted to a specific phase, but rather occurs consistently across all developmental stages. We believe these additional data significantly strengthen our conclusions.

Reviewer 3 Point 5. The sc-seq is done in rat and compared to published mouse data from corresponding stages. They conclude species specific differences in progenitor gene expression. I am unsure how appropriate this is. Are similar sequencing platforms used? Can they find similar results if using multiple dataset? There are other datasets that may be used to validate these findings beyond DiBella et al.

Response: We sincerely thank the reviewer for this insightful comment. We agree that establishing the validity of our analytical approach is crucial for the reader’s confidence in our findings. To address this, we have explicitly stated in the revised manuscript that both our rat scRNAseq data and the publicly available datasets were generated using consistent experimental platforms. This ensures that the integration process is technically sound.

Revised text (Page 13, Lines 16-18): “After quality control, we integrated these profiles with previously published mouse cortical cell data from corresponding neurogenic stages, which is prepared using the consistent platform with ours (35) (Figure S11).”

Furthermore, to ensure the robustness of our comparative analysis, we have incorporated an additional independent dataset (Ruan et al., PNAS 2021) in addition to the Di Bella et al. Nature 2021 data used in the original manuscript. We confirmed that the results obtained using this second dataset are highly consistent with our initial findings, further validating our conclusions across different studies (Figure S13A).

Reviewer 3 Point 6. Wnt ligand analysis requires validation in situ across developmental stages, to support their conclusions. Ideally they might consider doing some manipulations to provide context to this observation.

Response: We sincerely thank the reviewer for these insightful suggestions. We agree that validating the spatial expression patterns of Wnt ligands and confirming their expression in rat-specific RG, as suggested by our scRNAseq data, is crucial for strengthening our conclusions.

Regarding the expression of Wnt3a, a key ligand in cortical development: although immunohistochemical analysis clearly identified Wnt3a expression in the cortical hem, the expression levels in RG within the cortical area were substantially lower than those in the hem, making definitive visualization challenging. To complement these findings and provide more robust evidence, we performed the following additional experiments:

1. Validation of Wnt signaling levels: Using RNAscope-based in situ hybridization for Axin2, we successfully confirmed the elevated Wnt signaling levels in rat-specific RG (Figure 7C, D), consistent with our scRNAseq findings.
2. Elucidating strikingly high correlated expressions of Lmx1a and Wnt ligand genes in the rat cortical progenitors in our scRNAseq dataset (Figure S16B).
3. Functional analysis: To test the functional significance of this signaling, we inhibited Wnt signaling by electroporating dominant-negative TCF7L2 into rat RG at E15.5. This manipulation resulted in a subtype shift of the generated neurons toward an upper-layer identity (Figure 7I, J). These new results demonstrate that the rat-specific extension of high Wnt signaling levels serves as a fundamental mechanism for the prolonged production of deep-layer (DL) neurons. We are grateful to the reviewer for these suggestions; these additional data have significantly strengthened our core argument that the heterochronic regulation of Wnt signaling states drives the evolution of cortical neuronal composition.

__Reviewer 3 Point 7 __Minor concerns-1

Please separate images in Fig 1D it is very strange to have them all on top of each other.

Response: We sincerely thank the reviewer for this suggestion. As requested, we have provided individual channel images alongside the merged multicolor panels. We agree that this modification significantly enhances the clarity of our data and makes the results much easier to interpret.

__Reviewer 3 Point 8 __Minor concerns-2

Are data in Fig 4E Edu+Tbr1+EdU+? This should be clarified and would be most accurate.

Response: We appreciate the reviewer’s suggestion. We added the label of Y axes of the plots in Figure 4E-K. The procedure of cell count in these analyses are documented in the caption of Figure 4E-K, “Normalized counts of neurons colabeled for EdU and projection-specific markers, relative to the peak of EdU+ and marker+ cells.”.

__Reviewer 3 Point 9 __Minor concerns-3

Fig 4 graphs only have titles without Y axis. Please adjust location of title or repeat for clarity.

Response: We thank the reviewer for this helpful suggestion. To clarify the definition of the Y-axis, we have now added a descriptive label to the axis in the revised figure.

__Reviewer 3 Point 10 __Minor concerns-4

Fig 4A implies cumulative incorporation which I don't think is being performed here. They should clarify this in the figure.

Response: We appreciate the reviewer’s insightful comment. To avoid any potential misunderstanding regarding the additivity of the effect, we have revised the illustration in Figure 4A for greater clarity.

__Reviewer 3 Point 11 __Minor concerns-5

Fig 5 needs labels for the actual stages assayed, as illustrated in Fig 4A.

Response: We thank the reviewer for this helpful suggestion. Following your comment, we have added the developmental stage information (expressed as embryonic days) for both mice and rats in the revised manuscript.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

In this study the authors investigate differences between two closely related species, rats and mice, in terms of cortical development and neuronal composition. They first perform comparative analysis of cortical layers which revealed the density and markers of deep layer neurons of rats is disproportionately larger compared to adult mice. They then use retroviruses for lineage analysis from embryonic stages to P7. They find in general that there are temporal differences in when mice and rats produce upper versus deep layer neurons, with the process being protracted in rats. EdU injections were used to report differences in the timing of cortical neuron generation between species and they note no striking differences in IPs. Sc-sequencing of rat cortices at different stages was then used to measure temporal changes in gene expression and compared to published mouse data. They note that rats have sustained Wnt ligand expression in radial glia highlighting that as a potential mechanism of action.

Major concerns 1. Satb2 is also widely recognized as a deep layer marker. The authors need to perform analysis and quantification in Figs 1 and 4 with other II/III and IV markers such as Cux1 and Rorb. 2. Rats have larger cortices. Therefore, quantification of neurons should also be normalized to cortical thickness in Fig 1E and also represented with individual data points. 3. The clonal analysis in Figs 2 and 3 quantifies GFP and RFP and reports these as neurons. However, without using cell-specific markers, it seems the authors cannot exclude that some progeny are also glia derived from a radial glial progney. I don't expect all experiments to have this but they must have some measures of both populations to address this possibility. This needs to be addressed to build confidence in the conclusion that there is clonal production of neurons. Related to this, the relationship between position and fate is not always 1 to 1. The data summarized in Fig 2G are based on position and not using subtype markers. They should include assessment of markers as they do in Fig 4. 4. In Fig 5, the authors use PH3 as well as EdU to measure differences in indirect neurogenesis. Using EdU and Tbr2 they report more dividing IPs. However they need to measure this over the total number of Tbr2 cells as it is not normalized to differences in Tbr2 cells between species. Are there total differences in Tbr2+ cells when normalized to DAPI as well? Moreover, little analyses is performed to measure any impact on radial glia. As no striking differences were observed in IPs this leaves the cellular mechanism a bit unclear and begs the impact on radial glia. Measuring PH3+ cells in VZ and SVZ is not cell specific nor does it yield information to support the prolonged neurogenesis. 5. The sc-seq is done in rat and compared to published mouse data from corresponding stages. They conclude species specific differences in progenitor gene expression. I am unsure how appropriate this is. Are similar sequencing platforms used? Can they find similar results if using multiple dataset? There are other datasets that may be used to validate these findings beyond DiBella et al. 6. Wnt ligand analysis requires validation in situ across developmental stages, to support their conclusions. Ideally they might consider doing some manipulations to provide context to this observation.

Minor concerns 1. Please separate images in Fig 1D it is very strange to have them all on top of each other. 2. Are data in Fig 4E Edu+Tbr1+EdU+? This should be clarified and would be most accurate. 3. Fig 4 graphs only have titles without Y axis. Please adjust location of title or repeat for clarity. 4. Fig 4A implies cumulative incorporation which I don't think is being performed here. They should clarify this in the figure. 5. Fig 5 needs labels for the actual stages assayed, as illustrated in Fig 4A.

Significance

Strengths:

The finding that there are differences in cortical composition between rats and mice and that this is linked to prolonged neurogenesis in rats Use of careful and detailed lineage analysis to define differences in temporal production of neurons Inclusion of single cell sequencing

Limitations:

Largely descriptive Requires additional investigation to support some conclusions about neurons Concerns about inferring too much from single cell sequencing done by the authors but compared to publication

Advance: Finding that there are differences in neurogenesis between closely related species is interesting and provides insight into mechanisms of cortical evolution.

Audience: Evolution, cortical development

Expertise: Cortical development, evolution

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

Summary:

Yamauchi et al. performed a comparative anatomical analysis of the layer architecture in the primary somatosensory cortex across 8 mammalian species. Unlike primates, which show an expansion of upper layers (UL), rodents, especially rats, display a pronounced thickening of deep layers (DL). In this study they focus on comparing rats and mice, given the higher abundance of DL neuron subtypes in rats. Using histological analysis, they showed that rats possess significantly more DL neurons per cortical column than mice, while UL neuron counts remain similar. Clonal lineage tracing showed that rat radial glial (RG) progenitors generate more DL neurons, indicating species-specific differences in progenitor neurogenic activity. Birth dating assays confirmed an extended DL neurogenesis phase in rats, followed by a conserved UL generation phase. Single-cell RNA sequencing further revealed that rats maintain an early progenitor state longer than mice, marked by sustained expression of DL-associated genes. Specifically, rat RG progenitors exhibit prolonged and elevated expression of Wnt signaling genes, particularly Wnt ligands. Comparative analysis of published single-cell RNA-Seq across species highlighted that this extended Wnt-high period in rats is exceptional, suggesting a species-specific extension of a conserved neurogenic program.

Major comments:

This reviewer thinks the topic is exciting, and the experiments elegant, insightful and well described. The paper is well written and follows a very logical flow, the conclusion for each experiment is supported by the data and they are carefully stated. This reviewer really appreciated the summary illustration included as a panel in each figure, they think that this greatly enhanced the clarity and accessibility of the data presented, especially because species comparison can be difficult to follow.

In this reviewer's opinion, there are some aspects of the findings that the authors would need to clarify/address to explain in clarify the phenotype observed and to enhance the overall significance of this very well-made paper: 1. The introduction lacks sufficient background and fails to convey the significance of the study. Specifically, why the research was undertaken, what knowledge gap it addresses, and how the findings could be applied. Addressing these questions already in the introduction would enhance the impact of the work and broaden its readership. 2. In figure 5 the authors conclude that "differences in cell cycle kinetics and indirect neurogenesis are unlikely to be the primary factors driving the species-specific variation in DL neuron production. Instead, the temporal regulation of progenitor neurogenic competence, which determines the duration of the DL production phase, provides a more plausible explanation for the greater number of DL subtypes observed in rats". It is not clear to this reviewer how the authors come to this conclusion. Authors observe a significant proportion of mitotic cells in rat VZ from day 1, and a higher constant proportion of mitotic progenitors in SVZ rats compared to mouse (Figure 5C). This points to an early difference in mitotic progenitors that may also lead to increased IP numbers, and potentially an increased number in DL cells, even before day 1. In addition, the higher abundance of IPs in the G2/S phase (statistically significant in 4 of the 7 time points) (Figure 5F), would suggest that this difference might play a role in the species-specific variation of DL neuron production. The authors should estimate cell cycle length instead of just measuring proportions to conclude something about cell cycle kinetics. They can then model growth curves to predict the effect caused if there were differences in cell cycle length between equivalent cell types across species. 3. In Figure 6 the authors focus only on the mouse and rat datasets. Given the availability of datasets from primates that the author used already for Figure 7, it would give the reader a broader prospective if also these datasets would be integrated in the analysis done for Figure 6, particularly it would be interesting to integrate them in the pseudotime alignment of cortical progenitor. How do human and/or macaque early and late neurogenic phase would compare to mouse and rat in this model? 4. In Figures 6C and 6D, the authors distinguish between cycling and non-cycling NECs and RGCs. Could the authors clarify the rationale behind making this distinction? Could the authors comment on how they interpret the impact of cycling versus non-cycling states on species-specific non-uniform scaling? Do they consider the observed non-linear correspondences to be driven by differences in cell cycle activity? 5. For the non-uniform scaling in Figure 6F, the authors identify critical inflection points and mention that "the largest delay in rat progenitors occurring where Day 1 and Day 3 progenitors overlapped". It would be good if the authors could discuss what they think all the inflection points represents. How much can it be explained by the heterogeneity within progenitors per time point? There is a clear higher spread of histograms at days 3 and 5, and the histogram at day 5 almost overlaps with day 1. I wonder if the same conclusion about non-uniform scaling would be detected if the distance matrix was built separately for specific cell types, for example only looking at NECs or RGCs. 6. The authors conclude that the elevated and prolonged expression of Wnt-ligand genes in rat RGs extend the DL neurogenic window and contribute to rat-specific expansion of deep cortical layer. In order to validate this finding it would be good for the authors to perform a perturbation experiment and reduce Wnt signalling/ Axin 2 levels in rats or depleted the Lmx1a and Lhx2 double-positive population. 7. The authors conclude that Wnt signaling is a rat specific effect since they did not observe any clear temporal change in wnt receptors in gyrencephalic species, and only a subset of RG in rats co-express Lmx1a and Lhx2. However, specific Wnt ligands and receptors (Wnt5a, Fzd and Lrp6) seem to be upregulated in human as well (Fig 7G), non RG cells could act as wnt ligand inducers in other species, and it has not been demonstrated that Lmx1a and Lhx2 are the source for Wnt ligand production. I wonder if the authors can completely rule out a role for Wnt in the protracted neurogenesis of other species.

Minor comments:

The RNAscope experiment is currently qualitative. Is it the mRNA copy number per cell equal in both species but more cells are positive in rat, or are there differences in number of mRNA molecules as well? It is not indicated if the RNAscope probes are the same for mouse and rat.

Significance

How different species achieve such remarkable differences in brain shape and size remains poorly understood. A critical aspect of this process is the duration of the neurogenic phase: the period during which neural progenitors generate neurons. This phase tends to be extended in species with larger brains and contains multiple neuronal stem cell types in varying proportions. It is thought that this accounts for their increased neuronal numbers. In their search for mechanisms that prolong neurogenesis across species, the authors propose a rat-specific role for Wnt ligands in expanding the neurogenic period in the rat brain. Importantly, they rule out that this mechanism operates in other species, such as primates or ferrets, to achieve similar extensions.

The study is of high quality, incorporating rigorous lineage-tracing experiments in two species and single-cell RNA sequencing. Previous work established a role for Wnt signaling in regulating early neurogenesis in mice. Here, the authors characterize a novel population of radial glial cells (Lmx1a and Lhx2 double-positive) that may explain increased Wnt ligand secretion in rats. However, functional validation of this mechanism is still lacking. To strengthen its evolutionary relevance, it would be important to determine whether similar effects occur during earlier neural stages in other species (such as neuroepithelium thickening), or whether other cell types have co-opted the proposed Lmx1a-Lhx2 regulatory module in other species.

From the perspective of a researcher with a stem cell and developmental background focused on neural evo-devo, this manuscript represents a solid and novel contribution. The proposed model of a rat-specific mechanism for extending the neurogenic phase contrasts with the prevailing concept of convergence in mechanisms underlying species-specific cortical development. This raises intriguing questions about how multiple molecular pathways have been co-opted to achieve similar developmental outcomes. Furthermore, we know very little about what determines the duration of specific developmental processes. This work suggests that extended Wnt signaling may account for prolonged neurogenesis in rats compared to mice. Future studies should aim to validate the proposed rat-specific co-option of an Lmx1a-Wnt ligand cascade in cortical radial glia, potentially through relief of Lhx2-mediated repression of Lmx1a.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Yamauchi and colleagues explore how species-specific differences in timing of neurogenesis may contribute to cell composition in the mature brain, using rat and mouse cortex as a main model of study. They first estimate and compare among 8 mammalian species the number of cortical neurons corresponding to deep layer (DL) and upper layer (UL) neurons. They find a species-specific relative increase of DL/UL neurons in rats, compared with all other species tested. They then explore the cellular mechanisms underlying these differences in mouse and rat, using retrovirus-based clonal analyses and EdU nuclear labeling, as well as axonal projection retrograde tracing. They conclude that the increased number of DL neurons in the rat is correlated with an increase in the period of DL neuron generation at early stages of corticogenesis. They also report a lack of obvious difference in cell cycle kinetics and indirect vs direct neurogenesis that could explain the DL/UL differences. Finally, they perform comparative scRNAseq analysis in mouse vs rat embryonic cortical cells. This first confirms at the transcriptomic level an apparent prolonged period of early neurogenesis in the rat cortex. Moreover they find among modules of co-expression detectable at these stages an increase in genes corresponding to Wnt signalling, a pathway previously linked to increased self-renewal and delayed differentiation of radial glial progenitors. They thus conclude that the species-differences in neuronal number in the rat is linked to increased Wnt signaling at a critical time of corticogenesis.

Overall this is a thorough and elegant study focused on a timely and interesting topic. The data shown are convincing and carefully interpreted. I have however a couple of comments and questions to make the study fully clear and convincing.

• The authors describe cortical neuronal counts across several mammalian species, which is quite impressive, but the information on the methods of counting is lacking: how representative are the data used / shown; how many individuals / brains / sections were used for each species considered? Much more detailed description of the quantifications should be provicded to judge the validity of this first conclusion.
• The authors use several markers of cortical neuron identity to confirm their neuron number measurements, but from the data shown in Figure 1D,E it seems that only some markers (Satb2) show species-differences while others do not (CTIP2 / Tbr1). How do the authors explain this discrepancy - does this mean that it is mainly Satb2 neurons that are increased in number? But if so how to explain the relative increase in subcortical projections shown in Figure S7?
• The authors focus their study almost exclusively on somatosensory cortex, but can they comment on other areas (motor, visual for instance)? It would be nice to provide additional comparative data on other areas, at least for some of the parameters examined acros mouse and rat. Alternatively the authors should be more explicit in the abstract and description of the study that it is limited to a single area.
• The authors provide convincing evidence of increased Wnt signaling pathway in the rat. They should show more explicitely how other classical pathways of neurogenic balance / temporal patterning are expressed in their mouse and rat transcriptome data sets. These would include Notch, FGF, BMP, for which all the data should be available to provide meaningful species comparison.
• The alignment of mouse and rat trajectories is very nicely showing a delay at early-mid-corticogenesis. But there is also heterochronic transcriptome at latest stages (end of 5). How can this be interpreted? Does this mean potentially prolonged astrogliogenesis in the rat cortex?
• Figure 7: description implies that module 3 is a subset of module 4, but this is not obvious at all from the panels shown. Please clarify.

Significance

The topic of the study if of general interest and original, and the conclusions original and important. The approaches used are state of the art and applied in an elegant fashion to the topic. This study should be of broad interest to developmental neurobiologists, but also developmental biologists interesting in temporal patterning and developmental timing across species.

Transparent Peer Review

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

We thank the reviewers and editors for their careful evaluation of our manuscript and their positive comments on the importance and rigor of the work. Below you will find our point-by-point response to each reviewer's suggestions. We believe that we have addressed (in the response and the revised manuscript) all of the concerns. Please note that in some cases, we have numbered a reviewer's comments for clarity, however beyond this, we have not altered any of the reviewers' text.

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

Lo et al., report a high-throughput functional profiling study on the gene encoding for argininosuccinate synthase (ASS1), done in a yeast experimental system. The study design is robust (see lines 141-143, main text, Methods), whereby "approximately three to four independent transformants of each variant would be isolated and assayed." (lines 140 - 141, main text, Methods). Such a manner of analysis will allow for uncertainty of the functional readout for the tested variants to be accounted for.

This is an outstanding study providing insights on the functional landscape of ASS1. Functionally impaired ASS1 may cause citrullinemia type I, and disease severity varies according to the degree of enzyme impairment (line 30, main text; Abstract). Data from this study forms a valuable resource in allowing for functional interpretation of protein-altering ASS1 variants that could be newly identified from large-scale whole-genome sequencing efforts done in biobanks or national precision medicine programs. I have some suggestions for the Authors to consider:

1. The specific function of ASS1 is to condense L-citrulline and L-aspartate to form argininosuccinate. Instead of measuring either depletion of substrate or formation of product, the Authors elected to study 'growth' of the yeast cells. This is a broader phenotype which could be determined by other factors outside of ASS1. Whereas i agree that the experiments were beautifully done, the selection of an indirect phenotype such as ability of the yeast cells to grow could be more vigorously discussed.

We appreciate the reviewer's point regarding the indirect nature of growth as a functional readout. In our system, yeast growth is tightly and specifically coupled to ASS enzymatic activity. The strains used are isogenic and lack the native yeast argininosuccinate synthetase, such that arginine biosynthesis, and therefore yeast replication on minimal medium lacking arginine, depends exclusively on the activity of human ASS1. Under these defined and limiting conditions, growth provides a quantitative proxy for ASS1 function. However, we acknowledge that this assay does not resolve specific molecular mechanisms underlying reduced function, such as altered catalytic activity versus effects on protein stability. We have updated the text to clarify these points.

"While growth is an indirect phenotype relative to direct measurement of substrate turnover or product formation, it is tightly coupled to ASS enzymatic activity in this system and is expected to be impaired by amino acid substitutions that reduce catalytic activity or protein stability. Therefore, growth on minimal medium lacking arginine is a quantitative measure of ASS enzyme function, allowing the impact of ASS1 missense variants to be assessed at scale through a high-throughput growth assay, in a single isogenic strain background, under controlled, defined conditions that limit confounding factors unrelated to ASS1 activity. We expect that the assay will detect reductions in both catalytic activity and protein stability but will not distinguish between these mechanisms."

1. One of the key reasons why studies such as this one are valuable is due to the limitations of current variant classification methods that rely on 'conservation' status of amino acid residues to predict which variants might be 'pathogenic' and which variants might be 'likely benign'. However, there are serious limitations, and Figures 2 and 6 in the main text shows this clearly. Specifically, there is an appreciable number of variants that, despite being classified as "ClinVar Pathogenic", were shown by the assay to unlikely be functionally impaired. This should be discussed vigorously. Could these inconsistencies be potentially due to the read out (growth instead of a more direct evaluation of ASS1 function)?

We interpret this discrepancy as reflecting a sensitivity limitation of the growth-based readout rather than a fundamental disagreement between functional effect and clinical annotation. Specifically, we believe that our assay is unable to resolve the very mildest hypomorphic variants from true wild type, i.e., the residual activity of these variants is sufficient to fully support yeast growth under the conditions used. On this basis, we have chosen not to treat wild-type-like growth in our assay as informative for benignity; conversely, reduced growth provides evidence supporting pathogenicity (all clinically validated variants examined in this range are pathogenic).

We have revised the manuscript to clarify this point explicitly and to frame these variants as lying outside the effective resolution limit of the assay rather than representing true false positives. Additional discussion of this limitation and its implications is provided in our responses to Reviewer 2 (points 1 and 4) along with specific changes made to the text.

1. Figure 3 is very interesting, showing a continuum of functional readout ranging from 'wild-type' to 'null'. It is very interesting that the Authors used a threshold of less than 0.85 as functionally hypomorphic. What does this mean? It would be very nice if they have data from patients carrying two hypomorphic ASS1 alleles, and correlate their functional readout with severity of clinical presentation. The reader might be curious as to the clinical presentation of individuals carrying, for example, two ASS1 alleles with normalized growth of 0.7 to 0.8.

I hope you will find these suggestions helpful.

We thank the reviewer for this thoughtful comment. Figure 3 indeed illustrates a continuum of functional effects, and we agree that careful interpretation of the thresholds used is important. To clarify the rationale for the hypomorphic threshold, the interpretation of intermediate growth values, and to emphasize that these labels reflect only behavior in the functional assay, we have rewritten the relevant section of the Results:

"The normalized growth scores of the 2,193 variants tested in our functional assay form a clear bimodal distribution (Figure 3), with two distinct peaks corresponding to functional extremes, as is commonly reported in large-scale functional assays of protein function [9, 10]. The smaller peak, centered around the null control (normalized growth = 0), represents variants that fail to support growth in the assay (growth 0.85). Variants with growth values falling between these two peak-based thresholds display partial functional impairment and are classified as functionally hypomorphic (n = 323). Crucially, these classifications are entirely derived from the observed peaks in the distribution of growth values and reflect differences in functional activity under the assay conditions. They do not provide direct evidence for clinical pathogenicity or benignity and should not be used for clinical variant interpretation without proper benchmarking against clinical reference datasets, as implemented below within an OddsPath framework."

We agree with the reviewer that correlating functional measurements with clinical severity in individuals carrying two hypomorphic ASS1 alleles would be highly informative, particularly given that ASS1 deficiency is an autosomal recessive disorder. While mild hypomorphic variants (for example, variants with normalized growth values of 0.7-0.8 in our assay) could plausibly contribute to disease when paired with a complete loss-of-function allele, systematic analysis of combinatorial genotype effects and genotype-phenotype correlations is beyond the scope of the present study, which focuses on the functional effects of individual variants. We view this as an important direction for future work.

Reviewer #1 (Significance (Required)):

This is an outstanding study providing insights on the functional landscape of ASS1. Functionally impaired ASS1 may cause citrullinemia type I, and disease severity varies according to the degree of enzyme impairment (line 30, main text; Abstract). Data from this study forms a valuable resource in allowing for functional interpretation of protein-altering ASS1 variants that could be newly identified from large-scale whole-genome sequencing efforts done in biobanks or national precision medicine programs.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

In this manuscript, Lo et al characterize the phenotypic effect of ~90% of all possible ASS1 missense mutations using an elegant yeast-based system, and use this dataset to aid the interpretation of clinical ASS1 variants. Overall, the manuscript is well-written and the experimental data are interpretated rigorously. Of particular interest is the identification of pairs of deleterious alleles that rescue ASS1 activity in trans. My comments mainly pertain to the relevance of using a yeast screening methodology to infer functional effects of human ASS1 mutations.

1. Since human ASS1 is heterologously expressed in yeast for this mutational screen, direct comparison of native expression levels between human cells and yeast is not possible. Could the expression level of human ASS1 (driven by the pARG1 promoter) in yeast alter the measured fitness defect of each variant? For instance, if ASS1 expression in yeast is sufficiently high to mask modest reductions in catalytic activity, such variants may be misclassified as hypomorphic rather than amorphic. Conversely, if expression is intrinsically low, even mild catalytic impairments could appear deleterious. While it is helpful that the authors used non-human primate SNV data to calibrate their assay, experiments could be performed to directly address this possibility.

The nature of the relationship between yeast growth and availability of functional ASS1 could also influence the interpretation of results from the yeast-based screen. Does yeast growth scale proportionately with ASS1 enzymatic activity?

We completely agree that the expression level of human ASS1 in yeast could influence the measured fitness effects of individual variants. We expect the rank ordering of variants in our growth assay to reflect their relative enzymatic activity (i.e. a monotonic relationship) but acknowledge that the precise mapping between activity and growth is unknown and may include ceiling and floor effects that limit the assay's dynamic range. As the reviewer notes, under high expression conditions moderate loss-of-function variants could appear indistinguishable from wild type (ceiling effect), whereas under lower expression the same variants could behave closer to the null control (floor effect).

In our system, ASS1 is expressed from the pARG1 promoter, chosen under the assumption that the native expression level of ARG1 (the yeast ASS1 ortholog) is appropriately tuned for yeast growth. Crucially, rather than assuming a fixed mapping from assay growth to clinical pathogenicity (given potential nonlinearities in the relationship between ASS function and growth) we benchmark the assay against external data, including known pathogenic and benign variants and non-human primate SNVs, to calibrate thresholds and guide interpretation within an OddsPath framework. This benchmarking indicates that ceiling effects are likely present, with some mild loss-of-function pathogenic variants appearing indistinguishable from wild type in the growth assay. We explicitly account for this by not using high-growth scores as evidence toward benignity. We have made the following changes the manuscript:

"A subset of clinically pathogenic ASS1 variants exhibit near-wild-type growth in our yeast assay. In general, we expect a monotonic relationship between ASS function and yeast growth, but with the potential for floor and ceiling effects that constrain the assay's dynamic range. In this context, we interpret high-growth pathogenic variants as likely causing mild loss of function that cannot be distinguished from wild type in our assay"

"Based on these findings and given that 22/56 pathogenic variants show >85% growth, we conclude that growth above this threshold should not be used as evidence toward benignity."

1. It would be helpful to add an additional diagram to Figure 1A explaining how the screen was performed, in particular: when genotype and phenotype were measured, relative to plating on selective vs non-selective media? This is described in "Variant library sequence confirmation" and "Measuring the growth of individual isolates" of the Methods section but could also be distilled into a diagram.

We thank the reviewer for this helpful suggestion. We have updated Figure 1 by adding a new schematic panel (Figure 1C) that distills the experimental workflow into a visual overview. This diagram is intended to complement the detailed descriptions in the Methods and improve clarity for the reader.

1. The authors rationalize the biochemical consequences of ASS1 mutations in the context of ASS1 per se - for example, mutations in the active site pocket impair substrate binding and therefore catalytic activity, which is expected. Does ASS1 physically interact with other proteins in human cells, and could these interactions be altered in the presence of specific ASS1 mutations? Such effects may not be captured by performing mutational scanning in yeast.

We are not aware of any specific protein-protein interactions involving ASS that are required for its enzymatic function. However, we agree that ASS could engage in non-essential interactions with other human proteins that might be altered by specific missense variants and that such interactions would not necessarily be captured in a yeast-based assay.

Importantly, our complementation system depends on human ASS providing the essential enzymatic activity required for arginine biosynthesis in yeast. If ASS1 required obligate human-specific protein interactions to function, even the wild-type enzyme would fail to support yeast growth, which is clearly not the case. We therefore conclude that the assay robustly reports on the intrinsic enzymatic activity of ASS, while acknowledging that non-essential human-specific interactions may not be assessed. We have updated the manuscript to reflect this point.

"Importantly, successful functional complementation indicates that ASS enzymatic activity does not depend on any obligate human-specific protein interactions."

1. The authors note that only a small number (2/11) of mutations at the ASS1 monomer-monomer interface lead to growth defects in yeast. It would be helpful for the authors to discuss this further.

As discussed in response to the reviewer's comments on the relationship between ASS activity and yeast growth (point 1 above), we expect growth to be a monotonic but nonlinear function of enzymatic activity, with potential ceiling effects at high activity. Under this model, variants causing weak or moderate loss of function may remain indistinguishable from wild type when residual activity is sufficient to support normal growth. We favor this explanation for the observation that only 2/11 interface variants show reduced growth, as many pathogenic interface substitutions are associated with milder disease presentations, consistent with higher residual enzyme function. Consistent with this interpretation, variants affecting the active site, where substitutions are expected to cause large reductions in catalytic activity, are readily detected by the assay.

Although we cannot exclude partial buffering of dimerization defects in yeast, we interpret the reduced sensitivity to interface variants primarily as a general limitation of growth-based assays. Accordingly, our decision not to use growth >85% as evidence toward benignity is conservative relative to approaches that would classify high-growth variants as benign except at the monomer-monomer interface, avoiding reliance on structural subclassification and minimizing the risk of false benign interpretation. Reduced growth, by contrast, provides strong evidence of loss of ASS1 function and pathogenicity, validated under the OddsPath framework.

We have updated the Results and Discussion sections to clarify these points (also see response to the reviewer's point 1).

"A subset of clinically pathogenic ASS1 variants exhibit near-wild-type growth in our yeast assay. In general, we expect a monotonic relationship between ASS function and yeast growth, but with the potential for floor and ceiling effects that constrain the assay's dynamic range. In this context, we interpret high-growth pathogenic variants as likely causing mild loss of function that cannot be distinguished from wild type in our assay. Consistent with this view, many pathogenic variants with high assay growth are located at the monomer-monomer interface rather than the active site, and are associated with milder or later-onset clinical presentations, suggesting partial enzymatic impairment that is clinically relevant in humans but not resolved by the yeast assay."

"Based on these findings and given that 22/56 pathogenic variants show >85% growth, we conclude that growth above this threshold should not be used as evidence toward benignity. Notably, this approach is conservative relative to treating high-growth variants as benign except at the monomer-monomer interface, avoiding reliance on structural subclassification and minimizing the risk of false benign interpretation arising from assay ceiling effects. Conversely, the variants with

Reviewer #2 (Significance (Required)):

This study presents the first comprehensive mutational profiling of human ASS1 and would be of broad interest to clinical geneticists as well as those seeking biochemical insights into the enzymology of ASS1. The authors' use of a yeast system to profile human mutations would be particularly useful for researchers performing deep mutational scans, given that it provides functional insights in a rapid and inexpensive manner.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

Section 1 - Evidence, reproducibility, and clarity Summary This manuscript presents a comprehensive functional profiling of 2,193 ASS1 missense variants using a yeast complementation assay, providing valuable data for variant interpretation in the rare disease citrullinemia type I. The dataset is extensive, technically sound, and clinically relevant. The demonstration of intragenic complementation in ASS1 is novel and conceptually important. Overall, the study represents a substantial contribution to functional genomics and rare disease variant interpretation.

Major comments 1. This is an exciting paper as it can provide support to clinicians to make actionable decisions when diagnosing infants. I have a few major comments, but I want to emphasize the label of "functionally unimpaired" variants to be misleading. The authors explain that there are several pathogenic ClinVar variants that fall into this category (above the >.85 growth threshold) but I think this category needs a more specific name and I would ask the authors to reiterate the shortcomings of the assay again in the Discussion section.

We thank the reviewer for raising this important point. We agree that the label "functionally unimpaired" could be misleading if interpreted as implying clinical benignity rather than assay behavior. We have therefore clarified that this designation refers strictly to variant behavior in the yeast growth assay and does not imply absence of pathogenicity.

In addition, we have expanded the Discussion to explicitly address the existence of clinically pathogenic variants with high growth scores (>0.85), emphasizing that these likely reflect a ceiling effect of the assay and represent a key limitation for interpretation. This clarification reiterates that high-growth scores should not be used as evidence toward benignity, while reduced growth provides strong functional evidence of pathogenicity. Relevant revisions are described in our responses to Reviewers 1 and 2.

1. I think there's an important discussion to be had here, is the assay detecting variants that alter the function of ASS or is it detecting a complete ablation of enzymatic activity? The results might be strengthened with a follow-up experiment that identifies stably expressed ASS1 variants.

We agree with the review that distinguishing between stability and enzyme activity would be valuable information. Unfortunately, we do not currently have the resources to perform this type of large-scale study. We have acknowledged in the text that our assay does not distinguish between enzyme activity and protein stability:

"We expect that the assay will detect reductions in both catalytic activity and protein stability, but will not distinguish between these mechanisms."

At the very least, it would be great to see the authors replicate some of their interesting results from the high-throughput screen by down-selecting to ~12 variants of uncertain significance that could be newly considered pathogenic.

We have included new analysis of all 25 VUS variants falling in the pathogenic range of our assay (Supplemental Table S7). Reclassification under current guidelines (in the absence of our data) shifts six variants to Pathogenic/Likely Pathogenic and 11 more are reclassified to Likely Pathogenic with the application of our functional data as PS3_Supporting. The remaining eight VUS are all reclassified to Likely Pathogenic when inclusion of homozygous PrimateAI-benign variants allows the assay to satisfy full PS3 criteria.

1. I would ask the authors to provide more citations of the literature in the introduction of the manuscript. I would be especially interested in knowing more about human ASS being identified as a homolog of yeast ARG1, as they share little sequence similarity (27.5%) at the protein level. That said, I find the yeast complementation assay exciting.

We thank the reviewer for this suggestion. Human ASS and yeast Arg1 catalyze the same biochemical reaction and share approximately 49% amino acid sequence identity. We have revised the Introduction to clarify this relationship and to note explicitly that the Saccharomyces Genome Database (SGD) identifies the human gene encoding argininosuccinate synthase (ASS1) as the ortholog of yeast ARG1. An appropriate citation has been added to support this statement. The protein alignments have been provided as File S2.

"This assay is based on the ability of human ASS to functionally replace (complement) its yeast ortholog (Arg1) in S. cerevisiae (Saccharomyces Genome Database, 2026). Importantly, successful functional complementation indicates that ASS enzymatic activity does not depend on any obligate human-specific protein interactions. At the protein level, human ASS and yeast Arg1 display 49% sequence identity (File S2) and share identical enzymatic roles in converting citrulline and aspartate into argininisuccinate."

1. I appreciate the efforts made by the authors to share their work and make this study more reproducible, such as sharing the hASS1 and yASS1 plasmids being shared on NCBI Genbank (Line 121) and publishing the ONT reads on SRA (Line 154). I made a requests for additional data to be shared, such as the custom method/code for codon optimization and a table of Twist variant cassettes that were ordered. I would also love to see these results shared on MaveDB.org.

We thank the reviewer for these suggestions regarding data sharing and reproducibility. As requested, we have provided the custom codon optimization script as File S1 and the amino acid alignment used to perform codon harmonization as File S2. The sequence of the underlying variant cassette is included in the corresponding GenBank entry, and we have clarified this point in the legend of Figure 1. For each amino acid substitution, Twist Bioscience used a yeast-specific codon scheme with a single consistent codon per amino acid; accordingly, the sequence of each variant cassette can be inferred from the base construct and the specified amino acid change. A complete list of variant amino acid substitutions used in this study is provided in Table S3.

1. I find this manuscript very exciting as the authors have a compelling assay that identifies pathogenic variants, but I was generally disappointed by the quality and organization of the figures. For example, Figure 4 provides very little insight, but could be dramatically improved with an overlay of the normalized growth score data or highlighting variants surrounding the substrate or ATP interfaces. There are some very interesting aspects of this manuscript that could be shine through with some polished figures.

We thank the reviewer for this feedback and agree that clear and well-organized figures are essential for conveying the key results of the study. In response, we have substantially revised Figure 4 by adding colored overlays showing residue conservation and median normalized growth scores (new panels Figure 4C and 4D), which more directly link structural context to functional outcomes and highlight patterns surrounding the active site and substrate interfaces.

I would also encourage the authors to generate a heatmap of the data represented in Figure 2 (see Fowler and Fields 2014 PMID 25075907, Figure 2), this would be more helpful reference to the readers.

The reviewer also suggested that a heatmap representation, similar to that used in Fowler and Fields (2014), might aid interpretation of the data shown in Figure 2. Because our dataset consists of sparse single-amino acid substitutions rather than a complete mutational scan, such heatmaps are inherently less dense and less effective at conveying patterns than in saturation mutagenesis studies. Nevertheless, to aid readers who may find this visualization useful, we have generated and included a single-nucleotide variant heatmap as Supplemental Figure S1.

My major comments are as follows: 6. Citations needed - especially in the introduction and for establishing that hASS is a homolog of yARG1

We have added the requested citations and clarified the ASS1-ARG1 orthology in the Introduction, as described in our response to point 3 above.

1. Generally, the authors do a nice job distinguishing the ASS1 gene from the ASS enzyme, though I found some ambiguities (Line 685). Please double-check the use of each throughout the manuscript.

We have edited the manuscript to ensure consistent and unambiguous use of gene and enzyme nomenclature throughout.

1. Generally, I'm confused about what strain was used for integrating all these variants, was is the arg1 knock-out strain from the yeast knockout collection or was it FY4? I think FY4 was used for the preliminary experiments, then the KO collection strain was used for making the variant library but I think this could be made more clear in the text and figures. Lines 226-229 describes introducing the hASS1 and yASS1 sequences into the native ARG1 locus in strain FY4, but the Fig1A image depicts the ASS1 variants going into arg1 KO locus. Fig1A should be moved to Fig2.

We agree that the strain construction steps were not described as clearly as they could have been. We have therefore clarified the strain construction workflow in the Materials & Methods and Results sections, as well as in the Figure 1 legend, to explicitly distinguish preliminary experiments performed in strain FY4 from construction of the variant library in the arg1 knockout background.

As we have also added an additional panel to Figure 1 that schematically explains how the screen was performed (per Reviewer #2's request), we believe that Figure 1A is appropriately placed and should remain in Figure 1.

1. Line 303 - "We classify these variants as 'functionally unimpaired'", this is not an accurate description of these variants as Figure 2 highlights 24 pathogenic ClinVar variants that would fall into this category of "functionally unimpaired". The yeast growth assay appears to capture pathogenic variants, but there is likely some nuance of human ASS functionality that is not being assessed here. I would make the language more specific, e.g. "complementary to Arg1" or "growth-compatible".

We agree that the label "functionally unimpaired" could be misinterpreted if read as implying clinical benignity. We have therefore clarified within the manuscript that this designation refers strictly to variant behavior in the yeast growth assay (i.e., wild-type-like growth under assay conditions) and does not imply absence of pathogenicity. We also expanded the Discussion to explicitly address the subset of clinically pathogenic variants with high growth scores (>0.85), consistent with a ceiling effect of the assay and a key limitation for interpretation. See response to reviewer #3 point 1. Relevant revisions are also discussed in our responses to Reviewers #1 and #2.

1. Lines 345-355 - It is interesting that there are variants that appear functional at the substrate interfacing sites. Is there anything common across these variants? Are they maintaining the polarity or hydrophobicity of the WT residue? Are any of these variants included in ClinVar or gnomAD? Are pathogenic variants found at any of these sites

Yes. For highly sensitive active-site residues that have few permissible variants, the vast majority of amino acid substitutions that do retain activity preserve key physicochemical properties of the wild-type residue, such as hydrophobicity or charge. We have added this important observation to the manuscript:

"Any variants at these sensitive residues that are permissive for activity in our assay retain hydrophobicity or charged states relative to the original amino acid side chain (Figure 5A & Table S5)."

None of these variants are present in ClinVar. Only L15V and E191D are present in gnomAD (Table S4).

1. Lines 423-430 - The OddsPath calculation would seem to rely heavily on the thresholds of .85 for normalized growth. The OddsPath calculation could be bolstered with some additional analysis that emphasizes the robustness to alternative thresholds.

We agree that the sensitivity of the OddsPath calculation to the choice of growth thresholds is an important consideration. In our assay, benign ClinVar variants and non-human primate variants are observed exclusively within the peak centered on wild-type growth, whereas clinically annotated variants falling below this peak are exclusively pathogenic. On this basis, we defined the upper boundary of the assay range interpreted as supporting pathogenicity as the lower boundary of the wild-type-centered peak in the growth distribution (as defined in Figure 3), rather than selecting a cutoff by direct optimization of the OddsPath. This choice reflects the observed concordance, in our dataset, between the onset of measurable functional impairment in the assay and clinical pathogenic annotation. Importantly, in practice the OddsPath value is locally robust to the precise placement of this boundary, remaining invariant across the range 0.82-0.88. Supporting our chosen threshold of 0.85, the lowest-growth benign or primate variant observed has a normalized growth value of 0.88, while the lowest growth observed among variants present as homozygotes in gnomAD was 0.86. We have clarified this rationale and analysis in the revised manuscript.

"Notably, the "Among all nine of the human ASS1 missense variants observed as homozygotes in gnomAD which were tested as amino acid substitutions in our assay, the lowest observed growth value was 0.86 (Ala258Val) consistent with the lower boundary of the PrimateAI variants which was a growth value of 0.87 (Ala81Thr) (Figure 6) and with our use of a 0.85 classification threshold."

"If we treat PrimateAI variants as benign (solely for OddsPath calculation purposes), the OddsPath for growth

1. Lines 432-441 - This is an interesting idea to use variants observed in primates, has ACMG weighed in on this? I understand that CTLN1 is an autosomal recessive disorder but I'd still be interested in seeing how the observed ASS1 missense variants in gnomAD perform in your growth assay, possibly a supplemental figure?

To our knowledge, the ACMG/AMP guidelines do not currently address the use of homozygous missense variants observed in non-human primates. We are currently in discussion with two ClinGen working groups to discuss the possibility of formalizing the use of this data source.

We agree that comparison with human population data is also important. Accordingly, total gnomAD allele counts and homozygous counts for all applicable ASS1 missense variants are provided in Table S4, and the growth behavior of ASS1 missense variants observed in the homozygous state in gnomAD is shown in Figure 6. These homozygous variants uniformly exhibit high growth in our assay, consistent with the absence of strong loss-of-function effects. We have updated the manuscript text to clarify these points.

Minor comments 1. Lines 53-59 - This paragraph needs to cite the literature, especially lines 56, 57, and 59 2. Line 61 - no need to repeat "citrullinemia type I", just use the abbreviation as it was introduced in the paragraph above 3. Lines 61-71 - again, this paragraph needs more literature citations 4. Line 62 - change to "results"

The changes suggested in points 1-4 have all been implemented in the revised manuscript.

1. Line 74-75 - "RUSP" acronym not needed as it's never used in the manuscript, the same goes for "HHS"

We agree that the acronyms "RUSP" and "HHS" are not reused elsewhere in the manuscript. We have nevertheless retained them at first mention, alongside the expanded names, because these acronyms are commonly used in newborn screening and public health policy contexts and may be more familiar to some readers than the expanded terms. We would be happy to remove the acronyms if preferred.

1. Line 86 - "ASS1" I think is referring to the enzyme and should just be "ASS"? If referring to the gene then italicize to "ASS1"
2. Lines 91-93 - It would be helpful to mention this is a functional screen in yeast
3. Line 101 - It would be helpful to the readers to define SD before using the acronym, consider changing to "minimal synthetic defined (SD) medium" and afterwards can refer to as "SD medium"
4. 109-114 - It would be great if you could share your method for designing the codon-harmonized yASS1 gene, consider sharing as a supplemental script or creating a GitHub repository linked to a Zenodo DOI for publication.

The changes suggested in points 6-9 have all been implemented in the revised manuscript. The codon harmonization script has been provided as File S1.

1. Lines 135-137 - I think it's helpful to provide a full table of the cassettes ordered from Twist as well as the primers used to amplify them, consider a supplemental table.

Details of Twist cassette and the primer sequences used for amplification have been added to the Materials & Methods.

1. Line 138 - "standard methods" is a bit vague, I'm guessing this is a Geitz and Schiestl 2007 LiAc/ssDNA protocol (PMID 17401334)? Also, was ClonNAT used to select for natMX colonies?

The reviewer is correct about which protocol was used, and we have added the citation. We have also clarified that selection was carried out based on resistance to nourseothricin.

1. Line 150 - change to "sequence the entire open reading frame, as previously described [4]."
2. Line 222-223 - remove "replace" and just use "complement" (and remove the parenthesis)
3. Line 249 - It would be great to see a supplemental alignment of the hASS1 and yASS1 sequences.
4. Line 261 - spelling "citrullemia" should be corrected to "citrullinemia"
5. Line 280 - "using Oxford Nanopore sequencing" is a bit vague, I suggest specifying the equipment used (e.g. Oxford Nanopore Technologies MinION platform) or simplify to "via long-read sequencing (see Materials & Methods)"

The changes suggested in points 12-16 have all been implemented in the revised manuscript. An alignment of the ASS and Arg1 protein sequences has been provided as File S2.

1. Line 287-289 - It would be great to see the average number of isolates per variant, as well as a plot of the variant growth estimate vs individual isolate growth.

We agree with the reviewer that conveying measurement precision is important. The number of isolates assayed per variant is provided in Table S4, and we have added explicit mention of this in the text. Because variants were assayed with a mixture of 1, 2, or {greater than or equal to}3 independent isolates, a scatterplot of variant-level growth estimates versus individual isolate measurements would be difficult to interpret and potentially misleading. Instead, we report standard error estimates for each variant in Table S4, derived from the linear model used to estimate growth effects, which more appropriately summarizes measurement uncertainty given the experimental design.

1. Lines 324-25 - consider removing the last sentence of this paragraph, it is redundant as the following paragraph starts with the same statement.

We have removed this sentence.

1. Lines 327-335 - This is interesting and would benefit from its own subpanel or plot in which the normalized growth score is plotted against variants that are at conserved or diverse residues in human ASS, and see if there's a statistical difference in score between the two groupings.

As suggested by the reviewer, we have added Supplemental Figure 2 (Figure S2) in which the normalized growth score of each variant is plotted against the conservation of the corresponding residue, as measured by ConSurf. The manuscript already includes a statistical analysis of the relationship between residue conservation and functional impact, showing that amorphic variants occur significantly more frequently at highly conserved residues than unimpaired variants do (one-sided Fisher's exact test). We now refer to this new supplemental figure in the relevant Results section.

1. Lines 339-341 - As written, it is unclear if aspartate interacts with all of the same residues as citrulline or just Asn123 and Thr119.
2. Lines 345-355 - As with my above comment, I find this interesting and would
3. Line 353 - add a period to "al" in "Diez-Fernandex et al."

The issues raised in points 20 and 22 have all addressed. Point 21 appears to be truncated.

1. Figure 1 a. Remove "Figure" from the subpanels and show just "A" and "B" (as you do for Figure 4) and combine the two images into a single image. Also make this correction to Figure 5 and Figure 8. b. Panel A - I thought the hASS1 and yASS1 were dropped into FY4, not the arg1 KO strain. This needs clarification. c. Panel A - I'm assuming the natMX cassette contains its own promoter, you could use a right-angled arrow to indicate where the promotors are in your construct. d. Panel B - I'm not sure the bar graph is necessary, it would be more helpful to see calculations of the colony size (or growth curves for each strain) and plot the raw values (maybe pixel counts?) for each replicate rather than normalizing to yeast ARG1. I would be great to have a supplemental figure showing all the replicates side-by-side. e. Panel B - Would be helpful to denote the pathogenic and benign ClinVar variants with an icon or colored text.

f. Figure 1 Caption - make "A)" and "B)" bold.

We have implemented the requested changes in Figure 1 with the following exceptions. We have retained panels A and B as separate subfigures because they illustrate distinct experimental concepts. In addition, we respectfully disagree with point (d). The bar graph is intended to provide a clear, high-level comparison of functional complementation by hASS1 versus yASS1 and to illustrate the gross differences in growth between benign and pathogenic proof-of-principle variants. As the bar graph includes error bars for standard deviations, presenting raw colony size measurements or growth curves for individual replicates would substantially complicate the figure without materially improving interpretability for this purpose.

1. Figure 2 a. "Shown in magenta are amino acid substitutions corresponding to ClinVar pathogenic, pathogenic/likely pathogenic, and likely pathogenic variants" is repeated in the figure caption. b. "Shown in green are amino acid substitutions corresponding to ClinVar benign and likely benign variants." I don't see any green points. c. Identify the colors used for ASS1 substrate binding residues. d. This plot would benefit from a depiction of the human ASS secondary structure and any protein domains (nucleotide-binding domain, synthase domain, and C-terminal helix from Fig4B)

e. Line 685 675 - "ASS1" is being used in reference to the enzyme, is this correct or should it be "ASS"?

We have made the requested changes to Figure 2. The repeated caption text has been removed, and references to green points have been corrected to orange points to match the figure. The colors used to indicate ASS substrate-binding residues are explicitly described in the figure key. Secondary structure annotations have been added. References to the enzyme have been corrected to "ASS" rather than "ASS1" where appropriate.

1. Figure 3 a. Rename the "unimpaired" category as there are several pathogenic ClinVar variants that fall into this category.

To address this point, we have clarified the labeling by adding "in our yeast assay" to the figure legend, making explicit that the "unimpaired" category refers only to wild-type-like behavior under assay conditions and does not imply clinical benignity. See also response to Reviewer #3, Major Comment 1.

1. Figure 4 a. List the PDB or AlphaFold accession used for this structure b. Panel A - state which colors are used for to depict each monomer. It is confusing to see several shades of pink/purple used to depict a single monomer in Panel A. c. It is very difficult to make out the aspartate and citrulline substrates in the catalytic binding activity, consider making an inset zooming-in on this domain and displaying a ribbon diagram of the structure rather than the surface. d. Generally, it would be more helpful here to label any particular residues that were identified as pathogenic from your screen, or to overlay average grow scores per residue data onto the structure

We have implemented the requested changes to Figure 4. The relevant PDB/AlphaFold accession is now listed, and the colors used to depict each monomer in Panel A are clarified in the figure legend. An inset focusing on the active site has been added to improve visualization of the citrulline and aspartate substrates. In addition, we have added new panels (Figure 4C and 4D) overlaying pathogenic residues and average growth scores onto the structure to more directly link structural context with functional data.

1. Figure 5 a. Line 716 - Insert a page break to place Figure 5 on its own page b. I suggest using a heatmap for this type of plot, as it is very difficult to track which color corresponds to which residue.

c. Fig5A - This plot could be improved by identifying which residue positions interface with which substrate.

We have placed Figure 5 on its own page and added information to the legend identifying which residue positions interface with each substrate. We have retained the active-site variant strip charts raised in point (b), as we believe they effectively illustrate how the distribution of variant effects differs between residues. In addition, we have provided a supplemental heatmap showing variant growth across the entire protein (Figure S1), and individual variant scores for all residues are provided in Table S4.

1. Figure 7 a. Line 735 - Insert page break to place figure on a new page

List the PDB accession used for these images. c. For clarity I would mention "human ASS" in the figure title d. State the colors of the substrates e. Panels A and B could be combined into a single panel, making it easier to distinguish the active site and dimerization variants.

f. Could be interesting to get SASA scores for the ClinVar structural variants to determine if they are surface-accessible

We have implemented the requested changes in Figure 7 with the following exceptions. For point (e), there is no single orientation of the structure that allows a clear simultaneous view of both active-site and dimerization variants; accordingly, we have retained panels A and B as separate subfigures to preserve clarity. With respect to point (f), we agree that solvent accessibility analysis could be informative in other contexts. However, such an analysis does not integrate naturally with the functional and assay-based framework of the present study and was therefore not included.

1. Figure 8 a. Panel B - overlay a square frame in the larger protein structure that depicts where the below inset is focused, and frame inset image as well.

We have framed the inset image as requested. We did not add a corresponding frame to the full protein structure, as doing so obscured structural details in the region of interest.

Reviewer #3 (Significance (Required)):

Section 2 - Significance This study represents a substantial technical, functional, and translational advance in the interpretation of missense variation in ASS1, a gene of high clinical relevance for the rare disease citrullinemia type I. Its principal strength lies in the generation of an experimentally validated functional atlas of ASS1 missense variants that covers ~90% of all SNV-accessible substitutions. The scale, internal reproducibility, and careful benchmarking of the yeast complementation assay against known pathogenic and benign variants provide a robust foundation for identifying pathogenic ASS1 variants. Particularly strong aspects include the rigorous quality control of variant identities, the quantitative nature of the functional readout, and the thoughtful integration of results into the ACMG/AMP OddsPath framework. The discovery of intragenic complementation between variants affecting distinct structural regions of the enzyme is a notable conceptual and mechanistic contribution. Limitations include the assay's reduced sensitivity to variants impacting oligomerization or subtle folding defects, and the use of yeast as a heterologous system, which may mask disease-relevant mechanisms as several pathogenic ClinVar variants were found to be "functionally unimpaired". Future work extending functional testing to additional cellular contexts or expanding genotype-level combinatorial analyses would further enhance clinical applicability. Relative to prior studies, which have relied on small numbers of patient-derived variants or low-throughput biochemical assays, this work extends the field decisively by delivering a comprehensive, variant-resolved functional map for ASS1. To the best of my current knowledge, this is the first systematic functional screen of ASS1 at this scale and the first direct experimental demonstration that ASS active sites span multiple subunits, enabling intragenic complementation consistent with Crick and Orgel's classic variant sequestration model. As such, the advance is simultaneously technical (high-throughput functional genomics), mechanistic (defining structural contributors to catalysis and epistasis), and clinical (enabling evidence-based reclassification of VUS). I find the use of homozygous non-human primate variants as an orthogonal benign calibration set both creative and controversial, my hope would be that this manuscript will prompt a productive discussion.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary

This manuscript presents a comprehensive functional profiling of 2,193 ASS1 missense variants using a yeast complementation assay, providing valuable data for variant interpretation in the rare disease citrullinemia type I. The dataset is extensive, technically sound, and clinically relevant. The demonstration of intragenic complementation in ASS1 is novel and conceptually important. Overall, the study represents a substantial contribution to functional genomics and rare disease variant interpretation.

Major comments

This is an exciting paper as it can provide support to clinicians to make actionable decisions when diagnosing infants. I have a few major comments, but I want to emphasize the label of "functionally unimpaired" variants to be misleading. The authors explain that there are several pathogenic ClinVar variants that fall into this category (above the >.85 growth threshold) but I think this category needs a more specific name and I would ask the authors to reiterate the shortcomings of the assay again in the Discussion section. I think there's an important discussion to be had here, is the assay detecting variants that alter the function of ASS or is it detecting a complete ablation of enzymatic activity? The results might be strengthened with a follow-up experiment that identifies stably expressed ASS1 variants. At the very least, it would be great to see the authors replicate some of their interesting results from the high-throughput screen by down-selecting to ~12 variants of uncertain significance that could be newly considered pathogenic. I would ask the authors to provide more citations of the literature in the introduction of the manuscript. I would be especially interested in knowing more about human ASS being identified as a homolog of yeast ARG1, as they share little sequence similarity (27.5%) at the protein level. That said, I find the yeast complementation assay exciting. I appreciate the efforts made by the authors to share their work and make this study more reproducible, such as sharing the hASS1 and yASS1 plasmids being shared on NCBI Genbank (Line 121) and publishing the ONT reads on SRA (Line 154). I made a requests for additional data to be shared, such as the custom method/code for codon optimization and a table of Twist variant cassettes that were ordered. I would also love to see these results shared on MaveDB.org. I find this manuscript very exciting as the authors have a compelling assay that identifies pathogenic variants, but I was generally disappointed by the quality and organization of the figures. For example, Figure 4 provides very little insight, but could be dramatically improved with an overlay of the normalized growth score data or highlighting variants surrounding the substrate or ATP interfaces. There are some very interesting aspects of this manuscript that could be shine through with some polished figures. I would also encourage the authors to generate a heatmap of the data represented in Figure 2 (see Fowler and Fields 2014 PMID 25075907, Figure 2), this would be more helpful reference to the readers.

My major comments are as follows:

1. Citations needed - especially in the introduction and for establishing that hASS is a homolog of yARG1
2. Generally, the authors do a nice job distinguishing the ASS1 gene from the ASS enzyme, though I found some ambiguities (Line 685). Please double-check the use of each throughout the manuscript
3. Generally, I'm confused about what strain was used for integrating all these variants, was is the arg1 knock-out strain from the yeast knockout collection or was it FY4? I think FY4 was used for the preliminary experiments, then the KO collection strain was used for making the variant library but I think this could be made more clear in the text and figures. Lines 226-229 describes introducing the hASS1 and yASS1 sequences into the native ARG1 locus in strain FY4, but the Fig1A image depicts the ASS1 variants going into arg1 KO locus. Fig1A should be moved to Fig2.
4. Line 303 - "We classify these variants as 'functionally unimpaired'", this is not an accurate description of these variants as Figure 2 highlights 24 pathogenic ClinVar variants that would fall into this category of "functionally unimpaired". The yeast growth assay appears to capture pathogenic variants, but there is likely some nuance of human ASS functionality that is not being assessed here. I would make the language more specific, e.g. "complementary to Arg1" or "growth-compatible".
5. Lines 345-355 - It is interesting that there are variants that appear functional at the substrate interfacing sites. Is there anything common across these variants? Are they maintaining the polarity or hydrophobicity of the WT residue? Are any of these variants included in ClinVar or gnomAD? Are pathogenic variants found at any of these sites
6. Lines 423-430 - The OddsPath calculation would seem to rely heavily on the thresholds of <.05 and >.85 for normalized growth. The OddsPath calculation could be bolstered with some additional analysis that emphasizes the robustness to alternative thresholds.
7. Lines 432-441 - This is an interesting idea to use variants observed in primates, has ACMG weighed in on this? I understand that CTLN1 is an autosomal recessive disorder but I'd still be interested in seeing how the observed ASS1 missense variants in gnomAD perform in your growth assay, possibly a supplemental figure?

Minor comments

1. Lines 53-59 - This paragraph needs to cite the literature, especially lines 56, 57, and 59
2. Line 61 - no need to repeat "citrullinemia type I", just use the abbreviation as it was introduced in the paragraph above
3. Lines 61-71 - again, this paragraph needs more literature citations
4. Line 62 - change to "results"
5. Line 74-75 - "RUSP" acronym not needed as it's never used in the manuscript, the same goes for "HHS"
6. Line 86 - "ASS1" I think is referring to the enzyme and should just be "ASS"? If referring to the gene then italicize to "ASS1"
7. Lines 91-93 - It would be helpful to mention this is a functional screen in yeast
8. Line 101 - It would be helpful to the readers to define SD before using the acronym, consider changing to "minimal synthetic defined (SD) medium" and afterwards can refer to as "SD medium"
9. 109-114 - It would be great if you could share your method for designing the codon-harmonized yASS1 gene, consider sharing as a supplemental script or creating a GitHub repository linked to a Zenodo DOI for publication.
10. Lines 135-137 - I think it's helpful to provide a full table of the cassettes ordered from Twist as well as the primers used to amplify them, consider a supplemental table
11. Line 138 - "standard methods" is a bit vague, I'm guessing this is a Geitz and Schiestl 2007 LiAc/ssDNA protocol (PMID 17401334)? Also, was ClonNAT used to select for natMX colonies?
12. Line 150 - change to "sequence the entire open reading frame, as previously described [4]."
13. Line 222-223 - remove "replace" and just use "complement" (and remove the parenthesis)
14. Line 249 - It would be great to see a supplemental alignment of the hASS1 and yASS1 sequences
15. Line 261 - spelling "citrullemia" should be corrected to "citrullinemia"
16. Line 280 - "using Oxford Nanopore sequencing" is a bit vague, I suggest specifying the equipment used (e.g. Oxford Nanopore Technologies MinION platform) or simplify to "via long-read sequencing (see Materials & Methods)"
17. Line 287-289 - It would be great to see the average number of isolates per variant, as well as a plot of the variant growth estimate vs individual isolate growth
18. Lines 324-25 - consider removing the last sentence of this paragraph, it is redundant as the following paragraph starts with the same statement
19. Lines 327-335 - This is interesting and would benefit from its own subpanel or plot in which the normalized growth score is plotted against variants that are at conserved or diverse residues in human ASS, and see if there's a statistical difference in score between the two groupings
20. Lines 339-341 - As written, it is unclear if aspartate interacts with all of the same residues as citrulline or just Asn123 and Thr119.
21. Lines 345-355 - As with my above comment, I find this interesting and would
22. Line 353 - add a period to "al" in "Diez-Fernandex et al."
23. Figure 1

a. Remove "Figure" from the subpanels and show just "A" and "B" (as you do for Figure 4) and combine the two images into a single image. Also make this correction to Figure 5 and Figure 8

b. Panel A - I thought the hASS1 and yASS1 were dropped into FY4, not the arg1 KO strain. This needs clarification

c. Panel A - I'm assuming the natMX cassette contains its own promoter, you could use a right-angled arrow to indicate where the promotors are in your construct

d. Panel B - I'm not sure the bar graph is necessary, it would be more helpful to see calculations of the colony size (or growth curves for each strain) and plot the raw values (maybe pixel counts?) for each replicate rather than normalizing to yeast ARG1. I would be great to have a supplemental figure showing all the replicates side-by-side

e. Panel B - Would be helpful to denote the pathogenic and benign ClinVar variants with an icon or colored text

f. Figure 1 Caption - make "A)" and "B)" bold 24. Figure 2

a. "Shown in magenta are amino acid substitutions corresponding to ClinVar pathogenic, pathogenic/likely pathogenic, and likely pathogenic variants" is repeated in the figure caption

b. "Shown in green are amino acid substitutions corresponding to ClinVar benign and likely benign variants." I don't see any green points

c. Identify the colors used for ASS1 substrate binding residues

d. This plot would benefit from a depiction of the human ASS secondary structure and any protein domains (nucleotide-binding domain, synthase domain, and C-terminal helix from Fig4B)

e. Line 685 - "ASS1" is being used in reference to the enzyme, is this correct or should it be "ASS"? 25. Figure 3

a. Rename the "unimpaired" category as there are several pathogenic ClinVar variants that fall into this category 26. Figure 4

a. List the PDB or AlphaFold accession used for this structure

b. Panel A - state which colors are used for to depict each monomer. It is confusing to see several shades of pink/purple used to depict a single monomer in Panel A

c. It is very difficult to make out the aspartate and citrulline substrates in the catalytic binding activity, consider making an inset zooming-in on this domain and displaying a ribbon diagram of the structure rather than the surface.

d. Generally, it would be more helpful here to label any particular residues that were identified as pathogenic from your screen, or to overlay average grow scores per residue data onto the structure 27. Figure 5

a. Line 716 - Insert a page break to place Figure 5 on its own page

b. I suggest using a heatmap for this type of plot, as it is very difficult to track which color corresponds to which residue

c. Fig5A - This plot could be improved by identifying which residue positions interface with which substrate 28. Figure 7

a. Line 735 - Insert page break to place figure on a new page

b. List the PDB accession used for these images

c. For clarity I would mention "human ASS" in the figure title

d. State the colors of the substrates

e. Panels A and B could be combined into a single panel, making it easier to distinguish the active site and dimerization variants

f. Could be interesting to get SASA scores for the ClinVar structural variants to determine if they are surface-accessible 29. Figure 8

a. Panel B - overlay a square frame in the larger protein structure that depicts where the below inset is focused, and frame inset image as well.

Significance

This study represents a substantial technical, functional, and translational advance in the interpretation of missense variation in ASS1, a gene of high clinical relevance for the rare disease citrullinemia type I. Its principal strength lies in the generation of an experimentally validated functional atlas of ASS1 missense variants that covers ~90% of all SNV-accessible substitutions. The scale, internal reproducibility, and careful benchmarking of the yeast complementation assay against known pathogenic and benign variants provide a robust foundation for identifying pathogenic ASS1 variants. Particularly strong aspects include the rigorous quality control of variant identities, the quantitative nature of the functional readout, and the thoughtful integration of results into the ACMG/AMP OddsPath framework. The discovery of intragenic complementation between variants affecting distinct structural regions of the enzyme is a notable conceptual and mechanistic contribution. Limitations include the assay's reduced sensitivity to variants impacting oligomerization or subtle folding defects, and the use of yeast as a heterologous system, which may mask disease-relevant mechanisms as several pathogenic ClinVar variants were found to be "functionally unimpaired". Future work extending functional testing to additional cellular contexts or expanding genotype-level combinatorial analyses would further enhance clinical applicability.

Relative to prior studies, which have relied on small numbers of patient-derived variants or low-throughput biochemical assays, this work extends the field decisively by delivering a comprehensive, variant-resolved functional map for ASS1. To the best of my current knowledge, this is the first systematic functional screen of ASS1 at this scale and the first direct experimental demonstration that ASS active sites span multiple subunits, enabling intragenic complementation consistent with Crick and Orgel's classic variant sequestration model. As such, the advance is simultaneously technical (high-throughput functional genomics), mechanistic (defining structural contributors to catalysis and epistasis), and clinical (enabling evidence-based reclassification of VUS). I find the use of homozygous non-human primate variants as an orthogonal benign calibration set both creative and controversial, my hope would be that this manuscript will prompt a productive discussion.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

In this manuscript, Lo et al characterize the phenotypic effect of ~90% of all possible ASS1 missense mutations using an elegant yeast-based system, and use this dataset to aid the interpretation of clinical ASS1 variants. Overall, the manuscript is well-written and the experimental data are interpretated rigorously. Of particular interest is the identification of pairs of deleterious alleles that rescue ASS1 activity in trans. My comments mainly pertain to the relevance of using a yeast screening methodology to infer functional effects of human ASS1 mutations.

• Since human ASS1 is heterologously expressed in yeast for this mutational screen, direct comparison of native expression levels between human cells and yeast is not possible. Could the expression level of human ASS1 (driven by the pARG1 promoter) in yeast alter the measured fitness defect of each variant? For instance, if ASS1 expression in yeast is sufficiently high to mask modest reductions in catalytic activity, such variants may be misclassified as hypomorphic rather than amorphic. Conversely, if expression is intrinsically low, even mild catalytic impairments could appear deleterious. While it is helpful that the authors used non-human primate SNV data to calibrate their assay, experiments could be performed to directly address this possibility.
• The nature of the relationship between yeast growth and availability of functional ASS1 could also influence the interpretation of results from the yeast-based screen. Does yeast growth scale proportionately with ASS1 enzymatic activity?
• It would be helpful to add an additional diagram to Figure 1A explaining how the screen was performed, in particular: when genotype and phenotype were measured, relative to plating on selective vs non-selective media? This is described in "Variant library sequence confirmation" and "Measuring the growth of individual isolates" of the Methods section but could also be distilled into a diagram.
• The authors rationalize the biochemical consequences of ASS1 mutations in the context of ASS1 per se - for example, mutations in the active site pocket impair substrate binding and therefore catalytic activity, which is expected. Does ASS1 physically interact with other proteins in human cells, and could these interactions be altered in the presence of specific ASS1 mutations? Such effects may not be captured by performing mutational scanning in yeast.
• The authors note that only a small number (2/11) of mutations at the ASS1 monomer-monomer interface lead to growth defects in yeast. It would be helpful for the authors to discuss this further.

Significance

This study presents the first comprehensive mutational profiling of human ASS1 and would be of broad interest to clinical geneticists as well as those seeking biochemical insights into the enzymology of ASS1. The authors' use of a yeast system to profile human mutations would be particularly useful for researchers performing deep mutational scans, given that it provides functional insights in a rapid and inexpensive manner.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Lo et al., report a high-throughput functional profiling study on the gene encoding for argininosuccinate synthase (ASS1), done in a yeast experimental system. The study design is robust (see lines 141-143, main text, Methods), whereby "approximately three to four independent transformants of each variant would be isolated and assayed." (lines 140 - 141, main text, Methods). Such a manner of analysis will allow for uncertainty of the functional readout for the tested variants to be accounted for.

This is an outstanding study providing insights on the functional landscape of ASS1. Functionally impaired ASS1 may cause citrullinemia type I, and disease severity varies according to the degree of enzyme impairment (line 30, main text; Abstract). Data from this study forms a valuable resource in allowing for functional interpretation of protein-altering ASS1 variants that could be newly identified from large-scale whole-genome sequencing efforts done in biobanks or national precision medicine programs. I have some suggestions for the Authors to consider:

1. The specific function of ASS1 is to condense L-citrulline and L-aspartate to form argininosuccinate. Instead of measuring either depletion of substrate or formation of product, the Authors elected to study 'growth' of the yeast cells. This is a broader phenotype which could be determined by other factors outside of ASS1. Whereas i agree that the experiments were beautifully done, the selection of an indirect phenotype such as ability of the yeast cells to grow could be more vigorously discussed.
2. One of the key reasons why studies such as this one are valuable is due to the limitations of current variant classification methods that rely on 'conservation' status of amino acid residues to predict which variants might be 'pathogenic' and which variants might be 'likely benign'. However, there are serious limitations, and Figures 2 and 6 in the main text shows this clearly. Specifically, there is an appreciable number of variants that, despite being classified as "ClinVar Pathogenic", were shown by the assay to unlikely be functionally impaired. This should be discussed vigorously. Could these inconsistencies be potentially due to the read out (growth instead of a more direct evaluation of ASS1 function)?
3. Figure 3 is very interesting, showing a continuum of functional readout ranging from 'wild-type' to 'null'. It is very interesting that the Authors used a threshold of less than 0.85 as functionally hypomorphic. What does this mean? It would be very nice if they have data from patients carrying two hypomorphic ASS1 alleles, and correlate their functional readout with severity of clinical presentation. The reader might be curious as to the clinical presentation of individuals carrying, for example, two ASS1 alleles with normalized growth of 0.7 to 0.8.

I hope you will find these suggestions helpful.

Significance

This is an outstanding study providing insights on the functional landscape of ASS1. Functionally impaired ASS1 may cause citrullinemia type I, and disease severity varies according to the degree of enzyme impairment (line 30, main text; Abstract). Data from this study forms a valuable resource in allowing for functional interpretation of protein-altering ASS1 variants that could be newly identified from large-scale whole-genome sequencing efforts done in biobanks or national precision medicine programs.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

1. General Statements [optional]

We thank all three Reviewers for appreciating our work and for sharing constructive feedback to further enhance the quality of our study. It is really gratifying to read that the Reviewers believe that this work is interesting, novel and of interest to broad audience. Therefore, we believe that it will be suitable for a high profile journal. Further, the experiments suggested by the reviewers have added value to the work and have substantiated our findings. It is important to highlight that we have performed all the suggested experiments. Please find below the detailed point by point response to Reviewer’s Comments.

2. Point-by-point description of the revisions

Reviewer #1 (Evidence, reproducibility and clarity (Required):

• The manuscript entitled, "IP3R2 mediated inter-organelle Ca2+ signaling orchestrates melanophagy" is a rather diffuse study of the relationship between IP3R2 and melanin production. While this is an interesting and understudied area, the study lacks a clear focus. The model seems to be that IP3R2 is essential for mitochondrial calcium loading. And that its absence increases lysosomal calcium loading. There are also a number of incomplete and/or unconvincing links to autophagy/melanophagy, TMEM165, TRPML1 and even gene transcription. In this kind of diffuse study, each step needs to be convincing to get to the next one, which is not the case here. There are also references to altered proteasome function, despite the total absence of any direct data on the proteasome. Finally, I felt it was sometimes unclear whether the authors were referring to melanosomes or lysosomes at various points throughout the study.*

While I suspect that, somewhere in here, there are some novel relationships worthy of further investigation, this is a case where the many parts make the overall product less convincing. What effects here are directly relevant to IP3R2? This study should stop there, leaving investigations of peripheral factors for future investigations, as the further you get from where you start, the less clear what you are studying becomes. And the less direct.

Response: We thank the Reviewer for finding our study interesting and recognizing that this is an understudied area. Further, we appreciate the constructive feedback given by the Reviewer. We have addressed all the Reviewer’s comments. Please find below point-wise responses to the comments.

Specific Comments:

__ Comment 1.__ The separation of Figures 1F and 1J makes it impossible to assess the effect of αMSH on IP3R2 expression. This presentation makes interpretation difficult; a simple 4 lane Western would be more informative.

Response: We apologize to the Reviewer for not being very clear. Actually, we have separated these data sets because these are two independent experimental conditions. The Figure 1F illustrates data from the LD-based pigmentation model, whereas Supplementary Figure 1K (Previously Fig 1J) depicts data from α-MSH–induced pigmentation model.

Comment 2. One of the most attractive points made by this study is that there is a specific link between IP3R2 and melanin production. In my opinion, the null hypothesis is that this is just about the amount of IP3Rs expressed per cell. To reject this concept, the authors should show data demonstrating the relative expression of all 3 IP3Rs. Without this information, the null hypothesis that IP3R2 is the most expressed IP3R isoform and that's why its knockdown has the most dramatic effect cannot be rejected It would also be helpful to show where the different IP3Rs are expressed within the cell.

Response: We thank the Reviewer for raising this interesting point and for the constructive comment. As suggested, we would like to clarify that the relative expression of all three IP₃R isoforms has already been analyzed in our study. Specifically, in Figure 1B, we demonstrate the expression pattern of IP₃R isoforms in our experimental system, where IP₃R2 shows the highest expression level, followed by IP₃R3 and IP₃R1 (IP₃R2 > IP₃R3 > IP₃R1). Further, in the revised manuscript, we additionally analyzed publicly available datasets for IP₃Rs expression. “The Human Protein Atlas” reports a higher expression of IP₃R2 in melanocytes compared to the other IP₃R isoforms (Supplementary Fig 1A). Therefore, we agree with the Reviewer’s proposed concept that the relatively higher expression of IP₃R2 can be one of the important factors that regulate pigmentation levels. Indeed, our analysis of microarray dataset from African vs Caucasian skin revealed a greater IP₃R2 expression in African skin compared to Caucasian skin (__Figure 1L). __

With respect to subcellular localization, all three IP₃R isoforms are predominantly localized to the endoplasmic reticulum, consistent with their established role as ER-resident Ca²⁺ release channels. However, their expression levels are known to be highly cell and tissue specific (Bartok et al., Nature Communications 2019), supporting the idea that higher IP₃R2 levels play a functionally specialized role in melanogenesis.

Comment 3. It would be helpful to label Figs 3F-I with the conditions used. The description in the text is of increased LC3II levels, however, the ratio of LC3I to LC3II might be more meaningful. Irrespective, although the graph shows an increase in LC3II, the Western really doesn't show much. As a standalone finding, I don't find this figure to be very convincing; there are better options to demonstrate this proposed relationship between IP3R2 and autophagy than what is shown.

Response: We sincerely thank the Reviewer for this thoughtful and critical evaluation, which has helped us improve the clarity and precision of this analysis. To address this concern, in the revised manuscript, we have now labeled ‘LD’ in the Supplementary Fig 2A-B (Previously, Fig 4F-I) with the corresponding experimental conditions for clarity. In addition, we reanalyzed the data by calculating the LC3II/LC3I ratio in all the figures of the revised manuscript that include LC3II expression, which provides a more meaningful and robust assessment of autophagic flux. This revised analysis yields a clearer representation of LC3 dynamics and strengthens the interpretation of the western blotting data in support of the relationship between IP₃R2 and autophagy. Further, we have shown by confocal imaging that IP3R2 silencing significantly reduced GFP/RFP ratio of the pMRX-IP-GFP-LC3-RFP reporter system in comparison to control condition in Fig 4M-N to demonstrate the relationship between IP3R2 and autophagy. Collectively, these autophagy flux assays and biochemical experiments clearly demonstrate a direct relationship between IP3R2 and autophagy.

Comment 4. The following statement at the beginning of page 22 "We observed an impaired proteasomal degradation of critical melanogenic proteins localized on melanosomes in the IP3R2 knockdown condition" is insufficiently supported by data to be made. Even if I was convinced that autophagy was enhanced, there is no data of any kind about the proteasome in this manuscript.

Response: We appreciate the Reviewer’s careful scrutiny of this statement and the opportunity to clarify and strengthen our interpretation. To directly address the concern regarding proteasomal involvement, in the revised manuscript, we performed additional experiments using MG132, a well-established inhibitor of proteasomal degradation. These experiments were designed to assess whether the altered stability of melanogenic proteins observed upon IP₃R2 knockdown could be attributed to changes in proteasome-mediated turnover.

In the revised manuscript, our new data show that treatment with MG132 leads to a marked reduction in the levels of melanosome-associated melanogenic proteins, including GP100 and DCT, compared to the DMSO control (Fig. 4A–D). This response contrasts with that of non-melanosomal proteins, such as IP₃R2 and Calnexin, which are localized to the endoplasmic reticulum and exhibits increased accumulation upon MG132 treatment (Fig. 4E–H), consistent with canonical proteasomal inhibition. These differential outcomes suggest that melanosome-resident proteins respond distinctly to proteasomal blockade, likely due to their compartmentalized localization on melanosomes.

Previous studies have shown that impairment of proteasomal function can activate autophagy as a compensatory, cytoprotective mechanism (Williams et al, 2013; Li et al, 2019; Su & Wang, 2020; Pan et al, 2020). Indeed, we observed a significant increase in LC3II/LC3I levels in IP3R2 knockdown plus MG132 treatment condition in comparison to IP3R2 knockdown plus the DMSO control (Fig. 4I–J).

To investigate whether impairment of proteasomal degradation upon IP3R2 silencing alone or together with MG132 selectively triggers melanophagy, we assessed melanophagy using melanophagy reporter, mCherry-Tyrosinase-eGFP following IP3R2 silencing along with MG132 treatment. Our observations revealed an increase in melanophagy flux with IP3R2 silencing and MG132 treatment compared to siNT with DMSO control (Fig 5K-L). This suggests that IP3R2 silencing induced inhibition of proteasomal degradation activates melanophagy. Taken together, these findings indicate that compromised proteasomal degradation engages the autophagy machinery, providing a mechanistic link between proteasome dysfunction, enhanced autophagy, and altered melanogenic protein turnover.

Comment 5. In figure 5, the authors create a new ratiometric dye to detect melanosome stability based on the principle that tyrosinase is exclusively found in melanosomes. Unfortunately, there is no validation that this new construct is found exclusively in melanosomes upon expression. In addition, there is discussion about the pH of lysosomes, but not of melanosomes. Ultimately, this data cannot be considered at face value without any type of validation; I also note that the pictures lack sufficient detail to support identification of these structures as melanosomes. * While I maintain the above concerns, I note that, the data in supplemental figure 3 is MUCH more convincing than what is in the figure. Both the writing and the figure design should be rethought.*

Response: We appreciate the Reviewer’s thorough evaluation and constructive critique of Figure 5, which has helped us to better clarify and validate this aspect of the study. In the revised manuscript, we directly address the concern regarding the subcellular specificity of the ratiometric probes, we performed detailed colocalization analysis using established melanosome markers. Specifically, we assessed the localization of the melanophagy detection probes mCherry–Tyr–eGFP and tyrosinase–mKeimaN1 with the melanosome-resident protein GP100 detected by anti-HMB45 (Supplementary Fig 2E-F and 2K-L). These analyses revealed a very high degree of colocalization, reflected by strong Pearson’s correlation and overlap coefficients, thereby validating that the expressed probes are predominantly localized to melanosomes.

Regarding Lysosome/Melanosomal pH considerations, our melanophagy detection ratiometric probes: mCherry–Tyrosinase–eGFP (sensitive to acidic pH via eGFP) and tyrosinase mKeimaN1 (sensitive to acidic pH via Keima) are specifically designed to identify melanosome degradation, which happens upon melanosome fusion with lysosome. Consequently, the observed signal shifts indicate melanosome turnover rather than merely reflecting the lysosomal pH.

To further corroborate the microscopic observations, we performed biochemical assays to study melanophagy flux upon IP3R2 silencing. We employed Bafilomycin A1, an inhibitor of autophagosome-lysosome fusion, to examine melanosomal protein accumulation. Upon Bafilomycin A1 treatment, IP3R2 silenced cells showed enhanced accumulation of melanosomes, as indicated by elevated tyrosinase levels compared with siNT controls (Supplementary Fig 3C-D), indicating elevated melanophagy flux upon IP3R2 knockdown. In the revised manuscript, we employed additional melanophagy detection strategies to further strengthen our findings. Specifically, we used Retagliptin phosphate (RTG), a well-established selective inducer of melanophagy, and observed a marked increase in melanophagy using the mCherry–Tyrosinase–eGFP melanophagy probe (Supplementary Fig 2G-H). Additionally, we performed independent validation by assessing colocalization of the melanosome (recognized by anti-HMB45 ab that identifies melanosomal structural protein GP100) with LC3 (Supplementary Fig 3A-B). This analysis revealed a significant increase in melanosomes colocalization with LC3 upon IP₃R2 silencing compared to control conditions.

Collectively, these independent approaches clearly demonstrate that the melanophagy probes localize to melanosomes and detect melanophagy (by responding to melanosome fusion to lysosomes).

Comment 6. Given the increase in ER Ca2+ content after IP3R2 knockdown, ER calcium content should be emptied before attempting to estimate lysosomal Ca2+ content with GPN or Bafilomycin. Otherwise, the source of calcium is less than clear.

Response____: We appreciate the Reviewer’s careful consideration of Ca²⁺ source, which is critical for accurate interpretation of these experiments. Therefore, as suggested, in the revised manuscript, we conducted experiments involving Thapsigargin (Tg) pre-treatment to deplete ER Ca²⁺ reserves before examining lysosomal Ca²⁺ release using GPN or Bafilomycin (Supplementary Fig 6I-N). Even under these conditions, we noted increased lysosomal Ca²⁺ release in IP₃R2 knockdown cells, thus confirming that the observed Ca²⁺ signals originate from lysosomes rather than any remaining ER Ca²⁺. Importantly, this approach allowed us to minimize ER-derived Ca²⁺ contributions to changes in the lysosomal Ca²⁺ release.

Reviewer #1 (Significance (Required)):

The manuscript entitled, "IP3R2 mediated inter-organelle Ca2+ signaling orchestrates melanophagy" is a rather diffuse study of the relationship between IP3R2 and melanin production. While this is an interesting and understudied area, the study lacks a clear focus. The model seems to be that IP3R2 is essential for mitochondrial calcium loading. And that its absence increases lysosomal calcium loading. There are also a number of incomplete and/or unconvincing links to autophagy/melanophagy, TMEM165, TRPML1 and even gene transcription. In this kind of diffuse study, each step needs to be convincing to get to the next one, which is not the case here. There are also references to altered proteasome function, despite the total absence of any direct data on the proteasome. Finally, I felt it was sometimes unclear whether the authors were referring to melanosomes or lysosomes at various points throughout the study.

Response____: We thank the Reviewer for finding our work interesting and appreciating that this is an understudied field. Further, we thank him/her for the constructive feedback on our study. We have performed several additional experiments and significantly revised the manuscript to address all the comments of the Reviewer.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

In the present manuscript, Saurav et al. identify IP3R2-mediated ER calcium release as a key suppressor of melanophagy, thereby sustaining pigmentation in melanocytes. Using in vitro (B16 murine melanoma cells, primary human melanocytes) and in vivo (zebrafish) models, the authors report that IP3R2 expression is positively correlated with pigmentation. They then investigate the impact of IP3R2 knockdown and find that IP3R2 silencing enhances the stability of melanogenic proteins, while also inducing autophagic degradation of melanosomes (i.e., melanophagy). Concomitantly, they find that IP3R2 silencing decreases mitochondrial calcium uptake, increases lysosomal calcium loading, and lowers lysosomal pH. They propose a pathway wherein in IP3R2 knockdown cells impaired mitochondrial calcium uptake induces the activation of AMPK-ULK1, and increased lysosomal calcium activates TRPML1 via TMEM165 and closer proximity interactions between ER and lysosomes, TFEB nuclear translocation, and upregulation of melanophagy-related genes, namely OPTN and RCHY1. The work is placed within the context of emerging roles of organelle calcium signaling in pigmentation biology, where extracellular calcium influx pathways are known regulators, but the contribution of ER-mitochondria-lysosome crosstalk to melanosome turnover remains largely unknown.

Response____: We thank the Reviewer for appreciating our work and highlighting that the contribution of ER-mitochondria-lysosome crosstalk to melanosome turnover remains largely unappreciated.

Major comments:

Comment 1- The central finding is that IP3R2 knockdown induces melanophagy and reduces pigmentation. However, the manuscript does not identify any physiological or pathological context in which IP3R2 expression or activity is naturally downregulated in melanocytes. Without such context, the knockdown may represent an artificial perturbation that broadly alters ER calcium handling and triggers melanophagy as part of a general stress-induced autophagy response. This raises uncertainty about whether the pathway operates in vivo under normal or disease conditions. It would strengthen the study to identify upstream cues that reduce IP3R2 function and to test whether these also trigger melanophagy through the proposed mechanism.

Response____: We thank the Reviewer for asking such an important question. The Reviewer asked to identify any physiological or pathological context in which IP3R2 expression is naturally downregulated in melanocytes. To address this question, in the revised manuscript, we analyzed publicly available microarray datasets comparing skin samples from Caucasian and African populations (Yin et al., Experimental Dermatology 2014). This unbiased analysis revealed considerably lower IP₃R2 expression in the Caucasian skin as compared to African skin (Fig. 1L). This data support a physiological correlation between IP₃R2 expression and pigmentation level, reinforcing the physiological relevance of the proposed pathway.

Comment 2- While the data link IP3R2 knockdown to decreased pigmentation and increased melanophagy, the causality between altered organelle calcium dynamics and the melanophagy induction is inferred from correlation and partial rescue experiments. More direct interventions in the proposed downstream pathways (e.g., acute mitochondrial calcium uptake restoration, lysosomal calcium buffering) would strengthen mechanistic claims.

Response____: We appreciate the Reviewer’s recommendation on strengthening the mechanistic causality between organelle Ca²⁺ dynamics and melanophagy. As suggested, in the revised manuscript, we restored acute mitochondrial Ca²⁺ uptake by MCU over-expression in the IP₃R2 knockdown background, which resulted in a marked reduction in melanophagy along with increased mitochondrial Ca²⁺ uptake in comparison to control (Fig 6I-L). This data clearly demonstrates that downstream of IP₃R2 silencing mitochondrial Ca²⁺ restoration rescues the melanophagy phenotype thereby revealing a mechanistic causality between mitochondrial Ca²⁺ dynamics and melanophagy.

Similarly, to assess the causality between lysosomal Ca²⁺ dynamics and melanophagy, we silenced TMEM165 in the IP₃R2 knockdown background. Excitingly, upon TMEM165 knockdown we observed reduction in melanophagy, concomitant with decrease in lysosomal Ca²⁺ levels under IP₃R2 silencing conditions (Supplementary Fig 7I-L). Together, these direct manipulations support a causal role for altered organelle Ca²⁺ dynamics in driving melanophagy.

We believe that these experiments would have addressed the concern of the Reviewer. However, if there are any other specific experiments that the Reviewer would like us to perform, we would be happy to carry out them as well.

__Comment 3____- __Zebrafish assays convincingly show altered pigmentation with altered IP3R2 levels, but do not connect this to in vivo melanophagy measurements or TRPML1/TFEB activity, which would link the cell biology to organismal phenotype more directly.

Response____: We thank the Reviewer for appreciating our in vivo zenrafish experiments. Futher, we acknowledge the Reviewer’s point of linking the cellular mechanisms to organismal phenotypes in vivo. Therefore, as suggested, we activated TRPML1 in the zebrafish model system. In the revised manuscript, we investigated role of the TRPML1–TFEB axis in pigmentation in vivo by pharmacological activation of TRPML channels with MLSA1. The MLSA1 treatment resulted in a marked reduction in zebrafish pigmentation compared to vehicle-treated controls (Fig. 8M). This phenotypic change was further substantiated by quantitative melanin content assays, which confirmed a significant decrease in melanin levels following MLSA1 treatment (Fig. 8M–N). These in vivo findings support the involvement of TRPML1-mediated lysosomal signaling in pigmentation regulation.

Comment 4- The work suggests therapeutic potential for pigmentary disorders, but no disease models are tested. It is unclear whether the observed mechanisms operate under physiological stressors.

Response____: We appreciate the Reviewer’s comment regarding physiological relevance and disease context. As addressed in Comment 1, we examined publicly available human skin microarray datasets for IP₃R2 expression in Caucasian and African population. This analysis revealed a positive correlation between IP₃R2 expression and human skin pigmentation, supporting that modulation of IP₃R2 occurs under physiological conditions rather than representing an artificial perturbation.

While formal pigmentary disease models were not examined in this study, the observed correlation between IP₃R2 expression and physiological pigmentation differences along with our robust in vivo zebrafish data suggests that IP₃R2 plays an important role in physiological pigmentation. As highlighted by Reviewer 1 and Reviewer 3, the manuscript is already too long. Therefore, we plan to delineate the precise role of IP₃R2 in pigmentary disorders as an independent study.

Comment 5- The paradox between the observed enhanced stability of melanogenic proteins and increased melanophagy is insufficiently addressed. DCT, Tyrosinase and GP100 are all melanosome-associated and their stability or degradation is in prior literature often interpreted as reflecting melanosome biogenesis and turnover. This discrepancy needs to be resolved, as it complicates interpretation of melanophagy assays.

Response____: We appreciate the Reviewer’s careful consideration of this apparent paradox. This point was also raised by Reviewer 1. We have addressed the query in detail in response to Comment 4 of Reviewer 1. Briefly, the enhanced stability of melanosome-associated proteins reflects impaired proteasomal degradation and prolonged protein half-life, while the concurrent increase in melanophagy represents a compensatory turnover mechanism for degrading such dysfunctional melanosomes.

Thus, increased melanophagy and apparent stabilization of melanogenic proteins are not contradictory but instead represent parallel outcomes of disrupted proteostasis. This interpretation is supported by our proteasomal inhibition experiments (Fig 4A-H) and autophagy analyses (Fig 4I-P), which collectively reconcile the observed protein stability with enhanced melanosome turnover.

Comment 6- The authors propose that mitophagy and ER-phagy are reduced in IP3R2 knockdown cells, suggesting specific induction of melanophagy, but the rationale for why increased autophagic flux only targets melanosomes is insufficiently addressed. Also, these conclusions are solely based on Keima assays, and positive controls for mitophagy and ER-phagy are lacking.

Response: We appreciate the Reviewer’s critical assessment of the specificity of autophagic targeting in the IP₃R2 knockdown condition and the need for appropriate validation controls. In the revised manuscript, we have repeated both the mitophagy and ER-phagy assays with well-established positive controls. Carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) was employed as a positive control to robustly induce mitophagy (Supplementary Fig 4E-F), while 4-phenylbutyric acid (4PBA) was used as a positive control for ER-phagy/reticulophagy (Supplementary Fig 4G-H). Secondly, we have validated the microscopy data with biochemical assays by examining levels of ER (Fig 4E-H) and mitochondria resident protein MCU.

To provide a mechanistic rationale for the specific induction of melanophagy, we examined recently identified regulators of melanophagy, RCHY1 and OPTN (Lee et al., PNAS 2024). Bioinformatic analysis identified multiple TFEB binding sites on the promoters of both genes, which was supported by increased RCHY1 and OPTN expression following IP₃R2 knockdown. Further, in the revised manuscript, we performed additional loss-of-function experiments to demonstrate that co-silencing IP3R2 along with RCHY1 or OPTN significantly reduced melanophagy flux compared to IP₃R2 knockdown alone (Fig. 9H–K). Taken together, these data explain why enhanced autophagic flux downstream of IP₃R2 silencing is preferentially directed toward melanosomes.

Comment 7- The melanophagy probes are novel and validated with rapamycin/bafilomycin, but quantitative calibration of GFP/mCherry or Keima signal to actual lysosomal delivery rates is missing; photobleaching, pH heterogeneity (incl., observed decrease in lysosomal pH), and melanin autofluorescence (see below) could confound ratios. Also, side-by-side comparison with other melanophagy detection approaches (e.g., colocalization of melanosomes with LC3) is lacking.

__Response____: __We appreciate the Reviewer’s careful evaluation of the melanophagy probes and the potential technical confounders. In the revised manuscript, we have performed a variety of experiments to further characterize and validate the probes. First of all, the melanophagy detection ratiometric probes (mCherry–Tyrosinase–eGFP and tyrosinase mKeimaN1) are built on well-established and extensively validated backbones. Further, we used appropriate controls (empty vectors/non-targeting siRNAs/vehicle controls) in all experiments to analyze the relative fluorescence changes in the test condition v/s control. The confounding factors, if any, should be present for both test and control. Therefore, we initially did not perform side-by-side comparison with other melanophagy detection approaches.

In the revised manuscript, as suggested by the reviewer, we employed additional melanophagy detection strategies to further strengthen our findings. Specifically, we used Retagliptin phosphate (RTG), a well-established selective inducer of melanophagy, and observed a marked increase in melanophagy using the mCherry–Tyrosinase–eGFP melanophagy probe (Supplementary Fig 2G-H). Additionally, we performed independent validation by assessing colocalization of the melanosome (recognized by anti-HMB45 ab that identifies melanosomal structural protein GP100) with LC3 (Supplementary Fig 3A-B). This analysis revealed a significant increase in melanosomes colocalization with LC3 upon IP₃R2 silencing compared to control conditions. Further, to minimize the contribution of melanin autofluorescence, non-transfected cells were imaged under identical settings, and background signals obtained from these cells were subtracted during fluorescence quantitation from all acquired images. Potential effects of photobleaching and pH heterogeneity were minimized by uniform acquisition parameters and ratiometric analysis. Taken together, we believe these complementary approaches address the Reviewer’s concerns and reinforce the robustness of our melanophagy measurements.

Comment 8- Melanosomes exhibit broad autofluorescence, particularly upon excitation at 405-488 nm and extending into the red channel. This signal can overlap with the detection ranges for GFP, mCherry, and mKeima reporters, potentially confounding quantitative readouts unless appropriate controls (e.g., untransfected cells, spectral unmixing) are used. Throughout this manuscript, it is not addressed how melanosome autofluorescence was controlled for or excluded in the reported fluorescence measurements.

__Response____: __We apologize to the Reviewer for not clearly stating that melanosome autofluorescence was controlled by imaging non-transfected cells under identical settings, and these background signals were subtracted during quantitation from the acquired images. Specifically, to rigorously control this issue, autofluorescence was systematically evaluated using non-transfected control cells imaged under identical excitation and emission settings used for GFP, mCherry, and mKeima reporters. These controls allowed us to define the baseline autofluorescence profile arising from melanosomes across the relevant spectral ranges. These details are included in the methods section.

Comment 9- While OPTN and RCHY1 expression is elevated upon IP3R2 knockdown, functional engagement (e.g., OPTN localization to melanosomes, melanosome ubiquitination by RCHY1), or necessity (e.g., siRNA knockdown of these in the IP3R2-deficient background), are not tested.

Response: We appreciate the Reviewer’s point on establishing necessity of OPTN and RCHY1 in IP₃R2 knockdown–induced melanophagy. In the revised manuscript, we performed targeted loss of function analyses for both OPTN and RCHY1 in the IP₃R2-deficient background. We assessed melanophagy using the mCherry–Tyrosinase–eGFP melanophagy probe following co-silencing of IP₃R2 with either OPTN or RCHY1. Quantitative analysis revealed a significant reduction in melanophagy flux upon co-silencing of either gene compared to IP₃R2 silencing alone (Fig. 9H–K). These findings establish the functional requirement of OPTN and RCHY1 downstream of IP₃R2 loss to drive melanophagy. Since functional engagement of OPTN and RCHY1 on melanosomes is already well-established (Lee et al. PNAS 2024 and Park et al. Autophagy 2024), we have not repeated these experiments. Taken together, our data demonstrates that OPTN and RCHY1 are not only overexpressed but also act as critical mediators of melanophagy downstream of IP₃R2 silencing.

__Comment 10- __While siRNA/shRNA efficacy is shown, functional rescue with pore-dead mutants sometimes fails to return to control values. The possibility of partial off-target or compensatory effects is not fully excluded.

Response: We thank the Reviewer for raising for this point. In this study, we employed pore-dead mutants of IP₃R2 (IP₃R2-M) and TRPML1 (TRPML1-M), both of them are well characterized, widely validated and extensively used by a number of leading groups in the field. Upon meticulous literature analysis, we came across multiple studies wherein partial rescue effect was reported with these pore-dead mutants. Therefore, we believe it is not surprising that we are also observing partial rescue in some of our assays.

Actually, it is important to note that we observe rescue of the function and phenotype in every single experiment carried out with the mutants. We agree with the Reviewer that the extent of rescue is not up to control levels in few experiments. This can be attributed to the differences in the extend of expression of mutants across different experiments. However, we have validated the results with multiple independent approaches. Collectively, the use of multiple independent approaches along with genetic silencing, pharmacological inhibition/activation supports the specificity of the observed phenotypes.

Comment 11- The mitochondrial and lysosomal calcium measurements are largely endpoint peak quantifications; kinetic analyses and buffering capacity measurements would provide more mechanistic depth, especially for the TMEM165 contribution. Also, TMEM165 necessity for melanophagy induction upon IP3R2 knockdown has not been directly addressed.

Response: We appreciate the Reviewer’s request for greater mechanistic depth regarding organelle Ca²⁺ dynamics and the specific contribution of TMEM165. Consistent with this, we had previously demonstrated that TMEM165 silencing decreases lysosomal Ca²⁺ levels using Oregon BAPTA–dextran–based measurements (Supplementary Fig 7C-D), establishing its role in regulating lysosomal Ca²⁺ buffering. Building on this, in the revised manuscript, we performed kinetic analyses of lysosomal Ca²⁺ levels following IP₃R2 and TMEM165 silencing. These kinetic analyses validated our end point measurements that IP₃R2 knockdown leads to increase in lysosomal Ca²⁺ levels, whereas TMEM165 silencing results in decrease in lysosomal Ca²⁺ content in comparison to control. Therefore, highlighting distinct and opposing effects of IP₃R2 and TMEM165 on lysosomal Ca²⁺ kinetics.

Further, we directly evaluated the necessity of TMEM165 for melanophagy induction in the IP₃R2-deficient background. TMEM165 knockdown alone resulted in a significant reduction in melanophagy (Supplementary Fig 7G-H). Further, co-silencing of TMEM165 with IP₃R2 also attenuated melanophagy compared to IP₃R2 knockdown alone (Supplementary Fig 7K-L). Collectively, these kinetic Ca²⁺ assays and genetic loss-of-function analyses provide mechanistic depth to the organelle Ca²⁺ measurements and establish TMEM165 as a critical regulator of melanophagy downstream of IP₃R2 silencing.

Comment 12- The proximity ligation assay between VAP-A and LAMP1 is interpreted as showing increased ER-lysosome contacts in IP3R2 knockdown cells. However, additional controls are needed and quantitative TEM should be included to substantiate changes in organelle contact frequency and distance.

Response: We thank the Reviewer’s for his/her emphasis on strengthening the validation of the proximity ligation assay (PLA) findings and on providing ultrastructural evidence to support altered organelle interactions. The PLA data revealed a significant increase in VAP-A–LAMP1 interaction signals in IP₃R2-silenced cells compared to control conditions (Fig. 7L–M). In the revised manuscript, this increase was not observed upon treatment with bafilomycin A1, a specific inhibitor of lysosomal acidification, or when one of the primary antibodies was omitted, confirming the specificity of the PLA signal (Fig. 7L–M). These controls support the interpretation that IP₃R2 downregulation enhances ER–lysosome interactions.

To further substantiate the changes in organelle contact frequency and distance, we performed ultrastructural analyses using transmission electron microscopy (TEM). The quantitative TEM measurements revealed no significant change in the frequency of ER–mitochondria or ER–lysosome contacts upon IP₃R2 silencing (Fig. 7N–P). Similarly, ER–mitochondria distances remained unchanged. However, we observed a significant reduction in the distance between the ER and lysosomes in IP₃R2 knockdown cells compared to control (Fig. 7N, 7Q–R). Together, these complementary approaches demonstrate that IP₃R2 silencing specifically increases ER–lysosome proximity without altering overall contact frequency, thereby strengthening the conclusion that IP₃R2 regulates ER–lysosome coupling.

Comment 13- Some assays report small biological n (e.g., three independent experiments with relatively small per-condition cell counts).

__Response:____ __We appreciate the Reviewer’s comment regarding sample size. All experiments were performed with a minimum of three independent biological replicates, which is consistent with standard practice in the field. For imaging-based assays, multiple fields of view and cells were analyzed per condition in each independent experiment, and quantitative analyses were performed on pooled data across replicates. As suggested by the Reviewer, we have increased the cell numbers in some experiments. The detailed information on biological replicates and cell numbers analyzed is provided in the respective figure legends.

Minor comments:

• Comment 1- The title "IP3R2-mediated inter-organelle Ca2+ signaling orchestrates melanophagy" could be misread as indicating IP3R2 'promotes' melanophagy; consider rewording to make clear that IP3R2 suppresses melanophagy to maintain pigmentation. Similarly, the running title "IP3R2 negatively regulates melanophagy" would be clearer as "IP3R2 suppresses melanophagy".*

__Response____: __As suggested by the Reviewer, we have modified the title and running title in the revised manuscript.

Comment 2- Unify the framing of "positively regulates pigmentation" vs. "negatively regulates melanophagy" in the Introduction/Discussion.

Response: As recommended, we have unified the framing in the suggested sections.

Comment 3- Adding schematic flow diagrams summarizing each pathway at the end of relevant results (figure) sections could help accessibility.

Response____: __We appreciate the Reviewer’s suggestion to improve accessibility of the presented pathways. Accordingly, we have included schematic diagrams at the end of the relevant figures. These schematics summarize: (i) ER–mitochondria interactions in the context of melanophagy (__Fig. 6P); (ii) differences in Ca²⁺ and pH regulation between wild-type and IP₃R2-silenced cells (Fig. 7S); and (iii) TRPML1-mediated Ca²⁺ release driving melanophagy via TFEB translocation (Fig. 9L). Together, these diagrams provide a concise visual overview of the key mechanistic pathways described in the study.

Comment 4- While the introduction summarizes extracellular calcium signaling in pigmentation, there is less coverage of recent work on selective autophagy of other lysosome-related organelles (e.g., platelet dense granules, lytic granules), which could provide broader mechanistic context.

__Response____: __As suggested by the Reviewer, we have discussed selective autophagy of other lysosome-related organelles in the introduction.

Reviewer #2 (Significance (Required)):

This study addresses an important gap in pigmentation biology by identifying IP3R2-mediated ER calcium release as a suppressor of melanophagy and a positive regulator of pigmentation. The strongest aspects are the integration of in vitro and in vivo models, the multi-faceted mechanistic exploration linking altered organelle calcium dynamics to selective melanosome turnover, and the development of novel ratiometric fluorescent probes for live-cell melanophagy measurement. Conceptually, the work extends prior literature that has focused on extracellular calcium influx and melanosome biogenesis, revealing a new inter-organelle calcium signaling module that controls melanosome degradation via AMPK-ULK1 and TMEM165-TRPML1-TFEB pathways.

• However, several limitations reduce the strength of the mechanistic claims. Some key pathway steps are inferred from correlation and partial rescue rather than direct necessity/sufficiency tests (e.g., mitochondrial calcium uptake restoration, lysosomal calcium buffering). The paradoxical observation that IP3R2 knockdown both increases melanophagy and stabilizes melanosome-resident protein (DCT, Tyrosinase, GP100) is not resolved, complicating interpretation of the melanophagy assays. The specificity for melanophagy over other selective autophagy pathways is asserted but not fully explained mechanistically, and positive controls for mitophagy/ER-phagy are missing. Potential technical confounds, such as melanin autofluorescence in the detection ranges of GFP, mCherry, and mKeima, are not explicitly addressed and alternative assays for these key data were insufficiently employed. In vivo results do not yet connect altered pigmentation to melanophagy readouts or downstream TRPML1/TFEB activation. Importantly, the study does not identify any physiological or pathological scenario in which IP3R2 expression or activity is naturally reduced in melanocytes. In the absence of such upstream cues, IP3R2 knockdown may represent an artificial perturbation that triggers melanophagy as part of a broader stress-induced autophagy response, raising questions about the in vivo relevance of the proposed pathway.*

• The work's primary audience is specialized, cell biologists, autophagy researchers, and pigmentation/skin biology specialists, but the mechanistic framework on organelle crosstalk and selective autophagy will interest a broader basic research readership, including those studying lysosome-related organelles in other systems. The ratiometric probes could be adapted for future melanophagy research, and the pathway insights may guide translational studies in pigmentary disorders or melanoma. My expertise is in mitochondrial and lysosomal calcium signaling, autophagy, and microscopy-based functional assays; I do not have detailed expertise in zebrafish developmental genetics, though the phenotypic analysis appears sound.*

Response____: We thank the Reviewer for appreciating our work and stating that our study “addresses an important gap in pigmentation biology”. Further, we thank him/her for believing that this work will be of interest to a broad basic research readership. Moreover, we thank him/her for valuing the importance and potential significance of the ratio-metric melanophagy probes generated in this study. Finally, we acknowledge the Reviewer’s constructive feedback on our study, which has helped us in enhancing the quality of our manuscript. We have performed variety of additional in vitro experiments, in vivo zebrafish studies and have significantly revised the manuscript to address all the comments of the Reviewer.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

This is a robust and extensive study showing that IP3R2 selectively initiates a calcium signalling pathway leading to melanophagy, that is the degradation of melanosomes. This reduces pigmentation and UV light protection. A strength of the paper is that it combines detailed cellular studies with in viva studies in the zebrafish model. They show that knockdown of IP3R2 reverses this process perhaps leading to a strategy to enhance melanosome number and hence to afford protection from UV irradiation. The authors use a battery of fluorescent probes (mainly genetically encoded reporters) in investigate the signalling cascade leading to melanophagy or its reduction. This involves reports for a number of different organelles involved in this process. The experiments are generally well performed with clear controls for the probes in many cases. My main issue is the panels contain too much data which may obscure the message, and a good deal could be moved to supplementary data. The manuscript investigates many mechanisms in distinct organelles which is remarkable for a two author paper. Particularly interesting was the design of novel fluorescent protein reporters for melanophagy itself. One area not explored is ion fluxes across melanosomes themselves which are lysosome-related organelles and may exhibit similar properties and signalsomes of lysosomes.

Specifically, the authors show that a REDUCTION of IP3R2-mediated calcium release leads to a calcium flux from the ER by a different mechanism (possibly via TMBIM6). This increases calcium loading of the lysosome via TMEM165, at the expense of calcium transfer to mitochondria, and an acidification.

• This leads to TRPML1 activation and the lysosomal calcium release activates TFEB translocation to the nucleus increases the transcription of autophagy/melanophagy genes and activation of the AMPK-ULK1 pathway (rather than mTOR). This is a complex pathway and evidence is presented for many of the steps involved.*

• This is a tour de force investigating organelle communication during the process of melanophagy, that is little understood. It highlights many important organelle ion transport events that are important findings in their own right. For example, the importance of TMEM165 in calcium filling of lysosomes.*

Response____: We thank the Reviewer for appreciating our study and thinking that it is a robust and extensive study in a highly understudied area. We appreciate the Reviewer’s acknowledgement that our manuscript combines detailed cellular studies with in vivo studies in the zebrafish model. Further, we thank the Reviewer for his/her constructive feedback on our work.

__ Major points:__

Comment 1- The authors state that TPC activation does not activate TFEB translocation the nucleus. This is now not the case and should be at least looked at. What is the role of endolysosomal channels on the melanosomes themselves in melanophagy.

Response____: We appreciate the Reviewer’s comment regarding the potential contribution of TPC channels to TFEB activation and melanophagy. In the revised manuscript, we assessed Ca²⁺ release from TPC2 under IP₃R2 knockdown conditions using the selective TPC2 agonist TPC2-A1-N (Supplementary Fig 9G-H). Additionally, we evaluated TFEB nuclear translocation following TPC2-mediated Ca²⁺ release using TPC2-A1-N (Supplementary Fig 9I-J). Our analyses revealed no significant differences in TPC2 activity or TFEB nuclear translocation upon IP₃R2 silencing compared to control conditions. These findings suggest that, in our system, TPC2-mediated Ca²⁺ signaling does not contribute significantly to TFEB activation or melanophagy downstream of IP₃R2 silencing, indicating a more prominent role for TRPML1-dependent Ca²⁺ signaling in this context.

Comment 2- How does reduction in IP3R2 mediated calcium fluxes enhance lysosomal acidity?

Response____: We thank the Reviewer’s question regarding the mechanistic link between reduced IP₃R2-mediated Ca²⁺ flux and enhanced lysosomal acidity. In the revised manuscript, we show that IP₃R2 silencing results in a significant upregulation of the lysosomal proton pump H⁺-ATPase subunits: ATPV0D1 and ATP6V1H (Supplementary Fig 6E-F). Increased H⁺-ATPase expression is expected to promote proton influx into the lysosomal lumen, thereby enhancing lysosomal acidification. These findings provide a mechanistic basis for how IP₃R2 silencing can drive increased lysosomal acidity.

Comment 3- What mediates the ER source for calcium filling of lysosomes?

Response____: We appreciate the Reviewer’s interest in the mechanism underlying ER to lysosome Ca²⁺ transfer. Recently, an independent study also reported that IP₃R2 silencing enhances lysosomal Ca²⁺ levels and lysosomal Ca²⁺ release (Zheng et al. Cell 2022). Literature suggests that lysosomal Ca²⁺ refilling is depend on Ca²⁺ fluxes originating from the endoplasmic reticulum, particularly through ER Ca²⁺ leak pathways at ER–lysosome contact sites. In this context, ER-resident Ca²⁺ leak channels such as TMBIM6 (also known as Bax inhibitor-1) play an important role in maintaining basal cytosolic Ca²⁺ levels that can be subsequently taken up by lysosomes (Kim et al. Autophagy 2020). TMBIM6-mediated Ca²⁺ leak from the ER provides a continuous, low-level Ca²⁺ source that supports lysosomal Ca²⁺ loading, (Kim et al. Autophagy 2020). This mechanism allows lysosomes to replenish their Ca²⁺ stores via Ca²⁺ uptake systems operating at ER–lysosome contact sites. Thus, ER Ca²⁺ leak channels represent a key conduit linking ER Ca²⁺ homeostasis to lysosomal Ca²⁺ filling and function.

Recently, lysosome localized TMEM165 was identified to play an important role in Ca²⁺ filling of lysosomes (Zajac et al. Science Advances 2024). Here, in our study, we observe that TMEM165 drives lysosomal Ca²⁺ influx in melanocytes.

Comment 4- Oregon-green-dextran is not a great probe for lysosomal calcium. Its Kd is 170nM and even in the acidic environment this may be lowered to low micromolar which may not be great for measuring changes around luminal concentrations of around 500uM. Additionally, it is usual to correct for pH effects simultaneously since the dye is also a pH reporter and has been used as such. However, I take the point that they still see an increase in fluorescence whilst pH falls probably indicating an increase in luminal lysosomal calcium confirmed by increased perilysosomal calcium.

Response____: We thank the Reviewer for the careful and balanced assessment of the Oregon Green–dextran measurements. We appreciate the acknowledgment that, despite the known limitations of this probe and its pH sensitivity, the observed increase in fluorescence concurrent with reduced lysosomal pH is consistent with elevated luminal lysosomal Ca²⁺ levels. We are grateful for this positive interpretation, which strengthens our conclusions when considered alongside the large amount of supporting data.

Comment 5- The major point is to reduce the number of main data panels with consigment of some controls perhaps to supplementary. This would increase the comprehensibility of the paper.

Response____: We thank the Reviewer for this constructive and positive suggestion. We appreciate the emphasis on reducing the data in the main figures. Therefore, as suggested, we have moved considerable data to the supplementary figures. However, due to the additional experiments performed to address the concerns of other Reviewers, the main data panels may still look little busy. We sincerely think that the Reviewer would understand our situation.

Minor points

Comment 1- Fig 10 needs a clear legend with symbols in the diagram explained. eg ER calcium release proteins.

Response____: We thank the Reviewer for this helpful and constructive comment. Therefore, we have revised the Figure 10 legend to clearly explain all symbols used in the schematic illustration.

Reviewer #3 (Significance (Required)):

This is a tour de force investigating organelle communication during the process of melanophagy, that is little understood. It highlights many important organelle ion transport events that are important findings in their own right. For example, the importance of TMEM165 in calcium filling of lysosomes.

Response____: We sincerely thank the Reviewer for considering our work as “a tour de force investigation” and appreciating that our study presents several important organelle ion transport events.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

This is a robust and extensive study showing that IP3R2 selectively initiates a calcium signalling pathway leading to melanophagy, that is the degradation of melanosomes. This reduces pigmentation and UV light protection. A strength of the paper is that it combines detailed cellular studies with in viva studies in the zebrafish model. They show that knockdown of IP3R2 reverses this process perhaps leading to a strategy to enhance melanosome number and hence to afford protection from UV irradiation. The authors use a battery of fluorescent probes (mainly genetically encoded reporters) in investigate the signalling cascade leading to melanophagy or its reduction. This involves reportes for a number of different organelles involved in this process. The experiments are generally well performed with clear controls for the probes in many cases. My main issue is the panels conatin to much data which may obscure the message, and a good deal could be moved to supplementary data. The manuscript investigates many mechanisms in distinct organelles which is remarkable for a two author paper. Particularly interesting was the design of novel fluorescent protein reporters for melanophagy itself. One area not explored is ion fluxes across melanosomes themselves which are lysosome-related organelles and may exhibit similar properties and signalsomes of lysosomes. Specifically the authors show that a REDUCTION of IP3R2-mediated calcium release leads to a calcium flux from the ER by a different mechanism (possibly via TMBIM6). This increases calcium loading of the lysosome via TMEM165, at the expense of calcium transfer to mitochondria, and an acidification. This leads to TRPML1 activation and the lysosomal calcium release activates TFEB translocation to the nucleus increases the transcription of autophagy/melanophagy genes and activation of the AMPK-ULK1 pathway (rather than mTOR). This is a complex pathway and evidence is presented for many of the steps involved.

This is a tour de force investigating organelle communication during the process of melanophagy, that is little understood. It highlights many important organelle ion transport events that are important finmdings in their own right. For example, the importance of TMEM165 in calcium filling of lysosomes.

Major points:

1. The authors state that TPC activation does not activate TFEB translocation the the nucleus. This is now not the case and should be at least looked at. What is the role of endolysosomal channels on the melanosomes themselves in melanophagy.
2. How does reduction in IP3R2 mediated calcium fluxes enhance lysosomal acidity?
3. What mediates the ER source for calcium filling of lysosomes?
4. Oregon-green-dextran is not a great probe for lysosomal calcium. Its Kd is 170nM and even in the acidic environment this may be lowered to low micromolar which may not be great for meaduring changes around luminal concentrations of around 500uM. Additionally, it is usual to correct for pH effects simulataneously since the dye is also a pH reporter and has been used as such. However, I take the point that they still see an increase in fluorescence whilst pH falls probably indicating an increase in luminal lysosomal calcium confirmed by increased perilysosomal calcium.

The major point is to reduce the number of main data panels with consigment of some controls perhaps to supplementary. This would increase the comprehensibility of the paper.

Minor points

1. Fig 10 needs a clear legend with symbols in the diagram explained. eg ER calcium release proteins

Significance

This is a tour de force investigating organelle communication during the process of melanophagy, that is little understood. It highlights many important organelle ion transport events that are important findings in their own right. For example, the importance of TMEM165 in calcium filling of lysosomes.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

In the present manuscript, Saurav et al. identify IP3R2-mediated ER calcium release as a key suppressor of melanophagy, thereby sustaining pigmentation in melanocytes. Using in vitro (B16 murine melanoma cells, primary human melanocytes) and in vivo (zebrafish) models, the authors report that IP3R2 expression is positively correlated with pigmentation. They then investigate the impact of IP3R2 knockdown and find that IP3R2 silencing enhances the stability of melanogenic proteins, while also inducing autophagic degradation of melanosomes (i.e., melanophagy). Concomitantly, they find that IP3R2 silencing decreases mitochondrial calcium uptake, increases lysosomal calcium loading, and lowers lysosomal pH. They propose a pathway wherein in IP3R2 knockdown cells impaired mitochondrial calcium uptake induces the activation of AMPK-ULK1, and increased lysosomal calcium activates TRPML1 via TMEM165 and closer proximity interactions between ER and lysosomes, TFEB nuclear translocation, and upregulation of melanophagy-related genes, namely OPTN and RCHY1. The work is placed within the context of emerging roles of organelle calcium signaling in pigmentation biology, where extracellular calcium influx pathways are known regulators, but the contribution of ER-mitochondria-lysosome crosstalk to melanosome turnover remains largely unknown.

Major comments:

• The central finding is that IP3R2 knockdown induces melanophagy and reduces pigmentation. However, the manuscript does not identify any physiological or pathological context in which IP3R2 expression or activity is naturally downregulated in melanocytes. Without such context, the knockdown may represent an artificial perturbation that broadly alters ER calcium handling and triggers melanophagy as part of a general stress-induced autophagy response. This raises uncertainty about whether the pathway operates in vivo under normal or disease conditions. It would strengthen the study to identify upstream cues that reduce IP3R2 function and to test whether these also trigger melanophagy through the proposed mechanism.
• While the data link IP3R2 knockdown to decreased pigmentation and increased melanophagy, the causality between altered organelle calcium dynamics and the melanophagy induction is inferred from correlation and partial rescue experiments. More direct interventions in the proposed downstream pathways (e.g., acute mitochondrial calcium uptake restoration, lysosomal calcium buffering) would strengthen mechanistic claims.
• Zebrafish assays convincingly show altered pigmentation with altered IP3R2 levels, but do not connect this to in vivo melanophagy measurements or TRPML1/TFEB activity, which would link the cell biology to organismal phenotype more directly.
• The work suggests therapeutic potential for pigmentary disorders, but no disease models are tested. It is unclear whether the observed mechanisms operate under physiological stressors.
• The paradox between the observed enhanced stability of melanogenic proteins and increased melanophagy is insufficiently addressed. DCT, Tyrosinase and GP100 are all melanosome-associated and their stability or degradation is in prior literature often interpreted as reflecting melanosome biogenesis and turnover. This discrepancy needs to be resolved, as it complicates interpretation of melanophagy assays.
• The authors propose that mitophagy and ER-phagy are reduced in IP3R2 knockdown cells, suggesting specific induction of melanophagy, but the rationale for why increased autophagic flux only targets melanosomes is insufficiently addressed. Also, these conclusions are solely based on Keima assays, and positive controls for mitophagy and ER-phagy are lacking.
• The melanophagy probes are novel and validated with rapamycin/bafilomycin, but quantitative calibration of GFP/mCherry or Keima signal to actual lysosomal delivery rates is missing; photobleaching, pH heterogeneity (incl., observed decrease in lysosomal pH), and melanin autofluorescence (see below) could confound ratios. Also, side-by-side comparison with other melanophagy detection approaches (e.g., colocalization of melanosomes with LC3) is lacking.
• Melanosomes exhibit broad autofluorescence, particularly upon excitation at 405-488 nm and extending into the red channel. This signal can overlap with the detection ranges for GFP, mCherry, and mKeima reporters, potentially confounding quantitative readouts unless appropriate controls (e.g., untransfected cells, spectral unmixing) are used. Throughout this manuscript, it is not addressed how melanosome autofluorescence was controlled for or excluded in the reported fluorescence measurements.
• While OPTN and RCHY1 expression is elevated upon IP3R2 knockdown, functional engagement (e.g., OPTN localization to melanosomes, melanosome ubiquitination by RCHY1), or necessity (e.g., siRNA knockdown of these in the IP3R2-deficient background), are not tested.
• While siRNA/shRNA efficacy is shown, functional rescue with pore-dead mutants sometimes fails to return to control values. The possibility of partial off-target or compensatory effects is not fully excluded.
• The mitochondrial and lysosomal calcium measurements are largely endpoint peak quantifications; kinetic analyses and buffering capacity measurements would provide more mechanistic depth, especially for the TMEM165 contribution. Also, TMEM165 necessity for melanophagy induction upon IP3R2 knockdown has not been directly addressed.
• The proximity ligation assay between VAP-A and LAMP1 is interpreted as showing increased ER-lysosome contacts in IP3R2 knockdown cells. However, additional controls are needed and quantitative TEM should be included to substantiate changes in organelle contact frequency and distance.
• Some assays report small biological n (e.g., three independent experiments with relatively small per-condition cell counts).

Minor comments:

• The title "IP3R2-mediated inter-organelle Ca2+ signaling orchestrates melanophagy" could be misread as indicating IP3R2 'promotes' melanophagy; consider rewording to make clear that IP3R2 suppresses melanophagy to maintain pigmentation. Similarly, the running title "IP3R2 negatively regulates melanophagy" would be clearer as "IP3R2 suppresses melanophagy".
• Unify the framing of "positively regulates pigmentation" vs. "negatively regulates melanophagy" in the Introduction/Discussion.
• Adding schematic flow diagrams summarizing each pathway at the end of relevant results (figure) sections could help accessibility.
• While the introduction summarizes extracellular calcium signaling in pigmentation, there is less coverage of recent work on selective autophagy of other lysosome-related organelles (e.g., platelet dense granules, lytic granules), which could provide broader mechanistic context.

Significance

This study addresses an important gap in pigmentation biology by identifying IP3R2-mediated ER calcium release as a suppressor of melanophagy and a positive regulator of pigmentation. The strongest aspects are the integration of in vitro and in vivo models, the multi-faceted mechanistic exploration linking altered organelle calcium dynamics to selective melanosome turnover, and the development of novel ratiometric fluorescent probes for live-cell melanophagy measurement. Conceptually, the work extends prior literature that has focused on extracellular calcium influx and melanosome biogenesis, revealing a new inter-organelle calcium signaling module that controls melanosome degradation via AMPK-ULK1 and TMEM165-TRPML1-TFEB pathways.

However, several limitations reduce the strength of the mechanistic claims. Some key pathway steps are inferred from correlation and partial rescue rather than direct necessity/sufficiency tests (e.g., mitochondrial calcium uptake restoration, lysosomal calcium buffering). The paradoxical observation that IP3R2 knockdown both increases melanophagy and stabilizes melanosome-resident proteisn (DCT, Tyrosinase, GP100) is not resolved, complicating interpretation of the melanophagy assays. The specificity for melanophagy over other selective autophagy pathways is asserted but not fully explained mechanistically, and positive controls for mitophagy/ER-phagy are missing. Potential technical confounds, such as melanin autofluorescence in the detection ranges of GFP, mCherry, and mKeima, are not explicitly addressed and alternative assays for these key data were insufficiently employed. In vivo results do not yet connect altered pigmentation to melanophagy readouts or downstream TRPML1/TFEB activation. Importantly, the study does not identify any physiological or pathological scenario in which IP3R2 expression or activity is naturally reduced in melanocytes. In the absence of such upstream cues, IP3R2 knockdown may represent an artificial perturbation that triggers melanophagy as part of a broader stress-induced autophagy response, raising questions about the in vivo relevance of the proposed pathway.

The work's primary audience is specialized, cell biologists, autophagy researchers, and pigmentation/skin biology specialists, but the mechanistic framework on organelle crosstalk and selective autophagy will interest a broader basic research readership, including those studying lysosome-related organelles in other systems. The ratiometric probes could be adapted for future melanophagy research, and the pathway insights may guide translational studies in pigmentary disorders or melanoma. My expertise is in mitochondrial and lysosomal calcium signaling, autophagy, and microscopy-based functional assays; I do not have detailed expertise in zebrafish developmental genetics, though the phenotypic analysis appears sound.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

The manuscript entitled, "IP3R2 mediated inter-organelle Ca2+ signaling orchestrates melanophagy" is a rather diffuse study of the relationship between IP3R2 and melanin production. While this is an interesting and understudied area, the study lacks a clear focus. The model seems to be that IP3R2 is essential for mitochondrial calcium loading. And that its absence increases lysosomal calcium loading. There are also a number of incomplete and/or unconvincing links to autophagy/melanophagy, TMEM165, TRPML1 and even gene transcription. In this kind of diffuse study, each step needs to be convincing to get to the next one, which is not the case here. There are also references to altered proteasome function, despite the total absence of any direct data on the proteasome. Finally, I felt it was sometimes unclear whether the authors were referring to melanosomes or lysosomes at various points throughout the study. While I suspect that, somewhere in here, there are some novel relationships worthy of further investigation, this is a case where the many parts make the overall product less convincing. What effects here are directly relevant to IP3R2? This study should stop there, leaving investigations of peripheral factors for future investigations, as the further you get from where you start, the less clear what you are studying becomes. And the less direct.

Specific Comments:

1. The separation of Figures 1F and 1J makes it impossible to assess the effect of αMSH on IP3R2 expression. This presentation makes interpretation difficult; a simple 4 lane Western would be more informative
2. One of the most attractive points made by this study is that there is a specific link between IP3R2 and melanin production. In my opinion, the null hypothesis is that this is just about the amount of IP3Rs expressed per cell. To reject this concept, the authors should show data demonstrating the relative expression of all 3 IP3Rs. Without this information, the null hypothesis that IP3R2 is the most expressed IP3R isoform and that's why its knockdown has the most dramatic effect cannot be rejected It would also be helpful to show where the different IP3Rs are expressed within the cell.
3. It would be helpful to label Figs 3F-I with the conditions used. The description in the text is of increased LC3II levels, however, the ratio of LC3I to LC3II might be more meaningful. Irrespective, although the graph shows an increase in LC3II, the Western really doesn't show much. As a standalone finding, I don't find this figure to be very convincing; there are better options to demonstrate this proposed relationship between IP3R2 and autophagy than what is shown.
4. The following statement at the beginning of page 22 "We observed an impaired proteasomal degradation of critical melanogenic proteins localized on melanosomes in the IP3R2 knockdown condition" is insufficiently supported by data to be made. Even if I was convinced that autophagy was enhanced, there is no data of any kind about the proteasome in this manuscript.
5. In figure 5, the authors create a new ratiometric dye to detect melanosome stability based on the principle that tyrosinase is exclusively found in melanosomes. Unfortunately, there is no validation that this new construct is found exclusively in melanosomes upon expression. In addition, there is discussion about the pH of lysosomes, but not of melanosomes. Ultimately, this data cannot be considered at face value without any type of validation; I also note that the pictures lack sufficient detail to support identification of these stuctures asmelanosomes.

While I maintain the above concerns, I note that, the data in supplemental figure 3 is MUCH more convincing than what is in the figure. Both the writing and the figure design should be rethought. 6. Given the increase in ER Ca2+ content after IP3R2 knockdown, ER calcium content should be emptied before attempting to estimate lysosomal Ca2+ content with GPN or Bafilomycin. Otherwise, the source of calcium is less than clear.

Significance

The manuscript entitled, "IP3R2 mediated inter-organelle Ca2+ signaling orchestrates melanophagy" is a rather diffuse study of the relationship between IP3R2 and melanin production. While this is an interesting and understudied area, the study lacks a clear focus. The model seems to be that IP3R2 is essential for mitochondrial calcium loading. And that its absence increases lysosomal calcium loading. There are also a number of incomplete and/or unconvincing links to autophagy/melanophagy, TMEM165, TRPML1 and even gene transcription. In this kind of diffuse study, each step needs to be convincing to get to the next one, which is not the case here. There are also references to altered proteasome function, despite the total absence of any direct data on the proteasome. Finally, I felt it was sometimes unclear whether the authors were referring to melanosomes or lysosomes at various points throughout the study.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

We are grateful to the Review Commons reviewers for their constructive feedback, which has significantly strengthened the manuscript. In response, we have performed additional experiments, revised and expanded multiple figures, incorporated new statistical and functional analyses, and carefully edited the text to improve clarity and precision. A detailed point-by-point response to all reviewer comments, together with a summary of revised figures, is provided.

To address the reviewers' suggestions, we have conducted additional experiments that are now incorporated into new figures, or we have added new images to several existing figures where appropriate.

For this reason, please note that all figures have been renumbered to improve clarity and facilitate cross-referencing throughout the text. As recommended by Referee #3, all figure legends have been thoroughly revised to reflect these updates and are now labeled following the standard A-Z panel format, enhancing readability and ensuring easier identification. In addition, all figure legends now include the sample size for each statistical analysis.

For clarity and ease of reference, we provide below a comprehensive list of all figures included in the revised version. Figures that have undergone modifications are underlined.

Figure 1____. The first spermatogenesis wave in prepuberal mice.

This figure now includes amplified images of representative spermatocytes and a summary schematic illustrating the timeline of spermatogenesis. In addition, it now presents the statistical analysis of spermatocyte quantification to support the visual data.

__Figure 2.____ Cilia emerge across all stages of prophase I in spermatocytes during the first spermatogenesis wave. __

The images of this figure remain unchanged from the original submission, but all the graphs present now the statistical analysis of spermatocyte quantification.

Figure 3. Ultrastructure and markers of prepuberal meiotic cilia.

This figure remains unchanged from the original submission; however, we have replaced the ARL3-labelled spermatocyte image (A) with one displaying a clearer and more representative signal.

__Figure 4. Testicular tissue presents spermatocyte cysts in prepuberal mice and adult humans. __

This figure remains unchanged from the original submission.

__Figure 5. Cilia and flagella dynamics are correlated during prepuberal meiosis. __

This figure remains unchanged from the original submission.

__Figure 6. Comparative proteomics identifies potential regulators of ciliogenesis and flagellogenesis. __

This figure remains unchanged from the original submission.

Figure 7.____ Deciliation induces persistence of DNA damage in meiosis.

This figure has been substantially revised and now includes additional experiments analyzing chloral hydrate treatment, aimed at more accurately assessing DNA damage under both control and treated conditions. Images F-I and graph J are new.

Figure 8____. Aurora kinase A is a regulator of cilia disassembly in meiosis.

This figure is remodelled as the original version contained a mistake in previous panel II, for this, graph in new Fig.8 I has been corrected. In addition, it now contains additional data of αTubulin staining in arrested ciliated metaphases I after AURKA inhibition (new panel L1´).

__Figure 9. Schematic representation of the prepuberal versus adult seminiferous epithelium. __

This figure remains unchanged from the original submission.

__Supplementary Figure 1. Meiotic stages during the first meiotic wave. __

This figure remains unchanged from the original submission.

__Supplementary Figure 2 (new)____. __

This is a new figure that includes additional data requested by the reviewers. It includes additional markers of cilia in spermatocytes (glutamylated Tubulin/GT335), and the control data of cilia markers in non-ciliated spermatocytes. It also includes now the separated quantification of ciliated spermatocytes for each stage, as requested by reviewers, complementing graphs included in Figure 2.

Please note that with the inclusion of this new Supplementary Figure 2, the numbering of subsequent supplementary figures has been updated accordingly.

Supplementary Figure 3 (previously Suppl. Fig. 2)__. Ultrastructure of prophase I spermatocytes. __

This figure is equal in content to the original submission, but some annotations have been included.

Supplementary Figure 4 (previously Suppl. Fig. 3).__ Meiotic centrosome under the electron microscope. __

This figure remains unchanged from the original submission, but additional annotations have been included.

Supplementary Figure 5 (previously Suppl. Fig. 4)__. Human testis contains ciliated spermatocytes. __

This figure has been revised and now includes additional H2AX staining to better determine the stage of ciliated spermatocytes and improve their identification.

Supplementary Figure 6 (previously Suppl. Fig. 5). GLI1 and GLI3 readouts of Hedgehog signalling are not visibly affected in prepuberal mouse testes.

This figure has been remodeled and now includes the quantification of GLI1 and GLI3 and its corresponding statistical analysis. It also includes the control data for Tubulin, instead of GADPH.

Supplementary Figure 7 (previously Suppl. Fig. 6)__. CH and MLN8237 optimization protocol. __

This figure has been remodeled to incorporate control experiments using 1-hour organotypic culture treatment.

Supplementary Figure 8 (previously Suppl. Fig. 7)__. Tracking first meiosis wave with EdU pulse injection during prepubertal meiosis. __This figure remains unchanged from the original submission.

Supplementary Figure 9 (previously Suppl. Fig. 8)__. PLK1 and AURKA inhibition in cultured spermatocytes. __

This figure has been remodeled and now includes additional data on spindle detection in control and AURKA-inhibited spermatocytes (both ciliated and non ciliated).

DETAILED POINT-BY-POINT RESPONSE TO THE REVIEWERS

We will submit both the PDF version of the revised manuscript and the Word file with tracked changes relative to the original submission. Each modification made in response to reviewers' suggestions is annotated in the Word document within the corresponding section of the text. all new figures have also been uploaded to the system.

Response to the Referee #1

In this manuscript by Perez-Moreno et al., titled "The dynamics of ciliogenesis in prepubertal mouse meiosis reveal new clues about testicular maturation during puberty", the authors characterize the development of primary cilia during meiosis in juvenile male mice. The authors catalog a variety of testicular changes that occur as juvenile mice age, such as changes in testis weight and germ cell-type composition. They next show that meiotic prophase cells initially lack cilia, and ciliated meiotic prophase cells are detected after 20 days postpartum, coinciding with the time when post-meiotic spermatids within the developing testes acquire flagella. They describe that germ cells in juvenile mice harbor cilia at all substages of meiotic prophase, in contrast to adults where only zygotene stage meiotic cells harbor cilia. The authors also document that cilia in juvenile mice are longer than those in adults. They characterize cilia composition and structure by immunofluorescence and EM, highlighting that cilia polymerization may initially begin inside the cell, followed by extension beyond the cell membrane. Additionally, they demonstrate ciliated cells can be detected in adult human testes. The authors next perform proteomic analyses of whole testes from juvenile mice at multiple ages, which may not provide direct information about the extremely small numbers of ciliated meiotic cells in the testis, and is lacking follow up experiments, but does serve as a valuable resource for the community. Finally, the authors use a seminiferous tubule culturing system to show that chemical inhibition of Aurora kinase A likely inhibits cilia depolymerization upon meiotic prophase I exit and leads to an accumulation of metaphase-like cells harboring cilia. They also assess meiotic recombination progression using their culturing system, but this is less convincing.

Author response: We sincerely thank Ref #1 for the thorough and thoughtful evaluation of our manuscript. We are particularly grateful for the reviewer's careful reading and constructive feedback, which have helped us refine several sections of the text and strengthen our discussion. All comments and suggestions have been carefully considered and addressed, as detailed below.

__Major comments: __

1. There are a few issues with the experimental set up for assessing the effects of cilia depolymerization on DNA repair (Figure 7-II). First, how were mid pachytene cells identified and differentiated from early pachytene cells (which would have higher levels of gH2AX) in this experiment? I suggest either using H1t staining (to differentiate early/mid vs late pachytene) or the extent of sex chromosome synapsis. This would ensure that the authors are comparing similarly staged cells in control and treated samples. Second, what were the gH2AX levels at the starting point of this experiment? A more convincing set up would be if the authors measure gH2AX immediately after culturing in early and late cells (early would have higher gH2AX, late would have lower gH2AX), and then again after 24hrs in late cells (upon repair disruption the sampled late cells would have high gH2AX). This would allow them to compare the decline in gH2AX (i.e., repair progression) in control vs treated samples. Also, it would be informative to know the starting gH2AX levels in ciliated vs non-ciliated cells as they may vary.

Response:

We thank Ref #1 for this valuable comment, which significantly contributed to improving both the design and interpretation of the cilia depolymerization assay.

Following this suggestion, we repeated the experiment including 1-hour (immediately after culturing), and 24-hour cultures for both control and chloral hydrate (CH)-treated samples (n = 3 biological replicates). To ensure accurate staging, we now employ triple immunolabelling for γH2AX, SYCP3, and H1T, allowing clear distinction of zygotene (H1T−), early pachytene (H1T−), and late pachytene (H1T+) cells. The revised data (Figure 7) now provide a more complete and statistically robust analysis of DNA damage dynamics. These results confirm that CH-induced deciliation leads to persistence of the γH2AX signal at 24 hours, indicating impaired DNA repair progression in pachytene spermatocytes. The new images and graphs are included in the revised Figure 7.

Regarding the reviewer's final point about the comparison of γH2AX levels between ciliated and non-ciliated cells, we regret that direct comparison of γH2AX levels between ciliated and non-ciliated cells is not technically feasible. To preserve cilia integrity, all cilia-related imaging is performed using the squash technique, which maintains the three-dimensional structure of the cilia but does not allow reliable quantification of DNA damage markers due to nuclear distortion. Conversely, the nuclear spreading technique, used for DNA damage assessment, provides optimal visualization of repair foci but results in the loss of cilia due to cytoplasmic disruption during the hypotonic step. Given that spermatocytes in juvenile testes form developmentally synchronized cytoplasmic cysts, we consider that analyzing a statistically representative number of spermatocytes offers a valid and biologically meaningful measure of tissue-level effects.

In conclusion, we believe that the additional experiments and clarifications included in revised Figure 7 strengthen our conclusion that cilia depolymerization compromises DNA repair during meiosis. Further functional confirmation will be pursued in future works, since we are currently generating a conditional genetic model for a ciliopathy in our laboratory.

The authors analyze meiotic progression in cells cultured with/without AURKA inhibition in Figure 8-III and conclude that the distribution of prophase I cells does not change upon treatment. Is Figure 8-III A and B the same data? The legend text is incorrect, so it's hard to follow. Figure 8-III A shows a depletion of EdU-labelled pachytene cells upon treatment. Moreover, the conclusion that a higher proportion of ciliated zygotene cells upon treatment (Figure 8-II C) suggests that AURKA inhibition delays cilia depolymerization (page 13 line 444) does not make sense to me.

Response:

We thank Ref#1 for identifying this issue and for the careful examination of Figure 8. We discovered that the submitted version of Figure 8 contained a mismatch between the figure legend and the figure panels. The legend text was correct; however, the figure inadvertently included a non-corresponding graph (previously panel II-A), which actually belonged to Supplementary Figure 7 in the original submission. We apologize for this mistake.

This error has been corrected in the revised version. The updated Figure 8 now accurately presents the distribution of EdU-labelled spermatocytes across prophase I substages in control and AURKA-inhibited cultures (previously Figure 8-II B, now Figure 8-A). The corrected data show no significant differences in the proportions of EdU-labelled spermatocytes among prophase I substages after 24 hours of AURKA inhibition, confirming that meiotic progression is not delayed and that no accumulation of zygotene cells occurs under this treatment. Therefore, the observed increase in ciliated zygotene spermatocytes upon AURKA inhibition (new Figure 8 H-I) is best explained by a delay in cilia disassembly, rather than by an arrest or slowdown in meiotic progression. The figure legend and main text have been revised accordingly.

How do the authors know that there is a monopolar spindle in Figure 8-IV treated samples? Perhaps the authors can use a different Tubulin antibody (that does not detect only acetylated Tubulin) to show that there is a monopolar spindle.

Response:

We appreciate Ref#1 for this excellent suggestion. In the original submission (lines 446-447), we described that ciliated metaphase I spermatocytes in AURKA-inhibited samples exhibited monopolar spindle phenotypes. This description was based on previous reports showing that AURKA or PLK1 inhibition produces metaphases with monopolar spindles characterized by aberrant yet characteristic SYCP3 patterns, abnormal chromatin compaction, and circular bivalent alignment around non-migrated centrosomes (1). In our study, we observed SYCP3 staining consistent with these characteristic features of monopolar metaphases I.

However, we agree with Ref #1 that this could be better sustained with data. Following the reviewer's suggestion, we performed additional immunostaining using α-Tubulin, which labels total microtubules rather than only the acetylated fraction. For clarity purposes, the revised Figure 8 now includes α-Tubulin staining in the same ciliated metaphase I cells shown in the original submission, confirming the presence of defective microtubule polymerization and defective spindle organization. For clarity, we now refer to these ciliated metaphases I as "arrested MI". This new data further support our conclusion that AURKA inhibition disrupts spindle bipolarization and prevents cilia depolymerization, indicating that cilia maintenance and bipolar spindle organization are mechanistically incompatible events during male meiosis. The abstract, results, and discussion section has been expanded accordingly, emphasizing that the persistence of cilia may interfere with microtubule polymerization and centrosome separation under AURKA inhibition. The Discussion has been expanded to emphasize that persistence of cilia may interfere with centrosome separation and microtubule polymerization, contrasting with invertebrate systems -e.g. Drosophila (2) and P. brassicae (3)- in which meiotic cilia persist through metaphase I without impairing bipolar spindle assembly.

1. Alfaro, et al. EMBO Rep 22, (2021). DOI: 15252/embr.202051030 (PMID: 33615693)
2. Riparbelli et al . Dev Cell (2012) DOI: 1016/j.devcel.2012.05.024 (PMID: 22898783)
3. Gottardo et al, Cytoskeleton (Hoboken) (2023) DOI: 1002/cm.21755 (PMID: 37036073)

The authors state in the abstract that they provide evidence suggesting that centrosome migration and cilia depolymerization are mutually exclusive events during meiosis. This is not convincing with the data present in the current manuscript. I suggest amending this statement in the abstract.

Response:

We thank Ref#1 for this valuable observation, with which we fully agree. To avoid overstatement, the original statement has been removed from the Abstract, Results, and Discussion, and replaced with a more accurate formulation indicating that cilia maintenance and bipolar spindle formation are mutually exclusive events during mouse meiosis.

This revised statement is now directly supported by the new data presented in Figure 8, which demonstrate that AURKA inhibition prevents both spindle bipolarization and cilia depolymerization. We are grateful to the reviewer for highlighting this important clarification.

Minor comments:

The presence of cilia in all stages of meiotic prophase I in juvenile mice is intriguing. Why is the cellular distribution and length of cilia different in prepubertal mice compared to adults (where shorter cilia are present only in zygotene cells)? What is the relevance of these developmental differences? Do cilia serve prophase I functions in juvenile mice (in leptotene, pachytene etc.) that are perhaps absent in adults?

Related to the above point, what is the relevance of the absence of cilia during the first meiotic wave? If cilia serve a critical function during prophase I (for instance, facilitating DSB repair), does the lack of cilia during the first wave imply differing cilia (and repair) requirements during the first vs latter spermatogenesis waves?

In my opinion, these would be interesting points to discuss in the discussion section.

Response:

We thank the reviewer for these thoughtful observations, which we agree are indeed intriguing.

We believe that our findings likely reflect a developmental role for primary cilia during testicular maturation. We hypothesize that primary cilia at this stage might act as signaling organelles, receiving cues from Sertoli cells or neighboring spermatocytes and transmitting them through the cytoplasmic cysts shared by spermatocytes. Such intercellular communication could be essential for coordinating tissue maturation and meiotic entry during puberty. Although speculative, this hypothesis aligns with the established role of primary cilia as sensory and signaling hubs for GPCR and RTK pathways regulating cell differentiation and developmental patterning in multiple tissues (e.g., 1, 2). The Discussion section has been expanded to include these considerations.

1. Goetz et al, Nat Rev Genet (2010)- DOI: 1038/nrg2774 (PMID: 20395968)
2. Naturky et al , Cell (2019) DOI: 1038/s41580-019-0116-4 (PMID: 30948801) Our study focuses on the first spermatogenic wave, which represents the transition from the juvenile to the reproductive phase. It is therefore plausible that the transient presence of longer cilia during this period reflects a developmental requirement for external signaling that becomes dispensable in the mature testis. Given that this is only the second study to date examining mammalian meiotic cilia, there remains a vast area of research to explore. We plan to address potential signaling cascades involved in these processes in future studies.

On the other hand, while we cannot confirm that the cilia observed in zygotene spermatocytes persist until pachytene within the same cell, it is reasonable to speculate that they do, serving as longer-lasting signaling structures that facilitate testicular development during the critical pubertal window. In addition, the observation of ciliated spermatocytes at all prophase I substages at 20 dpp, together with our proteomic data, supports the idea that the emergence of meiotic cilia exerts a significant developmental impact on testicular maturation.

In summary, although we cannot yet define specific prophase I functions for meiotic cilia in juvenile spermatocytes, our data demonstrate that the first meiotic wave differs from later waves in cilia dynamics, suggesting distinct regulatory requirements between puberty and adulthood. These findings underscore the importance of considering developmental context when using the first meiotic wave as a model for studying spermatogenesis.

The authors state on page 9 lines 286-288 that the presence of cytoplasmic continuity via intercellular bridges (between developmentally synchronous spermatocytes) hints towards a mechanism that links cilia and flagella formation. Please clarify this statement. While the correlation between the timing of appearance of cilia and flagella in cells that are located within the same segment of the seminiferous tubule may be hinting towards some shared regulation, how would cytoplasmic continuity participate in this regulation? Especially since the cytoplasmic continuity is not between the developmentally distinct cells acquiring the cilia and flagella?

Response:

We thank Ref#1 for this excellent question and for the opportunity to clarify our statement.

The presence of intercellular bridges between spermatocytes is well known and has long been proposed to support germ cell communication and synchronization (1,2) as well as sharing mRNA (3) and organelles (4). A classic example is the Akap gene, located on the X chromosome and essential for the formation of the sperm fibrous sheath; cytoplasmic continuity through intercellular bridges allows Akap-derived products to be shared between X- and Y-bearing spermatids, thereby maintaining phenotypic balance despite transcriptional asymmetry (5). In addition, more recent work has further demonstrated that these bridges are critical for synchronizing meiotic progression and for processes such as synapsis, double-strand break repair, and transposon repression (6).

In this context, and considering our proteomic data (Figure 6), our statement did not intend to imply direct cytoplasmic exchange between ciliated and flagellated cells. Although our current methods do not allow comprehensive tracing of cytoplasmic continuity from the basal to the luminal compartment of the seminiferous epithelium, we plan to address this limitation using high-resolution 3D and ultrastructural imaging approaches in future studies.

Based on our current data, we propose that cytoplasmic continuity within developmentally synchronized spermatocyte cysts could facilitate the coordinated regulation of ciliogenesis, and similarly enable the sharing of regulatory factors controlling flagellogenesis within spermatid cysts. This coordination may occur through the diffusion of centrosomal or ciliary proteins, mRNAs, or signaling intermediates involved in the regulation of microtubule dynamics. However, we cannot exclude the possibility that such cytoplasmic continuity extends across all spermatocytes derived from the same spermatogonial clone, potentially providing a larger regulatory network.]] This mechanism could help explain the temporal correlation we observe between the appearance of meiotic cilia and the onset of flagella formation in adjacent spermatids within the same seminiferous segment.

We have revised the Discussion to explicitly clarify this interpretation and to note that, although hypothetical, it is consistent with established literature on cytoplasmic continuity and germ cell coordination.

1. Dym, et al. * Reprod.*(1971) DOI: 10.1093/biolreprod/4.2.195 (PMID: 4107186)
2. Braun et al. Nature. (1989) DOI: 1038/337373a0 (PMID: 2911388)
3. Greenbaum et al. * Natl. Acad. Sci. USA*(2006). DOI: 10.1073/pnas.0505123103 (PMID: 16549803)
4. Ventelä et al. Mol Biol Cell. (2003) DOI: 1091/mbc.e02-10-0647 (PMID: 12857863)
5. Turner et al. Journal of Biological Chemistry (1998). DOI: 1074/jbc.273.48.32135 (PMID: 9822690)
6. Sorkin, et al. Nat Commun (2025). DOI: 1038/s41467-025-56742-9 (PMID: 39929837) *note: due to manuscript-length limitations, not all cited references can be included in the text; they are listed here to substantiate our response.

Individual germ cells in H&E-stained testis sections in Figure 1-II are difficult to see. I suggest adding zoomed-in images where spermatocytes/round spermatids/elongated spermatids are clearly distinguishable.

Response:

Ref#1 is very right in this suggestion. We have revised Figure 1 to improve the quality of the H&E-stained testis sections and have added zoomed-in panels where spermatocytes, round spermatids, and elongated spermatids are clearly distinguishable. These additions significantly enhance the clarity and interpretability of the figure.

In Figure 2-II B, the authors document that most ciliated spermatocytes in juvenile mice are pachytene. Is this because most meiotic cells are pachytene? Please clarify. If the data are available (perhaps could be adapted from Figure 1-III), it would be informative to see a graph representing what proportions of each meiotic prophase substages have cilia.

Response:

We thank the reviewer for this valuable observation. Indeed, the predominance of ciliated pachytene spermatocytes reflects the fact that most meiotic cells in juvenile testes are at the pachytene stage (Figure 1). We have clarified this point in the text and have added a new supplementary figure (Supplementary Figure 2, new figure) presenting a graph showing the proportion of spermatocytes at each prophase I substage that possess primary cilia. This visualization provides a clearer quantitative overview of ciliation dynamics across meiotic substages.

I suggest annotating the EM images in Sup Figure 2 and 3 to make it easier to interpret.

Response:

We thank the reviewer for this helpful suggestion. We have now added annotations to the EM images in Supplementary Figures 3 and 4 to facilitate their interpretation. These visual guides help readers more easily identify the relevant ultrastructural features described in the text.

The authors claim that the ratio between GLI3-FL and GLI3-R is stable across their analyzed developmental window in whole testis immunoblots shown in Sup Figure 5. Quantifying the bands and normalizing to the loading control would help strengthen this claim as it hard to interpret the immunoblot in its current form.

Response:

We thank the reviewer for this valuable suggestion. Following this recommendation, Supplementary Figure 5 has been revised to include quantification of GLI1 and GLI3 protein levels, normalized to the loading control.

After quantification, we observed statistically significant differences across developmental stages. Specifically, GLI1 expression is slightly higher at 21 dpp compared to 8 dpp. For GLI3, we performed two complementary analyses:

• Total GLI3 protein (sum of full-length and repressor forms normalized to loading control) shows a progressive decrease during development, with the lowest levels at 60 dpp (Supplementary Figure 5D).
• GLI3 activation status, assessed as the GLI3-FL/GLI3-R ratio, is highest during the 19-21 dpp window, compared to 8 dpp and 60 dpp. Although these results suggest a possible transient activation of GLI3 during testicular maturation, we caution that this cannot automatically be attributed to increased Hedgehog signaling, as GLI3 processing can also be affected by other processes, such as changes in ciliogenesis. Furthermore, because the analysis was performed on whole-testis protein extracts, these changes cannot be specifically assigned to ciliated spermatocytes.

We have expanded the Discussion to address these findings and to highlight the potential involvement of the Desert Hedgehog (DHH) pathway, which plays key roles in testicular development, Sertoli-germ cell communication, and spermatogenesis (1, 2, 3). We plan to investigate these pathways further in future studies.

1. Bitgood et al. Curr Biol. (1996). DOI: 1016/s0960-9822(02)00480-3 (PMID: 8805249)
2. Clark et al. Biol Reprod. (2000) DOI: 1095/biolreprod63.6.1825 (PMID: 11090455)
3. O'Hara et al. BMC Dev Biol. (2011) DOI: 1186/1471-213X-11-72 (PMID: 22132805) *note: due to manuscript-length limitations, not all cited references can be included in the text; they are listed here to substantiate our response.

There are a few typos throughout the manuscript. Some examples: page 5 line 172, Figure 3-I legend text, Sup Figure 5-II callouts, Figure 8-III legend, page 15 line 508, page 17 line 580, page 18 line 611.

Response:

We thank the reviewer for detecting this. All typographical errors have been corrected, and figure callouts have been reviewed for consistency.

Response to the Referee #2

This study focuses on the dynamic changes of ciliogenesis during meiosis in prepubertal mice. It was found that primary cilia are not an intrinsic feature of the first wave of meiosis (initiating at 8 dpp); instead, they begin to polymerize at 20 dpp (after the completion of the first wave of meiosis) and are present in all stages of prophase I. Moreover, prepubertal cilia (with an average length of 21.96 μm) are significantly longer than adult cilia (10 μm). The emergence of cilia coincides temporally with flagellogenesis, suggesting a regulatory association in the formation of axonemes between the two. Functional experiments showed that disruption of cilia by chloral hydrate (CH) delays DNA repair, while the AURKA inhibitor (MLN8237) delays cilia disassembly, and centrosome migration and cilia depolymerization are mutually exclusive events. These findings represent the first detailed description of the spatiotemporal regulation and potential roles of cilia during early testicular maturation in mice. The discovery of this phenomenon is interesting; however, there are certain limitations in functional research.

We thank Referee #2 for their careful reading of the manuscript and for highlighting important limitations regarding functional interpretation.

Our primary objective in this study was to provide a rigorous structural, temporal, and developmental characterization of meiotic ciliogenesis in the mammalian testis, a process for which almost no prior data exist. Given this lack of foundational information, we focused on establishing when, where, and in which meiotic stages primary cilia form during prepubertal development, and on identifying candidate regulatory pathways using complementary imaging, proteomic, and pharmacological approaches.

We agree that genetic ablation models would provide the most direct means to test ciliary function during spermatogenesis. However, we believe that such functional analyses must be preceded by a detailed developmental and phenotypic framework, which was previously unavailable. The present study therefore represents a necessary first step, defining the dynamics, ultrastructure, and molecular context of meiotic cilia during the transition from juvenile to adult spermatogenesis. We are currently generating conditional genetic models to directly address functional mechanisms in future work.

Regarding the temporal coincidence between the emergence of meiotic cilia and the onset of flagellogenesis, we do not interpret this observation as evidence of stochastic or non-functional protein expression. Rather, we present it as a developmental correlation that may reflect shared regulatory constraints on axonemal assembly during testicular maturation. We have clarified in the revised manuscript that this relationship is descriptive and hypothesis-generating, and we avoid assigning direct causal roles.

With respect to the proteomic analysis, we agree that proteomics alone cannot establish function. Our intent was not to assign causality, but to provide a developmental, hypothesis-generating dataset identifying candidate regulators that are enriched at the precise developmental window when both meiotic cilia and spermatid flagella first emerge. We have revised the text to explicitly frame these data as a resource for future mechanistic studies, rather than as direct functional evidence.

Taken together, we believe that the revised manuscript now more accurately reflects the scope and limitations of the study, while providing a robust and much-needed developmental framework for future genetic and functional analyses of meiotic ciliogenesis in mammals. We would be happy to further clarify any aspect of these interpretations if the reviewer or editor considers it helpful.

Major points:

1. The prepubertal cilia in spermatocytes discovered by the authors lack specific genetic ablation to block their formation, making it impossible to evaluate whether such cilia truly have functions. Because neither in the first wave of spermatogenesis nor in adult spermatogenesis does this type of cilium seem to be essential. In addition, the authors also imply that the formation of such cilia appears to be synchronized with the formation of sperm flagella. This suggests that the production of such cilia may merely be transient protein expression noise rather than a functionally meaningful cellular structure.

Response:

We agree that a genetic ablation model would represent the ideal approach to directly test cilia function in spermatogenesis. However, given the complete absence of prior data describing the dynamics of ciliogenesis during testis development, our priority in this study was to establish a rigorous structural and temporal characterization of this process in the main mammalian model organism, the mouse. This systematic and rigorous phenotypic characterization is a necessary first step before any functional genetics could be meaningfully interpreted.

To our knowledge, this study represents the first comprehensive analysis of ciliogenesis during prepubertal mouse meiosis, extending our previous work on adult spermatogenesis (1). Beyond these two contributions, only four additional studies have addressed meiotic cilia-two in zebrafish (2, 3), with Mytlys et al. also providing preliminary observations relevant to prepubertal male meiosis that we discuss in the present work, one in Drosophila (4) and a recent one in butterfly (5). No additional information exists for mammalian gametogenesis to date.

1. López-Jiménez et al. Cells (2022) DOI: 10.3390/cells12010142 (PMID: 36611937)
2. Mytlis et al. Science (2022) DOI: 10.1126/science.abh3104 (PMID: 35549308)
3. Xie et al. J Mol Cell Biol (2022) DOI: 10.1093/jmcb/mjac049 (PMID: 35981808)
4. Riparbelli et al . Dev Cell (2012) DOI: 10.1016/j.devcel.2012.05.024 (PMID: 22898783)
5. Gottardo et al, Cytoskeleton (Hoboken) (2023) DOI: 10.1002/cm.21755 (PMID: 37036073) We therefore consider this descriptive and analytical foundation to be essential before the development of functional genetic models. Indeed, we are currently generating a conditional genetic model for a ciliopathy in our laboratory. These studies are ongoing and will directly address the type of mechanistic questions raised here, but they extend well beyond the scope and feasible timeframe of the present manuscript.

We thus maintain that the present work constitutes a necessary and timely contribution, providing a robust reference dataset that will facilitate and guide future functional studies in the field of cilia and meiosis.

Taking this into account, we would be very pleased to address any additional, concrete suggestions from Ref#2 that could further strengthen the current version of the manuscript

The high expression of axoneme assembly regulators such as TRiC complex and IFT proteins identified by proteomic analysis is not particularly significant. This time point is precisely the critical period for spermatids to assemble flagella, and TRiC, as a newly discovered component of flagellar axonemes, is reasonably highly expressed at this time. No intrinsic connection with the argument of this paper is observed. In fact, this testicular proteomics has little significance.

Response:

We appreciate this comment but respectfully disagree with the reviewer's interpretation of our proteomic data. To our knowledge, this is the first proteomic study explicitly focused on identifying ciliary regulators during testicular development at the precise window (19-21 dpp) when both meiotic cilia and spermatid flagella first emerge.

While Piprek et al (1) analyzed the expression of primary cilia in developing gonads, proteomic data specifically covering the developmental transition at 19-21 dpp were not previously available. Furthermore, a recent cell-sorting study (2), detected expression of cilia proteins in pachytene spermatocytes compared to round spermatids, but did not explore their functional relevance or integrate these data with developmental timing or histological context.

In contrast, our dataset integrates histological staging, high-resolution microscopy, and quantitative proteomics, revealing a set of candidate regulators (including DCAF7, DYRK1A, TUBB3, TUBB4B, and TRiC) potentially involved in cilia-flagella coordination. We view this as a hypothesis-generating resource that outlines specific proteins and pathways for future mechanistic studies on both ciliogenesis and flagellogenesis in the testis.

Although we fully agree that proteomics alone cannot establish causal function, we believe that dismissing these data as having little significance overlooks their value as the first molecular map of the testis at the developmental window when axonemal structures arise. Our dataset provides, for the first time, an integrated view of proteins associated with ciliary and flagellar structures at the developmental stage when both axonemal organelles first appear. We thus believe that our proteomic dataset represents an important and novel contribution to the understanding of testicular development and ciliary biology.

Considering this, we would again welcome any specific suggestions from Ref#2 on additional analyses or clarifications that could make the relevance of this dataset even clearer to readers.

1. Piprek et al. Int J Dev Biol. (2019) doi: 10.1387/ijdb.190049rp (PMID: 32149371).
2. Fang et al. Chromosoma. (1981) doi: 10.1007/BF00285768 (PMID: 7227045). Response to the Referee #3

In "The dynamics of ciliogenesis in prepubertal mouse meiosis reveals new clues about testicular development" Pérez-Moreno, et al. explore primary cilia in prepubertal mouse spermatocytes. Using a combination of microscopy, proteomics, and pharmacological perturbations, the authors carefully characterize prepubertal spermatocyte cilia, providing foundational work regarding meiotic cilia in the developing mammalian testis.

Response: We sincerely thank Ref#3 for their positive assessment of our work and for the thoughtful suggestions that have helped us strengthen the manuscript. We are pleased that the reviewer recognizes both the novelty and the relevance of our study in providing foundational insights into meiotic ciliogenesis during prepubertal testicular development. All specific comments have been carefully considered and addressed as detailed below.

Major concerns:

1. The authors provide evidence consistent with cilia not being present in a larger percentage of spermatocytes or in other cells in the testis. The combination of electron microscopy and acetylated tubulin antibody staining establishes the presence of cilia; however, proving a negative is challenging. While acetylated tubulin is certainly a common marker of cilia, it is not in some cilia such as those in neurons. The authors should use at least one additional cilia marker to better support their claim of cilia being absent.

Response:

We thank the reviewer for this helpful suggestion. In the revised version, we have strengthened the evidence for cilia identification by including an additional ciliary marker, glutamylated tubulin (GT335), in combination with acetylated tubulin and ARL13B (which were included in the original submission). These data are now presented in the new Supplementary Figure 2, which also includes an example of a non-ciliated spermatocyte showing absence of both ARL13B and AcTub signals.

Taken together, these markers provide a more comprehensive validation of cilia detection and confirm the absence of ciliary labelling in non-ciliated spermatocytes.

The conclusion that IFT88 localizes to centrosomes is premature as key controls for the IFT88 antibody staining are lacking. Centrosomes are notoriously "sticky", often sowing non-specific antibody staining. The authors must include controls to demonstrate the specificity of the staining they observe such as staining in a genetic mutant or an antigen competition assay.

Response:

We appreciate the reviewer's concern and fully agree that antibody specificity is critical when interpreting centrosomal localization. The IFT88 antibody used in our study is commercially available and has been extensively validated in the literature as both a cilia marker (1, 2), and a centrosome marker in somatic cells (3). Labelling of IFT88 in centrosomes has also been previously described using other antibodies (4, 5). In our material, the IFT88 signal consistently appears at one of the duplicated centrosomes and at both spindle poles-patterns identical to those reported in somatic cells. We therefore consider the reported meiotic IFT88 staining as specific and biologically reliable.

That said, we agree that genetic validation would provide the most definitive confirmation. We would like to inform that we are currently since we are currently generating a conditional genetic model for a ciliopathy in our laboratory that will directly assess both antibody specificity and functional consequences of cilia loss during meiosis. These experiments are in progress and will be reported in a follow-up study.

1. Wong et al. Science (2015). DOI: 1126/science.aaa5111 (PMID: 25931445)
2. Ocbina et al. Nat Genet (2011). DOI: 1038/ng.832 (PMID: 21552265)
3. Vitre et al. EMBO Rep (2020). DOI: 15252/embr.201949234 (PMID: 32270908)
4. Robert A. et al. J Cell Sci (2007). DOI: 1242/jcs.03366 (PMID: 17264151)
5. Singla et al, Developmental Cell (2010). DOI: 10.1016/j.devcel.2009.12.022 (PMID: 20230748) *note: due to manuscript-length limitations, not all cited references can be included in the text; they are listed here to substantiate our response.

There are many inconsistent statements throughout the paper regarding the timing of the first wave of spermatogenesis. For example, the authors state that round spermatids can be detected at 21dpp on line 161, but on line 180, say round spermatids can be detected a 19dpp. Not only does this lead to confusion, but such discrepancies undermine the validity of the rest of the paper. A summary graphic displaying key events and their timing in the first wave of spermatogenesis would be instrumental for reader comprehension and could be used by the authors to ensure consistent claims throughout the paper.

Response:

We thank the reviewer for identifying this inconsistency and apologize for the confusion. We confirm that early round spermatids first appear at 19 dpp, as shown in the quantitative data (Figure 1J). This can be detected in squashed spermatocyte preparations, where individual spermatocytes and spermatids can be accurately quantified. The original text contained an imprecise reference to the histological image of 21 dpp (previous line 161), since certain H&E sections did not clearly show all cell types simultaneously. However, we have now revised Figure 1, improving the image quality and adding a zoomed-in panel highlighting early round spermatids. Image for 19 dpp mice in Fig 1D shows early, yet still aflagellated spermatids. The first ciliated spermatocytes and the earliest flagellated spermatids are observed at 20 dpp. This has been clarified in the text.

In addition, we also thank the reviewer for the suggestion of adding a summary graphic, which we agree greatly facilitates reader comprehension. We have added a new schematic summary (Figure 1K) illustrating the key stages and timing of the first spermatogenic wave.

In the proteomics experiments, it is unclear why the authors assume that changes in protein expression are predominantly due to changes within the germ cells in the developing testis. The analysis is on whole testes including both the somatic and germ cells, which makes it possible that protein expression changes in somatic cells drive the results. The authors need to justify why and how the conclusions drawn from this analysis warrant such an assumption.

Response:

We agree with the reviewer that our proteomic analysis was performed on whole testis samples, which contain both germ and somatic cells. Although isolation of pure spermatocyte populations by FACS would provide higher resolution, obtaining sufficient prepubertal material for such analysis would require an extremely large number of animals. To remain compliant with the 3Rs principle for animal experimentation, we therefore used whole-testis samples from three biological replicates per age.

We acknowledge that our assumption-that the main differences arise from germ cells-is a simplification. However, germ cells constitute the vast majority of testicular cells during this developmental window and are the population undergoing major compositional changes between 15 dpp and adulthood. It is therefore reasonable to expect that a substantial fraction of the observed proteomic changes reflects alterations in germ cells. We have clarified this point in the revised text and have added a statement noting that changes in somatic cells could also contribute to the proteomic profiles.

The authors should provide details on how proteins were categorized as being involved in ciliogenesis or flagellogenesis, specifically in the distinction criteria. It is not clear how the categorizations were determined or whether they are valid. Thus, no one can repeat this analysis or perform this analysis on other datasets they might want to compare.

Response:

We thank the reviewer for this opportunity to clarify our approach. The categorization of protein as being involved in ciliogenesis or flagellogenesis was based on their Gene Ontology (GO) cellular component annotations obtained from the PANTHER database (Version 19.0), using the gene IDs of the Differentially Expressed Proteins (DEPs). Specifically, we used the GO terms cilium (GO:0005929) and motile cilium (GO:0031514). Since motile cilium is a subcategory of cilium, proteins annotated only with the general cilium term, but not included under motile cilium, were considered to be associated with primary cilia or with shared structural components common to different types of cilia. These GO terms are represented in the bottom panel of the Figure 6.

This information has been added to the Methods section and referenced in the Results for transparency and reproducibility.

In the pharmacological studies, the authors conclude that the phenotypes they observe (DNA damage and reduced pachytene spermatocytes) are due to loss of or persistence of cilia. This overinterprets the experiment. Chloral hydrate and MLN8237 certainly impact ciliation as claimed, but have additional cellular effects. Thus, it is possible that the observed phenotypes were not a direct result of cilia manipulation. Either additional controls must address this or the conclusions need to be more specific and toned down.

Response:

We thank the reviewer for this fair observation and have taken steps to strengthen and refine our interpretation. In the revised version, we now include data from 1-hour and 24-hour cultures for both control and chloral hydrate (CH)-treated samples (n = 3 biological replicates). The triple immunolabelling with γH2AX, SYCP3, and H1T allows accurate staging of zygotene (H1T⁻), early pachytene (H1T⁻), and late pachytene (H1T⁺) spermatocytes.

The revised Figure 7 now provides a more complete and statistically supported analysis of DNA damage dynamics, confirming that CH-induced deciliation leads to persistent γH2AX signal at 24 hours, indicative of delayed or defective DNA repair progression. We have also toned down our interpretation in the Discussion, acknowledging that CH could affect other cellular pathways.

As mentioned before, the conditional genetic model that we are currently generating will allow us to evaluate the role of cilia in meiotic DNA repair in a more direct and specific way.

Assuming the conclusions of the pharmacological studies hold true with the proper controls, the authors still conflate their findings with meiotic defects. Meiosis is not directly assayed, which makes this conclusion an overstatement of the data. The conclusions need to be rephrased to accurately reflect the data.

Response:

We agree that this aspect required clarification. As noted above, we have refined both the Results and Discussion sections to make clear that our assays specifically targeted meiotic spermatocytes.

We now present data for meiotic stages at zygotene, early pachytene and late pachytene. This is demonstrated with the labelling for SYCP3 and H1T, both specific marker for meiosis that are not detectable in non meiotic cells. We believe that this is indeed a way to assay the meiotic cells, however, we have specified now in the text that we are analysing potential defects in meiosis progression. We are sorry if this was not properly explained in the original manuscript: it is now rephrased in the new version both in the results and discussion section.

It is not clear why the authors chose not to use widely accepted assays of Hedgehog signaling. Traditionally, pathway activation is measured by transcriptional output, not GLI protein expression because transcription factor expression does not necessarily reflect transcription levels of target genes.

Response:

We agree with the reviewer that measuring mRNA levels of Hedgehog pathway target genes, typically GLI1 and PTCH1, is the most common method for measuring pathway activation, and is widely accepted by researchers in the field. However, the methods we use in this manuscript (GLI1 and GLI3 immunoblots) are also quite common and widely accepted:

Regarding GLI1 immunoblot, many articles have used this method to monitor Hedgehog signaling, since GLI1 protein levels have repeatedly been shown to also go up upon pathway activation, and down upon pathway inhibition, mirroring the behavior of GLI1 mRNA. Here are a few publications that exemplify this point:

• Banday et al. 2025 Nat Commun. DOI: 10.1038/s41467-025-56632-0 (PMID: 39894896)
• Shi et al 2022 JCI Insight DOI: 10.1172/jci.insight.149626 (PMID: 35041619)
• Deng et al. 2019 eLife, DOI: 10.7554/eLife.50208 (PMID: 31482846)
• Zhu et al. 2019 Nat Commun, DOI: 10.1038/s41467-019-10739-3 (PMID: 31253779)
• Caparros-Martin et al 2013 Hum Mol Genet, DOI: 10.1093/hmg/dds409 (PMID: 23026747) *note: due to manuscript-length limitations, not all cited references can be included in the text; they are listed here to substantiate our response.

As for GLI3 immunoblot, Hedgehog pathway activation is well known to inhibit GLI3 proteolytic processing from its full length form (GLI3-FL) to its transcriptional repressor (GLI3-R), and such processing is also commonly used to monitor Hedgehog signal transduction, of which the following are but a few examples:

• Pedraza et al 2025 eLife, DOI: 10.7554/eLife.100328 (PMID: 40956303)
• Somatilaka et al 2020 Dev Cell, DOI: 10.1016/j.devcel.2020.06.034 (PMID: 32702291)
• Infante et al 2018, Nat Commun, DOI: 10.1038/s41467-018-03339-0 (PMID: 29515120)
• Wang et al 2017 Dev Biol DOI: 10.1016/j.ydbio.2017.08.003 (PMID: 28800946)
• Singh et al 2015 J Biol Chem DOI: 10.1074/jbc.M115.665810 (PMID: 26451044) *note: due to manuscript-length limitations, not all cited references can be included in the text; they are listed here to substantiate our response.

In summary, we think that we have used two well established markers to look at Hedgehog signaling (three, if we include the immunofluorescence analysis of SMO, which we could not detect in meiotic cilia).

These Hh pathway analyses did not provide any convincing evidence that the prepubertal cilia we describe here are actively involved in this pathway, even though Hh signaling is cilia-dependent and is known to be active in the male germline (Sahin et al 2014 Andrology PMID: 24574096; Mäkelä et al 2011 Reproduction PMID: 21893610; Bitgood et al 1996 Curr Biol. PMID: 8805249).

That said, we fully agree that our current analyses do not allow us to draw definitive conclusions regarding Hedgehog pathway activity in meiotic cilia, and we now state this explicitly in the revised Discussion.

Also in the Hedgehog pathway experiment, it is confusing that the authors report no detection of SMO yet detect little to no expression of GLIR in their western blot. Undetectable SMO indicates Hedgehog signaling is inactive, which results in high levels of GLIR. The impact of this is that it is not clear what is going on with Hh signaling in this system.

Response:

It is true that, when Hh signaling is inactive (and hence SMO not ciliary), the GLI3FL/GLI3R ratio tends to be low.

Although our data in prepuberal mouse testes show a strong reduction in total GLI3 protein levels (GLI3FL+GLI3R) as these mice grow older, this downregulation of total GLI3 occurs without any major changes in the GLI3FL/GLI3R ratio, which is only modestly affected (suppl. Figure 6).

Hence, since it is the ratio that correlates with Hh signaling rather than total levels, we do not think that the GLI3R reduction we see is incompatible with our non-detection of SMO in cilia: it seems more likely that overall GLI3 expression is being downregulated in developing testes via a Hh-independent mechanism.

Also potentially relevant here is the fact that some cell types depend more on GLI2 than on GLI3 for Hh signaling. For instance, in mouse embryos, Hh-mediated neural tube patterning relies more heavily on GLI2 processing into a transcriptional activator than on the inhibition of GLI3 processing into a repressor. In contrast, the opposite is true during Hh-mediated limb bud patterning (Nieuwenhuis and Hui 2005 Clin Genet. PMID: 15691355). We have not looked at GLI2, but it is conceivable that it could play a bigger role than GLI3 in our model.

Moreover, several forms of GLI-independent non-canonical Hh signaling have been described, and they could potentially play a role in our model, too (Robbins et al 2012 Sci Signal. PMID: 23074268).

We have revised the discussion to clarify some of these points.

All in all, we agree that our findings regarding Hh signaling are not conclusive, but we still think they add important pieces to the puzzle that will help guide future studies.

There are multiple instances where it is not clear whether the authors performed statistical analysis on their data, specifically when comparing the percent composition of a population. The authors need to include appropriate statistical tests to make claims regarding this data. While the authors state some impressive sample sizes, once evaluated in individual categories (eg specific cell type and age) the sample sizes of evaluated cilia are as low as 15, which is likely underpowered. The authors need to state the n for each analysis in the figures or legends.

We thank the reviewer for highlighting this important issue. We have now included the sample size (n) for every analysis directly in the figure legends. Although this adds length, it improves transparency and reproducibility.

Regarding the doubts of Ref#3 about the different sample sizes, the number of spermatocytes quantified in each stage is in agreement with their distribution in meiosis (example, pachytene lasts for 10 days this stage is widely represented in the preparations, while its is much difficult to quantify metaphases I that are less present because the stage itself lasts for less than 24hours). Taking this into account, we ensured that all analyses remain statistically valid and representative, applying the appropriate statistical tests for each dataset. These details are now clearly indicated in the revised figures and legends.

Minor concerns:

1. The phrase "lactating male" is used throughout the paper and is not correct. We assume this term to mean male pups that have yet to be weaned from their lactating mother, but "lactating male" suggests a rare disorder requiring medical intervention. Perhaps "pre-weaning males" is what the authors meant.

Response:

We thank the reviewer for noticing this terminology error. The expression has been corrected to "pre-weaning males" throughout the manuscript.

The convention used to label the figures in this paper is confusing and difficult to read as there are multiple panels with the same letter in the same figure (albeit distinct sections). Labeling panels in the standard A-Z format is preferred. "Panel Z" is easier to identify than "panel III-E".

Response:

We thank the reviewer for this suggestion. All figures have been relabelled using the standard A-Z panel format, ensuring consistency and easier readability across the manuscript.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary:

In "The dynamics of ciliogenesis in prepubertal mouse meiosis reveals new clues about testicular development" Pérez-Moreno, et al. explore primary cilia in prepubertal mouse spermatocytes. Using a combination of microscopy, proteomics, and pharmacological perturbations, the authors carefully characterize prepubertal spermatocyte cilia, providing foundational work regarding meiotic cilia in the developing mammalian testis.

Major concerns:

1. The authors provide evidence consistent with cilia not being present in a larger percentage of spermatocytes or in other cells in the testis. The combination of electron microscopy and acetylated tubulin antibody staining establishes the presence of cilia; however, proving a negative is challenging. While acetylated tubulin is certainly a common marker of cilia, it is not in some cilia such as those in neurons. The authors should use at least one additional cilia marker to better support their claim of cilia being absent.

2. The conclusion that IFT88 localizes to centrosomes is premature as key controls for the IFT88 antibody staining are lacking. Centrosomes are notoriously "sticky", often sowing non-specific antibody staining. The authors must include controls to demonstrate the specificity of the staining they observe such as staining in a genetic mutant or an antigen competition assay.

3. There are many inconsistent statements throughout the paper regarding the timing of the first wave of spermatogenesis. For example, the authors state that round spermatids can be detected at 21dpp on line 161, but on line 180, say round spermatids can be detected a 19dpp. Not only does this lead to confusion, but such discrepancies undermine the validity of the rest of the paper. A summary graphic displaying key events and their timing in the first wave of spermatogenesis would be instrumental for reader comprehension and could be used by the authors to ensure consistent claims throughout the paper.

4. In the proteomics experiments, it is unclear why the authors assume that changes in protein expression are predominantly due to changes within the germ cells in the developing testis. The analysis is on whole testes including both the somatic and germ cells, which makes it possible that protein expression changes in somatic cells drive the results. The authors need to justify why and how the conclusions drawn from this analysis warrant such an assumption.

5. The authors should provide details on how proteins were categorized as being involved in ciliogenesis or flagellogenesis, specifically in the distinction criteria. It is not clear how the categorizations were determined or whether they are valid. Thus, no one can repeat this analysis or perform this analysis on other datasets they might want to compare.

6. In the pharmacological studies, the authors conclude that the phenotypes they observe (DNA damage and reduced pachytene spermatocytes) are due to loss of or persistence of cilia. This overinterprets the experiment. Chloral hydrate and MLN8237 certainly impact ciliation as claimed, but have additional cellular effects. Thus, it is possible that the observed phenotypes were not a direct result of cilia manipulation. Either additional controls must address this or the conclusions need to be more specific and toned down.

7. Assuming the conclusions of the pharmacological studies hold true with the proper controls, the authors still conflate their findings with meiotic defects. Meiosis is not directly assayed, which makes this conclusion an overstatement of the data. The conclusions need to be rephrased to accurately reflect the data.

8. It is not clear why the authors chose not to use widely accepted assays of Hedgehog signaling. Traditionally, pathway activation is measured by transcriptional output, not GLI protein expression because transcription factor expression does not necessarily reflect transcription levels of target genes.

9. Also in the Hedgehog pathway experiment, it is confusing that the authors report no detection of SMO yet detect little to no expression of GLIR in their western blot. Undetectable SMO indicates Hedgehog signaling is inactive, which results in high levels of GLIR. The impact of this is that it is not clear what is going on with Hh signaling in this system.

10. There are multiple instances where it is not clear whether the authors performed statistical analysis on their data, specifically when comparing the percent composition of a population. The authors need to include appropriate statistical tests to make claims regarding this data. While the authors state some impressive sample sizes, once evaluated in individual categories (eg specific cell type and age) the sample sizes of evaluated cilia are as low as 15, which is likely underpowered. The authors need to state the n for each analysis in the figures or legends.

Minor concerns:

1. The phrase "lactating male" is used throughout the paper and is not correct. We assume this term to mean male pups that have yet to be weaned from their lactating mother, but "lactating male" suggests a rare disorder requiring medical intervention. Perhaps "pre-weaning males" is what the authors meant.

2. The convention used to label the figures in this paper is confusing and difficult to read as there are multiple panels with the same letter in the same figure (albeit distinct sections). Labeling panels in the standard A-Z format is preferred. "Panel Z" is easier to identify than "panel III-E".

Significance

Overall, this is a well-done body of work that deserves recognition for the novel and implicative discoveries it presents. Assuming the conclusions hold true following appropriate statistical analysis and rephrasing, this paper would report the first documented evidence of meiotic cilia in the developing mammalian testis with sufficient rigor to become the foundational work on this topic.

This paper will be of interest to communities focused on germ cell development, cilia, and Hedgehog signaling. It may prompt a new perspective on Desert Hedgehog signaling as it pertains to spermatogenesis. Further, this work will be of interest to those studying male fertility, as it highlights the potential role of cilia in spermatogenesis.

Further, the proteomic analysis presented has the potential to invoke hypotheses and experimentation investigating the role of several proteins with previously uncharacterized roles in ciliogenesis, flagellogenesis, and/or spermatogenesis. The finding that the onset of ciliogenesis and flagellogenesis appear to be temporally linked has the potential to prompt research regarding shared molecular mechanisms dictating axonemal formation. We believe this paper has the potential to have an impact in its respective field, underscored by the exquisite microscopy and detailed characterization of meiotic cilia.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

This study focuses on the dynamic changes of ciliogenesis during meiosis in prepubertal mice. It was found that primary cilia are not an intrinsic feature of the first wave of meiosis (initiating at 8 dpp); instead, they begin to polymerize at 20 dpp (after the completion of the first wave of meiosis) and are present in all stages of prophase I. Moreover, prepubertal cilia (with an average length of 21.96 μm) are significantly longer than adult cilia (10 μm). The emergence of cilia coincides temporally with flagellogenesis, suggesting a regulatory association in the formation of axonemes between the two. Functional experiments showed that disruption of cilia by chloral hydrate (CH) delays DNA repair, while the AURKA inhibitor (MLN8237) delays cilia disassembly, and centrosome migration and cilia depolymerization are mutually exclusive events. These findings represent the first detailed description of the spatiotemporal regulation and potential roles of cilia during early testicular maturation in mice. The discovery of this phenomenon is interesting; however, there are certain limitations in functional research.

Major points:

1. The prepubertal cilia in spermatocytes discovered by the authors lack specific genetic ablation to block their formation, making it impossible to evaluate whether such cilia truly have functions. Because neither in the first wave of spermatogenesis nor in adult spermatogenesis does this type of cilium seem to be essential. In addition, the authors also imply that the formation of such cilia appears to be synchronized with the formation of sperm flagella. This suggests that the production of such cilia may merely be transient protein expression noise rather than a functionally meaningful cellular structure.

2. The high expression of axoneme assembly regulators such as TRiC complex and IFT proteins identified by proteomic analysis is not particularly significant. This time point is precisely the critical period for spermatids to assemble flagella, and TRiC, as a newly discovered component of flagellar axonemes, is reasonably highly expressed at this time. No intrinsic connection with the argument of this paper is observed. In fact, this testicular proteomics has little significance.

Significance

Strengths: The discovery of a very interesting time window for ciliary growth in spermatocytes.

Weaknesses: Insufficient analysis of the function of such cilia.

Readers: Developmental biologists, reproductive biologists

My expertise: Spermatogenesis, genetics

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

In this manuscript by Perez-Moreno et al., titled "The dynamics of ciliogenesis in prepubertal mouse meiosis reveal new clues about testicular maturation during puberty", the authors characterize the development of primary cilia during meiosis in juvenile male mice. The authors catalog a variety of testicular changes that occur as juvenile mice age, such as changes in testis weight and germ cell-type composition. They next show that meiotic prophase cells initially lack cilia, and ciliated meiotic prophase cells are detected after 20 days postpartum, coinciding with the time when post-meiotic spermatids within the developing testes acquire flagella. They describe that germ cells in juvenile mice harbor cilia at all substages of meiotic prophase, in contrast to adults where only zygotene stage meiotic cells harbor cilia. The authors also document that cilia in juvenile mice are longer than those in adults. They characterize cilia composition and structure by immunofluorescence and EM, highlighting that cilia polymerization may initially begin inside the cell, followed by extension beyond the cell membrane. Additionally, they demonstrate ciliated cells can be detected in adult human testes. The authors next perform proteomic analyses of whole testes from juvenile mice at multiple ages, which may not provide direct information about the extremely small numbers of ciliated meiotic cells in the testis, and is lacking follow up experiments, but does serve as a valuable resource for the community. Finally, the authors use a seminiferous tubule culturing system to show that chemical inhibition of Aurora kinase A likely inhibits cilia depolymerization upon meiotic prophase I exit and leads to an accumulation of metaphase-like cells harboring cilia. They also assess meiotic recombination progression using their culturing system, but this is less convincing.

Few suggestions/comments are listed below:

Major comments

1. There are a few issues with the experimental set up for assessing the effects of cilia depolymerization on DNA repair (Figure 7-II). First, how were mid pachytene cells identified and differentiated from early pachytene cells (which would have higher levels of gH2AX) in this experiment? I suggest either using H1t staining (to differentiate early/mid vs late pachytene) or the extent of sex chromosome synapsis. This would ensure that the authors are comparing similarly staged cells in control and treated samples. Second, what were the gH2AX levels at the starting point of this experiment? A more convincing set up would be if the authors measure gH2AX immediately after culturing in early and late cells (early would have higher gH2AX, late would have lower gH2AX), and then again after 24hrs in late cells (upon repair disruption the sampled late cells would have high gH2AX). This would allow them to compare the decline in gH2AX (i.e., repair progression) in control vs treated samples. Also, it would be informative to know the starting gH2AX levels in ciliated vs non-ciliated cells as they may vary.

2. The authors analyze meiotic progression in cells cultured with/without AURKA inhibition in Figure 8-III and conclude that the distribution of prophase I cells does not change upon treatment. Is Figure 8-III A and B the same data? The legend text is incorrect, so it's hard to follow. Figure 8-III A shows a depletion of EdU-labelled pachytene cells upon treatment. Moreover, the conclusion that a higher proportion of ciliated zygotene cells upon treatment (Figure 8-II C) suggests that AURKA inhibition delays cilia depolymerization (page 13 line 444) does not make sense to me.

3. How do the authors know that there is a monopolar spindle in Figure 8-IV treated samples? Perhaps the authors can use a different Tubulin antibody (that does not detect only acetylated Tubulin) to show that there is a monopolar spindle.

4. The authors state in the abstract that they provide evidence suggesting that centrosome migration and cilia depolymerization are mutually exclusive events during meiosis. This is not convincing with the data present in the current manuscript. I suggest amending this statement in the abstract.

Minor comments

1. The presence of cilia in all stages of meiotic prophase I in juvenile mice is intriguing. Why is the cellular distribution and length of cilia different in prepubertal mice compared to adults (where shorter cilia are present only in zygotene cells)? What is the relevance of these developmental differences? Do cilia serve prophase I functions in juvenile mice (in leptotene, pachytene etc.) that are perhaps absent in adults?

Related to the above point, what is the relevance of the absence of cilia during the first meiotic wave? If cilia serve a critical function during prophase I (for instance, facilitating DSB repair), does the lack of cilia during the first wave imply differing cilia (and repair) requirements during the first vs latter spermatogenesis waves?

In my opinion, these would be interesting points to discuss in the discussion section.

1. The authors state on page 9 lines 286-288 that the presence of cytoplasmic continuity via intercellular bridges (between developmentally synchronous spermatocytes) hints towards a mechanism that links cilia and flagella formation. Please clarify this statement. While the correlation between the timing of appearance of cilia and flagella in cells that are located within the same segment of the seminiferous tubule may be hinting towards some shared regulation, how would cytoplasmic continuity participate in this regulation? Especially since the cytoplasmic continuity is not between the developmentally distinct cells acquiring the cilia and flagella?

2. Individual germ cells in H&E-stained testis sections in Figure 1-II are difficult to see. I suggest adding zoomed-in images where spermatocytes/round spermatids/elongated spermatids are clearly distinguishable.

3. In Figure 2-II B, the authors document that most ciliated spermatocytes in juvenile mice are pachytene. Is this because most meiotic cells are pachytene? Please clarify. If the data are available (perhaps could be adapted from Figure 1-III), it would be informative to see a graph representing what proportions of each meiotic prophase substages have cilia.

4. I suggest annotating the EM images in Sup Figure 2 and 3 to make it easier to interpret.

5. The authors claim that the ratio between GLI3-FL and GLI3-R is stable across their analyzed developmental window in whole testis immunoblots shown in Sup Figure 5. Quantifying the bands and normalizing to the loading control would help strengthen this claim as it hard to interpret the immunoblot in its current form.

6. There are a few typos throughout the manuscript. Some examples: page 5 line 172, Figure 3-I legend text, Sup Figure 5-II callouts, Figure 8-III legend, page 15 line 508, page 17 line 580, page 18 line 611.

Significance

This work provides new information about an important but poorly understood cellular structure present in meiotic cells, the primary cilium. More generally, this work expands on our understanding of testis development in juvenile mice. The microscopy images presented here are beautiful. The work is mostly descriptive but lays the groundwork for future investigations. I believe that this study would of interest to the germ cell, meiosis, and spermatogenesis communities, and with a few modifications, is suitable for publication.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1 (R1)

R1 General statement: Here, Escalera-Maurer and colleagues, present an up-to-date distribution of homologues of Hok toxic proteins belonging to the well-annotated, but otherwise functionally obscure, hok/Sok type I toxin-antitoxin system, across the RefSeq database. Although such computational analyses have been done in the past, the authors here find many more hok homologs than described before, and they categorise their distribution based on whether they are encoded on chromosomes, plasmids, or (pro)phages. These computational analyses are in general tricky with T1TAs, as their toxins are quite short (~50 amino acids, as is the case for Hok), which is why the authors here used three separate approaches to expand their search (nucleotide-level BLAST, protein-homology, or both combined with Infernal). The authors cluster the Hok homologues they find based on a 60% sequence identity cut-off (expanding the known clusters in the process), and proceeded to test 31 candidates belonging to 15 sequence-clusters for their toxicity in Salmonella Typhimurium LT2, showing that 30/31 were toxic upon induction. An interesting finding from their endeavours is that hok/Sok homologues are enriched within prophages and large plasmids, but are not enriched near bacterial anti-phage defense systems (in contrast to the SymE/SymR T1TA). The findings suggest that hok/Sok are indeed sometimes linked to phage and plasmid biology, although they might not be antiphage defenses per se (they have been clearly shown in the past to be addiction modules, and this is still clearly true).

Authors' answer to R1 General statement: __We do not state here that hok/Sok are not anti-phage defense systems, but we simply observe that they do not cluster with anti-phage defense systems. We have also observed (unpublished data) that known defense systems do not systematically cluster together with other defense systems. Therefore, strong association with other defense systems would have been a strong indication of their function in phage defense but the fact that we did not observe any association with defense systems does not exclude they are involved in phage defense. __

R1_C1: My expertise lies towards the experimental side of the authors' work, I thus cannot comment on the accuracy/robustness of the computational analyses performed here. The authors do a fine job in clearly stating their findings overall; I could follow most of the conclusions, and I deemed that most of them were supported by their work. Additionally, I find that this paper is a missed opportunity to uncover even more novel biology connected to the interesting hok/Sok T1TAs. The paper does not provide a new framework to think about what is the function of the chromosomal/prophage hok/Sok T1TA systems, although I realize that this is very difficult to accomplish, especially when considering that hok/Sok systems have been around in the literature for almost 40 years.

Authors' answer to R1_C1: We agree with the reviewer, as we indeed performed this analysis having in mind to clarify the role of hok/Sok systems. However, we still believe that our strong survey of Hok loci put in light their enrichment in various mobile genetic elements, such as prophage and large conjugative plasmids, which is indubitably linked to their function. In addition, our study will guide future experimental efforts in uncovering the function of these systems, for example by helping researchers to select relevant homologs to test for a specific function.__ __

R1_C2: My major comment is in regard to the Hok toxicity assays (Fig. 2). The authors state in the discussion that "Hok peptides originating from chromosomes are as toxic as those from plasmids", but I believe that the way that they tested their constructs might not have allowed them to see toxicity differences between the two groups. Specifically, using the multi-copy plasmid pAZ3 (pBR322 origin of replication; ~15-20 plasmid copies per chromosome) to induce the different Hok toxin homologues in Salmonella Typhimurium LT2 with arabinose might have masked toxicity differences that would otherwise be apparent on the chromosomal expression-level.

Some of the authors themselves have previously used the FASTBAC-Seq method to study the Hok homologue from plasmid R1, a useful technique during which a toxin is integrated in the chromosome, in order to study their toxicity under natural levels of expression. I believe that an ideal scenario would be to apply FASTBAC-seq to some of the 31 Hok homologues described here (e.g., a subset of plasmidic vs chromosomal Hok homologues) to shed light on potential toxicity differences between the Hok clusters. This would increase the value of the presented study.

Alternatively, the authors could employ an L-arabinose concentration gradient to titrate the expression levels of the Hok toxins in order to potentially see different toxicity levels from the different homologues. However, this is not going to work in the system as they are using it now for two reasons:

1. a) the S. Typhimurium LT2 (STm) used here has its arabinose utilization operon intact (araBAD), which means that Salmonella can catabolize arabinose to use it as a carbon source. This catabolization process interferes with the arabinose induction (i.e., Salmonella eats arabinose instead of using it as the Hok inducer). To ameliorate this, the authors could delete the araBAD operon in STm, rendering STm incapable of catabolizing arabinose, and repeat the experiments in that strain. Or use E. coli BW25113 as the expression host, which already has the araBAD operon deleted (it is not clear to me why the different Hok homologues would not be toxic in E. coli, as the different Hok homologues are widely diverse in sequence, as the authors found here).
2. b) Even with the araBAD operon deleted, the arabinose induction would be bimodally on or off in the population, due to the bimodal expression of the arabinose transporter (AraE; see Khlebnikov et al., 2002). This would again not allow for titratable arabinose-inducible expression from different concentrations of arabinose. The solution for this would be to co-express a separate plasmid with araE, which would render every cell the same in regards to arabinose permeability, and thus the system would be titratable (as explained in Khlebnikov et al., 2002). Therefore, if the authors would be interested to go towards this route, they would have to first delete the araBAD from STm, then transform STm with an araE plasmid, and redo the experiments. In addition, I would propose to the authors to use the drop plate method (agar plate-based), which is more sensitive compared to the liquid assays employed here.

Having said all that, I understand that all this experimental work would be strenuous and time-consuming, and although I would like to see it happen, this is not my paper. I would be content therefore if the authors toned down the claim that plasmidic vs chromosomal Hok homologues have the same toxicity, and discuss that chromosomal levels of toxicity are an important caveat that has not been explored here.

__Authors' answer to R1_C2: __ We thank the reviewer for the detailed suggestion on how to better assess toxicity differences by using an araBAD deletion mutant overexpressing araE. We repeated the arabinose induction assays using drop assays and strain BW25223 with plasmid pJAT13araE and our pAZ3 based plasmid carrying Hok CDS homologs. However, we obtained similar data, not being able to distinguish between the toxicity of chromosomal versus plasmidic CDS, even using different concentration of Arabinose. This is probably because low concentration of the Hok protein are sufficient for activity, but here we are bypassing all post-transcriptional silencing by the native Hok mRNAs by expressing directly the protein, and we are using a multicopy plasmid. We now included 0.01% arabinose induction drop assays in the manuscript as the data obtained with other arabinose concentration did not provide new information. In any case, we are still not accessing the native expression levels for the following reasons 1/ chromosomal level of toxicity were not explored here and 2/ only the toxicity of the coding sequence but not the full mRNA was tested. Indeed, we do not know the exact sequence of the hok homolog mRNAs and this is beyond the scope of the study. These remarks were clearly added in the discussion.

We agree that the sentence "Hok peptides originating from chromosomes are as toxic as those from plasmids" was too strong and we have added the caveats of our experimental design in the discussion. While we indeed did not compare the toxicity of the peptides, we still showed that chromosomal Hok can be toxic upon overexpression, which would not be the case if the sequences were degenerated.

The reviewer also suggests the use of the FASTBAC-Seq method, that we previously used to study Hok from the R1 plasmid, which is a method to study toxic type I toxins at the native expression level. While FASTBAC-Seq identifies loss-of-function mutants of the systems, it does not allow to determine a difference of toxicity between systems per se. In addition, FASTBAC-Seq was always done in the context of the full mRNA, not only the coding sequence, and these sequences are presently unknown for most homologs.

Other comments:

__R1_C3: __a) There is barely any discussion of the Sok component (RNA antitoxin) of the homologues; why is that? Could you please discuss Sok differences across the homologues, or at least explain why this is not discussed at all in the paper (e.g., in the discussion)?

Authors' answer to R1_C3: __It is not trivial to identify the Sok RNA sequence, this is why it was not done in this study, a paragraph was added in the discussion explaining this. __

__R1_C4: __b) In the results section, the Hok clusters are referred to as 62 in number ("Because Hok sequences were too short and variable to construct a meaningful phylogenetic tree, we clustered the Hok sequences with a 60% identity threshold and obtained 62 clusters"), but then in the discussion section, the cluster number becomes 74 ("We highlighted the high sequence variability within Hok peptides by obtaining a total of 74 clusters with 60% identity (Fig. S7)."). Which one is the right number, and why is there a discrepancy?

Authors' answer to R1_C4: We apologize for the discrepancy between the number. The first number corresponded to the Hok hits from the refSeq and we then added the Hok hits from the plasmid and virus databases (performed later in the manuscript). We clarified this information both in the result and discussion texts (61 clusters from RefSeq and 79 in total, 74 was a typo).__ __

__R1 Significance: __The most well-clarified aspect of the paper presented here is the distribution of Hok homologues, with the novel aspect of the location in which the hok/Sok T1TAs reside (i.e., chromosome, plasmid, or phage). There is room for the molecular genetics part to be developed further, as I discussed earlier, however this study is the most up-to-date characterization of the diversity of Hok homologues, and will be of interest to the T1TA and the general toxin-antitoxin field.

__Reviewer #2 (R2) __

R2 General statement: The authors examined how the Hok toxins are spread across bacterial genomes. The manuscript including its figures is hard to read and understand. I commented figure 1 in details, but similar comments apply to the other figures. Overall, the data lack clarity and precision. Finding information about sequences, clusters in the supplementary materials was not easy. The manuscript should be thoroughly revised. In addition, I believe that other aspects should be developed to expand the interest of the study, such as the co-occurrence of multiple systems in chromosomes, on plasmids and whether they are able to crosstalk. This might provide some evolutionary insights into the biology of these toxins.

__Authors' answer to R2 General statement: __We designed all figures according to established standards for scientific data visualization, although we recognize that different presentations may work better for different audiences. In our detailed response to Figure 1A, we explain how UpSet plots are constructed and interpreted, which we hope clarifies the visualization approach for the full dataset. We are open to discussing specific improvements if the reviewer has suggestions for enhanced clarity. To address concerns about accessibility, we want to clarify that all sequences are compiled in Table S1 with their clus100 identifiers, making them easy to locate. We are open to reorganizing supplementary materials if a different structure would be more user-friendly. Finally, we agree that an extensive analysis of co-occurrences and crosstalks would be valuable. However, predicting crosstalk bioinformatically for all genomes presents challenges, as it would require predicting RNA:RNA interactions between hok mRNA and Sok sequences, which are currently unknown. Given these limitations, this analysis was beyond the scope of the current study.

R2_C1: The introduction lacks information regarding the Hok protein (size, structure prediction, localization) as well as a bit of explanation about the reason of looking at these toxins. The description of the potential roles should be a bit expanded.

Authors' answer to R2_C1: Following the comment from the reviewer, we have provided additional information about Hok in the introduction.

__R2_C2: __When the authors talk about 'loci', they mean genes encoding Hok homologs if I understand correctly. They did not look for the Sok sequences (hok-sok loci).

__Author's answer to R2_C2: __Indeed, we did not look for the Sok sequences and we are only describing Hok homologs loci, that could either encode or lack a Sok homolog.

__R2_C3: __It is not clear what the authors did with the sequences for which they could not detect a start codon and a SD (although it is unusual to refer to SD in the context of protein sequence)

Authors' answer to R2_C3: The peptides were annotated by extending the initial hit until the first start codon. Therefore, all annotated peptides have a start codon. Shine-Dalgarno sequences were annotated when confidently predicted, to provide additional information. Sequences were not excluded based on the presence or absence of the SD.

__R2_C4: __Figure 1A is not clear. The total of the bars equal 32,532 which is the number of 'loci' detected by the combination of the different methods. However, it is not clear to me how many are redundant. For instance, I suppose that all the 8483 sequences that were retrieved using blastn and Infernal were retrieved using MMseqs2, blastn and Infernal. So, what is the actual number of sequences that were found? When the authors talk about 1264 distinct peptides, what do they mean? What are the numbers on the X axis (18209, 2260, 27728)?

Author's answer to R2_C4: Figure A1 is a very typical "UpSet" plot, as indicated in the legend (A. Lex, N. Gehlenborg, H. Strobelt, R. Vuillemot and H. Pfister, "UpSet: Visualization of Intersecting Sets," in IEEE Transactions on Visualization and Computer Graphics, vol. 20, no. 12, pp. 1983-1992, 31 Dec. 2014, doi: 10.1109/TVCG.2014.2346248). Those plots are a data visualization method for showing data with more than two intersecting sets. The Hok sequence hits were obtained by 3 different methods stated on the rows (MMseqs2, blastn and Infernal, therefore the number 18209 is the number of hits by the MMseqs2, 22680 the number of hits by blastn and 27728 the number of hits by Infernal). The columns show the intersections between these three sets. For example, the mentioned 8483 sequences (second column) were only found by blastn and Infernal but not by MMseqs2. The actual total number of sequences found is indeed 32 532. The 1264 distinct peptides are peptides with different sequences. After removing false positives, degenerated sequences and small peptides, we obtained 1264 unique Hok sequences that are found in the 32532 bacterial loci.

__R2_C5: __About Infernal: first the authors are stating that only 8% of the sequences are lost when not considering the mRNA structure - which they seem to consider as negligeable. Then in the next section, they state that Infernal is the best tool at identifying clusters that are not detected otherwise. Seems a bit contradictory.

__Authors' answer to R2_C5: __We appreciate the reviewer pointing out this apparent contradiction, we have clarified this part in the revised manuscript. Infernal uses both sequence and structure information simultaneously for homology detection. While only 8% of Infernal's hits are detected uniquely when structural information was considered, these sequences account for 9 additional clusters with notably high sequence diversity, which would otherwise have been undetected. Therefore, we believe that Infernal is the best tool to capture novel cluster diversity.

__R2_C6: __Cluster determination. The threshold was put at 60% identity. What is the rationale for the 60% identity? Given that the Hok sequences (like toxins and antitoxins from TA systems in general) are highly variable, this leads to a high number of clusters. I'm not sure of the relevance of these clusters. Are there any other criteria to define clusters?

Authors' answer to R2_C6: We selected 60% identity as a balance between capturing sequence diversity and generating interpretable results. We also tested 70, 80 and 90% and obtained 128, 221, 377 clusters, respectively, which would be too many for a meaningful visualization and interpretation. The best clustering method would be constructing a phylogenetic tree. However, as explained in the discussion, because the high sequence diversity prevented the construction of a reliable phylogenetic tree, clustering was used as an alternative strategy to identify and interpret patterns of sequence variability.

__R2_C7: __The authors claim that most of the Hok diversity is found on chromosomes. However, the number of chromosomal Hok is higher than that located on plasmids, which might be related to the different sizes of the different replicons ie, chromosomes being larger than plasmids. Is there a way to normalize by determining the density per size?

Authors' answer to R2_C7: We do not claim that chromosomes contain most of Hok diversity, as this would be indeed influenced by biases in the databases. We are just describing that we found most of the diversity in chromosomes, but we cannot conclude whether this is a true representation of the frequencies in nature.__ __

R2_C8: '46 of the 62 clusters contained 10 or less distinct sequences and might be in the process of degenerating'. The authors also linked this with SD detection. Please explain. From what was indicated earlier, I understand that sequences with premature stop codons or short sequences (Authors' answer to R2_C8: We did not remove sequences for which we could not predict the SD. Indeed, lacking SD is a sign that the hok mRNA might not be able to play its biological role and would be indicative that the sequences have degenerated. To evaluate this hypothesis, we experimentally tested 5 sequences without a predicted SD and two of those were not toxic (see Table S2). In order to assess if the low abundant clusters contained degenerated sequences we experimentally tested representatives from some of the clusters with only one Hok CDS and found most of them to be toxic.

R2_C9: 'Only 7.3% of the unique sequences were found on both plasmids and chromosomes'. From this observation, the authors conclude that 'there is little stable transfer from chromosomes to plasmids or vice-versa'. I don't understand what this means. Do they mean identical sequences? The fact that sequences differ from chromosomes to plasmids does not rule out 'stable transfer'. What do they actually mean by stable transfer? Once the gene is horizontally transferred, it is fixed and vertically transmitted? Same comments apply to the inter-genera horizontal transfer by plasmids.

__Authors' answer to R2_C9: __Due to the impossibility of constructing a reliable phylogenetic tree, we used identity of sequences across different localizations or genera as our marker for recent, stable transfer events. We define stable transfer as the persistence of sequences in an unchanged form following horizontal transfer; long enough to be detected in current databases. Our approach likely underestimates total transfer events, as sequences accumulating mutations after transfer would not be captured. We would expect to observe numerous identical sequences across plasmids and chromosomes if frequent exchange were occurring, unless rapid mutation after the transfer prevented their detection as identical sequences. We have added a sentence to clarify this in the manuscript and removed the term stable transfer.

__R2_C10: __I don't understand the next section about 'family'. What do the authors mean about 'family'? Genera? The same apply to the next section about the Y to C recoding. Did the authors do point mutations in the conserved amino acids/codons to test whether they are important for toxicity? Some Hok variants lacks some of the conserved amino acids and are toxic (under overexpression conditions in Salmonella). What about T18, C31 and E42?

Authors' answer to R2_C10: Families (Enterobacteriaceae, Vibrionaceae etc... ) and genera (Escherichia, Salmonella etc...) refer to the taxonomic categories. Following the reviewer comment, we experimentally assessed the toxicity of Hok from R1 plasmid after mutating the conserved amino acids to alanine residues. All the mutants were found to be toxic under our expression conditions.

__R2_C11: __The prevalence of Hok in chromosomes or on plasmids might depend on various confounding parameters, such as the size, number of sequences available among others. The authors should find methods to correct for all that.

Authors' answer to R2_C11: Normalization would indeed be needed if we were comparing the prevalence on chromosomes vs the prevalence on plasmids. Here, we do not claim that Hok homologs are more prevalent in plasmid or chromosomes and only describe where we found them.

__R2_C12: __Link with defense systems. The threshold was set at 20 kb. Why this threshold?

Authors' answer to R2_C12: The size of defense islands in a previous report was approximately 40 kb, by setting up a 20 kb threshold we searched for defense systems in a region of 40 kb adjacent to each of the homologs (https://doi.org/10.1126/science.aar4120). If the specific homolog was part of a defense island we would expect that it is less than 20 kb apart from any defense system.

__R2 Significance: __The paper in its current state appears to serve the role of a data repository rather than a thorough and original analysis. It requires extensive revisions before it can be of interest to experts in the toxin-antitoxin field.

__ ____Reviewer #3 (R3): __

R3 General statement: In the manuscript, "The Hok bacterial toxin: diversity, toxicity, distribution and genomic localization," by Escalera-Maurer et al., investigate the distribution of Hok type I toxin proteins across bacterial species. The Hok-Sok type I toxin-antitoxin system was first described on plasmids where it serves to maintain the plasmid in a population of bacterial cells: translation of the hok mRNA is prevented via the small antitoxin RNA Sok. Upon plasmid loss, with no new transcription of sok, the highly stable hok mRNA is translated into a small protein, killing the plasmid-less cell. Homologues to the system were identified in the chromosome of E. coli in the 1990s, and subsequent analyses have identified identical systems in other bacterial chromosomes, though they are close relatives to E. coli. Given the increased number of bacterial genomes sequenced, the group examined how widespread Hok may be across bacteria. They used a combination of BLASTn, MMseqs2 (protein) and Infernal (RNA) to identify, as best possible, all possible homologs. They then used sequence identity cut-offs to form Hok "clusters," and identified key features of the cluster as well as tested toxicity of overproduction of 31 homologs in a strain of Salmonella. Overall, though a variety of bioinformatic predictions and analyses, the manuscript identifies an expanded number of Hok members not previously identified and broaden the species it is found in, supported that Hok is not associate with defense systems, and provides additional support that horizontal transfer of hok genes is likely via plasmids (where hok is presumed to have originated).

Major comments: There are some areas of the text that are a bit too definitive (these can be fixed or better explained in the text) and a few questions raised about the analyses and interpretations.

Authors' answer to R3 Major Comment: As suggested by the reviewer, we rephrased parts of the manuscript.

__These are the specific comments: __

Introduction R3_C1: First paragraph: "Toxin production leads to the death of the cell encoding it" For many chromosomally encoded systems, toxicity has only been observed via artificial overexpression. This is an important point, as for many systems, a true biological function remains unknown. Further, add caveats regarding toxin function (for systems with validated function, they are involved in...). Again, there are still many questions for many t-at systems, in particular the Type I systems.

__Authors' answer to R3_C1: __Indeed, the function of type 1 TA, in particular chromosomal ones, is still a matter of debate. While for hok/Sok R1, we previously showed death by expression at the chromosomal level, this was not shown for all TA (Le Rhun et al., NAR, 2023). We added that it could lead to the death or growth arrest of the cell instead and added the reviewer changes to for the function part.

__R3_C2: __Introduction: type I's are more narrow in distribution, but much of this is due to their size and lack of biochemical domains. Again, please clarify more here.

__Authors' answer to R3_C2: __We added the reviewer suggestion to the text.

__R3_C3: __Introduction: while Hok's have been found on chromosomes, in E. coli strains, there is clear evidence that many are inactive. This comes up in the discussion, but it is worth including briefly in the introduction.

Authors' answer to R3_C3: We have now added in the introduction that in the K12 laboratory strain, most chromosomal hok/Sok were found to be inactive.

__R3_C4: __For the predicted transmembrane domain: it would be worth to include a box/indication as to where that is within the peptide (with the understanding it may not be exact). Is there more/less variation here? I'm assuming all clusters/family have a predicted TM domain?

__Authors' answer to R3_C4: __When predicting the TM domain using DeepTMHMM - 1.0 prediction (https://services.healthtech.dtu.dk/services/DeepTMHMM-1.0/), 227 out of the 1264 unique Hok sequence are predicted to have a TM (transmembrane), 7 a SP (signal peptide) and a TM and 1025 have a SP. When predicting the TM of the consensus sequence (most abundant amino-acid) shown in Fig. 1D, region A8 to L25 is predicted to be inserted in the membrane, with the Nterm inside and Cterm outside.

__R3_C5: __What is the cutoff for being a Hok? Did they take the "last hit" and use that in additional searches to see if more appeared? If that was done, and the search was exhaustive, this really important to add for the reader.

Authors' answer to R3_C5: The MMseqs2 search was performed using 5 iterations as indicated in the M&M, meaning that the hits of the one search were used to search the database again five time in a raw. Importantly, an attempt to increase the number of iterations to 10 did not significantly increase the number of hits. Therefore, at least for the MMseqs2 search in the RefSeq database, we are close to being exhaustive.

__R3_C6: __Figure S4: the authors state that there was no difference in the degree of toxicity between the clusters. There do appear to be some peptides tested that at the arabinose concentration used did not repress growth as immediately as others. If higher arabinose concentration is used, does that eliminate these differences? OR are many of these suppressors-if diluted back again, do they grow as if they are non-toxic in arabinose?

Authors' answer to R3_C6: As suggested by Reviewer 1 (R1_C2), we performed titration of arabinose in a system overexpressing araE in a ΔaraBAD but were not able to find difference of toxicity in our conditions, see also our answer to R1_C2.

__R3_C7: __Discussion: "because non-functional homologs are expected to quickly accumulate mutations..." is a bit problematic. Hok is highly regulated-as are some of the other well-described type I toxins. In MG1655, while the coding sequence may be intact, there are other mutations and/or insertion elements that prevent expression (and be extension, function. Given the lack of consensus data for type Is, it is best to provide more context for this. If the authors wish to argue that they should quickly accumulate mutations, it would be good to provide additional rates/evidence (even for other loci) from the Enterobacteriaceae.

__Authors' answer to R3_C7: __We agree this statement might need to be supported further. We have removed this sentence to address this concern.

__Minor comments: __

__R3_C8: __For the sequences used in the search: please provide the sequence used in addition to the reference to the T1TAdb. Was the full-length hok mRNA, including mok, used? Please provide the nucleic acid sequence (and include description of whether full-length, etc.) in Materials and Methods or in Supplemental.

__Authors' answer to R3_C8: __Sequences and code were deposited on https://gitub.u-bordeaux.fr/alerhun/Escalera-Maurer_2025. This files named curated_Hok.fasta and hok.fa, corresponding to Hok protein and mRNA sequences respectively are available in the file "T1TAdb input".

__R3_C9: __60% identity was used for clustering. Did this become a problem-meaning separation of same property amino acid?

__Authors' answer to R3_C9: __We checked amino acid signatures for each cluster (Fig S2), but could not find anything relevant.

__R3_C10: __Fig. S2: for the clusters shown, please add in HokB, HokE, etc., to better correspond to Figure 1 in the main text.

__Authors' answer to R3_C10: __The clusters were annotated according to the suggestion.

__R3_C11: __Fig S1: this figure is challenging to orient-what are the numbers (8_10_85)?

Authors' answer to R3_C11: The figure was generated using the CLANS tool, with each unique sequence retrieved by our analysis shown as a dot. Hok homologous sequences are in red and cluster together, the outlier clusters are annotated with the numbers corresponding to their 60% identity cluster. We understand that separating the number using an underscore could lead to confusion, therefore we have now separated the numbers using a coma.

__R3_C12: __Please make a separate table or sheet for the experimentally tested peptides. Table S1 is quite large and a separate table/sheet would make this easier to find. If possible, please give the files names a more descriptive title (Table S1 in the name for example). This may be an issue with Review Commons but the individual file names were non-descript and the descriptions on the webpage did not indicate what the file contained.

__Authors' answer to R3_C12: __We named the files Table S1 and File_S1 to S7. We added a table S2 with the experimentally tested peptides. Note that identical peptides can be sometime found in several bacterial loci.

__R3_C13: __Figure S9: the black arrow for Hok is hard to see-it appears that the long grey bar going through multiple loci is indicative of Hok. Perhaps label this differently to make it easier on the reader (the line initially seemed to be a formatting issue and not indicative of the position of Hok.

__Authors' answer to R3_C13: __We have now added a new label to indicate where is Hok, and clarified it in the figure legend.

__R3_C14: __While the authors focused on Hok for this approach, which is fine and appropriate, can they comment at all about where mok is there in these new clusters/sub-families? Sok potential?

__Authors' answer to R3_C14: __We added a paragraph about Mok in the discussion.

__R3 Significance: __Overall the paper is a sound bioinformatic exercise and is improved with the testing of numerous "new" Hok proteins. Most of the comments can be done with some clarifications and maybe some additional analyses and/or verification which should take minimal time. The authors are over-emphatic at points as indicated and need to be more careful and precise with their language.

In terms of advancement, it advances the distribution of these systems and adds to the depth of sub-classes. The audience will be more specialized to those who study these systems.

Expertise: I have been studying type I toxin-antitoxin systems since the mid-2000s. We published a study examining (and mentioned well by this article!) the distribution in chromosomes of type I toxin-antitoxin systems, identified brand-new systems (that were chromosomally-limited at the time). My lab has continued to study regulation of type I toxins and distribution of chromosomally-only-encoded systems (so not Hok).

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

In the manuscript, "The Hok bacterial toxin: diversity, toxicity, distribution and genomic localization," by Escalera-Maurer et al., investigate the distribution of Hok type I toxin proteins across bacterial species. The Hok-Sok type I toxin-antitoxin system was first described on plasmids where it serves to maintain the plasmid in a population of bacterial cells: translation of the hok mRNA is prevented via the small antitoxin RNA Sok. Upon plasmid loss, with no new transcription of sok, the highly stable hok mRNA is translated into a small protein, killing the plasmid-less cell. Homologues to the system were identified in the chromosome of E. coli in the 1990s, and subsequent analyses have identified identical systems in other bacterial chromosomes, though they are close relatives to E. coli. Given the increased number of bacterial genomes sequenced, the group examined how widespread Hok may be across bacteria. They used a combination of BLASTn, MMseqs2 (protein) and Infernal (RNA) to identify, as best possible, all possible homologs. They then used sequence identity cut-offs to form Hok "clusters," and identified key features of the cluster as well as tested toxicity of overproduction of 31 homologs in a strain of Salmonella. Overall, though a variety of bioinformatic predictions and analyses, the manuscript identifies an expanded number of Hok members not previously identified and broaden the species it is found in, supported that Hok is not associate with defense systems, and provides additional support that horizontal transfer of hok genes is likely via plasmids (where hok is presumed to have originated).

Major comments: There are some areas of the text that are a bit too definitive (these can be fixed or better explained in the text) and a few questions raised about the analyses and interpretations. These are the specific comments:

Introduction

First paragraph: "Toxin production leads to the death of the cell encoding it" For many chromosomally encoded systems, toxicity has only been observed via artificial overexpression. This is an important point, as for many systems, a true biological function remains unknown. Further, add caveats regarding toxin function (for systems with validated function, they are involved in...). Again, there are still many questions for many t-at systems, in particular the Type I systems. Introduction: type I's are more narrow in distribution, but much of this is due to their size and lack of biochemical domains. Again, please clarify more here.

Introduction: while Hok's have been found on chromosomes, in E. coli strains, there is clear evidence that many are inactive. This comes up in the discussion, but it is worth including briefly in the introduction.

For the predicted transmembrane domain: it would be worth to include a box/indication as to where that is within the peptide (with the understanding it may not be exact). Is there more/less variation here? I'm assuming all clusters/family have a predicted TM domain?

What is the cutoff for being a Hok? Did they take the "last hit" and use that in additional searches to see if more appeared? If that was done, and the search was exhaustive, this really important to add for the reader.

Figure S4: the authors state that there was no difference in the degree of toxicity between the clusters. There do appear to be some peptides tested that at the arabinose concentration used did not repress growth as immediately as others. If higher arabinose concentration is used, does that eliminate these differences? OR are many of these suppressors-if diluted back again, do they grow as if they are non-toxic in arabinose?

Discussion: "because non-functional homologs are expected to quickly accumulate mutations..." is a bit problematic. Hok is highly regulated-as are some of the other well-described type I toxins. In MG1655, while the coding sequence may be intact, there are other mutations and/or insertion elements that prevent expression (and be extension, function. Given the lack of consensus data for type Is, it is best to provide more context for this. If the authors wish to argue that they should quickly accumulate mutations, it would be good to provide additional rates/evidence (even for other loci) from the Enterobacteriaceae.

Minor comments:

For the sequences used in the search: please provide the sequence used in addition to the reference to the T1TAdb. Was the full-length hok mRNA, including mok, used? Please provide the nucleic acid sequence (and include description of whether full-length, etc.) in Materials and Methods or in Supplemental.

60% identity was used for clustering. Did this become a problem-meaning separation of same property amino acid? Fig. S2: for the clusters shown, please add in HokB, HokE, etc., to better correspond to Figure 1 in the main text.

Fig S1: this figure is challenging to orient-what are the numbers (8_10_85)?

Please make a separate table or sheet for the experimentally tested peptides. Table S1 is quite large and a separate table/sheet would make this easier to find. If possible, please give the files names a more descriptive title (Table S1 in the name for example). This may be an issue with Review Commons but the individual file names were non-descript and the descriptions on the webpage did not indicate what the file contained.

Figure S9: the black arrow for Hok is hard to see-it appears that the long grey bar going through multiple loci is indicative of Hok. Perhaps label this differently to make it easier on the reader (the line initially seemed to be a formatting issue and not indicative of the position of Hok.

While the authors focused on Hok for this approach, which is fine and appropriate, can they comment at all about where mok is there in these new clusters/sub-families? Sok potential?

Significance

Overall the paper is a sound bioinformatic exercise and is improved with the testing of numerous "new" Hok proteins. Most of the comments can be done with some clarifications and maybe some additional analyses and/or verification which should take minimal time. The authors are over-emphatic at points as indicated and need to be more careful and precise with their language.

In terms of advancement, it advances the distribution of these systems and adds to the depth of sub-classes.

The audience will be more specialized to those who study these systems.

Expertise: I have been studying type I toxin-antitoxin systems since the mid-2000s. We published a study examining (and mentioned well by this article!) the distribution in chromosomes of type I toxin-antitoxin systems, identified brand-new systems (that were chromosomally-limited at the time). My lab has continued to study regulation of type I toxins and distribution of chromosomally-only-encoded systems (so not Hok).

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

The authors examined how the Hok toxins are spread across bacterial genomes. The manuscript including its figures is hard to read and understand. I commented figure 1 in details, but similar comments apply to the other figures. Overall, the data lack clarity and precision. Finding information about sequences, clusters in the supplementary materials was not easy. The manuscript should be thoroughly revised. In addition, I believe that other aspects should be developed to expand the interest of the study, such as the co-occurrence of multiple systems in chromosomes, on plasmids and whether they are able to crosstalk. This might provide some evolutionary insights into the biology of these toxins.

Introduction:

The introduction lacks information regarding the Hok protein (size, structure prediction, localization) as well as a bit of explanation about the reason of looking at these toxins. The description of the potential roles should be a bit expanded.

Results:

When the authors talk about 'loci', they mean genes encoding Hok homologs if I understand correctly. They did not look for the Sok sequences (hok-sok loci).

It is not clear what the authors did with the sequences for which they could not detect a start codon and a SD (although it is unusual to refer to SD in the context of protein sequence)

Figure 1A is not clear. The total of the bars equal 32,532 which is the number of 'loci' detected by the combination of the different methods. However, it is not clear to me how many are redundant. For instance, I suppose that all the 8483 sequences that were retrieved using blastn and Infernal were retrieved using MMseqs2, blastn and Infernal. So, what is the actual number of sequences that were found? When the authors talk about 1264 distinct peptides, what do they mean? What are the numbers on the X axis (18209, 2260, 27728)?

About Infernal: first the authors are stating that only 8% of the sequences are lost when not considering the mRNA structure - which they seem to consider as negligeable. Then in the next section, they state that Infernal is the best tool at identifying clusters that are not detected otherwise. Seems a bit contradictory.

Cluster determination. The threshold was put at 60% identity. What is the rationale for the 60% identity? Given that the Hok sequences (like toxins and antitoxins from TA systems in general) are highly variable, this leads to a high number of clusters. I'm not sure of the relevance of these clusters. Are there any other criteria to define clusters?

The authors claim that most of the Hok diversity is found on chromosomes. However, the number of chromosomal Hok is higher than that located on plasmids, which might be related to the different sizes of the different replicons ie, chromosomes being larger than plasmids. Is there a way to normalize by determining the density per size?

'46 of the 62 clusters contained 10 or less distinct sequences and might be in the process of degenerating'. The authors also linked this with SD detection. Please explain. From what was indicated earlier, I understand that sequences with premature stop codons or short sequences (<40aa) were removed from the analysis earlier. Lacking an SD is a sign of decay? Were these sequences lacking SD not discarded before starting the analysis? Did the authors experimentally validate some of these sequences?

'Only 7.3% of the unique sequences were found on both plasmids and chromosomes'. From this observation, the authors conclude that 'there is little stable transfer from chromosomes to plasmids or vice-versa'. I don't understand what this means. Do they mean identical sequences? The fact that sequences differ from chromosomes to plasmids does not rule out 'stable transfer'. What do they actually mean by stable transfer? Once the gene is horizontally transferred, it is fixed and vertically transmitted? Same comments apply to the inter-genera horizontal transfer by plasmids.

I don't understand the next section about 'family'. What do the authors mean about 'family'? Genera? The same apply to the next section about the Y to C recoding. Did the authors do point mutations in the conserved amino acids/codons to test whether they are important for toxicity? Some Hok variants lacks some of the conserved amino acids and are toxic (under overexpression conditions in Salmonella). What about T18, C31 and E42?

The prevalence of Hok in chromosomes or on plasmids might depend on various confounding parameters, such as the size, number of sequences available among others. The authors should find methods to correct for all that.

Link with defense systems. The threshold was set at 20 kb. Why this threshold?

Significance

The paper in its current state appears to serve the role of a data repository rather than a thorough and original analysis. It requires extensive revisions before it can be of interest to experts in the toxin-antitoxin field.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Here, Escalera-Maurer and colleagues, present an up-to-date distribution of homologues of Hok toxic proteins belonging to the well-annotated, but otherwise functionally obscure, hok/Sok type I toxin-antitoxin system, across the RefSeq database. Although such computational analyses have been done in the past, the authors here find many more hok homologs than described before, and they categorise their distribution based on whether they are encoded on chromosomes, plasmids, or (pro)phages. These computational analyses are in general tricky with T1TAs, as their toxins are quite short (~50 amino acids, as is the case for Hok), which is why the authors here used three separate approaches to expand their search (nucleotide-level BLAST, protein-homology, or both combined with Infernal). The authors cluster the Hok homologues they find based on a 60% sequence identity cut-off (expanding the known clusters in the process), and proceeded to test 31 candidates belonging to 15 sequence-clusters for their toxicity in Salmonella Typhimurium LT2, showing that 30/31 were toxic upon induction. An interesting finding from their endeavours is that hok/Sok homologues are enriched within prophages and large plasmids, but are not enriched near bacterial anti-phage defense systems (in contrast to the SymE/SymR T1TA). The findings suggest that hok/Sok are indeed sometimes linked to phage and plasmid biology, although they might not be antiphage defenses per se (they have been clearly shown in the past to be addiction modules, and this is still clearly true).

My expertise lies towards the experimental side of the authors' work, I thus cannot comment on the accuracy/robustness of the computational analyses performed here. The authors do a fine job in clearly stating their findings overall; I could follow most of the conclusions, and I deemed that most of them were supported by their work. Additionally, I find that this paper is a missed opportunity to uncover even more novel biology connected to the interesting hok/Sok T1TAs. The paper does not provide a new framework to think about what is the function of the chromosomal/prophage hok/Sok T1TA systems, although I realize that this is very difficult to accomplish, especially when considering that hok/Sok systems have been around in the literature for almost 40 years.

My major comment is in regard to the Hok toxicity assays (Fig. 2). The authors state in the discussion that "Hok peptides originating from chromosomes are as toxic as those from plasmids", but I believe that the way that they tested their constructs might not have allowed them to see toxicity differences between the two groups. Specifically, using the multi-copy plasmid pAZ3 (pBR322 origin of replication; ~15-20 plasmid copies per chromosome) to induce the different Hok toxin homologues in Salmonella Typhimurium LT2 with arabinose might have masked toxicity differences that would otherwise be apparent on the chromosomal expression-level.

Some of the authors themselves have previously used the FASTBAC-Seq method to study the Hok homologue from plasmid R1, a useful technique during which a toxin is integrated in the chromosome, in order to study their toxicity under natural levels of expression. I believe that an ideal scenario would be to apply FASTBAC-seq to some of the 31 Hok homologues described here (e.g., a subset of plasmidic vs chromosomal Hok homologues) to shed light on potential toxicity differences between the Hok clusters. This would increase the value of the presented study.

Alternatively, the authors could employ an L-arabinose concentration gradient to titrate the expression levels of the Hok toxins in order to potentially see different toxicity levels from the different homologues. However, this is not going to work in the system as they are using it now for two reasons:

a) the S. Typhimurium LT2 (STm) used here has its arabinose utilization operon intact (araBAD), which means that Salmonella can catabolize arabinose to use it as a carbon source. This catabolization process interferes with the arabinose induction (i.e., Salmonella eats arabinose instead of using it as the Hok inducer). To ameliorate this, the authors could delete the araBAD operon in STm, rendering STm incapable of catabolizing arabinose, and repeat the experiments in that strain. Or use E. coli BW25113 as the expression host, which already has the araBAD operon deleted (it is not clear to me why the different Hok homologues would not be toxic in E. coli, as the different Hok homologues are widely diverse in sequence, as the authors found here).

b) Even with the araBAD operon deleted, the arabinose induction would be bimodally on or off in the population, due to the bimodal expression of the arabinose transporter (AraE; see Khlebnikov et al., 2002). This would again not allow for titratable arabinose-inducible expression from different concentrations of arabinose. The solution for this would be to co-express a separate plasmid with araE, which would render every cell the same in regards to arabinose permeability, and thus the system would be titratable (as explained in Khlebnikov et al., 2002).

Therefore, if the authors would be interested to go towards this route, they would have to first delete the araBAD from STm, then transform STm with an araE plasmid, and redo the experiments. In addition, I would propose to the authors to use the drop plate method (agar plate-based), which is more sensitive compared to the liquid assays employed here.

Having said all that, I understand that all this experimental work would be strenuous and time-consuming, and although I would like to see it happen, this is not my paper. I would be content therefore if the authors toned down the claim that plasmidic vs chromosomal Hok homologues have the same toxicity, and discuss that chromosomal levels of toxicity are an important caveat that has not been explored here.

Other comments:

a) There is barely any discussion of the Sok component (RNA antitoxin) of the homologues; why is that? Could you please discuss Sok differences across the homologues, or at least explain why this is not discussed at all in the paper (e.g., in the discussion)?

b) In the results section, the Hok clusters are referred to as 62 in number ("Because Hok sequences were too short and variable to construct a meaningful phylogenetic tree, we clustered the Hok sequences with a 60% identity threshold and obtained 62 clusters"), but then in the discussion section, the cluster number becomes 74 ("We highlighted the high sequence variability within Hok peptides by obtaining a total of 74 clusters with 60% identity (Fig. S7)."). Which one is the right number, and why is there a discrepancy?

Significance

The most well-clarified aspect of the paper presented here is the distribution of Hok homologues, with the novel aspect of the location in which the hok/Sok T1TAs reside (i.e., chromosome, plasmid, or phage). There is room for the molecular genetics part to be developed further, as I discussed earlier, however this study is the most up-to-date characterization of the diversity of Hok homologues, and will be of interest to the T1TA and the general toxin-antitoxin field.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

__Reply to the Reviewers __

We thank the Reviewers for their positive assessment and recognition of the paper achievements. The insightful comments will strengthen the data and manuscript.

Referee #1* *

Minor comments

1. Fig 1B - add arrows showing mRNAs being translated or not (the latter mentioned in line 113 is not so easy to see). We have magnified the inset of the colocalisation in the right column; we added arrows and arrowheads to differentiate colocalised and non-colocalised bcd with translating SunTag.

2. Fig 2A - add a sentence explaining why 1,6HD, 2,5HD and NaCl disrupt P bodies. *

We have added the information on the use of 1,6HD, 2,5HD, and NaCl to disrupt P-bodies as below. Revised line 158: “To further show that bcd storage in P bodies is required for translational repression, we treated mature eggs with chemicals known to disrupt RNP granule integrity (31, 37, 69-72). Previous work has shown that the physical properties of P bodies in mature Drosophila oocytes can be shifted from an arrested to a more liquid-like state by addition of the aliphatic alcohol hexanediol (HD) (Sankaranarayanan et al., 2021, Ribbeck and Görlich, 2002; Kroschwald et al., 2017). While 1,6 HD has been widely used to probe the physical state of phase-separated condensates both in vivo and in vitro (Alberti et al., 2019; McSwiggen et al., 2019; Gao et al., 2022), in some cells it appears to have unwanted cellular consequences (Ulianov et al., 2021). These include a potentially lethal cellular consequences that may indirectly affect the ability of condensates to form (Kroschwald et al., 2017) and wider cellular implications thought to alter the activity of kinases (Düster et al., 2021). While we did not observe any noticeable cellular issues in mature Drosophila oocytes with 1,6 HD, we also used 2,5 HD, known to be less problematic in most tissues (Ulianov et al., 2021) and the monovalent salt sodium chloride (NaCl), which changes electrostatic interactions (Sankaranarayanan et al., 2021).”

*Fig 4C - explain in the legend what the white lines drawn over the image represent. And why is there such an obvious distinction in the staining where suddenly the DAPI is much more evident (is the image from tile scans)? *

Figure 4C is the tile scan image of a n.c.10 embryo and the white line classified the image into four quadrants. We used this image to quantify the extent of bcd (magenta) colocalisation to SunTag (green) in the anterior and posterior domains of the embryo in the bar graph shown in panel C’. There is a formatting error in the image. We will correct this in the revised version. We will also include the details of white lines in the legends. Finally, based on further reviewer comments, in the revised version this data is shifted to the supplementary information.

• Line 215 - 'We did not see any significant differences in the translation of bcd based on their position, however, there appears an enhanced translation of bcd localised basally to the nuclei (Figure S5).' Since the difference is not significant, I do not think the authors should conclude that translation is enhanced basally. *

We agree with the reviewer. In this preliminary revision we have changed this statement to: “We did not see any differences in the translation of bcd based on their position with respect to the nuclei position (Figure S5)” (revised line 238-239).

*Line 218: 'The interphase nuclei and their subsequent mitotic divisions appeared to displace bcd towards the apical surface (Figure S6B).' Greater explanation is needed in the legend to Fig S6B to support this statement as the data just seem to show a nuclear division - I would have thought an apical-basal view is needed to conclude this. *

We have rearranged this figure and shown in clarity the apical-basal view of the blastoderm nuclei and the displacement of bcd from the surface of the blastoderm in Figure S8.

New Figure S8: n.c.8 - pre-cortical migration; n.c.12,14- post cortical migration; Mitosis stages of n.c.9-10. The cortical interphase nuclei at n.c. 12,14 displaces bcd. The nuclear area (DAPI, cyan) does not show any bcd particles (magenta) indicated by blue stars. The mitotic nuclei (yellow arrowheads, yellow stars) displace bcd along the plane of nuclear division (doubled headed yellow arrows).

Fig 5B - the authors compare Bcd protein distribution across developmental time. However, in the early time points cytoplasmic Bcd is measured (presumably as it does not appear nuclear until nc8 onwards) and compare the distribution to nuclear Bcd intensities from nc9 onwards. Is most/all of the Bcd protein nuclear localised form nc9 to validate the nuclear quantitation? Does the distribution look the same if total Bcd protein is measured per volume rather than just the nuclear signal? Are the authors assuming a constant fast rate of nuclear import?

From n.c.8 onwards, the Bcd signal in interphase nuclei builds up, with the nuclear intensity becoming very high compared to cytoplasmic Bcd. However, we do see significant Bcd signal in the cytoplasm (i.e., above background). In earlier work, gradients of the nuclear Bcd and nuclear-import mutant Bcd overlapped closely (Figure 1B, Grimm et al., 2010). This essentially suggests the nuclear Bcd gradient reflects the corresponding gradient of cytoplasmic Bcd. Further, the nuclear import of Bcd occurs rapidly after photobleaching (Gregor et al., 2007). Based on these observations, and our own measurements, prior to n.c. 9, the cytoplasmic gradient is likely a good approximation of the overall shape, whereas post n.c. 9 the Bcd signal is largely nuclear localised. Further, the overall profile is not dependent on the nuclear volume.

• Line 259 - 'We then asked if considering the spatiotemporal pattern of bcd translation' - the authors should clarify what new information was included in the model. Similarly in line 286, 'By including more realistic bcd mRNA translation' - what does this actually mean? In line 346, 'We see that the original SDD model .... was too simple.' It would be nice to compare the outputs from the original vs modified SDD models to support the statement that the original model was too simple. *

We will improve the linking of the results to the model. The important point is that when and where Bcd production occurs is more faithfully used, compared with previous approximations. By including more realistic production domains, we can replicate the observed Bcd gradient within the SDD paradigm without resorting to more complex models.

Fig S1A - clarify what the difference is between the 2 +HD panels shown.__ __

The two +HD panels at stage 14 indicate that upon the addition of HD, there are no particles in 70% of the embryos, and 30% show reduced particles. We will add this information to the figure legend.

• Fig S2E - the graph axis label/legend says it is intensity/molecule. Since intensity/molecule is higher in the anterior for bcd RNAs, is this because there are clumps of mRNAs (in which case it's actually intensity/puncta)? *

The density of mRNA is very high in the anterior pole; there is a chance that more than one bcd particle is within the imaged puncta (due to optical resolution limitations). We will change the y-axis to average intensity per molecule to average intensity per puncta.

• Fig S4 - I think this line is included in error: '(B) The line plots of bcd spread on the Dorsal vs. Ventral surfaces.'*

Yes, we will correct this in the revision.

• In B, D, E - is the plot depth from the dorsal surface? I would have preferred to see actual mRNA numbers rather than normalised mRNAs. In Fig S4D moderate, from 10um onwards there are virtually no mRNA counts based on the normalised value, but what is the actual number? The equivalent % translated data in Fig S4E look noisy so I wonder if this is due to there being a tiny mRNA number. The same is true for Figs S4D, E 10um+ in the low region.*

Beyond 10um from the dorsal surface, the number of bcdsun10 counts is very low. It becomes negligible at the moderate and low domains. We will attach the actual counts of mRNA in all these domains as a supplementary table in the revised version.

General assessment Strengths are: 1) the data are of high quality; 2) the study advances the field by directly visualising Bcd mRNA translation during early Drosophila development; 3) the data showing re-localisation of bcd mRNAs to P bodies nc14 provides new mechanistic insight into its degradation; 4) a new SDD model for Bcd gradient formation is presented. Limitations of the study are: 1) there was already strong evidence (but no direct demonstration) that bcd mRNA translation was associated with release from P bodies at egg activation; 2) it is not totally clear to me how exactly the modified SDD model varies from the original one both in terms of parameters included and model output.

This is the first direct demonstration of the translation of bcd mRNA released as a single mRNA from P bodies. Previously, we have shown that P bodies disruption releases single bcd from the condensates (31). We have captured a comprehensive understanding of the status of individual bcd translation events, from their release from P bodies at the end of oocyte maturation until the end of blastoderm formation.

The underlying SDD model – that of localised production, diffusion, and degradation – is still the same (up to spatially varying diffusion). Yet the model as originally formulated did not fit all aspects of the data, especially with regards to the system dynamics. Here, we demonstrate that by including more accurate approximations of when and where Bcd is produced, we can explain the formation of the Bcd morphogen gradient without recourse to any further mechanism.

Referee #2

1. Line 114: The authors claim to have validated the SunTag using a fluorescent reporter, but do not show any data. Ref 60 is a general reference to the SunTag, and not the Bcd results in this paper. Perhaps place their data into a supplemental figure or movie? To show the validation of our bcdSun32 line, we have composed a new Figure S1 that shows the translating bcdSun32 (magenta) colocalising to the ScFV-mSGFP2 (green). Yellow arrowheads in the zoom (right panel) points to the translating bcdSun32 (magenta) and red arrowheads points to the untranslated bcdSun32. In addition, we have also shown the validation of bcdSun32 with the anti-GCN4 staining in the main Figure 1B.

Further, we have dedicated supplementary Figure S3 (previously Figure S2) for the validation of our bcdSun10 construct. Briefly, bcdSun10 is inserted into att40 site of chr.2. We did a rescue experiment, where bcdSun10 rescued the lethality of homozygous bcdE1 null mutant. We then performed a colocalisation experiment using smFISH, where we demonstrated that almost all bcd in the anterior pole are of type bcdSun10. We targeted specific fluorescent FISH probes against 10xSunTag sequence (magenta, Figure S2A) and bcd coding sequence (magenta, Figure S2A). Upon colocalisation, we found ~90% of the mRNA are of bcdSun10 type. The remaining 10% could likely be contributed by the noise level (Figure S2B). We will make sure these points are clear in the revised manuscript.

Line 128 and Fig. 1E: The claim that bcd becomes dispersed is difficult to verify by looking at the image. The language could also be more precise. What does it mean to lose tight association? Perhaps the authors could quantify the distribution, and summarize it by a length scale parameter? This is one of the main claims of the paper (cf. Line 23 of the abstract) but it is described vaguely and tersely here.

We have changed the text from, “We also confirmed that bcd becomes dispersed, losing its tight association with the anterior cortex (Figure 1E) (31)” to, “We also confirmed that bcd is released from the anterior cortex at egg activation (Figure 1E) (31, 21).” (Revised line 131).

The release of bcd mRNA at egg activation was first shown in 2008 (Ref 21, Figure 4, D-E) and again in 2021 (Ref 31, Figure 7 B and E). The main point in line 127-128, “P bodies disassembled and bcd was no longer colocalised with P bodies” and the novel aspect of line 23 is “translation observed”. The distribution of bcd mRNA after egg activation was not the point of this section. We have improved the writing in the revision to make this clearer.

Line 146, Fig. 1G: This is a really important figure in the paper, but it is confusing because it seems the authors use the word "translation," when they mean "presence of Bcd protein." In other places in the paper, the authors give the impression that "bcd translation" means translation in progress (assayed by the colocalization of GCN4 and bcd mRNA). However, in Fig. 1G, the focus is only on GCN4. Detecting Bcd protein only at the anterior does not mean that translation happens only at the anterior (e.g., diffusion or spatially-restricted degradation could be in play).

In Figure 1G, we have shown only the “translated” Bcd by staining with a-GCN4. We have changed line 146 from, “Consistent with previous findings, we only observed bcd translation at the anterior of the activated egg and early embryo (Figure 1G-H) (3, 68)” to, “Consistent with previous findings, we only observed the presence of Bcd protein at the anterior of the activated egg and early embryo (Figure 1G-H) (3, 68). (Revised line 151-153). We will use “translating bcd” or “bcd in translation” where we show colocalisation of bcd with BcdSun10 or BcdSun32 elsewhere in the manuscript.

We did not mean to claim that translation occurred only in the anterior pole. We show that the abundance of bcd is very high in the anterior pole (in agreement with previous work) and that this is where the majority of observed translation events took place. Indeed, we have also shown that posteriorly localised mRNAs have the same BcdSun10 intensity per bcd puncta from the posterior pole (Figure 3B & 4C’ and Figure S2 E), but these are much fewer in number.

*It would also be helpful to show a plot with quantification of Bcd detection (or translation) on the y-axis and a continuous AP coordinate on the x-axis, instead of just two points (anterior and posterior poles, the latter of which is uninteresting because observing no Bcd at the posterior pole is expected). *

In Figure 1G,H, our aim was to test whether release from P bodies allowed for bcd mRNA to be translated. We used the presence of Bcd protein at the anterior domain of the oocytes to show this. The posterior pole was included as an internal control. To show the spatial distribution of bcd mRNA and its translation, we used early blastoderm (Figure 3, Figure S4).

• *

Another issue with Fig. 1G is that the A and P panels presumably have different brightness and contrast. If not, just from looking at the A and P panels, the conclusion would be that Bcd protein is diffuse (and abundant) in the posterior and concentrated into puncta in the anterior. The authors should either make the brightness and contrast consistent or state that the P panel had a much higher brightness than the A panel.

We agree with this shortcoming. We have now added the following to Figure 1 legend to clarify this observation. “G: Representative fixed 10 µm Z-stack images (from 10 samples) showing BcdSun32 protein (anti-GCN4) is only present at the anterior of an in vitro activated egg or early embryo 30-minute post fertilization. BcdSun32 protein is not detected in these samples at the posterior pole (image contrast increased to highlight the lack of distinct particles at the posterior). BcdSun32 protein is also not detected at the anterior or posterior of a mature oocyte or an in vitro activated egg incubated with NS8953 (images have the contrast increased to highlight the lack of distinct particles). Scale bar: 20 mm; zoom 2 mm.” (Revised line 623).

• Line 176: This section is very confusing, because at this point the authors already addressed the spatial localization of translation in Fig. 1G,H (see my above comment). However, here it seems the authors have switched the definition of translation back to "translation in progress." Therefore, the confusion here could be eliminated by addressing the above point.*

In the revised version, we will use Bcd protein when shown with anti-GCN4 staining. We will use “translating bcd” or “bcd in translation” where we show colocalisation of bcd with a-GCN4 (BcdSun10 or BcdSun32). We will change this in the corresponding text.

Line 185: The sentence here is seemingly contradictory: "most...within 100 microns" implies that at least some are beyond 100 microns, while the sentence ends with "[none]...more than 100 microns." The language could perhaps be altered to be less vague/contradictory.

We will clarify this in the revised version. There are few particles visible beyond 100 um. In the lower panel of Figure 3B, the posterior domain shows few particles. However, their actual number compared to bcd counts within the 100 um is negligible (Figure3C). Nonetheless, the few bcd particles we observe do seem to be under translation (quantified in Figure 4C’ and Figure S2E).

• Line 204: It would be really nice to have quantification of the translation events, such as curves of rate of translation as a function of a continuous AP coordinate, and a curve for each nc.*__ __

In the revised version we will provide the results quantifying the translation events across the anterior- posterior axis. This will provide a clarity to the presence of bcd and their translation in the posterior domain with time.

Our colocalisation analysis is semi-automated. It includes an automated counting of the individual bcd particle counts and a manual judgement of the colocalised BcdSun10 protein (distinct spots, above noise) to bcd particles (Figure S3D). The bcd particle counts ran into thousands in each cyan square box (measuring 50um radius and ~ 20um deep from the dorsal surface). We selected three such boxes covering 150um (continuously) from the anterior pole across A-P axis and 20um deep of the flattened embryo mounts across D-V axis (Figure 3A-C, Figure S4). We have also scanned scarce particles in the posterior; however, bcd counts are very low compared to the anterior. Further, in Figure 4 we have repeated the same technique to measure translation of bcd particles in embryos at different nuclear cycles.

We have also shown continuous intensity measurements of bcd particles with their respective BcdSun10 gradient in Figure 5 across the A-P axis at different nuclear cycles. Here, we know BcdSun10 intensity is not only from the “translating” bcd (colocalised BcdSun10 to bcd particles) but also from the translated BcdSun10 freely diffusing (non-colocalised BcdSun10 to bcd particles). As asked by the reviewer, in the revised version we will add bcd counts and their translation status from anterior to posterior axis for each of the nuclear cycles.

In our future work, we planned to generate MS2 tagged bcdSun10 to measure the rates of translation in live across all nuclear cycles.

• *

*Line 209 and Fig 4C: The authors use the terms "intensity of translation events" or "translation intensity" without clearly defining them. From the figure (specifically from the y-axis label), it looks like the authors are quantifying the intensity per molecule (which is not clearly the same thing as "translation intensity"), but it would be nice if that were stated explicitly. *

In the relevant result section, we have changed the results text to “the intensity of translation events” for explaining the results of Figure 4C’.

• In addition, the authors again quantify only two points. This is a continuously frustrating part of the manuscript, which applies to nearly all figures where the authors looked only at two points in space. At a typical sample size of N = 3, it seems well within time constraints to image at multiple points along the AP axis.*__ __

In addition to the quantification shown at the anterior and posterior locations of the embryo in the Figure 3 and 4, we will show in the revised version, the quantification of translation events across all locations from the anterior to the posterior. We will use three embryos for each nuclear cycle from n.c.1 to 14.

• Furthermore, it sounds like the authors are saying the "translation intensity" is the same in anterior and the posterior, which is counterintuitive. The expectation is that translation would be undetectable at the posterior end, in part because bcd mRNA would not be present. (Note that this expectation is even acknowledged by the authors on Line 185, which I comment on above, and also on Line 197). There should also be very low levels of Bcd protein (possibly undetectable) at the posterior pole. As such, the authors should explain how they think their claim of the same "translation intensities" in the anterior vs posterior fits into the bigger picture of what we know about Bcd and what they have already stated in the manuscript. They should also explain how they observed enough molecules to quantify at the posterior end. The authors should also disclose how many points are in each box in the boxplot. For example, the sample size is N = 3 embryos. In just three embryos, how many bcd/GCN4 colocalizations did the authors observe at the posterior end of the embryo?*

In n.c.4 in Figure3, we saw few bcd particles in the posterior. However, at n.c.10 in Figure 4C’ the number of posterior bcd particles are higher than at the early stages. We have quantified them in Figure 4C’. We will clarify this from the new set of quantification we are undertaking now to quantify translation across the A-P axis in the revision.

Finally, we will also provide the number of bcd particle counts and their colocalisation with a-GCN4 as a supplementary table.

• Line 215: The sentence that starts on this line seems self-contradictory: I cannot tell whether or not there is a difference in translation based on position. *

We have not observed any difference in the translation of bcd particles depending on the position along the Z-axis. We will edit this in our revised version.

• Line 229: Long-ranged is a relative term. From the graph, one could state there is some spatial extent to the mRNA gradient, so it is unclear what the authors mean when they say it is not "long-ranged." Could the mRNA gradient be quantified, such as with a spatial length scale? This would provide more information for readers to make their own conclusions about whether it is long-ranged.*

We have quantified the bcd mRNA gradient for each n.c. (Figure 5B-C); absolute bcd intensities in Figure 5B, left panel and the normalised intensities in Figure 5C. The length of the mRNA spread appears constant with the half-length maximum of ~75um across all nuclear cycles. Our conclusion of a long ranged Bcd gradient is based on the comparisons of the half-length maximum measurements of bcd particles and BcdSun10 (Figure 5D).

*Line 230: When the authors claim the Bcd gradient is steeper earlier, a quantification of the spatial extent (exponential decay length scale) would be appropriate. Indeed, lambda as a function of time would be beneficial. It should also be placed in context of earlier papers that claim the spatial length scale is constant. *

We will show this effectively from the live movies of bcdSun10/nanos-scFv-sGFP2 in the revised version.

• Lines 235-236: The two sentences that start on these two lines are vague and seemingly contradictory. The first sentence says there is a spatial shift, but the second sentence sounds like it is saying there is no spatial change. The language could be more precise to explain the conclusions. *

We agree with the reviewer. We will edit this in revision.

Minor comments

1. • Line 81: Probably meant "evolutionarily conserved" * Yes, we have changed, “P bodies are an evolutionarily cytoplasmic RNP granule” to, “P bodies are an evolutionarily conserved cytoplasmic RNP granule.”(Revised line 84-85).

*Figure 1 legend: part B says "from 15 samples" but also says N = 20. Which is it, or do these numbers refer to different things? *

We have edited this from, “early embryo (from 15 samples)” to, “early embryo (from 20 samples)”. (Revised line 602).

• Line 217: migration of what? *

Edited to “cortical nuclear migration”.

• Line 228: "early embryo" is vague. The authors should give specific time points or nuclear cycle numbers.*

Edited to “nuclear cycles 1-8”.

• Line 301: Other locations in the paper say 75 microns or 100 microns. *

We will make the changes. It is 100 um.

• Fig. 5: all images should be oriented such that the dorsal midline is on the upper half of the embryo/image. *

We will flip the image to match.

• Fig. 5B: There are light tan and/or light orange curves (behind the bold curves) that are not explained. *

It is the standard deviation. This will be explained.

• Fig. 5C: the plot says "normalized" but nowhere do the authors describe what the curves are normalized to. There is also no explanation for what the broad areas of light color correspond to.*__ __

Normalised to the bcd intensity maxima. This will be explained.

Significance

The results, if upheld, are highly significant, as they are foundational measurements addressing a longstanding question of how morphogen gradients are formed, using Bcd (the foundational morphogen gradient) as a model. They also address fundamental questions in genetics and molecular biology: namely, control of mRNA distribution and translation.__ __

We thank Reviewer 2 for highlighting the importance of our work in the field. We are confident that we address the issues raised by Reviewer 2 with the new set of quantifications we are currently working on.

Referee #3

1. • It is not evident from the main results and methods text that the new SDD model incorporates the phenomenon reported in figure 4B. From my reading, the parameter beta accounts for the Bcd translation rate, which according to figure 7B(ii) effectively switches from off to on around fertilization and thereafter remains constant. Figure 4B shows that the fraction of bcd mRNA engaged in translation decreases beginning around NC12/13, and this is one of the more powerful results that comes from monitoring translation in addition to RNA localization/abundance/stability. My expectation based on figure 4B would be that parameter beta should decrease over time beginning around 90-100 minutes and approach zero by ~150 minutes. This rate could be fit to the experimental data that yields figure 4B. The modeling should be repeated while including this information. This is a good observation. Currently, the reduced rate of bcd translation is modelled by incorporating an increased rate of bcd *mRNA degradation. Of course, this could also be reduced by a change in the rate of translation directly. As stated already, the beta parameter is the least well characterised. In the revision, we will include a model where beta changes but not the mRNA degradation rate. We will improve the discussion to make this point clearer.

2. The presentation of the SDD model should be expanded to address how well the characteristic decay length fits A) measured Bcd protein distributions, B) measured at different nuclear cycles. This would strengthen the claim that the new SDD model better captures gradient dynamics given the addition of translation and RNA distribution. These experimental data already exist as reported in Figure 5. In the current Figure 7, panels D and D' add little to the story and could be moved to a supplement if the authors want to include it (in any case, please fix the typo on the time axis of fig 7D' to read "hours"). The model per cell cycle and the comparison of experimental and modeled decay lengths could replace current D and D'.*

Originally, we kept discussion of the SDD model only to core points. It is clear from all Reviewers that expanding this discussion is important. In the revision, we will refocus Figure 7 on describing new results that we can learn. As outlined in the responses above, this paper reveals an important insight: the SDD model – with suitable modifications such as temporally restricted Bcd production – can explain all observed properties of Bcd gradient formation. Other mechanisms – such as bcd mRNA gradients – are not required.

• The exposition of the manuscript would benefit significantly by including a section either in the introduction or the appropriate section of the results that defines the competing models for gradient formation. In the current version, these models are only cited, and the key details only come out late (e.g., lines 302 onward, in the Discussion). Nevertheless, some of the results are presented as if in dialog with these models, but it reads as a one-sided conversation. For instance: Figure 3. The undercurrent in this figure is the RNA-gradient model. In the context of this model, the results clearly show that translation of bcd is restricted to the anterior. Without this context, Figure 3 could read as a fairly unremarkable observation that translation occurs wherever there is mRNA. Restructuring the manuscript to explicitly name competing models and to address how experimental results support or detract from each competing model would greatly enhance the impact of the exposition.*

We thank the reviewer for this suggestion. We will add the current models of Bcd gradient formation in the introduction section and will change the narrative of results in the section explaining the models.

(4A) Related to point 3: The entire results text surrounding Figure 2 should be revised to include more detail about A) what specific hypotheses are being tested; and B) to critically evaluate the limitations of the experimental approaches used to evaluate these hypotheses. Hexanediol and high salt conditions are not named explicitly in the text, but the text touts these as "chemicals" that "disrupt P-body integrity." This implies that the treatments are specific to P-bodies. Neither of these approaches are only disrupting P Body integrity. This does not invalidate this approach, but the manuscript needs to state what hypothesis HD and NaCl treatment addresses, and acknowledge the caveats of the approach (such as the non-specificity and the assumptions about the mechanism of action for HD).

We have made the following edits to resolve this point. Revised line 158: “To further show that bcd storage in P bodies is required for translational repression, we treated mature eggs with chemicals known to disrupt RNP granule integrity (31, 37, 69-72). Previous work has shown that the physical properties of P bodies in mature Drosophila oocytes can be shifted from an arrested to a more liquid-like state by addition of the aliphatic alcohol hexanediol (HD) (Sankaranarayanan et al., 2021, Ribbeck and Görlich, 2002; Kroschwald et al., 2017). While 1,6 HD has been widely used to probe the physical state of phase-separated condensates both in vivo and in vitro (Alberti et al., 2019; McSwiggen et al., 2019; Gao et al., 2022), in some cells it appears to have unwanted cellular consequences (Ulianov et al., 2021). These include a potentially lethal cellular consequences that may indirectly affect the ability of condensates to form (Kroschwald et al., 2017) and wider cellular implications thought to alter the activity of kinases (Düster et al., 2021). While we did not observe any noticeable cellular issues in mature Drosophila oocytes with 1,6 HD, we also used 2,5 HD, known to be less problematic in most tissues (Ulianov et al., 2021) and the monovalent salt sodium chloride (NaCl), which changes electrostatic interactions (Sankaranarayanan et al., 2021).”

(4B) Continuing the comment above: it is good that the authors checked that HD and NaCl treatment does not cause egg activation. But no one outside of the field of Drosophila egg activation knows what the 2-minute bleach test is and shouldn't have to delve into the literature to understand this sentence. Please explain in one sentence that "if eggs are activated, then x happens following a short exposure to bleach (citations). We exposed HD and NaCl treated eggs to bleach and observed... ."

We have made the following edits to resolve this point. Revised line 174: “After treating mature eggs with these solutions, we observed BcdSun32 protein in the oocyte anterior (Figure 2A-B). One caveat to this experiment could be that treating mature eggs with these chemicals results in egg activation which would in turn generate Bcd protein. To eliminate this possibility, we first screened for phenotypic egg activation markers, including swelling and a change in the chorion (73). We also applied the classic approach of bleaching eggs for two minutes which causes lysis of unactivated eggs (74). All chemically treated eggs failed this bleaching test meaning they were not activated (74). While we unable to rule out non-specific actions of these treatments, these experiments corroborate that storage in P bodies that adopt an arrested physical state is crucial to maintain bcd translational repression (31).”

(4C) Continuing the comment above: The section of the results related to the endos mutation needs additional information. It is not apparent to the average reader how the endos mutation results in changes in RNP granules, nor what the expected outcome of such an effect would "further test the model" set up by the HD and NaCl experiments. The average reader needs more hand-holding throughout this entire section (related to figure 2) to follow the exposition of the results.

We have made the following edits to resolve this point. Edited line 185: “Finally, we used a genetic manipulation to change the physical state of P bodies in mature oocytes. Mutations in Drosophila Endosulfine (Endos), which is part of the conserved phosphoprotein ⍺-endosulfine (ENSA) family (75), caused a liquid-like P body state after oocyte maturation, similar to that observed with chemical treatment (Figure 2C) (31). This temporal effect matched the known roles of Endos as the master regulator of oocyte maturation (75, 76). endos mutant oocytes lost the colocalisation of bcd mRNA and P bodies, concurrent with P bodies becoming less viscous during oocyte maturation (Figure 2D, Figure S1). Particle size and position analysis showed that bcd mRNA prematurely exhibits an embryo distribution in these mutants (Figure 2E). Due to genetic and antibody constraints, we are unable to test for translation of bcd in the endos mutant. However, it follows that bcd observed in this diffuse distribution outside of P bodies would be translationally active (Figure 2E-F).”

• (4D) Continuing the comment above: The average reader also needs a better explanation of what hypothesis is being tested in Figure 1 with the pharmacological inhibition of calcium. *

We have made the following edits to resolve this point. Revised line 138: “We next sought to maintain the relationship between bcd mRNA and P bodies through egg activation. This would act as a control to further test if colocalisation of bcd to P bodies was necessary for its translational repression. Previous work has shown that a calcium wave is required at egg activation for further development (references to add Kaneuchi et al., 2015; York-Anderson et al., 2019; Hu and Wolfner, 2019). Chemical treatment with NS8593 disrupts this calcium wave, while other phenotypic markers of egg activation are still observed (58). Using NS8593 to disrupt the calcium wave in the activated egg, we show P bodies are retained during ex vivo egg activation (Figure 1E). In these treated eggs, bcd mRNA remains colocalised with the retained P bodies (Figure 1F). Based on these results and previous observations (31, 66), we hypothesised that the loss of colocalisation between bcd and P bodies correlates with bcd translation.”

*It is unclear why Bcd translation could not be measured in the endos mutant background, but it would be necessary to measure Bcd translation in the endos background. If genotypically it is not possible/inconvenient to invoke the suntag reporter in the endos background, would it not be sufficient to immunostain against Bcd itself? Different Bcd antisera have recently been reported and distributed by the Wieschaus and the Zeitlinger groups. *

We have recently received the Bcd antibody from the Zeitlinger group. This has not been shown to work for immunostaining. It remains unclear if it will be successful in this capacity, but we are currently testing it and will include this experiment in the revision if successful.

*Figure 4 overall is glorious, but there is a problem with panel C. What are the white lines? Why does the intensity for the green and magenta channel change abruptly in the middle of the embryo? *

These white lines divide the embryo into 4 compartments. We used this method to quantify the intensity of Bcd translation with respect to the bcd puncta. We will correct this image as there is a problem in formatting.

*It is noted that neither the methods section or the supplement does not contain any mention of how the modeling was performed. How was parameter beta fit? At least a brief section should be added to the methods describing how beta was fit (pending adjustments suggested in comment 1 above). A platinum-level addition would include a modeling supplement that reports the sensitivity of model outcomes to changes in parameters. *

We apologise for this omission and will include full methodological details in the revision.

Minor Comments:

1. • Line 28: "Source-Diffusion-Degradation" should be changed to "Synthesis-..."* We will edit in the revised version.

*Line 39: "blastocyst" should be "blastoderm stage embryo". *

We will edit in the revised version.

• Line 81: "P bodies are an evolutionarily cytoplasmic RNP granule." is "conserved" missing here? *

We will edit in the revised version.

• Throughout the manuscript, there should be better reporting of the imaged genotypes and whether the suntag is being visualized by indirect immunostaining of fixed tissues or through an encoded nanobody-GFP fusion. *

We will explain in detail in the revised version.

• Figure 1G: Why is the background staining so different across conditions? Is this a normalization artifact?*__ __

We agree with this shortcoming. We have now added the following to the figure legend to clarify this observation. “G: Representative fixed 10 µm Z-stack images (from 10 samples) showing BcdSun32 protein (anti-GCN4) is only present at the anterior of an in vitro activated egg or early embryo 30-minute post fertilization. BcdSun32 protein is not detected in these samples at the posterior pole (image contrast increased to highlight the lack of distinct particles at the posterior). BcdSun32 protein is also not detected at the anterior or posterior of a mature oocyte or an in vitro activated egg incubated with NS8953 (images have the contrast increased to highlight the lack of distinct particles). Scale bar: 20 mm; zoom 2 mm.” (Revised line 623).

Figure 2 legend: what is +Sch in the x-axis labels of figure 2B? The legend says that 2B is the quantification of the data in 2A, but there is no (presumed control) +Sch image in 2A.__ __

Thank you for this suggestion we have added the data to Figure 2A.

• Figure 5A largely repeats information presented in figure 4A. Please consider moving to a supplement. Also, please re-orient embryos to follow the convention that dorsal-most surfaces be presented on the top of the displayed images. *

Thank you for this suggestion. We will consider moving Figure 5A to the supplementary.

• The lower-case roman numerals referred to in the text for figure 7B are not included in the corresponding figure panel. *

We will edit in the revised version.

• Figure 7C y-axis typo (concentration). *

We will edit in the revised version.

• Line 222: "make a long-range functional gradient": more accurate to say, "but also marks mature, Bcd protein which resolves in the expected long-range gradient." *

We will edit in the revised version.

• Methods: Please check that all buffers referred to as acronyms are both compositionally defined in the reagents table, and that full names are written out at the time of first mention in the presented order. For instance, Schneider's media is referred to a few times before defining the acronym about midway through the methods section.*__ __

We have added to Figure 2B: “Quantification of experiments shown in A. The number of oocytes that displayed Bcd protein at the anterior as measured by the presence of BcdSun32 at the anterior of the oocyte, but not the posterior. Schneider’s Insect Medium (+Sch) used as a negative control. N = 30 oocytes for each treatment. Scale bar: 5 um.” (Revised line 646).

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

This is a review of "Dynamics of bicoid mRNA localization and translation dictate morphogen gradient formation" by Athilingam et al. In this manuscript, the authors perform quantification of mRNA localization and translation of bicoid, spanning oogenesis through the maternal to zygotic transition, yielding a definitive characterization of Bicoid gradient formation. The experiments, analysis, and interpretation are on the whole performed rigorously. I very much enjoyed this paper, partly for incorporating the aspects of bcd regulation during oogenesis, which compared to embryonic function of bcd is relatively under-studied. Also valuable is improving the characterization of how bcd expression is shut down at NC14. I have several major comments for revision, and a few minor comments. I should stress that none of the major comments are terrible but are intended to improve the impact/readability/flow of this nice manuscript. With the exception of a straightforward immunostaining experiment, all major comments constitute reworking of the model or the text.

Major Comments:

1) It is not evident from the main results and methods text that the new SDD model incorporates the phenomenon reported in figure 4B. From my reading, the parameter beta accounts for the Bcd translation rate, which according to figure 7B(ii) effectively switches from off to on around fertilization and thereafter remains constant. Figure 4B shows that the fraction of bcd mRNA engaged in translation decreases beginning around NC12/13, and this is one of the more powerful results that comes from monitoring translation in addition to RNA localization/abundance/stability. My expectation based on figure 4B would be that parameter beta should decrease over time beginning around 90-100 minutes and approach zero by ~150 minutes. This rate could be fit to the experimental data that yields figure 4B. The modeling should be repeated while including this information.

2) The presentation of the SDD model should be expanded to address how well the characteristic decay length fits A) measured Bcd protein distributions, B) measured at different nuclear cycles. This would strengthen the claim that the new SDD model better captures gradient dynamics given the addition of translation and RNA distribution. These experimental data already exist as reported in Figure 5. In the current Figure 7, panels D and D' add little to the story and could be moved to a supplement if the authors want to include it (in any case, please fix the typo on the time axis of fig 7D' to read "hours"). The model per cell cycle and the comparison of experimental and modeled decay lengths could replace current D and D'.

3) The exposition of the manuscript would benefit significantly by including a section either in the introduction or the appropriate section of the results that defines the competing models for gradient formation. In the current version, these models are only cited, and the key details only come out late (e.g., lines 302 onward, in the Discussion). Nevertheless, some of the results are presented as if in dialog with these models, but it reads as a one-sided conversation. For instance: Figure 3. The undercurrent in this figure is the RNA-gradient model. In the context of this model, the results clearly show that translation of bcd is restricted to the anterior. Without this context, Figure 3 could read as a fairly unremarkable observation that translation occurs wherever there is mRNA. Restructuring the manuscript to explicitly name competing models and to address how experimental results support or detract from each competing model would greatly enhance the impact of the exposition.

4A) Related to point 3: The entire results text surrounding Figure 2 should be revised to include more detail about A) what specific hypotheses are being tested; and B) to critically evaluate the limitations of the experimental approaches used to evaluate these hypotheses. Hexanediol and high salt conditions are not named explicitly in the text, but the text touts these as "chemicals" that "disrupt P-body integrity." This implies that the treatments are specific to P-bodies. Neither of these approaches are only disrupting P Body integrity. This does not invalidate this approach, but the manuscript needs to state what hypothesis HD and NaCl treatment addresses, and acknowledge the caveats of the approach (such as the non-specificity and the assumptions about the mechanism of action for HD).

4B) Continuing the comment above: it is good that the authors checked that HD and NaCl treatment does not cause egg activation. But no one outside of the field of Drosophila egg activation knows what the 2-minute bleach test is and shouldn't have to delve into the literature to understand this sentence. Please explain in one sentence that "if eggs are activated, then x happens following a short exposure to bleach (citations). We exposed HD and NaCl treated eggs to bleach and observed... ."

4C) Continuing the comment above: The section of the results related to the endos mutation needs additional information. It is not apparent to the average reader how the endos mutation results in changes in RNP granules, nor what the expected outcome of such an effect would "further test the model" set up by the HD and NaCl experiments. The average reader needs more hand-holding throughout this entire section (related to figure 2) to follow the exposition of the results.

4D) Continuing the comment above: The average reader also needs a better explanation of what hypothesis is being tested in Figure 1 with the pharmacological inhibition of calcium.

5) It is unclear why Bcd translation could not be measured in the endos mutant background, but it would be necessary to measure Bcd translation in the endos background. If genotypically it is not possible/inconvenient to invoke the suntag reporter in the endos background, would it not be sufficient to immunostain against Bcd itself? Different Bcd antisera have recently been reported and distributed by the Wieschaus and the Zeitlinger groups.

6) Figure 4 overall is glorious, but there is a problem with panel C. What are the white lines? Why does the intensity for the green and magenta channel change abruptly in the middle of the embryo?

7) It is noted that neither the methods section or the supplement does not contain any mention of how the modeling was performed. How was parameter beta fit? At least a brief section should be added to the methods describing how beta was fit (pending adjustments suggested in comment 1 above). A platinum-level addition would include a modeling supplement that reports the sensitivity of model outcomes to changes in parameters.

Minor Comments:

• Line 28: "Source-Diffusion-Degradation" should be changed to "Synthesis-..."
• Line 39: "blastocyst" should be "blastoderm stage embryo".
• Line 81: "P bodies are an evolutionarily cytoplasmic RNP granule." is "conserved" missing here?
• Throughout the manuscript, there should be better reporting of the imaged genotypes and whether the suntag is being visualized by indirect immunostaining of fixed tissues or through an encoded nanobody-GFP fusion.
• Figure 1G: Why is the background staining so different across conditions? Is this a normalization artifact?
• Figure 2 legend: what is +Sch in the x-axis labels of figure 2B? The legend says that 2B is the quantification of the data in 2A, but there is no (presumed control) +Sch image in 2A.
• Figure 5A largely repeats information presented in figure 4A. Please consider moving to a supplement. Also, please re-orient embryos to follow the convention that dorsal-most surfaces be presented on the top of the displayed images.
• The lower-case roman numerals referred to in the text for figure 7B are not included in the corresponding figure panel.
• Figure 7C y-axis typo (concentration).
• Line 222: "make a long-range functional gradient": more accurate to say, "but also marks mature, Bcd protein which resolves in the expected long-range gradient."
• Methods: Please check that all buffers referred to as acronyms are both compositionally defined in the reagents table, and that full names are written out at the time of first mention in the presented order. For instance, Schneider's media is referred to a few times before defining the acronym about midway through the methods section.

Referees cross-commenting

OK, We've been asked to comment on each others' reviews. I am reviewer 3. We have not been asked, as far as I can tell, to come up with a consensus review.

Overall, I feel that we are all generally enthusiastic about this manuscript. From most to least enthusiastic, we have reviewer 1, 3, and finally 2. But all three of us are apparently advocating positively and encouraging revision and clarification because, as we all agree, these results are important to publish.

Consensus Strengths:

1. The experimental approach is elegant, rigorous, and innovative, especially the real-time visualization of Bcd translation.
2. The data provide new mechanistic insight into when and where bcd is translated and how this changes over developmental time.
3. The relocalization of bcd mRNAs to P bodies during nc14 and the implications for RNA degradation are particularly compelling.
4. The manuscript establishes a path toward refining reaction-diffusion models of morphogen gradients using direct measurements of translation dynamics.

I agree with all of Reviewer 1's minor points.

I agree with Reviewer 2's points about:

• Showing the SunTag validation data using the fluorescent reporter.
• Clarifying the noted "translation" vs. "protein" issues. This bothered me too, but I wasn't able to articulate the issue as well as done here. This major issue summarizes several of the Reviewer's comments.
• Generally tightening the precision with which the results are discussed.

Overall: we have all provided favorable reviews that require mostly tightening of the text, showing some control datasets, maybe quantifying more points across the AP axis, and presenting the SDD model more comprehensively (comparing with old/translation-agnostic model, reporting characteristic decay lengths at different nuclear cycles, incorporating the reported change in translation rate across nuclear cycles (if this survives the clarification of what 'translation' means per Reviewer 2's comments), and perhaps providing more methodological detail on how parameters were fit).

Significance

The importance of this study is at several levels. For the developmental biologist, it addresses important mechanisms of translational control and RNA stability over the functional lifetime of a single, critical biological cue that governs embryonic patterning. Not only do the experiments provide quantification of these features, but also point to likely candidates (P-bodies) for gating bcd's translation in the narrow window between egg activation and cellular blastoderm. For the biophysically-inclined, this adds critical quantitative information of translational state that allows for further refining computational models for how this manifestation of a reaction-diffusion system actually comes together in a complex biological context.

The primary audience for this work will be the two groups above: developmental biologists and scientists interested in the quantitative modeling of biological phenomena.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

In this manuscript by Athilingam et al., the authors are studying the translation of the morphogen Bicoid (Bcd), which is in anterior-posterior patterning of the blastoderm Drosophila embryo. They have used an array of sunTag elements in the 5' UTR of Bcd to detect the localization of translation. They found that, not only is Bcd not translated until egg activation, but it can only be translated at the anterior pole, even though bcd mRNA has a broader spatial distribution.

In general, the paper uses a cutting-edge methodology to address one of the foundational questions of the best-studied morphogen gradient: namely, what is the spatial distribution of the Bcd source? Together with the dynamics of its spreading (which they addressed in a separate study in 2024) and Bcd degradation, their results point to a modified form of the synthesis/diffusion/degradation (SDD) model of Bcd gradient formation, which they have analyzed in the final subsection of the results. However, there are several major issues that erode the validity and impact of the paper, most of which can be put into the category of vague explanations, missing information, or contradictory statements, making it hard to understand/verify what conclusions can be drawn. This is also coupled with vague figures and captions. We describe these, and a few minor issues, in detail below:

• Line 114: The authors claim to have validated the SunTag using a fluorescent reporter, but do not show any data. Ref 60 is a general reference to the SunTag, and not the Bcd results in this paper. Perhaps place their data into a supplemental figure or movie?
• Line 128 and Fig. 1E: The claim that bcd becomes dispersed is difficult to verify by looking at the image. The language could also be more precise. What does it mean to lose tight association? Perhaps the authors could quantify the distribution, and summarize it by a length scale parameter? This is one of the main claims of the paper (cf. Line 23 of the abstract) but it is described vaguely and tersely here.
• Line 146, Fig. 1G: This is a really important figure in the paper, but it is confusing because it seems the authors use the word "translation," when they mean "presence of Bcd protein." In other places in the paper, the authors give the impression that "bcd translation" means translation in progress (assayed by the colocalization of GCN4 and bcd mRNA). However, in Fig. 1G, the focus is only on GCN4. Detecting Bcd protein only at the anterior does not mean that translation happens only at the anterior (e.g., diffusion or spatially-restricted degradation could be in play).

It would also be helpful to show a plot with quantification of Bcd detection (or translation) on the y-axis and a continuous AP coordinate on the x-axis, instead of just two points (anterior and posterior poles, the latter of which is uninteresting because observing no Bcd at the posterior pole is expected).

Another issue with Fig. 1G is that the A and P panels presumably have different brightness and contrast. If not, just from looking at the A and P panels, the conclusion would be that Bcd protein is diffuse (and abundant) in the posterior and concentrated into puncta in the anterior. The authors should either make the brightness and contrast consistent or state that the P panel had a much higher brightness than the A panel.

• Line 176: This section is very confusing, because at this point the authors already addressed the spatial localization of translation in Fig. 1G,H (see my above comment). However, here it seems the authors have switched the definition of translation back to "translation in progress." Therefore, the confusion here could be eliminated by addressing the above point.
• Line 185: The sentence here is seemingly contradictory: "most...within 100 microns" implies that at least some are beyond 100 microns, while the sentence ends with "[none]...more than 100 microns." The language could perhaps be altered to be less vague/contradictory.
• Line 204: It would be really nice to have quantification of the translation events, such as curves of rate of translation as a function of a continuous AP coordinate, and a curve for each nc.
• Line 209 and Fig 4C: The authors use the terms "intensity of translation events" or "translation intensity" without clearly defining them. From the figure (specifically from the y-axis label), it looks like the authors are quantifying the intensity per molecule (which is not clearly the same thing as "translation intensity"), but it would be nice if that were stated explicitly.

In addition, the authors again quantify only two points. This is a continuously frustrating part of the manuscript, which applies to nearly all figures where the authors looked only at two points in space. At a typical sample size of N = 3, it seems well within time constraints to image at multiple points along the AP axis.

Furthermore, it sounds like the authors are saying the "translation intensity" is the same in anterior and the posterior, which is counterintuitive. The expectation is that translation would be undetectable at the posterior end, in part because bcd mRNA would not be present. (Note that this expectation is even acknowledged by the authors on Line 185, which I comment on above, and also on Line 197). There should also be very low levels of Bcd protein (possibly undetectable) at the posterior pole. As such, the authors should explain how they think their claim of the same "translation intensities" in the anterior vs posterior fits into the bigger picture of what we know about Bcd and what they have already stated in the manuscript. They should also explain how they observed enough molecules to quantify at the posterior end. The authors should also disclose how many points are in each box in the boxplot. For example, the sample size is N = 3 embryos. In just three embryos, how many bcd/GCN4 colocalizations did the authors observe at the posterior end of the embryo?

• Line 215: The sentence that starts on this line seems self-contradictory: I cannot tell whether or not there is a difference in translation based on position.
• Line 229: Long-ranged is a relative term. From the graph, one could state there is some spatial extent to the mRNA gradient, so it is unclear what the authors mean when they say it is not "long-ranged." Could the mRNA gradient be quantified, such as with a spatial length scale? This would provide more information for readers to make their own conclusions about whether it is long-ranged.
• Line 230: When the authors claim the Bcd gradient is steeper earlier, a quantification of the spatial extent (exponential decay length scale) would be appropriate. Indeed, lambda as a function of time would be beneficial. It should also be placed in context of earlier papers that claim the spatial length scale is constant.
• Lines 235-236: The two sentences that start on these two lines are vague and seemingly contradictory. The first sentence says there is a spatial shift, but the second sentence sounds like it is saying there is no spatial change. The language could be more precise to explain the conclusions.

Minor issues/typos (still must be addressed for content):

• Line 81: Probably meant "evolutionarily conserved"
• Figure 1 legend: part B says "from 15 samples" but also says N = 20. Which is it, or do these numbers refer to different things?
• Line 217: migration of what?
• Line 228: "early embryo" is vague. The authors should give specific time points or nuclear cycle numbers.
• Line 301: Other locations in the paper say 75 microns or 100 microns.
• Fig. 5: all images should be oriented such that the dorsal midline is on the upper half of the embryo/image.
• Fig. 5B: There are light tan and/or light orange curves (behind the bold curves) that are not explained.
• Fig. 5C: the plot says "normalized" but nowhere do the authors describe what the curves are normalized to. There is also no explanation for what the broad areas of light color correspond to.

Referees cross-commenting

This is Reviewer 2. Yes, I am enthusiastic about the work: it is a much needed set of experiments and it fits well into the overall goal of quantitatively understanding the processes that establish the Bcd gradient. My main concern(s) about this paper is the loose and vague way they described their experiments and the interpretations. My hope is they will use the revision as an opportunity to more precisely explain their work.

Other than that, I am in agreement with the other reviewers on the need to revise for clarity and publish this important work.

Significance

The results, if upheld, are highly significant, as they are foundational measurements addressing a longstanding question of how morphogen gradients are formed, using Bcd (the foundational morphogen gradient) as a model. They also address fundamental questions in genetics and molecular biology: namely, control of mRNA distribution and translation.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

In this paper the authors use the Suntag system to visualise bcd mRNA translation in the Drosophila embryo. They elucidate the relationship between bcd mRNA translation and P body localisation. In the oocyte, bcd mRNAs are localised in P bodies and translationally repressed, but upon egg activation bcd mRNAs are released from P bodies and translated. In addition, during mid-nc14, bcd mRNAs become localised to embryonic P bodies and degraded. The authors use their data to modify the Synthesis, Diffusion, Degradation model of Bcd gradient formation, which recapitulates the Bcd gradient detected experimentally.

Overall, I think the data are of high quality and support the authors' conclusions. I only have minor comments, as follows:

Fig 1B - add arrows showing mRNAs being translated or not (the latter mentioned in line 113 is not so easy to see).

Fig 2A - add a sentence explaining why 1,6HD, 2,5HD and NaCl disrupt P bodies.

Fig 4C - explain in the legend what the white lines drawn over the image represent. And why is there such an obvious distinction in the staining where suddenly the DAPI is much more evident (is the image from tile scans)?

Line 215 - 'We did not see any significant differences in the translation of bcd based on their position, however, there appears an enhanced translation of bcd localised basally to the nuclei (Figure S5).' Since the difference is not significant, I do not think the authors should conclude that translation is enhanced basally.

Line 218: 'The interphase nuclei and their subsequent mitotic divisions appeared to displace bcd towards the apical surface (Figure S6B).' Greater explanation is needed in the legend to Fig S6B to support this statement as the data just seem to show a nuclear division - I would have thought an apical-basal view is needed to conclude this.

Fig 5B - the authors compare Bcd protein distribution across developmental time. However, in the early time points cytoplasmic Bcd is measured (presumably as it does not appear nuclear until nc8 onwards) and compare the distribution to nuclear Bcd intensities from nc9 onwards. Is most/all of the Bcd protein nuclear localised form nc9 to validate the nuclear quantitation? Does the distribution look the same if total Bcd protein is measured per volume rather than just the nuclear signal? Are the authors assuming a constant fast rate of nuclear import?

Line 259 - 'We then asked if considering the spatiotemporal pattern of bcd translation' - the authors should clarify what new information was included in the model. Similarly in line 286, 'By including more realistic bcd mRNA translation' - what does this actually mean? In line 346, 'We see that the original SDD model .... was too simple.' It would be nice to compare the outputs from the original vs modified SDD models to support the statement that the original model was too simple.

Fig S1A - clarify what the difference is between the 2 +HD panels shown.

Fig S2E - the graph axis label/legend says it is intensity/molecule. Since intensity/molecule is higher in the anterior for bcd RNAs, is this because there are clumps of mRNAs (in which case it's actually intensity/puncta)?

Fig S4 - I think this line is included in error: '(B) The line plots of bcd spread on the Dorsal vs. Ventral surfaces.' In B, D, E - is the plot depth from the dorsal surface? I would have preferred to see actual mRNA numbers rather than normalised mRNAs. In Fig S4D moderate, from 10um onwards there are virtually no mRNA counts based on the normalised value, but what is the actual number? The equivalent % translated data in Fig S4E look noisy so I wonder if this is due to there being a tiny mRNA number. The same is true for Figs S4D, E 10um+ in the low region.

Referees cross-commenting

I think the concerns raised by reviewers 2 and 3 are valid, and that it is feasible for the authors to address all the reviewers' concerns in order to improve the manuscript.

Significance

General assessment

Strengths are: 1) the data are of high quality; 2) the study advances the field by directly visualising Bcd mRNA translation during early Drosophila development; 3) the data showing re-localisation of bcd mRNAs to P bodies nc14 provides new mechanistic insight into its degradation; 4) a new SDD model for Bcd gradient formation is presented. Limitations of the study are: 1) there was already strong evidence (but no direct demonstration) that bcd mRNA translation was associated with release from P bodies at egg activation; 2) it is not totally clear to me how exactly the modified SDD model varies from the original one both in terms of parameters included and model output.

Advance

The advance is conceptual, technical and mechanistic.

Audience

The results will be important to a broad range of researchers interested in the formation of developmental morphogen gradients and the post-transcriptional regulation of gene expression, particularly the relationship with P bodies.

My expertise

Wetlab developmental biologist

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

__Reviewer #1 __

*This study "Interpreting the Effects of DNA Polymerase Variants at the Structural Level" comprises an in-depth analysis of protein sequence variants in two DNA polymerase enzymes with particular emphasis on deducing the mechanistic impact in the context of cancer. The authors identify numerous variants for prioritisation in further studies, and showcase the effectiveness of integrating various data sources for inferring the mechanistic impact of variants. *

*All the comments below are minor, I think the manuscript is exceptionally well written. *

*> The main body of the manuscript has almost as much emphasis on usage of the MAVISp tool as analysis of the polymerase variants. I don't think this is an issue, as an illustrated example of proper usage is very handy. I do, however, think that the title and abstract should better reflect this emphasis. E.g. "Interpreting the Effects of DNA Polymerase Variants at the Structural Level with MAVISp". This would make the paper more discoverable to people interested in learning about the tool. *

We have changed the manuscript title according to the reviewer’s suggestions, and the current title is “Interpreting the Effects of DNA Polymerase Variants at the Structural Level using MAVISp and molecular dynamics simulations.”

• *

*> Figure 1. I don't believe there is much value in showing the intersection between the datasets (especially since the in-silico saturation dataset intersects perfectly with all the others). As an alternative, I suggest a flow-chart or similar visual overview of the analysis pipeline. *

• *

We moved the former Figure 1 to SI. We decided to keep it at least in SI because it provides guidance on the number of variants relative to the total reported across the different disease-related datasets annotated with the MAVISp toolkit. On the other hand, the suggestion of a visual scheme for the pipeline followed in the analyses is a great idea. We have thus added Figure 1, which illustrates the pipeline workflows for analysis of known pathogenic variants and for discovery of VUS and other unknown variants, as suggested by the reviewer.

*> Please note in the MAVISp dot-plot figure legends that the second key refers to the colour of the X-axis labels rather than the dots *

We have revised the code that produces the dotplot so the second key is placed closer to the x-axis and clearer to read.

Missing figure reference (Figure XXX) at the bottom of page 16

We apologize for this mistake. Figures, contents, and the order have changed significantly to address all reviewers’ comments; this statement is no longer included. Also, we have carefully proofread the final version of the manuscript before resubmitting it.

__Reviewer #2 __

• *

This manuscript reports a comprehensive study of POLE and POLD1 annotated clinical variants using a recently developed framework, MAVISp, that leverages scores and classifications from evolutionary-based variant effect predictors. The resource can be useful for the community. However, I have a number of major concerns regarding the methodology, the presentation of the results.

*** On the choice of tools in MAVISp and interpretation of their outputs *

- Based on the ProteinGym benchmark: https://proteingym.org/benchmarks*, GEMME outperforms EVE for predicting the pathogenicity of ClinVar mutations, with an AUC of 0.919 for GEMME compared to 0.914 for EVE. Thus, it is not clear for me why the authors chose to put more emphasis on EVE for predicting mutation pathogenicity. It seems that GEMME can better predict this property, without any adaptation or training on clinical labels. *

• *

We appreciate this comment, but we should not exclude EVE entirely from our data collection or from VEP coverage under MAVISp, based on a difference in AUC of 0.005. It was not our intention to place more emphasis on EVE predictions, and we have revised it accordingly. We would like to clarify the workflow we use for applications of the MAVISp framework in “discovery mode,” i.e., for variants not reported as pathogenic in ClinVar. This relies on AlphaMissense to prioritize the pathogenic variants and then retain further only the ones that also have an impact according to DeMaSk, which provides further indication for loss/gain-of-fitness. DeMaSk nicely fits the MAVISp framework, as it was trained on data from experimental deep mutational scans, which we generally import in the EXPERIMENTAL_DATA module. We have revised the text to make this clearer. GEMME and EVE (or REVEL) can be used for complementary analysis in the discovery workflow. Other users of MAVISp data might want to combine them with a different design, and they have access to all the original scores in the MAVISp database CSV file and the code for downstream analysis to do so. The choice for our MAVISp discovery workflow is mainly dictated by the fact that we have noticed we do not always have full coverage of all variants in many protein instances for EVE, GEMME, and REVEL. In particular, since the reviewer highlights GEMME over EVE, GEMME is currently unavailable for a few cases in the MAVISp database. This is because we need to rely on an external web server to collect the data, which slows down data collection on our end.

Additionally, we have encountered instances where GEMME was unable to provide an output for inclusion in the MAVISp entries. When we designed the workflow for variant characterization in focused studies, we also made practical considerations. We are also exploring the possibility of using pre-calculated GEMME scores from

https://datadryad.org/dataset/doi:10.5061/dryad.vdncjsz1s, but we encountered some challenges at the moment that deserve further investigations and considerations. For example, MAVISp annotations rely on the canonical isoform as reported in Uniprot, which can lead to mismatches with the GeMME pre-computed scores. So far, we have identified a couple of entries whose canonical isoforms no longer match the one in the pre-computed GEMME score dataset. Another limitation is the absence of the original MSA files in the dataset, which we would need for a more in-depth comparison with the ones we used for our calculations. We are facing some challenges in reproducing the MSA output from MMseq2-based ColabFold protocol in this context that need to be solved first. Overall, the dataset shows potential for integration into MAVISp, but we need to define the inclusion criteria and compare it with the existing results in more detail.

Additionally, since the principle behind MAVISp is to provide a framework rooted in protein structure, AlphaMissense was the most reasonable choice for us as the primary indicator among the VEPs for our discovery workflow, and it has performed reasonably well in this case study and others.

Of course, our discovery design is one of the many applications and designs that could be envisioned using the data provided and collected by MAVISp. We also include all raw scores in the database's final CSV files, allowing other end users to decide how to use them in their own computational design. The design choice we made for the discovery phase of focused studies, using MAVISp to identify variants of interest for further studies, has been applied in other publications (see https://elelab.gitbook.io/mavisp/overview/publications-that-used-mavisp-data) in some cases together with experiments. It is also a fair choice for the application, as the ultimate goal is to provide a catalog of variants for further studies that may have a potentially damaging impact, along with a corresponding structural mechanism.

We have now revised the results section text where Table 1 is cited to clarify this. We also revised the terminology because we are using the VEPs' capability to predict damaging variants, rather than the pathogenic variants themselves. Experiments on disease models should validate our predictions before concluding whether a variant is pathogenic in a disease context, and we want to avoid misunderstandings among readers regarding our stance on this matter.

- Which of the predictors, among AM, EVE, GEMME, and DeMaSK, provide a classification of variants and which ones provide continuous scores? This should be clarified in the text. If some predictors do not output a classification, then evaluating their performance on a classification task is unfair. The MAVISp framework sets thresholds on the predicted scores to perform the classification and it is unclear from reading the manuscript whether these thresholds are optimal nor whether using universal cutoff values is pertinent. For instance, for GEMME, a recent study shows that fitting a Gaussian mixture to the predicted score distribution yields higher accuracy than setting a universal threshold (https://doi.org/10.1101/2025.02.09.637326*). Along this line, for predictors that do not provide a classification, I am not convinced of the benefit for the users of having access to only binary labels, instead of the continuous scores. The users currently do not have any idea of whether each variant is borderline (close to theshold) or confident (far from threshold). *

We agree with the reviewer, and this is due to us not being sufficiently clear in the manuscript. We have now revised the first part of the results to clarify this and to explain how we use the MAVISp data for application to focused studies, where the goal is to identify the most interesting variants that are potentially damaging and have a linked structural mechanism. Of course, there are other applications for leveraging the data in the database. We do offer scores to variants instead of just classification labels in the MAVISp csv file. They can be accessed, together with the full dataset, through the MAVISp website and reused for any applications.

Additionally, we used the scores in the revised manuscript for the VUS variant ranking (Figure 5), applying a strategy recently designed as an addition to the downstream analysis tool kit of MAVISp (​​https://github.com/ELELAB/MAVISp_downstream_analysis), thereby allowing the scores themselves to be taken into account. Also, in the final part of the manuscript, the VEP scores have been used to introduce the ACMG-like classification of the variants in response to reviewer 3 (Figure 9 and Tables S3-S4). We absolutely agree that it is informative to keep the continuous scores, and we have never overlooked this aspect. However, we also need a strategy with a simpler classification to highlight the most interesting variants among thousands or more to start an exploration. This is why we included the support with dotplots and lolliplots, for example. Our purpose here is to identify, among many cases, those with a potentially damaging signature (and thus we need a binary classification for simplicity). Next, we evaluate whether this signature entails a fitness effect (with DeMaSk), and finally, retain only the cases we can identify with a structural mechanism to study further.

The thresholds we set as the default for data analysis of dotplots in GEMME and DeMaSk are discussed in __Supplementary Text S3 __of the original MAVISp article. In brief, we carried out an ROC analysis against the scores for known pathogenic and benign variants in ClinVar with review status higher than 2. For applicative purposes, one could design other strategies to analyze the MAVISp data too; it is not limited to the workflow we decided to set as the primary one for our focused studies, as already mentioned above.

We have now also included classification based on the GMM model applied to GEMME scores for POLE and POLD1, so it can be evaluated against other designs for our protein of interest (see Table 1 in the revised version). The method section has been revised to include this part, and the ProteoCast pre-print is cited as a reference. We have not yet officially included this classification in the MAVISp database because we must first follow internal protocols to meet the inclusion criteria for new methods or analyses. We will do so by performing a similar comparison on the entire MAVISp dataset and focusing on high-quality variants, as ClinVar annotations, as we did to set the current thresholds for GEMME in Supplementary Table S3 of the original MAVISp article. We need to allocate time and resources to this pilot, which is scheduled for Q1 2026.

** On the presentation and impact of the results

• While reading the manuscript, it is difficult to grasp the main messages. The text contains abundant discussion about the potential caveats of the framework, the care that should be taken in interpreting the results, and the dependency on the clinical context. Although these aspects are certainly important, this extensive discussion (spread throughout the manuscript) obscures the results. Moreover, the way variants are catalogued throughout the text makes it difficult to grasp key highlights. The reader is left unsure about whether the framework can actually help the clinical practitioners.

We have revised the text to make it easier to read, including additional MD simulations of three variants of interest and more downstream analyses to clarify the mechanisms of action. We also added a recap of the most interesting variants and their associated mechanisms, along with the ranking of the variants using the different features available in the MAVISp csv file for the VUS. We hope that this makes it more accessible and valuable. In the original publication, Table 2 aimed to provide a summary of the interesting variants, and we have revised it now in light of the ranking results and the additional analyses that allow us to clarify the mechanisms of action further. We have also introduced__ Figure 9 and Tables S3 and S4__, which present data on ACMG-like classification for VUS that can fall into the likely pathogenic or benign categories.

• In many cases, the authors state that experimental validation is required to validate the results. Could they be more explicit on the experimental design and the expected outcome?

We have added a section on the point above at pages 21 and 30, where, alongside the summary of mechanisms per variant, we propose the experimental readouts to use based on known MAVE assays or assays that could be designed.

• AlphaMissense seems to tend to over-predict pathogenicity. Could the authors comment on that?

We are unsure whether this comment relates to our specific case or to a general feature of AlphaMissense.

In the latest iteration of our small benchmarking dataset for POLE and POLD1 (as shown in the paper), we achieve a sensitivity of 1 and a balanced specificity of 0.96 for AlphaMissense, which suggests that AlphaMissense does not over-predict pathogenicity very significantly in these proteins, predicting true negatives (i.e., non-pathogenic) mutations quite accurately. As performance was sufficient in our case, we deemed recalibrating the classification threshold for AlphaMissense unnecessary.

We are aware that this is not necessarily the case for every gene, e.g., it has been shown that AlphaMissense shows lower specificity in some cases (see e.g. 10.3389/fgene.2024.1487608, 10.1038/s41375-023-02116-3). This is also why we found it essential to evaluate its performance with its recommended classification on a gene-specific basis, as done here. In the future, we will keep a critical eye on our predictors to understand whether they are suitable for the specific case of study, or whether they require threshold recalibration or the use of a different predictor.

** On specific variants

• The mention of H1066R, H1068, and D1068Y is very confusing. There seems to be a confusion between residue numbers and amino acid types.

We have revised the text for typos and errors. This part of the text changed, so these specific variants are no longer mentioned.

• A major limitation of the 3D modeling is this impossibility to include Zn2+ coordination by cysteine residues. This limitation holds for both POLE and POLD1. Could the authors comment on the implication of this limitation for interpreting the mechanistic impact of variants. In particular, there are several variants reported in the study that consist in gain of cysteines. The authors discuss the potential impact of some of these mutations on the structural stability but not that on Zn coordination or the formation of disulphide bridges.

This is a great suggestion. We had, for a long time, a plan in the pipeline to include a module to tackle changes in cysteines. We have now used this occasion to include a new module that allows identifying mutations: 1) that are likely to disrupt native disulphide bridges and annotate them as damaging or 2) potential de novo formation of disulphide bridges upon a mutation of a residue to a cysteine, also annotated as damaging with respect to the original functionality. We also included a step that evaluates if the protein target is eligible for the analysis based on the cellular localization, since in specific compartments the redox condition (such as the nucleus) would not favour disulfide bridges. The module has been added to MAVISp, and we are collecting data with the module for the existing entries in the database to be able to release them at one of the following updates. More details are on the website in the Documentation section (https://services.healthtech.dtu.dk/services/MAVISp-1.0/). We could not apply the module to POLE and POLD1 since they are nuclear proteins, and it would not be meaningful to look into this structural aspect either in connection with loss of native cysteines or de novo disulfide bridge formation upon mutations that change a wild-type residue to a cysteine.

We would like to clarify that the structures we use, as it is a focused study rather than high-throughput data collection for the first inclusion in the MAVISp database, have been modelled with zinc at the correct position. It is just the first layer of high-throughput collection with MAVISp, which uses models without cofactors unless the biocurator attempts to model them or we move to collect further data for research studies (as done here). Prompted by this confusion, we have now added a field to the metadata of a MAVISp entry indicating the cofactor state. Nevertheless, the RaSP stability prediction does not account for the cofactor's presence, even when it is bound in the model. This is discussed in the Method Section. We thus did not further analyze the variants in sites directly coordinating the metal groups due to these limitations.

• MAVISp does not identify any mechanistic effect for a substantial portion of variants labelled as pathogenic. Could the authors comment on this point?

We are not sure how to interpret this question. It can be read two ways. Either the reviewer is asking about the known pathogenic ClinVar variants without mechanistic indicators, or more generally, the ones that we label “pathogenic” in discovery (we actually refer to more usually damaging in the dotplots), and for which we cannot associate a mechanism.

Overall, as a general consideration, it would be challenging to envision a mechanism for each variant predicted to be functionally damaging. For example, in the case of POLE and POLD1, we still lack models of complexes that did not meet the quality-control and inclusion criteria for the binding-free-energy scheme used by the LOCAL INTERACTION module. Also, when it comes to effects on catalysis or to analyzing effects in more detail at the cofactor sites, we could miss effects that would require QM/MM calculations. Other points we have not yet covered include cases related to changes in protein abundance due to degron exposure for degradation, which is one of the mechanistic indicators we are currently developing. Moreover, we used only unbiased molecular simulations of the free protein, and we would need future studies with enhanced sampling approaches and longer timescales to better address conformational changes and changes in the population of different protein conformational states induced by the mutation (including DNA). This can be handled formally by the MAVISp framework using metadynamics approaches, but it would be outside the scope of this work and is a direction for future studies on a subset of variants to investigate in even greater detail.

Furthermore, modifications related to PTM differ from phosphorylations. Anyway, our scope is to use the platform to provide structure-based characterization of either known pathogenic variants or potentially damaging ones predicted by VEPs, and focus on more detailed analyses of those. As we develop MAVISp further and design new modules, we will also be able to tackle other mechanistic aspects. This discussion, however, is more relevant to the MAVISp method paper itself.

Moreover, none of the variants discussed are associated with allosteric effect. Is this expected?

.

In general, allosteric mutations are rare. Nevertheless, in these case studies, the size of the proteins under investigation also poses some challenges for the underlying coarse-grain model used in the simple mode to generate the allosteric signalling map, as we have found it performs best on protein structures below 1000 residues

__Reviewer #3 (Evidence, reproducibility and clarity (Required)): __

The manuscript utilized the MAVISp framework to characterize 64,429 missense variants (43,415 in POLE and 21,014 in POLD1) through computational saturation mutagenesis. The authors integrate protein stability predictions with pathogenicity predictors to provide mechanistic insights into DNA polymerase variants relevant to cancer predisposition and immunotherapy response. There are discussions of known PPAP-associated variants and somatic cancer mutations in the context of known data and some proposed variants of interest (which are not validated).

Major comments:

I was unaware of the MAVISp framework. It concerns me that alebit this paper has a lot of technical details about the framework, its not the paper about the framework. I did look into the paper https://www.biorxiv.org/content/10.1101/2022.10.22.513328v5 which keeps benign updated (version five now) for three years, but I do not see a peer reviewed version. It would be unfair of me to peer review the underlying framework of the work but together with the previous comments, I am a bit concerned.

We have intentionally left the MAVISp resource paper as a living pre-print until we have sufficient data in the database that could be useful to the rest of the community. We have been actively revising the manuscript, thanks to comments from users in previous versions, to ensure it provides a solid resource. We had attempted approximately one and a half years ago a submission to a high-impact journal and even addressed the reviewers’ comments there. Still, we did not receive feedback for a long time, and ultimately, we were not sent to the reviewers again despite more than six months of work on our side. After that, we realized that we would benefit from collecting a larger dataset, and we invested time and effort in that and submitted again for revision, this time through Review Commons in the Summer of 2025. Anyway, the paper has been peer-reviewed by three reviewers through Review Commons. We submitted the revised version and response to reviewers, and it is now under revision with Protein Science. The reviewers’ comments and our responses can be found in the “Latested Referred Preprints” on the Review Commons website with the date of 17th of October 2025.

We would also like to clarify another point on this. In our experience, it is common practice to keep sofware on BioRxiv even for a long and to bring it to a more complete form in parallel with the community already applying it. This allows feedback from peers in a broad manner. We had similar experiences with MoonlightR, where the first publications with applications within the TCGA-PanCancer papers came before the publication of the tool itself, and the same has been for any of our main workflows, such as MutateX or RosettaDDGPrediction, which are widely used by the community. Finally, it can be considered that the MAVISp framework has already been used in different published peer-review studies (since 2023), attesting to its integrity and potential. Here, the reviewer can read more about the studies that used MAVISp data or modules: https://elelab.gitbook.io/mavisp/overview/publications-that-used-mavisp-data

For example, the authors are using AlphaFold models to predict DDG values. Delgado et al. (2025, Bioinformatics) explicitly tested FoldX on such models and concluded that "AlphaFold2 models are not suitable for point mutation ΔΔG estimation" after observing a correlation of 0.06 between experimental and calculated values. AlphaFold's own documentation states it "has not been validated for predicting the effect of mutations". Pak et al. (2023, PLOS ONE) showed correlation between AlphaFold confidence metrics and experimental ΔΔG of -0.17. Needless to say that these concerns seriously undermine the validity of a major part of the study.

We appreciate the reviewer’s comments and would like to clarify a point regarding the MAVISp STABILITY module, which we believe may have been misunderstood. Based on the studies cited by the reviewer, which critique the use of AF-generated mutant structures for assessing stability effects, we understand that this assumption may have led to the concern.

The STABILITY module utilises three in silico tools (FoldX, Rosetta, and RaSP) to assess changes in protein stability resulting from missense mutations. Importantly, the input to these assessments consists of AF models of the WT protein structures, not of AF-generated mutant structures. The mutants are generated using the FoldX and Rosetta protocols, along with estimates of the changes in free energy. For further details and clarification, we kindly refer the reviewer to the MAVISp original publication.

Also, one should consider the goal of our use of free energy calculations: not to identify the exact ΔΔG values, but to correlate with data from in vitro or biophysical experiments, such as those from cellular experiments like MAVE. We, other researchers, have shown that we have a good agreement in the MAVISp paper (case study on PTEN as an example in the original MAVISp publication and https://pmc.ncbi.nlm.nih.gov/articles/PMC5980760/ https://pubmed.ncbi.nlm.nih.gov/28422960/,10.7554/eLife.49138). Also, we had, before even designing the STABILITY module for MAVISp, verified that we can use WT structures from AlphaFold (upon proper trimming and quality control with Prockech) instead of experimental structure without compromising accuracy in the publications of the two main protocols of the STABILITY module (MutateX and RosettaDDGPrediction and a case study on p53, https://doi.org/10.1093/bib/bbac074,https://doi.org/10.1002/pro.4527). In the focused studies, we also carefully consider whether the prediction is at a site with a low pLDDT score or surrounded by other sites with a low pLDDT score before reaching any conclusions. The pLDDT score is reported in the MAVISp csv file exactly to be used for flagging variants or looking closer at them, as we discuss in this study (see, for example, Figure 2). Additionally, it should be noted that we employ a consensus approach across the two classes of methods in MAVISp to account for their limitations arising from their empirical energy function or backbone stiffness. Furthermore, in the focused studies, we also collected molecular dynamics simulations for the ensemble mode and reassessed the stability on different conformations from the trajectory to compensate for the issues with backbone stiffness of FoldX, RaSP, and Rosetta ΔΔG protocols.

I have to add that this is also true for the technical choices: Several integrated predictors (DeMaSk, GEMME) are outperformed by newer methods according to benchmarking studies (https://www.embopress.org/doi/full/10.15252/msb.202211474). AlphaMissense, while state-of-the-art, shows substantial overcalling of pathogenic variants. could ensemble meta-predictors (REVEL, BayesDel) improve accuracy?

The MAVISP framework includes REVEL as one of the VEPs available for data analysis. In this way, we were representing one of the ensemble meta-predictors. This is explained in the MAVISp original paper. We were not aware of BayesDel, which we will consider for one of the next pilot projects to assess new tools for the framework (see more details below on how we generally proceed). Currently, we cannot use REVEL for all variants because we do not necessarily have genomic coordinates for them. We retrieve genomic-level variants corresponding to our protein variants from mutation databases, where available (e.g., ClinVar, COSMIC, or CbioPortal). However, as we strive to cover every possible mutation, several of the variants in MAVISp are not in the database, which means we do not have the corresponding genomic variation for those, limiting our ability to annotate them with VEPs. In the future (see GitHub issue https://github.com/ELELAB/cancermuts/issues/235), we will revise the code to identify the genomic variants that could give rise to each protein mutation of interest, thereby increasing the coverage of VEP annotations.

We can see from the work cited by the reviewer that ESM-1v, EVE, and DeepSequence are among the top performers, whereas reviewer 2 cited another work in which GEMME outperforms EVE. We have been covering all of them, except ESM-1v, in our framework. We are planning to evaluate for inclusion in MAVISP some of the new top-performing predictors, including ESM-1v, in Q2 2026 (according to the protocol described later in this answer), which is why it is not available yet.

In our discovery protocol (i.e., when we work on VUS or variants not classified in ClinVar), we generally use AlphaMissense as the first indicator of potentially damaging variants. EVE, REVEL, or GEMME could be used in the case that AlphaMissense data are missing or as a second layer of evidence in the case we want, for example, to select a smaller pool of variants for experimental validation in a protein target with too many uncharacterized variants and too many that pass the evaluation with our discovery workflow. Finally, we rely on DeMaSk, as it also provides information on possible loss- or gain-of-fitness signatures to further filter the variant of interest for the search of mechanistic indicators. Since the MAVISp framework is modular, other users may want to use the data differently and design a different workflow. They have access to them (scores and classifications) through the web portal. The fact that we combine AlphaMissense with DeMaSk could yield final results after further variant filtering and mitigate the issue that AlphaMissense risks over-predicting pathogenicity.

In general, we work to keep MAVISp up-to-date, and we have developed a protocol for the inclusion of new methodologies in the available module before generating and releasing data with new tools in the database. In particular, we perform comparative studies using data already available in the database to evaluate the performance of new approaches against that of the tools already included. Depending on the module, we use different golden standards that we are also curating in parallel, and it would make sense to apply for that specific module. For example, if the question is to evaluate VEP, we would compare it against ClinVar known variants with good review status. If the VEP performs better than the currently included ones, we can include it as an additional source of annotations and evaluate whether we could change the protocol for the discovery/characterization of variants. We operate similarly for the structural modules. For example, for stability, we are importing experimental data from MAVE assays on protein abundance and use them as a golden standard where we evaluate new approaches against the current FoldX and Rosetta-based consensus for changes in folding free energies. Instead, If we find evidence that suggests switching to a new method or integrating it would be beneficial, we will do so as a result of these investigations. An example of our working mode for evaluating tools for inclusion in the framework is illustrated by how we handled the comparison between RaSP and Rosetta in the MAVISp original article (Supplementary file S2) before officially switching to RaSP for high-throughput data collection. We still maintain Rosetta, especially in focused studies, to validate further variants classified as uncertain.

*Further, I found the web site of the framework, where I looked for the data on these models, rather user unfriendly. Selecting POLD1, POLD2, or POLE tells me I am viewing entries A2ML1, ABCB11, ABCB6 respectively, when I search for POL and then click: these are the first three entries of the table, bot the what I click on. displaying the whole table and clicking on POLD1, gets me to POLD1. However, when I selected "Damaging mutations on structure" I get "Could not fetch protein structure model from the AlphaFold Protein Structure Database". Many other features are not working (Safari or Chrome, in a Mac). That is a concern for the usability of the dataset. *

• *

We have been able to reproduce the bugs identified by the reviewer and have fixed them. The second was connected to recent updates on the AlphaFold Protein Structure Database. We are not really sure how to work and act on the “other features that are not working” due to lack of specificity in this comment. Still, we have worked to make the website more robust: the coauthors of this work and other colleagues in the MAVISp team have extensively tested it across different proteins and with various browsers and operating systems, and we have fixed all identified issues. We also have a GitHub repository where users can open issues to share problems they have been experiencing with the website, which we will fix as promptly as we can (https://www.github.com/ELELAB/MAVISp), as we do for any of the tools we develop and maintain. If the reviewer were to come across other specific problems with the website, we recommend to (anonymously) open issues on the MAVISp repository so that they can be described more in detail and dealt with appropriately.

This comment seems more related to the MAVISP paper itself than to the POLE and POLD1 entries. We have been doing several revisions to the web app to improve it over time. We are also afraid that the reviewer consulted it during one of these changes, and we hope it will be better now. For POLE and POLD1, the CSV files were, in any case, also available through the MAVISp website itself (https://services.healthtech.dtu.dk/services/MAVISp-1.0/), as well as in the OSF repository connected to this paper (https://osf.io/z8x4j/overview), in case the reader needed to consult them or as a reference for the analyses reported in this paper.

Albeit this is a thorough analysis with the existing tools, and the authors make some sparse attempts to put the mutants classification in context with examples, the work stays descriptive for know effects in literature, or point out that e.g. "further functional and in vitro assays are required". The examples are not presented in a systematic way, or in an appealing manner. Thus, what this manuscript adds to the web site is unclear. It is a description of content, which could be at least more appealing if examples woudl be more clearly outlined in a conceptual framework, and illustrated more consistently. For exmaple I read in the middle of mage 16 "One such example is the F931S (p.Phe931Ser) variant (Figure 5A)" and then I see "F931 forms contacts with D626, a critical residue for the coordination of Mg2+ which is essential for the correct orientation of the incoming nucleotide (Figure XXX)". Figure 5B is not XXX as this has just many mutations labeled. These issues are very discouraging. I woudl recommend to put much more effort in examples, put them in clearer paragraphs, and decribe results rather than the methodology. Doing both in an intemigled way, clearly does not work for me.

We have revised the storyline to make it more straightforward for the reader, focusing on the essential messages and avoiding excessive description in the results section, instead conveying the key points directly. We also included new simulation data on three variants and downstream analyses of other variants. We revised the section to focus less on methodologies and more on the actual biological results. We have also added a ranking approach for the VUS and an ACMG-like classification to facilitate the identification of the most important results.

Additionally, we included a summary Table (Table 2) and Figure 9 that present the main findings on the VUS, and we discussed in the text the possible associated experimental validation.

We also do not fully understand the reviewer’s comment “the work stays descriptive for know effects in literature”. We agree that we should make a better effort to write the results in a logical and easy-to-follow manner, without risking the reader getting lost in too many details, and with more dedicated subsections. However, the paper does not describe just known effects in the literature. We had, in the previous version, a section aimed at identifying mechanistic indicators for ClinVar-reported variants that are also (in some cases) functionally characterized. This is true, but it is the very first part of the results, and it is still adding structure-based knowledge to these variants. After this, we also reported predicted results with mechanisms for VUS and variants in other databases. We took the opportunity in this revised version to elaborate more on the results of the variants reported in COSMIC and cBioPortal.

We are afraid that we also do not fully understand the reviewer's comment on the fact that “Thus, what this manuscript adds to the website is unclear.” We have generated POLE and POLD1 data with the MAVISp toolkit in both ensemble and simple mode, and the whole pool of local interactions with other proteins and DNA, specifically for this publication. It should be acknowledged that we have generated new data in ensemble mode, which relies on all-atom microsecond molecular dynamics simulations, and additional modules for the simple mode, including calculations with the flexddg protocol of Rosetta, which is also computationally demanding, to provide a comprehensive overview of the effects of variants in POLE and POLD1. The two proteins were available in the database only in simple mode with the basic default modules, and the remaining data were collected during this research article. This can also be inferred by the references in the csv file of the ensemble mode, which refer only to the DOI of the pre-print of this article. This entails a substantial effort in computing and analysis. The website is the repository for data that researchers collect using the MAVISp protocols or modules; in our opinion, it cannot replace a research project. We designed the database to store the data generated by the framework for others to consult and use for various purposes (e.g., biological studies, preparing datasets for benchmarking approaches against existing ones, or using features for machine learning applications). The entry point in the database is the simple mode, along with some compulsory modules (VEPs, STABILITY, PTM, EFOLDMINE, SASA). After this initial entry point, a biocurator or a team of researchers can decide to expand data coverage by moving into the other modules. Still, at some point, one would need to design focused studies to have a comprehensive overview of the effects on specific targets, as we did here, or, for example, in the publication https://doi.org/10.1016/j.bbadis.2024.167260.

Furthermore, there are analyses here, especially in the simulations, that are not directly available from consulting the database; in these cases, one needs to use other resources beyond MAVISp to investigate further the mechanisms underlying the predicted mechanistic indicators. We also included simulations of mutant variants to validate the hypothesis further. And another example is the analysis of the effects on the splicing site that is not covered by a structure-based framework, such as MAVISp, but is still an essential aspect in the analysis of the variants' effects.

Will the community find this analysis useful?

The analysis provided here will be helpful, especially for researchers interested in experimental studies of these enzymes, because they have throughout the study an extensive portfolio of structural data to consult, including a ranked list of variants by class of effect. We originally started designing MAVISp because we realized it was needed by our experimental collaborators, both in cellular biology and in more clinical research, whenever they needed to predict or simulate variants, and we expanded the concept into a robust, versatile framework for broader use. Especially for those genes where extensive MAVE data are not available (as in this case), having a set of variants to test experimentally is crucial support, as it provides the potential mechanism behind the predicted damaging variant.

How many ClinVar VUS could be reclassified using MAVISp data under current ACMG/AMP guidelines?

• *

The ACMG/AMP variant classification guidelines, to the best of our knowledge, include computational evidence (PP3/BP4) and well-established functional studies (PS3/BS3). Because MAVISp provides multi-level mechanistic predictions derived from structural modelling, these data formally fall within the PP3/BP4 computational category. They cannot be used to reclassify ClinVar VUS independently under ACMG/AMP rules. This is not really the goal of our framework, which is to provide a structure-based framework for investigating potentially damaging variants predicted by VEPs. However, the suggestion of the reviewer is something we wanted to explore too in general with MAVISp data, and we failed because of a lack of time. We checked the requirements for PP3, BP4, and PM1 and developed a classifier for VUS reported in ClinVar, using MAVISp features in accordance with the ACMG/AMP guidelines. Using ClinVar pathogenic and benign variants with at least a review status of 1 for calibration, we obtained thresholds for all MAVISp-supported VEPs (REVEL, AlphaMissense, EVE, GEMME, and DeMaSk). These thresholds were then applied to all ClinVar VUS to determine PP3 (pathogenic-supporting) and BP4 (benign-supporting) evidence. In parallel, we constructed a PM1-like mechanistic evidence category that integrates MAVISp structural stability, protein–protein interactions, DNA interactions, long-range allosteric paths, functional sites, and PTM-mediated regulatory effects. Variants classified as damaging in MAVISp according to such criteria were assigned PM1-like support. These evidence tags provide mechanistic insight to support VUS classification for polymerase proofreading genes. The workflow and complete annotated VUS table are now included in the revised manuscript and in the OSF repository. Although these findings cannot formally reclassify variants under ACMG/AMP criteria, they provide prioritization for PS3/BS3 experimental validation and highlight variants that are likely to be reclassified once supporting functional evidence becomes available.

How do MAVISp predictions meet calibrated thresholds, as in https://genomemedicine.biomedcentral.com/articles/10.1186/s13073-023-01234-y* for the exonuclease domain of POLE and POLD1? *

• *

Mur et al. (Genome Medicine 2023) restricted their ACMG/AMP recommendations to the exonuclease domain (ED) because (i) nearly all known pathogenic germline variants in POLE/POLD1 cluster within the ED, (ii) the ED has a well-characterised structure–function architecture, and (iii) sufficient pathogenic and benign variants exist only within the ED to support empirical calibration. To mirror this approach, we performed the calibration workflow exclusively on ED variants (POLE residues 268–471; POLD1 residues 304–533). For these ED-restricted variants, we recalibrated all MAVISp-derived computational predictors (REVEL, AlphaMissense, EVE, GEMME, DeMaSk) using ClinVar P/LP and B/LB variants. We applied the resulting POLE/POLD1-specific thresholds to all ClinVar VUS within the ED. We also applied our PM1-like structural/functional evidence exclusively to ED variants. The results of this ED-specific analysis are now reported in the revised manuscript (Figure 9 Supplementary Tables S3 and S4), as also explained in the response to the previous question. This ensures that MAVISp predictions are applied in a manner that is consistent with the principles of Mur et al. and ACMG/AMP variant interpretation.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

The manuscript used the MAVISp framework to characterize 64,429 missense variants (43,415 in POLE, 21,014 in POLD1) through computational saturation mutagenesis. The authors integrate protein stability predictions with pathogenicity predictors to provide mechanistic insights into DNA polymerase variants relevant to cancer predisposition and immunotherapy response. There are discussions of known PPAP-associated variants and somatic cancer mutations in the context of known data and some proposed variants of interest (which are not validated).

Major comments:

I was unaware of the MAVISp framework. It concerns me that alebit this paper has a lot of technical details about the framework, its not the paper about the framework. I did look into the paper https://www.biorxiv.org/content/10.1101/2022.10.22.513328v5 which keeps benign updated (version five now) for three years, but I do not see a peer reviewed version. It would be unfair of me to peer review the underlying framework of the work but together with the previous comments, I am a bit concerned. For example, the authors are using AlphaFold models to predict DDG values. Delgado et al. (2025, Bioinformatics) explicitly tested FoldX on such models and concluded that "AlphaFold2 models are not suitable for point mutation ΔΔG estimation" afte observing a correlation of 0.06 between experimental and calculated values. AlphaFold's own documentation states it "has not been validated for predicting the effect of mutations". Pak et al. (2023, PLOS ONE) showed correlation between AlphaFold confidence metrics and experimental ΔΔG of -0.17. Needless to say that these concerns seriously undermine the validity of a major part of the study. I have to add tha this is also true for toher technical choices: Several integrated predictors (DeMaSk, GEMME) are outperformed by newer methods according to benchmarking studies (https://www.embopress.org/doi/full/10.15252/msb.202211474). AlphaMissense, while state-of-the-art, shows substantial overcalling of pathogenic variants. could ensemble meta-predictors (REVEL, BayesDel) improve accuracy?

Further, I found the web site of the framework, where I looked for the data on these models, rather user unfriendly. Selecting POLD1, POLD2, or POLE tells me I am viewing entries A2ML1, ABCB11, ABCB6 respectively, when I search for POL and then click: these are the first three entries of the table, bot the what I click on. displaying the whole table and clicking on POLD1, gets me to POLD1. However, when I selected "Damaging mutations on structure" I get "Could not fetch protein structure model from the AlphaFold Protein Structure Database". Many other features are not working (Safari or Chrome, in a Mac). That is a concern for the usability of the dataset.

Albeit this is a thorough analysis with the existing tools, and the authors make some sparse attempts to put the mutants classification in context with examples, the work stays descriptive for know effects in literature, or point out that e.g. "further functional and in vitro assays are required". The examples are not presented in a systematic way, or in an appealing manner. Thus, what this manuscript adds to the web site is unclear. It is a description of content, which could be at least more appealing if examples woudl be more clearly outlined in a conceptual framework, and illustrated more consistently. For exmaple I read in the middle of mage 16 "One such example is the F931S (p.Phe931Ser) variant (Figure 5A)" and then I see "F931 forms contacts with D626, a critical residue for the coordination of Mg2+ which is essential for the correct orientation of the incoming nucleotide (Figure XXX)". Figure 5B is not XXX as this has just many mutations labeled. These issues are very discouraging. I woudl recommend to put much more effort in examples, put them in clearer paragraphs, and decribe results rather than the methodology. Doing both in an intemigled way, clearly does not work for me.

Will the community find this analysis useful? How many ClinVar VUS could be reclassified using MAVISp data under current ACMG/AMP guidelines? How do MAVISp predictions meet calibrated thresholds as in https://genomemedicine.biomedcentral.com/articles/10.1186/s13073-023-01234-y for the exonuclease domain of POLE and POLD1? Such questions might undermien teh appear of the work and coudl been looked into.

Referee cross-commenting

I agree with all the comments raised by reviewer 2; she/he elaborates more on some issues I brought up too briefly (e.g. the choice of GEMME) while other issues that I made more comments about are also mentioned. I only want to note that the statement "A major limitation of the 3D modeling is this impossibility to include Zn2+ coordination by cysteine residues" is not accurate, as there are many 3D structure prediction tools and modeling tools that are capable og handling zinc ions coordinated by cysteines.

While I respect that Referee 1 is clearly more positive and less concerned by methodological issues, I note that while I agree that "The authors identify numerous variants for prioritisation in further studies" (albeit in a sparse and not well organised manner in my view), I am not convinced by the present manuscript that "the effectiveness of integrating various data sources for inferring the mechanistic impact of variants" is really shown: there are hypotheses generated, but none are tested, so the effectiveness of the approach remains to be proven in my view.

I still view this as a thorough study and a very brave attempt to be integrative and inclusive, but several methodological limitations and lack of concrete novel insight, seriously dampen my enthusiasm.

Significance

Strengths:

A very comprehensive analysis of POLE and POLD1 missense variants (64,429 total), approximately 600-fold more coverage than the ~100 experimentally characterized variants in the PolED database. The multi-layered MAVISp approach provides mechanistic interpretability beyond simple pathogenic/benign classifications, potentially valuable for understanding variant effects on stability, DNA binding, protein interactions, and allosteric communication. The clinical context is highly relevant given POLE/POLD1 roles in disease.

Limitations:

The methodological concerns were outlined above. No solid new insight examples in a validated manner. Examples of how the datasets can be really used are not well-organised as they appear in the context of the approach in perplexed manner.

Advance:

The advance is primarily technical and database-driven rather than conceptually novel. Scale, Multi-dimensional assessment, Mechanistic insight and consideration of Clinical framework integration is a clear advance.

Audience:

The audience is the POLDPOLE experts; I however doubt if clinical scientists will find the paper useful, especially in the context of the absence of a dedicated resource and the fact that the entried in the MAVISp web-toold are not easily and intuitively accessible and clinical requirements(eg Integration with ACMG/AMP classification frameworks) are not clearly met.

Reviewer expertise: I am a structural biologist with experience in structure analysis of experimental and predicted models, but no specific expertise or interest in polymerases.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

This manuscript reports a comprehensive study of POLE and POLD1 annotated clinical variants using a recently developed framework, MAVISp, that leverages scores and classifications from evolutionary-based variant effect predictors. The resource can be useful for the community. However, I have a number of major concerns regarding the methodology, the presentation of the results and the impact of the work.

On the choice of tools in MAVISp and interpretation of their outputs

• Based on the ProteinGym benchmark: https://proteingym.org/benchmarks, GEMME outperforms EVE for predicting the pathogenicity of ClinVar mutations, with an AUC of 0.919 for GEMME compared to 0.914 for EVE. Thus, it is not clear for me why the authors chose to put more emphasis on EVE for predicting mutation pathogenicity. It seems that GEMME can better predict this property, without any adaptation or training on clinical labels.
• Which of the predictors, among AM, EVE, GEMME, and DeMaSK, provide a classification of variants and which ones provide continuous scores? This should be clarified in the text. If some predictors do not output a classification, then evaluating their performance on a classification task is unfair. I would guess that the MAVISp framework sets thresholds on the predicted scores to perform the classification and it is unclear from reading the manuscript whether these thresholds are optimal nor whether using universal cutoff values is pertinent. For instance, for GEMME, a recent study shows that fitting a Gaussian mixture to the predicted score distribution yields higher accuracy than setting a universal threshold (https://doi.org/10.1101/2025.02.09.637326). Along this line, for predictors that do not provide a classification, I am not convinced of the benefit for the users of having access to only binary labels, instead of the continuous scores. The users currently do not have any idea of whether each variant is borderline (close to theshold) or confident (far from threshold).

On the presentation and impact of the results

• While reading the manuscript, it is difficult to grasp the main messages. The text contains abundant discussion about the potential caveats of the framework, the care that should be taken in interpreting the results and the dependency on the clinical context. Although these aspects are certainly important, this extensive discussion (spread throughout the manuscript) obscures the results. Moreover, the way variants are catalogued throughout the text makes it difficult to grasp key highlights. The reader is left unsure about whether the framework can actually help the clinical practitionners.
• In many cases, the authors state that experimental validation is required to validate the results. Could they be more explicit on the experimental design and the expected outcome?
• AlphaMissense seems to have a tendency to over-predict pathogenicity. Could the authors comment on that?

On specific variants

• The mention of H1066R, H1068, and D1068Y is very confusing. There seems to be a confusion between residue numbers and amino acid types.
• A major limitation of the 3D modeling is this impossibility to include Zn2+ coordination by cysteine residues. This limitation holds for both POLE and POLD1. Could the authors comment on the implication of this limitation for interpreting the mechanistic impact of variants. In particular, there are several variants reported in the study that consist in gains of cysteines. The authors discuss the potential impact of some of these mutations on the structural stability but not that on Zn coordination or the formation of disulphide bridges.
• MAVISp does not identify any mechanistic effect for a substantial portion of variants labelled as pathogenic. Could the authors comment on this point? Moreover, none of the variant discussed are associated with allosteric effect, is this expected?

Referee cross-commenting

I agree with the comments and overall assessment of Reviewer 3. I would like to take this opportunity to clarify that I did not meant 3D modelling of Zinc ion coordination by Cys is impossible in general. I wanted to emphasise that the exclusion some Zinc-binding sites in the present study is a limitation.

Significance

The work's strength is its comprehensive analysis. The weaknesses are a methodology that does not seem mature and with output that are still difficult to predict. In addition, it seems that a lot of expertise and manual curation based on metadata (phenotype, functional state...) is needed for the users to benefit from the analysis. The manuscript reads a bit like a catalogue from where it is difficult to understand to what extent the results are significant and impactful.

I have expertise in computational modelling, protein sequence-structure-function relationship and prediction of variant effects.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

This study "Interpreting the Effects of DNA Polymerase Variants at the Structural Level" comprises an in-depth analysis of protein sequence variants in two DNA polymerase enzymes with particular emphasis on deducing the mechanistic impact in the context of cancer. The authors identify numerous variants for prioritisation in further studies, and showcase the effectiveness of integrating various data sources for inferring the mechanistic impact of variants.

All the comments below are minor, I think the manuscript is exceptionally well written.

• The main body of the manuscript has almost as much emphasis on usage of the MAVISp tool as analysis of the polymerase variants. I don't think this is an issue, as an illustrated example of proper usage is very handy. I do however, think that the title and abstract should better reflect this emphasis. E.g. "Interpreting the Effects of DNA Polymerase Variants at the Structural Level with MAVISp". This would make the paper more discoverable to people interested in learning about the tool.
• Figure 1. I don't believe there is much value in showing the intersection between the datasets (especially since the in-silico saturation dataset intersects perfectly with all the others). As an alternative, I suggest a flow-chart or similar visual overview of the analysis pipeline.
• Please note in the MAVISp dot-plot figure legends that the second key refers to the colour of the X-axis labels rather than the dots
• Missing figure reference (Figure XXX) at the bottom of page 16

Significance

In addition to identifying a large number of variants in POLE and POLD1 that are good candidates for further investigation, this study acts as a showcase for how evidence from different sources can be combined in a context-dependent manner (in this case, cancer). In terms of limitations, the lack of structures for POLD1 proved a hindrance several times in this study, obscuring some potentially pathogenic variants.

This manuscript bears some similarity to the MAVISp paper (10.1101/2022.10.22.513328v6), which gives a number of brief examples of analyses that can be conducted with the pipeline. This paper differs in that it shows a full start-to-end analysis and goes into considerable detail about possible mechanisms of pathogenesis. The mechanistic detail and potential clinical relevance set this paper apart.

This paper would most likely interest those involved in VUS interpretation, both in a clinical and research capacity. Those specialised in DNA polymerase enzymes would also likely find this paper of interest. This paper may also be useful for future research into the impact of the highlighted variants.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

The authors do not wish to provide a response at this time.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Wiesner et al. use a combination of state-of-the-art imaging techniques to visualize the exocytosis of vesicles labeled with vamp2-phluorin. This work builds on previous findings of the group and aims to quantify vesicle exocytosis along the axon and relate it to the location of actin rings (MPS). Exocytosis indeed occurs in axonal and dendritic regions ; however, at a significantly lower rate than in presynaptic terminals. Exceptionally, the AIS shows a remarkably high exocytosis rate compared to other axonal regions. Perturbation of the MPS with swinholide increases the nonsynaptic release of vamp2-phluorin. The spots supporting exocytosis along the axon lack spectrin but are spatially segregated from regions used for CCP formation.

This work takes advantage of last-generation optical microscopy approaches to provide a quantitative analysis of exocytosis along the axon in nonsynaptic regions. Findings are solid, and the segregation of spots supporting exocytosis and endocytosis is intriguing. However, it is unclear to me whether the results obtained reflect a general mechanism or if they are biased by the experimental approach. Specifically, I have these major comments:

1) Use of the term "spontaneous." I understand that the term "spontaneous" refers to exocytosis that "just" occurs. But exocytosis cannot be evaluated without considering electrical activity. Vamp2-phluorin has been extensively used to investigate neurotransmitter release. Since spontaneous neurotransmitter release occurs in the absence of action potentials, it is important to know how the rates of exocytosis are affected after incubation with TTX. These experiments are necessary to show if vesicles are indeed released spontaneously or if they require the presence of action potentials.

2) The rates of spontaneous exocytosis are expressed in μm² /hour because these events are quite infrequent. According to the methods section, typical recording times are 5 minutes or less (lines 522-533). It would be more appropriate to express values per minute to establish comparisons with other works. The goal is to understand what sort of vesicles are being exocytosed. This is a key question that must be addressed before exploring other aspects such as the relationship to spectrin or endocytosis. If the authors can provide more information about the types of vesicles being exocytosed, this work becomes very relevant. Since I am aware of the technical difficulties associated with this, some suggestions are: use a vamp2-apex2-phluorin construct and confirm vesicle identity by EM, or, use iGluSnFR to confirm neurotransmitter release along axons.

Minor comments:

1) Since culture conditions promote synapse formation, could spontaneous exocytosis found along axons related to synapse formation? This aspect could be tested by co-staining with PSD-95 after fixation.

Significance

Significance:

This is state-of-the-art study of the cell biology of neurons. The works demonstrates that vesicles are exocytosed along the axon and describes the molecular characteristics of the cytoskeletal elements involved.

General assessment:

The main strength of the study is the quality and diversity of imaging approaches used. The main limitation is defining the type of vesicle being exocytosed. It is important to know if vesicles imaged contain neurotransmitters.

Advance:

This paper is technically sound and provides interesting new concepts about how exocytosis occurs in nonsynaptic regions.

Audience:

This paper is appropriate for an audience familiarized with cell biology or cellular and molecular neuroscience

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

The manuscript by Wiesner et al examines the non-synaptic exocytosis of vesicles in axon initial segment (AIS) as well as proximal and distal axons. Using VAMP2-pHluorin authors convincingly demonstrated that exocytosis occurs in AIS, along the axon shaft, cell body and dendrite. Upon perturbing membrane-associated periodic skeleton (MPS) pharmacologically, exocytic events at the AIS seem to increase, suggesting a potential inhibition of exocytosis at the baseline by MPS. To test whether exocytosis occurs in the area where the MPS is disorganized, authors developed a novel correlative live-cell/super resolution microscopy where exocytic events are identified live by HiLo imaging, followed by fixing the cells and imaging spectrins using SMLM platform and identified the repetitive spectrin map with SReD detector. Using this approach, they have identified that exocytosis in proximal and distal axons occurs at the membrane area with less spectrin. This area is distinct from the clathrin-enriched area where the group has previously identified as the endocytic sites. The strength of the paper lies within the imaging techniques. However, for publication, the following concerns should be addressed.

Major concerns

1) The absence of spectrin mesh. Unlike their previous paper using platinum replica EM where filaments are clearly visible, they are using the antibodies against one end of the spectrin, and therefore, they can visualize the periodic distribution of spectrin ends but cannot visualize its meshwork in this study. In addition to this limitation, at thicker processes, molecules below and above the focal plane may or may not be visible, potentially creating the spectrin-less area at the center. Thus, the conclusion regarding the absence of spectrin meshes at the exocytic sites is not well supported.

2) The rate of exocytosis among controls. The rate of exocytosis at the AIS does not match between Fig. 2C and 3B (1.08 vs 0.64 events/um2/hour). Although the increase in the rate by swinA is relative to the DMSO control, the rate in swinA-treated neurons can be said similar to the control in Fig. 2C. So it is equally likely that DMSO is affecting the rate, rather than swinA. They need an additional control group with no treatment.

3) Alignment of pHluorin with SMLM images. Since the interpretation depends highly on the perfect alignment of live-cell images with SMLM data and fixation can alter the ultrastructure [PMC7339343], using internal structures like mitochondria as fiducials would be more helpful. However, adding discussion would suffice.

Minor concerns

1) The data presented here do not support the claim that actin perturbation favors non synaptic exocytosis (line 205). Please revise the sentence.

Significance

General assessment: The strength of the paper is in the imaging techniques. Visualizing the exocytic sites along the axons relative to the MPS is novel. The limitation of the paper is the lack of an approach to fully visualize spectrin networks.

Advance: If they can provide more convincing data demonstrating that exocytic sites are devoid of the spectrin meshwork, this paper will establish a novel concept regarding how non-synaptic exocytosis occurs along the axon.

Audience: Researchers working in the neuronal cell biology field will be the main audience of the manuscript.

Reviewer's expertise: Neuronal cell biology, exocytosis and endocytosis, and imaging

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Summary

This manuscript investigates non-synaptic exocytosis along the axon shaft and examines how the submembrane actin-spectrin skeleton shapes the distribution of exocytic sites. Using cultured hippocampal neurons expressing VAMP2-pHluorin, the authors map spontaneous exocytosis along axons and apply a correlative live-cell/super-resolution imaging workflow to visualize the nanoscale organization of spectrin gaps relative to exocytic hotspots. They report that axonal shaft exocytosis is enriched at the AIS, that perturbing the actin-spectrin lattice alters shaft exocytosis, and that exocytic sites generally correspond to spectrin-free regions. The methodological quality and imaging data are excellent and represent a strength of the study.

Major comments

The manuscript presents compelling imaging, but several major claims require additional experimental evidence or clarification. The most critical issue concerns the distinction between synaptic and non-synaptic exocytic events. In Figure 1, synaptophysin is used to define synapses, but this is insufficient since synaptophysin is a presynaptic marker and does not confirm the presence of a postsynaptic compartment. The classification of exocytic events as synaptic, therefore, requires co-localization with postsynaptic markers such as PSD95 or Homer. Without this, the paper's main conceptual distinction is not fully supported. Figure 6 requires revision because endocytosis needs to be assessed using a synaptic vesicle-specific assay. A synaptotagmin luminal-domain antibody uptake experiment is recommended, as it would allow precise identification of bona fide SV recycling. This is essential to conclude whether the reported endocytic events reflect synaptic vesicle turnover. The nature of the vesicles undergoing exocytosis along the shaft and at the AIS also remains unresolved. It will be essential to determine whether the authors are observing exocytosis of synaptic vesicles (e.g., VGLUT-positive) or large dense-core vesicles (e.g., BDNF-containing). This can be addressed straightforwardly using available pHluorin-tagged constructs (VGLUT-pHluorin, BDNF-pHluorin). Calcium dependence of the AIS exocytic events should be evaluated. Experiments removing extracellular calcium, blocking voltage-gated calcium channels, or depolarizing neurons (for example, with elevated KCl) would clarify whether these correspond to classical calcium-triggered SV fusion. These requested experiments are realistic in scope and can generally be completed in a few weeks. The imaging, analysis, and methodological descriptions are of high quality, although more information on replication, sample size, and statistical treatment would improve reproducibility.

Minor comments

Clarification of the criteria used to classify spectrin gaps versus clathrin clearings would be helpful. Some figure legends require more detailed acquisition parameters. A clearer description of the image registration and alignment steps in the correlative pipeline would improve transparency. Prior literature on non-synaptic axonal exocytosis and on AIS trafficking could be cited more extensively. The figures are generally high quality, and a schematic summarizing the main findings might help readers.

Referees cross-commenting

Our reviews are pretty consistent overall, I think. Major requests relate to calcium/depolarization dependency, and I would like to insist on the synaptic vs extra-synaptic and SV vs LDCV issues.

Significance

General assessment: This is a technically strong study addressing an interesting and timely question in neuronal cell biology. The imaging quality and methodological innovation are clear strengths. Significant limitations include insufficient distinction between synaptic and non-synaptic release, lack of characterization of vesicle identity, and unclear calcium dependence of AIS exocytosis. Addressing these points will significantly improve the rigor and impact of the study. Advance: The work provides new insights into the spatial organization of axonal exocytosis and its relationship to the actin-spectrin skeleton. The correlative imaging pipeline is valuable. However, the conceptual advance depends on resolving the synaptic/non-synaptic distinction and identifying the vesicle populations involved. Audience: The study will primarily interest a specialized audience in cellular neuroscience, membrane trafficking, and axonal biology. With the recommended revisions, it will also appeal to a broader neurobiology readership interested in nanoscale cytoskeletal organization, synaptic physiology, and axonal signaling.

Field of expertise: neuronal membrane trafficking, synaptic vesicle cycling, autophagy and endolysosomal pathways, cytoskeletal organization. I do not claim expertise in advanced optical engineering, but feel comfortable evaluating the biological interpretations and trafficking mechanisms.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

We were very pleased to see the very positive evaluation of our work by all 3 reviewers and appreciate their constructive comments and suggestions. We have now addressed all reviewers’ comments by making changes and clarifications to the manuscript.

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

In the present manuscript, the authors present an in-depth study on the effect of a heat-shock response on the ability of yeast to regain viability after quiescence when their ability to respire is inhibited. They nicely demonstrate that these effects correlate with the measured diffusion coefficients, providing deeper insight into the (at least partially) responsible environmental stress response and the molecular players involved. This work is an important contribution to the growing (or resurging) field of the physical properties of the cell.

We thank this reviewer for this very positive evaluation.

My two main comments are the following:

• The authors determine the diffusion coefficients from the MSD, as well as further analyze them all the way up to the confinement size. As far as I can judge from the manuscript, these analyses are for 2D systems and were initially developed for processes on membranes. How does this change for 3D systems? I understand that for a straightforward qualitative comparison of apparent MSD, this assumption is acceptable, but it may deviate more strongly with the additional analyses the authors present.

This is indeed an important point, and the reviewer is correct that the trajectories are analyzed in 2D (x,y) while the cytoplasm is a 3D environment. We fully agree that this requires careful interpretation, particularly for metrics beyond the short-lag diffusion coefficient.

First, for the diffusion coefficient, it is well established that for isotropic 3D motion the movements in all three dimensions are independent of each other and the projected 2D MSD satisfies:

= 4*D*τ

Thus, estimating from the short-lag slope of the 2D MSD yields the correct diffusivity of the underlying 3D process (up to standard experimental corrections such as localization error and motion blur). This approach is therefore widely used in cytoplasmic SPT and GEM studies, including in yeast, and is not restricted to membrane diffusion [1, 2].

Regarding confinement-related metrics derived from longer time lags, we agree that these were originally developed and most rigorously interpreted for 2D systems. In our study, these quantities are intentionally used as effective in-plane (x,y) descriptors of particle motion rather than as a full reconstruction of a 3D confinement geometry. Mapping a 2D MSD plateau to an absolute 3D confinement size depends on assumptions about geometry and isotropy and cannot be done uniquely without full 3D tracking. Nevertheless, MSD-based analyses have been successfully extended to explicitly model and quantify 3D confined diffusion in previous studies, provided that full 3D trajectories or well-defined confinement geometries are available. [2, 3]

[1] Gómez-García, P.A., Portillo-Ledesma, S., Neguembor, M.V., Pesaresi, M., Oweis, W., Rohrlich, T., Wieser, S., Meshorer, E., Schlick, T., Cosma, M.P., Lakadamyali, M., 2021. Mesoscale Modeling and Single-Nucleosome Tracking Reveal Remodeling of Clutch Folding and Dynamics in Stem Cell Differentiation. Cell Rep. 34. https://doi.org/10.1016/j.celrep.2020.108614

[2] Delarue, M., Brittingham, G.P., Pfeffer, S., Surovtsev, I. V., Pinglay, S., Kennedy, K.J., Schaffer, M., Gutierrez, J.I., Sang, D., Poterewicz, G., Chung, J.K., Plitzko, J.M., Groves, J.T., Jacobs-Wagner, C., Engel, B.D., Holt, L.J., 2018. mTORC1 Controls Phase Separation and the Biophysical Properties of the Cytoplasm by Tuning Crowding. Cell 174, 338-349.e20.

[3] Lerner, J., Gómez-García, P.A., McCarthy, R.L., Liu, Z., Lakadamyali, M., Zaret, K.S., 2020. Two-parameter single-molecule analysis for measurement of chromatin mobility. STAR Protoc 1.

Importantly, we do not assume perfect isotropy of the yeast cytoplasm. Local anisotropies are expected due to organelles, crowding heterogeneity, and cell geometry. However, the system is sufficiently close to isotropic at the length and time scales probed that the extracted confinement radius is highly reproducible across independent biological replicates. In our experiments, we observe consistent radius of confinements across three replicates, indicating that any bias introduced by partial anisotropy or projection into 2D is systematic and small.

Based on the observed reproducibility and the finite depth of field of our measurements (~100 nm), we estimate that potential errors in the absolute values of confinement-related parameters arising from 2D projection and incomplete isotropy are on the order of We have now clarified this point explicitly in the Methods section, emphasizing that confinement parameters are effective 2D measures, that the cytoplasm is not assumed to be perfectly isotropic, and that the conclusions rely on consistent, comparative measurements obtained under identical imaging and analysis conditions. The updated Methods paragraph is as follows:

[…] Trajectory analysis: Radius of Confinement

The radius of confinement was obtained only for the subgroup of confined trajectories. It quantifies the degree of confinement by estimating the radius of the 2D area explored by the particle in the imaging plane, which serves as a proxy measurement for the 3D volume that it explores. It was measured by fitting a circle-confined diffusion model to the TE-MSD (ensemble of all trajectories) (Wieser and Schütz, 2008).

TE-MSD = R^2 * (1 - exp(-4*D*t_lag/R^2)) + O

where R is the radius of confinement and D is the diffusion coefficient at short timescales. O is an offset value that comes from the localization precision limit inherent to localization-based microscopy methods.

Trajectories were analyzed in the imaging plane (x,y), and confinement metrics were therefore derived from 2D MSDs. Although particles diffuse in a three-dimensional cytoplasmic environment, projection onto 2D does not bias estimation of the short-lag diffusion coefficient for isotropic motion, since the projected MSD follows ⟨Δr_xy²(τ)⟩ = 4Dτ. However, confinement-related parameters derived from longer lag times should be interpreted as effective in-plane descriptors of mobility rather than as a direct reconstruction of a full 3D confinement geometry. Mapping a 2D MSD plateau to an absolute 3D confinement size would require explicit assumptions about geometry or full 3D tracking. Our conclusions rely on comparative analyses performed under identical imaging and analysis conditions, and the extracted confinement radii were highly reproducible across biological replicates, indicating that any bias introduced by 2D projection or moderate anisotropy is systematic and does not affect the validity of the relative differences reported.

• The authors show data in the supporting information where the GEMs provide larger foci after stress with longer imaging times. Could the authors provide the images of the shorter imaging times that they use? That seems a more equal comparison than Figure C. It is also unclear to me why fixed cells are used in Figure C, as well as the meaning of the x-axis. In line with this, can the authors exclude that GEMs dimerize/oligomerize after stress, and therefore display a lower diffusion coefficient?

We are happy to include the images acquired at a shorter time interval and have done so (Fig S2A). We apologize for insufficiently explaining the GEM intensity experiment shown in Figure S2C. The fixation was done to immobilize the GEMs, since they are rapidly diffusing in live cell imaging and the diffusion speed relative to camera exposure time will impact the brightness (any movement of a particle during exposure causes the signal on the detector to become “blurred” and reduces the intensity per pixel). Hence, GEM brightness does not solely reflect the monomer or potential aggregate/multimer state, but is also affected by diffusion speed and exposure time: faster moving GEMs will generally appear dimmer than slower moving ones, since the signal detection during the acquisition time is reduced by the particle movement. Another effect is that, since GEMs are moving in live cell imaging, they have a probability of spatially overlapping, enhancing the signal levels of the single detected spots.

We have quantified the brightness distribution in the different conditions to detect aggregation or multimerization of GEMs, which we expect to be visible as a shoulder on the Gaussian curve. The x-axis shows the intensity which we have determined for each trajectory. We chose to assess GEM intensity in the frame with the highest intensity, and to take the “Total” intensity, meaning we sum up the intensity of the pixels within the Point Spread Function (PSF) of each localization in that frame.

To clarify these points, we have extended the description of this experiment in the Results and Methods sections:

Results:

[...] Additional evidence for this comes from the observation that imaging GEMs at a lower frame rate (i.e., longer exposure time of 100 ms) showed a uniformly diffuse signal in SCD, whereas distinct foci appeared under starvation conditions (Figures S2A and S2B). This might suggest that GEMs aggregate in starvation. However, imaging GEMs at a faster frame rate (used for SPT, 30 ms exposure time) shows GEMs freely diffusing in all conditions (Figure S2A). Furthermore, analyzing GEM particle intensities in fixed cells, to eliminate motion blur-induced intensity attenuation, showed uniform GEM brightness distributions in all conditions (Figure S2C). Rather than aggregates, the bright foci thus represent immobile, single GEM particles that are confined and appear brighter during long exposure times due to their confinement in low-diffusive compartments. [...]

Methods:

[...] Trajectory analysis: Track Total Intensity

To assess GEM brightness, we determined the intensity of each trajectory in fixed cells. Cell fixation eliminates the motion blur-induced intensity attenuation, which would otherwise confound the GEM brightness depending on the movement speed and confinement. For each individual particle trajectory, the frame with the highest signal intensity of the localized particle was determined and the sum of the pixel intensities of the particle in that frame was calculated as the “Track Total Intensity”. In fixed cells, the GEM intensities were comparable in all conditions (Figure S2C). All GEM intensity histograms show a single, bell-shaped distribution of intensities with no indication of several GEM particles aggregating into brighter foci. [...]

Other comments: - For the precision of the language, the authors equate ribosome content with macromolecular crowding, with the diffusion of the GEMs throughout, and this becomes more conflated in the discussion, where it is compared to viscosity and macromolecular crowding effects, e.g., translation. Is it macromolecular crowding, mesoscale crowding, nano-rheology, or ribosome crowding? What is measured precisely?

We agree that careful and consistent nomenclature is important and thank the reviewer for bringing this point to our attention. We believe our manuscript maintains the proper distinctions of the terms diffusion, crowding and viscosity. We refer to what we study with the GEM single-particle tracking consistently as “(cytoplasmic) diffusion”. In Figure 2, we add “crowding” as an additional term since we observe a change in ribosome concentration and we affect the cytoplasmic crowdedness with a hyperosmotic shock. Our in-depth analysis of the confined and unconfined trajectory diffusion suggested that the cytoplasm is not simply globally affected by crowding or viscosity, but contains regions or compartments that trap GEM. Apart from Figure 2, we do not use the term viscosity or crowding, and we only return to “crowding” in the Discussion, either in reference to the aforementioned experiments from Figure 2 (ribosome concentration, hyperosmotic shock) or when discussing studies from cited works.

However, we did not use the term “macromolecular crowding” consistently and simplified it to “crowding” in a few instances. To be more precise, we now specify “macromolecular crowding” instead of “crowding” wherever applicable; namely in the text referring to Figure 2, where we specifically assess macromolecular crowding.

• In the EM images, the ribosomes seem smaller after starvation. Is that correct, and how should we interpret this? Is this due to an increased number of monosomes?

This is an important point, and it indeed appears that in SCD some ribosomes are close together, potentially as polysomes. In SC, the ribosomes appear more distinctly separated from each other, which would be expected due to the polysome collapse that occurs in starvation. However, the apparent size of individual ribosomes is identical in both conditions. Unfortunately, the resolution is not good enough to accurately measure the sizes of the ribosomes and clearly determine their monomer/polysome state.

• The authors refer to recent work on how biochemical reactions, such as translation, are determined by the cytoplasm. There is some older work on this, see for example in bacteria https://doi.org/10.1073/pnas.1310377110, and also in vitro here DOI: 10.1021/acssynbio.0c00330

We thank this reviewer for pointing out these publications and have included them in this group of citations.

• On the section of correlating diffusion and survival outcomes (bottom page 12), it is mentioned that the lowered diffusion could enhance aggregation. However, literature indicates that the opposite is true in buffer; lower diffusion reduces aggregation (also nucleation is inversely proportional to the viscosity).

This is a valuable point and we have happily expanded on it in the Discussion section. It is true that chemical assays have demonstrated that higher viscosity and slower diffusion decrease nucleation and aggregate formation. However, in vitro studies that alter diffusion through crowding changes have revealed a complex relation between crowding and aggregation propensity. The basic idea is that the excluded volume effect decreases aggregation by stabilization of the more compact, folded state. But the opposite effect, precluded protein folding, has also been ascribed to the excluded volume effect. As of now, studies with different crowders (dextran, ficoll, PEG, etc.) demonstrated increased or reduced protein aggregation upon crowding [1, 2, 3, 4]. The variable effect on aggregation seems to be not only based on the protein that is studied, but also the properties of the crowder (charges, shape, size), the interaction of the crowder with the protein, and the mixture of crowders [5].

Even though the relationship between crowding and protein aggregation is complex, we speculate that lower diffusion in our more crowded cells could cause protein aggregation, because these starvation conditions are known to induce the formation of protein fibrils and the condensation of mRNA and proteins.

[1] Uversky, V.N., M. Cooper, E., Bower, K.S., Li, J., Fink, A.L., 2002. Accelerated α-synuclein fibrillation in crowded milieu. FEBS Lett. 515, 99–103. https://doi.org/10.1016/S0014-5793(02)02446-8

[2] Munishkina, L.A., Cooper, E.M., Uversky, V.N., Fink, A.L., 2004. The effect of macromolecular crowding on protein aggregation and amyloid fibril formation. J. Mol. Recognit. 17, 456–464. https://doi.org/10.1002/jmr.699

[3] Biswas, S., Bhadra, A., Lakhera, S., Soni, M., Panuganti, V., Jain, S., Roy, I., 2021. Molecular crowding accelerates aggregation of α-synuclein by altering its folding pathway. Eur. Biophys. J. https://doi.org/10.1007/s00249-020-01486-1

[4] Mittal, S., Singh, L.R., 2014. Macromolecular crowding decelerates aggregation of a β-rich protein, bovine carbonic anhydrase: a case study. J. Biochem. 156, 273–282. https://doi.org/10.1093/jb/mvu039

[5] Kuznetsova, I.M., Zaslavsky, B.Y., Breydo, L., Turoverov, K.K., Uversky, V.N., 2015. Beyond the excluded volume effects: Mechanistic complexity of the crowded milieu. Molecules 20, 1377–1409. https://doi.org/10.3390/molecules20011377

To be more precise, we have therefore extended our Discussion section. We believe part of this additional discussion fits better in an earlier section, where we specifically discuss how the cytoplasmic properties, and specifically crowding, have been linked to filament/condensate formation. The updated paragraphs are as follows:

[...] Additional cytoplasmic rearrangements occur upon energy depletion, including filament formation or the formation of biomolecular condensates (Narayanaswamy et al., 2009; Noree et al., 2010; Petrovska et al., 2014; Prouteau et al., 2017; Riback et al., 2017; Saad et al., 2017; Marini et al., 2020; Stoddard et al., 2020; Cereghetti et al., 2021) highlighting a broader reorganization of the cytoplasm that could further affect the diffusion of macromolecules. In turn, the amount of crowding might also influence the propensity to form condensates and filaments (Heidenreich et al., 2020). Interestingly, in vitro studies have demonstrated a complex, dual effect of crowding on protein fibrillation and aggregation, in suppressing or accelerating it (Uversky et al., 2002; Munishkina et al., 2004; Mittal and Singh, 2014; Biswas et al., 2021). This appears to be dependent not only on the protein of study, but the properties of the crowder (size, charge, shape) and the specific mixture of crowders (Kuznetsova et al., 2015). [...]

[...] By contrast, extremely low diffusion, as seen in the absence of respiration in glucose starvation, might irreversibly impair cellular functions due to limited movement of proteins and RNA in and out of certain compartments, cellular territories and condensates. Such a model is supported by our analysis of how lower diffusion is the result of confined spaces becoming more prevalent, creating compartments that can trap macromolecules. As previously mentioned, increased crowding and reorganization of the cytoplasm have been linked to condensation and fibril formation of proteins, and, in certain in vitro contexts, accelerated aggregation. This state of crowding-induced low diffusion might therefore enhance protein aggregation or preclude the refolding of damaged proteins, which could disrupt proteostasis and lead to toxic aggregates that are a hallmark of the aging process (López-Otín et al., 2013). Together, these effects on proteins, RNA and other macromolecules likely lead to loss of cell fitness and irreversible arrest of the cells, preventing their reentry into the cell division cycle. [...]

Reviewer #1 (Significance (Required)):

General assessment: Strengths: It is a comprehensive study that provides a wealth of information and insight into the intricacies of a field that has received considerable attention, and its views are evolving rapidly. Weaknesses: It may suffer from some overinterpretation of diffusion data. Advance: The significant advance is that the molecular response pathway and precise molecular players are connected to the biophysical response of cells to starvation/quiescence. The dependence of diffusion on starvation has received considerable attention (Jacobs-Wagner, Cell, 2014; the current authors in eLife, 2016; and more recent investigations by Holt, Delarue, and others). Still, the authors take the next step and demonstrate how quiescence, and particularly how the history of a cell affects it, correlates strongly with the diffusion. As far as I can tell, this is new. As mentioned, the molecular insights into the pathways are exceptionally strong from my perspective. From personal experience, this work is also very important for researchers outside of the field from a practical standpoint: Do your measurements change when you stress cells by walking to a microscope? And even if you incubate them there, your measurement outcome will change. In my experience, this is a crucial point, and the cell's history is often overlooked. Audience: Broad -- biophysicists, molecular biologists, cell biologists, biotechnologists. My field of expertise: Biophysics.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

This manuscript addresses an important and longstanding question in the field: how eukaryotic cells remodel themselves to enter and survive quiescence, particularly under nutrient limitation and energy stress. The authors combine tools from biophysics, proteomics, stress signaling, and functional genomics to propose that stress-induced cytoplasmic reorganization, rather than ATP availability per se, is critical for long-term survival when respiration is impaired. The topic is timely, the experiments are generally well executed, and the initial phenomenology is compelling. The paper begins with a set of clear and convincing figures that establish an interesting and biologically important phenotype: when cells are shifted into glucose starvation, they can survive long term only if respiration is functional. Blocking respiration with Antimycin A (AntA) severely compromises viability. One straightforward hypothesis is that this defect simply reflects a failure to generate sufficient ATP. The authors, however, show that a 30-minute heat shock (HS) before glucose withdrawal in the presence of AntA largely rescues survival, even though cellular ATP levels remain critically low. In parallel, they use very well-executed GEM single-particle tracking experiments to demonstrate that cytoplasmic particle mobility decreases markedly in glucose-starved, respiration-deficient cells, and that this diffusion defect is also rescued by the pre-HS, again without restoring ATP. Together, these initial figures strongly support the idea that stress-induced remodeling of the cytoplasm, rather than ATP levels per se, is a key determinant of whether cells can enter and maintain a viable quiescent state. The authors then propose that this protective effect of HS is mediated by induction of the environmental stress response (ESR) and by resulting changes in protein expression. To test whether new protein synthesis is required, they pre-treat cells with cycloheximide during the HS and recovery period. This treatment largely, although not completely, abrogates the beneficial effect of HS on survival and diffusion in AntA-treated, glucose-starved cells. This is a strong experiment and supports the idea that HS-induced synthesis of specific proteins is important for protection, while also hinting that some cycloheximide-insensitive or pre-existing components may contribute. To identify the relevant proteins, the authors turn to global proteomic analysis, comparing multiple conditions: glucose starvation (SC), heat shock followed by glucose starvation (HS SC), glucose starvation plus AntA (SC + AntA), and heat shock followed by glucose starvation plus AntA (HS SC + AntA), each at 1 and 20 hours. This is where, in my view, the story becomes significantly harder to follow. The text for Figure 3 relies almost entirely on GO term enrichment, with very little description of individual proteins or even basic quantitative summaries of the dataset. For example, the authors never clearly state how many proteins were robustly quantified, nor what fraction of the proteome that represents. Without this foundational information, it is difficult to evaluate the strength and generality of their conclusions. Related to this, the GO analysis in Figure 3F reports "significant" enrichment for categories such as ribosomes or translation, yet the underlying number of proteins making up these enrichments is not shown. From the volcano plots, it appears that only a very small number of proteins change in some conditions (e.g., SC 20 h), and yet GO terms appear with extremely strong q-values. This is confusing: how can such strong enrichment occur if only a handful of proteins are changing? At minimum, the authors should provide: • the number of significantly up- or down-regulated proteins in each comparison • the number of proteins contributing to each enriched GO category • the magnitude of the changes for these proteins Because the absolute number of significantly changing proteins appears small in several conditions, the current heavy reliance on GO analysis feels unwarranted and potentially misleading. In such cases, it would likely be more informative to list all differentially abundant proteins-either in supplementary materials or in a main-text table-and briefly describe the most relevant ones, rather than relying on broad category labels. Figure 3F, in particular, needs substantially more explanation. A related issue appears in Figure 3G (and the associated text), where the authors emphasize that the proteomic response to HS + AntA and the response to long-term glucose starvation are distinct. While this conclusion is plausible, the analysis also shows a subset of proteins that are upregulated in both conditions. These overlapping proteins may, in fact, represent the core protective module that enables survival in quiescence. The authors do not discuss these proteins at all; instead, they are effectively dismissed in favor of the "distinct responses" narrative. I encourage the authors to identify and discuss these overlapping proteins explicitly. Are they chaperones, proteasome components, antioxidant enzymes, or other classical stress-response factors? Even if the global proteomes differ, the overlapping subset could be highly informative about the minimal set of proteins required to stabilize the cytoplasm and support entry into quiescence. The SATAY screen is a major strength of the paper, as it moves from correlative proteomics to functional genetic analysis. The approach appears well-controlled, but key information is missing: How many unique insertions were obtained? Was the library saturating? What was the read distribution and coverage? The authors also discuss only a small subset of the screen hits. The volcano plots show many additional genes that are not addressed. What categories do these fall into? Are they informative about pathways beyond Ras/PKA and Msn2/4? Presenting a fuller analysis would strengthen the mechanistic interpretation. The parts of the SATAY analysis that are discussed are solid. The screen implicates the Ras/PKA signaling axis and Msn2/4 in survival under HS-preconditioned, respiration-deficient starvation, and the authors validate these hits with targeted survival assays. The correspondence between genetic perturbations and changes in cytoplasmic diffusion is an intriguing connection. However, the analysis stops short of identifying the downstream effector proteins that actually produce the biophysical benefits observed. The manuscript then returns to the idea that improved cytoplasmic diffusion and reduced confinement may be essential for survival. This is an appealing hypothesis, but the evidence remains correlative. It is still unclear whether biophysical rescue is the cause of improved survival or simply a downstream marker of a properly induced stress response. What remains missing is deeper integration of the proteomics and SATAY data to identify which proteins are likely responsible for the adaptive changes in cytoplasmic organization. Overexpression of promising candidates-such as chaperones or proteostasis factors found in the overlap between HS and long-term starvation responses-could help determine whether any single protein or small group of proteins can phenocopy the HS-induced rescue. Importantly, many of the comments above are intentionally broad: the manuscript does not simply require small clarifications but would benefit from substantial expansion and deepening of the analysis. The observations are compelling, but the mechanistic chain connecting ESR activation → proteomic remodeling → cytoplasmic biophysics → survival remains insufficiently developed in the current draft. Clearer quantitative reporting, fuller presentation of the data, and more thoughtful interpretation would significantly strengthen the manuscript.

We thank reviewer 2 for this very thoughtful evaluation of our manuscript. We agree that expanding the descriptions and analysis of the presented data will improve the manuscript. Importantly, we now provide the proteomics data and the SATAY screen in an accessible format as supplementary materials. We address the individual points below.

Summary of Major Issues That Need to Be Addressed • Quantitative clarity in the proteomics o State how many proteins were quantified. o Report the numbers of significantly changing proteins in each condition. o Identify the proteins underlying each GO term and provide effect sizes.

We have now included a supplemental table containing label-free protein abundances for all 3308 reproducibly quantified proteins across all nine conditions (Supplemental Table S4). In addition, we added a sentence to the main text specifying both the number of reproducibly identified proteins and the approximate coverage of the yeast proteome.

For the comparison of protein abundances between the different stress conditions and logarithmically growing SCD cells, we now indicate the number of significantly changed proteins in the legend of Figure 3E. Furthermore, we include a heatmap of standardized protein abundances for all proteins that were significantly changed in at least one stress condition (Supplemental File S1) and provide all pairwise comparison results in the supplemental table (Supplemental Table S5). This new Supplemental File S1 replaces the previous Supplemental File S1, which had a stricter cutoff, showing all proteins with an abundance change greater than 2 standard deviations.

The information requested by the reviewer regarding GO term analysis is indeed important and was missing in the original version. We now report, for each GO term, the number of proteins in the top or bottom 10% of differentially abundant proteins and provide the corresponding effect size, calculated as the ratio of the observed to expected hits (Figure 3F).

• Over-reliance on GO analysis o Provide explicit lists of differentially expressed proteins. o Indicate whether enrichment results are meaningful given the small number of hits.

We appreciate this reviewer’s comment and agree that the presentation of the proteomic data in Figure 3 relies strongly on GO term enrichment, with limited description of individual proteins. Our primary goal for the proteomic analysis was to characterize the cellular response to stress at a global level rather than to focus on individual proteins or stress-specific details. We therefore intentionally opted for a broader, more coarse-grained analysis to not overcomplicate the manuscript and maintain accessibility for a broad readership.

That said, we agree that the underlying data should be made fully accessible. We have therefore expanded the supplemental materials to include a heatmap of all proteins that were significantly changed in at least one condition (Supplemental File S1), as well as comprehensive tables reporting protein abundances and pairwise differences across all stress conditions (Supplemental Tables S4 and S5). These additions provide direct access to the protein-level data while preserving the clarity of the main text.

With respect to GO term analysis, to avoid overinterpretation driven by small protein sets and better comparability across different conditions, we always performed the GO enrichment based on the top and bottom 10% changed proteins. This is already stated in the legend of Figure 3F and in the Methods section. We have now added the key missing parameters of the analysis to Figure 3F (see response above). Given that the analysis identifies multiple GO terms generally associated with the environmental stress response and that these terms exhibit coordinated behavior across conditions (Figure S3A), we are confident that the conclusions drawn from this analysis are robust.

• Overlooked overlapping proteins o Analyze and discuss the subset of proteins upregulated both by HS and by long-term starvation. o These may represent the core factors enabling survival.

Indeed, we agree that the overlapping proteins that are observed in our Figure 3G analysis should be presented. Perhaps surprisingly, these proteins (Hxt5, Sps19, Atg8, Aim17, Put1, Fmp45, YNL194C) have diverse functions and have so far not been implemented in the environmental stress response.

In the Results section, we now mention and briefly discuss the four that are present in both time points of the HS SC +AntA condition. We now mention all of them in the figure legend.

The modified text from the Results section is as follows:

[...] Furthermore, the proteins that are enriched in long-term starvation (SC 20 h vs. SCD) and those enriched in pre-HS respiration-deficient starvation (HS SC +AntA 1 h vs. SCD; HS SC +AntA 20 h vs. SCD) are poorly correlated and there is only a small overlap of factors that are significantly upregulated in all conditions (Figure 3G). These proteins are Aim17, Put1, Fmp45 and YNL194C. Aim17 is a mitochondrial protein of unknown function and Put1 is a mitochondrial proline dehydrogenase. Fmp45 and YNL194C are paralogous membrane proteins involved in cell wall organization. Focusing on the broad proteomic adaptation, we looked at the Gene Ontology (GO) terms of the proteomic changes across all conditions, and observed that long-term starvation (SC 20) leads to the upregulation of a few groups of proteins, mostly involved in respiratory activity and rewiring of the metabolism (Figure S3A). [...]

We greatly appreciate the suggestion to do an overexpression experiment. However, the overlapping proteins are not significant hits in the SATAY, suggesting that they are individually not required for the survival rescue although their overexpression might benefit survival.

We have therefore chosen to keep a broad perspective on the proteomics results and investigate instead the SATAY results in more detail, since they inherently contain functional relevance to survival. Overall, we feel that the overexpression of those (individually or as a group) would extend beyond the scope of our current manuscript.

• SATAY analysis needs fuller presentation o Provide insertion numbers, coverage, and basic library statistics. o Discuss additional hits beyond the Ras/PKA/Msn2/4 pathways. o Integrate SATAY results more deeply with proteomics.

We have added the insertion numbers and genome coverage percentages to the Methods section as follows:

[...] SATAY Screen: Analysis and Plotting

Sequencing detected the following total unique transposon numbers: 690’935 (A1), 558’932 (HA1), and 359’935 (HA4d) unique transposons. The transposon insertions in the different genes yielded the following genome coverages: 96.3% (A1), 94.5% (HA1) and 89.3% (HA4). For each gene [...]

We now also provide the SATAY screen data as Supplemental Table S6.

In the Results section, we mention some additional hits from the SATAY screen (ribosome biogenesis, mitochondrial respiration) but then shift our focus to the ESR genes. We now add a comment to the ribosome biogenesis genes before going to the ESR:

[...] The screen revealed several highly significant gene disruptions that promote or impair the HS-mediated rescue of respiration-deficient, glucose-starved cells (Figure 4A, Supplemental Table S6). The most significant gene hits that impair survival in 4 d HS SC +AntA when disrupted are involved in a variety of cellular processes, including ribosome biogenesis (e.g., ARX1, BUD22, RRP6), mitochondrial respiration (e.g., CBR1, COX23, ETR1), and ESR (e.g., MSN2, PSR2, YAP1). Intriguingly, the ribosome biogenesis genes being crucial for survival suggests that new ribosomes might have to be produced to ensure proper translational response during the HS. Notable among the ESR genes are MSN2 and, less significantly scored, MSN4, the master regulators of the ESR. [...]

To deepen the discussion on the lack of overlap between the SATAY screen and the proteomics, we have added a sentence highlighting that the SATAY screen detected the main regulators of the ESR, and the proteomics revealed its downstream targets involved in proteostasis and other stress proteins, and therefore these two data sets do both point to the ESR as the crucial response behind the HS-induced rescue. The modified Discussion text is as follows:

[...] Furthermore, the signaling genes that scored highly in the SATAY screen are often regulated through their activity rather than their abundance. Plausibly, their downstream target proteins are differentially expressed, whereas disrupting the regulators themselves leads to strong survival phenotypes. Similar observations have been made in other stress conditions, where fitness-relevant genes showed little overlap with genes with upregulated expression (Birrell et al., 2002; Giaever et al., 2002). Nonetheless, the SATAY screen revealed the principal regulators of the ESR while the proteomic analysis detected many of the ESR downstream targets involved in proteostasis and oxidative stress, demonstrating a functional convergence on the ESR in both data sets. [...]

• Mechanistic depth remains limited o Clarify whether cytoplasmic biophysical rescue is causal or downstream. o Test whether overexpression of candidate proteins can mimic HS-induced protection. o Expand the discussion of potential mechanisms using insights from both datasets.

Indeed, the specific mechanism(s) that govern the cytoplasmic properties in our conditions are currently not known, preventing us from manipulating the cytoplasmic properties and confirming a causal relationship. To uncover the mechanisms, extensive follow-up studies on ESR genes and/or proteins would be required, going beyond the scope of this manuscript. Furthermore, our ongoing follow-up studies are pointing towards redundancy of some potential regulation of the cytoplasmic diffusion, further complicating the analysis.

The suggested overexpression experiment is addressed in a previous comment where the overlapping proteins are mentioned.

Reviewer #2 (Significance (Required)):

This manuscript addresses a fundamental and timely question in cell biology: how eukaryotic cells remodel themselves to enter and survive quiescence, particularly under conditions of nutrient depletion and compromised energy production. Although quiescence has been studied for decades, the mechanisms that link metabolic state, stress signaling, and the physical properties of the cytoplasm remain incompletely understood. This work brings together biophysical measurements, global proteomics, and unbiased genetic screening in an ambitious effort to illuminate how cells maintain viability when respiration-and thus efficient ATP generation-is disrupted. A key conceptual contribution of this study is the demonstration that ATP levels alone do not dictate survival during starvation. Rather, the ability of cells to mount an appropriate stress response and reorganize the cytoplasm appears to be crucial. The early figures provide compelling evidence that heat shock preconditioning can rescue both viability and cytoplasmic mobility in respiration-deficient cells, even when ATP remains low. This finding is notable because it challenges the widely held assumption that energy charge is the primary determinant of successful entry into quiescence. If strengthened by deeper mechanistic analysis, this insight could reshape how the field views energy stress and cellular dormancy. The identification of the Ras/PKA-Msn2/4 axis as a key regulatory node is also significant, as it connects quiescence survival to well-established nutrient and stress signaling pathways. The integration of a genome-wide SATAY screen adds functional depth and offers the potential to uncover specific downstream effectors that remodel the cytoplasm or stabilize cellular structures during prolonged stress. Finally, the manuscript touches on a concept that is gaining traction across many subfields of biology: that the biophysical state of the cytoplasm is a regulated and physiologically meaningful parameter, not merely a passive consequence of metabolic decline. Understanding how cells tune macromolecular crowding, diffusion, and spatial organization during quiescence could have broad implications beyond yeast, including in stem cell biology, microbial dormancy, cancer cell persistence, and aging. Overall, the questions addressed are important, and the study has the potential to make a meaningful conceptual contribution. However, realizing that impact will require clearer and deeper mechanistic analysis-particularly in the proteomics and SATAY sections-to convincingly identify the specific factors and pathways that mediate the cytoplasmic remodeling underlying survival.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

Summary. Yeast haploid cells enter quiescence during nutrient deprivation, undergoing major metabolic, transcriptional and biophysical changes. In particular, quiescent cells remodel their cytoplasm, increasing macromolecular crowding and reducing diffusion. Respiration is known to be essential for entry into quiescence and long-term survival.

In this study, the authors discovered that respiration is not intrinsically required for yeast to survive glucose-starvation-induced quiescence. In particular, they found that a short heat shock before starvation restores survival even in the absence of respiration (Antimycin A treatment), demonstrating that a stress-induced adaptation can bypass the respiratory requirement. This rescue occurs without ATP recovery and relies on de novo protein synthesis. This stress-induced adaptation also rescues quiescent-like biophysical properties of the cytoplasm (increased crowding) that are normally prevented in non-respiring cells, which are thought to be relevant for cell survival . Proteomics reveals that heat shock induces a distinct stress-response proteome enriched in proteostasis factors. A genetic screen reveals that Ras/PKA inhibition and Msn2/4 activation enable this protective reprogramming. Altogether this highlights the importance and complexity of stress adaptation to quiescence establishment.

This is an excellent paper in all aspects. I have no major points besides the data accessibility, below.

We thank this reviewer for this very positive evaluation.

Main comments. - It would be nice to have the MS data available as Excel files for the community, and uploaded to repositories such as PRIDE. Description of the MS data is a bit expedited to serve the purpose of the paper (clustering to evaluate the similarity of proteomic profiles between conditions, GO term enrichment) so having the full data available might help.

We agree that the MS data should be accessible. The label-free protein abundances for the reproducibly quantified proteins across all nine conditions (Supplemental Table S4) and the pairwise comparisons shown in Figure 3E (Supplemental Table S5) are now included as supplementary Excel files. The MS data is currently not on PRIDE but we will deposit it there upon publication of our manuscript.

• Same thing for the SATAY screen. The data is summarized in Fig 4B but I believe that the data should be provided.

We agree that the SATAY screen results should be accessible as well, and we have now included the data as Supplemental Table S6.

Minor comments and questions. -I believe that in graphs, the X axis should start at 0 to avoid confusion about the strength of the effect (eg. Fig 2B)

We thank reviewer 3 for pointing this out, and we have re-evaluated the axis limits of all plots. As suggested, we have adjusted the x-axis in Fig 2B to start at 0 to better highlight the strength of the effect. For our Radius of Confinement and %Confined Trajectories graphs, we believe adjusting the y-axis to start and end at the same values will allow better comparison across figures. However, we chose not to set those y-axes to start at 0, since our measurements lie in a range which is covered by these axes, and these plots would simply include blank space if set to start at 0.

-I found that using imaging of GEMs at low frequency to reveal cytoplasmic crowding heterogeneity very interesting. Quiescent cells are known to accumulate many "bodies" as discussed in the text, would any of those co-localize with GEM foci?

Indeed, the imaging at low frequency has revealed that fluorescently-tagged proteins might become trapped in certain regions of the cytoplasm, allowing their detection at conventional imaging frequencies. It is very likely that a similar effect occurs for other cytoplasmic “bodies”, which become visible not only through protein accumulation in a single body but also through low mobility. We have not performed any colocalization experiment with known “bodies” (such as P-bodies or stress granules). Therefore, we do not know if any stress-induced “bodies” are confined to the same spaces as GEMs. However, we would expect at best an incomplete colocalization based on the observation that glucose starvation-induced “bodies” are generally present in a higher percentage of cells than the GEM foci we observe, i.e. it is unlikely that all “bodies” overlap with a GEM focus. It might be interesting to perform such colocalization experiments in follow-up studies, but we feel that such an analysis would go beyond the current scope of this manuscript.

Reviewer #3 (Significance (Required)):

General assessment, advances in the field This is an excellent study. The key finding of this paper, ie. that heat shock can compensate for lack of respiration for entry into quiescence, challenges the current views on quiescence establishment. It describes an alternative program that contributes to cell viability upon C source depletion, with details on the proteomic changes occurring in this condition and some of the genetic basis of this pathway. The study is well designed and controlled, the conclusions are in line with the obtained results and very well discussed and placed in perspective. Experimentally, the authors combine several experimental approaches including live-cell single-particle tracking of GEM nanoparticles to quantify cytoplasmic diffusion, FIB-SEM ultrastructural imaging of the cytoplasm to measure macromolecular crowding, proteomics to map stress-induced protein changes and genome-wide SATAY transposon mutagenesis to identify genes required for survival in respiration-deficient cells. The limitations are: -we don't know how this stress program facilitates survival in the absence of restoration of ATP levels. The data suggest that protein homeostasis is involved (chaperones and proteasome up-regulated upon stress, reduced ribosomal and translation-associated proteins down-regulated in the absence of respiration) but the mechanism remains elusive. -the relationships between cytoplasmic crowding and quiescence establishment remain correlative. Yet, the authors provide another pathway to favour viability upon quiescence establishment (with HS) whose activation also displays an increased crowding and reduction of cytoplasmic movement, further consolidating this link. Both of these points are adequately discussed in the manuscript. None of these points should preclude publication of this study, in my opinion.

Audience. This study would be of interest to researchers in the field of quiescence, biophysics, proteostasis, stress response, nutrient signaling and yeast biology.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary.

Yeast haploid cells enter quiescence during nutrient deprivation, undergoing major metabolic, transcriptional and biophysical changes. In particular, quiescent cells remodel their cytoplasm, increasing macromolecular crowding and reducing diffusion. Respiration is known to be essential for entry into quiescence and long-term survival.

In this study, the authors discovered that respiration is not intrinsically required for yeast to survive glucose-starvation-induced quiescence. In particular, they found that a short heat shock before starvation restores survival even in the absence of respiration (Antimycin A treatment), demonstrating that a stress-induced adaptation can bypass the respiratory requirement. This rescue occurs without ATP recovery and relies on de novo protein synthesis. This stress-induced adaptation also rescues quiescent-like biophysical properties of the cytoplasm (increased crowding) that are normally prevented in non-respiring cells, which are thought to be relevant for cell survival . Proteomics reveals that heat shock induces a distinct stress-response proteome enriched in proteostasis factors. A genetic screen reveals that Ras/PKA inhibition and Msn2/4 activation enable this protective reprogramming. Altogether this highlights the importance and complexity of stress adaptation to quiescence establishment.

This is an excellent paper in all aspects. I have no major points besides the data accessibility, below.

Main comments.

• It would be nice to have the MS data available as Excel files for the community, and uploaded to repositories such as PRIDE. Description of the MS data is a bit expedited to serve the purpose of the paper (clustering to evaluate the similarity of proteomic profiles between conditions, GO term enrichment) so having the full data available might help.
• Same thing for the SATAY screen. The data is summarised in Fig 4B but I believe that the data should be provided.

Minor comments and questions.

• I believe that in graphs, the X axis should start at 0 to avoid confusion about the strength of the effect (eg. Fig 2B)
• I found that using imaging of GEMs at low frequency to reveal cytoplasmic crowding heterogeneity very interesting. Quiescent cells are known to accumulate many "bodies" as discussed in the text, would any of those co-localize with GEM foci?

Significance

General assessment, advances in the field

This is an excellent study. The key finding of this paper, ie. that heat shock can compensate for lack of respiration for entry into quiescence, challenges the current views on quiescence establishment. It describes an alternative program that contributes to cell viability upon C source depletion, with details on the proteomic changes occurring in this condition and some of the genetic basis of this pathway. The study is well designed and controlled, the conclusions are in line with the obtained results and very well discussed and placed in perspective. Experimentally, the authors combine several experimental approaches including live-cell single-particle tracking of GEM nanoparticles to quantify cytoplasmic diffusion, FIB-SEM ultrastructural imaging of the cytoplasm to measure macromolecular crowding, proteomics to map stress-induced protein changes and genome-wide SATAY transposon mutagenesis to identify genes required for survival in respiration-deficient cells.

The limitations are:

• we don't know how this stress program facilitates survival in the absence of restoration of ATP levels. The data suggest that protein homeostasis is involved (chaperones and proteasome up-regulated upon stress, reduced ribosomal and translation-associated proteins down-regulated in the absence of respiration) but the mechanism remains elusive.
• the relationships between cytoplasmic crowding and quiescence establishment remain correlative. Yet, the authors provide another pathway to favour viability upon quiescence establishment (with HS) whose activation also displays an increased crowding and reduction of cytoplasmic movement, further consolidating this link. Both of these points are adequately discussed in the manuscript. None of these points should preclude publication of this study, in my opinion.

Audience.

This study would be of interest to researchers in the field of quiescence, biophysics, proteostasis, stress response, nutrient signaling and yeast biology.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

This manuscript addresses an important and longstanding question in the field: how eukaryotic cells remodel themselves to enter and survive quiescence, particularly under nutrient limitation and energy stress. The authors combine tools from biophysics, proteomics, stress signaling, and functional genomics to propose that stress-induced cytoplasmic reorganization, rather than ATP availability per se, is critical for long-term survival when respiration is impaired. The topic is timely, the experiments are generally well executed, and the initial phenomenology is compelling. The paper begins with a set of clear and convincing figures that establish an interesting and biologically important phenotype: when cells are shifted into glucose starvation, they can survive long term only if respiration is functional. Blocking respiration with Antimycin A (AntA) severely compromises viability. One straightforward hypothesis is that this defect simply reflects a failure to generate sufficient ATP. The authors, however, show that a 30-minute heat shock (HS) before glucose withdrawal in the presence of AntA largely rescues survival, even though cellular ATP levels remain critically low. In parallel, they use very well-executed GEM single-particle tracking experiments to demonstrate that cytoplasmic particle mobility decreases markedly in glucose-starved, respiration-deficient cells, and that this diffusion defect is also rescued by the pre-HS, again without restoring ATP. Together, these initial figures strongly support the idea that stress-induced remodeling of the cytoplasm, rather than ATP levels per se, is a key determinant of whether cells can enter and maintain a viable quiescent state. The authors then propose that this protective effect of HS is mediated by induction of the environmental stress response (ESR) and by resulting changes in protein expression. To test whether new protein synthesis is required, they pre-treat cells with cycloheximide during the HS and recovery period. This treatment largely, although not completely, abrogates the beneficial effect of HS on survival and diffusion in AntA-treated, glucose-starved cells. This is a strong experiment and supports the idea that HS-induced synthesis of specific proteins is important for protection, while also hinting that some cycloheximide-insensitive or pre-existing components may contribute. To identify the relevant proteins, the authors turn to global proteomic analysis, comparing multiple conditions: glucose starvation (SC), heat shock followed by glucose starvation (HS SC), glucose starvation plus AntA (SC + AntA), and heat shock followed by glucose starvation plus AntA (HS SC + AntA), each at 1 and 20 hours. This is where, in my view, the story becomes significantly harder to follow. The text for Figure 3 relies almost entirely on GO term enrichment, with very little description of individual proteins or even basic quantitative summaries of the dataset. For example, the authors never clearly state how many proteins were robustly quantified, nor what fraction of the proteome that represents. Without this foundational information, it is difficult to evaluate the strength and generality of their conclusions.

Related to this, the GO analysis in Figure 3F reports "significant" enrichment for categories such as ribosomes or translation, yet the underlying number of proteins making up these enrichments is not shown. From the volcano plots, it appears that only a very small number of proteins change in some conditions (e.g., SC 20 h), and yet GO terms appear with extremely strong q-values. This is confusing: how can such strong enrichment occur if only a handful of proteins are changing? At minimum, the authors should provide:

• the number of significantly up- or down-regulated proteins in each comparison
• the number of proteins contributing to each enriched GO category
• the magnitude of the changes for these proteins

Because the absolute number of significantly changing proteins appears small in several conditions, the current heavy reliance on GO analysis feels unwarranted and potentially misleading. In such cases, it would likely be more informative to list all differentially abundant proteins-either in supplementary materials or in a main-text table-and briefly describe the most relevant ones, rather than relying on broad category labels. Figure 3F, in particular, needs substantially more explanation. A related issue appears in Figure 3G (and the associated text), where the authors emphasize that the proteomic response to HS + AntA and the response to long-term glucose starvation are distinct. While this conclusion is plausible, the analysis also shows a subset of proteins that are upregulated in both conditions. These overlapping proteins may, in fact, represent the core protective module that enables survival in quiescence. The authors do not discuss these proteins at all; instead, they are effectively dismissed in favor of the "distinct responses" narrative. I encourage the authors to identify and discuss these overlapping proteins explicitly. Are they chaperones, proteasome components, antioxidant enzymes, or other classical stress-response factors? Even if the global proteomes differ, the overlapping subset could be highly informative about the minimal set of proteins required to stabilize the cytoplasm and support entry into quiescence. The SATAY screen is a major strength of the paper, as it moves from correlative proteomics to functional genetic analysis. The approach appears well-controlled, but key information is missing: How many unique insertions were obtained? Was the library saturating? What was the read distribution and coverage? The authors also discuss only a small subset of the screen hits. The volcano plots show many additional genes that are not addressed. What categories do these fall into? Are they informative about pathways beyond Ras/PKA and Msn2/4? Presenting a fuller analysis would strengthen the mechanistic interpretation. The parts of the SATAY analysis that are discussed are solid. The screen implicates the Ras/PKA signaling axis and Msn2/4 in survival under HS-preconditioned, respiration-deficient starvation, and the authors validate these hits with targeted survival assays. The correspondence between genetic perturbations and changes in cytoplasmic diffusion is an intriguing connection. However, the analysis stops short of identifying the downstream effector proteins that actually produce the biophysical benefits observed. The manuscript then returns to the idea that improved cytoplasmic diffusion and reduced confinement may be essential for survival. This is an appealing hypothesis, but the evidence remains correlative. It is still unclear whether biophysical rescue is the cause of improved survival or simply a downstream marker of a properly induced stress response. What remains missing is deeper integration of the proteomics and SATAY data to identify which proteins are likely responsible for the adaptive changes in cytoplasmic organization. Overexpression of promising candidates-such as chaperones or proteostasis factors found in the overlap between HS and long-term starvation responses-could help determine whether any single protein or small group of proteins can phenocopy the HS-induced rescue. Importantly, many of the comments above are intentionally broad: the manuscript does not simply require small clarifications but would benefit from substantial expansion and deepening of the analysis. The observations are compelling, but the mechanistic chain connecting ESR activation → proteomic remodeling → cytoplasmic biophysics → survival remains insufficiently developed in the current draft. Clearer quantitative reporting, fuller presentation of the data, and more thoughtful interpretation would significantly strengthen the manuscript.

Summary of Major Issues That Need to Be Addressed

Quantitative clarity in the proteomics

• State how many proteins were quantified.
• Report the numbers of significantly changing proteins in each condition.
• Identify the proteins underlying each GO term and provide effect sizes.

Over-reliance on GO analysis

• Provide explicit lists of differentially expressed proteins.
• Indicate whether enrichment results are meaningful given the small number of hits.

Overlooked overlapping proteins

• Analyze and discuss the subset of proteins upregulated both by HS and by long-term starvation.
• These may represent the core factors enabling survival.

SATAY analysis needs fuller presentation

• Provide insertion numbers, coverage, and basic library statistics.
• Discuss additional hits beyond the Ras/PKA/Msn2/4 pathways.
• Integrate SATAY results more deeply with proteomics.

Mechanistic depth remains limited

• Clarify whether cytoplasmic biophysical rescue is causal or downstream.
• Test whether overexpression of candidate proteins can mimic HS-induced protection.
• Expand the discussion of potential mechanisms using insights from both datasets.

Significance

This manuscript addresses a fundamental and timely question in cell biology: how eukaryotic cells remodel themselves to enter and survive quiescence, particularly under conditions of nutrient depletion and compromised energy production. Although quiescence has been studied for decades, the mechanisms that link metabolic state, stress signaling, and the physical properties of the cytoplasm remain incompletely understood. This work brings together biophysical measurements, global proteomics, and unbiased genetic screening in an ambitious effort to illuminate how cells maintain viability when respiration-and thus efficient ATP generation-is disrupted. A key conceptual contribution of this study is the demonstration that ATP levels alone do not dictate survival during starvation. Rather, the ability of cells to mount an appropriate stress response and reorganize the cytoplasm appears to be crucial. The early figures provide compelling evidence that heat shock preconditioning can rescue both viability and cytoplasmic mobility in respiration-deficient cells, even when ATP remains low. This finding is notable because it challenges the widely held assumption that energy charge is the primary determinant of successful entry into quiescence. If strengthened by deeper mechanistic analysis, this insight could reshape how the field views energy stress and cellular dormancy.

The identification of the Ras/PKA-Msn2/4 axis as a key regulatory node is also significant, as it connects quiescence survival to well-established nutrient and stress signaling pathways. The integration of a genome-wide SATAY screen adds functional depth and offers the potential to uncover specific downstream effectors that remodel the cytoplasm or stabilize cellular structures during prolonged stress. Finally, the manuscript touches on a concept that is gaining traction across many subfields of biology: that the biophysical state of the cytoplasm is a regulated and physiologically meaningful parameter, not merely a passive consequence of metabolic decline. Understanding how cells tune macromolecular crowding, diffusion, and spatial organization during quiescence could have broad implications beyond yeast, including in stem cell biology, microbial dormancy, cancer cell persistence, and aging.

Overall, the questions addressed are important, and the study has the potential to make a meaningful conceptual contribution. However, realizing that impact will require clearer and deeper mechanistic analysis-particularly in the proteomics and SATAY sections-to convincingly identify the specific factors and pathways that mediate the cytoplasmic remodeling underlying survival.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

In the present manuscript, the authors present an in-depth study on the effect of a heat-shock response on the ability of yeast to regain viability after quiescence when their ability to respire is inhibited. They nicely demonstrate that these effects correlate with the measured diffusion coefficients, providing deeper insight into the (at least partially) responsible environmental stress response and the molecular players involved. This work is an important contribution to the growing (or resurging) field of the physical properties of the cell.

My two main comments are the following:

• The authors determine the diffusion coefficients from the MSD, as well as further analyze them all the way up to the confinement size. As far as I can judge from the manuscript, these analyses are for 2D systems and were initially developed for processes on membranes. How does this change for 3D systems? I understand that for a straightforward qualitative comparison of apparent MSD, this assumption is acceptable, but it may deviate more strongly with the additional analyses the authors present.
• The authors show data in the supporting information where the GEMs provide larger foci after stress with longer imaging times. Could the authors provide the images of the shorter imaging times that they use? That seems a more equal comparison than Figure C. It is also unclear to me why fixed cells are used in Figure C, as well as the meaning of the x-axis. In line with this, can the authors exclude that GEMs dimerize/oligomerize after stress, and therefore display a lower diffusion coefficient?

Other comments:

• For the precision of the language, the authors equate ribosome content with macromolecular crowding, with the diffusion of the GEMs throughout, and this becomes more conflated in the discussion, where it is compared to viscosity and macromolecular crowding effects, e.g., translation. Is it macromolecular crowding, mesoscale crowding, nano-rheology, or ribosome crowding? What is measured precisely?
• In the EM images, the ribosomes seem smaller after starvation. Is that correct, and how should we interpret this? Is this due to an increased number of monosomes?
• The authors refer to recent work on how biochemical reactions, such as translation, are determined by the cytoplasm. There is some older work on this, see for example in bacteria https://doi.org/10.1073/pnas.1310377110, and also in vitro here DOI: 10.1021/acssynbio.0c00330
• On the section of correlating diffusion and survival outcomes (bottom page 12), it is mentioned that the lowered diffusion could enhance aggregation. However, literature indicates that the opposite is true in buffer; lower diffusion reduces aggregation (also nucleation is inversely proportional to the viscosity).

Significance

General assessment:

Strengths: It is a comprehensive study that provides a wealth of information and insight into the intricacies of a field that has received considerable attention, and its views are evolving rapidly.

Weaknesses: It may suffer from some overinterpretation of diffusion data.

Advance:

The significant advance is that the molecular response pathway and precise molecular players are connected to the biophysical response of cells to starvation/quiescence. The dependence of diffusion on starvation has received considerable attention (Jacobs-Wagner, Cell, 2014; the current authors in eLife, 2016; and more recent investigations by Holt, Delarue, and others). Still, the authors take the next step and demonstrate how quiescence, and particularly how the history of a cell affects it, correlates strongly with the diffusion. As far as I can tell, this is new. As mentioned, the molecular insights into the pathways are exceptionally strong from my perspective. From personal experience, this work is also very important for researchers outside of the field from a practical standpoint: Do your measurements change when you stress cells by walking to a microscope? And even if you incubate them there, your measurement outcome will change. In my experience, this is a crucial point, and the cell's history is often overlooked.

Audience:

Broad -- biophysicists, molecular biologists, cell biologists, biotechnologists.

My field of expertise: Biophysics.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

We thank reviewers for the general positive feedback and insightful suggestions. Reviewers found that our study “provides a rich resource of potential E3-sensor interactions and represents a conceptual and technical advance for the field” and that our “key conclusions are convincing and interesting”. Reviewers suggested both editorial changes to improve the narrative of the manuscript and additional experiments to strengthen the conclusions of the study. We agree with both types of suggestions and decided to modify our manuscript accordingly.

Reviewer #1 (Evidence, reproducibility and clarity (Required)): 

The authors present a rational, AlphaFold-based strategy to systematically identify interactions between human nucleic acid sensors and SPRY-containing proteins. Their findings demonstrate that SPRY domains encode substrate-specific recognition patterns that govern immune responses: TRIM25-ZAP in antiviral defense and restricts LNP-encapsulated RNA, while Riplet-RIG-I for the IFNB1 production and restricts lipofection. They further dissect residue-level contributions to the ZAP-TRIM25 interface by integrating structural predictions with experimental validation. 

Specific comments.  1. The title of this manuscript appears quite broad given that this study mostly focuses on just TRIM25-ZAP and Riplet-RIG-I pairs. 

We agree that the original title was broader than the main mechanistic focus of the study. We will therefore revise the title to better reflect that the manuscript primarily dissects SPRY-domain–mediated specificity in the TRIM25-ZAP and Riplet-RIG-I interactions (identified through our AlphaFold-based screening framework), while retaining the broader screening context. Proposed new title: "SPRY domains encode ubiquitin ligase specificity for ZAP and RIG-I"

In Figure 1b, several predicted interaction scores appear inconsistent with previously reported experimental interactions. For instance, KHNYN has been experimentally validated as a TRIM25-interacting protein, yet its interaction score is notably low in your computational results. Could the authors clarify whether this discrepancy arises because the TRIM25 SPRY domain does not significantly contribute to KHNYN binding? 

We thank the reviewer for raising this point. To our knowledge, published data only support co-immunoprecipitation of TRIM25 and KHNYN in ZAP-deficient in cells (PMID: 31284899), but this does not by itself demonstrate a direct binary interaction, as the association could be mediated by other factors. Consistent with this, our AlphaFold-based screen predicts a low interaction score between KHNYN and TRIM25, suggesting that this may not be a direct protein-protein interaction. Nevertheless, we concede that our approach may have missed interactions that are governed by a small number of interacting residues. We added the following sentences on the limitation of this approach for such interactions in our discussion:

• “While our screen revealed novel interactions between SPRY domain containing proteins and innate immune sensors, it is plausible that certain interactions were missed. Interactions that rely on a small number of contacting residues or interactions that may be mediated by a third binding partner are likely to score poorly in our approach. Future optimization of our algorithm will improve the detection of such interactions.”*

In Figure 2c, the authors provide intriguing examples for shared targets by SPRY proteins with quite low homology, and distinct target profiles by nearly identical SPRY domains. However, the underlying mechanisms responsible for these observations are not discussed. 

This is an important point. At present, we cannot assign a single definitive mechanism for every example, but there are several plausible explanations consistent with our framework. First, our analysis indicates that substrate recognition is often driven by a limited subset of residues at the interaction surface, such that distinct sequences can converge on similar three-dimensional interface chemistry, while small local differences can shift binding preferences. Second, we note that although a large fraction of predicted contacting residues are within SPRY domains, other domains can also contribute to interaction and substrate recognition, which could modulate binding profiles even when SPRY sequences are near-identical. Third, the Pearson’s correlation coefficient was calculated all scores, which may include structures with low confidence scores or low interaction scores

In Figure 3e and 3f, the authors state that the Riplet-T25 SPRY chimeric protein showed enhanced AlphaFold predicted interaction with ZAP, and validated the interaction experimentally. However, the Alphafold also predicted that an increased interaction for the T25-Riplet chimera, although this mutant failed to be co-precipitated with ZAP. How do the authors reconcile this discrepancy between prediction and experimental outcome? 

The reviewer noticed an important, nuanced result in Fig. 3e. AlphaFold predicts that the TRIM25 chimera containing the Riplet SPRY domain (T25–Riplet) has a higher interaction score with ZAP than Riplet alone (Fig. 3e), yet this chimera is not recovered in ZAP co-immunoprecipitation (Fig. 3f). We reconcile this by emphasising that our framework uses an empirically benchmarked threshold: known SPRY–sensor interactions typically score >2.5, and we therefore adopted >2.5 as the cutoff for “high-confidence” candidate interactions. While the T25–Riplet chimera shows an increased score relative to Riplet, its score remains below this >2.5 cutoff in Fig. 3e (which reports interaction scores of the chimeras against ZAP). Therefore, the model is consistent with the experimental outcome: AlphaFold suggests some degree of interface compatibility, but not at a level we would classify as a robust/predictive interaction under our validated threshold. We clarified this point in the Results section to explicitly note that sub-threshold “increases” should be interpreted cautiously:

“Using the T25-RipletSPRY instead of the Riplet protein, predicted a higher interaction score despite the lack of specific pull-down between this chimera and ZAP; importantly, this interaction score is below our defined threshold (2.5), highlighting the importance of benchmarking predicted scores against known interactions.”

It is curious if the authors explain why TRIM25 consistently appears as two bands in many of the presented figures. 

We have also wondered about this observation as well. Other studies report that the double band pattern in western blots of TRIM25 (PMID: 17392790, 28060952, 21292167) and it is believed to be a product of non-degradative self-ubiquitination of TRIM25, primarily acting on the K117 residue (PMID: 21292167). We will add a brief description of this phenomenon in the figure legend.

In Figure 4b, the authors show that treatment with a proteasome inhibitor increased RIG-I ligand-induced IFNB1 expression and propose that RIG-I may undergo rapid degradation following its interaction with Riplet. However, the evidence supporting this claim is weak. The authors should demonstrate: (1) that RIG-I is indeed degraded via the proteasome, and (2) whether RIG-I undergoes K48-linked ubiquitination. Mutational analysis of putative ubiquitination sites in RIG-I would help clarify its contribution to the observed IFN responses. 

This is an important point and we are currently performing experiments addressing these questions. Specifically we will provide evidence of (1) whether RIG-I is degraded after activation using a combination of western blotting and pharmacological inhibition of the proteosome/translation machinery; (2) whether RIG-I goes K48- or K63-mediated ubiquitination by performing coIPs of RIG-I in the presence of HA-Ub wildtype or the commonly used HA-Ub K48 and K63 mutants (PMID: 15728840); and (3) whether lysine-to-arginine substitution of key residues impacts RIG-I ubiquitination/degradation.

Figure 5 c-g: why do the authors show ZAP-L, but not ZAP-S? 

Both ZAP-S and ZAP-L isoforms contain identical N-terminal domains, which is the region that interacts with TRIM25. Therefore, we assumed that the interaction between TRIM25 and ZAP-L would be similar to TRIM25-ZAP-S. However, to test this assumption, we will generate equivalent mutations in ZAP-S and perform similar co-immunoprecipitation experiments.

Reviewer #1 (Significance (Required)): 

This manuscript starts with the AlphaFold-based screening of interactions between human nucleic acid sensors and SPRY-containing proteins. However, the authors then just focused on TRIM25-ZAP and Riplet-RIG-I, whose interactions have been well demonstrated previously, although other protein interactions were not further explored. Also, the information on the evolutional aspects of TRIM25, ZAP, Riplet and RIG-I did not lead to clear conclusions. The differential contribution of TRIM25-ZAP and Riplet-RIG-I in LNP- and lipofectamine-transduced RNAs is interesting, although data shown in Fig.6 are expected from previous studies, and are not so relevant to other data in this study.  Therefore, the study is not well integrated, although pieces are interesting.  This study may attract researchers in innate 

My expertise is innate immunity and RNA biology. 

Reviewer #2 (Evidence, reproducibility and clarity (Required)): 

The paper describes the discovery of unknown E3-RNA sensor interactions from a large scale in silico prediction screen, as well as the clarification of previously described E3-sensor interactions. These findings extend previous work showing ancient relationships between nucleic acid sensors and RING E3s (e.g. PMID: 33373584), which also described the RIPLET-RIG-I pairing identified in the present screen. 

The interactions focused on are shown to have functional implications for immune signaling pathways, and stability implications for the bound sensor. The argument for the screen is that E3-target interactions are often too transient to detect biochemically. While possibly true, several of the pairings are confirmed by co-IP, with either WT E3 or a catalytically deficient E3 (known elsewhere as a 'substrate trap'). 

The key conclusions are convincing and interesting; in particular, the conserved interactions between RIPLET and RIG-I, and TRIM25 and ZAP. The hypothesis that the two E3s arose from a common ancestor is intriguing, and the use of chimeras in functional experiments suggest that the length of the coiled coil domains contributes to substrate targeting. The most interesting observation IMO is that showing that RNA vaccines can be sensed by orthogonal sensor/E3 pathways, depending on transfection method, suggesting that distinct entry routes are surveyed by different sensors. These experiments are well performed as E3 manipulation phenocopies sensor manipulation, supporting that the in silico approach will ID relevant pairings. 

Including the PAE plots for some of the key interactions would be helpful, as a lot of the interaction confidence metrics are hidden in interaction 'scores'. Fig. 1b heatmap is presented as a row max, so it is difficult to calibrate one E3 against another. The raw data from e.g. fig. 1c would be a valuable addition. This would also help orientate future studies predicting similar protein-protein interactions. 

We agree with the reviewer and we will provide the raw values for the interaction scores and PAE maps as supplementary data to be included in the final publication.

Figure 1 appears to just use the isolated SPRY domain for screening - were full-length proteins used?

The data in Figure 1 was generated using full-length proteins, but it will be interesting to test if a similar screen with SPRY domains alone can replicate the predicted interactions. We will repeat this using SPRY domain sequences.

How many copies of the FL protein were used. TRIM5 employs a low affinity, high avidity binding method; do binding patterns change when the valency of either component is altered? The Alphafold approach perhaps selects for high affinity binders? I don't expect many more experiments to be done here, but commenting on this would be useful. __ __

This is a rational consideration that we overlooked. We included in our discussion a comment on the limitation of this approach in the context of multimeric assemblies:

“Similarly, the oligomeric nature of some SPRY-containing proteins [22] is likely to impact the correct placement of these domains and, therefore, impact the predicted interaction score. Future optimization of our algorithm will improve the detection of such interactions.”

The TRIM25 -Riplet PRYSPRY swap experiments in Figure 3 are very informative and powerful. Some more detail on domain boundaries used would be helpful, including AF predictions of what these chimeras look like with respect to their natural counterparts. 

We agree with the reviewer about the need to explicitly define domain boundaries. We will include as supplemental information a comparison of the AF prediction of these chimeras in relation to the native proteins.

While AF can place confidence metrics on domain-domain interactions, SPRY containing proteins are themselves often comprised of regions of high structural confidence (e.g. many available PDBs for RINGs, coils and SPRYs) but their relative arrangement within the molecule is unpredictable. Do isolated SPRYs show any better/worse binding to targets? 

This is a good point as well, and this can be addressed by repeating the AlphaFold screen using only SPRY domain proteins rather than full-length protein (as described above).

Technically, fig. 1f does not show that TRIM58 destabilises OAS1, as there is no condition with OAS1 alone. Perhaps alter the text to reflect this or repeat with the necessary control. The direction of the text is fine, as Fig. 1g provides a striking result, but 1f needs attention. 

The reviewer raises an important consideration. To address this, we will repeat the experiment using a OAS1 alone condition, as suggested.

Fig. 2c - for clarity, please specify the meaning of the connecting lines between the bait 'hits' in the legend. What does the correlation coefficient relate to exactly? % similarity, is this across the whole molecule, or the PRYSPRY (presumably the latter would be a more useful comparison). And it is well established that single variations in SPRY variable loops can toggle binding, so this could be better referenced in the text. It would also be helpful to see e.g. dissimilar PRYSPRYs binding a common target, as PAE plots in the supplementary. Do any shared motifs occur at the variable loops between dissimilar SPRY molecules? 

We agree that this figure could be clearer. The connecting lines in Fig. 2c indicate protein-protein predictions with common sensors, i.e. connecting lines between the interaction score of ASH2L-MDA5 and the interaction score of TRIM51-MDA5. We only use % similarity of the SPRY domain alone, not the whole molecule. We have modified the figure legend to clarify this point and we include the PAE maps as supplementary information, as requested.

Fig. 2i - Bat RIG-I binds both TRIM25 and Riplet? This is in contrast to the predicted directionality in 2h? 

The reviewer astutely noted that, in Fig.2i, pulling down bat RIG-I co-immunoprecipitated with both bat Riplet and bat TRIM25, while AlphaFold predictions only suggest a Riplet-RIG-I interaction. However, while bat Riplet and bat TRIM25 express at comparable levels in the input sample, bat Riplet was far more enriched in RIG-I pulldowns than bat TRIM25. Our interpretation of this data is that, indeed, bat Riplet-RIG-I interaction is more powerful than TRIM25-Rig-I.

Fig. 3a-b, Sup Fig. 3c,d - IFNB1 transcript normalised to 3p-hRNA transfection in control NTC cells - the presentation chosen obscures the baseline IFNB1 levels in the different siRNA transfections. What is the fold induction of IFNB1 in the different cell lines? 

We will include the fold induction in each cell line as supplementary information.

Fig. 3g - RLUs of EV-A71 is only tested in TRIM25 KO cells transfected with the Riplet T25 chimera. The full panel of cDNAs (parental E3s and the inverse 25-riplet swap) should be tested in parallel to confirm the effect is specific to TRIM25 PRYSPRY. 

This is a great suggestion that will help clarify the role of different domains of TRIM25 in its antiviral activity. We are currently generating cell-line stably expressing these truncations and will perform the suggested experiment.

Fig. 4b - time point of 3p-hRNA transfection? Y-axis label suggested normalisation to NTC - incorrect label? What is the effect of bortezomib on IFNB1 mRNA in mock treated cells? 

We thank the reviewer for spotting this typo, we have known corrected the axis label. We harvested cellular mRNA 8h post-transfection. Bortezomib-treatment slightly reduced the background expression of IFNB1 mRNA, but this signal is very close to the detection limit that it is difficult to draw conclusions. Nevertheless, the addition of bortezomib did not increase IFNB1 mRNA expression in the absence of a stimulus.

Fig. 4g - these experiments would benefit from an untransfected control cell to clarify how cDNA expression impacts sensor stability. 

We agree with the reviewer and we will include an untransfected control.

There seems to be an inverse correlation between sensor degradation and signaling output - is that the summary of Fig. 4? On the one hand, sensor degradation attenuates functional output (Fig. 4b), and the E3 that degrades the sensor is required for sensor function; on the other hand, changing coil-length in the E3 disables sensor degradation (Fig. 4g) but and enhances signaling response (Fig. 3j). Do the chimeras of panel Fig. g, h influence IFNB1 expression in the assay from Fig. 3j - this experiment would marry RIG-I expression with signal output. 

This is an interesting experiment. We are in the process of generating a TRIM25-/- Riplet-/- cell line, which we will use to reconstitute with the chimeras mentioned above and perform the requested experiment.

The data is generally clear. To facilitate their interpretation and for clarity, Western blots require size markers and Co-IPs should indicate the flag-/ha-epitope tags. Would make fig. 2 i-j much clearer, particularly given apparent co-migration of IgG (heavy chain?) and riplet, and the lack of control IPs. 

We agree that contextual markings will improve the interpretation of these results. We will add size markers to the western blots in fig2 in order to improve clarity.

The figure legends could provide more detail. 

We will add additional experimental details (such as time points) to the figure legends.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

The paper describes the discovery of unknown E3-RNA sensor interactions from a large scale in silico prediction screen, as well as the clarification of previously described E3-sensor interactions. These findings extend previous work showing ancient relationships between nucleic acid sensors and RING E3s (e.g. PMID: 33373584), which also described the RIPLET-RIG-I pairing identified in the present screen.

The interactions focused on are shown to have functional implications for immune signaling pathways, and stability implications for the bound sensor. The argument for the screen is that E3-target interactions are often too transient to detect biochemically. While possibly true, several of the pairings are confirmed by co-IP, with either WT E3 or a catalytically deficient E3 (known elsewhere as a 'substrate trap').

The key conclusions are convincing and interesting; in particular, the conserved interactions between RIPLET and RIG-I, and TRIM25 and ZAP. The hypothesis that the two E3s arose from a common ancestor is intriguing, and the use of chimeras in functional experiments suggest that the length of the coiled coil domains contributes to substrate targeting. The most interesting observation IMO is that showing that RNA vaccines can be sensed by orthogonal sensor/E3 pathways, depending on transfection method, suggesting that distinct entry routes are surveyed by different sensors. These experiments are well performed as E3 manipulation phenocopies sensor manipulation, supporting that the in silico approach will ID relevant pairings.

Including the PAE plots for some of the key interactions would be helpful, as a lot of the interaction confidence metrics are hidden in interaction 'scores'. Fig. 1b heatmap is presented as a row max, so it is difficult to calibrate one E3 against another. The raw data from e.g. fig. 1c would be a valuable addition. This would also help orientate future studies predicting similar protein-protein interactions.

Figure 1 appears to just use the isolated SPRY domain for screening - were full-length proteins used? How many copies of the FL protein were used. TRIM5 employs a low affinity, high avidity binding method; do binding patterns change when the valency of either component is altered? The Alphafold approach perhaps selects for high affinity binders? I don't expect many more experiments to be done here, but commenting on this would be useful.

The TRIM25 -Riplet PRYSPRY swap experiments in Figure 3 are very informative and powerful. Some more detail on domain boundaries used would be helpful, including AF predictions of what these chimeras look like with respect to their natural counterparts.

While AF can place confidence metrics on domain-domain interactions, SPRY containing proteins are themselves often comprised of regions of high structural confidence (e.g. many available PDBs for RINGs, coils and SPRYs) but their relative arrangement within the molecule is unpredictable. Do isolated SPRYs show any better/worse binding to targets?

Technically, fig. 1f does not show that TRIM58 destabilises OAS1, as there is no condition with OAS1 alone. Perhaps alter the text to reflect this or repeat with the necessary control. The direction of the text is fine, as Fig. 1g provides a striking result, but 1f needs attention.

Fig. 2c - for clarity, please specify the meaning of the connecting lines between the bait 'hits' in the legend. What does the correlation coefficient relate to exactly? % similarity, is this across the whole molecule, or the PRYSPRY (presumably the latter would be a more useful comparison). And it is well established that single variations in SPRY variable loops can toggle binding, so this could be better referenced in the text. It would also be helpful to see e.g. dissimilar PRYSPRYs binding a common target, as PAE plots in the supplementary. Do any shared motifs occur at the variable loops between dissimilar SPRY molecules?

Fig. 2i - Bat RIG-I binds both TRIM25 and Riplet? This is in contrast to the predicted directionality in 2h?

Fig. 3a-b, Sup Fig. 3c,d - IFNB1 transcript normalised to 3p-hRNA transfection in control NTC cells - the presentation chosen obscures the baseline IFNB1 levels in the different siRNA transfections. What is the fold induction of IFNB1 in the different cell lines?

Fig. 3g - RLUs of EV-A71 is only tested in TRIM25 KO cells transfected with the Riplet T25 chimera. The full panel of cDNAs (parental E3s and the inverse 25-riplet swap) should be tested in parallel to confirm the effect is specific to TRIM25 PRYSPRY.

Fig. 4b - time point of 3p-hRNA transfection? Y-axis label suggested normalisation to NTC - incorrect label? What is the effect of bortezomib on IFNB1 mRNA in mock treated cells?

Fig. 4g - these experiments would benefit from an untransfected control cell to clarify how cDNA expression impacts sensor stability.

There seems to be an inverse correlation between sensor degradation and signaling output - is that the summary of Fig. 4? On the one hand, sensor degradation attenuates functional output (Fig. 4b), and the E3 that degrades the sensor is required for sensor function; on the other hand, changing coil-length in the E3 disables sensor degradation (Fig. 4g) but and enhances signaling response (Fig. 3j). Do the chimeras of panel Fig. g, h influence IFNB1 expression in the assay from Fig. 3j - this experiment would marry RIG-I expression with signal output.

The data is generally clear. To facilitate their interpretation and for clarity, Western blots require size markers and Co-IPs should indicate the flag-/ha-epitope tags. Would make fig. 2 i-j much clearer, particularly given apparent co-migration of IgG (heavy chain?) and riplet, and the lack of control IPs.

The figure legends could provide more detail.

Significance

The paper provides a rich resource of potential E3-sensor interactions and represents a conceptual and technical advance for the field. The authors take a novel approach to identify these pairings. Several pairings are validated in CoIPs, and two pairings (T25-ZAP, RIPLET-RIG-I) are studied in detail. Many E3s - including the TRIM proteins which comprise the bulk of E3s studied here - are purported to regulate key nucleic acid sensors in the literature, but the large scale approach taken here provides evidence that the pairings are really quite specific. The findings also supports prior work showing that the PRYSPRY domain (here called the SPRY) is a functionally plastic module that through variable loops can bind a range of different protein substrates.

The paper will be most interesting to the innate immune field, those working on nucleic acid sensing, and those looking at innate responses to RNA vaccines.

Regulation of E3 ubiquitin ligases, viral RNA sensing

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

The authors present a rational, AlphaFold-based strategy to systematically identify interactions between human nucleic acid sensors and SPRY-containing proteins. Their findings demonstrate that SPRY domains encode substrate-specific recognition patterns that govern immune responses: TRIM25-ZAP in antiviral defense and restricts LNP-encapsulated RNA, while Riplet-RIG-I for the IFNB1 production and restricts lipofection. They further dissect residue-level contributions to the ZAP-TRIM25 interface by integrating structural predictions with experimental validation.

Specific comments.

1. The title of this manuscript appears quite broad given that this study mostly focuses on just TRIM25-ZAP and Riplet-RIG-I pairs.
2. In Figure 1b, several predicted interaction scores appear inconsistent with previously reported experimental interactions. For instance, KHNYN has been experimentally validated as a TRIM25-interacting protein, yet its interaction score is notably low in your computational results. Could the authors clarify whether this discrepancy arises because the TRIM25 SPRY domain does not significantly contribute to KHNYN binding?
3. In Figure 2c, the authors provide intriguing examples for shared targets by SPRY proteins with quite low homology, and distinct target profiles by nearly identical SPRY domains. However, the underlying mechanisms responsible for these observations are not discussed.
4. In Figure 3e and 3f, the authors state that the Riplet-T25 SPRY chimeric protein showed enhanced AlphaFold predicted interaction with ZAP, and validated the interaction experimentally. However, the Alphafold also predicted that an increased interaction for the T25-Riplet chimera, although this mutant failed to be co-precipitated with ZAP. How do the authors reconcile this discrepancy between prediction and experimental outcome?
5. It is curious if the authors explain why TRIM25 consistently appears as two bands in many of the presented figures.
6. In Figure 4b, the authors show that treatment with a proteasome inhibitor increased RIG-I ligand-induced IFNB1 expression and propose that RIG-I may undergo rapid degradation following its interaction with Riplet. However, the evidence supporting this claim is weak. The authors should demonstrate: (1) that RIG-I is indeed degraded via the proteasome, and (2) whether RIG-I undergoes K48-linked ubiquitination. Mutational analysis of putative ubiquitination sites in RIG-I would help clarify its contribution to the observed IFN responses.
7. Figure 5 c-g: why do the authors show ZAP-L, but not ZAP-S?

Significance

This manuscript starts with the AlphaFold-based screening of interactions between human nucleic acid sensors and SPRY-containing proteins. However, the authors then just focused on TRIM25-ZAP and Riplet-RIG-I, whose interactions have been well demonstrated previously, although other protein interactions were not further explored. Also, the information on the evolutional aspects of TRIM25, ZAP, Riplet and RIG-I did not lead to clear conclusions. The differential contribution of TRIM25-ZAP and Riplet-RIG-I in LNP- and lipofectamine-transduced RNAs is interesting, although data shown in Fig.6 are expected from previous studies, and are not so relevant to other data in this study. Therefore, the study is not well integrated, although pieces are interesting. This study may attract researchers in innate

My expertise is innate immunity and RNA biology.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity (Required)): The key conclusions are solid. All the claims are supported by quality data. The content is rich, and no additional experiment is needed. The data and methods are properly presented for reproduction. The experiments are adequately replicated. One comment on statistical analysis is listed below.* *

__Summary:_ ___ This manuscript investigates how Drosophila immune pathways contribute to defense against a range of filamentous fungi with distinct ecological strategies. The work provides novel insights into Toll pathway activation through pattern recognition receptors and danger signals, relative roles of melanization, phagocytosis, and effects of antimicrobial peptides, and particularly the immune evasion strategy of E. muscae via protoplast formation. These findings are of broad relevance to insect immunology, host-pathogen interactions, and evolutionary biology. * The study is well designed, the experiments are carefully executed, and the manuscript is clearly written. It is novel to demonstrate that E. muscae evades immune recognition via protoplast formation. However, some aspects of clarity and discussion of limitations could be improved before publication.** *

We thank the reviewer of the positive assessment of our manuscript.We thank the reviewer of the positive assessment of our manuscript.

Major comments: 1) The Abstract is informative but a bit too long. Consider condensing some sentences and highlighting the novel contributions (e.g., role of protoplasts in immune evasion.).* *

Good points. We have reduced the abstract. The sentence is 'Our study also reveals that the fly-specific obligate fungus Entomophthora muscae employs a vegetative development strategy, protoplasts, to hide from the host immune response.'

We believe that the role of protoplasts is already mentioned in the abstract.

2) The Results may use more mechanistic links. For instance, the section on E. muscae immune evasion could more explicitly connect the morphological findings (protoplasts, lack of cell wall) with specific immune recognition failures.* *

Our article is a comparison of Drosophila host defense against fungi with various life styles. This obviously complexify the presentation of the results. We have made the maximum of effort to explain our data with clarity. We believe that having two successive sections entitled 'Natural infection with E. muscae barely induces the Toll pathway' followed by ' __Entomophthora muscae hides from the host immune response using a vegetative development strategy'____ __expose well the idea that E. muscae has a specific hiding strategy. We did not change this part.

3) Please clarify statistical analyses used for survival data (e.g., log-rank tests, multiple testing corrections). * We have clarified the statistical analysis in the method part. The sentence is 'Statistical significance of survival data was calculated with a log-rank test (Mantel-Cox test) comparing each genotype to w*1118 flies'.

__Minor comments:____ __ Abstract: 1) "The infection outcome depends on the complex interplay between insect immune defenses and fungal adaptive strategies." could be simplified to: "Infection outcomes depend on the interplay between insect immunity and fungal adaptation." 2) Replace "our study uncovers" with "we show" for more concise phrasing. Reduce phrases like "our study reveals" or 'we conclude" in other parts of the manuscript. * Results: p. 5: phrase "survival upon natural infection... reveals the major contribution" → reword to avoid passive tone. p. 10: clarify "vesicles push the membrane outwards" with more precise terminology (e.g., budding, extrusion). * Discussion: p. 20: streamline sentence beginning "These observations provide a mechanistic basis..." (currently too dense).

We have taken in consideration all these comments. Note that we removed in the revised version the sentence "The infection outcome depends on the complex interplay between insect immune defenses and fungal adaptive strategies." To shorten the abstract, we have removed the sentence 'These observations provide a mechanistic basis for future exploration.'

**Referee cross-commenting*** *

I agree with the comments of the other two reviewers.* *

__Reviewer #1 (Significance (Required)):____ __

This manuscript investigates how Drosophila immune pathways contribute to defense against a range of filamentous fungi with distinct ecological strategies (generalists, specialists, opportunists). By leveraging a comprehensive panel of genetically defined fly lines and standardized infections, the authors provide a demonstration that the Toll pathway is the predominant systemic antifungal defense, extending classical findings into a comparative framework across fungal lifestyles. The work provides novel insights into Toll pathway activation through GNBP3 and fungal proteases sensed by Psh, while also dissecting the relative contributions of melanization, phagocytosis, and antimicrobial peptides to host protection. Of particular note is the compelling demonstration that the fly specialist E. muscae can evade immune recognition through protoplast-like vegetative forms, minimizing cell-wall exposure and thereby escaping Toll activation.* *

My expertise and limitations: * Insect biochemistry and molecular biology, with particular focus on innate immunity, serine protease cascades, melanization, and host-pathogen interactions. I also have experience with genetic, biochemical, and functional approaches to dissecting immune signaling pathways in model insects. However, I do not have sufficient expertise to critically evaluate advanced statistical analyses.** *

__Reviewer #2 (Evidence, reproducibility and clarity (Required)):____ __

In this work the authors describe the contribution of distinct immune responses in Drosophila melanogaster to systemic and natural infections with 5 fungal species with different lifestyles some being generalists infecting a broad range of insects while others being more specialists or opportunistic. The authors used several well characterized Drosophila mutants of the Toll, Imd, phagocytosis and melanization responses to address this question. They show that Toll pathway is the key player in anti-fungal resistance in both natural and septic infections, whereas melanization plays a minor role mainly during natural infections possibly to limit fungal invasion through the cuticle. The authors show elegantly using different combinations of mutants for antimicrobial peptides genes with antifungal activities that Bomanins and Daisho (1 and 2) are the main Toll effectors mediating resistance to fungi but the authors did not find specific fungus-by-gene interaction, but rather antifungal peptides seem to act in a more general fashion against the fungi tested with significant redundancies between certain classes. Interestingly the authors show that while generalists like Beauveria and Metarhizium strongly activate the Toll pathway, the specialist E. muscae weakly activates the pathway and the opportunistic A. fumigatus does not activate the pathway, indicating that certain fungal species are able to evade sensing by immune pathways. In the context of the Toll activation, the sensor protease Psh and not GNBP3 seem to be the main trigger of the pathway.* *

__Minor comments____ __ This is an interesting work that compares the contributions of different arms of the fly immune response to 5 fungal species with diverse lifestyles. The use of different lines with different combinations of mutant genes is a strength to highlight the relative contribution of each immune response. Some of the data obtained is intriguing and warrants more future investigations such as the distinct phenotypes of ModSp and GNBP3 mutants in E. muscae infections. The methodology is robust and the conclusions are supported with good experimental evidence. I do not see any major concerns with the work. I just have some minor comments listed below* *

We thank the reviewer for the positive comments on our manuscript. 1- Statistical significance should be indicated on Figures 1 and 2, although it appears in the legend.

We have added statistical significance on Figures 1 and 2.

2- It is not very accurate to use the term resistance of the different mutants to infections with the diverse fungal species in Figures 1 and 2 especially that the authors have reported only survival data in these figures and have not measured fungal proliferation in infected flies (although they did that in later figures). It is more accurate to mention that the mutants flies have different levels of tolerance rather than resistance to fungal infections.* *

We agree that we cannot use the term 'resistance' in Figures 1 and 2, since this term has now a more restricted meaning in the community. We have replaced the term 'resistance' by 'host defense' or 'surviving' through the text to avoid the confusion, except when the bacterial load was monitored.

3- The authors show that Toll is over-activated in PPO1/PPO2 double mutant possibly through a negative feedback mechanism. However, there could be another explanation for this observation: For instance, the increased fungal proliferation in the PPO double mutant results in increased protease secretion by fungi enhancing Psh activation! Also, how can fungi manage to proliferate in this double mutant if Toll is overactivated? Could it be that Toll overactivation is triggering a fitness cost?* *

The reviewer raises a good point. It is difficult to reconcile the susceptibility of PPO1/2 mutants to fungi taking in consideration the higher Toll activation. The higher activation of Toll could be deleterious and We clearly observed higher Toll pathway activation in PPO1/2 flies upon clean injury (Fig. S9C) or injection of dead spores (data not shown). Thus, this higher expression cannot be only explained as a consequence of higher fungal growth.

4- In Lines 654-655, it is not accurate to say that E. muscae protoplasts are not detected by the immune response since E. muscae natural infections triggers Drs expression at 24 hpi and there is possibly some melanization taking place since PPO1 and PPO2 are required for defense against this fungus. A more accurate explanation is that this fungus is possibly more resistant to the effectors of the host immune response than the other fungi. I think a major point that the authors might have missed to consider in the discussion of their data is that the different fungi used herein may exhibit different levels of resilience to the effector reactions of the host such as AMPs and melanin deposition* *

*The observation that injection of E. muscae protoplasts do not trigger an immune response above the level of clean injury is a strong argument that support our view that E. muscae protoplasts are not immunogenic. The reviewer is correct by underlying the small but significant induction of Drs at 24h post natural infection. We hypothesize that this could be due to mechanical injury associated with the entry of E. muscae. We have added a sentence to underline the possibility raised by the reviewer: 'Although we cannot rule out that the high pathogenicity of E. muscae may be partly due to the fungus's increased resilience, we favor the interpretation that it is instead mainly driven by its capacity to evade immune detection.'

__Reviewer #2 (Significance (Required)):____ __

Although the importance of Toll pathway and melanization in antifungal immunity is not new per se, this work adds to this knowledge by showing that Toll has the upper hand in anti-fungal immunity and that the strength of Toll pathway activation and its effector capacity may vary depending on the type of invading fungus. The work also highlights that certain fungi may employ a delayed switch to hyphal growth to reduce the presence of cell wall sugars as a mechanism to evade immune recognition. Overall, this work significantly adds to the knowledge of Drosophila immunity and raises some interesting questions related to the evolution of host-pathogen interactions and to the complex functions of serine protease cascades regulating Toll and melanization. This work will be of interest to a broad audience in the field of host-pathogen interactions *

__Reviewer #3 (Evidence, reproducibility and clarity (Required)):____ __

This is a clearly written manuscript on the immune effector mechanisms regulating Drosophila melanogaster host defense against a broad range of fungal pathogens, including entomopathogenic and saprophytic filamentous fungi. The authors systematically dissect the contribution of major arms of Drosophila immunity, including cellular and humoral responses and melanization and potential mechanisms of cross talk using genetic tools and reporter lines. They also go into detail to characterize the contribution of upstream activators of these responses by fungal PAMPs and the role of antimicrobial effectors (AMPs) in fly susceptibility. * They conclude for no important role of phagocytosis in host defense. Instead, they find important contributions of Toll pathway mainly through the detection of fungal proteases by Persephone rather than b-glucan detection by GNBP3. They also demonstrate that Toll activation is proportional to the virulence of the fungal pathogen, showing little activation of this response by Aspergillus fumigatus. Finally, they identify melanization as another line of host defense that restricts pathogen dissemination and protects fly from invasive fungal disease. A very interesting part of this study is the identification of a virulence strategy of the obligate fungus Entomophthora muscae, which employs a vegetative development strategy, by making protoplast that avoid immune recognition by masking immunostimulatory cell wall molecules to avoid immune recognition by Toll pathway until the very last stage of invasive growth. Overall, this is a very interesting study on host-pathogen interplay in Drosophila, shedding light onto novel pathogenetic mechanism employed by entomopathogenic fungi to adapt to their hosts.** *

We thank the reviewer for his positive assessment.

__Major comments for the authors:____ __ 1. The use of reporter fungal strains to capture the dynamic interplay of the pathogen and the different arms of the immune system precludes firm conclusions on the contribution of various immune response to infection. This should be emphasized in the discussion* *

Unfortunately, we did not fully understand this point. Note that we monitored both survival and when possible fungal load (B. Beauveria, E. muscae and M. anisopliae for Toll; and B. Beauveria, and M. anisopliae for melanization) allowing to state that Toll and Melanization are contributing to host defense by limiting fungal growth.

2. The route of infection and the method employed to inject fungal spores has an impact on the effector pathways being activated. For example, pricking introduces spores less efficiently in the hemolymph compared to microinjection. The inoculum size in case of microinjection also has profound impact in understanding the role of cellular and humoral immunity during the infection course. For example, the lack of Toll activation in the natural infection with A. fumigatus does not mean that this pathway is not important in host defense against this pathogen.

We fully agree and expected to clarify this different outcome between septic injury and natural infection. In the case of A. fumigatus, we confirm that Toll is important upon systemic infection but not natural infection because this fungus has a limited ability to penetrate insect by the natural route. We have clarified this in the text by adding the sentence: 'The low Toll pathway activation by A. fumigatus is likely due the weak ability of this fungus to penetrate insect by the natural route.'.

3. The use of total KO strains does not preclude the cross talk of cellular and humoral immunity and consequently potential defects in cellular immunity upon deletion of a master regulator of the Toll pathway or even its downstream effectors

The observation that Toll deficient mutants are almost as susceptibility as mutant flies lacking all the four immune modules (△ITPM ) to the five fungal pathogens point to a major role of this pathway. In a previous study (Ryckebusch et al Elife 2025), we have shown that the four immune pathways largely work independently as phagocytosis was still observed in Toll deficient mutant.

4. Did the authors validate that NimC11; Eater1 flies are not able to phagocytose fungal spores?

In the first version of this manuscript, we did not validate that NimC1;eater flies are phagocytic deficient also for Fungal spores although our manuscript assumed it. To address the comment of the reviewer, we have extended our study to better characterize the role of the cellular immune response to fungal infection (See new Figure S1).

Our new results show that NimC1;eater deficient flies have defect in binding to M. anisopliae GFP spores (New Supplement Figure S1E,F). We did not see clear evidence of internalization. Thus, we conclude that the use of NimC1;eater flies is adequate to study the role of the cellular response. We have monitored the survival of hemoless flies that lack nearly all plasmatocytes due to the over-expression of the proapoptotic gene Bax, to natural infection and septic injury with B. bassiana and M. anisopliae. This new piece of data (described in New Supplementary Figure S1A-D) show that hemoless flies display a wild-type survival to B. Bassiana and a mild susceptibility to M. anisopliae consistent with our previous statement that the cellular response is less important than the humoral response. In the revised version, we have added this new piece of data and nuanced our statement on the role of the cellular response to fungal infection.

5. Is it possible that entomopathogenic fungi bypass phagocytosis as a virulence strategy by inducing large size germinating cells, which are not phagocytosed?

Indeed, there are several studies have showed that entomopathogenic fungi have evolved sophisticated strategies to evade or survive phagocytosis.

• Once fungal spores (conidia) germinate, penetrate host tegument and reach the hemocoel, fungi existwithin the hemocoel in the forms of blastospores with thinner cell walls than conidia (M. anisopliae, M. rileyi, B. bassiana), and cell wall-free protoplasts (E. muscae). Wang and St Leger (2006) had demonstrated that host hemocytes can recognize and ingest conidia of M. anisopliae, but this capacity is lost on production of blastospore, because of its ability to avoid detection depending on the cell surface hydrophobic protein gene Mcl1 that is expressed within 20 min of the fungal pathogen contacting hemolymph.
• Other studieshave shown that blastospores of B. bassiana and M. anisopliae can be phagocytosed at the early stages of infection but manage to emerge from host cells and continue to propagate. Growing hyphal bodies can deform the plasmatocyte cell membrane (Gillespie et al., 2000; Hung and Boucias, 1992; Vilcinskas et al., 1997). Studies have also shown that during the infection process of entomopathogenic fungi in insects, the hemocyte count gradually decreases. For instance, during the infection of Thitarodes xiaojinensis by Ophiocordyceps sinensis, blastospores are the initial cell type present in the host hemocoel and remained for 5 months or more before transformation into hypha, which finally led to host death; and the increase in blastospores quantity coincidence with a decline in hemocyte count (Liu et al., 2019; Li et al., 2020).<br /> In a new set of experiments, we tested the ability of plasmatocytes to phagocytose M. anisopliae-GFP spores. We observed that plasmatocytes bind to the spores, but we did not obtain clear evidence of internalization (New Figure S1E,F). However, this assay was not sufficient to conclusively determine whether plasmatocytes internalize M. anisopliae spores, as GFP fluorescence may be quenched in acidic intracellular compartments. Because entomopathogenic fungi can affect hemocyte abundance, we also monitored the expression level of Hml, a hemocyte-specific marker, in flies following natural infection with B. bassiana, M. anisopliae, M. rileyi, and E. muscae at 2, 3, and 5 days post-infection (see figure below). We did not observe a reduction in hemocyte levels for any of these fungi except M. anisopliae. This suggests that M. anisopliae may reduce hemocyte numbers as a strategy to circumvent the cellular immune response. These results, although promising, were not included in the revised version of the manuscript, as a thorough analysis of the cellular immune response would require a dedicated study on its own.

Figure: Expression of Hml by RT-qPCR upon natural infection with entomopathogenic fungi (figure not included in the revised manuscript)

6. Is it possible that fungal toxins kill phagocytes during germination?

There are indeed evidences that fungal toxins destruxins (DTXs) induce ultrastructural alterations of circulating plasmatocytes and sessile haemocytes of Galleria mellonella larvae. DTXs contribute to the fungal infection process by a true immune-inhibitory effect. This is evidenced by two key findings: first, the germination rate of injected Aspergillus niger spores was slightly but significantly enhanced; second, during incubation, the fungus demonstrated a greater ability to escape from the haemocyte-formed granuloma envelope (Vilcinskas et al., 1997; Vey et al., 2002). But in Drosophila, Destruxin does not appear to affect Drosophila cellular immune responses in vivo. Phagocytosis of E. coli bacterial particles in Destruxin-injected flies appeared to be the same as that seen in PBS-injected flies. The proliferation of bacteria in the Destruxin-injected flies was due to the lower expression of antimicrobial peptide genes suggesting that Destruxin A specifically suppressed the humoral immune response in Drosophila (Pal et al., 2007), which is consistent with major role of antimicrobial peptides in survival to fungi. This point is now discussed in the discussion with a new section on the cellular response to fungal infection.

__Reviewer #3 (Significance (Required)):____ __

This is an important work that provide new information on virulence mechanisms of entomopathogenic fungi and the host immune responses that mediate host protection. The authors should address my comments in the discussion and provide some additional evidence by using reporter fungal strains for hemocytes on whether these fungal pathogens completely bypass phagocytosis to invade the host. Therefore, rather than claiming that phagocytosis is not important it should be clarified whether phagocytes are directly involved in host defense or whether the fungus changes its cell wall surface to avoid this line of host defense. My expertise is on phagocyte biology and host-fungal interaction on human fungal pathogens.

We have added more information showing that plasmatocytes of NimC1;eater larvae fail to bind to spores of M. anisopliae suggesting that this line provides an appropriate tool to assess phagocytosis. We have also analyzed the survival of flies depleted for plasmatocytes via the over-expression of bax, which revealed a mild role for plasmatocyte in defense against M. anisopliae but not B. bassiana. By performing additional experiments, we realized that analyzing the role of cellular immunity in host defense against these five fungi would require much more work and is beyond the scope of this study. We have however added in the revised version a para in the discussion on the the cellular response.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

This is a clearly written manuscript on the immune effector mechanisms regulating Drosophila melanogaster host defense against a broad range of fungal pathogens, including entomopathogenic and saprophytic filamentous fungi. The authors systematically dissect the contribution of major arms of Drosophila immunity, including cellular and humoral responses and melanization and potential mechanisms of cross talk using genetic tools and reporter lines. They also go into detail to characterize the contribution of upstream activators of these responses by fungal PAMPs and the role of antimicrobial effectors (AMPs) in fly susceptibility.

They conclude for no important role of phagocytosis in host defense. Instead, they find important contributions of Toll pathway mainly through the detection of fungal proteases by Persephone rather than b-glucan detection by GNBP3. They also demonstrate that Toll activation is proportional to the virulence of the fungal pathogen, showing little activation of this response by Aspergillus fumigatus. Finally, they identify melanization as another line of host defense that restricts pathogen dissemination and protects fly from invasive fungal disease. A very interesting part of this study is the identification of a virulence strategy of the obligate fungus Entomophthora muscae, which employs a vegetative development strategy, by making protoplast that avoid immune recognition by masking immunostimulatory cell wall molecules to avoid immune recognition by Toll pathway until the very last stage of invasive growth. Overall, this is a very interesting study on host-pathogen interplay in Drosophila, shedding light onto novel pathogenetic mechanism employed by entomopathogenic fungi to adapt to their hosts.

Major comments for the authors:

1. The use of reporter fungal strains to capture the dynamic interplay of the pathogen and the different arms of the immune system precludes firm conclusions on the contribution of various immune response to infection. This should be emphasized in the discussion
2. The route of infection and the method employed to inject fungal spores has an impact on the effector pathways being activated. For example, pricking introduces spores less efficiently in the hemolymph compared to microinjection. The inoculum size in case of microinjection also has profound impact in understanding the role of cellular and humoral immunity during the infection course. For example, the lack of Toll activation in the natural infection with A. fumigatus does not mean that this pathway is not important in host defense against this pathogen.
3. The use of total KO strains does not preclude the cross talk of cellular and humoral immunity and consequently potential defects in cellular immunity upon deletion of a master regulator of the Toll pathway or even its downstream effectors
4. Did the authors validate that NimC11; Eater1 flies are not able to phagocytose fungal spores?
5. Is it possible that entomopathogenic fungi bypass phagocytosis as a virulence strategy by inducing large size germinating cells, which are not phagocytosed?
6. Is it possible that fungal toxins kill phagocytes during germination?

Significance

This is an important work that provide new information on virulence mechanisms of entomopathogenic fungi and the host immune responses that mediate host protection. The authors should address my comments in the discussion and provide some additional evidence by using reporter fungal strains for hemocytes on whether these fungal pathogens completely bypass phagogytosis to invade the host. Therefore, rather than claiming that phagocytosis is not important it should be clarified whether phagocytes are directly involved in host defense or whether the fungus changes its cell wall surface to avoid this line of host defense. My expertise is on phagocyte biology and host-fungal interaction on human fungal pathogens.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

In this work the authors describe the contribution of distinct immune responses in Drosophila melanogaster to systemic and natural infections with 5 fungal species with different lifestyles some being generalists infecting a broad range of insects while others being more specialists or opportunistic. The authors used several well characterized Drosophila mutants of the Toll, Imd, phagocytosis and melanization responses to address this question. They show that Toll pathway is the key player in anti-fungal resistance in both natural and septic infections, whereas melanization plays a minor role mainly during natural infections possibly to limit fungal invasion through the cuticle. The authors show elegantly using different combinations of mutants for antimicrobial peptides genes with antifungal activities that Bomanins and Daisho (1 and 2) are the main Toll effectors mediating resistance to fungi but the authors did not find specific fungus-by-gene interaction, but rather antifungal peptides seem to act in a more general fashion against the fungi tested with significant redundancies between certain classes. Interestingly the authors show that while generalists like Beauveria and Metarhizium strongly activate the Toll pathway, the specialist E. muscae weakly activates the pathway and the opportunistic A. fumigatus does not activate the pathway, indicating that certain fungal species are able to evade sensing by immune pathways. In the context of the Toll activation, the sensor protease Psh and not GNBP3 seem to be the main trigger of the pathway.

Minor comments

This is an interesting work that compares the contributions of different arms of the fly immune response to 5 fungal species with diverse lifestyles. The use of different lines with different combinations of mutant genes is a strength to highlight the relative contribution of each immune response. Some of the data obtained is intriguing and warrants more future investigations such as the distinct phenotypes of ModSp and GNBP3 mutants in E. muscae infections. The methodology is robust and the conclusions are supported with good experimental evidence. I do not see any major concerns with the work. I just have some minor comments listed below

1. Statistical significance should be indicated on Figures 1 and 2, although it appears in the legend.
2. It is not very accurate to use the term resistance of the different mutants to infections with the diverse fungal species in Figures 1 and 2 especially that the authors have reported only survival data in these figures and have not measured fungal proliferation in infected flies (although they did that in later figures). It is more accurate to mention that the mutants flies have different levels of tolerance rather than resistance to fungal infections.
3. The authors show that Toll is over-activated in PPO1/PPO2 double mutant possibly through a negative feedback mechanism. However, there could be another explanation for this observation: For instance, the increased fungal proliferation in the PPO double mutant results in increased protease secretion by fungi enhancing Psh activation! Also, how can fungi manage to proliferate in this double mutant if Toll is overactivated? Could it be that Toll overactivation is triggering a fitness cost?
4. In Lines 654-655, it is not accurate to say that E. muscae protoplasts are not detected by the immune response since E. muscae natural infections triggers Drs expression at 24 hpi and there is possibly some melanization taking place since PPO1 and PPO2 are required for defense against this fungus. A more accurate explanation is that this fungus is possibly more resistant to the effectors of the host immune response than the other fungi. I think a major point that the authors might have missed to consider in the discussion of their data is that the different fungi used herein may exhibit different levels of resilience to the effector reactions of the host such as AMPs and melanin deposition

Significance

Although the importance of Toll pathway and melanization in antifungal immunity is not new per se, this work adds to this knowledge by showing that Toll has the upper hand in anti-fungal immunity and that the strength of Toll pathway activation and its effector capacity may vary depending on the type of invading fungus. The work also highlights that certain fungi may employ a delayed switch to hyphal growth to reduce the presence of cell wall sugars as a mechanism to evade immune recognition. Overall, this work significantly adds to the knowledge of Drosophila immunity and raises some interesting questions related to the evolution of host-pathogen interactions and to the complex functions of serine protease cascades regulating Toll and melanization. This work will be of interest to a broad audience in the field of host-pathogen interactions

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

The key conclusions are solid. All the claims are supported by quality data. The content is rich, and no additional experiment is needed. The data and methods are properly presented for reproduction. The experiments are adequately replicated. One comment on statistical analysis is listed below.

Summary:

This manuscript investigates how Drosophila immune pathways contribute to defense against a range of filamentous fungi with distinct ecological strategies. The work providesovel insights into Toll pathway activation through pattern recognition receptors and danger signals, relative roles of melanization, phagocytosis, and effects of antimicrobial peptides, and particularly the immune evasion strategy of E. muscae via protoplast formation. These findings are of broad relevance to insect immunology, host-pathogen interactions, and evolutionary biology. The study is well designed, the experiments are carefully executed, and the manuscript is clearly written. It is novel to demonstrate that E. muscae evades immune recognition via protoplast formation. However, some aspects of clarity and discussion of limitations could be improved before publication.

Major comments:

1. The Abstract is informative but a bit too long. Consider condensing some sentences and highlighting the novel contributions (e.g., role of protoplasts in immune evasion.).
2. The Results may use more mechanistic links. For instance, the section on E. muscae immune evasion could more explicitly connect the morphological findings (protoplasts, lack of cell wall) with specific immune recognition failures.
3. Please clarify statistical analyses used for survival data (e.g., log-rank tests, multiple testing corrections).

Minor comments:

Abstract: 1) "The infection outcome depends on the complex interplay between insect immune defenses and fungal adaptive strategies." could be simplified to: "Infection outcomes depend on the interplay between insect immunity and fungal adaptation." 2) Replace "our study uncovers" with "we show" for more concise phrasing. Reduce phrases like "our study reveals" or 'we conclude" in other parts of the manuscript. Results: p. 5: phrase "survival upon natural infection... reveals the major contribution" → reword to avoid passive tone. p. 10: clarify "vesicles push the membrane outwards" with more precise terminology (e.g., budding, extrusion). Discussion: p. 20: streamline sentence beginning "These observations provide a mechanistic basis..." (currently too dense).

Referee cross-commenting

I agree with the comments of the other two reviewers.

Significance

This manuscript investigates how Drosophila immune pathways contribute to defense against a range of filamentous fungi with distinct ecological strategies (generalists, specialists, opportunists). By leveraging a comprehensive panel of genetically defined fly lines and standardized infections, the authors provide a demonstration that the Toll pathway is the predominant systemic antifungal defense, extending classical findings into a comparative framework across fungal lifestyles. The work provides novel insights into Toll pathway activation through GNBP3 and fungal proteases sensed by Psh, while also dissecting the relative contributions of melanization, phagocytosis, and antimicrobial peptides to host protection. Of particular note is the compelling demonstration that the fly specialist E. muscae can evade immune recognition through protoplast-like vegetative forms, minimizing cell-wall exposure and thereby escaping Toll activation.

My expertise and limitations:

Insect biochemistry and molecular biology, with particular focus on innate immunity, serine protease cascades, melanization, and host-pathogen interactions. I also have experience with genetic, biochemical, and functional approaches to dissecting immune signaling pathways in model insects. However, I do not have sufficient expertise to critically evaluate advanced statistical analyses.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Response to Reviewer 1:

The authors introduce G2PT, a hierarchical graph transformer model that integrates genetic variants (SNPs), gene annotations, and multigenic systems (Gene Ontology) to predict and interpret complex traits.

We thank the reviewer for this accurate summary of our approach and contributions.

Major Comments:

Comment 1-1. Insufficient Specification of Model Architecture: The description of the "hierarchical graph transformer" lacks technical depth. Key implementation details are missing: how node embeddings are initialized for SNPs, genes, and systems; how graph connectivity is defined at each level (e.g., adjacency matrices used in Equations 5-9, the sparsity); justification for the choice of embedding dimension and number of attention heads, including any sensitivity analysis; and the architecture of the feed-forward neural networks (e.g., number of layers, activation functions, and hidden dimensions).

__Reply 1-1. __As requested, we have expanded the technical description of the model architecture, including the hierarchical graph transformer (HiGT), in the Materials and Methods section. Details regarding node initialization and hierarchical connectivity are now included in the new paragraph "Model Initialization and Graph Construction." Specifically, all node embeddings corresponding to SNPs, genes, and ontology-defined systems are initialized using uniform Xavier initialization (Glorot and Bengio, 2010).

We have also clarified our hyperparameter optimization strategy. Learning rate, weight decay, hidden (embedding) dimension, and the number of attention heads were selected via grid search, as summarized in new Supplementary Fig. 8, reproduced below. Based on both performance and computational efficiency, we adopted four attention heads-consistent with the configuration commonly used in academic transformer models (Vaswani et al., 2017) (the original Transformer used eight).

Regarding the feed-forward neural network, we follow the standard Transformer architecture consisting of two position-wise layers with hidden dimension four times larger than the node embedding size and a GeLU nonlinear activation function (Hendrycks and Gimpel, 2016). This configuration is widely established in the literature and functions as an intermediate processing step following attention; therefore, it is not a focus of hyperparameter tuning. All corresponding updates have been incorporated into the revised Methods section for clarity and completeness.

Comment 1-2. No Simulation Studies to Validate Epistasis Detection: The ground truth epistasis interaction should use the ones that have been manually validated by literature. The central claim of discovering epistatic interactions relies heavily on the model's attention mechanism and downstream statistical filtering. However, no simulation studies are presented to validate that G2PT can reliably detect epistasis when ground-truth interactions are known. Demonstrating robust detection of non-additive interactions under varying genetic architectures and noise levels in simulated genotype-phenotype datasets is essential to substantiate the method's core capability.

Reply 1-2. We agree that a simulation of epistasis detection using the G2PT model is a worthy addition to the manuscript. Accordingly, we have now incorporated a new section in the Results titled "Validation of Epistasis through Simulation Studies", which includes two new figures reproduced below (Supplementary Fig. 6 and Fig. 5). We have also added a new Methods section to describe this simulation study under the heading "Epistasis Simulation". These simulation studies show that G2PT recovers epistatic gene pairs with high fidelity when these pairs are coherent with the systems ontology (c.f. 'ontology coherence' in Supplementary Fig. 6, which reflects the probability that both SNPs are assigned to the same leaf system). Furthermore, G2PT outcompetes previous tools, such as PLINK-epistasis, which do not use knowledge of the systems hierarchy in the same way (Supplementary Fig 6b-d). Using simulation parameters consistent with current genome-wide association studies (n = 400,000) and understanding of heritability (h2 = 0.3 to 0.5) (Bloom et al. 2015; Speed and Evans 2023), we find that approximately 10% of all epistatic SNP pairs can be recovered at a precision of 50% (Fig. 5). We have provided the source code for this simulation study in our GitHub repository (https://github.com/idekerlab/G2PT/blob/master/Epistasis_simulation.ipynb)

Comment 1-3. Lack of Justification for Model Complexity and Missing Ablation Insights: While Supplementary Figure 2 presents ablation studies, the manuscript needs to justify the high computational cost (168 GPU hours using 4×A30 GPUs) of the full model. It remains unclear how much performance gain is specifically due to reverse propagation (Equations 8-9), which is claimed to capture biological context. The benefit of using a full Gene Ontology hierarchy versus a flat system list is not quantified. There is also no comparison between bidirectional versus unidirectional propagation. Overall, the added complexity is not empirically shown to be necessary

Reply 1-3. We thank the reviewer for prompting a clearer justification of complexity and ablations. We have now revised the Results to (i) quantify the specific value of the ontology and reverse propagation, and (ii) explain why a flat SNP→system model is computationally and biologically sub-optimal. We have added new ablation results to compare bidirectional (forward+reverse) versus forward-only propagation. Reverse propagation has little effect when epistatic pairs are within one system (ontology coherence ρ=1.0) but substantially improves retrieval when interactions span related systems (e.g., ρ≈0.8) (Figure reproduced below) A flat design scores a dense genes×systems map, ignoring known sparsity (sparse SNP→gene assignments; sparse ontology edges) and losing multi-scale context; our hierarchical formulation restricts computation to observed edges (SNP→gene→system) and aggregates signals across levels, yielding better efficiency and biological fidelity.

Comment 1-4. Non-Equivalent Benchmarking Against PRS Methods: Figure 2 compares G2PT to polygenic risk score (PRS) methods such as LDpred2 and Lassosum, but G2PT is run only on SNPs pre-filtered by marginal association (p-values between 10⁻⁵ and 10⁻⁸), while the PRS methods use genome-wide SNPs. This introduces a strong bias in G2PT's favor by effectively removing noise. A fair comparison would require: (a) running LDpred2 and Lassosum on the same pre-filtered SNP sets as G2PT, or (b) running G2PT on genome-wide or LD-pruned SNP sets. The reported superior performance of G2PT may be driven primarily by this input filtering, not the model architecture.

Reply 1-4. We appreciate the reviewer's concern regarding benchmarking equivalence. In response, we have extended our analyses to include PRS-CS (Ge et al., 2019) and SBayesRC (Zheng et al., 2024), two state-of-the-art Bayesian shrinkage methods comparable to LDpred2 and Lassosum. Although we initially attempted to run LDpred2 and Lassosum under all SNP-filtering conditions, their computational requirements at UK Biobank scale proved prohibitively time consuming. We therefore focused on PRS-CS and SBayesRC, which offer similar modeling principles with greater computational tractability. These methods have now been run at matched SNP-filtering conditions to our original study. The new results demonstrate that G2PT consistently outperforms PRS-CS and SBayesRC (new Fig. 2, reproduced below), indicating that its performance advantage is not solely attributable to SNP pre-filtering but also to its hierarchical attention-based architecture.

Comment 1-5: No Details on Hyperparameter Optimization: Although the manuscript mentions grid search for hyperparameter tuning, it provides no information about which parameters were optimized (e.g., learning rate, dropout rate, weight decay, attention dropout, FFNN dimensions), what search space was explored, or what final values were selected. There is also no assessment of how sensitive the model's performance is to these choices. Better transparency would help facilitate reproducibility

Reply 1-5. We agree with the reviewer and have expanded the manuscript to include full details of hyperparameter optimization. As described in the revised Methods section, we performed a grid search over learning rate {10−3,10−4,10−5} hidden dimension {64,128} and weight decay {0,10−5,10−3}. The results, summarized in Supplementary Fig. 8 (reproduced above), show that model performance is most sensitive to the learning rate, while hidden dimension and weight decay exert more moderate effects. Based on these findings, we selected a learning rate of 10−5, hidden dimension of 64, and weight decay of 10−3 for all subsequent experiments. Although a hidden dimension of 128 slightly improved performance, we adopted 64 to balance predictive accuracy with computational efficiency.

Comment 1-6. Absence of Control for Key Confounders: In interpreting attention scores as reflecting genetic relevance (e.g., the role of the immunoglobulin system), the model includes only age, sex, and genetic principal components as covariates. Important confounders such as BMI, alcohol use, or medication (e.g., statins) have not been controlled for. Since TG/HDL levels are strongly influenced by environment and lifestyle, it is entirely plausible that some high-attention features reflect environmental tagging, not biological causality.

Reply 1-6. In the current framework, we included age, sex, and genetic principal components to account for demographic and population-structure effects, focusing on genetic contributions within a controlled baseline. We acknowledge that non-genetic covariates can influence downstream biological states and may indirectly shape attention at the gene or system level. Accurately modeling such effects requires an extended framework where environmental variables directly modulate gene and system embeddings rather than being implicitly absorbed by the attention mechanism. We have clarified these limitations in the Discussion along with plans to incorporate explicit confounder modeling in future extensions of G2PT.

Comment 1-7. Oversimplified Treatment of SNP-to-Gene Mapping: The SNP-to-gene mapping strategy combines cS2G, eQTL, and nearest-gene annotations, but the limitations of this approach are not adequately addressed. The manuscript does not specify how conflicts between methods are resolved or what fraction of SNPs map ambiguously to multiple genes. Supplementary Figure 2 shows model performance degrades when using only nearest-gene mapping, but there is no systematic analysis of how mapping uncertainties propagate through the hierarchy and affect attention or interpretation.

Reply 1-7. In the revision (Results), we have clarified how conflicts between cS2G, eQTL, and nearest-gene annotations are resolved, and we have reported the proportion of SNPs that map to multiple genes across these three annotation approaches. We note that the hierarchical attention mechanism enables the model to prioritize among alternative gene mappings in a data-driven manner, and this is a major strength of the approach. As shown in Fig. 3 (Results, reproduced below), SNP-to-gene attention weights reveal dominant linkages, reducing the impact of mapping uncertainty on interpretation. We now explicitly describe this mechanism and acknowledge that further work in probabilistic mapping and fine-mapping approaches is a valuable future direction for improving resolution and interpretability.

"For SNPs with several potential SNP-to-gene mappings (Methods), we found that G2PT often prioritized one of these genes in particular due to its membership in a high-attention system. For example, the chr11q23.3 locus contains multiple genes including the APOA1/C3/A4/A5 gene cluster (Fig. 3c) which is well-known to govern lipid transport, an important system for G2PT predictions (Fig. 3a). Due to high linkage disequilibrium in the region, all of its associated SNPs had multiple alternative gene mappings available. For example, SNP rs1145189 mapped not only to APOA5 but to the more proximal BUD13, a gene functioning in spliceosomal assembly (a system receiving substantially lower G2PT attention). Here, the relevant information flow learned by G2PT was from rs1145189 to APOA5 to lipid transport and protein-lipid complex remodeling (Fig. 3c; and conversely, deprioritizing BUD13 as an effector gene for TG/HDL). We found that this particular genetic flow was corroborated by exome sequencing, which implicates APOA5 but not BUD13 in regulation of TG/HDL, using data that were not available to G2PT. Similarly, two other SNPs at this locus - rs518547 and rs11216169 - had potential mappings to their closest gene SIK3, where they reside within an intron, but also to regulatory elements for the more distant lipid transport genes APOC3 and APOA4. Here, G2PT preferentially weighted the mappings to APOC3 and APOA4 rather than to SIK3 (Fig. 3c)."

Comment 1-8. Naive Scoring of System Importance: The method used to quantify the biological relevance of systems (i.e., correlating attention scores with predicted phenotype values) risks circular reasoning. Since the model is trained to optimize prediction, systems that contribute strongly to prediction will naturally show high correlation-even if they are not biologically causal. No comparison is made with established gene set enrichment methods applied to GWAS summary statistics. The approach lacks an independent benchmark to validate that the "important" systems are biologically meaningful.

Reply 1-8. As requested, we compared G2PT's system-level importance scores with results from MAGMA competitive gene-set analysis, an established enrichment approach. This analysis indeed shows significant correlation between the systems identified by the two approaches (ρ = 0.26, p .01; Supplementary Table. 2), reflecting a shared emphasis on canonical lipid processes. We also observed systems detected by G2PT but not strongly detected by MAGMA's linear enrichment model-for example, the lipopolysaccharide-mediated signaling pathway (Kalita et al. 2022)

Comment 1-9. No External Validation to Assess Generalizability. All evaluations are performed using cross-validation within the UK Biobank. There is no assessment of generalizability to independent cohorts or diverse ancestries. Given population structure, genotyping platform, and phenotype measurement variability, external validation is essential before claiming the method is suitable for broader use in polygenic risk assessment.

Reply 1-9. To externally validate the G2PT model requires individual level genotype data with paired TG/HDL measurements, sample size at the scale of the UK Biobank, and GPU access to this data. Thus, we approached the All of Us program, a large and diverse cohort with individual level data and T2D conditions with HbA1C measurements. We first processed the All of Us genotype and phenotype data as we had processed UKBB data (Methods), resulting in 41,849 participants with T2D and 80,491 without T2D across various ethnicities. We then transferred the trained T2D G2PT model to the AoU Workbench and evaluated its performance. The model demonstrated robust discriminative capability with an explained variance of 0.025, as shown in the new Fig. 2d, (reproduced above).

Comment 1-10. Computational Burden and Scalability Are Not Addressed: The paper notes that training the model requires 168 GPU hours on 4×A30 GPUs for just ~5,000 SNPs. However, there is no discussion of whether G2PT can scale to larger SNP sets (e.g., genome-wide imputed data) or more complex biological hierarchies (e.g., Reactome pathways). Without addressing scalability, the model's applicability to real-world, large-scale genomic datasets remains unclear.

Reply 1-10. We have addressed scalability with both engineering optimizations and new scalability experiments. First, we refactored the model to use the xFormer memory-efficient attention for the hierarchical graph transformer (Lefaudeux et al., 2022), which also helps full parallelization of training, reducing bottlenecks. Second, we added a scaling study with progressively increasing SNP count. On 4×A30 GPUs, end-to-end training time for the 5k-SNP setting decreased from 4000 to 400 min. (approximately 7 GPU-hours, ×10). These new results are given in Supplementary Fig. 7, reproduced below.

Minor Comment:

Comment 1-11. Attention Weights as Mechanistic Insight: The paper equates high attention scores with biological importance, for example in highlighting the immunoglobulin system. There is no causal validation showing that altering the highlighted SNPs, genes, or systems has an actual effect on TG/HDL. Attention weights in transformer models are known to sometimes reflect spurious correlations, especially in high-dimensional settings. The correlation between attention scores and predictions (Supplementary Fig. 3a,b) does not constitute biological evidence. The interpretability claims can be restated without supporting functional or causal validation.

Reply 1-11. We thank the reviewer for this thoughtful comment. We agree that attention weights are not causal evidence. In the revision, we (1) reframe attention-based findings as hypothesis-generating rather than mechanistic, and (2) add an explicit limitation noting that correlations between attention scores and predictions do not constitute biological validation.

Response to Reviewer 2:

This manuscript describes the introduction of the Genotype-to-Phenotype Transformer (G2PT), described by the authors as "a framework for modeling hierarchical information flow among variants, genes, multigenic systems, and phenotypes." The authors used the ratio TG/HDL as a trait for proof of concept of this tool.

This is a potentially interesting computational tool of interest to bioinformaticians, computational genomicists, and biologists.

We thank the reviewer for their overall positive assessment of our study.

Comment 2-1. The rationale for choosing the TG/HDL ratio for this proof of concept analysis is not well justified beyond it being a marker for insulin resistance. Overall the use of a ratio may be problematic (see below). Analyses of TG and HDL separately as individual quantitative traits would be of interest. And an analysis of a dichotomous clinical trait (T2DM or CAD) would also be of great interest.

Reply 2-1. We thank the reviewer for this suggestion. In the revised manuscript, we have expanded our analyses beyond the TG/HDL ratio to include TG and HDL as individual quantitative traits (Fig. 2, reproduced below). These additional analyses demonstrate that G2PT captures predictive signals robustly across each lipid component, not solely through their ratio. Furthermore, to address the reviewer's interest in clinical outcomes, we incorporated an analysis of type 2 diabetes (T2D) as a dichotomous trait of direct clinical relevance. Collectively, these results strengthen the rationale for our chosen phenotype and show that the G2PT framework generalizes effectively across quantitative and binary traits, consistently outperforming advanced PRS and machine learning benchmarks.

Comment 2-2. The approach to mapping SNPs to genes does not incorporate the most advanced approaches. This should be described in more detail.

Reply 2-2. We agree that the choice of SNP-to-gene mapping materially affects both performance and interpretability-indeed, our epistasis simulations suggest that more accurate mappings can improve recovery and localization. In this proof-of-concept work we use a straightforward, modular mapping sufficient to demonstrate the modeling framework, and we have clarified this in the Methods. The architecture is designed to plug-and-play alternative SNP-to-gene maps (e.g., eQTL/colocalization-based assignments, promoter-capture Hi-C). A dedicated follow-up study will systematically compare these alternatives and quantify their impact on attribution and downstream discovery.

Comment 2-3. The example of gene prioritization at the A1/C3/A4/A5 gene locus is not particularly illuminating, as the prioritized genes are already well-known to influence TG and HDL-C levels and the TG/HDL ratio. Can the authors provide an example where G2PT prioritized a gene at a locus that is not already a well-known regulator of TG and HDL metabolism?

Reply 2-3. We thank the reviewer for this suggestion. We have revised the manuscript to de-emphasize the well-established APOA1 locus and instead highlight the less expected "Positive regulation of immunoglobulin production" system (Figure 3a,b, Discussion). Here our model prioritizes the gene TNFSF13 based on specific variants that are not previously associated with TG or HDL (e.g., rs5030405, rs1858406, shown in blue). This finding points to an intriguing, non-canonical link between B-cell regulation and lipid metabolism. While full exploration of this finding is beyond the scope of the present methods paper, this example demonstrates G2PT's ability to identify novel, high-priority candidates in atypical systems.

Comment 2-4. The identification of epistatic interactions is a potentially interesting application of G2PT. However, suppl table 1 shows a very limited number of such interactions with even fewer genes, and most of these are well established biological interactions (such as LPL/apoA5). The TGFB1 and FKBP1A interaction is interesting and should be discussed. What is needed for increasing the number of potential interactions, greater power?

Reply 2-4. We are glad the reviewer appreciates the use of the G2PT model to identify epistatic interactions. We have now discussed a potential mechanism of epistasis between TGFB1 and FKBP1A in the protein dephosphorylation system (Discussion). In addition, we have addressed the reviewer's question about statistical power through extensive epistasis simulations (Fig. 5 and Supplementary Fig. 6), which show that G2PT's detection ability scales strongly with sample size-1,000 samples are insufficient, performance improves at 5,000, and power becomes reliable at 100,000. Realistic simulations (Fig. 5b-d) further demonstrate that under biologically plausible architectures, G2PT can robustly recover specific interactions even within complex genetic backgrounds

Comment 2-5. Furthermore, the use of the TG/HDL ratio for the assessment of epistatic interactions may be problematic. For example, if one SNP affected only TG and the other only HDL-C, it would appear to be an epistatic interaction with regard to the ratio, although the biological epistasis may be limited to non-existent.

Reply 2-5. We have greatly expanded the example phenotypes modeled in our study, Please see our reply 2-1 above.

Response to Reviewer 3:

This manuscript by Lee et al provides a sensible and powerful approach to polygenic score prediction. The model aggregates information from SNPs to genes to systems, using a transformer based architecture, which appears to increase predictive performance, produce interpretable outputs of genes and systems that underlie risk, and identify candidates for epistasis tests.

I think the manuscript is clear and well written, and conducted via state-of-the-art approaches. I don't have any concerns regarding the claims that are made.

We thank the reviewer for their very positive assessment of our study.

Major comments:

Comment 3-1. Specifically, lipid based traits are perhaps the most well-powered and the most biologically coherent; they are also very well-studied biologically and thus overrepresented in the gene ontology. It is unclear whether this approach will work as well for a trait like Schizophrenia for which the underlying pathways are not as well captured in existing ontologies. The authors anticipate this in their limitations section, and I am not expecting them to solve every issue with this, but it would be nice to expand the testing a little bit beyond only this one trait.

Reply 3-1. We appreciate the reviewer's suggestion to expand beyond a single lipid trait. In the revised manuscript, we have included analyses of additional phenotypes, including low-density lipoprotein (LDL) and T2D (Fig. 2). These additions demonstrate the broader applicability of our framework beyond a single trait class.

Comment 3-2. It also seems like the authors have not compared their method to the truly latest PRS methods, such as PRS-CSx and SBayesR. I would suggest adding some of the methods shown to be the best from this recent paper: https://www.nature.com/articles/s41598-025-02903-1

Reply 3-2. We agree these are important comparators. Accordingly, we have extended our comparison to include PRS‑CS (Ge et al., 2019) and SBayesRC (Zheng et al., 2024), following its strong performance demonstrated in recent benchmarking studies (see Figure 2 above). We confirmed that G2PT outperforms advanced PRS methods for all TG/HDL ratio, LDL, and T2D phenotypes.

Comment 3-3. Another major comment regards whether this method could be applied to traits with just GWAS summary statistics, rather than individual level data. This would not enable identification of specific methods underlying an individual, but it could still learn SNP based weights that could be mapped to genes and systems that could help explain risk when the model is applied to individuals (kind of like a pretraining step?)

Reply 3-3. We appreciate this suggestion. While SNP weights from GWAS summary statistics could, in principle, serve as informative priors for attention values, incorporating them would require a sophisticated mathematical formulation that is beyond the scope of this study. Our current framework also relies on individual-level genotype and phenotype data to capture multilevel information flow and individual-specific variation.

Minor comments:

Comment 3-4. Why the need to constrain to a small number of SNPs? Is it just computational cost? If so, what would happen as power increases and more SNPs exceed the thresholds used?

Reply 3-4. Yes, it's about computational cost, but we've now modified the code for improved computational efficiency. First, we refactored the model to use the xFormer memory-efficient attention for the hierarchical graph transformer (Lefaudeux et al., 2022), which also helps full parallelization of training, reducing bottleneck effects. Second, we added a scaling study of the impact of varying SNP count. On 4×A30 GPUs, end-to-end training time for the 5k-SNP setting decreased from 65 hours to 7 GPU-hours (×9). We expect performance can potentially increase if more SNPs are provided to the model based on Fig. 2 (reproduced above). With the optimized implementation, users can raise SNP thresholds as power increases; the expected behavior is improved accuracy up to a plateau, while hierarchical sparsity maintains training tractability and ensures well-regularized results.

Comment 3-5. What type of sample size/power does this method require to work well? If others were to use it, how many SNPs/samples would be needed to obtain good performance?

Reply 3-5. To address this comment, we quantified performance as a function of training size by subsampling the cohort and retraining G2PT with identical architecture and SNP set. New Supplementary Fig. 3 (reproduced below) shows monotonic gains with sample size across three representative phenotypes. We found that stable performance is reached by ~100k samples. These trends hold for continuous traits (TG/HDL, LDL) and more modestly for a binary trait (T2D), consistent with lower per-sample information for case-control settings.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

This manuscript by Lee et al provides a sensible and powerful approach to polygenic score prediction. The model aggregates information from SNPs to genes to systems, using a transformer based architecture, which appears to increase predictive performance, produce interpretable outputs of genes and systems that underlie risk, and identify candidates for epistasis tests.

I think the manuscript is clear and well written, and conducted via state-of-the-art approaches. I don't have any concerns regarding the claims that are made.

My two major comments regard a question about how well this will work when compared to other approaches for other traits besides TG:HDL. Specifically, lipid based traits are perhaps the most well-powered and the most biologically coherent; they are also very well-studied biologically and thus overrepresented in the gene ontology. It is unclear whether this approach will work as well for a trait like Schizophrenia for which the underlying pathways are not as well captured in existing ontologies. The authors anticipate this in their limitations section, and I am not expecting them to solve every issue with this, but it would be nice to expand the testing a little bit beyond only this one trait.

Therefore, I would suggest that the authors test a limited number of additional traits that are not lipid based traits, and ideally not metabolic traits, to see how their model behaves. I would pick well-powered GWAS with a lot of associations but from a different phenotypic category

It also seems like the authors have not compared their method to the truly latest PRS methods, such as PRS-CSx and SBayesR. I would suggest adding some of the methods shown to be the best from this recent paper: https://www.nature.com/articles/s41598-025-02903-1

Another major comment regards whether this method could be applied to traits with just GWAS summary statistics, rather than individual level data. This would not enable identification of specific methods underlying an individual, but it could still learn SNP based weights that could be mapped to genes and systems that could help explain risk when the model is applied to individuals (kind of like a pretraining step?)

Other minor comments:

Why the need to constrain to a small number of SNPs? Is it just computational cost? If so, what would happen as power increases and more SNPs exceed the thresholds used?

What type of sample size/power does this method require to work well? If others were to use it, how many SNPs/samples would be needed to obtain good performance?

Will this work just as well for binary diseases? Is this a straightforward extension of the method or does it require more work?

Since I think a lot of geneticists will read it, more intuition as to how attention weights map to parameters geneticists think about would be useful, in particular how the graphics in Fig 3 are made (this may be second nature to ML experts but may not be obvious to statistical geneticists)

The authors claim that G2PT identifies epistatic interactions. Is this true or does it just identify pairs of SNPs that could be subsequently tested for epistasis?

Significance

This study does a great job of marrying the latest (interesting) technologies in AI/ML with a specific problem in statistical genetics. The clarity of presentation and interpretability of the model are strong. The main areas for improvement are to clarify how general this approach is -- will it work for other traits, is it truly better than the latest PRS methods, and what are the specifics of the GWAS it requires (sample size, individual-level data, power, type of trait)

I think the main advance is therefore currently conceptual, but not yet practical, unless more performance comparisons were done.

It seems like the main audience would be geneticists, since I suspect most AI/ML researchers are familiar with this type of approach. If there are fundamental innovations in applying transformers in this specific way to genetics, that would be good to highlight in more depth.

My expertise: statistical genetics and computer science, familiar with DNNs but not a practitioner in them.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

This manuscript describes the introduction of the Genotype-to-Phenotype Transformer (G2PT), described by the authors as "a framework for modeling hierarchical information flow among variants, genes, multigenic systems, and phenotypes." The authors used the ratio TG/HDL as a trait for proof of concept of this tool.

Specific comments:

1. The rationale for choosing the TG/HDL ratio for this proof of concept analysis is not well justified beyond it being a marker for insulin resistance. Overall the use of a ratio may be problematic (see below). Analyses of TG and HDL separately as individual quantitative traits would be of interest. And an analysis of a dichotomous clinical trait (T2DM or CAD) would also be of great interest.
2. The approach to mapping SNPs to genes does not incorporate the most advanced approaches. This should be described in more detail.
3. The example of gene prioritization at the A1/C3/A4/A5 gene locus is not particularly illuminating, as the prioritized genes are already well-known to influence TG and HDL-C levels and the TG/HDL ratio. Can the authors provide an example where G2PT prioritized a gene at a locus that is not already a well-known regulator of TG and HDL metabolism?
4. The identification of epistatic interactions is a potentially interesting application of G2PT. However, suppl table 1 shows a very limited number of such interactions with even fewer genes, and most of these are well established biological interactions (such as LPL/apoA5). The TGFB1 and FKBP1A interaction is interesting and should be discussed. What is needed for increasing the number of potential interactions, greater power?
5. Furthermore, the use of the TG/HDL ratio for the assessment of epistatic interactions may be problematic. For example, if one SNP affected only TG and the other only HDL-C, it would appear to be an epistatic interaction with regard to the ratio, although the biological epistasis may be limited to non-existent.

Significance

This is a potentially interesting computational tool of interest to bioinformaticians, computational genomicists, and biologists.

The proof of concept offered here using a single ratio is not sufficient to conclude its potential utility.

My expertise is in genetics and molecular mechanisms of lipid metabolism.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

The authors introduce G2PT, a hierarchical graph transformer model that integrates genetic variants (SNPs), gene annotations, and multigenic systems (Gene Ontology) to predict and interpret complex traits.

Major Comments:

1. Insufficient Specification of Model Architecture: The description of the "hierarchical graph transformer" lacks technical depth. Key implementation details are missing: how node embeddings are initialized for SNPs, genes, and systems; how graph connectivity is defined at each level (e.g., adjacency matrices used in Equations 5-9, the sparsity); justification for the choice of embedding dimension and number of attention heads, including any sensitivity analysis; and the architecture of the feed-forward neural networks (e.g., number of layers, activation functions, and hidden dimensions).
2. No Simulation Studies to Validate Epistasis Detection: The ground truth epistasis interaction should use the ones that have been manually validated by literature. The central claim of discovering epistatic interactions relies heavily on the model's attention mechanism and downstream statistical filtering. However, no simulation studies are presented to validate that G2PT can reliably detect epistasis when ground-truth interactions are known. Demonstrating robust detection of non-additive interactions under varying genetic architectures and noise levels in simulated genotype-phenotype datasets is essential to substantiate the method's core capability.
3. Lack of Justification for Model Complexity and Missing Ablation Insights: While Supplementary Figure 2 presents ablation studies, the manuscript needs to justify the high computational cost (168 GPU hours using 4×A30 GPUs) of the full model. It remains unclear how much performance gain is specifically due to reverse propagation (Equations 8-9), which is claimed to capture biological context. The benefit of using a full Gene Ontology hierarchy versus a flat system list is not quantified. There is also no comparison between bidirectional versus unidirectional propagation. Overall, the added complexity is not empirically shown to be necessary.
4. Non-Equivalent Benchmarking Against PRS Methods: Figure 2 compares G2PT to polygenic risk score (PRS) methods such as LDpred2 and Lassosum, but G2PT is run only on SNPs pre-filtered by marginal association (p-values between 10⁻⁵ and 10⁻⁸), while the PRS methods use genome-wide SNPs. This introduces a strong bias in G2PT's favor by effectively removing noise. A fair comparison would require: (a) running LDpred2 and Lassosum on the same pre-filtered SNP sets as G2PT, or (b) running G2PT on genome-wide or LD-pruned SNP sets. The reported superior performance of G2PT may be driven primarily by this input filtering, not the model architecture.
5. No Details on Hyperparameter Optimization: Although the manuscript mentions grid search for hyperparameter tuning, it provides no information about which parameters were optimized (e.g., learning rate, dropout rate, weight decay, attention dropout, FFNN dimensions), what search space was explored, or what final values were selected. There is also no assessment of how sensitive the model's performance is to these choices. Better transparency would help facilitate reproducibility
6. Absence of Control for Key Confounders: In interpreting attention scores as reflecting genetic relevance (e.g., the role of the immunoglobulin system), the model includes only age, sex, and genetic principal components as covariates. Important confounders such as BMI, alcohol use, or medication (e.g., statins) have not been controlled for. Since TG/HDL levels are strongly influenced by environment and lifestyle, it is entirely plausible that some high-attention features reflect environmental tagging, not biological causality.
7. Oversimplified Treatment of SNP-to-Gene Mapping: The SNP-to-gene mapping strategy combines cS2G, eQTL, and nearest-gene annotations, but the limitations of this approach are not adequately addressed. The manuscript does not specify how conflicts between methods are resolved or what fraction of SNPs map ambiguously to multiple genes. Supplementary Figure 2 shows model performance degrades when using only nearest-gene mapping, but there is no systematic analysis of how mapping uncertainties propagate through the hierarchy and affect attention or interpretation.
8. Naive Scoring of System Importance: The method used to quantify the biological relevance of systems (i.e., correlating attention scores with predicted phenotype values) risks circular reasoning. Since the model is trained to optimize prediction, systems that contribute strongly to prediction will naturally show high correlation-even if they are not biologically causal. No comparison is made with established gene set enrichment methods applied to GWAS summary statistics. The approach lacks an independent benchmark to validate that the "important" systems are biologically meaningful.
9. No External Validation to Assess Generalizability: All evaluations are performed using cross-validation within the UK Biobank. There is no assessment of generalizability to independent cohorts or diverse ancestries. Given population structure, genotyping platform, and phenotype measurement variability, external validation is essential before claiming the method is suitable for broader use in polygenic risk assessment.
10. Computational Burden and Scalability Are Not Addressed: The paper notes that training the model requires 168 GPU hours on 4×A30 GPUs for just ~5,000 SNPs. However, there is no discussion of whether G2PT can scale to larger SNP sets (e.g., genome-wide imputed data) or more complex biological hierarchies (e.g., Reactome pathways). Without addressing scalability, the model's applicability to real-world, large-scale genomic datasets remains unclear.

Minor:

1. Attention Weights as Mechanistic Insight: The paper equates high attention scores with biological importance, for example in highlighting the immunoglobulin system. There is no causal validation showing that altering the highlighted SNPs, genes, or systems has an actual effect on TG/HDL. Attention weights in transformer models are known to sometimes reflect spurious correlations, especially in high-dimensional settings. The correlation between attention scores and predictions (Supplementary Fig. 3a,b) does not constitute biological evidence. The interpretability claims can be restated without supporting functional or causal validation.

Significance

Novelty

This work presents novelty by introducing the first transformer-based model that integrates the GO hierarchy to enable bidirectional mapping between genotype and phenotype. Additionally, the use of attention mechanisms to screen for epistasis offers a novel and computationally efficient alternative to traditional exhaustive SNP-SNP interaction tests.

Impact

Target Audience

• Specialized: Computational biologists working on interpretable machine learning methods in genomics.
• Broader: Geneticists investigating polygenic traits and drug developers focusing on pathway-level therapeutic targets.

Limitations vs. Contributions

While the work presents a clear conceptual advance by incorporating hierarchical biological priors and attention mechanisms, the technical contribution is somewhat limited by its validation on a single trait and the absence of simulation-based benchmarking. Nevertheless, the framework shows potential if extended to other traits and experimentally validated.

Overall Assessment

Recommendation: Major Revision

Strengths:

• Predictive performance appears strong.
• The use of biological priors enables interpretability at the pathway level.

Major Weaknesses:

• The current validation is limited to a single trait, restricting generalizability.
• The manuscript lacks a complete and clear description of the model architecture.
• No simulations are provided to assess the method's ability to recover known epistatic interactions or pathways.

Reviewer Expertise: Machine learning applications in genomics and genetics.

Transparent Peer Review

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1

Evidence, reproducibility and clarity

__Summary

Köver et al. examine the genetic and environmental underpinnings of multicellular-like phenotypes (MLPs) in fission yeast, studying 57 natural isolates of Schizosaccharomyces pombe. They uncover that a noteworthy subset of these isolates can develop MLPs, with the extent of these phenotypes varying according to growth media. Among these, two strains demonstrate pronounced MLP across a range of conditions. By genetically manipulating one strain with an MLP phenotype (distinct from the previously mentioned two strains), they provide evidence that genes such as MBX2 and SRB11 play a direct role in MLP formation, strengthening their genetic mapping findings. The study also reveals that while some key genes and their phenotypic effects are strikingly similar between budding and fission yeast, other aspects of MLP formation are not conserved, which is an intriguing finding.

Overall, the manuscript is well-written, dense yet logically structured, and the figures are well presented. The combination of phenotypic, genetic, and bioinformatics analyses, particularly from wet lab experiments, is commendable. The study addresses a significant gap in our understanding, primarily explored in budding yeast, by providing comprehensive data on MLP diversity in fission yeast and the interplay of genetic and environmental factors.

In summary, I enjoyed reading the manuscript and have only a few minor suggestions to strengthen the paper:

Minor revisions:

1. Although this may seem like a minor revision, but it is a crucial point. Please make sure that all raw data used to generate figures, run stats, sequence data, and scripts used to run data analysis are made publicly available. Provide relevant accession numbers and links to public data repositories. It is important that others can download the various types of data that went into the major conclusions of this paper in order to replicate your analysis or expand upon the scope of this work. I am not sure if the journal has a policy regarding this, but it should be followed to allow for transparency and reproducibility of the research.__

Reply: We very much agree with the reviewer that sharing raw data and scripts is an essential part of open science. All code and data are deposited to Github (https://github.com/BKover99/S.-Pombe-MLPs) and Figshare (https://figshare.com/articles/software/S_-Pombe-MLPs/25750980), which have now been updated to reflect our revisions. Additionally, the sequenced genomes have been deposited to ENA (PRJEB69522). Where external data was used, it was properly referenced and specifically included in Supplementary Table 3.

Two out of 57 strains exhibit strong and consistent MLP across multiple environments. Providing more information on these strains (JB914 and JB953), such as their natural habitats and distinct appearances of their MLP phenotypes under varying conditions, would provide valuable insights.

First, a brief discussion highlighting what differentiates these two strains from the rest would be helpful for readers (e.g. insight into their unique genetic and environmental background that might be linked to the MLP phenotype).

Additionally, culture tube and microscopy images of these strains, similar to those presented for JB759 in Figure 2A, can be included in the supplementary materials. My reasoning is that these images could help illustrate variation or lack thereof in aggregative group size across different media.

Reply: We thank the reviewer for highlighting this issue. Our further investigation into these strains has added additional interesting insights. JB914 and JB953 were isolated from molasses in Jamaica and the exudate of Eucalyptus in Australia, respectively, though it remains unclear whether these environments are related or even selective for the ability of these strains to form MLPs. We note that the environment from which a strain is isolated is an incomplete way of assessing its ecology. Indeed, recent research suggests that the primary habitat of S. pombe is honeybee honey and suggests that bees, which may be attracted to a number of sugary substances, may be a vector by which fission yeast are transported (1). Therefore, isolation from a particular nectar or food production environment might not reflect significant ecological differences. We now refer to the location of strain isolation in the manuscript text (lines 208-209).

However, there is more to learn from the genetic backgrounds of these two strains. We found that JB914 possesses the same variant in srb11 causally related to MLPs as JB759, the MLP-forming parental strain for our QTL analysis. To understand whether the appearance of this variant in these two strains derived from a single mutation event or was a case of convergent evolution, we analysed homology between the genomes of JB759 and JB914, focusing specifically on that variant. We found an approximately 20kb region of homology between JB759 and JB914 surrounding the srb11 truncation variant, in contrast to the majority of the genome, which does not share homology between those two strains (New Supplementary Figure 9A, B)). This result suggests that, while the two strains are largely unrelated, that specific region shares a recent common ancestor and is likely a result of interbreeding across strains.

Importantly, this analysis further emphasizes the point that the srb11 variant segregates with the MLP-forming phenotype. We conclude this because none of the other strains similar to JB759 (either across the whole genome, or specifically in the region surrounding srb11) exhibit MLPs (New Supplementary Figure 9C). This thereby further complements our QTL analysis on the significance of this variant. We have added this analysis to the manuscript text (lines 337-349).

Furthermore, we searched other strains which exhibited MLPs in our experiments (e.g. JB953) for frame shifts, insertions or deletions in any other genes in the CKM module or in the genes that were identified in our deletion library screen as adhesive, and did not identify any severe mutations falling into coding regions (other than the srb11 truncation in JB914 and JB759). This indicates that MLPs in these other strains may be caused by differences in regulatory regions surrounding these genes, or variants in other genes that were not identified in our screen. We have added this analysis to our manuscript (lines 424-425) and Supplementary Table 13.

We agree that microscopy and culture tube images of JB914 and JB953 may give insight into the nature of the MLPs exhibited by those strains. We have included such images of cultures grown in YES, EMM and EMM-Phosphate media in our revision (Lines 207-208, Supplementary Figures 4 and 5). These images are consistent with our adhesion assay screen and show that JB914 and JB953 are adhesive at the microscopic level in the relevant conditions (EMM or EMM-Phosphate).

The phenotypic outcome of overexpressing MXB2 is striking, as shown in Supplementary Figure 4C. Incorporating at least one of the culture tube images depicting large flocs into the main text, perhaps adjacent to Figure 3 panel D, would improve the visual appeal and highlight this key finding (at the moment those images are only shown in the supplementary materials).

Reply: We thank the reviewer for this suggestion. In response to Reviewer 2's suggestion to overexpress mbx2 in YES, we created new mbx2 overexpression strains that could overexpress mbx2 in YES, which was not possible in our previous strain in which mbx2 overexpression was triggered by removal of thymine from the media. We have replaced our original data from Figure 3D with data from the new mbx2 overexpression experiment, including flask images.

I know that the authors discuss the knowledge gap in the intro and results, but the abstract does not mention this critical gap. Please stress this critical gap (i.e., MLPs understudied in fission yeast) with a brief sentence in the abstract. Similarly, please consider writing a brief concluding sentence summarizing the paper's most significant finding referring to the knowledge gap would provide a clearer takeaway message for the reader - the abstract ends abruptly without any conclusion.

Reply: We agree and have now emphasized the critical gap in our abstract:

"As MLP formation remains understudied in fission yeast compared to budding yeast, we aimed to narrow this gap." at lines 18-19.

Additionally, we added the following final sentence to give the reader a clearer takeaway message:

"Our findings provide a comprehensive genetic survey of MLP formation in fission yeast, and a functional description of a causal mutation that drives MLP formation in nature." at lines 31-32.

1. The observation that strains with adhesive phenotypes have a lower growth rate compared to non-adhesive strains is a noteworthy point (lines 532-535). This represents yet another example of this classical trade-off. This point could be emphasized in the Discussion or alongside the relevant result, with a brief speculative explanation for this phenomenon.

Reply: We agree that the nature of the trade-off between MLP formation is an interesting discussion point that could arise from our work. Understanding this trade-off is made more complicated by the fact that growth is always condition-dependent, and measuring growth in strains exhibiting MLPs is non-trivial, as adhesion to labware and thick clumps of cells separated by regions of cell-free media can add variability. Nonetheless, there has been some previous work on this problem. In S. cerevisiae, it was shown that larger group size correlates with slower growth rate (3), and that flocculating cells grow more slowly (4). In S. cerevisiae, cAMP, a signalling molecule heavily involved in regulating growth in response to nutrient availability, also regulates filamentation (5). However, the relationship between flocculation and slow growth is not consistent in the literature. In some settings overexpressing the flocculins FLO8, FLO5, and FLO10 results in slower growth (6), while in others it does not (7). In addition, ethanol production has been shown to improve for biofilms (7).

Furthermore, in S. cerevisiae, MLP-forming cells grow better in low sucrose concentrations (8) and under various stress conditions (4). Flocculating cells have also shown faster fermentation in media containing common industrial bioproduction inhibitors, despite slower fermentation than non-flocculating cells in non-inhibitory media (9). However, any consequence of this possible advantage on growth has not been characterised.

In S. pombe, there is less work on this topic; however, it has been shown that deletions of rpl3201 and rpl3202, which code for ribosomal proteins, cause flocculation and slow growth (10). In that case, it is not clear if there is any causal relationship between slow growth and flocculation or if they are both parallel consequences of the ribosomal pathway disruption. We have added some of these points to the portion of the discussion that discusses this tradeoff (Lines 477-499).

To get a better understanding of this tradeoff in our system, we took several approaches. First, we added a supporting analysis (New Supplementary Figure 12B), using published growth data based on measurements on agar plates for the S. pombe gene deletion library (11). There, the authors defined a set of deletion strains that grow more slowly on EMM than the wild-type lab strain. We found that our MLP hit strains were significantly enriched in this "EMM-slow" category. This information is now included in the manuscript (Lines 409-413, New Supplementary Figure 12B).

It is, however, possible that for the assays from that work, the appearance of slow growth on solid agar in adhesive cells could be partially artifactual. Indeed, we have observed that adhesive cells tend to stick to flasks and, when grown on agar plates, cells in the same colony can stick to one another rather than to inoculation loops or pin pads. Both of these dynamics can reduce initial inoculation densities. This is less of a concern for our adhesion assay and Figures 2E, 5B, and 5F, because our before-wash intensity was done with a 7x7 pinned square about 10x10 mm2. Nonetheless, as we wanted to make a point about srb10 and srb11 mutants growing faster than other deletion mutants that exhibit MLP-formation, we also conducted growth assays in liquid media (New Figure 5F).

We observed that srb10Δ and srb11Δ strains (which exhibit MLPs in EMM) show growth curves similar to wild-type cells in minimal (EMM) and rich media (YES). On the other hand, other strains that grow similarly to wild type cells in YES, such as tlg2Δ and rpa12Δ, grow much more slowly in EMM when they clump together. There are also some strains, mus7Δ and kgd2Δ, that grow more slowly in both YES and EMM but are only adhesive in EMM.

The text mentions two lab strains, JB22 and JB50, displaying strong adhesion under phosphate starvation (lines 525-526), yet the data point for JB22 in Figure 2C is not labeled.

Reply: We agree that highlighting JB22 on the figure is crucial, given that it was mentioned in the main text. JB22 is now highlighted in green on Fig 2C.

1. Although I generally avoid commenting on formatting, I found the manuscript to be dense. As mentioned above, I truly enjoyed reading it! But I couldn't help but think of ways to make the manuscript more concise for readers. The Results section spans nine pages (excluding figure captions), and the Discussion is five pages long. The main text contains 6 figures with approximately 27 panels and 32 plots and Venn diagrams, while the supplementary material has 11 figures with 22 panels and about 59 plots. Altogether, the manuscript comprises 17 figures, 49 panels, and roughly 91 plots and Venn diagrams! While I will not request any changes, I encourage the authors to consider streamlining the text/data where possible to focus on the core theme of the study.

We thank the reviewer for these suggestions and have reorganised some of our figures and text to appear less dense. We have also added several figures and panels in response to reviewer comments. While we endeavor to make our points clear and concise in the main figures, we believe that it is important to retain key supplementary figures so that an interested reader can evaluate the data in more detail:

A summary of our major changes to the figures is below, and we also provide a manuscript with changes tracked for the reviewers' convenience:

Fig 2:

Added Panel E in response to reviewer comments. Fig 3:

Removed axes for pfl3 and pfl7 from Fig 3C, as the point was made by the other genes displayed (mbx2, pfl8 and gsf2) Replaced Fig 3D with similar data from an improved experiment in response to reviewer comments. Added New Fig 3F from Original Supp Fig 5 Fig 5:

Moved Original Fig 5A to New Supp Fig 10A. Added New Fig 5F in response to reviewer comments. Original Supp Fig 4 / New Supp Fig 6:

Removed mbx2 overexpression images from Original Fig 4C, to be replaced by new overexpression data and images in New Fig 3D. Added flask images for srb10 and srb11 deletion mutants from Original Supp Fig 5A to New Supp Fig 6C. Added microscope image for srb11 deletion mutant from Ooriginal Supp Fig 5A to New Supp Fig 6C. Added adhesion assay results from Original Supp Fig 5C to New Supp Fig 6C. Added New Supp Fig 6D in response to review Original Supp Fig 5

Removed this figure. Original Supp Fig 5A and 5B were moved to New Supp Fig 6. Original Supp Fig 5B was removed to make the manuscript more concise. Original Supp Figs 6, 7 and 8 were combined into New Supp Fig 8.

Original Supp Fig 6A and 6B are now New Supp Fig 8A and 8B. Original Supp Fig 7 is now New Supp Fig 8C. Original Supp Fig 8A is now New Supp Fig 8D and 8E. Original Supp Fig 8B is now New Supp Fig 8F Original Supp Fig 9/New Supp Fig 10

Added Original Fig 5A as new Supp Fig 10A. Original Supp Fig 11/New Supp Fig 12

Removed Original Fig 11B and the relevant text to make the manuscript more concise. Added New Supp Fig 12B in response to reviewer comments. New Supplementary Figures added in response to reviewer comments:

New Supp Fig 4: Microscopy images of natural isolates. New Supp Fig 5: Flask images of natural isolates New Supp Fig 7: Microscopy and flask images of mbx2 overexpression strains. New Supp Fig 9: Genomic comparisons between JB759 and the MLP-forming wild isolate, JB914. Removed some less relevant points from our discussion, to reduce the length.

Added new Supplementary Tables:

Supplementary Table 13: Variants in candidate genes. Added in response to reviewer comments Supplementary Table 14: List of plasmids used in the study.

**Referees cross-commenting**

There are many useful recommendations from all the other reviewers that will help improve the final product. Once those points are revised, I think this will be a nice paper of interest to folks interested in natural variation in MLPs and its genetic background.

Significance

My expertise: evolutionary genetics, evolution of multicellularity, yeast genetics, experimental evolution

Overall, the manuscript is well-written, dense yet logically structured, and the figures are well presented. The combination of phenotypic, genetic, and bioinformatics analyses, particularly from wet lab experiments, is commendable. The study addresses a significant gap in our understanding, primarily explored in budding yeast, by providing comprehensive data on MLP diversity in fission yeast and the interplay of genetic and environmental factors.

In summary, I enjoyed reading the manuscript and have only a few minor suggestions to strengthen the paper.

Reviewer #2

Evidence, reproducibility and clarity

REVIEWER COMMENTS

Yeast species, including fission yeast and budding yeast, could form multicellular-like phenotypes (MLP). In this work, Kӧvér and colleagues found most proteins involved in MLP formation are not functionally conserved between S. pombe and budding yeast by bioinformatic analysis. The authors analyzed 57 natural S. pombe isolates and found MLP formation to widely vary across different nutrient and drug conditions. The authors demonstrate that MLP formation correlated with expression levels of the transcription factor gene mbx2 and several flocculins. The authors also show that Cdk8 kinase module and srub11 deletions also resulted in MLP formation. The experimental design is logic, the manuscript is well-written and organized. I have a few concerns that should be addressed before the publication.

Major points:

1) Line 61-62, how did the authors grow yeast cells in the liquid medium? Shaking or static? If shaking, the nutrient should be even distributed in the medium.

If static culture, most single yeast cells could precipitate on the bottom, how do you address the advantage of flocculation for increasing the sedimentation? In addition, under static culture, the bottom will have less air than the up medium, how to balance the air and nutrients?

Reply: In line 61-62 we stated that "Similarly, flocculation could increase sedimentation in liquid media, thereby assisting the search for more nutrient-rich or less stressful environments (4)".

Our intent was to speculate on the advantages of multicellular-like growth, and cited a review article which has mentioned sedimentation. After further consideration, we decided that this is a minor point and is rather speculative, and removed it altogether from the manuscript.

In response to the Reviewer's question about how cells were grown in liquid medium, throughout the paper we used shaking cultures for our flocculation assays and for pre-cultures. We have made this more clear in the text where it was ambiguous (e.g. line 189, throughout the methods section, and in the legend of Fig. 2A).

2) Line 555, it will be interesting to test whether overexpression of mbx2 could cause flocculation in YES medium. In Figure 3D, the authors use two control strains, but only one mbx2 OE strain, mbx2 OE should be tested in both strains. In addition, did the authors transform empty plasmid into the control strains, please indicate in the figure.

In this experiment, mbx2 was overexpressed using a thiamine-repressible nmt1 promoter, which is a standard construct in fission yeast studies. Assaying MLP formation was not feasible in YES with this strain, because YES is a rich media made up of yeast extract which contains thiamine. Thus, we could not remove thiamine from the media to trigger mbx2 overexpression.

In order to test the influence of mbx2 overexpression in YES, we constructed strains in which mbx2 was integrated into the genome and expression was driven by the rpl2102 promoter, which has been shown to provide constitutive moderate expression levels (12). We observed strong flocculation in both EMM and YES (Fig 3D, New Supplementary Figure 7) . We did not see strong flocculation in a control in which GFP was expressed under the rpl2102 promoter. The flocculation phenotype was so strong that our original adhesion assay protocol required modification for this experiment, including resuspension in 10 mM EDTA before repinning (Methods). We observed strong adhesion for the mbx2 overexpression strains (Fig 3D), but not for control strains in YES. We could not check adhesion in EMM for those strains because cells pinned on EMM did not survive resuspension in EDTA.

We performed these experiments in two backgrounds, 968 h90 (JB50), which is one of the parental strains of the segregant library analysed in Figure 3 and 972 h- (JB22), which is an appropriate background for the gene deletion collection.

We have replaced the data from the original Figure 3D with the new adhesion assay and added New Supplementary Figure 7 to the manuscript (Lines 236-244).

This result also helped us to further refine our model for the pathway. We can now say that the repression of MLPs in rich media must act via Mbx2, as overexpression of mbx2 is sufficient to abolish it, and is likely to act transcriptionally (if it acted on the protein level, the mild overexpression would likely not have led to the phenotype) (Figure 6, Lines 554-556 in the discussion)

3) Line 600-601, the authors may do the backcross of srb11Δ::Kan to exclude the possibility caused by other mutations.

Reply: We thank the reviewer for noticing our concern about suppressor mutations arising in the srb11Δ strain obtained from our deletion library. This initial concern arose following the observation that while qualitatively the srb11Δ::Kan and srb11Δ(CRISPR) strains were both strongly adhesive, there was a minor quantitative difference in their adhesion.

As we obtained this strain from an h+ deletion library strain backcrossed with a prototrophic h- strain (JB22) in order to restore auxotrophies (13), the chances for a suppressor mutation to arise are very low. We have therefore removed that language from our text. We now suspect that a more likely explanation for this small difference could be the strain background, as our CRISPR engineered strain was made in a JB50 background which has the h90 mating type, while the deletion library strains are h- without auxotrophic markers.

We would like to emphasize, however, that despite this quantitative difference in the adhesion phenotype between the two srb11Δ strains, they both have a large increase in the adhesion phenotype relative to the respective wild-type strains. To address this point, we have removed the unnecessary statistical comparison of these two deletion strains and focused on their qualitatively high levels of adhesion in the text (lines 267-269) and in our Revised Supplementary Figure 6D.

Minor points:

1) Line 506, what are the growth conditions of cells in Figure 2A? Did the authors use the liquid or solid medium? Please mention in the Methods or figure legends.

Reply: We have updated the manuscript to include the relevant details in the text (line 189), figure caption for Fig. 2A and in the methods section (lines 829-831).

2) Line 533-535, please explain why the strains exhibiting strong adhesion have a decreased growth rate. Is there any related research? Please add some references.

Reply: Please see reply to Reviewer 1, comment 5.

**Referees cross-commenting**

I agree with most of the comments from other reviewers. This publication may indeed be of interest to a minor area. But the results and the interpretations of the data are interesting and warranted, the findings are scientifically important.

Significance

The authors did many large-scale screens and bioinformatic analyses. The experiments in the manuscript are generally logical and sound. This study is useful for deciphering the mechanism of multicellular-like phenotype formation in the fission yeast, with some implications for some other organisms.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

Summary: Using a variety of targeted and genome wide analyses, the authors investigate the basis for "multicellular-like phenotypes" in S. pombe. Authors developed several methodologies to detect and quantify "multicellular-like phenotypes" (flocculation, aggregation...) and defined genes involved in these processes in laboratory and wild S. pombe.

SECTION A - Evidence, reproducibility and clarity

This is a very solid manuscript that is well-written and supported by convincing data. While one can imagine many additional experiments, the manuscript stands on its own and presents a quite exhaustive analysis of the area. I commend the author for their rigorous work and clear presentation. They are only a few minor points that warrant comments or corrections: - Supplementary Figure 1 is a typical example of the "necessity" to have statistics and P-values everywhere. The data are convincing but what is the evidence that the Filtering assay and the Plate-reader assay values should be linearly related? Lets imagine that Plate-reader assay value is proportional to the square of the Filtering assay value. What would be the Pearson R and P-value in this case? What is most appropriate? Why would one use a linear correlation? What is the "real" significance?

Reply: We thank the reviewer for pointing out that the data in Supplementary Figure 1 does not appear to be linear and, therefore, reporting the Pearson correlation coefficient may not be the best way to represent the relationship between the two assays. The nonlinear nature of this data could indicate that

The filtering assay saturates before the plate reader assay, and is less able to distinguish between strains that flocculate strongly and The filtering assay may be more sensitive for strains that show lower levels of flocculation. In general, we observed fewer strains with intermediate phenotypes for both assays, making it difficult to ascertain the true relationship between them; however, we believe that the key result is that the strains with the highest level of flocculation have the highest values in both assays. To capture this aspect of the data, we now report the Spearman correlation which is non-parametric and indicates how similar the ranking of each strain is based on both assays. With the alternative hypothesis being that the correlation is > 0, we report a Spearman correlation coefficient of 0.24 and a P-value of 0.04 (lines 823-826)

• Minor points: * They are several "personal communications" in the manuscript (page 11, page 18, page 23). It should be checked whether this is accepted in the journal that publishes this manuscript.

Reply: We thank the reviewer for highlighting this issue. We had three instances of "personal communications" in our original submission.

The first instance was an acknowledgement for advice on our DNA extraction protocol from Dan Jeffares. We now include this in the Acknowledgements section instead.

The second communication with Angad Garg described that they observed flocculation while growing cells in phosphate starvation conditions, which was not reported in their publication (14). Though we appreciate their willingness to share unpublished data with us, we have removed this observation from our manuscript and instead rely only on our own observations and arguments based on their published RNA-seq data to make our point.

The third personal communication with Olivia Hillson supplements a minor hypothesis, namely that deletion of SPNCRNA.781 might cause MLP formation by affecting the promoter of hsr1, for which we had access to unpublished ChIP-seq data, showing its binding to flocculins. Recently published work from a different group (15) also suggests this link between hsr1 and flocculation and is now discussed in our manuscript instead of the result based on unpublished data obtained from personal communication at Lines 397-398.

* Page 4 check "a few regulators"

Reply: For clarity, this has now been changed to "several regulatory proteins" at Line 108. The specific proteins we are referring to are highlighted in Figure 1C.

* Page 19, line 567: "remaining 8 strains" may be confusing as Material and Methods states "remaining 10 strains".

Reply: Two of the 10 strains were found to be redundant after sequencing as explained in the Methods (Lines 930-934). Therefore, we only added 8 new strains to the analysis. We thank the reviewer for highlighting this as a potential source of misunderstanding, and clarified this point in the text (Lines 247-250 and in the methods).

**Referees cross-commenting**

I concur with most comments. Overall, the reviewers agree that this is a solid piece of work that could benefit from minor modifications and should be published. I reiterate that, for me, despite its quality, this publication will only be of interest to specialists.

Reviewer #3 (Significance (Required)):

A limited number of studies have investigated "multicellular-like phenotypes" in S. pombe. This manuscript brings therefore new and solid information. Yet, despite an impressive amount of work, our conceptual advance in understanding this process and its phylogenetic conservation remains limited. This is probably best illustrated in the figure 6 that summarize the study and contains 3 question marks and an additional unknown mechanism. (Most of the solid arrows in this figure correspond to interactions within the Mediator complex that were well known before this study.) In addition, while only few studies have been published in this area, the authors' findings are often only bringing additional support to already published observations. Overall, while this manuscript will be of interest to a restricted group of aficionados, it will most likely not attract the attention of a wide readership.

__ Reviewer #4 (Evidence, reproducibility and clarity (Required)):__

In this manuscript, the authors explore how multicellular-like phenotypes (MLPs) arise in the fission yeast S. pombe. Although yeasts are characterized as unicellular fungi, diverse species show MLPs, including filamentous growth on agar plates and flocculation in liquid media. MLPs may provide certain advantages in nutritionally poor conditions and protection against external challenges, upon which natural selection can then act. Previous work on MLPs has mostly been carried out in the budding yeasts S. cerevisiae and C. albicans, and little was known about these behaviors in S. pombe. The authors thus set out to investigate both genetic and environmental regulators of MLP formation.

First, their analysis of published data revealed a limited number of shared regulators of MLP between S. pombe, S. cerevisiae, and C. albicans, although the cell adhesion proteins themselves are largely not conserved. Next, the authors screened a set of non-clonal natural isolates using two high-throughput assays that they developed and found that MLPs vary in strains and depending on nutrient conditions. Focusing on a natural isolate that showed both adhesion on agar plates and flocculation in liquid medium, they then analyzed a segregant library generated from this and a laboratory strain using their assays. Using QTL analysis, they uncovered a frameshift in the srb11 gene, which encodes a subunit of the Mediator complex, as the likely causal inducer of MLP. This was confirmed by additional analyses of strains lacking srb11 or other members of Mediator. Furthermore, the authors showed that loss of srb11 function resulted in the upregulation of the Mbx2 transcription factor, which was both necessary and sufficient for MLP formation in this background. Finally, screening of two additional yeast strain collections (gene and long intergenic non-coding RNA deletion) identified both known and novel regulators representing different pathways that may be involved in MLP formation.

Altogether, this study provides new perspectives into our understanding of the diverse inputs that regulate multicellular-like phenotypes in yeast.

Major comments:

• The methods for screening for adhesion and flocculation are well described, with representative figures that show plates and flasks. However, there are few microscopy images of cells, and it would be interesting and helpful for the reader to have an idea of how cells look when they exhibit MLPs. For instance, are there any differences in cell shape or size when strains present different degrees of adhesion or flocculation? In addition, the authors mention that mutants with strong adhesion generally had lower colony density and are likely to be slower growing. Although their analyses suggest otherwise (page 22), this has a potential for introducing error in their observations, and including images of the adhesion/flocculation phenotypes may provide further support for their conclusions. I suggest that the authors present microscopy images 1) similar to what is shown for JB759 in Figure 2A and 2) of cells growing on agar in the adhesion assay. This could be included for the different Mediator subunit deletions that they tested, where there appear to be varying phenotypes. It could also be informative for a subset of the 31 high-confidence candidates that they identified in their screen.

Reply: We thank the reviewer for highlighting the need for further microscopic characterisation of MLP forming strains. We therefore now include images of JB914, JB953 (New Supplementary Figures 4, Figure 2E) in liquid media in EMM, EMM-Phosphate, and YES; an srb11 deletion strain (Figure 3F), and mbx2 overexpression strains (New Supplementary Figure 7).

• Upon identifying a frameshift in srb11 that is responsible for the MLP, the authors assessed whether deletion of other Mediator subunits would result in the same phenotype. They found that srb10 and srb11 deletions both flocculate and show adhesion, while other mutants had milder phenotypes. However, the authors also found that a new deletion of srb11 that they generated had a stronger adhesion phenotype than the srb11 deletion from the prototrophic deletion library, which was attributed this the accumulation of suppressor mutations in the strains of the deletion collection. As the authors make clear distinctions between the phenotypes of different Mediator mutants, I suggest generating and analyzing "clean" deletions of the 6 other subunits that they tested. This would strengthen their conclusion and help to rule out accumulated suppressors as the cause of the differences in the observed phenotypes.

Reply: We thank the reviewer for noticing our concern about suppressor mutations in the manuscript. As we describe above in response to a similar question from reviewer 2, as the prototrophic deletion library from which we extracted the Mediator deletion strains had been backcrossed during its construction (13), we no longer suspect that small difference between the srb11Δ::Kan strain from the deletion library and the newly created srb11Δ (CRISPR) strains is due to suppressor mutations. Rather, we think they may be a result of the difference in genetic background and possibly mating type between the two strains. We also want to emphasize that this difference is small compared to the difference between the adhesion ratios of the srb11Δ strains and their respective control strains.

Nevertheless, we made clean, independent Mediator mutants for 5 out of 6 Mediator genes tested (med10Δ, med13Δ, med19Δ, med27Δ, and srb10Δ) as well as an additional mutant that we didn't have in our library, med12Δ (Figure R9). When running the assay on these new strains we got an overall lower dynamic range, possibly due to variations in the water flow rate relative to the first assay. However, we saw a strong phenotype for both library and our own srb10Δ and CRISPR srb11Δ strains. We did not see a significant increase in adhesion for the other Mediator deletion mutants in EMM relative to wild type with the exception of for med10Δ in both the library strain and for our clean mutant, for which we did not observe a phenotype in our previous experiment. We included the experiment for the newly created mutants as New Supplementary Figure S6E and described them in lines 276-281 in our revised manuscript.

Minor comments:

• One point that recurs in the manuscript is the idea that mutations that give rise to strong MLPs also generally lead to slower growth, representing a potential trade-off. This idea could be reinforced with measurements of growth rate or generation time by optical density or cell number, for instance, rather than comparisons of colony density. Also, it would be interesting to mention if the slow growth phenotype is only observed in MLP-inducing conditions or also in rich medium.

Reply: As described above in response to item 5 from Reviewer 1, we have conducted growth assays in liquid media for srb10Δ, srb11Δ, and other mutants from our adhesion screen (tlg2Δ, rpa12Δ, mus7Δ and kgd2Δ) that showed a similar phenotype to those genes in both minimal (EMM) and rich (YES) media. We observe that in rich media, srb10Δ and srb11Δ cells grow similarly to control strains, and they exhibit a lower decrease in growth rate than the other similarly adhesive strains. Both mus7Δ and kgd2Δ cells grow more slowly, even in rich media.

We have also added data on the tradeoff between growth and adhesion based on growth on solid media from (11) for all mutants identified in our screen (New Supp Fig 12B)).

Thus, the relationship between slow growth and clumpiness depends on the mutation, and specifically, mutations of the Mediator, including those to srb11 and srb10, seem to decrease the impact of any tradeoff between growth and adhesion.

• The authors show that the MLPs of the srb10 and srb11 deletions occur through mbx2 upregulation. Do the varying strengths of the phenotypes of the strains lacking different Mediator subunits correlate with mbx2 levels in these backgrounds?

Reply: There is some evidence from previous work that the relationship between the strength of the MLPs and the expression of mbx2 may not be perfectly proportional. In (16), med12Δ had a higher (though qualitatively comparable) level of mbx2 upregulation than srb10Δ (New Supp Fig 8E), even though that paper reported a milder phenotype for med12Δ than for srb10Δ cells. We did not observe a significant increase in adhesion in our med12Δ strain (New Supp Fig 6D). This suggests that in the case of these mutants, it is not simply the level of mbx2 that controls MLP formation, but that there are likely additional regulatory mechanisms. We have added some discussion on this context in the manuscript (lines 545-547).

**Referees cross-commenting**

I agree overall with the comments and suggestions from the other reviewers. The revision would require only minor modifications. The paper is interesting both for the combination of methodologies used and its findings, and I believe that it would benefit a growing community of researchers.

Reviewer #4 (Significance (Required)):

This study employed a variety of methods that allowed the authors to uncover previously unknown regulators of MLPs. Taking advantage of the diversity of natural fission yeast isolates as well as the constructed gene and non-coding RNA deletion collections, the authors identified novel genetic determinants that give rise to MLPs, opening new avenues into this exciting area of research. The overall conclusions of the work are solid and supported by the reported results and analyses. This study will be appreciated by a broad audience of readers who are interested in understanding how organisms respond to environmental challenges as well as how MLPs may result in emergent properties that play key roles in these responses. Some of the limitations of the work are described above, with recommendations for addressing these points.

Keywords for my field of expertise: fission yeast, cell cycle, transcription, replication.

References for Response to Reviews

1. Brysch-Herzberg M, Jia GS, Seidel M, Assali I, Du LL. Insights into the ecology of Schizosaccharomyces species in natural and artificial habitats. Antonie Van Leeuwenhoek. 2022 May 1;115(5):661-95.
2. Jeffares DC, Rallis C, Rieux A, Speed D, Převorovský M, Mourier T, et al. The genomic and phenotypic diversity of Schizosaccharomyces pombe. Nat Genet. 2015 Mar;47(3):235-41.
3. Ratcliff WC, Denison RF, Borrello M, Travisano M. Experimental evolution of multicellularity. Proc Natl Acad Sci. 2012 Jan 31;109(5):1595-600.
4. Smukalla S, Caldara M, Pochet N, Beauvais A, Guadagnini S, Yan C, et al. FLO1 is a variable green beard gene that drives biofilm-like cooperation in budding yeast. Cell. 2008 Nov 14;135(4):726-37.
5. Lorenz MC, Heitman J. Yeast pseudohyphal growth is regulated by GPA2, a G protein alpha homolog. EMBO J. 1997 Dec 1;16(23):7008-18.
6. Ignacia DGL, Bennis NX, Wheeler C, Tu LCL, Keijzer J, Cardoso CC, et al. Functional analysis of Saccharomyces cerevisiae FLO genes through optogenetic control. FEMS Yeast Res. 2025 Sept 24;25:foaf057.
7. Wang Z, Xu W, Gao Y, Zha M, Zhang D, Peng X, et al. Engineering Saccharomyces cerevisiae for improved biofilm formation and ethanol production in continuous fermentation. Biotechnol Biofuels Bioprod. 2023 July 31;16(1):119.
8. Koschwanez JH, Foster KR, Murray AW. Improved use of a public good selects for the evolution of undifferentiated multicellularity. eLife. 2013 Apr 2;2:e00367.
9. Westman JO, Mapelli V, Taherzadeh MJ, Franzén CJ. Flocculation Causes Inhibitor Tolerance in Saccharomyces cerevisiae for Second-Generation Bioethanol Production. Appl Environ Microbiol. 2014 Nov;80(22):6908-18.
10. Li R, Li X, Sun L, Chen F, Liu Z, Gu Y, et al. Reduction of Ribosome Level Triggers Flocculation of Fission Yeast Cells. Eukaryot Cell. 2013 Mar;12(3):450-9.
11. Rodríguez-López M, Bordin N, Lees J, Scholes H, Hassan S, Saintain Q, et al. Broad functional profiling of fission yeast proteins using phenomics and machine learning. Marston AL, James DE, editors. eLife. 2023 Oct 3;12:RP88229.
12. Hebra T, Smrčková H, Elkatmis B, Převorovský M, Pluskal T. POMBOX: A Fission Yeast Cloning Toolkit for Molecular and Synthetic Biology. ACS Synth Biol. 2024 Feb 16;13(2):558-67.
13. Malecki M, Bähler J. Identifying genes required for respiratory growth of fission yeast. Wellcome Open Res. 2016 Nov 15;1:12.
14. Garg A, Sanchez AM, Miele M, Schwer B, Shuman S. Cellular responses to long-term phosphate starvation of fission yeast: Maf1 determines fate choice between quiescence and death associated with aberrant tRNA biogenesis. Nucleic Acids Res. 2023 Feb 16;51(7):3094-115.
15. Ohsawa S, Schwaiger M, Iesmantavicius V, Hashimoto R, Moriyama H, Matoba H, et al. Nitrogen signaling factor triggers a respiration-like gene expression program in fission yeast. EMBO J. 2024 Oct 15;43(20):4604-24.
16. Linder T, Rasmussen NN, Samuelsen CO, Chatzidaki E, Baraznenok V, Beve J, et al. Two conserved modules of Schizosaccharomyces pombe Mediator regulate distinct cellular pathways. Nucleic Acids Res. 2008 May;36(8):2489-504.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #4

Evidence, reproducibility and clarity

In this manuscript, the authors explore how multicellular-like phenotypes (MLPs) arise in the fission yeast S. pombe. Although yeasts are characterized as unicellular fungi, diverse species show MLPs, including filamentous growth on agar plates and flocculation in liquid media. MLPs may provide certain advantages in nutritionally poor conditions and protection against external challenges, upon which natural selection can then act. Previous work on MLPs has mostly been carried out in the budding yeasts S. cerevisiae and C. albicans, and little was known about these behaviors in S. pombe. The authors thus set out to investigate both genetic and environmental regulators of MLP formation.

First, their analysis of published data revealed a limited number of shared regulators of MLP between S. pombe, S. cerevisiae, and C. albicans, although the cell adhesion proteins themselves are largely not conserved. Next, the authors screened a set of non-clonal natural isolates using two high-throughput assays that they developed and found that MLPs vary in strains and depending on nutrient conditions. Focusing on a natural isolate that showed both adhesion on agar plates and flocculation in liquid medium, they then analyzed a segregant library generated from this and a laboratory strain using their assays. Using QTL analysis, they uncovered a frameshift in the srb11 gene, which encodes a subunit of the Mediator complex, as the likely causal inducer of MLP. This was confirmed by additional analyses of strains lacking srb11 or other members of Mediator. Furthermore, the authors showed that loss of srb11 function resulted in the upregulation of the Mbx2 transcription factor, which was both necessary and sufficient for MLP formation in this background. Finally, screening of two additional yeast strain collections (gene and long intergenic non-coding RNA deletion) identified both known and novel regulators representing different pathways that may be involved in MLP formation.

Altogether, this study provides new perspectives into our understanding of the diverse inputs that regulate multicellular-like phenotypes in yeast.

Major comments:

• The methods for screening for adhesion and flocculation are well described, with representative figures that show plates and flasks. However, there are few microscopy images of cells, and it would be interesting and helpful for the reader to have an idea of how cells look when they exhibit MLPs. For instance, are there any differences in cell shape or size when strains present different degrees of adhesion or flocculation? In addition, the authors mention that mutants with strong adhesion generally had lower colony density and are likely to be slower growing. Although their analyses suggest otherwise (page 22), this has a potential for introducing error in their observations, and including images of the adhesion/flocculation phenotypes may provide further support for their conclusions. I suggest that the authors present microscopy images 1) similar to what is shown for JB759 in Figure 2A and 2) of cells growing on agar in the adhesion assay. This could be included for the different Mediator subunit deletions that they tested, where there appear to be varying phenotypes. It could also be informative for a subset of the 31 high-confidence candidates that they identified in their screen.
• Upon identifying a frameshift in srb11 that is responsible for the MLP, the authors assessed whether deletion of other Mediator subunits would result in the same phenotype. They found that srb10 and srb11 deletions both flocculate and show adhesion, while other mutants had milder phenotypes. However, the authors also found that a new deletion of srb11 that they generated had a stronger adhesion phenotype than the srb11 deletion from the prototrophic deletion library, which was attributed this the accumulation of suppressor mutations in the strains of the deletion collection. As the authors make clear distinctions between the phenotypes of different Mediator mutants, I suggest generating and analyzing "clean" deletions of the 6 other subunits that they tested. This would strengthen their conclusion and help to rule out accumulated suppressors as the cause of the differences in the observed phenotypes.

Minor comments:

• One point that recurs in the manuscript is the idea that mutations that give rise to strong MLPs also generally lead to slower growth, representing a potential trade-off. This idea could be reinforced with measurements of growth rate or generation time by optical density or cell number, for instance, rather than comparisons of colony density. Also, it would be interesting to mention if the slow growth phenotype is only observed in MLP-inducing conditions or also in rich medium.
• The authors show that the MLPs of the srb10 and srb11 deletions occur through mbx2 upregulation. Do the varying strengths of the phenotypes of the strains lacking different Mediator subunits correlate with mbx2 levels in these backgrounds?

Referees cross-commenting

I agree overall with the comments and suggestions from the other reviewers. The revision would require only minor modifications. The paper is interesting both for the combination of methodologies used and its findings, and I believe that it would benefit a growing community of researchers.

Significance

This study employed a variety of methods that allowed the authors to uncover previously unknown regulators of MLPs. Taking advantage of the diversity of natural fission yeast isolates as well as the constructed gene and non-coding RNA deletion collections, the authors identified novel genetic determinants that give rise to MLPs, opening new avenues into this exciting area of research. The overall conclusions of the work are solid and supported by the reported results and analyses. This study will be appreciated by a broad audience of readers who are interested in understanding how organisms respond to environmental challenges as well as how MLPs may result in emergent properties that play key roles in these responses. Some of the limitations of the work are described above, with recommendations for addressing these points.

Keywords for my field of expertise: fission yeast, cell cycle, transcription, replication.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary:

Using a variety of targeted and genome wide analyses, the authors investigate the basis for "multicellular-like phenotypes" in S. pombe. Authors developed several methodologies to detect and quantify "multicellular-like phenotypes" (flocculation, aggregation...) and defined genes involved in these processes in laboratory and wild S. pombe.

SECTION A - Evidence, reproducibility and clarity

This is a very solid manuscript that is well-written and supported by convincing data. While one can imagine many additional experiments, the manuscript stands on its own and presents a quite exhaustive analysis of the area. I commend the author for their rigorous work and clear presentation. They are only a few minor points that warrant comments or corrections:

• Supplementary Figure 1 is a typical example of the "necessity" to have statistics and P-values everywhere. The data are convincing but what is the evidence that the Filtering assay and the Plate-reader assay values should be linearly related? Lets imagine that Plate-reader assay value is proportional to the square of the Filtering assay value. What would be the Pearson R and P-value in this case? What is most appropriate? Why would one use a linear correlation? What is the "real" significance?

Minor points:

• They are several "personal communications" in the manuscript (page 11, page 18, page 23). It should be checked whether this is accepted in the journal that publishes this manuscript.
• Page 4 check "a few regulators"
• Page 19, line 567: "remaining 8 strains" may be confusing as Material and Methods states "remaining 10 strains".

Referees cross-commenting

I concur with most comments. Overall, the reviewers agree that this is a solid piece of work that could benefit from minor modifications and should be published. I reiterate that, for me, despite its quality, this publication will only be of interest to specialists.

Significance

A limited number of studies have investigated "multicellular-like phenotypes" in S. pombe. This manuscript brings therefore new and solid information. Yet, despite an impressive amount of work, our conceptual advance in understanding this process and its phylogenetic conservation remains limited. This is probably best illustrated in the figure 6 that summarize the study and contains 3 question marks and an additional unknown mechanism. (Most of the solid arrows in this figure correspond to interactions within the Mediator complex that were well known before this study.) In addition, while only few studies have been published in this area, the authors' findings are often only bringing additional support to already published observations. Overall, while this manuscript will be of interest to a restricted group of aficionados, it will most likely not attract the attention of a wide readership.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

Yeast species, including fission yeast and budding yeast, could form multicellular-like phenotypes (MLP). In this work, Kӧvér and colleagues found most proteins involved in MLP formation are not functionally conserved between S. pombe and budding yeast by bioinformatic analysis. The authors analyzed 57 natural S. pombe isolates and found MLP formation to widely vary across different nutrient and drug conditions. The authors demonstrate that MLP formation correlated with expression levels of the transcription factor gene mbx2 and several flocculins. The authors also show that Cdk8 kinase module and srub11 deletions also resulted in MLP formation. The experimental design is logic, the manuscript is well-written and organized. I have a few concerns that should be addressed before the publication.

Major points:

1. Line 61-62, how did the authors grow yeast cells in the liquid medium? Shaking or static? If shaking, the nutrient should be even distributed in the medium. If static culture, most single yeast cells could precipitate on the bottom, how do you address the advantage of flocculation for increasing the sedimentation? In addition, under static culture, the bottom will have less air than the up medium, how to balance the air and nutrients?
2. Line 555, it will be interesting to test whether overexpression of mbx2 could cause flocculation in YES medium. In Figure 3D, the authors use two control strains, but only one mbx2 OE strain, mbx2 OE should be tested in both strains. In addition, did the authors transform empty plasmid into the control strains, please indicate in the figure.
3. Line 600-601, the authors may do the backcross of srb11Δ::Kan to exclude the possibility caused by other mutations.

Minor points:

1. Line 506, what are the growth conditions of cells in Figure 2A? Did the authors use the liquid or solid medium? Please mention in the Methods or figure legends.
2. Line 533-535, please explain why the strains exhibiting strong adhesion have a decreased growth rate. Is there any related research? Please add some references.

Referees cross-commenting

I agree with most of the comments from other reviewers. This publication may indeed be of interest to a minor area. But the results and the interpretations of the data are interesting and warranted, the findings are scientifically important.

Significance

The authors did many large-scale screens and bioinformatic analyses. The experiments in the manuscript are generally logical and sound. This study is useful for deciphering the mechanism of multicellular-like phenotype formation in the fission yeast, with some implications for some other organisms.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Summary

Köver et al. examine the genetic and environmental underpinnings of multicellular-like phenotypes (MLPs) in fission yeast, studying 57 natural isolates of Schizosaccharomyces pombe. They uncover that a noteworthy subset of these isolates can develop MLPs, with the extent of these phenotypes varying according to growth media. Among these, two strains demonstrate pronounced MLP across a range of conditions. By genetically manipulating one strain with an MLP phenotype (distinct from the previously mentioned two strains), they provide evidence that genes such as MBX2 and SRB11 play a direct role in MLP formation, strengthening their genetic mapping findings. The study also reveals that while some key genes and their phenotypic effects are strikingly similar between budding and fission yeast, other aspects of MLP formation are not conserved, which is an intriguing finding.

Overall, the manuscript is well-written, dense yet logically structured, and the figures are well presented. The combination of phenotypic, genetic, and bioinformatics analyses, particularly from wet lab experiments, is commendable. The study addresses a significant gap in our understanding, primarily explored in budding yeast, by providing comprehensive data on MLP diversity in fission yeast and the interplay of genetic and environmental factors.

In summary, I enjoyed reading the manuscript and have only a few minor suggestions to strengthen the paper:

Minor revisions:

1. Although this may seem like a minor revision, but it is a crucial point. Please make sure that all raw data used to generate figures, run stats, sequence data, and scripts used to run data analysis are made publicly available. Provide relevant accession numbers and links to public data repositories. It is important that others can download the various types of data that went into the major conclusions of this paper in order to replicate your analysis or expand upon the scope of this work. I am not sure if the journal has a policy regarding this, but it should be followed to allow for transparency and reproducibility of the research.
2. Two out of 57 strains exhibit strong and consistent MLP across multiple environments. Providing more information on these strains (JB914 and JB953), such as their natural habitats and distinct appearances of their MLP phenotypes under varying conditions, would provide valuable insights.

First, a brief discussion highlighting what differentiates these two strains from the rest would be helpful for readers (e.g. insight into their unique genetic and environmental background that might be linked to the MLP phenotype).

Additionally, culture tube and microscopy images of these strains, similar to those presented for JB759 in Figure 2A, can be included in the supplementary materials. My reasoning is that these images could help illustrate variation or lack thereof in aggregative group size across different media. 3. The phenotypic outcome of overexpressing MXB2 is striking, as shown in Supplementary Figure 4C. Incorporating at least one of the culture tube images depicting large flocs into the main text, perhaps adjacent to Figure 3 panel D, would improve the visual appeal and highlight this key finding (at the moment those images are only shown in the supplementary materials). 4. I know that the authors discuss the knowledge gap in the intro and results, but the abstract does not mention this critical gap. Please stress this critical gap (i.e., MLPs understudied in fission yeast) with a brief sentence in the abstract. Similarly, please consider writing a brief concluding sentence summarizing the paper's most significant finding referring to the knowledge gap would provide a clearer takeaway message for the reader - the abstract ends abruptly without any conclusion. 5. The observation that strains with adhesive phenotypes have a lower growth rate compared to non-adhesive strains is a noteworthy point (lines 532-535). This represents yet another example of this classical trade-off. This point could be emphasized in the Discussion or alongside the relevant result, with a brief speculative explanation for this phenomenon. 6. The text mentions two lab strains, JB22 and JB50, displaying strong adhesion under phosphate starvation (lines 525-526), yet the data point for JB22 in Figure 2C is not labeled. 7. Although I generally avoid commenting on formatting, I found the manuscript to be dense. As mentioned above, I truly enjoyed reading it! But I couldn't help but think of ways to make the manuscript more concise for readers. The Results section spans nine pages (excluding figure captions), and the Discussion is five pages long. The main text contains 6 figures with approximately 27 panels and 32 plots and Venn diagrams, while the supplementary material has 11 figures with 22 panels and about 59 plots. Altogether, the manuscript comprises 17 figures, 49 panels, and roughly 91 plots and Venn diagrams! While I will not request any changes, I encourage the authors to consider streamlining the text/data where possible to focus on the core theme of the study.

Referees cross-commenting

There are many useful recommendations from all the other reviewers that will help improve the final product. Once those points are revised, I think this will be a nice paper of interest to folks interested in natural variation in MLPs and its genetic background.

Significance

My expertise: evolutionary genetics, evolution of multicellularity, yeast genetics, experimental evolution

Overall, the manuscript is well-written, dense yet logically structured, and the figures are well presented. The combination of phenotypic, genetic, and bioinformatics analyses, particularly from wet lab experiments, is commendable. The study addresses a significant gap in our understanding, primarily explored in budding yeast, by providing comprehensive data on MLP diversity in fission yeast and the interplay of genetic and environmental factors.

In summary, I enjoyed reading the manuscript and have only a few minor suggestions to strengthen the paper.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1

Evidence, reproducibility and clarity

The manuscript by Wu and Griffin describes a mechanism where CHD4 and BRG1, two chromatin remodelling enzymes, have antagonistic functions to regulate extracellular matrix (ECM) plasmin activity and sterile inflammatory phenotype in the endothelial cells of the developing liver. As a follow up from a previous study, the authors investigate the phenotype of embryonic-lethal endothelial-specific CHD4-knockout, leading to liver phenotype and embryo death, and the rescue of this phenotype when subsequently BRG1 is knocked-out also in the endothelium. First, the authors show that the increase in plasmin activator uPAR (which leads to ECM degradation) in CHD4-KO embryos can be rescued by BRG1-KO, and that both CHD4 and BRG1 interact with the uPAR promoter. However, the authors demonstrate that reducing plasminogen by genetic knockout is unable to rescue the CHD4-KO embryos alone, suggesting an additional mechanism. By RNAseq analysis, the authors identify sterile inflammation as another potential contributor to the lethal phenotype of CHD4-KO embryos through increased expression of ICAM-1 in endothelial cells, also showing binding of both chromatin remodellers to ICAM-1 promoter. Finally, the authors use nonsteroidal anti-inflammatory drug carprofen, alone or in combination with plasminogen genetic knockout, and demonstrate CHD4-KO lethal embryonic phenotype rescue with the combination of plasminogen reduction and inflammation reduction, highlighting the synergistic role of both ECM degradation and sterile inflammation in this genetic KO.

The findings of the manuscript are interesting, experiments well controlled and paper well written. While the work is of potential specialist interest to the field of liver development, there are several issues which authors should address before this paper can be published:

Major issues:

1. The authors still see embryonic lethality of some embryos with endothelial BRG1-KO or combined endothelial CHD4/BRG1-KO - could the authors please show or at least comment in the discussion why those animals are dying?

We observed no dead Brg1-ECko or Brg1/Chd4-ECdko embryos by E14.5. However, at E17.5, there was an 18.8% lethality rate for Brg1-ECko mutants and a 12.5% rate for Brg1/Chd4-ECdko mutants (Fig. 1B). The reasons behind the incomplete rescue of Brg1/Chd4-ECdko embryos and the cause of death in Brg1-ECko mutants remain unknown, as we have mentioned in the revised discussion (see lines 311-316).

1. In the qRT-PCR results Fig.2c, what is each dot?

Each dot represents transcripts acquired from a separate embryo. We have modified the figure legend for clarification.

1. In the same figure, I would expect that in CHD4-KO there is no CHD4 transcript, and in BRG1-KO there is no BRG1 transcript, rather than the reduction shown, which seems quite noisy (though significant) - is it this a result of normalisation? Or is indeed only a certain amount of the transcript reduced?

The VE-Cadherin Cre mouse line utilized in this study is reported to have progressive Cre expression and activity from E8.5 to E13.5 and only to reach full penetrance across all vasculature at E14.51. The liver sinusoidal ECs (LSECs) analyzed in Fig. 2C were isolated at E12.5, before Cre activity reached its full penetrance. This is likely the primary cause of the variability in gene excision seen in this panel.

1. In the same figure, is the statistical testing performed before or after normalisation? This can introduce errors if done after normalisation.

Normalization was performed before statistical analysis to combine relative transcript counts from embryos harvested in multiple litters. This is now clarified in our methods (see lines 486-489).

1. In some cases, the authors show immunofluorescence images but do not specify how many biological replicates this represents (e.g. Fig.1d, 4c-d). This should be added.

We have updated the legends for Figs. 1E, 4C-D, and 6E-F, as suggested.

1. I also encourage the authors to present a supplementary figure with at least one other biological replicate shown for imaging data (optional).

We appreciated this suggestion but opted not to add additional supplemental figures, which might have been confusing to readers.

1. The plasminogen reduction by genetic modulation results in drastic changes to the embryos' appearance - is this a whole embryo KO or endothelial-specific KO? Can authors at least comment on the differences?

The plasminogen-deficient embryos used in this study were global knockouts; this is now clarified on line 177. The Chd4-ECko embryos with varying degrees of plasminogen deficiency that are shown in Fig. 2F were dissected at E17.5, which is ~3 days after the typical time of death for Chd4-ECko embryos. This explains why the dead and partially resorbed mutants in Fig. 2F look so different from their control (Plg-/-) littermate and from the E14.5 Chd4-ECko embryos shown in Fig. 1C.

1. In Fig.2b, do I understand correctly only 1 sample was analysed with different areas plotted on the graph? If so, this experiment should be repeated on another set of embryos to be robust, and data plotted as a mean of each embryo (rather than areas).

Each dot represents the mean value obtained after quantifying 4 fluorescent areas within a liver section from a single embryo. The N number indicates the number of embryos used from each genotype. We have updated the figure legend accordingly.

1. Also in some graphs, authors specify that it was more than n>x embryos, but then - what are the dots on the graph representing? Each embryo? This should be specified (e.g. Fig.2b-c, but please check this in all the figure legends).

Thank you for this question. We have worked to clarify the legends for all our graphs. Overall, for graphs related to embryos, each dot represents data from a single embryo. Since the sample sizes vary across genotypes, we used the smallest sample size taken from the mutant groups when listing our minimum N.

1. "we found Plaur was the only gene that was induced in CHD4-ECko LSECs at E12.5 (Figure S3D)." - I am not sure this is correct, as gene Plau is also increased in 2/3 samples?

Although Plau transcripts were also increased in Chd4-ECko LSECs compared to control samples, our statistical analysis showed a p-value of 0.0564, which was deemed non-significant according to our cutoff criteria of p

1. I find the title and the running title somewhat misleading and too broad; the authors should specify more detail in the title about the content of the paper - the current statement of the title is somewhat true but shown only for one genetic model and not confirmed for all types of "lethal embryonic liver degeneration".

We have updated the title to incorporate this suggestion. The revised title is ‘Plasmin activity and sterile inflammation synergize to promote lethal embryonic liver degeneration in endothelial chromatin remodeler mutants.’ The revised running title is ‘Plasmin and inflammation in endothelial mutant livers.’

Minor issues:

1. If an animal licence was used, its number should be specified in the ethics or methods section

We have added this information to the methods (see line 383).

1. In fig.3g it is very hard to see each of the samples, could authors try to improve this graph for clarity using colours-or split Y axis - or both?

We have revised Fig. 3G to include a split y-axis, as suggested.

1. "This indicates that ECs can play a pro-inflammatory role in embryonic livers and highlights the need for tight regulation to ensure normal liver growth." This sentence for me is misleading, EC are producing inflammatory signals only during the CHD4-KO according to the author's data, and authors do not show such data in normal homeostasis condition. Actually, the pro-inflammatory role here seems detrimental, and ECs should not exhibit it for correct development. The authors should rephrase this to be clearer.

The detrimental inflammation observed when Chd4 was deleted in ECs indicates that endothelial CHD4 normally suppresses inflammation during liver development (Fig. 3F-G, and 4A-B). When endothelial CHD4 functions properly, there is no excessive cytokine activation and inflammation. We have modified the sentence to help clarify this information (see lines 295-297).

Significance

General assessment: The study is well controlled and well written. The findings are interesting. The limitation of the findings is only 1 combination genetic model being studied, and it is unclear if the synergistic effect of sterile inflammation and ECM degradation is broadly applicable to other models, where embryo dies because of liver failure.

Advance: The study makes an incremental advance, following up findings from a previous study. However, it is conceptually interesting.

Audience: The audience for this manuscript would be a liver development specialist. However, broader concepts could also be applicable to liver disease.

Expertise: I research in the field of liver regeneration and disease.

__Reviewer #2 __

Evidence, reproducibility and clarity

In essence, Wu et Al find that Chd4 mutant mice exhibit embryonic liver degeneration due to uPA-mediated plasmin hyperactivity and an ICAM-1-driven hyperinflammation and that additional mutation of BRG1 opposes this liver degeneration, possibly via ICAM-1.

Generally, this is an excellent manuscript with a very logical sequence of experiments, although it has shortcomings such as validating their findings in an independent system, ideally human, and further establishing the translational relevance. Establishing translational relevance through mechanistic experiments that identify specific inflammatory tissue pathways, such as by blocking ICAM-1 and TNF-alpha, could also define developmental aberrations as a model for broader (patho)physiology and thereby enhance the impact on the field.

Major

1. The embryonic and postnatal survival data of Chd4-ECko and Brg1/Chd4-Ecdko mice should be included in Fig. 1

We revised Fig. 1 to add representative photos and lethality rates for control and mutant embryos at E17.5 (see new Fig. 1B). All Chd4-ECko embryos we dissected at E17.5 were dead, which was consistent with our previous report2. Although Brg1/Chd4-ECdko embryos were largely rescued at E17.5, these mutants still die soon after birth due to lung development issues, as we previously reported3.

1. What is the impact of Chd4-ECko and Brg1/Chd4-ECdko on the multicellular microenvironment? At a minimum, IF or spatial transcriptomics for hepatocyte and biliary markers, pericytes, and other mesenchymal cells would be recommended. Can there be a distinction made on what type of endothelial cell is affected? (sinusoidal lineage, vs. venous vs. lymphatic)

To assess whether the multicellular microenvironment of Chd4-ECko livers was altered, we performed immunostaining for various cellular markers from E12.5 to E14.5. These markers included LYVE-1 for liver sinusoids; PROX1 and E-cadherin (ECAD) for hepatocytes; CD41 for platelets and megakaryocytes; CD45 for leukocytes; CD68 and F4/80 for macrophages; MPO for neutrophils; TER119 for erythroid cells; and a-smooth muscle actin (SMA) for pericytes and smooth muscle cells (see Fig. 4D and__ Fig. R1*__). Across all the images we examined, no obvious cell-type-specific differences were observed between control and mutant livers.

Biliary epithelial cells, which begin to differentiate at approximately E15.54, were also assessed using cytokeratin 19 (CK19) immunostaining; however, no CK19-positive cells were detected in control livers at E14.5 (see Fig. R2*). Note that although LYVE-1 is also expressed by lymphatic endothelial cells, lymphatic vessels are not yet established in the liver at E14.52. Therefore, LYVE-1 staining is appropriate for identifying liver sinusoidal ECs at this stage of development. Our data indicate that the affected vasculature in Chd4-ECko livers is predominantly localized to the liver periphery (see Fig. 1D), which LYVE-1 staining shows to be mostly populated by sinusoidal vessels (Fig. R1B and R1F).

*Please see uploaded Response to Reviewers PDF for Figures R1 and R2

1. The experiments showing how endothelial Chd4 loss leads to a hyperinflammatory endothelial-and potentially hepatoblast-state are important. However, the relevance of immune cell infiltration in the hematopoietic-developing liver remains unclear. Which immune cells are presumably recruited to inflame the microenvironment then? Bone-marrow-derived? This aspect would benefit from experimental clarification, for example, using migration and/or direct co-culture versus indirect cell co-culture-ideally with or without ICAM-1 blockade-in vitro assays to determine if direct crosstalk with the CD45+ immune cell compartment explains the hyperinflammatory endothelia phenotype.

In mice, the first hematopoietic cells emerge in the yolk sac at E7.55. Subsequently, embryonic hematopoiesis takes place in the aorta-gonad-mesonephros (AGM) region and the placenta, before immature hematopoietic cells migrate to the fetal liver. After E11.0, the fetal liver becomes the main hematopoietic organ, supporting the expansion and differentiation of hematopoietic stem and progenitor cells into all mature blood cell lineages5-8. Around E16.5, hematopoietic cells migrate to the bone marrow9, so the bone marrow is not a relevant source of infiltrating immune cells in our E12.5-14.5 Chd4-ECko mutants. We therefore examined immune cell populations, including leukocytes, macrophages, and neutrophils, in Chd4-ECko livers. No enrichment of specific immune cell types was observed in Chd4-ECko livers compared with controls at E13.5-14.5 (Fig. R1). Since immune cells develop within fetal livers at this stage, these findings suggest that they are locally activated rather than recruited to Chd4-ECko livers. Moreover, because fetal livers contain a heterogeneous mixture of immature and mature hematopoietic and immune cells, appropriate in vitro cell models to assess immune cell activation in this context are currently lacking. We have added comments to the introduction to address some of these points (see lines 66-68).

1. Related to the previous comment: Can the authors validate their findings in an independent, ideally human, cell-based system?

To explore this, we analyzed PLAUR and ICAM1 transcripts following CHD4 and/or BRG1 knockdown in primary human umbilical vein endothelial cells (HUVECs) for 48 hours. No antagonistic regulation of either gene was detected in HUVECs (Fig. R3*). Moreover, while Icam1 transcription was antagonistically regulated by CHD4 and BRG1 in the mouse MS1 EC line (see Fig. 5A), transcriptional regulation of Plaur by these remodelers was observed only in isolated LSECs and not in cultured MS1 cells. Together, these findings demonstrate that BRG1 and CHD4 play context-specific roles when regulating Icam1 and Plaur transcription in different EC types. Furthermore, in vitro versus in vivo EC environments may additionally influence BRG1 and CHD4 activity.

*Please see uploaded Response to Reviewers PDF for Figure R3

1. Identifying the specific hematopoietic/immune subset could further increase the paper's impact, as it would more definitively clarify the mechanism in the developing endothelial niche.

Please see our response to question # 3.

1. Also, can the authors show experimentally whether, conversely, Chd4 overexpression can limit an endothelial-type of inflammatory liver injury?

We agree that exploring this suggestion would provide useful insights. However, we currently lack a genetic or inducible endothelial-specific Chd4 overexpression model, which makes it challenging to link our embryonic findings to the context of adult liver injury. For now, our study demonstrates that hepatic ECs regulate sterile inflammation to support embryonic liver development. Future development of appropriate genetic tools will allow us to determine if the role of endothelial CHD4 that is demonstrated in the current study is recapitulated in adult inflammatory liver injury models.

Minor

1. A separate figure panel for Chd4fl/fl; Vav-Cre+ appears reasonable, instead of being shown as a table.

Thank you. Please see our new Fig. S1, which includes representative images (and lethality rates) of control and Chd4fl/fl;Vav-Cre+ embryos at E18.5.

Significance:

Generally, this is an excellent manuscript with a strong developmental biology focus, and its translational relevance is not immediately apparent; however, establishing such a link could significantly increase its impact. For example, the significance of these findings in ischemia-reperfusion injury, SOS/VOD, and sepsis could offer therapeutic avenues to stabilize endothelial function.

The advance is the elegant discovery of a multifactorial endothelial-stabilizing mechanism in development, although its applicability to scenarios beyond developmental mutation remains unknown.

The strengths are the clear and transparent experimental interrogation. Rightfully, the authors acknowledge that there would be a benefit in finalizing inflammatory blockade, genetic or antibody-mediated, to pin down the mechanistic circuit.

The reviewer's expertise is: childhood liver diseases, developmental liver organoid generation, stem cells (iPSCs), cell reprogramming

Reviewer #3

Evidence, reproducibility and clarity:

1. Wu et al. report antagonistic roles for chromatin remodelers Chd4 and Brg1, in endothelial cells, during liver development. There is a major flaw in the study which makes it difficult to interpret the conclusions. The genotypes of the mouse models used are flawed. The comparison should be made between two single knockouts (Chd4 single, Brg1 single), double mutants (Chd4/Brg1) and proper controls. For both "single KO", one allele of the other gene is also deleted - Chd4 -Ecko has one allele of Brg1 deleted and vice versa. Also, the proper control should be Chd4 fl/flBrg1fl/fl without the Cre. Since 3 alleles (not just two that belong to the same gene) are deleted in a single knockout, it is impossible to assign the effect to one gene.

We acknowledge the fact that the single Brg1 and Chd4 EC knockouts in this study each carry a heterozygous deletion allele for the other remodeler (exact genotypes are shown in Fig. 1A). The mating strategy that yielded these mutants was chosen for three reasons. First, we have found that genetic background influences the embryonic phenotypes of these chromatin remodeler mutants3. Moreover, embryonic development at the stages analyzed in this study occurs quickly and requires precise timing for comparative analysis between genotypes. Therefore, it is most rigorous to study littermates when comparing single- and double-mutant embryos for BRG1 and CHD4. To achieve this, we used Brg1fl/fl;Chd4fl/fl females rather than Brg1fl/+;Chd4fl/+ females for timed matings. Although the former females cannot produce single knockout embryos without a compound heterozygous allele of the other remodeler, these females allowed us to generate single- and double-knockouts at a rate of 1/8 embryos. If we had used Brg1fl/+;Chd4fl/+ females for timed matings, we would have been able to generate “clean” single mutants with wildtype alleles of the other remodeler, but the single- and double-knockout generation rate would have been 1/32 embryos. This would have been an impractical mutant generation rate for this study. Second, our prior research demonstrates that heterozygous deletion of Chd4 or Brg1 does not produce the liver phenotypes seen with the respective homozygous deletions2,3. Third, the complete lethality of Chd4-ECko (Brg1fl/+;Chd4fl/fl;VE-cadherin-Cre+) mutants in this study demonstrates that deleting one allele of Brg1 cannot rescue Chd4-related lethality.

As for controls in this study, we saw no evidence of phenotypes or of any gene deletion in our Cre- embryos (either in this study or in previous ones analyzing similar phenotypes2,3). Therefore, we used Cre- embryos for controls because they were generated at a 1/2 rate by our timed matings, which boosted our output for analyses.

Specific points

1. Fig 2c Plaur transcript - no statistical comparison between 2nd and 4th column, Chd4 Ecko vs double mutant. If there is not statistical difference, does not explain the rescue in double mutants

Thank you for the suggestion. We have included a comparison between Chd4-ECko and Brg1/Chd4-ECdko in our revised Fig 2C. The Kruskal-Wallis test showed a significant difference between the Chd4-Ecko and Brg1/Chd4-ECdkogroups (p=0.016). This indicates that Plaur induction in Chd4-Ecko LSECs is rescued in Brg1/Chd4-ECdko LSECs.

1. Fig 2e. Comparison should be made between Plg-/- Chd4 fl/fl and Plg-/- Chd4 fl/fl Cre, not other genotypes

This experiment aims to determine whether different levels of plasminogen (Plg) reduction can rescue the lethality caused by Chd4 deletion. To do this, we set up the mating strategy shown in Fig. 2E to produce appropriate littermate controls and to compare lethality among Plg+/+;Chd4-ECko, Plg+/-;Chd4-ECko, and Plg-/-;Chd4-ECko embryos. This comparison would not have been possible with embryos generated only from mice on a Plg-/- background.

1. Fig. 4. How does Chd4 or Brg1 activity in endothelial cells lead to Icam1 activation in epithelial cells?

Since cytokines like IFNg, TNFa, and IL1b can induce ICAM-1 expression in hepatocytes10, we speculate that ICAM-1 expression in hepatoblasts (ECAD+ cells in Fig. 4D) was induced by the elevated TNFa and IL1b produced in Chd4-ECko livers (Fig. 3G).

1. Mice used in Figure 5 are Cdf4 fl/+ and Cdf4 fl/fl, no Brg1 deletion. The authors improperly compare these to Chd4-Ecko which have one allele of Brg1 deleted. The rescue needs to be done in the same genotype Chd4-Ecko.

Please note that data from Fig. 5 were generated from cultured ECs (MS1 cells).

Significance

Wu et al. report antagonistic roles for chromatin remodelers Chd4 and Brg1, in endothelial cells, during liver development. There is a major flaw in the study which makes it difficult to interpret the conclusions. Genotypes that were chosen for the study make the data not interpretable

Please see our response to your Question #1

In summary, we have included the following changes to this revised manuscript:

• New Figure 1B: Representative images and lethality rates for control, Chd4-ECko, Brg1-ECko, and Brg1/Chd4-ECdko embryos at E17.5.
• New Figure 2C: qRT-PCR analysis of Chd4, Brg1, and Plaur gene transcripts in E12.5 control and mutant LSECs.
• Regraphing of Figure 3G: qRT-PCR analysis of Tnf, Il6, and Il1b gene transcripts in E14.5 control and mutant livers.
• New Figure S1: Representative images and lethality rates for control, Chd4fl/+;Vav-Cre+, and Chd4fl/fl;Vav-Cre+embryos at E18.5. References for this revision:

Alva JA, Zovein AC, Monvoisin A, Murphy T, Salazar A, Harvey NL, Carmeliet P, Iruela-Arispe ML. VE-Cadherin-Cre-recombinase Transgenic Mouse: A Tool for Lineage Analysis and Gene Deletion in Endothelial Cells. Dev Dyn. 2006;235:759-767. doi: 10.1002/dvdy.20643 Crosswhite PL, Podsiadlowska JJ, Curtis CD, Gao S, Xia L, Srinivasan RS, Griffin CT. CHD4-regulated plasmin activation impacts lymphovenous hemostasis and hepatic vascular integrity. J Clin Invest. 2016;126:2254-2266. doi: 10.1172/JCI84652 Wu ML, Wheeler K, Silasi R, Lupu F, Griffin CT. Endothelial Chromatin-Remodeling Enzymes Regulate the Production of Critical ECM Components During Murine Lung Development. Arterioscler Thromb Vasc Biol. 2024;44:1784-1798. doi: 10.1161/ATVBAHA.124.320881 Shiojiri N, Inujima S, Ishikawa K, Terada K, Mori M. Cell lineage analysis during liver development using the spfash-heterozygous mouse. Lab Invest. 2001;81:17-25. doi: 10.1038/labinvest.3780208 Soares-da-Silva F, Peixoto M, Cumano A, Pinto-do OP. Crosstalk Between the Hepatic and Hematopoietic Systems During Embryonic Development. Front Cell Dev Biol. 2020;8:612. doi: 10.3389/fcell.2020.00612 Ema H, Nakauchi H. Expansion of hematopoietic stem cells in the developing liver of a mouse embryo. Blood. 2000;95:2284-2288. Kieusseian A, Brunet de la Grange P, Burlen-Defranoux O, Godin I, Cumano A. Immature hematopoietic stem cells undergo maturation in the fetal liver. Development. 2012;139:3521-3530. doi: 10.1242/dev.079210 Freitas-Lopes MA, Mafra K, David BA, Carvalho-Gontijo R, Menezes GB. Differential Location and Distribution of Hepatic Immune Cells. Cells. 2017;6. doi: 10.3390/cells6040048 Christensen JL, Wright DE, Wagers AJ, Weissman IL. Circulation and chemotaxis of fetal hematopoietic stem cells. PLoS Biol. 2004;2:E75. doi: 10.1371/journal.pbio.0020075 Satoh S, Nussler AK, Liu ZZ, Thomson AW. Proinflammatory cytokines and endotoxin stimulate ICAM-1 gene expression and secretion by normal human hepatocytes. Immunology. 1994;82:571-576.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Wu et al. report antagonistic roles for chromatin remodelers Chd4 and Brg1, in endothelial cells, during liver development. There is a major flaw in the study which makes it difficult to interpret the conclusions. The genotypes of the mouse models used are flawed. The comparison should be made between two single knockouts (Chd4 single, Brg1 single), double mutants (Chd4/Brg1) and proper controls. For both "single KO", one allele of the other gene is also deleted - Chd4 -Ecko has one allele of Brg1 deleted and vice versa. Also, the proper control should be Chd4 fl/flBrg1fl/fl without the Cre. Since 3 alleles (not just two that belong to the same gene) are deleted in single knockout it is impossible to assign the effect on one gene.

Specific points

1. Fig 2c Plaur transcript - no statistical comparison between 2nd and 4th column, Chd4 Ecko vs double mutant. If there is not statistical difference, does not explain the rescue in double mutants
2. Fig 2e. Comparison should be made between Plg-/- Chd4 fl/fl and Plg-/- Chd4 fl/fl Cre, not other genotypes
3. Fig. 4. How does Chd4 or Brg1 activity in endothelial cells lead to Icam1 activation in epithelial cells?
4. Mice used in Figure 5 are Cdf4 fl/+ and Cdf4 fl/fl, , no Brg1 deletion. The authors improperly compare these to Chd4-Ecko which have one allele of Brg1 deleted. The rescue need s to be done in the same genotype Chd4-Ecko.

Significance

Wu et al. report antagonistic roles for chromatin remodelers Chd4 and Brg1, in endothelial cells, during liver development. There is a major flaw in the study which makes it difficult to interpret the conclusions. Genotypes that were chosen for the study make the data not interpretable

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

In essence, Wu et Al find that Chd4 mutant mice exhibit embryonic liver degeneration due to uPA-mediated plasmin hyperactivity and an ICAM-1-driven hyperinflammation and that additional mutation of BRG1 opposes this liver degeneration, possibly via ICAM-1.

Generally, this is an excellent manuscript with a very logical sequence of experiments, although it has shortcomings such as validating their findings in an independent system, ideally human, and further establishing the translational relevance. Establishing translational relevance through mechanistic experiments that identify specific inflammatory tissue pathways, such as by blocking ICAM-1 and TNF-alpha, could also define developmental aberrations as a model for broader (patho)physiology and thereby enhance the impact on the field.

Major

• The embryonic and postnatal survival data of Chd4-ECko and Brg1/Chd4-Ecdko mice should be included in Fig. 1
• What is the impact of Chd4-ECko and Brg1/Chd4-ECdko on the multicellular microenvironment? At a minimum, IF or spatial transcriptomics for hepatocyte and biliary markers, pericytes, and other mesenchymal cells would be recommended. Can there be a distinction made on what type of endothelial cell is affected? (sinusoidal lineage, vs. venous vs. lymphatic)
• The experiments showing how endothelial Chd4 loss leads to a hyperinflammatory endothelial-and potentially hepatoblast-state are important. However, the relevance of immune cell infiltration in the hematopoietic-developing liver remains unclear. Which immune cells are presumably recruited to inflame the microenvironment then? Bone-marrow-derived? This aspect would benefit from experimental clarification, for example, using migration and/or direct co-culture versus indirect cell co-culture-ideally with or without ICAM-1 blockade-in vitro assays to determine if direct crosstalk with the CD45+ immune cell compartment explains the hyperinflammatory endothelia phenotype.
• Related to the previous comment: Can the authors validate their findings in an independent, ideally human, cell-based system?
• Identifying the specific hematopoietic/immune subset could further increase the paper's impact, as it would more definitively clarify the mechanism in the developing endothelial niche.
• Also, can the authors show experimentally whether, conversely, Chd4 overexpression can limit an endothelial-type of inflammatory liver injury?

Minor

• A separate figure panel for Chd4fl/fl; Vav-Cre+ appears reasonable, instead of being shown as a table.

Significance

Generally, this is an excellent manuscript with a strong developmental biology focus, and its translational relevance is not immediately apparent; however, establishing such a link could significantly increase its impact. For example, the significance of these findings in ischemia-reperfusion injury, SOS/VOD, and sepsis could offer therapeutic avenues to stabilize endothelial function.

The advance is the elegant discovery of a multifactorial endothelial-stabilizing mechanism in development, although its applicability to scenarios beyond developmental mutation remains unknown.

The strengths are the clear and transparent experimental interrogation. Rightfully, the authors acknowledge that there would be a benefit in finalizing inflammatory blockade, genetic or antibody-mediated, to pin down the mechanistic circuit.

The reviewer's expertise is: childhood liver diseases, developmental liver organoid generation, stem cells (iPSCs), cell reprogramming

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

The manuscript by Wu and Griffin describes a mechanism where CHD4 and BRG1, two chromatin remodelling enzymes, have antagonistic functions to regulate extracellular matrix (ECM) plasmin activity and sterile inflammatory phenotype in the endothelial cells of the developing liver. As a follow up from a previous study, the authors investigate the phenotype of embryonic-lethal endothelial-specific CHD4-knockout, leading to liver phenotype and embryo death, and the rescue of this phenotype when subsequently BRG1 is knocked-out also in the endothelium. First, the authors show that the increase in plasmin activator uPAR (which leads to ECM degradation) in CHD4-KO embryos can be rescued by BRG1-KO, and that both CHD4 and BRG1 interact with the uPAR promoter. However, the authors demonstrate that reducing plasminogen by genetic knockout is unable to rescue the CHD4-KO embryos alone, suggesting an additional mechanism. By RNAseq analysis, the authors identify sterile inflammation as another potential contributor to the lethal phenotype of CHD4-KO embryos through increased expression of ICAM-1 in endothelial cells, also showing binding of both chromatin remodellers to ICAM-1 promoter. Finally, the authors use nonsteroidal anti-inflammatory drug carprofen, alone or in combination with plasminogen genetic knockout, and demonstrate CHD4-KO lethal embryonic phenotype rescue with the combination of plasminogen reduction and inflammation reduction, highlighting the synergistic role of both ECM degradation and sterile inflammation in this genetic KO.

The findings of the manuscript are interesting, experiments well controlled and paper well written. While the work is of potential specialist interest to the field of liver development, there are several issues which authors should address before this paper can be published:

Major issues:

• The authors still see embryonic lethality of some embryos with endothelial BRG1-KO or combined endothelial CHD4/BRG1-KO - could the authors please show or at least comment in the discussion why those animals are dying?
• In the qRT-PCR results Fig.2c, what is each dot?
• In the same figure, I would expect that in CHD4-KO there is no CHD4 transcript, and in BRG1-KO there is no BRG1 transcript, rather than the reduction shown, which seems quite noisy (though significant) - is it this a result of normalisation? Or is indeed only a certain amount of the transcript reduced?
• In the same figure, is the statistical testing performed before or after normalisation? This can introduce errors if done after normalisation.
• In some cases, the authors show immunofluorescence images but do not specify how many biological replicates this represents (e.g. Fig.1d, 4c-d). This should be added.
• I also encourage the authors to present a supplementary figure with at least one other biological replicate shown for imaging data (optional).
• The plasminogen reduction by genetic modulation results in drastic changes to the embryos' appearance - is this a whole embryo KO or endothelial-specific KO? Can authors at least comment on the differences?
• In Fig.2b, do I understand correctly only 1 sample was analysed with different areas plotted on the graph? If so, this experiment should be repeated on another set of embryos to be robust, and data plotted as a mean of each embryo (rather than areas).
• Also in some graphs, authors specify that it was more than n>x embryos, but then - what are the dots on the graph representing? Each embryo? This should be specified (e.g. Fig.2b-c, but please check this in all the figure legends).
• "we found Plaur was the only gene that was induced in CHD4-ECko LSECs at E12.5 (Figure S3D)." - I am not sure this is correct, as gene Plau is also increased in 2/3 samples?
• I find the title and the running title somewhat misleading and too broad; the authors should specify more detail in the title about the content of the paper - the current statement of the title is somewhat true but shown only for one genetic model and not confirmed for all types of "lethal embryonic liver degeneration".

Minor issues:

• If an animal licence was used, its number should be specified in the ethics or methods section
• In fig.3g it is very hard to see each of the samples, could authors try to improve this graph for clarity using colours-or split Y axis - or both?
• "This indicates that ECs can play a pro-inflammatory role in embryonic livers and highlights the need for tight regulation to ensure normal liver growth." This sentence for me is misleading, EC are producing inflammatory signals only during the CHD4-KO according to the author's data, and authors do not show such data in normal homeostasis condition. Actually, the pro-inflammatory role here seems detrimental, and ECs should not exhibit it for correct development. The authors should rephrase this to be clearer.

Significance

General assessment: The study is well controlled and well written. The findings are interesting. The limitation of the findings is only 1 combination genetic model being studied, and it is unclear if the synergistic effect of sterile inflammation and ECM degradation is broadly applicable to other models, where embryo dies because of liver failure.

Advance: The study makes an incremental advance, following up findings from a previous study. However, it is conceptually interesting.

Audience: The audience for this manuscript would be a liver development specialist. However, broader concepts could also be applicable to liver disease.

Expertise: I research in the field of liver regeneration and disease.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

SECTION A - Evidence, Reproducibility, and Clarity Summary The study investigates the neurodevelopmental impact of trisomy 21 on human cortical excitatory neurons derived from induced pluripotent stem cells (hiPSCs). Key findings include a modest reduction in spontaneous firing, a marked deficit in synchronized bursting, decreased neuronal connectivity, and altered ion channel expression-particularly a downregulation of voltage‐gated potassium channels and HCN1. These conclusions are supported by a combination of in vitro calcium imaging, electrophysiological recordings, viral monosynaptic tracing, RNA sequencing, and in vivo transplantation with two‐photon imaging.

Major Comments • Convincing Nature of Key Conclusions: The study's conclusions are generally well supported by a diverse set of experimental approaches. However, certain claims regarding the intrinsic properties of the excitatory network would benefit from further qualification. In particular, the assertion that reduced synchronization is solely attributable to altered ion channel expression might be considered somewhat preliminary without additional corroborative experiments.

1.1) We agree with the reviewer and now write in the abstract: 'Together, these findings demonstrate long-lasting impairments in human cortical excitatory neuron network function associated with Trisomy 21 .' And in the Introduction: 'Collectively, the observed changes in ion channel expression, neuronal connectivity, and network activity synchronization may contribute to functional differences relevant to the cognitive and intellectual features associated with Down syndrome.'

• One major limitation of the current experimental design is the reliance on predominantly excitatory neuronal cultures derived from hiPSCs. Although the authors convincingly demonstrate differences in network synchronization and connectivity between trisomic (TS21) and control neurons, the almost exclusive focus on excitatory cells limits the physiological relevance of the in vitro network. In the developing cortex, interneurons and astrocytes play crucial roles in modulating network excitability, synaptogenesis, and plasticity. Therefore, incorporating these cell types-either through co-culture systems or through directed differentiation protocols that yield a more heterogeneous neuronal population-could help to determine whether the observed deficits are intrinsic to excitatory neurons or are compounded by a lack of proper inhibitory regulation and glial support. 1.2) Thank you for this thoughtful comment. We agree that interneurons and astrocytes are crucial for network function. To clarify, astrocytes are generated in this culture system, as we previously reported in our characterisation of the timecourse of network development using this approach (Kirwan et al., Development 2025). However, our primary goal was to first isolate and define the cell-autonomous defects intrinsic to TS21 excitatory neurons, minimizing the complexity introduced by additional neuronal types. This focused approach was chosen also because engineering a stable co-culture system with reproducible excitatory/inhibitory (E/I) proportions is a significant undertaking that extends beyond the scope of this initial investigation, and has proven challenging to date for the field. By establishing this foundational phenotype, our work complements prior studies on interneuron and glial contributions. Future studies building on this work will be essential to dissect the more complex, non-cell-autonomous effects within a heterogeneous network. Importantly, since our initial submission, two highly relevant preprints have emerged-including a notable study from the Geschwind laboratory at UCLA (Vuong et al., bioRxiv, 2025; Risgaard et al., bioRxiv, 2025), as well as our own complementary study Lattke et al, under revision, that highlight widespread transcriptional changes in excitatory cells of the human fetal DS cortex, providing strong validation for our central findings. This convergence of results from multiple groups underscores the timeliness and importance of our work.

• Furthermore, the assessment of neuronal connectivity via pseudotyped rabies virus tracing, while innovative, has inherent limitations. The quantification of connectivity as a ratio of red-to-green fluorescence pixels may be influenced by differential viral infection efficiencies, variations in the expression levels of the TVA receptor, or even by the lower basal activity levels observed in TS21 cultures. Complementary approaches-such as electron microscopy for synaptic density analysis or functional connectivity measurements using multi-electrode arrays (MEAs)-could provide additional structural and functional insights that would validate the rabies tracing data. 1.3) Thank you for this constructive feedback. While we cannot formally exclude that TS21 cells might express the TVA receptor at lower levels due to generalized gene dysregulation, we infected all WT and TS21 cultures in parallel using identical virus preparations and titers to minimize technical variability. Crucially, we also addressed the potential confound of differential basal activity by performing the rabies tracing under TTX incubation (see Suppl. Fig. 7), which blocks network activity and ensures that viral spread reflects structural connectivity alone.

While complementary methods like EM or MEA could provide additional insight, they fall outside the scope of the current study. We are confident that our rigorous controls validate our use of the rabies tracing method to assess structural connectivity.

• Qualification of Claims: Some conclusions, particularly those linking specific ion channel dysregulation (e.g., HCN1 loss) directly to network deficits, might be better presented as preliminary. The authors could temper their language to indicate that while the evidence is suggestive, the mechanistic link remains to be fully established. 1.4) We have revised the text to more clearly indicate that the link between HCN1 dysregulation and network deficits is correlative and remains to be fully established. While our ex vivo recordings suggest altered Ih-like currents consistent with reduced HCN1 expression, we now present these findings as preliminary and hypothesis-generating, pending further functional validation. We write in the discussion: However, further targeted functional validation will be needed to confirm a causal link.

• Need for Additional Experiments: Additional experiments that could further consolidate the current findings include: o Inclusion of Inhibitory Neurons or Co-culture Systems: Incorporating interneurons or astrocytes would help determine whether the observed deficits are solely intrinsic to excitatory neurons. See 1.2 o Alternative Connectivity Assessments: Complementing the rabies virus tracing with electron microscopy or multi-electrode array (MEA) recordings would add structural and functional validation of the connectivity differences. See 1.3 o Extended Temporal Profiling: Monitoring network activity over a longer developmental window would clarify whether the observed deficits represent a delay or a permanent alteration in network maturation. 1.5) In vivo we were able to track the cells for up to five months post-transplantation supporting the interpretation of a permanent alteration.

• Reproducibility and Statistical Rigor: The methods and data presentation are largely clear, with adequate replication and appropriate statistical analyses. Nonetheless, a more detailed description of the experimental replicates, particularly regarding the viral tracing and in vivo transplantation studies, would enhance reproducibility. The availability of raw data and scripts for calcium imaging analysis would also further support independent verification. We thank the reviewer for these suggestions and we now provide a more detailed description of replicates. We also add the raw data.

Minor Comments • Experimental Details: Minor revisions could include clarifying the infection efficiency and expression levels of the viral constructs used in connectivity assays to rule out technical variability.

See 1.3

• Literature Context: The authors reference prior studies appropriately; however, integrating a brief discussion comparing their findings with alternative DS models (e.g., organoids or other hiPSC-derived systems) would improve contextual clarity. We thank the reviewer for this helpful suggestion. We have now added a brief discussion comparing our findings with those reported in alternative Down syndrome models, including brain organoids and other hiPSC-derived systems. This addition helps to contextualize our results within the broader field and highlights the unique strengths and limitations of our in vitro and in vivo xenograft approach. We write: 'Our findings align with and extend previous studies using alternative Down syndrome models, such as brain organoids and other hiPSC-derived systems. Organoid models have provided valuable insights into early neurodevelopmental phenotypes in DS, including altered interneuron proportions (Xu et al Cell Stem Cell 2019) but also suggest that variability across isogenic lines can overshadow subtle trisomy 21 neurodevelopmental phenotypes (Czerminski et al Front in Neurosci 2023). However, these systems often lack the structural complexity, vascularization, and long-term maturation achievable in vivo. By using a xenotransplantation model, we were able to assess the maturation and functional properties of human neurons within a physiologically relevant environment over extended time frames, offering complementary insights into DS-associated circuit dysfunction (Huo et al Stem Cell Reports 2018; Real et al., 2018).

• Presentation and Clarity: Figures are generally clear,.But the manuscript contains a minor labeling error. On page 13, the figure is erroneously labeled as "Fig6A", whereas, based on the context and corresponding data, it should be "Fig5A". I recommend that the authors correct this mistake to ensure consistency and avoid potential confusion for readers. Thank you for pointing this out. This has been corrected in the revised manuscript.

Reviewer #1 (Significance (Required)):

SECTION B - Significance • Nature and Significance of the Advance: The work offers a substantial conceptual advance by providing a mechanistic link between trisomy 21 and impaired neuronal network synchronization. Technically, the study integrates state-of-the-art imaging, electrophysiology, and transcriptomic profiling, thereby offering a multifaceted view of DS-related neural dysfunction. Clinically, the findings have the potential to inform future therapeutic strategies targeting network connectivity and ion channel function in Down syndrome.

We thank the reviewer for this very supportive comment.

• Context in the Existing Literature: The study builds on previous observations of altered network activity in DS patients and DS mouse models (e.g., altered EEG synchronization and reduced synaptic connectivity). It extends these findings to human-derived neuronal models, thus bridging a gap between clinical observations and molecular/cellular mechanisms. Relevant literature includes studies on DS neurodevelopment and the role of ion channels in synaptic maturation. • Target Audience: The reported findings will be of interest to researchers in neurodevelopmental disorders, Down syndrome, and ion channel physiology. Additionally, the study may attract the attention of those working on hiPSC-derived models of neurological diseases, as well as clinicians interested in the pathophysiology of DS. • Keywords and Field Contextualization: Keywords: Down syndrome, trisomy 21, neuronal connectivity, synchronized network activity, hiPSC-derived cortical neurons, ion channel dysregulation.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

Summary The manuscript by Peter et al., reports on the neuronal activity and connectivity of iPSC-derived human cortical neurons from Down syndrome (DS) that is caused by caused by trisomy of the human chromosome 21 (TS21). Major points: Although the manuscript is potentially interesting, the results appear somehow preliminary and need to be corroborated by control experiments and quantifications of effects to fully sustain the conclusions. (1) The authors have not assessed the percentage of WT and TS21 cells that acquire a neuronal or glia identity in their cultures. Indeed, the origin of alterations in network activity and connectivity observed in TS21 neurons could simply derive from reduced number of neurons arising from TS21 iPSC. Alternatively, the same alteration in network activity and connectivity could derive from a multitude of other factors including deficits in neuronal development, neurite extension, or intrinsic electrophysiological properties. In the current version of the manuscript, none of these has been investigated. 2.1) We thank the reviewer for this thoughtful comment. In response, we included an in vivo characterization of cell-type proportions at the same time points where we observed network activity defects using in vivo calcium imaging (see Supplementary Fig. 6).

Previous work has identified several cellular and molecular phenotypes in human cells, postmortem tissue, and mouse models-including those mentioned by the reviewer. In this study, our focus was on investigating neural network activity, intrinsic electrophysiological properties both in vitro and in vivo, and preliminary bulk RNA sequencing. We have also independently measured cell proportions in the human fetal cortex and conducted a more extensive transcriptomic analysis of Ts21 versus control cells in a separate study (Lattke et al., under revision). We observed a reduction of RORB/FOXP1-expressing Layer 4 neurons in the human fetal cortex at midgestation, as well as increased GFAP+ cells, reduced progenitors and a non significant reduction of Cux2+ cells in late stage DS human cell transplants, along with a gene network dysregulation specifically affecting excitatory neurons (Lattke et al., under revision). Here, we provide complementary findings, demonstrating reduced excitatory neuron network connectivity in vitro and decreased neural network synchronised activity in both in vitro and in vivo models (see also 2.8). We agree with the reviewer that this could be for a number of reasons, both cell autonomous (channel expression and/or function) or non-autonomous (connectivity and/or network composition - as reflected in differences in proportions of SATB2+ neurons generated in TS21 cortical differentiations).

(2) Electrophysiological properties of TS21 and WT neurons at day 53/54 in vitro indicate an extremely immature stage of development (i.e. RMP between -36 and -27 mV with most of the cells firing a single action potential after current injection) in the utilized culture conditions: This is far from ideal for in vitro neuronal-network studies. Finally, reduced activity of HCN1 channels should be confirmed by specific recordings isolating or blocking the related current.

2.2) Thank you for this thoughtful comment. We have also conducted ex vivo electrophysiological recordings and found that the neurons exhibit relatively immature properties, consistent with the known slow developmental trajectory of human neuron cultures. In light of this and the absence of direct confirmatory evidence, we now refer to the observed reduction in HCN1 as preliminary.

Main points highlighting the preliminary character of the study. 1) In Figure 1 immunofluorescence images of the neuronal differentiation markers (Tbr1, Ctip2 and Tuj1) are showed. However, no quantification of the percentage of cells expressing these markers for WT and TS21 neurons is reported. On the other hand, simple inspection of the representative images clearly seams to indicate a difference between the two genotypes, with TS21 cultures showing lower number of cells expressing neuronal markers. This quantification should be corroborated by a similar staining for an astrocyte marker (GFAP, but not S100b since is triplicated in DS). This is an extremely important point since it is obvious that any change in the percentage of neurons (or the neuron/astrocyte ratio) in the cultures will strongly affect the resulting network activity (shown in Figure 2) and the connectivity (showed in Figure 4). Possibly, the quantification should be done at the same time points of the calcium imaging experiments.

2.3) See 2.1. We included an in vivo characterization of cell-type proportions at the same time points where we observed network activity defects using in vivo calcium imaging. (see Supplementary Fig. 6).

2) In Figure 2 the authors show some calcium imaging traces of WT and TS21 cultures at different time points. However, they again do not show any quantification of neuronal activity. A power spectra analysis is shown in Supplementary Figure 2, but only for WT cultures, while in Supplementary Figure 3 a comparison between WT and Ts21 power spectra is done, but only at the 50 day time point, while difference in synchrony are assessed at 60 days. At minimum, the author should include in main Figure 2 the quantification of the mean calcium event rate and mean event amplitude at the different time points and the power spectra analysis for both WT and TS21 cultures at the same timepoints.

2.4) We thank the reviewer for this comment. We now add the power spectra analysis in the main Figure 2 and quantification of the mean calcium burst rate and mean event amplitude in SuppFig. 4.

Of note, the synchronized neuronal activity is present in WT cultures at day 60, but totally lost at subsequent time-points (70 and 80 days). The results of this later time points are different from previous data from the same lab (Kirwan et al., 2015). How might these data be explained? It would be important to rule out any potential issues with the health of the culture that could explain the loss of neuronal activity.It would be beneficial to check cell viability at the different time points to exclude possible confounding factors ? A propidium staining or a MTT assay would strongly improve the soundness of the calcium data.

2.5) We thank the reviewer for this important observation. The difference from the findings reported in Kirwan et al., 2015 is due to the use of a different neuronal differentiation medium in the current study (BrainPhys versus N2B27). BrainPhys medium supports robust early network activity compared to N2B27 (onset before day 60 in BrainPhys, post-day 60 in N2B27), resulting in an earlier decline in synchrony at later stages (day 70-80 in BrainPhys, compared with day 90-100 in N2B27). Importantly, in our in vivo xenograft model, burst activity is sustained up to at least 5 months post-transplantation (mpt), indicating that the neurons retain the capacity for network activity over extended periods in a more physiological environment. We adapted the text accordingly.

3) In Figure 3 there is no quantification of the number and/or density of transplanted neurons for WT and TS21, but only representative images. As above, inspection of the representative images seems to show a decrease in cells labeled by the Tbr1 neuronal marker for TS21 cells. Moreover, the in vivo calcium imaging of transplanted WT and TS21 cells lacks most of the quantification normally done in calcium imaging experiments. Are the event rate and event amplitude different between WT and TS21 neurons ? The measure of neuronal synchrony by mean pixel correlation is not well explained, but it looks somehow simplistic. Neuronal synchrony can be more precisely measured by cross-correlation analysis or spike time tiling coefficients on the traces from single-neuron ROI rather than on all pixels in the field of view, as apparently was done here.

2.6) We thank the reviewer for these valuable points. We now include quantification of the number and density of transplanted neurons for both WT and Ts21 grafts in Extended Data Figure 5 (see 2.1).

Regarding the in vivo calcium imaging, we appreciate the reviewer's suggestion to include additional standard metrics. We have quantified the event rate in Real et al 2018. These analyses reveal that Ts21 neurons show a reduction in event rate.

We agree that our initial description of the synchrony analysis using mean pixel correlation was not sufficiently detailed. We have now clarified this in the Methods and Results, and we acknowledge its limitations. Importantly, we note that the reduced synchronisation is a highly consistent phenotype, observed across at least six independent donor pairs, different differentiation protocols, and both in vitro (and in two independent labs) and in vivo settings. As suggested, future studies using ROI-based approaches-such as cross-correlation or spike-time tiling coefficients-would provide a more refined characterization of synchrony at the single-neuron level (Sintes et al, in preparation). We now include this point in the discussion.

4) The results on reduced neuronal connectivity in Figure 3 look very striking. However, these results should be accompanied by control experiments to verify the number of neuronal cells and neurite extension in WT and Ts21 cultures. These two parameters could indeed strongly influence the results. As the cultures appear to grow in clusters, bright-field images and TuJ1 staining of the cultures will also greatly help to understand the degree of morphological interconnection between the clusters.

We now add Tuj1 staining in Supplementary figure 10.

5) The authors performed RNA-seq experiments on day 50 cultures. Why the authors do not show the complete differential gene expression analysis, but only a small subset of genes? A comprehensive volcano plot and the complete list of identified genes with logFC and FDR values would be helpful. If possible, comparison of the present data (particularly on KCN and HCN expression changes) with published and publicly available expression datasets of other human or human Down syndrome iPSC-derived neurons or human Down syndrome brains will greatly increase the soundness of the present findings. In addition, the gene ontology (GO) results are mentioned in the text, but are not presented. Showing the complete GO analysis for both up and downregulated genes will help the reader to better understand the RNA-seq results. Notably, the results shown in Supplementary Figure on GRIN2A and GRIN2B expression (with values of 300-700 counts versus 2000-4000 counts, respectively) clearly indicate that in both WT and TS21 cultures the NMDA developmental switch has not occurred yet at the 50 days timepoint.

We now show volcano plots in Supplementary Fig. 11.

6) The measure of hyperpolarization-activated currents shown in Figure 5 lack proper control experiments. First, the hyperpolarizing current in TS21 cells do not reach a steady-state as the controls. The two curves are therefore hard to compare. To exclude possible difference in kinetic activation, the authors should have prolonged the current injection period (1-2 seconds). Second, to ultimately prove that such currents are mediated by HCN channels in WT cells the authors should perform some control experiments with a specific HCN blocker. A good example of a suitable protocol, with also current blockers to exclude all other possible current contributions, is the one reported in Matt et al Cell. Mol. Life Sci. 68, 125-137 (2011).

2.7) We thank the reviewer for this detailed and helpful comment. We agree that to definitively identify the recorded currents as Ih, it would be necessary to isolate them pharmacologically using specific HCN channel blockers and appropriate controls, such as those described in Matt et al., Cell. Mol. Life Sci. Unfortunately, due to current constraints, we no longer have access to the animals used in this study and cannot allocate the necessary time or resources, we are unable to perform the additional experiments at this stage.

However, our goal here was to use electrophysiological recordings as an indication of altered HCN channel activity, which we then support with molecular evidence. We now emphasize this point more clearly in the revised manuscript.

7) The manuscript lacks information on the statistical analysis used. Also, the numerosity of samples is not clear. Were the dots shown in some graph technical replicates from a single neuronal induction or were all independent neuronal inductions or a mix of the two ? Please clarify.

We now clarify the numbers in the Figure legend.

8) The method section lacks important information to guarantee reproducibility. Just a few examples: • Only electrophysiology methods for slice are reported, but not for in vitro culture.

We now clarify these details in the methods.

• Details on Laminin coating is lacking. What concentration was used ? Was poly-ornithine or poly-lysine used before Laminin coating ? We now clarify these details in the methods.

• How long cells were switched to BrainPhys medium before calcium imaging ? We now clarify these details in the methods.

Minor point/typos etc.

Introduction • Page 4 line 6: in the line "Trisomy 21 in humans commonly results in a range in developmental and morphological changes in the forebrain ..." "in" could be replaced by "of". We have fixed this. • Page 5 line 2: please remove "an" before the word "another". We have fixed this. • Page 5 line 2: please replace "ecitatory" with "excitatory". We have fixed this typo.

Results • Page 10 line 25: The concept of "pixel-wise" appears for the first time in this section and could be better introduced to facilitate the understanding of the experiment. • In the "results" section, page 11 line 1 and 4, references are made to "Figure 4D" and "4F," but these figures do not appear to be present in the figure section. Upon reviewing the rest of the section, the data seem to refer to "Figure 3D" and "3E." We have fixed this. Discussion • Page 15 line 20: please replace "synchronised" with "synchronized". We have fixed this typo. • Page 16 line 11: please replace "T21" with "TS21". We have fixed this typo. Methods • Page 19 line 12: "Pens/Strep" has to be replaced by Pen/Strep. We have fixed this typo. • Page 20 line 20: "Tocris Biocience" has to be replaced by "Tocris Bioscience". We have fixed this typo. • Page 21 line 2: "Addegene" has to be replaced by "Addgene". We have fixed this typo. Figures • Figure 3: the schematic experimental design (Fig. 3A) could be enlarged to match the width of the images/graphs below. We have fixed this. • Figure 5: the reviewer suggests resizing/repositioning the graphs in Fig. 1A so that they match the width of those below. We have fixed this. • Figure S1D: In all the figures of the paper, the respective controls for the TS21 1 and TS21 2 lines are labelled as "WT1/WT2," while in these graphs, they are called "Ctrl1" and "Ctrl2." To ensure consistency throughout the paper, it is suggested to change the names in these graphs. We have fixed this. • Figure S4L: The graph is not very clear, especially regarding the significance reported at -50 pA, please modify the graphical visualization and/or add a legend in the caption. We have fixed this.

Reviewer #2 (Significance (Required)):

Nature and significance of the advance for the field. The results presented in the manuscript are potentially interesting and useful, but not completely novel (currents deregulation has already been highlighted in mouse models of Down Syndrome).

2.8) We thank the reviewer for this comment. While we agree that current deregulation has been observed in mouse models of Down syndrome, the novelty and significance of our study lie in demonstrating these alterations directly in human neurons using both in vitro and in vivo xenograft models.

This is a critical advance because the human cortex has distinct developmental and functional properties not fully recapitulated in mice. In fact, three recent studies have already highlighted significant defects mainly in excitatory neurons within the fetal human DS cortex (Vuong et al., bioRxiv, 2025; Risgaard et al., bioRxiv, 2025; Lattke et al, under revision). Our work builds directly on these observations by providing, for the first time, an electrophysiological and network-level characterization of these human-specific deficits.

Our findings thus provide translationally relevant insight that is not merely confirmatory but extends previous work by grounding it in a human cellular context.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

Summary

The manuscript by Peter et al., reports on the neuronal activity and connectivity of iPSC-derived human cortical neurons from Down syndrome (DS) that is caused by caused by trisomy of the human chromosome 21 (TS21).

Major points:

Although the manuscript is potentially interesting, the results appear somehow preliminary and need to be corroborated by control experiments and quantifications of effects to fully sustain the conclusions.

(1) The authors have not assessed the percentage of WT and TS21 cells that acquire a neuronal or glia identity in their cultures. Indeed, the origin of alterations in network activity and connectivity observed in TS21 neurons could simply derive from reduced number of neurons arising from TS21 iPSC. Alternatively, the same alteration in network activity and connectivity could derive from a multitude of other factors including deficits in neuronal development, neurite extension, or intrinsic electrophysiological properties. In the current version of the manuscript, none of these has been investigated.

(2) Electrophysiological properties of TS21 and WT neurons at day 53/54 in vitro indicate an extremely immature stage of development (i.e. RMP between -36 and -27 mV with most of the cells firing a single action potential after current injection) in the utilized culture conditions: This is far from ideal for in vitro neuronal-network studies. Finally, reduced activity of HCN1 channels should be confirmed by specific recordings isolating or blocking the related current.

Main points highlighting the preliminary character of the study.

1) In Figure 1 immunofluorescence images of the neuronal differentiation markers (Tbr1, Ctip2 and Tuj1) are showed. However, no quantification of the percentage of cells expressing these markers for WT and TS21 neurons is reported. On the other hand, simple inspection of the representative images clearly seams to indicate a difference between the two genotypes, with TS21 cultures showing lower number of cells expressing neuronal markers. This quantification should be corroborated by a similar staining for an astrocyte marker (GFAP, but not S100b since is triplicated in DS). This is an extremely important point since it is obvious that any change in the percentage of neurons (or the neuron/astrocyte ratio) in the cultures will strongly affect the resulting network activity (shown in Figure 2) and the connectivity (showed in Figure 4). Possibly, the quantification should be done at the same time points of the calcium imaging experiments.

2) In Figure 2 the authors show some calcium imaging traces of WT and TS21 cultures at different time points. However, they again do not show any quantification of neuronal activity. A power spectra analysis is shown in Supplementary Figure 2, but only for WT cultures, while in Supplementary Figure 3 a comparison between WT and Ts21 power spectra is done, but only at the 50 day time point, while difference in synchrony are assessed at 60 days. At minimum, the author should include in main Figure 2 the quantification of the mean calcium event rate and mean event amplitude at the different time points and the power spectra analysis for both WT and TS21 cultures at the same timepoints.

Of note, the synchronized neuronal activity is present in WT cultures at day 60, but totally lost at subsequent time-points (70 and 80 days). The results of this later time points are different from previous data from the same lab (Kirwan et al., 2015). How might these data be explained? It would be important to rule out any potential issues with the health of the culture that could explain the loss of neuronal activity.It would be beneficial to check cell viability at the different time points to exclude possible confounding factors ? A propidium staining or a MTT assay would strongly improve the soundness of the calcium data.

3) In Figure 3 there is no quantification of the number and/or density of transplanted neurons for WT and TS21, but only representative images. As above, inspection of the representative images seems to show a decrease in cells labeled by the Tbr1 neuronal marker for TS21 cells. Moreover, the in vivo calcium imaging of transplanted WT and TS21 cells lacks most of the quantification normally done in calcium imaging experiments. Are the event rate and event amplitude different between WT and TS21 neurons ? The measure of neuronal synchrony by mean pixel correlation is not well explained, but it looks somehow simplistic. Neuronal synchrony can be more precisely measured by cross-correlation analysis or spike time tiling coefficients on the traces from single-neuron ROI rather than on all pixels in the field of view, as apparently was done here.

4) The results on reduced neuronal connectivity in Figure 3 look very striking. However, these results should be accompanied by control experiments to verify the number of neuronal cells and neurite extension in WT and Ts21 cultures. These two parameters could indeed strongly influence the results. As the cultures appear to grow in clusters, bright-field images and TuJ1 staining of the cultures will also greatly help to understand the degree of morphological interconnection between the clusters.

5) The authors performed RNA-seq experiments on day 50 cultures. Why the authors do not show the complete differential gene expression analysis, but only a small subset of genes? A comprehensive volcano plot and the complete list of identified genes with logFC and FDR values would be helpful. If possible, comparison of the present data (particularly on KCN and HCN expression changes) with published and publicly available expression datasets of other human or human Down syndrome iPSC-derived neurons or human Down syndrome brains will greatly increase the soundness of the present findings. In addition, the gene ontology (GO) results are mentioned in the text, but are not presented. Showing the complete GO analysis for both up and downregulated genes will help the reader to better understand the RNA-seq results. Notably, the results shown in Supplementary Figure on GRIN2A and GRIN2B expression (with values of 300-700 counts versus 2000-4000 counts, respectively) clearly indicate that in both WT and TS21 cultures the NMDA developmental switch has not occurred yet at the 50 days timepoint.

6) The measure of hyperpolarization-activated currents shown in Figure 5 lack proper control experiments. First, the hyperpolarizing current in TS21 cells do not reach a steady-state as the controls. The two curves are therefore hard to compare. To exclude possible difference in kinetic activation, the authors should have prolonged the current injection period (1-2 seconds). Second, to ultimately prove that such currents are mediated by HCN channels in WT cells the authors should perform some control experiments with a specific HCN blocker. A good example of a suitable protocol, with also current blockers to exclude all other possible current contributions, is the one reported in Matt et al Cell. Mol. Life Sci. 68, 125-137 (2011).

7) The manuscript lacks information on the statistical analysis used. Also, the numerosity of samples is not clear. Were the dots shown in some graph technical replicates from a single neuronal induction or were all independent neuronal inductions or a mix of the two ? Please clarify.

8) The method section lacks important information to guarantee reproducibility. Just a few examples: - Only electrophysiology methods for slice are reported, but not for in vitro culture. - Details on Laminin coating is lacking. What concentration was used ? Was poly-ornithine or poly-lysine used before Laminin coating ? - How long cells were switched to BrainPhys medium before calcium imaging ?

Minor point/typos etc.

Introduction

• Page 4 line 6: in the line "Trisomy 21 in humans commonly results in a range in developmental and morphological changes in the forebrain ..." "in" could be replaced by "of".
• Page 5 line 2: please remove "an" before the word "another".
• Page 5 line 2: please replace "ecitatory" with "excitatory"

Results

• Page 10 line 25: The concept of "pixel-wise" appears for the first time in this section and could be better introduced to facilitate the understanding of the experiment.
• In the "results" section, page 11 line 1 and 4, references are made to "Figure 4D" and "4F," but these figures do not appear to be present in the figure section. Upon reviewing the rest of the section, the data seem to refer to "Figure 3D" and "3E."

Discussion

• Page 15 line 20: please replace "synchronised" with "synchronized".
• Page 16 line 11: please replace "T21" with "TS21".

Methods

• Page 19 line 12: "Pens/Strep" has to be replaced by Pen/Strep.
• Page 20 line 20: "Tocris Biocience" has to be replaced by "Tocris Bioscience".
• Page 21 line 2: "Addegene" has to be replaced by "Addgene".

Figures

• Figure 3: the schematic experimental design (Fig. 3A) could be enlarged to match the width of the images/graphs below.
• Figure 5: the reviewer suggests resizing/repositioning the graphs in Fig. 1A so that they match the width of those below.
• Figure S1D: In all the figures of the paper, the respective controls for the TS21 1 and TS21 2 lines are labelled as "WT1/WT2," while in these graphs, they are called "Ctrl1" and "Ctrl2." To ensure consistency throughout the paper, it is suggested to change the names in these graphs.
• Figure S4L: The graph is not very clear, especially regarding the significance reported at -50 pA, please modify the graphical visualization and/or add a legend in the caption.

Significance

Nature and significance of the advance for the field. The results presented in the manuscript are potentially interesting and useful, but not completely novel (currents deregulation has already been highlighted in mouse models of Down Syndrome).

Work in the context of the existing literature. This work follows the line of evidence that characterizes Down Syndrome in human neurons (Huo, H.-Q. et al. Stem Cell Rep. 10, 1251-1266 (2018); Briggs, J. A. et al. Etiology. Stem Cells 31, 467-478 (2013)), both in vitro and in xenotransplanted mice, by corrborating some important findings already found in animal models (Stern, S., Segal, M. & Moses, E. EBioMedicine 2, 1048-1062 (2015); Cramer, N. P., Xu, X., F. Haydar, T. & Galdzicki, Z. Physiol. Rep. 3, e12655 (2015); Stern, S., Keren, R., Kim, Y. & Moses, E. http://biorxiv.org/lookup/doi/10.1101/467522 (2018) doi:10.1101/467522.

Audience. Scientists in the field of pre-clinical biomedical research, especially those working on neurodevelopmental disorders and iPSC-based non-animal models.

Field of expertise. In vitro electrophysiology, Neurodevelopmental disorders, Down Syndrome, ips cells.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Summary

The study investigates the neurodevelopmental impact of trisomy 21 on human cortical excitatory neurons derived from induced pluripotent stem cells (hiPSCs). Key findings include a modest reduction in spontaneous firing, a marked deficit in synchronized bursting, decreased neuronal connectivity, and altered ion channel expression-particularly a downregulation of voltage‐gated potassium channels and HCN1. These conclusions are supported by a combination of in vitro calcium imaging, electrophysiological recordings, viral monosynaptic tracing, RNA sequencing, and in vivo transplantation with two‐photon imaging.

Major Comments

• Convincing Nature of Key Conclusions: The study's conclusions are generally well supported by a diverse set of experimental approaches. However, certain claims regarding the intrinsic properties of the excitatory network would benefit from further qualification. In particular, the assertion that reduced synchronization is solely attributable to altered ion channel expression might be considered somewhat preliminary without additional corroborative experiments.
• One major limitation of the current experimental design is the reliance on predominantly excitatory neuronal cultures derived from hiPSCs. Although the authors convincingly demonstrate differences in network synchronization and connectivity between trisomic (TS21) and control neurons, the almost exclusive focus on excitatory cells limits the physiological relevance of the in vitro network. In the developing cortex, interneurons and astrocytes play crucial roles in modulating network excitability, synaptogenesis, and plasticity. Therefore, incorporating these cell types-either through co-culture systems or through directed differentiation protocols that yield a more heterogeneous neuronal population-could help to determine whether the observed deficits are intrinsic to excitatory neurons or are compounded by a lack of proper inhibitory regulation and glial support.
• Furthermore, the assessment of neuronal connectivity via pseudotyped rabies virus tracing, while innovative, has inherent limitations. The quantification of connectivity as a ratio of red-to-green fluorescence pixels may be influenced by differential viral infection efficiencies, variations in the expression levels of the TVA receptor, or even by the lower basal activity levels observed in TS21 cultures. Complementary approaches-such as electron microscopy for synaptic density analysis or functional connectivity measurements using multi-electrode arrays (MEAs)-could provide additional structural and functional insights that would validate the rabies tracing data.
• Qualification of Claims: Some conclusions, particularly those linking specific ion channel dysregulation (e.g., HCN1 loss) directly to network deficits, might be better presented as preliminary. The authors could temper their language to indicate that while the evidence is suggestive, the mechanistic link remains to be fully established.
• Need for Additional Experiments: Additional experiments that could further consolidate the current findings include:
  • Inclusion of Inhibitory Neurons or Co-culture Systems: Incorporating interneurons or astrocytes would help determine whether the observed deficits are solely intrinsic to excitatory neurons.
  • Alternative Connectivity Assessments: Complementing the rabies virus tracing with electron microscopy or multi-electrode array (MEA) recordings would add structural and functional validation of the connectivity differences.
  • Extended Temporal Profiling: Monitoring network activity over a longer developmental window would clarify whether the observed deficits represent a delay or a permanent alteration in network maturation.
• Reproducibility and Statistical Rigor: The methods and data presentation are largely clear, with adequate replication and appropriate statistical analyses. Nonetheless, a more detailed description of the experimental replicates, particularly regarding the viral tracing and in vivo transplantation studies, would enhance reproducibility. The availability of raw data and scripts for calcium imaging analysis would also further support independent verification.

Minor Comments

• Experimental Details:

Minor revisions could include clarifying the infection efficiency and expression levels of the viral constructs used in connectivity assays to rule out technical variability. - Literature Context:

The authors reference prior studies appropriately; however, integrating a brief discussion comparing their findings with alternative DS models (e.g., organoids or other hiPSC-derived systems) would improve contextual clarity. - Presentation and Clarity:

Figures are generally clear,.But the manuscript contains a minor labeling error. On page 13, the figure is erroneously labeled as "Fig6A", whereas, based on the context and corresponding data, it should be "Fig5A". I recommend that the authors correct this mistake to ensure consistency and avoid potential confusion for readers.

Significance

• Nature and Significance of the Advance:

The work offers a substantial conceptual advance by providing a mechanistic link between trisomy 21 and impaired neuronal network synchronization. Technically, the study integrates state-of-the-art imaging, electrophysiology, and transcriptomic profiling, thereby offering a multifaceted view of DS-related neural dysfunction. Clinically, the findings have the potential to inform future therapeutic strategies targeting network connectivity and ion channel function in Down syndrome. - Context in the Existing Literature:

The study builds on previous observations of altered network activity in DS patients and DS mouse models (e.g., altered EEG synchronization and reduced synaptic connectivity). It extends these findings to human-derived neuronal models, thus bridging a gap between clinical observations and molecular/cellular mechanisms. Relevant literature includes studies on DS neurodevelopment and the role of ion channels in synaptic maturation. - Target Audience:

The reported findings will be of interest to researchers in neurodevelopmental disorders, Down syndrome, and ion channel physiology. Additionally, the study may attract the attention of those working on hiPSC-derived models of neurological diseases, as well as clinicians interested in the pathophysiology of DS. - Keywords and Field Contextualization:

Keywords: Down syndrome, trisomy 21, neuronal connectivity, synchronized network activity, hiPSC-derived cortical neurons, ion channel dysregulation.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Reply to the Reviewers

I thank the Referees for their...

Referee #1

1. The authors should provide more information when...

Responses + The typical domed appearance of a hydrocephalus-harboring skull is apparent as early as P4, as shown in a new side-by-side comparison of pups at that age (Fig. 1A). + Though this is not stated in the MS 2. Figure 6: Why has only...

Response: We expanded the comparison

Minor comments:

1. The text contains several...

Response: We added...

Referee #2

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

In this work Neupane et al used large-scale robust CRISPR-based gene activation and ablation screens to identify novel regulators of α-synuclein pathology in synucleinopathies using as read-out p-αSyn129 signals by high-throughput fluorescence microscopy. The authors reveal that mitochondrial protein OXR1 promotes Ser129-phosphorylated αSyn aggregation, while ER-associated EMC4 suppresses it via enhanced autophagic clearance, highlighting new possible mechanistic pathways in disease progression of alpha-synucleinopathies.

Major comments:

1. As correctly pointed out by the authors in Introduction p-Syn is associated with aggregates, but its functional role is far to be clear and both neuroprotective or pro-aggregations effects have been proposed. Further it has been shown that, physiological neuronal activity augments Ser129-phospho αSyn, which is a trigger for protein-protein interactions, which in turn is necessary for mediating αSyn function at the synapse (https://doi.org/10.1016/j.neuron.2023.11.020). As a consequence modulation of p- p-αSyn as possible therapeutical target for PD and synucleinopathies is quite a complicate matter. The assumption, on which the whole paper is based, that increase in p-αSyn equates to αSyn aggregation and disease progression is rather weak to this reviewer, unless further validation of it is provided. Indeed while the authors performed experiments on human iPSC-derived cortical and dopaminergic neurons on p-αSyn analysis, any measurement of αSyn aggregates/oligomers, and neuronal degeneration is provided. It is recommended to provid this experiments ideally using different tecnique like αSyn-Proximity Ligation Assay for measurements of oligomers, as it has been largely validated in autoptic brains of PD, MSA and DLB patients (doi: 10.1007/s00401-025-02871-w.), as well as cell viability/apoptosis and neurites degeneration measurements upon OXR1 and EMC4 modulation in iPSC derived cortical and dopaminergic neurons.
2. The authors claims in Results page 5: "The absence of cytoplasmic pSyn129 signal in HEK293 cells lacking α-Syn overexpression demonstrates that elevated α-Syn levels are essential to drive robust and rapid aggregation. Moreover, it indicates that the 81A antibody selectively recognizes de novo aggregates rather than the recombinant seeds". The fact that ab81A recognize deNovo aggregates and not rec seeds is quite speculative, not supported by data, and might rather indicate that ab 81A does not recognize aggregates. Thus this further implays that other technology like for example Seeding amplification assays are being employed by the authors in addition to p-αSyn129 signals in validation experiments for example in genetic PD (ideally GBA1 or LRRK2) IPSC-derived dopaminergic neurons.

Minor comments:

1. The strain-specific effects especially from patients-derived fibrils of OXR1 activation and EMC4 depletion on pSyn levels is rather weak in comparison with RAB3 and PIKFYVE (fig 3F-G) and therefore the expected relevance of these results especially in vivo in patients should be better clarified and modulated in discussion
2. In discussion authors write "We observed that OXR1 activation preferentially increases α-Syn aggregates phosphorylation (EP1536Y) in neuronal somata, suggesting that mitochondrial dysfunction exacerbates α-Syn phosphorylation in later-stage aggregates." This is quite a surprising result since distal axonal endings are particularly susceptible to mitochondrial impairments for anatomical and physiologically reasons and if p-αSyn129 accumulation is driven by mithocondrial disfunction as suggested by this paper, this should be detected in neurites as well. Please clarify.
3. Authors say that they targeted mitochondrial, trafficking, and motility (MTM) genes in human cellular models. While mitochondrial and trafficking is clear in the context of Parkinson and neurodegnerative disease, less clear is the motility genes. Please expand on this.

Significance

This is a well written, comprehensive study with a well characterized, robust CRISPR-based gene activation and ablation screening pipeline to identify novel regulators of α-synuclein pathology. Methodology is rigorous and clearly described and results are well presented. The major limitation relays in the validation experiments where only one main read-out that is p-αSyn129 fluorescence signal is employed, limiting the significance and impact of the presented results. I believe that the basic science community might benefit principally of the proposed methodology of a large high-throughput screening to modulate a large set of genes, a platform that in principle might be used also for other scientific questions.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

In the present study, Neupane et al. performed arrayed CRISPR activation and ablation screens, targeting genes related to mitochondria, trafficking and motility, to identify genes that modulate the presence of Ser 129 phosphorylated alpha-synuclein aggregates (pSyn129) upon administration of exogenous preformed alpha-synuclein fibrils. The screens have been performed in HEK cells stably overexpressing alpha-synuclein in two independent replicates, and hits have been further validated in induced pluripotent stem cell derived forebrain and dopaminergic neurons. Following functional validations, the authors conclude that enhancing the expression of OXR1 results in a modest increase in the number of pSyn129 puncta within cells, and their size, while partial loss of EMC4 expression reduces these puncta. To date some pre-print studies have used genome-wide CRISPR screening to identify modifiers of the accumulation of alpha-synuclein preformed fibrils in cells, suggesting the importance of uptake and endolysosomal trafficking for the propagation of alpha-synuclein aggregates in recipient cells. Although the topic is of interest in the field of Parkinson's disease and synucleinopathies in general, the readout of the present screen (presence of pSyn129) is not very sensitive and without investigating endogenous alpha-synuclein or cell homeostasis in neuronal models limits the stated conclusions.

Major comments:

• Please clarify whether the positive control genes RAB13 and PIKFYVE were nominated hits within the CRISPR screens. Specifically, the authors state that the positive control of the CRISPRa screen was RAB13, expected to reduce pSyn129 upon overexpression, nevertheless this gene does not appear as a hit in the CRISPRa volcano plot (although present in table S1 but not making the cutoff). In figure 2D, activation of RAB13 does not seem to impact the main readout phenotype. Moreover, in the CRISPRo screen, PIKFYVE was used, but this gene is also not presented as a hit linked to reducing pSyn129 in the CRISPRo plot. If these control genes do not come up as hits, it is difficult to support the conclusions of the screen.
• The effect size for screen hits presented in figure 2A/B is rather small. It is difficult to interpret the power of these findings in the absence of uptake efficiency controls, such as dextrans of appropriate molecular weights.
• The readout of the screen is not very sensitive, and it is unclear what it represents. Specifically, in Figure 2F, G the authors validate the hits OXR1 and EMC4, showing a small effect, albeit statistically significant. The authors should strengthen this data by adding more experiments addressing, for instance, what the pSyn spot area and spot intensity signify for the cell. Some experiments in a neuronal context are important, including SNCA KO as a negative control.
• It is unclear why the authors chose to follow up on the OXR1 and EMC4 hits. Please explain the rationale for follow-up studies.
• Generally, the notable difference in the number of pSyn129+ cells in the non-targeting across various experiments (including Fig.1G/I compared to Fig.2G/I or Fig. 3F/G or flow cytometry experiment) suggests the readout is not very sensitive.

For instance, in figure S3 it would be important to add an experiment controlling for cell number as opposed to LDH release, as the micrographs show some differences in cells number, e.g. in the ntg vs. EMC4 condition. - The data is not sufficient to suggest that OXR1 and EMC4 are strong modulators of alpha-synuclein aggregation, as the authors suggest based on figures 2 and 3 that show statistically significant difference and a rather small effect size. It is important to provide more insight into how these genes may affect endogenous alpha-synuclein and cellular homeostasis in more detail, especially in neuronal models. Further investigating the hits in this direction in additional genetic backgrounds would also increase the relevance of the findings, e.g. in SNCA triplication or GBA-PD neurons.<br /> In Fig. S8B the immunoblot analysis shows there may be an effect of EMC4 and OXR1 CRISPRa on α-synuclein levels; please quantify for both iPSC-derived cortical neurons and dopaminergic neurons. - The pattern of tyrosine hydroxylase staining in Figure 5F does not seem specific or as expected for iPSC-derived dopaminergic neurons. Furthermore, since endogenous SNCA expression is expected to be analogous to the expression of TH (with TH+ cells expressing higher SNCA), it would be important to compare pSyn129 between TH+ cells and/or relative to the TH+ area.

Minor comments:

• The authors report that RAB13 overexpression reduces pSyn129⁺ prevalence, whereas RAB13 ablation (CRISPRo screen) enhances pSyn129⁺ levels (Figures 2D-2E). Please revise as these specific figures show no effects for this gene.
• Please specify how many individual cells (approximately) were quantified in each figure legend.
• Figure 3F/G may be better as a supplemental figure since it does not add to the conclusions of the study.
• It would be good to clarify for the reader some of the genes that serve as positive controls for the screen's readout (as shown in Fig. S2D/G).
• It would be helpful to further clarify which cell type was used in each figure legend.

Significance

Important topic but their experimental design limits the significance of their findings. Hard to improve the work in a reasonable amount of time. Also many technical issues.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Summary:

This study by Neupane et al. investigates modulators of α-synuclein aggregation, focusing on Ser129-phosphorylated α-synuclein (pSyn129), a pathological hallmark of Parkinson's disease (PD). The authors performed high-content image-based, arrayed CRISPR activation (CRISPRa) and knockout (CRISPRo) screens targeting > 2300 genes related to mitochondrial function, intracellular trafficking, and cytoskeletal reorganization. Using α-Syn overexpressing HEK293 cells, they identified OXR1 and EMC4 as novel modulators of pSyn129 abundance. Key findings were that activation of the mitochondrial protein OXR1 increased pSyn129 by decreasing ATP levels, while ablation of the ER-associated protein EMC4 reduced pSyn129 by enhancing autophagic flux and lysosomal clearance. These findings were validated in human iPSC-derived cortical and dopaminergic neurons.

My major comments have to do with statistical methods and with significance of their findings.

Major comments:

Are the claims and the conclusions supported by the data or do they require additional experiments or analyses to support them?

The claims and conclusions are generally well-supported by the presented data. The dual CRISPRa/CRISPRo screen provides a robust initial discovery platform, and the validation in iPSC-derived neurons strengthens the findings and their translational relevance. The mechanistic insights into OXR1 (ATP levels) and EMC4 (autophagic flux, lysosomal clearance) are supported by the described experiments. The use of two antibodies (81A and EP1536Y) for pSyn129 also enhances confidence in the measurements. I had a few questions about the statistical methods. The main concern I have about methodology for the screen is whether the authors have corrected for multiple hypotheses in their discovery screen. This is not clear from the text, methods, or legends (for Figures 2A/2B/2C).

• Figure 1B suggests a very large range of activation (multiple orders of magnitude) in the initial screen. What is the relationship between level of expression change and functional effect across the screen? How upregulated/downregulated are OXR1 and EMC4 at the mRNA and protein levels?
• Supplemental Figure S2D: Why do the non-targeting controls differ from the majority of the CRISPRa genes? If I am reading the figure correctly, it seems strange that the vast majority of the CRISPRa gene targets reduces pSyn pathology relative to the non-targeting controls (which is why I am wondering whether the level of increased expression correlates with the level of functional effect).
• In Figure 2A/B/C, is the p-value adjusted in any way for multiple comparisons? If so, this should be indicated in the legend. If not, why not? (The potential for false positives in a screen is very large and requires correction for multiple comparisons.)
• Figure 3: It's interesting that different seeding materials have different effects. However, it's quite surprising that the authors find less seeding with MSA-derived material in both the CRISPRa and CRISPRo context. This contradicts the work of Peng and coauthors (PMID 29743672) who find that MSA-derived material is much more potent in seeding aggregates in a number of different cell types. Do the authors have any thoughts about why this is the case?
• Figure 7A: pSyn129 image in the non-targeting control is poor - the very bright dots look like artifact. Not clear why the authors don't corroborate with EP1536Y antibody as they do in Figure 5.
• Overall methodology: Are the pSyn inclusions soluble? This could be easily determined by performing 1% TritonX extraction, for example, and it helps us understand how "pathological" the inclusions are.
• OPTIONAL: The authors perform some interesting experiments looking at genes affected downstream by, for example, OXR1 over-expression. It would be useful to understand whether the upstream effect is dependent on downstream effect. This could be tested by performing double perturbations (e.g. OXR1 overexpression and CCL8 knockout or ALDOC upregulation).
• OPTIONAL: The link between EMC4 ablation and enhanced ER-driven autophagic flux/lysosomal clearance could be corroborated with additional experiments. E.g.: Does EMC4 normally inhibit this pathway? Or only in the context of aSyn fibril seeding?

Are the suggested experiments realistic in terms of time and resources?

The OPTIONAL experiments are generally feasible as they employ methods that the lab is already using in this paper.

Are the experiments adequately replicated and statistical analysis adequate?

See comment about multiple hypothesis testing above.

Significance

This is a well-designed, difficult-to-accomplish study that expands the landscape of pS129Syn modulators. The validation of the primary hits identified in HEK293 cells in iPSC-derived neurons gives the findings greater relevance.

Strengths:

• Novelty: Using an unbiased and high-throughput approach, the study identifies two novel regulators of α-Syn aggregation, namely OXR1 and EMC4.
• Methodological Rigor: The use of arrayed CRISPRa/CRISPRo screens with high-content imaging is powerful and difficult to accomplish. Methodologically, this is a tour de force.
• Orthogonal Validation: The use of multiple α-Syn fibril polymorphs/strains and different antibodies (81A, EP1536Y) strengthens the robustness of the findings.

Limitations:

• It's not clear to me that pSyn129 is the ultimate readout. At a minimum, we should know something about the solubility of the inclusions. Some panels (e.g. Figure 7A) are not very informative in terms of what the authors are calling pSyn129+.
• The study relies on in vitro cellular models. While iPSC-derived neurons are relevant, the complexity of the brain environment, including glial cell interactions is not fully captured. This is fine for an initial report, but it does limit the significance.
• OXR1 and EMC4 seem to be very generic modulators. It's not clear to me that their effects are specific to aSyn or to PD in any way - they might just be effects on very basic cellular functions that would be applicable to a number of stressors or proteinopathies. Maybe that is fine (we probably need to get rid of tau aggregates, too!), but I don't think the authors can claim that they have identified "organelle-specific genetic nodes of aSyn pathology" since they biased their screen towards mitochondria and they don't test any other pathological aggregates. Moreover, from a translational perspective, it's not clear to me that implicating the antioxidant pathway or lysosomal/autophagosomal pathways in the pathogenesis of PD is new, and it's not clear that the specific genes identified would make good therapeutic targets.