Author Response

Reviewer #2 (Public Review):

The manuscript by Mohebi et al. examines a critical open question regarding the interaction of cholinergic interneurons of the striatum and transmitter release from dopaminergic axons in behaving animals. Activation of cholinergic interneurons in the striatum can evoke dopamine release in brain slices and in vivo as measured with voltammetry. However, it remains an open question in what context and to what extent this acetylcholine-mediated dopamine occurs in behaving animals. Here, the authors argue that CIN activity triggers dopamine release in the nucleus accumbens which encodes the motivation to obtain a reward through increasing "ramps" of dopamine release. Their data suggest that the ramps are not reflected in the firing of dopaminergic neurons. Rather, they provide compelling evidence that the ramps of dopamine release correlate with ramps in cholinergic interneuron activity as measured with GCaMP6. What's more, the authors show that ACh-mediated dopamine release has no paired-pulse depression, a striking result that differs from all prior ex vivo brain slice data. The manuscript is extremely well written and the data are of very high quality. Overall, this study represents an important step forward in our understanding of how ACh-mediated dopamine release regulates behavior, and more broadly how axons can generate behaviors independently from somatic activity.

Major comments

1) The complete absence of any short-term plasticity in CIN-mediated dopamine release is a striking result that is important for the field. The authors should strengthen this result with additional quantitative analysis demonstrating the lack of STP. They have analyzed paired-pulse ratios, but they should analyze this for stimuli at the higher frequencies (4 Hz, etc) that are more physiologically relevant. For example, Fig 1e shows a CIN-evoked DA release at many optically-stimulated frequencies. The authors should quantify short-term plasticity by generating fits of the single stimulus signal and comparing the mathematical sum predicted from 4 stim DA signals at different frequencies to the recorded data. A similar analysis has been done with Ca signals (Koester and Sakmann, 2000).

Thank you for this very helpful suggestion. We have performed this analysis as recommended, and now confirm the lack of STP even at the higher frequencies (see new Supplementary Figure 1).

2) The authors show that optical activation of CINs results in DA release as measured by dLight. To clearly establish that these signals are generated by DA release driven by nicotinic receptors (and not a partial effect of some unknown artifact), it would be useful to show that the optical CIN-evoked dLight signals shown in Fig. 1 are inhibited by nicotinic receptor antagonists such as DHbE. This control experiment would significantly strengthen the result shown here.

We agree that combining drug manipulations with photometry would be useful, but as noted above this is not a methodology in our current technical repertoire.

3) Similarly, the authors show clear correlations between CIN activity and DA release during behavior. The authors should consider determining whether CINs play a causal role in triggering DA release during behavior. For example, does infusion of DHbE in the NAc prevent the light-mediated DA release during behavior? As an alternative hypothesis, some groups have been suggesting that CIN activity has almost no direct influence over DA. Therefore, testing whether a causal relationship exists between CINs and DA release would be an important experiment in addressing these two opposing viewpoints.

As noted above we are not currently able to combine drug manipulations with photometry in behaving animals.

4) The ramps that are described in this manuscript are an order of magnitude faster (increasing over 100s of milliseconds) than ramps described in other studies that occur over seconds. In fact, the two signals may be completely different functionally. Discussion of this topic would be helpful.

Dopamine ramps have indeed been reported over multiple different time scales, and as discussed in Berke 2018, this seems to reflect the duration of the approach behavior. We think further discussion of this topic is better saved for another paper, especially as we are now actively studying ramping over longer time scales (Krausz et al. 2023).

Reviewer #3 (Public Review):

This report by Mohebi et al. provides new answers to old questions by showing that the activity of striatal cholinergic interneurons (CINs) escalates progressively during specific reward-related behaviors and that this correlates with previously observed ramps in dopamine (DA) release in the nucleus accumbens core. The report is strong and provides evidence for the authors' hypothesis that DA ramps are independent of DA neuron activity, but are instead the result of CIN activity and corresponding acetylcholine (ACh) release. The authors further demonstrate that the fidelity of CIN activation and consequent driving of DA release is even more robust in vivo than observed ex vivo slice preparations, which is fundamental for understanding the role of ACh-DA interactions in behavior. The findings complement the authors' previous evidence ventral tegmental area (VTA) DA neuron firing patterns do not show a ramping pattern; the previously reported VTA data are appropriately included here (in Fig. 3) to illustrate the absence of VTA firing during the time-locked increases in CIN activity and DA release. The present studies stop short of showing a direct link between CIN activity and DA release, however, which would require examining DA release during behavior in the presence of an antagonist of nicotinic ACh receptors. The authors also extend the understanding of the regulation of DA release by acetylcholine (ACh) by showing that optical activation of CINs in vivo promotes DA release responses that do not attenuate with repetitive stimulation. This contrasts with previous results in ex vivo striatal slices in which ACh-evoked DA release has been found to decline progressively from rundown and/or receptor desensitization. The authors propose that in vivo, AChE may be more effective in curtailing local ACh levels than in slices because of the slightly lower temperature typically used for slice studies, as well as the use of superfusion that might facilitate some AChE washout (AChE inhibitors are still effective in slices, of course). Overall, the report not only provides evidence for the cellular substrate for DA ramps but also shows the robustness of ACh-driven DA release in vivo. A few points to strengthen the report are listed below.

1) The authors give a few details about how CINs were activated at the beginning of the results, but say only that DA dynamics were monitored using fiber photometry. Given that the methods are at the end, a brief summary should be given here to indicate whether this means direct monitoring of DA or indirect via GCaMP, for example. It would be helpful to note the sensor used in the abstract, as well. In this light, as it were, RdLight1 should be described upon the first mention.

We have now clarified in both abstract and text that we are using the direct DA sensor RdLight1.

2) The authors show that infusion of DHbE in the NAc likelihood of decisions to approach the center port, as did antagonism of DA receptors. This supports the authors' argument that ramping of CIN activity and consequent ACh release underlies observed ramps in DA release. However, to show a causal interaction requires testing whether the observed DA ramps are absent after DHbE infusion in the NAc, under the same conditions that attenuated behavior.

As noted above we are not currently able to combine drug manipulations with photometry in behaving animals.

3) In Fig. 3, the y-axis title for the upper panels should specify VTA, not simply "rate". This is stated in the legend, but should also be specified in the figure panel.

We have updated the y-axis titles in this figure.

4) A recent preprint in BioRxiv by AC Krok, NX Tritsch et al. shows a related correlation between ACh and DA release in vivo in a reward task, as well as differences in other conditions. This report shows also that cortical input to CINs indeed plays a role, as suggested in the concluding sections of the present report. Consideration of the data in the preprint in the context of the present results could be valuable for the field.

We have also noted those pre-prints with interest, even though they investigated different brain regions using different approaches. There are established differences between CIN-DA interactions in dorsal vs. ventral striatum that we suspect are relevant here. But given the rapid pace of developments in this subfield, we prefer not to speculate too much at this point and instead review the overall body of work once it is published.

Author Response

Reviewer #1 (Public Review):

“The abstract does not adequately summarize the content of the paper. There is no mention of stimulation, or bilateral connectivity, which is a large part of the paper. The names of all five species should appear in the abstract, not just X. laevis.”

In the revised manuscript, we have included all the names of the species and types of stimuli used to elicit fictive vocalizations in the abstract. In regard to bilateral connectivity, we believe that the reviewer was referring to the rostral-caudal connections between the parabrachial nucleus and nucleus ambiguus, which are critical for fast, but not for slow trill production. We have added this piece of information in the abstract. Furthermore, we have clarified the bilateral nature of the two central pattern generators (CPGs) in male X. laevis. In our previous study (Yamaguchi et al., 2017), we demonstrated that transections of the two commissures (one at the parabrachial nuclei level and the other at the nucleus ambiguus level) did not eliminate fictive advertisement calls in male X. laevis brains, indicating the presence of fast and slow trill CPGs in left and right hemi-brains of male X. laevis. This information was originally included in the results section (“Unilateral transection desynchronizes the fast clicks, but not the slow clicks across species”). We have now added this information to the introduction section to provide a clearer description of the anatomical organization of the two CPGs (p5, ln10 – 14).

“The conclusion that the "fast and slow CPGs identified in male X. laevis are conserved across species." is contradicted by the last paragraph, which states, "Fast trill-like CPGs are likely present only in fast clickers..." This inherent contradiction needs to be resolved.”

To resolve this contradiction, we have revised sentences in the abstract to clarify our findings. Specifically, we now state that “We found that even though the courtship calls of different Xenopus species vary in their click rates and duration, the CPGs used to generate clicks are conserved across species. The fast CPGs found in male X. laevis, which critically rely on reciprocal connections between the parabrachial nucleus and the nucleus ambiguus, are conserved among species that produce fast clicks. Similarly, the slow CPGs found in the caudal brainstem of male X. laevis are shared among species that produce slow clicks” (p2, ln 10 – 15) By making this change, we hope to provide greater clarity regarding our findings and help to resolve any contradictions.

“The testosterone results are over-emphasized.” “The conclusion that there is differential expression of testosterone receptors in the brain of each species is completely speculative and not supported by the data presented here.”

We have extensively revised our manuscript to ensure a more accurate interpretation of the results regarding testosterone experiments. Revised conclusions are outlined below:

Abstract: “In addition, our results suggest that testosterone plays a role in organizing fast CPGs in fast-click species, but it does not appear to have the same effect in slow-click species.” (p2, ln 15 – 17)

Introduction: “Additionally, we found that fast trill-like CPGs are present only in species that produce fast clicks and their presence appears to be regulated by testosterone in these species. “ (p6, ln 2 – 4)

Discussion: “However, this effect of testosterone appears to be limited to the fast clicker species. Male X. tropicalis, a slow clicker species, has been shown to have comparable plasma levels of testosterone to male X. laevis (mean plasma levels of testosterone of male X. laevis: 13 to 22ng/ml (Hecker et al., 2005; Hayes et al., 2010), male X. tropicalis: ~20ng/ml, (Olmstead et al., 2009)), yet the synapses between the PBN and laryngeal motoneurons in male X. tropicalis remained weak, and PBN showed little activity during fictive advertisement calls. These results suggest that testosterone acts differently on the central vocal pathways of fast and slow clickers, promoting the emergence of fast trill-like CPGs in X. laevis but not in X. tropicalis. Although further experiments with controlled testosterone levels are necessary to validate these results, we hypothesize that changes in the androgen receptors (e.g., expression patterns, ligand affinity) may have contributed to the divergence of fast and slow clickers.“ (p26, ln 13 – 24)

”The use of the word "development" implies embryology. Here, adults were treated and looked at 13 weeks later. There is no data presented about development. ”

In our revised manuscript, we have replaced the term “development” with “presence” or “acquisition” of neural circuitry to enhance the clarity and help to prevent any potential misunderstandings.

Author Response

Reviewer #1 (Public Review):

By the in vitro DNA damage response (DDR) assay with a defined DNA substrate using Xenopus extracts and in vitro binding assays with purified proteins, the authors nicely showed the role of APE1 (APEX1) in ATRIP recruitment for DDR activation, particularly a non-enzymatic (structural) role of APE1 in the binding to both ssDNAs and ATRIP. The results described in the paper are very convincing to support the authors' claim. However, these studies lack the quantification with proper statistics (and/or mentioning the reproducibility of the results). And, given the important discovery of APE1 in the DDR activation in vitro, it would be nice to demonstrate the role of APE1(APEX1) in ATR activation in vivo using siRNA-mediated knockdown of mammalian cells or yeast cells.

Thanks for the suggestion. As shown in our response to the #2 Essential Revisions, we have addressed this question by additional experiment and added extra description in our revised manuscript showing that APE1 is important for the ATR DDR following oxidative stress in culture human cell U2OS cells (Figure 1-figure supplement 1B). In addition, we have performed at least three independent experiments and statistical analysis to support our claims.

Reviewer #2 (Public Review):

ATM and Rad3-related (ATR) interact with ATRIP and plays a central role in DNA damage response. Previous studies have established the idea that ATR is recruited to RPA-covered ssDNA via ATRIP-RPA interaction. In this paper, the authors propose a new RPA-independent mechanism for ATR recruitment.

Thanks for the nice summary of our major findings from the manuscript.

Reviewer #3 (Public Review):

In this manuscript, the authors explore the mechanism of ATRIP recruitment to single-stranded DNA (ssDNA), which is important for ATR activation and the subsequent control of DNA repair and cell cycle progression. Using Xenopus egg extracts, in vitro interaction assays, and ssDNA constructs, the authors found that AP endonuclease 1 (APE1) plays a role in the recruitment of ATRIP to ssDNA independently of RPA. Moreover, APE1 domains are characterized for ssDNA, ATRIP, and RPA interaction, determining that the nuclease activities of APE1 are not required for this new mode of ATRIP recruitment. Overall, the work presented makes a compelling case for a novel role for APE1 in ATRIP recruitment that seems crucial for ATR activation (at least in the Xenopus system). The results are likely to have an important impact on our understanding of the determinants for activation of ATR signaling and cellular responses to DNA damage and replication stress. It remains unclear whether the findings will be extended to other organisms and be relevant for different types of DNA lesions. Also, there are several points of concern in the manuscript that require further clarification, especially regarding some of the quantitative analyses presented and the claimed importance of the RPA-independent mode of ATRIP recruitment for ATR activation.

We thank the reviewer’s overall positive evaluation of our initial submission. We have included additional experimental data using mammalian cells showing the significance of APE1 in the ATR DDR, and also additional discussion of other studies in the literature. We also provided further clarifications or responses to the major/minor concerns (please see below detailed responses). In particular, we revised the proposed model of APE1 in ATRIP recruitment and ATR DDR (Please see revised Figure 5).

Author Response

Reviewer #1 (Public Review):

The data on embryonic "ventral nerve cord" glia are generated from whole embryos, and even provided that the ventral nerve cord harbors 75% of all glia and thus the majority is ventral nerve cord, the data should not be called vnc-specific. The vnc-specific data set (adult CNS) that is already published (Allen et al., 2020) is strangely not even mentioned in the current manuscript. The idea of having a comprehensive description of glial transcriptional profiles is great - but I was missing the integration of the midline glial cells, which can be considered as ensheathing glial cells that - as the cortex glia - also express wrapper (Stork et al., 2009).

• We agree with Reviewer 1 that the embryonic glia dataset represents all glia and not just VNC glia. We have amended the text accordingly.

• We now cite the Allen et al., 2020. Apologies for this omission.

• Midline Glia:

The embryonic glial cells analysed in the previous version of our manuscript included only repo+ glia only and therefore did not include midline glia, which do not express repo (Jacobs, 2000). In the revised manuscript, we reanalysed the complete embryonic dataset and identified the midline glia based on known markers and in vivo validation (Figure 3 – figure supplement 1). We also provide a list of genes that show enriched expression in the midline glial cluster as a supplementary file (Source data file 1).

We performed hierarchical cluster analysis on midline glia, all embryonic repo+ glial clusters and embryonic neuronal clusters to determine the relationship of midline glia to other glia. Interestingly, midline glia formed an outgroup to both neurons and repo+ glia (Figure 3 – figure supplement 1F), suggesting that they are quite distinct from other (repo+) glial classes. This is expected given their mesectodermal origin (Kosman et al., 1991; Thomas et al., 1988). Indeed, although midline glia express wrapper, otherwise known as a cortex glia marker (Banerjee et al., 2017; Noordermeer et al., 1998; Stork et al., 2009), they do not resemble cortex glia in form or function but instead ensheath commissural axons and play critical roles in axon guidance and VNC morphogenesis (Jacobs, 2000). Midline glia have been characterised extensively by several groups (Hartenstein, 2011; Hidalgo, 2003; Jacobs, 2000; Kearney et al., 2004; Vasenkova et al., 2006; Wheeler et al., 2006), therefore, given their distinct origin and the ambiguity surrounding their functional classification, we instead focused our analyses on repo+ glia in this manuscript.

Unfortunately, I found most of what is reported in this work not to be entirely new. The classification of glial diversity in the adult brain was presented by the Meinerzhagen and Gaul labs (Edwards and Meinertzhagen, 2010; Edwards et al., 2012; Kremer et al., 2017). The description of two astrocyte-like cell types is a reduction of data that defined three morphologically distinct astrocyte-like cells (Peco et al., 2016), which is not discussed. Some other aspects were ignored, too. Two other morphological distinct types of ensheathing glia exist, ensheathing glia and ensheathing/wrapping or track-associated glia were described but this is not discussed (Kremer et al., 2017; Peco et al., 2016).

We respectfully disagree with Reviewer 1’s assessment that much of the work presented in not new. This work represents the first Drosophila glial cell atlas with thorough validation of cluster marker expression in vivo. It is also the first systematic exploration of the relationship between glial morphology and transcriptional signature, a controversial topic in the field of glial biology. We fully agree that much of the adult glial morphology had been characterised previously by the Meinerzhagen and Gaul labs among many others and we acknowledge this explicitly in our manuscript and in references to Figures 2 (one out of a total of 9 main figures). Indeed, it is because Drosophila glial morphology has been so well characterised that a comprehensive exploration of the relationship between morphology and transcriptional signature was even feasible. Moreover, our revised manuscript also provides more in-depth morphological characterisation and quantification of glial morphology and defines subclasses and morphologies not described previously (e.g. channel perineurial glia and astrocyte morphologies of the lobula and lobula plate). Indeed, even the channel subperineurial glia, which were identified based on lineage relationships, nuclear position and molecular markers, were not described in morphological terms.

The 3 distinct astrocyte populations defined in Peco et al., (2016) refer to cell body position and neuropil domains covered by astrocytes. We now include this categorisation in our quantification of astrocyte morphology (See response to (6) and Figure 1 – figure supplement 2) and discuss their relationship to the type 1 and type 2 astrocyte morphologies that we observed.

As well we now include the ensheathing/wrapping or tract ensheathing glia as a morphological category of ensheathing glia in the manuscript (Figure 1A,N,O).

Author Response

Reviewer #1 (Public Review):

This is a simulation study comparing the performance of two major approaches for dealing with “population structure” when carrying out Genome-wide Association Studies - Principal Component Analysis and Linear Mixed-effects Models - a subject of considerable practical importance. The author correctly notes that previous comparisons have been quite limited. In particular, any study not concluding that LMM was superior has relied on very simple models of structure.

The paper is clearly written and beautifully reviews the theoretical underpinnings (albeit in a manner that will be difficult to penetrate without deep knowledge of several fields). The simulations are well-designed and far better than previous studies. From a theoretical point of view, the work is somewhat limited by being strongly anchored in a very classical quantitative genetics framework that is focused on allele frequencies and inbreeding coefficients, and totally ignores coalescent theory, but this is a minor quibble. The simulations are limited by utilizing ridiculously small sample sizes by the standards of modern human GWAS. And of course, they do not include all the complexities of real data.

The quantitative genetics framework we used was ideal for motivating and interpreting LMMs in particular, since they model relatedness with a kinship matrix which consists of IBD probabilities, all of which arose from quantitative genetics.

We also added the following text to the discussion: “However, our conclusions are not expected to change with larger sample sizes, as cryptic family relatedness will continue to be abundant in such data, if not increase in abundance, and thus give LMMs an advantage over PCA (Henn et al., 2012; Shchur et al., 2018; Loh et al., 2018).”

The main conclusion of the study is that LMM really are generally superior - as expected on theoretical grounds. However, the authors do address whether switching to LMM really is practicable given the sample size and lack of data sharing that characterize human genetics. Nor is it clear whether the difference in performance matters in real life given that the entire framework used is an idealized one - the fact that real human data suffers from environmental confounders that are correlated with “ancestry” is not addressed, to take the most obvious example. That said, it is surely important to note that the approach routinely used by the majority of users (PCA with 10 PCs) is most used for historical reasons and has little theoretical or empirical justification.

We added simulations with environment effects correlated with ancestry, which we hope will make our study even more relevant as it does make our evaluations even more realistic than before. In the presence of environment effects, LMM without PCs remains among the best approaches, although occasionally LMM with PCs or PCA will perform slightly better. However, modeling environment directly (with the true variables) improves performance much more than by using PCs to model environment indirectly, so we believe that is not a strong reason for continuing to use PCs (in LMMs or otherwise) unless there is no choice.

We also added the following text to the discussion: “However, recent approaches not tested in this work have made LMMs more scalable and applicable to biobank-scale data (Loh et al., 2015; Zhou et al., 2018; Mbatchou et al., 2021), so one clear next step is carefully evaluating these approaches in simulations with larger sample sizes.” As stated earlier, we believe that the difference in performance between LMM and PCA will remain in larger sample sizes because cryptic relatedness is more prevalent in that setting.

We excluded the “lack of data sharing” point from our discussion because it does not align well with the goals of our manuscript. The current solution to the lack of data sharing is meta-analysis, but its use does not give PCA or LMM an inherent advantage, since it can be applied to the summary statistics of either (or even a combination of models, in theory). There is interesting recent work on “federated” PCA and LMM association (both versions exist), that allow a single model to be fit jointly to separate datasets (residing in different buildings across the world) as if they were combined into a single dataset. Thus, these issues do not explain or motivate why PCA or LMM should be used.

Reviewer #2 (Public Review):

Yao and Ochoa present a very nice paper examining the age-old question of whether LMM or PCA is a better way to adjust for structure (population, family, admixture). The authors provide a very nice and detailed overview of the previous research addressing this question, summarizing it in a table. They find that LMMs are generally better at accounting for population structure. However, I feel there are a couple of important factors that are missing. One is the consideration of environmental structure. Another is that the relationship between PCA and LMM is usually a bit more complicated in practice than depicted here, where the devil really lies in the details. Also, I think there are a couple of key reasons why LMMs haven’t been adapted as quickly as one might have expected, including case-control imbalance and cohort meta-analyses, which I feel the authors could point out. In fact, I believe LMMs have become sort of popular in recent years (e.g. Japan Biobank GWAS results).

We added environment simulations, which we agree was an important shortcoming of the previous version of our work.

We now discuss how the PCA and LMM connection can be more complicated in practice, but as the main difference is in how LD is handled, once that is correctly adjusted, PCs and random effects are still mostly modeling the same relatedness signals. Ultimately, our main conclusion is unchanged, namely that only LMMs can model family relatedness, which is their key advantage.

We briefly commented on case-control imbalance in our discussion (now made more clear), but since this involves binary traits, which we did not explicitly test in this work, it is out of scope.

Cohort meta-analysis does not influence whether to use PCA or LMM, since it can be performed with summary statistics from either model (and in theory even a combination of different models per cohort). The broad use of meta-analysis does not in itself prevent users from using PCA or LMM within individual cohorts. The use of meta-analysis is very interesting in its own right, but it is outside the scope of this work.

Reviewer #3 (Public Review):

This paper examines the relative performance of linear mixed models (LMMs), principal components (PCA), and their combination (PCA-LMM) for genetic association studies in human populations. The authors claim that previous papers examining this question are inadequate and that: (i) there remains confusion on which method is best and in which context, (ii) that the metrics used in previous evaluations were insufficient, and (iii) that the simulation settings used in previous papers were not comprehensive. To fix these problems the authors perform an extensive set of simulations within several frameworks and suggest two new metrics for evaluating performance.

Strengths:

The simulation framework used in this paper and the extensive number of simulations provide an opportunity to examine the relative properties of the three approaches (LMM, PCA, PCA-LMM) in a variety of contexts.

The parameters of the simulation framework are based on highly diverged populations, which is an increasingly common analysis choice that has not been examined in detail via simulation previously.

The evaluation metrics used in this paper are AUC and a test of the uniformity of the p-value distribution under the null. This is an improvement over some previous analyses which did not examine power and relied on less sensitive tests of type I error.

Weaknesses:

This paper has a limited set of population frameworks just like all papers before it. The breakdown of which method is best (LMM, PCA, PCA-LMM) will be a function of the simulation framework chosen.

Ameliorating this issue, we added additional simulations with low heritability and with environment effects. We are pleased to report that all of our conclusions hold at low heritability (h2 = 0.3), and for the most part under environment effects (which occasionally give LMM with PCs and PCA a small advantage, but often LMM with no PCs remains best, and we show PCs are no replacement for directly modeling these environment effects).

The frameworks chosen for this paper are certainly not comprehensive in contemporary human genetic studies. In fact, the authors make a number of unusual choices. For example, the populations in the simulated study have extremely large Fsts. While this is also a strength, the lack of more standard study designs is a weakness. More importantly, there is no simulation of family effects, which is the basis of many of the PCA-LMM papers reported in Table 1.

We now better motivate in the introduction our focus on association studies of multiethnic and admixed individuals, which are nowadays very common and which have greater FST values than earlier studies. In reference to higher simulated FSTs, we also now cite our recent work, which has found that many previous FST estimates are downwardly biased (Ochoa and Storey, 2021, 2019). We simulated data that was fit to each of our three real datasets using our unbiased methods, so those values that (understandably) appear high are actually more correct (for multiethnic populations such as those in 1000 Genomes, HGDP, etc) than previous estimates in the literature. In our previous work we also determined that only previous pairwise FST estimators are unbiased (under some conditions), and using a previous pairwise FST estimator (from Bhatia et al., 2013) we obtained equally high values between the most diverged human populations (values from a revised version of Ochoa and Storey, 2019 that isn’t on bioRxiv yet): In HGDP, the largest pairwise FST is 0.479, between Pima and PapuanSepik; In Human Origins, the largest estimate is 0.396, between Cabecar and Baining_Malasait; Lastly, in 1000 Genomes, the largest estimate is 0.135, between YRI and JPT. (1000 Genomes was generally less structured than HGDP and Human Origins, because the latter include more diverse populations.) Several previous estimates from the literature, all between one hunter-gatherer Sub-Saharan African subpopulation and one non-African subpopulation resulted in values of about 0.25 (Bowcock et al., 1991, Henn et al., 2011, Bergstrom et al., 2020). FST estimates are also greater from whole-genome sequencing versus array data (revised version of Ochoa and Storey, 2019).

Family (household) effects is a case where PCA is not expected to outperform LMM, though standard LMMs do not model this effect explicitly either and may not do much better. As this is a feature of family studies that ought to be absent in population studies (as usually only siblings are in the same household, and not more distant relatives), it is also not entirely relevant to the majority of our simulations. In these ways, including such a feature in our simulations does not align with the goals of this present work, but we agree this is an important framework that deserves more attention in future evaluations.

The discussion (and simulations) of LMM vs PCA, particularly LMMs with PCs as fixed effects misses the critical distinction of whether PCs are in-sample (in which case including PCs as fixed effects effectively serves as a preconditioner for the kinship matrix, speeding up iterative methods such as BOLT), or projections of individuals onto out-of-sample principal axes. There is also no discussion of LOO methods to address “proximal contamination”, also quite relevant in evaluating power as a function of the number of PCs.

We added the following to our discussion concerning out-of-sample PC projections: “We do not consider the case where samples are projected onto PCs estimated from an external sample (Prive et al., 2020), which is uncommon in association studies, and whose primary effect is shrinkage, so if all samples are projected then they are all equally affected and larger regression coefficients compensate for the shrinkage, although this will no longer be the case if only a portion of the sample is projected onto the PCs of the rest of the sample.”

We also added the following to the discussion concerning the LOCO approach: “Similarly, the leave-onechromosome-out (LOCO) approach for estimating kinship matrices for LMMs prevents the test locus and loci in LD with it from being modeled by the random effect as well, which is called”proximal contamination” (Lippert et al., 2011, Yang et al., 2014). While LOCO kinship estimates vary for each chromosome, they continue to model family relatedness, thus maintaining their key advantage over PCA.”

The same new discussion paragraph closes with the following thoughts concerning LOCO and related approaches: “LD effects must be adjusted for, if present, so in unfiltered data we advise the previous methods be applied. However, in this work, simulated genotypes do not have LD, and the real datasets were filtered to remove LD, so here there is no proximal contamination and LD confounding is minimized if present at all, so these evaluations may be considered the ideal situation where LD effects have been adjusted successfully, and in this setting LMM outperforms PCA. Overall, these alternative PCs or kinship matrices differ from their basic counterparts by either the extent to which LD influences the estimates (which may be a confounder in a small portion of the genome, by definition) or by sampling noise, neither of which are expected to change our key conclusion.”

Lastly, we added the following to a different discussion paragraph: “A different benefit for including PCs were recently reported for BOLT-LMM, which does not result in greater power but rather in reduced runtime, a property that may be specific to its use of scalable algorithms such as conjugate gradient and variational Bayes (Loh et al., 2018).”

There is no discussion/simulation of spatial/environmental effects or rare vs common PCs as raised in Zaidi et al 2020. There are some open questions here regarding relative performance the authors could have looked at. Same for LMMs with multiple GRMs corresponding to maf/ld bins and thresholded GRMs. For example, it would be helpful to know if multiple-GRM LMMs mitigate some of the problems raised in the Zaidi paper.

We added simulations with environment effects, which are based on a two-level hierarchy of population labels so they are spatial to the extent that these labels capture spatial relationships between populations. However, our small sample size data are not well suited to study rare variants and their structure, so its out of scope. (The sample size limitation is also covered in a new discussion paragraph.) We hope to tackle this very interesting question in future work.

We added the following paragraph to our discussion: “Another limitation of this work is ignoring rare variants, a necessity given our smaller sample sizes, where rare variant association is miscalibrated and underpowered. Using simulations mimicking the UK Biobank, recent work has found that rare variants can have a more pronounced structure than common variants, and that modeling this rare variant structure (with either PCA and LMM) may better model environment confounding, improve inflation in association studies, and ameliorate stratification in polygenic risk scores (Zaidi and Mathieson, 2020). Better modeling rare variants and their structure is a key next step in association studies.”

Author Response:

We thank the editors and reviewers for their assessment of our manuscript, and their agreement that we present compelling evidence for post-transcriptional regulation of AURKA through the 3’UTR.

In response to Reviewer 1, we acknowledge that much of our study is performed exclusively in U2OS cells, and that study of alternative polyadenylation in additional cell lines would serve to further generalize our findings. However, as U2OS are a well-known model cell line for cell cycle studies we believe our demonstration of cell cycle regulation of AURKA through its 3’UTR offers a depth of understanding that is perhaps of greater interest than confirming the existence of alternative AURKA 3’UTRs in additional cell lines, using our methods. We note that the recent rapid growth in RNA seq data resources allows easy confirmation of the broad existence of alternative polyadenylation events on a genome-wide scale. For example, AURKA-specific data extracted from a recent benchmark study of Nanopore long read RNA sequencing (Chen et al., 2021) clearly shows the existence of two distinct AURKA 3’UTRs differentially expressed between a number of different cancer cell lines. In addition, a recent study investigating the landscape of APA at single-cell resolution detected AURKA APA isoforms in HeLa and MDA-MB-468 cell lines (Wang et al., 2022). Their study further identifies AURKA among genes showing negative correlation between generalized distal polyA site usage index (gDPAU) and expression levels, meaning preference to use the proximal polyA site when expression levels increase, and include AURKA in the gene cluster showing slight increase in usage of the distal polyA site from G1 to M phase (Wang et al., 2022). Both studies are in support of the evidence presented in our manuscript.

We agree with Reviewer 2 that better information on translation rates would improve our understanding of the impact of translation regulation on AURKA levels. Some insight on the translation rate of AURKA in the cell cycle can be derived from inspection of the ribosome profiling dataset published by Tanenbaum et al., 2015. From their analysis, translation efficiency of AURKA mRNA in G2 is 1.59 times that in G1 and in G1 it is 0.69 times that in M phase, whilst in G2 it is 1.10 times higher than in M. Such data reveal a reversible increase in translation of AURKA mRNA, alongside other mitotic regulators, in preparation for M phase (Tanenbaum et al., 2015). These results are in accordance with our findings that translation rates contribute modestly to cell cycle changes in AURKA levels in normal cells, and we concur with Reviewer 3’s comment that the contribution of increased translation rate to AURKA levels at mitosis is less than the change in mRNA levels at this point in the cell cycle.

We think the significance of the regulatory mechanism we describe lies rather in the large effect it has on AURKA levels in interphase (when AURKA expression is normally repressed at both mRNA and translation rate). We hypothesise that it is interphase regulation that may be relevant to roles of AURKA in cancer (and to the association of APA with cancer) (Bertolin and Tramier, 2020; Naso et al., 2021). It is indeed the case that (i) AURKA regulation by miRNA, (ii) cooperation between APA and translation and (iii) cell-cycle dependent control of AURKA at the translation level, are already known. We believe the novelty of our study lies in drawing together these elements to provide new insight into AURKA regulation, using tools that allow similar investigation of other APA events, and contributing new ideas for future therapeutic interventions for disease proteins regulated via APA.

• Bertolin and Tramier (2020), Cell. Mol. Life Sci https://link.springer.com/article/10.1007/s00018-019-03310-2
• Chen et al. (2021), bioRxiv https://www.biorxiv.org/content/10.1101/2021.04.21.440736v1
• Naso et al. (2021), Oncogene https://www.nature.com/articles/s41388-021-01766-w
• Tanenbaum et al. (2015), eLife https://elifesciences.org/articles/07957
• Wang et al. (2022), PNAS https://www.pnas.org/doi/10.1073/pnas.2113504119

Author Response:

We would like to thank you for your thorough review of the manuscript. We will take all comments into account in the revised version of the manuscript. Please find below our provisional responses to your comments.

eLife assessment

This study reports useful information on the limits of the organotypic culture of neonatal mouse testes, which has been regarded as an experimental strategy that can be extended to humans in the clinical setting for the conservation and subsequent re-use of testicular tissue. The evidence that the culture of testicular fragments of 6.5-day-old mouse testes does not allow optimal differentiation of steroidogenic cells is compelling and would be useful to the scientific community in the field for further optimizations.

Thank you for this assessment. We will carefully consider all comments and make the requested revisions to improve the manuscript.

Public Reviews

Reviewer #1 (Public Review):

In this manuscript, the authors aimed to compare, from testis tissues at different ages from mice in vivo and after culture, multiple aspects of Leydig cells. These aspects included mRNA levels, proliferation, apoptosis, steroid levels, protein levels, etc. A lot of work was put into this manuscript in terms of experiments, systems, and approaches. However, as written the manuscript is incredibly difficult to follow. The Introduction and Results sections contain rather loosely organized lists of information that were altogether confusing. At the end of reading these sections, it was unclear what advance was provided by this work. The technical aspects of this work may be of interest to labs working on the specific topics of in vitro spermatogenesis for fertility preservation but fail to appeal to a broader readership. This may be best exemplified by the statements at the end of both the Abstract and Discussion which state that more work needs to be done to improve this system.

As explained below, we will rework and reorganize the manuscript to make it clearer, more meaningful and more precise. We believe that this work may be of interest to a broader readership. Indeed, the development of a model of in vitro spermatogenesis could be of interest for labs working on the specific period of puberty initiation, on germ and somatic cell maturation and on steroidogenesis during this period, and could even be useful for testing the toxicity of cancer therapies, drugs, chemicals and environmental agents (e.g. endocrine disruptors) on the developing testis.

Reviewer #2 (Public Review):

Preserving and restoring the fertility of prepubertal patients undergoing gonadotoxic treatments involves freezing testicular fragments and waking them up in a culture in the context of medically assisted procreation. This implies that spermatogenesis must be fully reproduced ex vivo. The parameters of this type of culture must be validated using non-human models. In this article, the authors make an extensive study of the quality of the organotypic culture of neonatal mouse testes, paying particular attention to the differentiation and endocrine function of Leydig cells. They show that fetal Leydig cells present at the start of culture fail to complete the differentiation process into adult Leydig cells, which has an impact on the nature of the steroids produced and even on the signaling of these hormones.

The authors make an extensive study of the different populations of Leydig cells which are supposed to succeed each other during the first month of life of the mouse to end up with a population of adult and fully functional cells. The authors combine quantitative in situ studies with more global analyzes (RT-QtPCR Western blot, hormonal assays), which range from gene to hormone. This study is well written and illustrated, the description of the methods is honest, the analyses systematic, and are accompanied by multiple relevant control conditions.

Since the aim of the study was to study Leydig cell differentiation in neonatal mouse testis cultures, the study is well conceived, the results answer the initial question and are not over-interpreted.

My main concern is to understand why the authors have undertaken so much work when they mention RNA extractions and western blot, that the necrotic central part had to be carefully removed. There is no information on how this parameter was considered for immunohistochemistry and steroid measurements. The authors describe the initial material as a quarter testis, but they don't mention the resulting size of the fragment. A brief review of the literature shows that if often the culture medium is crucial for the quality of the culture (and in particular the supplementations as discussed by the authors here), the size of the fragments is also a determining factor, especially for long cultures. The main limitation of the study is therefore that the authors cannot exclude that central necrosis can have harmful effects on the survival and/or the growth and/or the differentiation of the testis in culture. In this sense, the general interpretation that the authors make of their work is correct, the culture conditions are not optimized.

When using the organotypic culture system at a gas-liquid interphase, the central part of the testicular tissue becomes necrotic. As previously reported (Komeya et al., 2016), the central region receives insufficient nutrients and oxygen. In vitro spermatogenesis therefore only occurs in the seminiferous tubules present in the peripheral region. As in our previous publications and recent RNA-seq analyses (Dumont et al., 2023), the central necrotic area was removed so that transcript and protein levels in the healthy part of the samples (i.e. where in vitro spermatogenesis occurs) could be measured and compared with in vivo controls. For histochemical and immunohistochemical analyses, only seminiferous tubules located at the periphery of the cultured fragments (outside of the necrotic region) were analyzed. Steroid measurements were performed on the entire fragments.

The initial material was indeed a quarter testis, which represents approximately 0.75 mm3. No growth of the fragments was observed during the organotypic culture period. We agree with the reviewer that the composition of the culture medium is not the only parameter to be considered for the quality of the culture and that the size of the fragments is also a determining factor. We do not exclude that central necrosis can have harmful effects on the survival and/or the growth and/or the differentiation of the testis in culture. Optimization of the culture medium and culture design (so that the tissue center receives sufficient nutrients and oxygen) will be necessary to increase the yield of in vitro spermatogenesis.

Organotypic culture is currently trying to cross the doors of academic research laboratories to become a clinical tool, but it requires many adjustments and many quality controls. This study shows a perfect example of the pitfall often associated with this approach. The road is still long, but every piece of information is useful.

Reviewer #3 (Public Review):

Moutard, Laura, et al. investigated the gene expression and functional aspects of Leydig cells in a cryopreservation/long-term culture system. The authors found that critical genetic markers for Leydig cells were diminished when compared to the in-vivo testis. The testis also showed less androgen production and androgen responsiveness. Although they did not produce normal testosterone concentrations in basal media conditions, the cultured testis still remained highly responsive to gonadotrophin exposure, exhibiting a large increase in androgen production. Even after the hCG-dependent increase in testosterone, genetic markers of Leydig cells remained low, which means there is still a missing factor in the culture media that facilitates proper Leydig cell differentiation. Optimizing this testis culture protocol to help maintain proper Leydig cell differentiation could be useful for future human testis biopsy cultures, which will help preserve fertility and child cancer patients.

Methods: In line 226, there is mention that the central necrotic area was carefully removed before RNA extraction. This is particularly problematic for the inference of these results, especially for the RT-qPCR data. Was the central necrotic area consistent between all samples and variables (16 and 30FT)? How big was the area? This makes the in-vivo testis not a proper control for all comparisons. Leydig cells are not evenly distributed throughout the testis. A lot of Leydig cells can be found toward the center of the gonad, so the results might be driven by the loss of this region of the testis.

When using the organotypic culture system at a gas-liquid interphase, the central part of the testicular tissue becomes necrotic. As previously reported (Komeya et al., 2016), the central region receives insufficient nutrients and oxygen. In vitro spermatogenesis therefore only occurs in the seminiferous tubules present in the peripheral region. As in our previous publications and recent RNA-seq analyses (Dumont et al., 2023), the central necrotic area was removed so that transcript levels in the healthy part of the samples (i.e. where in vitro spermatogenesis occurs) could be measured and compared with in vivo controls. The transcript levels of the selected genes were of course normalized to housekeeping genes (Gapdh and Actb) or to the Leydig cell-specific gene Hsd3b1.

The central necrotic area was consistent between all samples and variables: it represents on average 16-27% of the explants.

Moreover, we would like to point out that the gonads were cut into four fragments before in vitro cultures. It is therefore the central part of these explants that was removed and not the central part of the gonads. The central part of the gonads was thus included in our analyses.

What did the morphology of the testis look like after culturing for 16 and 30 days? These images will help confirm that the culturing method is like the Nature paper Sato et al. 2011 and also give a sense of how big the necrotic region was and how it varied with culturing time.

This point will be addressed in the detailed responses to reviewers.

There are multiple comparisons being made. Bonferroni corrections on p-value should be done.

This point will be addressed in the detailed responses to reviewers.

Results: In the discussion, it is mentioned that IGF1 may be a missing factor in the media that could help Leydig cell differentiation. Have the authors tried this experiment? Improving this existing culturing method will be highly valuable.

The decreased Igf1 mRNA levels found in the present study are in line with the RNA-seq data of Yao et al., 2017. As mentioned in the Discussion section, the addition of IGF1 in the culture medium led to a modest increase in the percentages of round and elongated spermatids in cultured mouse testicular fragments (Yao et al., 2017). However, the effect of IGF1 supplementation on Leydig cell differentiation was not investigated. The supplementation of organotypic culture medium with IGF1 is currently being tested in our research team.

Add p-values and SEM for qPCR data. This was done for hormones, should be the same way for other results.

p-values and SEM are shown for both qPCR and hormone data.

Regarding all RT-qPCR data-There is a switch between 3bHSD and Actb/Gapdh as housekeeping genes. There does not seem to be as some have 3bHSD and others do not. Why do Igf1 and Dhh not use 3bHSD for housekeeping? If this is the method to be used, then 3bHSD should be used as housekeeping for the protein data, instead of ACTB. Also, based on Figure 1B and Figure 2A (Hsd3b1) there does not seem to be a strong correlation between Leydig cell # and the gene expression of Hsd3b1. If Hsd3b1 is to be used as a housekeeper and a proxy for Leydig cell number a correlation between these two measurements is necessary. If there is no correlation a housekeeping gene that is stable among all samples should be used. Sorting Leydig cells and then conducting qPCR would be optimal for these experiments.

Hsd3b1 was used as a housekeeping gene only to normalize the mRNA levels of Leydig cell-specific genes. Therefore, Igf1 and Dhh transcript levels were not normalized with Hsd3b1 since Igf1 is expressed by several cell types in the testis (Leydig cells, Sertoli cells, peritubular myoid cells) and Dhh is expressed by Sertoli cells.  

Regarding western blots, the expression of AR, CYP19 and FAAH could not be normalized with 3bHSD since AR is expressed by Leydig cells, Sertoli cells and peritubular myoid cells, CYP19 is expressed by Leydig cells and germ cells and FAAH is expressed by Sertoli cells. We will review the western blot results for CYP17A1.

As shown in Figure 1B, the number of Leydig cells per cm2 of testicular tissue is not significantly different between the different time points in vivo (6 d_pp_, 22 d_pp_ and 36 d_pp_), in vitro (D16 FT and D30 FT) and between the in vivo and in vitro conditions (22 d_pp_ versus D16 FT, 36 d_pp_ versus D30 FT). Similarly, our data in Figure 2A show that Hsd3b1 mRNA levels are not significantly different between the different time points in vivo (6 d_pp_, 22 d_pp_ and 36 d_pp_), in vitro (D16 FT and D30 FT) and between 22 d_pp_ and D16 FT. However, Hsd3b1_mRNA levels were significantly lower in D30 FT tissues compared to 36 d_pp. We will measure the correlation between the number of Leydig cells per cm2 of testicular tissue and Hsd3b1 mRNA levels, as suggested by the reviewer.

Figure 2A (CYP17a1): It is surprising that the CYP17a1 gene and protein expression is very different between D30FT and 36.5dpp, however, the immunostaining looks identical between all groups. Why is this? A lower magnification image of the testis might make it easier to see the differences in Cyp17a1 expression. Leydig cells commonly have autofluorescence and need a background quencher (TrueBlack) to visualize the true signal in Leydig cells. This might reveal the true differences in Cyp17a1.

This point will be addressed in the detailed responses to reviewers.

Figure 3D: there are large differences in estradiol concentration in the testis. Could it be that the testis is becoming more female-like? Leydig and Sertoli cells with more granulosa and theca cell features? Were any female markers investigated?

We show in the present study that the expression level of the Sertoli cell-specific gene Dhh is not reduced in organotypic cultures. We also previously found that the expression level of the Sertoli-cell specific gene Amh was not reduced in in vitro matured testicular tissues (Rondanino et al., 2017). Moreover, our recent unpublished data show that Sox9, a testis-specific transcription factor, is expressed in Sertoli cells in organotypic cultures. These results suggest that Sertoli cells are not becoming granulosa-like cells and that the testis is not becoming more female-like. Markers of granulosa and theca cells were not investigated.

Figure 3D and Figure 5A: It is hard to imagine that intratesticular estradiol is maintained for 16-30 days without sufficient CYP19 activity or substrate (testosterone). 6.5 dpp was the last day with abundant CYP19 expression, so is most of the estrogen synthesized on this first day and it sticks around? Are there differences in estradiol metabolizing enzymes? Is there an alternative mechanism for E production?

This point will be addressed in the detailed responses to reviewers.

Author Response

Reviewer #2 (Public Review):

In a neonatal model of bacterial meningitis induced by s.c. injection of E. coli, transcriptional changes were found across all major cell types including endothelial cells, fibroblasts and macrophages. Among macrophages, they describe 2 resident subsets and 2 inflammatory subsets. By immunohistochemistry of arachnoid and dura flatmounts, they show vascular changes upon infection, including clustering of CLDN5 and PECAM1, and disorganized capillary morphology, which was dependent on Tlr4 signaling but independent of arachnoid macrophages.

The manuscript would benefit from rewriting, it is not written in a concise manner and the rationale for experiments, time points for analyses and their conclusions are not clear. The model of s.c. bacterial infection is not well introduced and overall changes in the periphery, survival curves or bacterial counts (in the KO models) in the meninges/brain are not mentioned.

Thank you for those comments. We hope that the text is now more readable. We have added a separate section to describe the meninges model and added data on survival and E coli counts (Supplemental Figure 3).

Author Response

Reviewer #1 (Public Review):

This work puts forward a comprehensive characterisation of colorectal cancer (CCCRC), by classifying it into 4 subtypes with distinct TME features. It uses 10 public databases: 8 microarray datasets for the training of molecular classification and 2 RNAseq for validation (CRC-RNAseq) to identify the 4 subtypes using unsupervised machine learning (consensus clustering). These 4 subtypes were found to be somewhat distinct in terms of immune response and the possibilities for effective treatments. They found that one subtype may be more sensitive to chemotherapy, two to WNT pathway inhibitor SB216763 and Hedgehog pathway inhibitor vismodegib, and one to ICB treatment. They show an association with patient outcome in terms of PFS, validated in the validation cohort. They used histology to correspond the subtypes to known pathological types, as well as investigating their T cell makeup. They also investigated the genetic tumour evolution that may occur between the subtypes. A single-sample gene classifier was put forward as a way of identifying the class of cancer. The evidence for the main results of the work is convincing, but a few areas need to be clarified and extended.

In the determination of the 4 subtypes (C1-C4) the methodology is clear, and the definition of the training and validation data are clear and well presented. The techniques used are well suited to the problem. The performance of the classification as a predictor of prognosis is presented as KM curves of PFS and OS for the training and validation sets. The training data shows a significant log-rank p-value in both PFS and OS. The validation data shows a significant effect in PFS.

What follows is quite an exhaustive process of finding differences between the cohorts using a multitude of techniques and datasets, including genomics, epigenetics, transcriptomics, and proteomics. These sections are mainly descriptive and do add understanding to the classification, especially with regard to the T-cell populations that are invasive.

Improvements could be made to the latter sections of the main paper. The basis for the potential clinical responses of the subtypes is arrived at via a "pre-clinical model" based on 81 genes. It would benefit from clarification on what genes were used in model training and details of the final model. Similarly the description of the "Single-sample gene classifier" could be enhanced similarly with a better description of which genes are in the final classifier.

Thank you for taking the time to review our article and for your positive feedback. Your thorough evaluation of our work has been invaluable to us, and we appreciate your recognition of the effort we put into it.

1) The basis for the potential clinical responses of the subtypes is arrived at via a "pre-clinical model" based on 81 genes.

The exact details of the filtering criteria used to obtain the list of pre-clinical model genes have added to the Methods section of the study (Lines 1061-108, Lines 503-511) (Supplementary file 3a). To explore the treatment for each CCCRC subtype using cancer cell line drug-sensitivity experiments, we developed a pre-clinical model based on subtype-specific, cancer cell-intrinsic gene markers according to a previously published study (Eide et al., 2017). Firstly, the “limma” package was used to identify DEGs with FDR < 0.05 between each of the four subtypes and the remaining subtypes in the CRC-AFFY cohort. To identify subtype-specific genes in one of the subtypes, we excluded those that were found to be differentially expressed in comparisons between one of the other subtypes and the remaining subtypes. The upregulated subtype-specific genes (log2 (fold change, FC) > 0 and FDR < 0.05) was ranked based on their log2FC and selected the top 500 genes for further gene screening. Secondly, the GEP of human CRC tissues versus patient-derived xenografts (PDX) in the GSE35144 dataset by the R package “limma” was used to remove those genes associated with stromal and immune components. DEGs with FDR > 0.5 and log2 FC < 1 between human CRC tissues versus PDX were considered as cancer cell-intrinsic genes. Thirdly, we also utilized human CRC cell lines to obtained cancer cell-intrinsic genes. A total of 71 human CRC cell lines with RNAseq data (log2TPM) was obtained from the Genomics of Drug Sensitivity in Cancer (GDSC) database (https://depmap.org/portal/download/all/), 43 of which had dose-response curve (area under the curve, AUC) values. The MSI status, FGA and TMB information of CRC cell lines was obtained from cbioportal website (https://www.cbioportal.org/study/summary?id=ccle_broad_2019). RNAseq data for 71 human CRC cell lines was used to further determine the cancer cell-intrinsic genes and genes among the top 25% within (i) the 10−90 % percentile range of the largest expression values and (ii) the highest expression in at least three samples. The subtype-specific genes and cancer cell-intrinsic genes were intersected to generate the gene list for developing the pre-clinical model. The pre-clinical model was developed using the nearest template prediction (NTP) function of R package “CMScaller”, which can be applied to cross-tissues and cross-platform predictions (Hoshida, 2010). The GEP (log2TPM) of 71 human CRC cell lines normalized by the Z-score were input into the pre-clinical model, and the cell lines were divided into four CCCRC subtypes. (Lines 1061-1088)

Here we want to make a point that we changed from using the xgboost algorithm to using the NTP algorithm to build our pre-clinical model. Based on the genomic features of the cell line, we evaluated the reliability of the final pre-clinical model and found that the pre-clinical model built using the NTP algorithm is more reliable. As expected, the C4 subtype cell lines demonstrated the highest TMB values and MSI frequency while exhibiting the lowest FGA scores when compared to other subtypes (Figure 6-figure supplement 1G-I). In contrast, C1 and C3 subtype cell lines showed significantly higher FGA scores and significantly lower TMB values and MSI frequency. The C2 subtype cell lines had median FGA scores, TMB values, and MSI frequency. The pre-clinical model is publicly available at https://github.com/XiangkunWu/pre_clinical_model. (Lines 503-511)

2) Similarly the description of the "Single-sample gene classifier" could be enhanced similarly with a better description of which genes are in the final classifier.

We apologize for any confusion caused in our revised regarding the derivation of the CCCRC classifier. Specifically, we have added more details on the derivation of model genes and the establishment of the model, and ensured the availability of the CCCRC classifier. The method details and results of deriving the model genes and building the model are described next. (Lines 1102-1121) (Lines 562-579) (Supplementary file 3c)

In order to facilitate the widespread application of CCCRC classification system, we established a simple gene classifier to predict CCCRC subtypes. Firstly, we filtered genes based on their mean expression and variance in the CRC-AFFY cohort, and genes with expression and variance below the bottom 25% were removed. Then, we applied the Random Forest algorithm (RF) in the R package "caret" to perform feature selection on the CCCRC subtype-specific genes of the CRC-AFFY cohort. The top 20 most informative features for each subtype were ranked and selected based on the impurity measure generated by the algorithm. This allowed us to identify critical genes that are strongly associated with each CCCRC subtype and develop the CCCRC classifier. Next, we randomly divided the CRC-AFFY cohort into training and validation sets at a ratio of 7:3 using “createDataPartition” function provided in the R package "caret" (seed=123). The GEP was normalized with Z-scores prior to model training and validation. The CCCRC classifiers were trained with the top 80 subtype-specific genes using the RF, Support Vector Machine (SVM), eXtreme Gradient Boosting (xgboost), and Logistic Regression algorithms implemented in the R package "caret". Finally, we validated the CCCRC classifier on the GSE14333 and GSE17536 datasets, as well as the CRC-AFFY cohort. We evaluated the predictive performance of the CCCRC classifier by evaluating measures such as accuracy value and F1 score, which were generated using the " confusionMatrix " function provided in the R package "caret". (Lines 1102-1121)

We established the CCCRC classifier on the training set by utilizing multiple machine learning algorithms based on the GEP of 80 upregulated subtype-specific genes (Supplementary file 3c). Upon application to the test set, GSE14333, and GSE17536 datasets, the performance of the eXtreme Gradient Boosting (xgboost) algorithm was the best with the highest accuracy values and F1 scores compared to the Random Forest (RF), Support Vector Machine (SVM), and Logistic Regression algorithms (Figure 6-figure supplement 4). Notably, the CCCRC classifier based on the xgboost algorithm displayed robust performance across gene expression platforms, Affymetrix and RNA-sequencing platforms, exhibiting a balanced accuracy of > 80% for all subtypes (Supplementary file 3d). These findings demonstrated the stability and cross-platform applicability of our classifier. The CCCRC classifier based on the xgboost algorithm is publicly available at https://github.com/XiangkunWu/CCCRC_classifier, and the CCCRC subtype information of CRC patients can be obtained by directly inputting the GEP of 80 upregulated subtype-specific mRNA genes. The CCCRC classifier might facilitate the discovery of new biomarkers and the personalized treatment of clinical patients with CRC. (Lines 562-579)

Reviewer #2 (Public Review):

This study aimed to classify colorectal cancer (CRC) samples based on the expression of genes in selected gene lists, where the gene lists were chosen to represent aspects of the tumour microenvironment, tumour-associated immune cells, and tumour cells. The resulting clusters were then used to define a classifier, followed by a detailed description of molecular features of the tumours and tumour microenvironments assigned to each cluster. The authors claim this study is more "holistic" than previous work on CRC clustering/classifiers because they aimed to explicitly include additional components of the tumour microenvironment in both the clustering/classifier definition and in the subsequent description of molecular characteristics.

The CCCRC clustering and the resulting classifier presented in this paper are derived from published RNAseq studies. The multi-omics aspect of the work is restricted to smaller sample numbers for which both transcriptomic and another omics dataset were available in public resources and comprises a description or correlative analysis of each omics data type within each of the assigned CCCRC subtypes.

By applying solid computational methods to a compendium of published RNAseq datasets (n~1500 tumours), they found that tumour samples from colorectal cancers clustered into 4 subtypes ("CCCRC" subtypes) on the basis of 61 pre-defined gene expression signatures. These subtypes correlated with but did not correspond to, the previously described Consensus Molecular Subtypes (CMS) of colorectal tumours.

Other types of molecular data were available for some tumours, obtained from the same published resources: whole-slide images, mutations, tumour proteomics, and/or scRNAseq. The authors reanalysed these datasets using standard methods and drew correlations with the CCCRC subtypes they had assigned in this work. To (semi-)quantify immune infiltration characteristics from whole-slide images (WSI), they additionally performed automated segmentation in addition to review by pathologists, which in combination produced a convincing WSI-derived dataset.

In combination with existing CRC classifications, this study could facilitate future biomarker discoveries. This appears to be the authors' main claim, and the data and methods broadly support this claim.

Thank you for taking the time to review our article and for your positive feedback. Your thorough evaluation of our work has been invaluable to us, and we appreciate your recognition of the effort we put into it.

Some aspects of the work need to be clarified: 1) This work relies on the definition of 4 clusters of CRC tumours based on consensus clustering of the 61 gene lists, which in turn depends on the choice of clustering method and the choice of gene lists. Sufficient detail is provided about the gene lists and resulting clusters, but this paper does not show how robust the 4 clusters are to these choices; for example, the "Energy" gene list appears to be a relatively strong component of clusters C2 and C3.

Thank you very much for providing such detailed and insightful feedback.

1.1. The reviewer has raised a valid concern about the impact of gene list selection on the robustness of the clusters. To address this issue, we used the “pamr.predict” function of the R package “pamr” (Tibshirani et al., 2002) to extract centroids of each subtype that best represent each subtype and establish a PAMR classifier. PAM (Prediction Analysis of Microarrays) is a statistical technique to identify subsets of features that best characterize each class using nearest shrunken centroids (Tibshirani et al., 2002). The technique is general and can be used in many other classification problems. As shown in Figure 1-figure supplement 2E, a threshold of 0.566 with minimum 10-fold cross-validation error was selected to identify the 61 TME-related signatures that exhibit at least one non-zero difference between each subtype (seed = 11). These signatures were then used to construct a PAMR classifier with superior predictive capability, exhibiting an overall error rate of 15%. We used the established PAMR classifier to predict the CCCRC subtypes on the CRC-RNAseq cohort and the same four CCCRC subtypes were revealed, with similar patterns of differences in the TME components (Fig. S2F, G). This indicated that the 61 TME-related signatures best represent each subtype and are indispensable for achieving the identification of the four CCCRC subtypes. (Lines 161-168)

1.2. The reviewer has raised a valid concern about the impact of the clustering method selection on the robustness of the clusters.

We performed extensive data analysis attempts during our unsupervised clustering analysis, which primarily involved attempting various clustering methods, including K-means clustering, non-negative matrix factorization (NMF) clustering, and hierarchical clustering, as well as replacing different sources and categories of the TME-related signatures. To determine the optimal clustering method and TME panel, we evaluated whether the TME panel could reproduce the heterogeneity of TME, the stability of the clustering itself, the biological characteristics of the subtypes, the correlation between subtypes and prognosis, and the correlation between subtypes and microsatellite instability (MSI), consensus molecular subtypes (CMS) classification system, and other molecular subtype systems. Due to the abundance of exploratory data analysis results, we ultimately selected the best clustering method and TME panel combination for showcase.

1.3. Also, we analyzed the sensitivity analysis of the effect of TME-related signatures on the clustering results. Since the effect of removing one of the TME-related signatures on the clustering results was not well evaluated, we attempted to remove the entire category. We performed consensus clustering analysis again using the same parameters (partitioning around medoids (pam) clustering; "Pearson" distance; 1,000 iterations; from 2-6 clusters). When we conducted consensus clustering analysis using only immune-related signatures, we identified three subtypes: low (C2), moderate (C3), and high (C1) immune infiltration subtypes. When we included both immune-related and tumor-related signatures, we identified four subtypes: immunomodulatory (S1), cold (S2/S3), and immune-excluded (S4) subtypes. It appears that the immunosuppressed subtype in the CCCRC classification system may have been assigned to both S1 and S4 subtypes. Limiting the consensus clustering analysis to only immune-related or immune- and stroma-related signatures, as done in previous studies (Bagaev et al., 2021; He et al., 2018), did not allow reliable identification of all four CCCRC subtypes. These sensitivity analyses underscored the necessity of our well-designed TME panel to achieve the identification of the four CCCRC subtypes. (Lines 172-176) (Figure 1-figure supplement 4)

2) The authors examined whether their CCCRC classification showed differential disease progression in available retrospective cohorts of people treated with anti-PDL1 therapy. The authors presented this work as "significance of CCCRC in guiding the clinical treatment of colorectal cancer", but the data presented in this section cannot support clinical treatment decisions, which would require prospective studies and clinical trial designs. However, this section is potentially useful for generating hypotheses about potential biomarkers related to the CCCRC subtypes, and might, in the future with additional evidence, contribute to the design of a trial. The authors point out that additional experimental evidence would be required.

Thank you for your constructive suggestions. We agree that our retrospective analysis of the CCCRC classification in relation to disease progression under immune checkpoint blockade treatment does not directly support clinical treatment decisions. We acknowledge that additional experimental evidence would be required to fully support the use of the CCCRC classification as a clinical tool for guiding treatment decisions. We have highlighted in the corresponding section of the article that this research is pre-clinical and still requires substantial basic experiments and clinical trials to validate. (Lines 536, 751)

3) Other prognostic or predictive clinicopathological variables for colorectal cancer are not discussed in detail in the present work but are important for further work on the prognostic and predictive value of CRC molecular subtypes and biomarker derivation. Discrepancies in treatment response have previously been observed in separate CRC trials of biologically targeted agents with different chemotherapy backbones and other authors have suggested that treatment interactions with the tumour microenvironment might in part explain these discrepancies (e.g. Aderka (2019) PMID:31044725, and others).

3.1) Other prognostic or predictive clinicopathological variables for colorectal cancer are not discussed in detail in the present work but are important for further work on the prognostic and predictive value of CRC molecular subtypes and biomarker derivation.

Thank you for bringing up this point. We apologize for not analyzing other clinicopathological variables for colorectal cancer in more detail in my original work. We agree that these variables are important for further study of our CCCRC classification system to guide biomarker derivation and to guide clinical treatment decisions. We added in the article the relationship between CCCRC subtypes and clinicopathological variables, as well as the comparison with CMS subtypes (Lines 256-262, 661-666). In addition, we have identified a clerical error in our manuscript and have corrected it accordingly. Specifically, the use of PFS as the endpoint in some parts of the manuscript was a mistake and has been corrected to DFS. We would like to clarify that the endpoint for the CRC-AFFY and CRC-RNAseq cohorts is DFS and OS, while the endpoint for the GSE104645 dataset is PFS and OS. For the immune checkpoint blockade therapy cohort, the endpoint for PRJEB23709 (Gide) is PFS and OS, and for the GSE135222 (Jung) dataset, the endpoint is PFS. Progression Free Survival (PFS) refers to the time from randomization (or treatment initiation) to the first occurrence of disease progression or death from any cause. The definition of Disease-Free Survival (DFS) is the time from randomization to the appearance of evidence of disease recurrence.

We further analyzed the association of CCCRC subtypes with clinicopathological characteristics (Supplementary file 1f, Supplementary file 1g). We found that the C4 subtype was mostly diagnosed in right-sided CRC lesions and in females, which was consistent with the CMS1 subtype. The C1 and C3 subtypes were mainly observed in left-sided CRC lesions and in males, consistent with the CMS2 and CMS4 subtypes. The C3 subtype was strongly associated with more advanced tumor stages, which was the similarity to the CMS4 subtype, while the C4 subtype was associated with higher histopathologic grade, which was the similarity to the CMS1 subtype. Furthermore, our analysis using the Kaplan-Meier method demonstrated that patients with the C4 subtype had significantly higher disease-free survival (DFS) and overall survival (OS) compared to those with the C2 and C3 subtypes in the CRC-AFFY (Figure 1I, Figure 1-figure supplement 7A) and CRC-RNAseq cohorts (Figure 1-figure supplement 7B, C). Multivariate Cox proportional hazard regression analysis showed that the C4 subtype was an independent predictor of the best OS and DFS, whereas the C3 subtype was an independent predictor of the worst OS and DFS after adjustment for age, gender, tumor site, TNM stage, grade, adjuvant chemotherapy or not, MSI status, BRAF and KRAS mutations, and the CMS classification system in the combined cohort (the CRC-AFFY and CRC-RNAseq cohorts) (Supplementary file 1h). Considering that the C1, C2/C3, and C4 subtypes partially overlap with the CMS2, CMS4, and CMS1 subtypes, respectively, we also analyzed the prognostic differences between them in the combined cohort. We found that the DFS/OS of patients with the C1 subtype was worse than those with the CMS2 subtype (Figure 1-figure supplement 7D, E), the DFS/OS of patients with the C2 subtype was better than those with the CMS4 subtype (Figure 1-figure supplement 7F, G), the DFS/OS of patients with the C3 subtype was not significantly different from those with the CMS4 subtype (Figure 1-figure supplement 7F, G), and the DFS/OS of patients with the C4 subtype was significantly better than those with the CMS1 subtype (Figure 1-figure supplement 7H, I). Notably, the C2 subtype within the CMS4 subtype also had a better prognosis than the C3 subtype within the CMS4 subtype (Figure 1-figure supplement 7J, K). The above analysis demonstrated that the CCCRC classification system were closely associated with clinicopathological characteristics, were able to refine the CMS classification system and MSI status, as well as contributed to the understanding of the mechanisms underlying the different clinical phenotypes resulting from TME heterogeneity.

3.2) Discrepancies in treatment response have previously been observed in separate CRC trials of biologically targeted agents with different chemotherapy backbones and other authors have suggested that treatment interactions with the tumour microenvironment might in part explain these discrepancies (e.g. Aderka (2019) PMID:31044725, and others).

The reviewer's comments greatly contributed to the quality of our study. Aderka et al. discussed the reasons for the differences in the results of the CALGB/SWOG 80405 and FIRE-3 clinical trials, which may be related to differences in the chemotherapy backbone used and TME heterogeneity (Aderka et al., 2019). Both trials evaluated the combination of cetuximab or bevacizumab with a different chemotherapy backbone: in the CALGB/SWOG 80405 trial, 75% of patients received oxaliplatin, while in the FIRE-3 trial, all patients received irinotecan. The CCCRC classification system also facilitates the understanding of the differences in the results of the CALGB/SWOG 80405 and FIRE-3 clinical trials (Heinemann et al., 2014; Lenz et al., 2019). We have added this content to the discussion section of the article (Lines 753-777). Based on our examination of the results summarized in Figure 4 of the work by Aderka et al. (Aderka et al., 2019), we found that differences in the treatment outcomes of the CMS1 and CMS4 subtypes were the crucial factor behind the divergent results observed in the two clinical trials. The CMS1 and CMS4 subtypes have a microenvironment rich in CAFs. Our CCCRC classification results also showed that CMS1, in addition to mainly consisting of the C4 subtype, also contains a considerable number of the C2 subtype, while the CMS4 subtype mainly consists of the C2 and C3 subtypes. Furthermore, our study results indicated that the C2 subtype is suitable for chemotherapy in combination with bevacizumab, possibly because the combination can inhibit the CAFs and abnormal blood vessel formation in the microenvironment, thus alleviating the immune suppression of the immune cells. However, the C3 subtype is not suitable for chemotherapy in combination with bevacizumab because it only accumulates CAFs and abnormal blood vessel formation but lacks T cell infiltration. Therefore, we boldly speculate that the CMS1 and CMS4 subtypes in the CALGB/SWOG 80405 clinical trial may contain more C2 subtypes than those in the FIRE-3 clinical trial, leading to the CMS1 and CMS4 subtypes in the CALGB/SWOG 80405 clinical trial being more suitable for chemotherapy in combination with bevacizumab than cetuximab compared to the FIRE-3 clinical trial. Overall, the integration of CCCRC and CMS classification systems provides valuable insights for understanding the divergent outcomes of the two clinical trials (Lines 753-777).

Reviewer #3 (Public Review):

In their study: Comprehensive characterization of tumor microenvironment in colorectal cancer via histopathology-molecular analysis, Wu et al., aim to examine the contribution of the tumour microenvironment (TME) on biological and clinical heterogeneity in colorectal cancer (CRC).

To achieve this the authors use a vast array of publicly available datasets across a variety of biological modalities (transcriptomic, epigenetic, mutational). Using thoughtfully curated genesets the authors classify CRC into 4 holistic comprehensive characterised CRC (CCCRC) subtypes which comprise immune, stromal, and tumour features of CRC biology.

The authors investigate the association of their novel CCCRC subtypes with current "gold standard" classification schemes.

The authors' integration of deep learning methods for HE classification and subsequent association with "Tumor level" CCCRC subtypes is a refreshing addition to the study. Comment on the degree of heterogeneity observed in HE samples and correlation to the heterogeneity of CCCRC subtypes would be a welcomed addition. It is likely publicly available datasets from such platforms as 10X Genomic Visium would be available for this type of analysis.

Whilst one of the main outcomes of the study is the addition of another classification scheme to the field of colorectal cancer, the CCCRC scheme represents a holistic perspective on CRC classification.

The authors provide a welcomed graphical overview of the complex narrative of the study in Figure 7.

The authors focus on the classification of inter-patient heterogeneity and its associated predictive and prognostic utility. There appears to be a significant degree of overlap between immunosuppressive and immune excluded, and proliferative and immuno-modulatory signatures in Figure 1A. One of the major limitations of patient response to treatment is intra-patient heterogeneity, it would be nice for the authors to comment briefly on the degree of intra-patient heterogeneity of the CCCRC subtypes.

Overall the authors succeed in providing a holistic deep characterization of CRC from the perspective of a variety of biological modalities. The authors provide a novel classification scheme for the field of CRC which demonstrates prognostic and predictive utility, which would benefit from further validation from external datasets. The authors demonstrate a pathway for integration and interpretation of complex high-dimensional data into clinically translatable currency such as the H&E.

Thank you for taking the time to review our article and for your positive feedback. Your thorough evaluation of our work has been invaluable to us, and we appreciate your recognition of the effort we put into it.

1) Comment on the degree of intra-patient heterogeneity of CCCRC subtypes would be nice.

We have added intra-tumor heterogeneity analysis for each subtype (Lines 196-198). The level of intratumor heterogeneity (ITH) was significantly linked to poor prognosis and drug resistance (Caswell and Swanton, 2017). The ITH data used in our study for the CRC-RNAseq cohort was obtained from a previous study conducted by Thorsson et al. (Thorsson et al., 2018). As expected, the ITH of the C2 and C3 subtypes was higher than that of the other subtypes, while the ITH of the C4 subtype was the lowest (Figure 1F). Our analysis using the Kaplan-Meier method demonstrated that patients with the C4 subtype had significantly higher overall survival (OS) and disease-free survival (DFS) compared to those with the C2 and C3 subtypes. Furthermore, the C3 subtype was resistant to chemotherapy, cetuximab, bevacizumab, and ICB therapy. Our investigation of drug sensitivity data of cell lines also indicated that the C2 and C3 subtypes were generally not responsive to most drugs.

2) A significant degree of overlap between immunosuppressive and immune excluded, and proliferative and immuno-modulatory signatures in Figure 1A is apparent and should be commented upon.

Our research revealed that both C2 and C3 subtypes exhibited a high level of tumor stroma, while C1 and C4 subtypes were characterized by active DNA damage and repair and high tumor proliferation. Additionally, C2 and C4 subtypes had an abundance of immune components. This was consistent with our finding that there may be interconversion between the C1 and C4 subtypes, between the C4 and C2 subtypes, and between the C2 and C3 subtypes in this evolutionary pattern. The interconversion between C2 and C4 subtypes in this evolutionary pattern was the rarest situation, indicating that once the tumor enters the C2 subtype, it is difficult to reverse and will progress to the C3 subtype. (Lines 637-644)

3) It is likely publicly available datasets from such platforms as 10X Genomic Visium would be available for this type of analysis.

To investigate the spatial distribution relationship between four CCCRC subtypes of tumor cells, T cells, and stromal cells, we conducted a re-analysis of publicly available CRC spatial transcriptomics data (ST) obtained from the 10X website (https://www.10xgenomics.com/resources/datasets). The Space Ranger output files were then processed with Seurat (V4.1.1) (Hao et al., 2021) using SCTransform for normalization (Hafemeister and Satija, 2019). RunPCA were used to dimension reduction and RunUMAP to visualize the data. We used “ssGSEA” method implemented in the R package “GSVA” to score the six cell types (C1-C4 subtype cancer cells, mesenchymal cells, and T cells) (Hänzelmann et al., 2013). The “ssGSEA” method has been previously demonstrated to be highly reliable and suitable for ST data analysis (Wu et al., 2022). The cell-type-rich region was defined as the ssGSEA score of each cell type from one spot larger than the 75% quantile of this cell type. The markers for the six cell types are listed in the Supplementary file 1a and Supplementary file 3a. (Lines 1090-1102)

The Cytassist and Visium samples had a total of 9080 and 2660 spots, respectively. We used “ssGSEA” method to quantify the six cell subpopulations of each spot and also visualized only the spots corresponding to the top 25% of the score ranking for each cell type (Figure 6-figure supplement 2AB, Figure 6-figure supplement 3AB). In Cytassist samples, we observed different spatial distribution patterns of the four subtypes of tumor cells (Figure 6-figure supplement 2B). Specifically, the C3 subtype of tumor cells was predominantly located in the tumor periphery with an enrichment of mesenchymal cells and T cells (areas selected by black dashed circles). In contrast, the C4 subtype of tumor cells was mainly present in the center of the tumor, accompanied by the presence of T cells. The C1 and C2 subtypes of tumor cells were distributed in relatively uniform areas, mainly in the tumor periphery, with fewer mesenchymal cells and T cells. However, the distribution areas of C2 subtype and C3 subtype of tumor cells also partially were in overlap (the area selected by red dashed circles). The same distribution patterns can also be observed in the Visium sample (Figure 6-figure supplement 3B). Further analysis of the correlation between the ssGSEA scores of each cell type in the cell-type-rich regions and those of other cell types was conducted (Figure 6-figure supplement 2D, E, Figure 6-figure supplement 3D, E). We found that in the C3 subtype-rich region of tumor cells, the C3 subtype score of tumor cells was significantly positively correlated with the mesenchymal cell score, while in the T cell-rich region, the C3 subtype score of tumor cells was significantly negatively correlated with the T cell score. The C4 subtype score of tumor cells was significantly positively correlated with the T cell score and negatively correlated with the mesenchymal cell score in the C4 subtype-rich, T cell-rich, and mesenchymal cell-rich regions. The C1 subtype and C2 subtype scores of tumor cells were negatively correlated with mesenchymal cell and T cell scores. Overall, these results were generally consistent with previous histopathologic analysis findings. (Lines 538-562)

Author Response

Reviewer #1 (Public review):

1.0) This paper investigates the metabolic basis of a node, posterior cingulate cortex (PCC), in the default node network (DMN). They employed sophisticated MRI-PET methods to measure both BOLD and CMRglc changes (both magnitude and dynamics) during attention-demanding and working memory tasks. They found uncoupling of BOLD and CMRglc in PCC with these different tasks. The implications of these findings are poorly interpreted, with a conclusion that is purely based on other work independent of this study. Various suggestions could allow them to place some speculations in line with a stronger interpretation of their results.

This is one of several papers in recent years investigating the metabolic underpinnings of activated (or task-positive) and deactivated (or task-negative) cortical areas in the human brain. In this study, they used BOLD fMRI and glucose PET scan to examine the metabolic distinction of the default node network (DMN), which is known to be deactivated during attention-demanding tasks, with different types of cognitively demanding tasks. Unlike the BOLD response in posteromedial DMN which is consistently negative, they found that CMRglc of the posteromedial DMN (a task-negative network) is dependent on the metabolic demands of adjacent task-positive networks like the dorsal attention network (DAN) and frontoparietal network (FPN). With attention-demanding tasks (like Tetris) the BOLD and CMRglc are both downregulated in DMN (specifically the posterior cingulate cortex, PCC, a task-negative node of DMN), but working memory induces CMRglc increase in PCC and which is decoupled from the negative BOLD response in PCC.

We thank the reviewer for the constructive feedback and the possibility to improve our manuscript. We agree that the interpretation of the results should be strengthened to provide a stronger focus on our data. Regarding the uncoupling of BOLD and CMRGlu during working memory, we acknowledge the need to further elaborate on this topic in our discussion. These suggestions and comments have been incorporated into the revised manuscript as outlined below.

1.1) These complicated results are the main findings, and to provide a biological basis to these data they rather surprisingly, but without their own experimental evidence, conclude that the negative BOLD and negative CMRglc in PCC during attention-demanding tasks is due to decreased glutamate signaling (which was not measured in this study) and the negative BOLD and positive CMRglc in PCC during working memory is due to increased GABAergic activity (which was not measured in this study). It is rather surprising that without measurement, a conclusion is made which would at best be considered a hypothesis to be tested. Thus, independent of these hypothesized mechanisms, they need to summarize their results based on their own measurements in this study (see 3 for a hint).

Thank you for bringing up this point and for the insightful suggestion concerning point 3. We have now explicitly stated that the interpretation regarding glutamate and GABAergic signaling is of speculative nature as theses were not measured in the current work, moreover, we have substantially reduced this section. As such, we agree with the reviewer that this represents an interesting hypothesis to be tested in future work. For further details please see response to comments 1.3 and 1.4.

Discussion, page 16, line 341:

On the neurotransmitter level, one of the current hypotheses regarding BOLD deactivations proposes that CMRO2 and CBF are affected by the balance of the excitatory and inhibitory neurotransmitters, specifically GABA and glutamate (Buzsáki et al., 2007; Lauritzen et al., 2012; Sten et al., 2017). In the PCC, glutamate release prevents negative BOLD responses (Hu et al., 2013), whereas a lower glutamate/GABA ratio is associated with greater deactivation (Gu et al., 2019). As glutamate elicits proportional glucose consumption (Lundgaard et al., 2015; Zimmer et al., 2017), decreases in glutamate signaling in the pmDMN could indeed explain both, the decreased BOLD response and decreased CMRGlu during the Tetris® task. Conversely, increased GABA supports a negative BOLD response in the PCC (Hu et al., 2013), as do working memory tasks (Koush et al., 2021) and pharmacological stimulation with GABAergic benzodiazepines (Walter et al., 2016). In consequence, the observed dissociation between BOLD changes and CMRGlu during working memory could indeed result from metabolically expensive (Harris et al., 2012) GABAergic suppression of the BOLD signal (Stiernman et al., 2021). However, we need to emphasize that glutamate and GABAergic signaling was not measured in the current study, thus, the above interpretations are of speculative nature. Nonetheless, future work may test this promising hypothesis, e.g., using pharmacological alteration of GABAergic and glutamatergic signaling or optogenetic approaches modulating GABAergic interneuron activity.

Furthermore, to maintain a more concise discussion that is closer aligned with the measured results, we have removed the following paragraph:

Discussion, page 15, line 309:

The associations of these metabolic demands between the DMN and task-positive networks is also reflected in their distance along a connectivity gradient, which is hierarchically organized from unimodal sensory/motor to complex associative functions and the DMN being at the end of the processing stream (Margulies et al., 2016; Smallwood et al., 2021). A corresponding decrease in pmDMN glucose metabolism was observed for tasks that activate unimodal networks and the DAN, but not for the FPN. The inverse influence of attention and control networks on the pmDMN may therefore suggest that connectivity gradients are supported by the underlying energy metabolism.

1.2) It is mentioned that the FDG-PET scans allow quantitative CMRglc, both in terms of units of glucose use but also with high time resolution. Based on the method described, it isn't clear how this is possible. Important details of either prior work or their own work have been excluded that show how the time course of CMRglc (regardless of whether it's absolute or relative) can be compared with the BOLD time course. Furthermore, it is extremely difficult to conceive that quantitative CMRglc can be estimated without additional measurements (e.g., blood samples, etc). Significant methodological details have to be provided, which even should make their way to results given the importance of their BOLD-CMRglc coupling and decoupling in the same region.

We thank the reviewer for this important comment and apologize for the lack of clarity. We would like to emphasize that in the current work only spatial patterns of CMRGlu and BOLD signal changes were compared, but not the time course of these signals. The manuscript was edited throughout to clarify this point.

Introduction, page 5, line 110:

Studies using simultaneous fPET/fMRI have shown a strong spatial correspondence between the BOLD signal changes and glucose metabolism in several task-positive networks and across various tasks requiring different levels of cognitive engagement (Hahn et al., 2020, 2016; Jamadar et al., 2019; Rischka et al., 2018; Stiernman et al., 2021; Villien et al., 2014).

Introduction, page 5, line 123

Specifically, it is unknown whether the observed dissociation between patterns of metabolism and BOLD changes in the DMN generalizes for complex cognitive tasks, and whether this in turn depends on the brain networks supporting the task performance and their interaction with the DMN.

Results, page 7, line 143:

From this dataset (DS1) we evaluated the spatial overlap of negative task responses in the cerebral metabolic rate of glucose (CMRGlu quantified with the Patlak plot) and the BOLD signal specifically in the pmDMN. […] After that, the distinct spatial activation patterns across different tasks were used to quantitatively characterize the CMRGlu response of the pmDMN in DS1.

The method of functional PET (fPET) imaging indeed enables the evaluation of changes in glucose metabolism with a relatively high temporal resolution. That is, a conventional bolus application and subsequent quantification yield a single CMRGlu image per scan of about 60 min (typical frame length ~1-5 min) or a single SUV image from a static scan. In contrast, the constant infusion employed in fPET allows to assess baseline metabolism and changes induced by different tasks in a single scan by using a frame length currently down to 6-30 s (Rischka et al., 2018), where the latter was also used in the current study. A general description of the fPET approach is now also included in the manuscript.

Introduction, page 5, line 99:

In this context, functional PET (fPET) imaging represents a promising approach to investigate the dynamics of brain metabolism. fPET refers to the assessment of stimulation-induced changes in physiological processes such as glucose metabolism (Villien et al., 2014; Hahn et al., 2016) and neurotransmitter synthesis (Hahn et al., 2021) in a single scan. The temporal resolution of this approach of 6-30 s (Rischka et al., 2018) is considerably higher than that of a conventional bolus administration. This is achieved through the constant infusion of the radioligand, thereby providing free radioligand throughout the scan that is available to bind according to the actual task demands. Here, the term “functional” is used in analogy to fMRI, where paradigms are often presented in repeated blocks of stimulation, which can subsequently be assessed by the general linear model.

Regarding the absolute quantification of CMRGlu, arterial blood samples were obtained from all subjects of DS1. These were used for absolution quantification of CMRGlu with the Patlak plot. Full details were already provided in the methods section and are now also mentioned in the results.

Results, page 7, line 140:

Simultaneous fPET/fMRI data and arterial blood samples were acquired from 50 healthy participants during the performance of the video game Tetris®, a challenging cognitive task requiring rapid visuo spatial processing and motor coordination (Hahn et al., 2020; Klug et al., 2022). From this dataset (DS1) we evaluated the spatial overlap of negative task responses in the cerebral metabolic rate of glucose (CMRGlu quantified with the Patlak plot) and the BOLD signal specifically in the pmDMN.

Methods, page 19, line 399:

For glucose metabolism, these changes are absolutely quantified in μmol/100g/min with the arterial input function and the Patlak plot.

Methods, blood sampling, page 24, line 536:

Before the PET/MRI scan blood glucose levels were assessed as triplicate (Gluplasma). During the PET/MRI acquisitions manual arterial blood samples were drawn at 3, 4, 5, 14, 25, 36 and 47 min after the start of the radiotracer administration (Rischka et al., 2018). From these samples whole-blood and plasma activity were measured in a gamma counter (Wizard2, Perkin Elmer). The arterial input function was obtained by linear interpolation of the manual samples to match PET frames and multiplication with the average plasma-to-whole-blood ratio.

Methods, cerebral metabolic rate of glucose metabolism, page 25, line 561:

Quantification was carried out with the Patlak plot (t* fixed to 15 min) and the influx constant Ki was converted to CMRGlu as CMRGlu = Ki * Gluplasma / LC * 100 with LC being the lumped constant = 0.89 (Graham et al. 2002, Wienhard 2002).

1.3) It is surmised that the glutamatergic/GABAergic involvement of these metabolic differences in PCC is from another study, but what mechanism causes the BOLD signal to decrease in both stimuli? This is where the authors have to divulge the biophysical basis of the BOLD response. At the most basic level, the BOLD signal change (dS) can be positive or negative depending on the degree of coupling with changed blood flow (dCBF) and oxidative metabolism (dCMRO2) from resting condition. Unfortunately, neither CBF nor CMRO2 was measured in this study. In the absence of these additional measurements, the authors should at least discuss the basis of the BOLD response with regard to CBF and CMRO2. If we assume that both attention-demanding and working memory tasks decreased BOLD response in PCC in the same way, we have identical dCBF/dCMRO2 in PCC with both tasks, i.e., their results seem to suggest an alteration in aerobic glycolysis with different tasks. With attention-demanding tasks, CMRglc decreases similarly to CMRO2 decreases in PCC, whereas with working memory tasks, CMRglc increases differently from CMRO2 decreases. This suggests PCC may the oxygen to glucose index (OGI=CMRO2/CMRglc) would rise in PCC attention-demanding tasks, but fall in PCC with working memory tasks. This is obviously an implication rather than a conclusion as CBF or CMRO2 were not measured.

1.4) Given the missing attention that gives rise to the BOLD contrast mechanism, it is almost necessary to discuss the biophysical basis of BOLD contrast and specifically how metabolic changes have been linked to both increases and decreases in neuronal activity in the past. Although this type of work has largely been conducted in animal models, it seems that this topic needs to be discussed as well.

We would like to thank the reviewer for sharing these insightful ideas and for bringing up these aspects that indeed appear to be essential for the manuscript. Since the points 1.3. and 1.4 complement each other, we have combined them and created a shared response. To fully address the points, the following paragraphs were added to the manuscript.

Discussion, page 15, line 310:

Metabolic and neurophysiological considerations effects

The distinct relationships between BOLD and CMRGlu signals that emerge during specific tasks highlight the different physiological processes contributing to neuronal activation of cognitive processing (Goyal and Snyder, 2021; Singh, 2012). While CMRGlu measured by fPET provides an absolute indicator for glucose consumption, the BOLD signal reflects deoxyhemoglobin concentration, which depends on various factors, such as cerebral blood flow (CBF), cerebral blood volume (CBV) and the cerebral metabolic rate of oxygen (CMRO2) (Goense et al., 2016). In simple terms, the BOLD signal relates to the ratio of ∆CBF/∆CMRO2. Assuming that the observed BOLD decreases during Tetris® and WM emerge from the same mechanisms, this would result in a comparable ∆CBF/∆CMRO2 in the pmDMN for both tasks. Given that these types of tasks (external attention and cognitive control) elicit a reduction in CBF in the pmDMN (Shulman 97, Zou 2011), CMRO2 also decreases albeit to a lesser extent (Raichle 2001). Therefore, the respective metabolic processes can be described by their oxygen-to-glucose index (OGI), the ratio of CMRO2/CMRGlu. Accordingly, our results suggest two distinct pathways underlying BOLD deactivations in the pmDMN that differ regarding their OGI. During Tetris® there is a BOLD deactivation with a high OGI, resulting from a larger decrease in CMRGlu than CMRO2. This metabolically inactive state is in line with electrophysiological recordings in humans (Fox et al., 2018) and in non-human primates showing a decrease of neuronal activity in the pmDMN that covaries with the degree of exteroceptive vigilance (Shmuel et al., 2006; Bentley et al., 2016; Hayden et al., 2009). Therefore, we suggest that the negative BOLD response during external tasks reflects a reduction of neuronal activity and their respective metabolic demands. On the other hand, the relatively increased CMRGlu without the corresponding surge in CMRO2 hints at another kind of BOLD deactivation with a low OGI in the pmDMN during working memory, indicating energy supply by aerobic glycolysis (Vaishnavi et al., 2010; Blazey et al., 2019). Previous work in non-human primates has indeed suggested a differential coupling of neuronal activity to hemodynamic oxygen supply in this region (Bentley et al., 2016). Furthermore, tonic suppression of PCC neuronal spiking during task performance was punctuated by positive phasic responses (Hayden et al., 2009), which could indicate differences between both tasks also at the level of electrophysiologically measured activity.

Reviewer #2 (Public Review):

2.0) This paper provides an important and insightful investigation into patterns of activations that emerge in external task states. The authors use state-of-the-art methods and novel analytic approaches to establish that deactivations in the default mode network during external tasks are driven by activity in brain regions that are important in the current tasks (such as the visual or dorsal attention networks). It will be important in the future to understand whether this is a symmetrical phenomenon by studying this behaviour in states that maximize activity within the default mode network and also drive reductions in networks that are not relevant to these situations.

We thank the reviewer for the encouraging feedback and the constructive comments on our manuscript. We particularly appreciate the interest in the research and the insightful suggestions for future work.

Reviewer #3 (Public Review):

3.0) The authors report a study where, using multiple datasets with [18F]FDG PET bolus + continuous infusion ("functional PET") and BOLD fMRI data, they re-evaluate the metabolic and hemodynamic properties of the default mode network (DMN) in a task-evoked context, with a focus on posteromedial DMN due to its relevance for across-network integration. They show how posterior DMN is differently engaged depending on the chosen task: while visual and motor tasks lead to BOLD deactivations and glucose metabolic decrease, specifically in the dorsal posterior cingulate cortex (PCC) area, working memory tasks produce BOLD deactivations but metabolic increases, specifically in ventral PCC, as shown in their previous paper (Stiernman et al. 2021, https://doi.org/10.1073/pnas.2021913118). This aims to solve the controversies elicited by findings of both increased and decreased glucose consumption in the presence of BOLD deactivation in the DMN.

Additionally, they show how task-evoked glucose metabolism in posterior DMN seems to be shaped by that of the corresponding task-positive networks, with a positive link with dorsal attention and a negative link with frontoparietal network metabolism. This is explored using a type of directional connectivity analysis called "metabolic connectivity mapping", drawn from their previous work (Riedl et al. 2016, https://doi.org/10.1073/pnas.1513752113; Hahn et al. 2020, https://doi.org/10.7554/eLife.52443). They go on to speculate that concomitant BOLD deactivation and reductions in glucose expense might relate to decreased glutamatergic signaling, while BOLD deactivations accompanied by increased glucose consumption might depend on increased GABAergic neuronal activity.

This is a relevant topic because it not only shows how the DMN is flexibly engaged in different tasks but also allows us to better understand the complex relationships between BOLD fMRI and [18F]FDG PET signals, which are still not fully characterized to this day. Of course, while in resting state the situation is further complicated by the more uncertain physiological meaning of the resting BOLD signal, task-evoked states are expected to provide a more interpretable intermodal link between metabolism and hemodynamics, due to the known major changes in blood flow, blood volume, and glucose metabolism - which underlie BOLD and [18F]FDG signal changes - in response to neural activation. However, even in task states, there is not always a strong association between the two responses, as previously shown by the authors themselves (Rischka et al. 2018, https://doi.org/10.1016/j.neuroimage.2018.06.079). This is something I think the authors should stress out a little more, as they have previously done (Rischka et al. 2018, https://doi.org/10.1016/j.neuroimage.2018.06.079), both in the introduction and in reference to Figure 1, which shows clear differences between BOLD and [18F]FDG activations/deactivations (e.g., widespread negative responses in the cerebellum for [18F]FDG).

Overall, the analyses reported in the manuscript are simple and seem mostly sound, drawing from well-established methods in PET and fMRI activation studies, with additional approaches previously developed by some of the authors themselves (e.g., "metabolic connectivity mapping", Riedl et al. 2016, https://doi.org/10.1073/pnas.1513752113). Moreover, a clear strength of the paper is the high number of subjects, at least from a PET perspective, i.e., n = 50 for the Tetris task, plus group averages of previously published data for working memory (Stiernman et al. 2021, https://doi.org/10.1073/pnas.2021913118) and motor tasks (Hahn et al. 2018, https://doi.org/10.1007/s00429-017-1558-0).

The conclusions are in line with the results, and, though a little speculative, are potentially relevant for further exploration aimed at characterizing the neurotransmitter pathways underlying positive and negative BOLD and [18F]FDG responses. Moreover, the language is sufficiently clear to allow a proper understanding of the aims and the results, as well as the details of the analyses. As a side note, the title should probably be adjusted to "Task-evoked metabolic demands of the posteromedial default mode network are shaped by dorsal attention and frontoparietal control networks", to emphasize that the findings do not necessarily generalize to the resting state.

In conclusion, I am overall quite positive about this manuscript, which seems to nicely position itself within the existing literature, making some additional contributions.

We thank the reviewer for the thorough evaluation and the positive feedback on our manuscript, we appreciate the constructive and insightful suggestions. We agree that the differential spatial patterns of activation between the BOLD signal and CMRGlu response require further attention. To address this point in more detail, we have added the following information to the manuscript.

Introduction, page 5, line 110:

Studies using simultaneous fPET/fMRI have shown a strong spatial correspondence between the BOLD signal changes and glucose metabolism in several task-positive networks and across various tasks requiring different levels of cognitive engagement (Hahn et al., 2020, 2016; Jamadar et al., 2019; Rischka et al., 2018; Stiernman et al., 2021; Villien et al., 2014). […]. However, also regional differences in activation patterns have been observed previously between these modalities in these and previous studies (Wehrl et al., 2013). Moreover, a dissociation between BOLD changes (negative) and glucose metabolism (positive) has recently been observed even in the same region of the DMN during working memory (Stiernman et al., 2021), namely the posteromedial default mode network (pmDMN).

Results, caption Figure 1, page 8, line 173

White clusters represent the intersection of significant CMRGlu and BOLD signal changes, irrespective of direction. Note, that also relevant differences between both imaging parameters can be observed, such as decreased CMRGlu in the cerebellum (in both datasets), without changes in the BOLD signal.

We appreciate the reviewer’s proposal for the title as it raises awareness that the activation patterns reflect task-specific inference.

Title:

Task-evoked metabolic demands of the posteromedial default mode network are shaped by dorsal attention and frontoparietal control networks

We have limited the discussion of underlying neurotransmitter effects and explicitly mention that these are of speculative nature. For manuscript adaptation on this point, we would like to refer to points 1.1, 1.3, 1.4 that address this topic as well.

Author Response

Reviewer #1 (Public Review):

The authors assessed the association between exposures and obesity by environment-wide and epigenome-wide association studies. The strength of this study is that exposures, body mass index, and waist-hip ratio were measured three times from adolescence to early adulthood, and the associations were repeatedly evaluated. A weakness of this study is that a loose significance threshold was used for the epigenome-wide association study and only a small number of study subjects were measured in early adulthood. Since this is an observational study, the confounding effect should be considered when interpreting the exposures associated with obesity reported in this study.

Thank you very much for your positive comments and helpful suggestions. We agree that the study has the limitation of the loose significance threshold used for the epigenome-wide association study and the limited sample size in early adulthood. Following the reviewer’s suggestion, we have revised the threshold of significance in the epigenome-wide association study to 1×10-6. We have added more discussion on confounding, and we are more cautious in the interpretation of the results.

Reviewer #2 (Public Review):

Since this study is a long-term cohort study in children and adolescents, it is advisable to decide whether to highlight differences by age group or to show consistent effect after exposure. In particular, obesity and related diseases are closely related to socio-economic environmental factors, and its impact might be different according to age (group) at exposure.

Thank you very much for your insightful suggestions. We agree that the associations of exposures, including socio-economic and environmental factors, with obesity might vary by age, that is why we examined the associations of early life exposures with BMI and WHR at different ages. It is possible that the same exposure may impact obesity differently by age, so we also assessed the associations of exposures selected at earlier ages of BMI/WHR with BMI/WHR at older ages, and compared consistency in the direction of association at different time points. We have added more explanation in the introduction and methods:

(Introduction, paragraph 4)

“Considering that exposures related to obesity at the outset and at the end of puberty may be different, and the associations of the same exposure with obesity may vary by age, we conducted the EWAS at different ages.”

(Methods, Statistical analysis)

“Third, to assess whether the associations differed by age, we checked the associations for the selected exposures from the earlier age groups (~11.5 years and ~17.6 years) in the follow-up survey (n=308) at age ~23 years and compared the direction of associations with those at earlier age groups (~11.5 years and ~17.6 years). Associations with consistent directions of associations in earlier age groups (~11.5 years or ~17.6 years) with those at ~23 years suggest a consistent association by age.”

The part described in comparison with previous studies is a good attempt. However, some results are consistent with those of previous studies and some are not. This may be related to the time difference in socio-economic environmental factors rather than simply the difference between the West and China (Hong Kong). According to modernization/urbanization, changes in living environment, changes in family relationships, and changes in the care environment can also be factors especially in children.

Thank you very much for your positive comments and raising this interesting point. We totally agree that inconsistency with results of previous studies is not merely due to the difference between the West and China (Hong Kong), but also related to changes in structural socio-economic and environmental factors, as well as changes in living environment, family relationships, social and community network, housing and care environment, that affect individuals’ health. Hence, we provided the necessary clarification by adding the following sentences:

(Discussion, Strengths and Limitations)

“Fourth, the inconsistency between some of our findings and previous studies, such as chocolate, sweets, tea and coffee consumption, should be interpreted cautiously. It may not only reflect differences between the West and China (Hong Kong), but also may be due to changes in structural socio-economic and environmental factors, as well as changes in living environment, family relationships, social and community networks, housing and the care environment.”

In studying the effect of environment on gene expression, it can be thought that the influence of genes and the degree of expression might be different depending on the age of the subject (newborn, infant, infant, adolescent, adult) duration of exposure and these still need to be elucidated.

Thank you very much for raising this important point. We fully agree with you. It would be interesting to examine the association of gene expression at different ages with obesity. However, we only collected blood samples at the Biobank Clinical follow-up (age ~17.6 years), so in this study we only conducted the epigenome-wide association study for DNA methylation at ~17.6 years with BMI and WHR at ~23 years. We have added this in the limitation:

(Discussion, Strengths and limitations)

“Fifth, we only collected blood samples at the Biobank Clinical follow-up (age ~17.6 years), so we only conducted the epigenome-wide association study for DNA methylation at ~17.6 years. It would be worthwhile to examine the association of DNA methylation at different ages with obesity.”

Author Response

Reviewer #1 (Public Review):

This study presents a valuable comparison of fibre orientation estimates from three different modalities: diffusion MRI, scattered light imaging, and x-ray scattering. The comparison is interesting as each modality is sensitive to different aspects of tissue microstructure - water anisotropy, micron-scale structural coherence, and myelin lamella respectively. Where scattered light and x-ray imaging can be only applied ex vivo, diffusion MRI has in vivo applications but suffers from being an indirect estimate of the microstructure of interest. By acquiring all modalities in both a vervet monkey and human brain sample, the authors provide quantitative, pixel/voxel-wise comparisons of fibre orientation estimates within the same tissue samples. The authors show convincing agreement in fibre orientations from all three methods, giving confidence in the fidelity of the methods for neuroanatomical investigations. Differences are also observed: SLI is shown to have less reliable estimates of fibre inclination, and the CSD analysis presented overestimates the number of crossing fibre populations when compared to the microscopy methods, particularly in single fibre regions such as the corpus callosum, a known artefact in some diffusion analyses.

In the current PDF, it is very difficult to see fibre orientations in figures due to low resolution, limiting the reader's ability to assess the results. Higher-resolution images would provide more information and easier comparisons.

The methods are generally clear though some additional information is needed:

1) to specify the resolution that the orientations are compared in each figure and how data was up-/down-sampled for these comparisons respectively. For example, each SAXS pixel contains many SLI pixels. It is currently unclear whether the mean SLI orientation from a neighbourhood is equivalent to the SLI compared, or whether a comparison was made for each SLI pixel. Similarly, for the dMRI-microscopy comparisons.

2) I also could not follow why two SLI methods are presented in the methods: SLI scatterometry relating to Figure 2, and angular SLI relating to all other results. Further clarification is needed.

3) Since the quality of the data co-registration can strongly impact pixel/voxel-wise comparisons, quantification of the registration accuracy or overlays demonstrating the quality of the co-registration would be valuable.

A primary weakness of the work as a diffusion MRI validation study is that though diffusion MRI supports many different models to extract fibre orientations with different outputs, here only a single model is compared to the microscopy data, which may affect the generalisability of the results. Further, it only compares the primary orientations from the diffusion MRI and does not consider each fibre population's magnitude (density of fibres) or the orientation dispersion, both of which can influence downstream analyses.

The paper could be strengthened by a more detailed discussion on the differences between the imaging modalities - e.g. in terms of imaging resolution, signal-generating mechanisms, and sensitivity to specific aspects of the tissue microstructure - and how these differences may limit their application to specific neuroanatomical investigations, or ability to validate one another. For example, the microscopy sections are 80 microns thick whilst the diffusion voxel is 200 microns. I expect this could contribute to the difference in the number of fibre populations per voxel.

The hypothesis that dMRI signal contributions from extra-axonal water result in additional fibre populations could be investigated by running CSD on both low and high-b-value data (for example using the openly available MGH dataset, Fan 2016) where fewer secondary fibre populations should be observed at high b-value.

We sincerely thank Reviewer #1 for the constructive feedback, which helped us to significantly improve our manuscript. We hope to have done our best to address all concerns:

First, we regret the insufficient resolution of figures. The resolution must have been reduced during the submission process, when generating the pdf version of our manuscript. We have now submitted all figures as separate files with the highest possible resolution. In addition, all parameter maps are publicly available and can be opened and zoomed in, e.g. with ImageJ, to see the fiber orientations of individual image pixels.

As requested by the reviewer, we have modified our manuscript and added additional methods information.

1) Concerning the data up-/downsampling: We have now specified in each figure caption at which resolution the images were compared and added the following explanation to the newly named Methods section “Image registration and pixel-wise comparison”: To minimize loss of information, the pixel/voxel-wise comparisons were performed at the spacing of the highest resolution image, i.e. the lower-resolution diffusion MRI (dMRI) and small-angle X-ray scattering (SAXS) images were upscaled to match the higher-resolution scattered light imaging (SLI) images. As a result, the fiber orientation of one SAXS pixel (px=150µm) was compared to the fiber orientations of 50x50 SLI pixels (px=3µm), and not to the mean; similarly for comparisons with dMRI.

2) Concerning the two SLI methods: We have added the following explanation to the Methods section “Scattered Light Imaging” to clarify why we used two different methods: To generate the scattering patterns (upper Figure 2C), a time-consuming SLI scatterometry measurement was performed in which the sample was illuminated from 6,400 different angles, as described in Menzel et al. (2021b). This was necessary to achieve sufficiently resolved scattering patterns for a visual comparison with SAXS scattering patterns. The fiber orientations can also be extracted from the peak positions in the azimuthal profiles (cf. bottom Figure 2C), without taking the overall shape of the scattering patterns into account. Therefore, all other results were obtained from more time- and data-efficient angular SLI measurements in which the sample was illuminated from 24 different angles around a circle and the fiber orientations were derived from the peak positions in the resulting line profiles, as described in Menzel et al. (2021a).

3) Concerning the quality of the co-registration: We thank the reviewer for this comment. We agree that the accuracy of image registration has a high impact on pixel/voxel-wise comparisons and determines the quality of our cross-validation study. We have added a new Discussion section “Quality of cross-validation” and inserted a new figure (Figure 4–figure supplement 1) to demonstrate the accuracy of image registration, both for the vervet and human brain samples: The reference and registered images are shown both in direct comparison (top and middle images, respectively) and as overlays (bottom images), as suggested by the reviewer. Reference and registered images show good correspondence (white/gray matter boundaries coincide). Only the fornix of the vervet brain section is not aligned (it moved when re-mounting the sample) so that this region was evaluated separately, as described in the manuscript. We found standard linear transformations (scaling, rotation, and translation) to be sufficient for achieving a fair comparison between the different modalities, demonstrating the experimental feasibility of our approach. There might still be individual voxels that were not sufficiently well aligned, especially when comparing sections (SLI/SAXS) to volumetric measurements (dMRI). However, this would only increase the angular differences between the fiber orientations. Our results can therefore be considered as an upper bound. Using standard linear transformations, we could already show that in-plane crossing orientations from SAXS and SLI, and through-plane orientations from SAXS and dMRI correspond very well to each other.

We understand the focus of our work lying rather on the cross-validation/evaluation of light and X-ray scattering, in comparison to dMRI which is much longer established, than on a “diffusion MRI validation study”: the myelin specific SAXS orientations and crossings were cross-validated with the high-resolution SLI orientations, and SLI out-of-plane fibers were validated using SAXS/dMRI as ground truth data.

The reviewer rightly noted that we used a single analysis method to extract fiber orientations from dMRI data (based on the MRtrix3 dwi2response and dwi2fod commands, using the dhollander and msmt_csd algorithms, respectively). Although to our knowledge this method is one of the most widely used for deriving fiber orientations for subsequent tractography, it is true that other methods might yield different results and that we cannot draw conclusions for diffusion MRI in general. We have included these considerations in the newly named Discussion section “Comparison of SAXS and SLI fiber orientations to dMRI”.

It is also true that our comparison focused on primary dMRI orientations without taking fiber density or dispersion into account. We decided to do so because deriving such metrics from SLI or SAXS data has not been implemented yet. However, we expect this to happen in the following years, enriching future studies. We have also included these aspects in the Discussion section.

We agree with the reviewer that our paper could be strengthened by a more detailed discussion on the differences between the imaging modalities. We have added a paragraph to the new Discussion section “Quality of cross-validation”: We compared results from three different imaging techniques (SLI, SAXS, dMRI) which all have different signal-generating mechanisms and resolutions. The different resolutions should be taken into account when interpreting the comparative studies. To investigate the relationship between SLI peak distance and fiber inclination, we used dMRI/SAXS images with at least 50 times lower in-plane resolution as reference (Figure 6). This is sufficient to validate the theoretical predictions, but insufficient to validate individual pixel values. To validate crossing fiber orientations from SAXS, we used SLI images with 30 times higher in-plane resolution, leading to a broad distribution of angular differences (depending on the region), but the mean difference around zero is evidence for a good overall correspondence (Figure 4). Finally, when comparing fiber orientations in SAXS and SLI to dMRI (Figure 5), it should be taken into account that dMRI voxels (with 200µm size) contain more fiber layers than the corresponding SAXS or SLI voxels (with 80µm section thickness), so that dMRI voxels might include additional fiber populations not present in SAXS or SLI data. On the other hand, fiber orientations that occur both in dMRI and SAXS voxels – like the out-of-plane fiber orientations from SAXS and dMRI (e.g. Figure 6B-C) – can be considered as reliable, given the substantially different contrast-generating mechanisms.

Finally, we thank the reviewer for the suggestion to study different b-values (last comment). We agree that an analysis based on different b-values might yield different results. Especially, an analysis with high b-values is expected to be more specific to the fiber orientations, as most other components of the signal would have already been attenuated. To investigate this hypothesis, we have run a separate analysis with high b-values only (5 and 10ms/μm2) and added a new supplementary figure (Figure 5–figure supplement 4) that compares the results for all b-values to high b-values only. We found that the fiber orientation distributions are almost identical between all b-values and high b-values only.

Reviewer #2 (Public Review):

This work is a cross-validation of an x-ray tomography technique (SAXS) and an optical microscopy technique (SLI) for imaging axonal orientations ex vivo. These innovative methods were introduced in recent papers by the authors, who have teamed up here to compare them side-by-side on the same tissue samples for the first time. The two methods are both label-free (do not require staining) and they are quite complementary. SAXS can provide full 3D orientation measurements on intact tissue, but it operates at a mesoscopic resolution and requires access to a synchrotron. SLI can measure the orientations of multiple fascicles per voxel at a microscopic resolution and relies on more widely accessible equipment, but its accuracy suffers for fiber orientations perpendicular to the imaging plane and it requires tissue to be sectioned before it is imaged. Therefore it makes a lot of sense to explore the complementary strengths of these two techniques, and to use one to "fill in the blanks" of the other. The paper also compares the orientation measurements obtained with SAXS and SLI to those obtained with diffusion MRI. The latter provides only indirect measurements based on water diffusion, at a mesoscopic resolution somewhat lower than that of SAXS, but has the benefit of being feasible in vivo.

A limitation of this study is that conclusions on the comparison between SAXS and SLI are drawn from only 2 sections of a partial monkey brain sample and 2 sections of a partial human brain sample. Conclusions on diffusion MRI are drawn only on the 2 human sample sections. This is particularly an issue for the comparison to diffusion MRI, as the diffusion MRI voxels are wider than the section thickness, hence one cannot preclude that any orientations detected with diffusion MRI but not with SAXS and SLI come from the portion of the voxel that is missing from the corresponding SAXS/SLI section.

The stated aim of the paper is to provide a framework for combining the complementary benefits of SAXS and SLI, rather than simply presenting the results of a cross-validation study. This is a significant and ambitious aim. However, in order for this to serve as a framework, there would have to be clear prescriptions for how researchers interested in obtaining ground-truth measurements of axonal orientations would do so by using these two methods in tandem. This is not adequately developed in the paper in its present form. For example, the results show reasonable agreement between SAXS and SLI orientations when fibers lie within the SLI imaging plane and decreasing agreement for fibers with increasing through-plane inclination. How would the two methods be combined in voxels where they disagree? Would one use SLI orientations in voxels with fewer through-plane fibers and SAXS orientations in voxels with more through-plane fibers? How would voxels be assigned to each category? How would the orientation vectors from the two modalities be composed and how would the resolution difference between the two be handled? When the through-plane measurement of SLI is unreliable, is its in-plane measurement still reliable? That is if there were one mainly in-plane and one mainly through-plane fiber population, would the orientation of the former still be measured correctly by SLI? There is also considerable agreement reported here between through-plane orientations obtained with SAXS and diffusion MRI. Would this mean that diffusion MRI itself could be used to supplement SLI with through-plane orientations? Any clear set of prescriptions along these lines would represent a framework for imaging orientations by combining modalities. This, however, would require detailed steps for how to perform the combination and use the multi- vs. uni-modal framework to reconstruct connectional anatomy.

A key advantage of SAXS is that it can be performed on intact samples, i.e., before any nonlinear distortions of the tissue are introduced by sectioning. Thus it can provide an undistorted reference, with contrast on axonal orientations that would be absent in, say, a structural MRI of comparable resolution. This contrast could be used to drive registration of the distorted SLI sections to an undistorted SAXS volume, and therefore is a key way in which the two techniques can complement each other. Here, however, this is not explored, as SAXS is performed after sectioning. It is not clear if this is the authors' prescription for how a combined SAXS/SLI framework would be implemented, or if it was done specifically for this study.

First, it would seem that SAXS on the intact sample would be lower maintenance, requiring less setup time and hence potentially less overall beamtime than performing SAXS on each section separately. This would make it more practical for routine deployment beyond a few sections.

Second, because the SAXS data are now nonlinearly distorted, they cannot be affinely aligned to the MRI volumes. While, in principle, performing both SAXS and SLI on the sections may facilitate the comparison between the two, having to unmount, rehydrate, and remount the sections in between may negate this advantage, as now there is no guarantee that SAXS and SLI can be affinely registered to each other. Here all these registration steps are performed affinely, so it is unclear to which extent the computed errors between modalities are characterizing the inherent limitations of the respective contrasts, or limitations of the registration technique. Some of the alignment is performed manually, for example, specific regions of the images are realigned by hand, and the slice of the diffusion MRI volume that is aligned to the SAXS/SLI sections is chosen by hand. Again, for this to serve as a framework that can be deployed on whole samples, there would have to be clear prescriptions for how to perform these steps robustly, how to ensure that the MRI can be acquired in a coordinate frame parallel to the sections, etc.

Finally, the paper puts forth a general conclusion that diffusion MRI overestimates the number of fiber populations per voxel, on the basis of small ODF peaks appearing perpendicular to the main ODF peaks. Of all conclusions in the paper, this is the least convincingly supported by evidence. First, these small perpendicular peaks are a known artifact, which would be typically eliminated by ignoring ODF peaks below a certain amplitude, a common practice in diffusion tractography algorithms. The authors refrain from using an amplitude threshold, with the rationale that it may also remove true diffusion orientations. However, they apply a threshold when they detect SLI peaks (a rather stringent 8% of the maximum). Second, the explanation that these artifactual peaks may appear due to vessel walls is not convincing. Vasculature is sparse. A single vessel wall will not impact the diffusion signal in the same way as a bundle of parallel axons. In an axon bundle, water molecule displacements are restricted in all directions except parallel to the axons. A single vessel wall in a voxel will not have the same effect on displacements (which are much smaller than the size of the voxel). From Figure 5, it looks like there would be at most 1-2 of these vessels in a diffusion MRI voxel, and they would not be in all voxels. This cannot explain the widespread appearance of these small artifactual peaks. Third, many ODF reconstruction methods have parameters that can be adjusted to make these artifactual peaks more or less prominent. The default parameters may be optimal for in vivo but not ex vivo data, due to the effects of fixation. In light of these concerns, I would caution against making such a general statement about all diffusion MRI in the human brain, especially on the basis of a single diffusion reconstruction method applied to a single location in one brain.

We sincerely thank Reviewer #2 for the constructive feedback, which helped us to significantly improve our manuscript. We hope to have done our best to address all concerns:

First, regarding the limited number of tissue sections used for our study (second paragraph):

It is true that we only evaluated a limited number of samples – mainly due to the limited beam time available for SAXS experiments. We believe that the main conclusions concerning the cross-validation of SAXS crossing fibers and SLI out-of-plane fibers still remain valid.

The reviewer correctly points out that the dMRI voxels (with 200um size) are wider than the section thickness (80um) so that additional fiber orientations detected with dMRI might come from the portion of voxels missing in the corresponding SAXS/SLI measurement. We have added a clarifying paragraph in the newly named Discussion section “Comparison of SAXS and SLI fiber orientations to dMRI” as well as in the new Discussion section “Quality of cross-validation”. Nevertheless, we do not expect additional fiber orientations in comparable homogeneous regions like the corpus callosum, and fiber orientations that occur both in dMRI and SAXS/SLI – like the out-of-plane fiber orientations from dMRI and SAXS (e.g. Figure 6B-C) – can be considered as reliable, given the substantially different contrast-mechanisms of the microscopy and dMRI techniques.

Concerning the aim of our paper and the questions raised by the reviewer in the third paragraph:

We understand that the term “framework” is not the appropriate word in this context, as it can raise false expectations. Our aim was rather to provide a basis (“groundwork”) to enable combined measurements of SLI/SAXS (and dMRI) on the same tissue samples and cross-validate the techniques (the crossing fiber orientations in SAXS and the through-plane fiber orientations in SLI have not been validated using other techniques so far). We have changed the wording throughout the manuscript, explaining that we focused on laying the “groundwork” instead of providing a “framework”, and reformulated the corresponding sentences.

Our aspiration was to provide a protocol how the complementary imaging techniques can be performed on the same tissue sample. When talking about a “combination” of techniques, we were referring to combined measurements (i.e. measurements on the same sample), and not to a combined analysis (e.g. in form of combined parameter maps and fiber orientation vectors). The latter, while very much needed in the field, would require many more and heterogeneous samples, and work beyond the scope of this manuscript, which we hope to perform in the future. Along these lines, we have removed the term “combined” throughout the manuscript, and wrote e.g. “measurements of SLI and 3D-sSAXS on the same tissue sample” instead of “combined measurements of SLI and 3D-sSAXS” to avoid confusion.

However, it is of course a valid question how SAXS and SLI can be combined in voxels where they disagree, how the orientation vectors can be composed, and how the resolution difference between the methods can be handled. We have added a new Discussion section “Towards a combination of SLI, SAXS, and dMRI” to elaborate on how a combined analysis (e.g. in form of combined fiber orientation maps) can be achieved and what challenges we are facing.

Concerning the reviewer’s question if the orientation of an in-plane fiber population would be correctly measured by SLI if there was another through-plane fiber population: We only evaluated regions belonging to a single fiber population (SLI azimuthal profiles with one or two dominant peaks) and regions belonging two in-plane crossing fiber populations (SLI azimuthal profiles with two dominant peak pairs). Voxels containing both in-plane and through-plane fibers were excluded from the analysis. The determined in-plane SLI orientations can thus be considered as reliable. We have added these aspects to the new Discussion section

“Quality of cross-validation”.

Regarding the reviewer’s question if dMRI itself could be used to supplement SLI with through-plane orientations: Diffusion MRI could indeed be used as a reference to enhance the interpretation of through-plane fiber orientations from SLI measurements. One disadvantage over SAXS is the lower resolution and that it cannot directly be performed on the same tissue section as SLI. These aspects have also been added to the new

Discussion section.

Concerning the reviewer’s suggestion to perform SAXS before sectioning and the problem of image registration (fourth paragraph):

It is true that SAXS tensor tomography can be applied to larger tissue volumes and that it is not limited to tissue sections. However, the reconstruction of crossing fibers has so far only been realized in sections (Georgiadis et al., 2022) and not in intact samples. As we wanted to cross-validate these fiber crossings using SLI as reference, we decided to perform the SAXS measurements on the same tissue sections as the SLI measurements. A comparison to results from SAXS tensor tomography might still be interesting in the future. We have added these considerations to the new Discussion section “Towards a combination of SLI, SAXS, and dMRI”.

It is also true that cutting a section from a brain tissue sample might introduce non-linear distortions; in particular, it is challenging to identify this particular section in the original tissue volume; unmounting and remounting of an already existing section introduces much less distortions. We have added a new figure (Figure 4–figure supplement 1) which shows that a co-registration with linear transformations (scaling, rotation, and translation) is already sufficient to allow for a fair comparison between the different image modalities, both for vervet and human brain samples. Only the fornix of the vervet brain section moved during remounting of the sample, and was therefore evaluated separately, as described in the manuscript. In any case, even if the angular differences in some image pixels were larger due to an imperfect co-registration, a perfect co-registration would only yield even smaller differences. Hence, the reported angular differences can be considered as upper bound, demonstrating that SAXS and SLI fiber orientations show already a very good correspondence. We have added a corresponding paragraph to the new Discussion section “Quality of cross-registration”.

Finally, we agree that a clear prescription would be necessary to enable combined analysis on whole tissue samples. As mentioned further above, our aim was to provide the groundwork for combined measurements on the same tissue sample and cross-validate the different techniques, and not to provide combined fiber orientation maps or similar. We have added our thoughts on how to combine the different image modalities to the new Discussion section “Towards a combination of SLI, SAXS, and dMRI”.

Concerning the final concern of the reviewer that an overestimation of the number of fiber populations per voxel is not sufficiently supported (last paragraph):

We understand this concern and have removed all phrases that could be understood as generalized claims for MRI, including any reference to fiber orientations overestimation. Furthermore, we have extended the Discussion to indicate the non-generalizability of our results.

Regarding the first point that the minor perpendicular ODF peaks could be removed by applying a suitable amplitude threshold: This is a valid remark and was discussed partly in the first version of the manuscript, when referring to increasing the threshold of secondary lobes prior to running tractography algorithms and to the problem that it might decrease the sensitivity for the cases where there exist actual but less prominent secondary fiber populations. We have extended the Discussion to address the concerns of the reviewer.

Regarding the second point that the minor ODF peaks are probably not caused by vessel walls: We thank the reviewer for the valid remarks and have removed all mentions of blood vessels in the manuscript, including the arrows in Figure 5H.

Regarding the third point that parameters can be adjusted to make the artifactual peaks more/less prominent, and that default parameters might be optimal for in vivo but not ex vivo data: We have added the remark that model parameters can be fine-tuned to decrease the percentage of false-positives to the Discussion.

Finally, it is true that we only used a single diffusion reconstruction method and measured only a single location in one human brain with dMRI. As mentioned at the very beginning, the number of samples was limited, and we included the reviewer’s concerns in the newly named Discussion section “Comparison of SAXS and SLI fiber orientations to dMRI”. For the main purposes of the paper like the cross-validation of out-of-plane fibers in SAXS/SLI, the dMRI data was still sufficient as we could show a good correspondence between dMRI/SAXS in these regions.

Author Response

Reviewer #1 (Public Review):

The study tackles the topic of male harm (sexual selection favoring male reproductive strategies that incur a reduction of female fitness) from an interesting angle. The authors put emphasis on using wild-collected populations and studying them within their normal thermal range of reproductive conditions. Where previous studies have used temperature variation as a proxy for stressful environmental change, this approach should instead clarify what can be the role of male harm on female fitness in natural conditions. A minor caveat regarding this point is the fact the polygamy treatment also has a heavily male-biased sex ratio (3:1). The authors argue that this sex ratio is within the range of normal variation in that species, but it is likely that the average is still (1:1) in natural populations and using a male-biased sex ratio could magnify the intensity of male harm. This does not undermine the conclusions regarding the temperature sensitivity of sexual conflict but should be acknowledged.

The authors find that varying temperature within a range found in natural conditions affects the reproductive interactions between males and females, particularly through male-harm mechanisms. Male harm, measured as a reduction in lifetime reproductive success (LRS) from monogamy to polygamy settings is present at 20C, stronger at 24, and absent or undetectable at 28C. Female senescence is always faster in the polygamy mating systems as compared to monogamy, but the effect appears strongest at 20C. Mating behaviors of males and females in these different settings are used to attempt to uncover underlying mechanisms of the sensitivity of male harm to temperature.

A weakness of the manuscript in its current form is the lack of clarity about the experimental design, which makes understanding the results a long and involved procedure, even for someone who is familiar with the field. If the authors consider revising the manuscript, I suggest giving a better overview of the experimental design(s) earlier in the manuscript, perhaps supported by a diagram or flowchart. I also suggest structuring the results better to aid the reader (e.g., make clearer distinctions between results that come from the different experiments). Finally, some additional figures and statistical tests corrected for multiple testing would help get a better feel of some aspects of the dataset.

I believe that the conclusions are generally justified and the results overall convincing. Overall, this is an impressive study with a lot of dimensions to it. Its complexity is a challenge and may require additional effort from the authors to make it easier to access. The core of the question is answered by LRS measures, but the authors have also provided a wealth of behavioral data as well as other fitness components. The manuscript could be greatly improved by putting more effort into linking the different metrics together to track down potential mechanisms for the observed variation in male-harm-induced reduction in female LRS. The discussion would also benefit from considering the female side of the sexual conflict coevolution arms race.

We are thankful for the nice words and constructive appraisal of our work. As stated above, reviews like this are extraordinarily helpful. The reviewer mentions four main points that we have addressed:

1. We now expand a bit on the justification to use a (3:1) male-biased sex ratio in the methods section (lines 150-155). We also acknowledge potential limitations of this design in the discussion (lines 563-571).
2. To clarify the methods, we have placed this section before the results. This, in itself, has significantly improved the clarity of the manuscript. We have also substantially re-written the methods and results (including adding some tables) to streamline the text while providing all the necessary details, and have also included several diagrams to illustrate all our experiments (in the SM, see Figs. S1.1 to S1.5) along with a general schematic figure of the general design that we present early on in the main text (in the introduction, see Fig. 1).
3. As suggested, we have re-run all analyses using the Benjamini-Hochberg procedure in order to correct for inflation of type I error rate due to multiple testing. We have also included in the SM a complementary set of models that also test for this via post hoc Tukey contrasts. Both these approached corroborate our initial findings, and thus contribute to strengthen our results.
4. We now explicitly discuss the female side of things in the discussion (lines 636-647).

Reviewer #2 (Public Review):

Londoño-Nieto et al. investigated the influence of temperature on the form and intensity of sexual conflict in Drosophila melanogaster. They aimed to test the effect of naturally occurring temperature fluctuations on a wild population of Drosophila while disentangling pre- and postcopulatory episodes of sexual conflict. To this end, they exposed females to males under monogamy or polyandry, hence manipulating the degree of male harm experienced by females. The effect of temperature was explored by exposing these groups to 20, 24, or 28{degree sign}C. They found that female fitness suffered from male harm most at 24{degree sign}C and less at the other two temperatures. Interestingly, pre- and postcopulatory episodes of sexual conflict were affected differently by temperature. Overall, these data suggest that the relationship between sexual conflict and temperature can be strong and complex. Hence, these results can have important implications for the impact of sexual conflict on population viability, especially in light of the climate crisis.

We want to thank the reviewer for the time invested in reading and reviewing our work. We are glad to read that the reviewer found our results interesting and considered our study to be of importance to the field.

This paper tackles a highly relevant question using an established model organism for sexual conflict and contains a rich dataset obtained using a series of carefully planned experiments and analysed in an appropriate way. Importantly, the authors used biologically meaningful temperatures and mating treatments, which increases the relevance of the data. The main conclusions are well supported by the data. Nevertheless, the devil is in the detail, and given the way the authors frame their study (i.e. testing a natural population under naturally occurring temperature fluctuations) and their results (i.e. sexual conflict is buffered by temperature effects in the wild) there are some limitations to be considered:

We appreciate the positive feedback! The reviewer identified potential limitations and made good suggestions that have only served to improve our manuscript considerably, for which we are very grateful. Details follow on how we have dealt with each specific comment.

1) The authors frame their study as addressing the question of how sexual conflict reacts to naturally occurring temperature fluctuations in the wild. Nevertheless, the population used in this experiment had been kept for nearly 3 years in the laboratory prior to the experiment. Importantly, the authors ensured that the laboratory population maintained genetic diversity, by regularly crossing wild lines into it. Nevertheless, this population remained for some time in the laboratory under standardized conditions. The applied temperature fluctuations are in a biologically meaningful range (though only during the reproductive season), but it remains unclear if the applied fluctuations were in a standardized way (i.e. pre-programmed) or included random fluctuations (i.e. a more natural setting). This laboratory setup has certainly clear advantages, for example, it enables the exclusion of any effects other than the temperature on sexual conflict. Nevertheless, how these will then ultimately play out in the wild could be a different story.

Agree. We clarify now that we meant pre-programmed fluctuations and acknowledge this limitation in the methods (lines 124-131).

2) The authors highlight clearly that temperature fluctuations in the wild might play an important part in how sexual conflict plays out in natural populations. This very interesting and highly relevant point might lead the reader to assume that this is what was actually tested in the experiment. Nevertheless, in the experiments, different constant temperatures were applied to the flies, while only the stock population was kept at a fluctuating temperature regime. Hence, the influence of fluctuations during episodes of sexual conflict remains untested. While the present data show that sexual conflict can be modulated by temperature, the effect of naturally occurring fluctuations on the net cost of sexual conflict to a population remains unclear.

Again, a fair point that we acknowledge in the current version (lines 571-575). “Second, our treatment temperatures were stable, designed to study how coarse-grain changes in temperature across the adult lifespan of flies may influence how sexual conflict unfolds in nature. Thus, future studies will need to encompass how fine-grained fluctuation (i.e., repeated variation of temperature across an adult’s lifespan) may affect male harm for a more comprehensive picture of temperature effects on sexual conflict in the wild”.

3) The authors conclude that the effect of sexual conflict can be buffered by temperature in the wild. In general, I agree with this, although a more conservative way of framing this would be to say that temperature modulates or moderates sexual conflict instead of buffers it. If there really is a buffering effect of temperature in the wild remains to be tested, I believe. This will depend on how actual changes in temperature affect this dynamic (see point 2). In addition, I think another interesting open question is what the mechanism behind the observed differences might be. Are male and female interests really more aligned at different temperatures (i.e. males plastically reduce harm)? This would really buffer the harm of sexual conflict at those temperatures. Nevertheless, alternatively, males might not be perfectly adapted to manipulate the female optimally at lower or higher temperatures. This would mean that if the temperatures change, males might evolve to increase the manipulation of females, and hence the scope for sexual conflict might not change in the end under this scenario. Nevertheless, as the authors themselves state: 'An intriguing possibility is thus that SFPs are more effective at lowering female re-mating rates at warm temperatures, thereby buffering these costs.' Therefore, a temperature-dependent increase in the effectiveness of male manipulation might counterintuitively reduce sexual conflict in this species.

We echo both points in the current version of the paper (see lines 633-655).

4) In the end the authors argue that the climate crisis might have 'unexpected positive consequences via its effect on male harm'. Sexual conflict is indeed widespread, but it takes many different forms (as has been nicely described in the introduction of this paper). Because the studied system seems to be quite a specific example, it is questionable how far spread this phenomenon is in nature. In addition, it remains unclear how male harm will evolve in response to the climate crisis (see point 3). Finally, the relative fitness of females increased in the present experiment, as the tested range was within the reproductive optimum of the species. Nevertheless, the relative importance of the positive effect of sexual conflict on fitness outside of optimal temperatures seems questionable.

Agree. Altogether, we have tried to tone down our conclusions regarding the implication of our results for a climate change scenario, and acknowledge all the points highlighted by the reviewer in the current version of the manuscript (see lines 563-575).

Nonetheless, I believe these results to be of exceeding interest to the scientific community and of importance to the field. It opens up many potential research directions and adds further data to the fascinating field of sexual conflict, SFPs, and male harm in Drosophila.

We are thrilled to read that the reviewer found our study of exceeding interest.

Reviewer #3 (Public Review):

In this paper, the authors explore the effects of the environment, specifically temperature, on male harm to females. Male harm is the phenomenon where males reduce female fitness in polyandrous systems, where a single female may mate with multiple males. The selection of males to increase their reproductive success in male-male competition can lead to genetic conflict that increases male fitness at the expense of female fitness. Typically, male harm has been studied in single environments under optimal conditions. However, there is an increasing focus on the effect of the environment on fitness costs of male harm to females, as a way to better understand the effect of male harm on population fitness in more realistic ecological contexts. In this paper, the authors add to these studies by exploring the effect of temperature on male harm and female fitness, using the fruit fly Drosophila melanogaster, as a model system. They find that temperature affects the impact of male harm on female fitness, with male harm having the greatest effect at 24˚C relative to 20˚C and 28˚C. The authors then go on to disentangle how temperature affects the various components of male harm that impact female fitness (e.g. harassment, ejaculate toxicity). The paper demonstrates that male harm depends on ecological context, which has implications for understanding its impact on population fitness under realistic ecological scenarios, particularly with respect to climate change.

The strength of the paper is that it demonstrates that male harm (presented as differences in female life reproductive success between monogamous and polyandrous matings) changes with temperature. The authors dissect this general observation by showing that different aspects of precopulatory reproductive behavior, for example, male-male aggression, copulation rate, and female rejection rate, also change with temperature. Further, they demonstrate that correlates for male ejaculate quality also change with temperature, suggesting that temperature also affects postcopulatory mechanisms of male harm.

The weakness of the paper is that the method and results section are difficult to follow, which negatively impacts the interpretation of the data. The experiments are complex and need to be for what the authors are studying. Nevertheless, the paper is written in a way that makes it challenging for the reader to fully understand how precisely the experiments were conducted. Further, the authors do not explain clearly how some of the experiments relate to the phenomenon ostensibly being assayed. For example, a more detailed explanation of why mating duration and remating latency are assays for ejaculate quality in the context of sperm competition would be very helpful in interpreting the data. Further, a clearer explanation of the statistical analyses conducted

Thank you for the positive, detailed and constructive review. We agree with all the weaknesses laid out and we have strived to address all of them in the current version. This includes a mayor rearrangement, structuring and re-write of the methods and results section and extra statistical analyses. Please find the details below.

Author Response

Reviewer #1 (Public Review):

The authors generated a detailed single-cell RNAseq dataset for the microfilariae stage of the human nematode parasite Brugia malayi. This is an impressive and important achievement, given that it is difficult to obtain sufficient material from human parasites and the microfilariae are protected by a chitin sheath. The authors collected microfilariae from jirds and carefully worked out a protocol of digestion, dissociation and filtering, to obtain single-cell material for sequencing.

The single-cell resource was complemented with a dataset derived from FACS-sorted large secretory cells, allowing the identification of several specific proteins expressed in this unique microfilarial cell-type important for immune evasion.

The authors also generated new data for secretory cells of Caenorhabditis elegans and concluded that there is limited similarity between the composition of Brugia and C. elegans secretory cell types.

In a further set of experiments, the authors analysed gene expression changes in dissociated Brugia cells to the commonly used anthelminthic drug ivermectin. This revealed specific gene expression changes across various cell types, providing new insights into how the drug effects the parasite.

Finally, the authors developed a method to keep dissociated Brugia cells alive in culture for two days. This method will aid cellular studies of this parasite.

The authors may want to explore the new resource in more detail to reach more specific biological conclusions. For example, the authors mention that the large secretory cells are critical to parasite survival and immune evasion. With a more complete list of genes expressed in these cells the authors could try to reach more specific conclusions or predictions. Are there newly identified secreted factors that could contribute to immune evasion? It would be important to read in more detail about such proteins (including an analysis of the sequences and phylogenies), especially if the authors could identify new candidates as potential vaccine or diagnostic targets. Likewise, can the data be used to understand in more detail the mechanism of immune evasion or ivermectin action?

Thank you for this comment. We have since added a source data file with the list of secretory cell DEGs along with gene ontology (GO) analysis. We have added a main figure to the revised manuscript that takes a deeper look at transcripts enriched in the secretory cell compared to other annotated cell types. Lastly, we included a deeper look at the paralogous expansion of C2H2 transcription factors that localize near exclusively to the secretory cell. This family of transcription factors is diverse and the significant presence in the secretory cell may play a role in adapting to varying host environmental conditions or in the expression of proteins contributing to immune evasion. Our single-cell data show specific transcriptional shifts in cells expressing putative IVM targets and recapitulate changes identified in whole-parasite drug exposure experiments and highlight the importance of cell connectivity to the in vitro phenotype. These supplemental analyses of the secretory cell will seed future lines of investigation about secretion and aid in further dissecting anthelmintic mode of action.

The authors searched for known secreted proteins, including antigens, vaccine targets, and diagnostic markers and mapped the expression of these to the single-cell atlas. It is not clear from the paper how comprehensive previous studies to identify secretory proteins were. With the new resource in hand, the authors could look at all secreted proteins (with a signal peptide) expressed in the ES and other cells. The paper would benefit from a more comprehensive overview of the classes of secretory proteins and their expression.

Thank you for this suggestion. We have completed a computational prediction of signal peptides in differentially expressed secretory cell transcripts (Figure 4) and show that although there is an enrichment of signal peptide-containing sequences enriched in the secretory cell compared to other cell types, less than half of the proteins identified contained signal peptide sequences.

This was unsurprising as most secreted proteins identified in the literature (diagnostic and vaccine targets) do not have a signal peptides. The routes of exit for these prominent circulating targets remain murky. We also carried out transmembrane prediction on protein-coding genes that are differentially expressed in the secretory cell (and other cell types) and note that some of these are established components of exosome-like vesicles, emerging as important players in host modulation. This additional analysis has been added to a new figure (Figure 4) and the accompanying results section.

The authors show that an abundance of C2H2 transcription factors is localizing almost exclusively to the secretory cells. It would be useful to see a classification of these proteins and phylogenetic analysis relating them to C2H2 from C. elegans and other animals.

The C. elegans genome contains 106 annotated C2H2 zinc finger transcription factors. Based on a reverse phylogenetic approach, we identified a total of 241 orthologous C2H2 zinc finger transcription factors in B. malayi, many of which exhibit strong and/or exclusive expression in the secretory cell. This analysis has been added to an additional figure (Figure 4) describing the secretory cell in more depth alongside signal peptide and transmembrane domain analysis of differentially expressed genes in the secretory cell compared to other identified major cell types.

In general, a more detailed bioinformatic analysis of secretory products and more discussions of potential functions (e.g. serpins etc.) would make the paper more interesting and could stimulate more mechanistic thinking.

Author Response

Reviewer #1 (Public Review):

Tunneling nanotubes, contrary to exosomes, directly connect remote cells and have been shown to allow the transfer of material between cells, including cellular organelles and RNAs. However, whether sorting mechanisms exist that allow to specifically transfer subspecies of RNAs, especially of mRNA, has not been shown, and the transcriptional consequences of RNA transfer have not been addressed yet.

Using cocultures (or mix or single cultures as controls) of human MCF7 breast cancer cell line, and immortalized mouse embryo fibroblasts (MEFs), followed by separation of human and mouse cells by cell sorting, the authors performed deep sequencing of the human mRNAs detected in mouse cells. An accurate analysis of the transferred material shows that all donor cell mRNAs transfer in a manner that correlates with their expression level, with less than 1% of total mRNA being transferred in acceptor cells.

These results show that the process of RNA transfer is nonselective and that the consequences on the cells receiving the RNAs should depend on the phenotype of the sending cells.

Although we did not address this last point in the original paper, we concur with this statement since we presented evidence to this effect in our previous publication (Haimovich et al., 2017) and which we discussed in the in the original Discussion section (lines 498-508 in the original manuscript; lines 529-539 in the revised manuscript). We have now amended the Introduction (line 91 of the modified manuscript) to reflect this idea.

These results are complemented by the last part of the manuscript where the authors convincingly show that the coculture of the two cell lines results in significant transcriptomic changes in acceptor MEF cells that could become CAF-like cells.

Reviewer #2 (Public Review):

In this manuscript, the authors characterize the extent of RNA transfer between cells in culture, with an emphasis on trying to identify RNAs that are transferred through tunneling nanotubes (TNTs). They use an in vitro human-mouse cell co-culture model, consisting of mouse embryonic fibroblasts and human MCF7 breast cancer cells. They take advantage of the CD326 cell surface molecule, which is specifically expressed on MCF7 cells, to separate the two cell populations using magnetic beads conjugated to anti-CD326 antibodies, followed by deep sequencing to identify human RNAs present in mouse cells. They identify many 'transferred' RNAs. Further analysis of sequencing data together with experiments using synthetic reporters indicate that RNA transfer is non-selective, that the amount of transfer strongly correlates with the level of expression in donor cells, and does not appear to require specific RNA motifs. The authors also note that co-culture with MCF7 cells leads to significant changes in the MEF transcriptome.

The experiments are overall carefully designed, and the data are clearly and quite carefully presented to point out limitations in interpretation and to distinguish speculations from experimental conclusions.

We thank the reviewer for this comment.

It should however be kept in mind that it is unclear to what extent these limitations influence the conclusions reached. For example, the identification of transferred RNAs relies on the purity of the isolated cell populations ad, while the authors provide some supporting evidence for this, nevertheless potential caveats remain. For instance, the isolated MEF samples used for analysis appear to lack single MCF7 cells, but still contain components, labeled as 'double stained' and 'unstained' cells, which are uncharacterized. The authors present some arguments as to why these would not contribute to 'transferred' reads, but given the low level of detectable transferred RNAs, and the unclear origin of these components, whether they influence the results could be debatable.

It is unlikely that these populations contributed to the human mRNA signals in the MEFs, since the percentage of these populations was substantially higher in the “Mix” samples than in the “Co-culture” samples. We now added the following text (lines 174-181 in the revised manuscript) which clarifies this point: “In addition, we found small sub-populations of double-stained and unstained cells within the purified populations that we suspect are mostly MEFs (see Methods). These sub-populations were greater in the Mix-derived MEFs vs. the Co-culture-derived MEFs (i.e. 0.08% and 0.03% double-stained, and 2.8% and 2.67% unstained in Mix samples vs. 0% and 0.03% double-stained, and 1% and 1.9% unstained in the Co-culture samples). As a consequence, if these double-stained and unstained cells had contributed to the background of human reads in the MEFs, we would’ve expected to have many more human reads in the Mix-derived MEFs.” However, this was not the case, rather we observed a 6.6-fold increase in human RNA presence in the Co-culture-derived MEFs (versus that in the Mix-derived MEFs) after subtraction of the single culture background. In addition, we note that the level of detectable human RNAs in the MEFs is not low, rather it is the percentage of human RNA that undergoes transfer that is low.

Furthermore, the small number of replicates (2 replicates for the genome-wide studies and 1 replicate for most of the subsequent experiments) minimizes the confidence in the conclusions.

We apologize for not stating it clearly that the smFISH, RT-qPCR ,and quadrapod experiments were all performed in 2 replicates. This information has now been added to the figure legends.

In this context, it is also notable that the profile of transferred RNAs between the two replicates of co-cultured samples appears quite different by PCA analysis. It is thus conceivable that there might be specificity in the RNA 'transferome', influenced by unknown experimental variables, which is though masked when averaging those samples in subsequent analyses.

We have replied to Reviewer #1 on this issue. PCA analysis (Figure 2B) of the heat map data (Figure 2A) reveals the similarity between the different samples, whereby 78% of the variability in the data is revealed by PC1 and 6.7% by PC2. Given that PC2 measures only 6.7% of the variation in the data, it likely results from small differences in the individual co-culture samples (such differences are often observed within replicas of RNA-seq experiments) and not via major differences in the measured transferomes. This indicates that the co-culture samples were overall quite similar as can also be observed from the heat map shown in Figure 2A, as differentiated from the controls (e.g. Mix, Single culture). Thus, we do not believe that further replicas will greatly change the results showing the abundant presence of human RNAs in the mouse cells after subtraction of the Mix background. We included additional sentences in the text and figure legend to clarify this point (lines 208-212 in the revised manuscript).

While the manuscript emphasizes the role of TNTs in RNA transfer, the actual involvement of TNTs relies solely on the observation that potential TNTs form between co-cultured cells. Other means of transfer, such as through engulfment or phagocytosis of cell fragments, could still possibly contribute.

While it is possible that transfer might occur through other means, our earlier paper (Haimovich et al., 2017) showed that engulfed apoptotic bodies rarely contribute to mRNA transfer, even upon near-100% of donor cell death. Moreover, RNAs in apoptotic bodies found in acceptor cells can be clearly identified by smFISH, as the RNAs are tightly clumped together. Likewise, our quadrapod experiments (Figure 6-figure supplement 1) might have revealed RNA transfer if engulfment of cell fragments had occurred.

Furthermore, the dependence of mRNA transfer on direct cell-to-cell contact is demonstrated for 5 RNAs and extrapolated to transcriptome-wide RNA transfer, an assumption which might, or might not, be valid.

We concur that we extrapolate from the few validated examples and have now added the following text (line 604-611 in the revised manuscript): “We validated several examples of transferred mRNAs that transfer via a contact-dependent mechanism, likely TNTs (Figure 6 and Figure 6-figure supplements 1 and 3), and extrapolate from them to the entire transcriptome. Although it is possible that some or many mRNAs transfer by means other than TNTs, we think it unlikely, since the results on TNT-mediated cell-to-cell transfer in both this and our previous publication (Haimovich, 2017), as well as by others (Ortin-Martinez et al., 2021; Su and Igyarto, 2019), tested a variety of mRNAs from different families and which localize to various sub-cellular localizations. This indicates that the pathway we have uncovered is more general than the few examples presented here.” In addition, we now cite in the Discussion (lines 611-621 in the revised manuscript) a new pre-print recently posted to bioRxiv that shows similar results of mRNA transfer in a human-mouse cells co-culture model.

Finally, the results on gene expression changes induced by co-culture (Figures 7, 8) are of unclear relevance. As the authors point out, it is uncertain whether RNA transfer or other paracrine or adhesion-mediated signaling events, underlie these changes. It is therefore not easy to see how these results relate to the rest of the presented work. Furthermore, while the authors expand on the potential significance of changes observed in genes related to cancer-associated fibroblasts or to immunity-related genes, these remain speculative and untested.

We concur that the part of the paper regarding the consequences of co-culture (upon the endogenous transcriptome) does not clarify the specific contribution of the “transferome” to the phenomenon. Future co-culture studies measuring transcriptome-wide transfer using the quadrapod co-culture system versus cell-cell contact co-culture could be performed. Yet, to make the distinction between TNT-dependent and -independent effects when cells are in contact will require further mechanistic knowledge of TNT-mediated mRNA transfer, which is beyond the scope of this paper. Nevertheless, we believe that the data on the endogenous gene expression in co-culture is important and could be useful to the cancer research community outside the context of the transferome information.

Overall, the manuscript presents evidence indicating that RNA is transferred non-selectively in co-cultured cells, under specific conditions and between the cell types tested. The impact of the work is reduced by the lack of mechanistic understanding underlying this transfer and the uncertainty of whether this phenomenon has any subsequent physiological relevance.

Our global analysis of TNT-mediated transfer (the transferome) is only a second step towards understanding this important and only recently identified process (i.e. the first step). Obviously, we would be happy to gain more mechanistic insight and knowledge of physiological relevance. We are currently working on several projects to try and answer some of these questions, but as one can understand, these are technically challenging, and have not yet come to fruition.

Author Response

Reviewer #1 (Public Review):

The human genetic variant Dantu increases the surface tension of red blood cells making it hard for malaria parasites to invade. This was shown beautifully by Kariuki et al in 2020 (doi.org/10.1038/s41586-020-2726-6) by analysing blood from children using in vitro assays with cultured malaria parasites. Now Kariuki et al show that parasite growth is indeed restricted in vivo by infecting Dantu adults under controlled conditions with cryopreserved Plasmodium falciparum sporozoites and analysing parasite growth by qPCR. The authors compare parasite growth, peak parasitaemia and if / when treatment was sought for malaria symptoms between non-Dantu (111) and Dantu heterozygous (27) and homozygous (3) participants. Dantu either completely prevented malaria parasite detection in the blood (for 21 days) or slowed down parasite growth considerably.

The authors present compelling in vivo evidence that Dantu conveys protection by preventing malaria parasites from establishing a blood-stage infection. Because the effect on parasite growth is crystal clear the link to uncomplicated malaria follows - no/less parasites leads to less participants experiencing malaria symptoms and seeking treatment. It should however be noted that the paper does not show that Dantu reduces symptomatology at identical parasite densities to non-Dantu. Its protective effect seems to be purely parasitological.

Given that all volunteers were exposed to malaria prior to being experimentally infected (in various transmission settings ranging from low to high) the authors state that they adjusted for factors like schizont antibody concentration in their multi-variate analysis. More details on the assumptions and which dependent / independent variables were included would benefit interpretation. It would be also good to see if Dantu individuals were spread homogeneously across all transmission settings - if e.g. they all had history of intense malaria exposure and thus strong pre-existing anti-malaria immunity this might account in part for reduced parasite growth when compared to non-Dantu from lower transmission settings. Being able to de-convolute the effect of pre-existing immunity from Dantu would strengthen the paper.

Thank you for the positive feedback and summary of the key findings. We absolutely agree that breaking down the impact of Dantu genotype by transmission would have been very interesting, but the sample numbers for some of the genotypic groups were simply too small to make stratification by area of residence meaningful. Instead, to address the core issue of whether prior immunity is a complicating factor in our analysis, we used measurements of antibodies to whole schizont extract as a proxy indicator of transmission setting or “malaria exposure” in our multivariate analyses. There was no difference in anti-schizont antibody levels across Dantu genotype groups – these data are now included in Figure 3 – figure supplement 1, as requested. This suggests that differences in pre-existing anti-malaria immunity between Dantu and non-Dantu cannot explain the differences seen in our current study. Regarding the comment about assumptions and variables in the multivariate analysis, we have added more details as requested, as outlined in further detail in subsequent points below.

The authors also presents data on other red cell polymorphisms known to modulate malaria infection and improve outcome: G6PD, blood group O, alpha thalassaemia and ATP2B4. However, no statistically significant differences between non-carriers and hetero/homozygous individuals were observed. This is probably because these mutations exert their effect not directly on parasite growth but modulate disease symptoms when parasite burden is high - which cannot be investigated in controlled human malaria infection settings as ethical considerations mandate treatment of all volunteers at parasite densities >500 parasites/ ul or any parasitaemia with symptoms. Controlled infections need to be complemented with other methods to understand the protective impact of genetic polymorphisms.

We thank the reviewer for this helpful observation with which we completely agree. To acknowledge this issue, we have added some consideration of this point to the Discussion section of the revised manuscript, within the sub-section that discusses protective mechanisms of other red cell polymorphisms on page 14.

Author Response

Reviewer #2 (Public Review):

Despite high bone mineral density, increased fracture risk has been associated with T2D in humans. In this study, the authors established a model that could mimic some aspects of T2D in mice and then study bone turnover and metabolism in detail.

Strengths

This is an exciting study, the methods are detailed and well done, and the results are presented coherently and support the conclusions.

Previous work from Dr. Long's group over this last decade has established a requirement for glycolysis in osteoblast differentiation. They showed the requirement for glycolysis not only for the anabolic action of PTH but also as an effector downstream of Wnt signaling. Using the T2D mouse model they have generated, they test if manipulating glycolysis and oxidative phosphorylation can rescue some of the detrimental effects on bone in this model. They use several novel approaches, they use glucose-labeling studies that are relatively underutilized, and it provides some insights into defective TCA cycle. They also utilize BMSCs that have been sorted for performing single-cell sequencing studies to identify specific populations modified with T2D. Unfortunately, the results are modest and need some clarification on what these populations add to the story.

We appreciate the positive comments. Although T2D had only modest effect on the relative pool size of each cell population, the changes in metabolic pathways (glycolysis and oxphos) in several clusters were notable and provided support to the central notion that T2D altered cellular metabolism in osteoblast-lineage and other bone marrow cells.

The authors use two approaches: a drug (Metformin) and a number of mouse genetic models to over-express genes involved in the glycolytic pathway using Dox inducible models. The results with overexpressing HIF1 and PFKFB3 show a potential rescue of bone defects with T2D, and Glut1 overexpression does not rescue T2D-induced bone loss.

Concerns

The authors have generated several overexpression models to manipulate the glycolytic pathway to recuse T2D-induced bone loss. The use of DOX in drinking water has been shown to affect mitochondrial metabolism. Did the authors control for these effects? Since both the groups of mice got the DOX in drinking water, there is internal control.

The experiments were controlled for any potential effects of DOX per se as all animals were subjected to the same DOX regimen.

Only one of the rescue experiments had control with the Chow diet. There are some studies that have shown a high-fat diet to be protective of bone loss in TID models.

We have now added the chow diet control for the Hif1a rescue experiment as well (Fig. 7).

The use of metformin to correct metabolic dysfunction and, thereby, bone mass is an exciting result. Did the authors test to see if they had in any way rescued this phenotype because of reducing ROS levels? The decrease in OxsPhos seen with the seahorse experiments suggests there could be mitochondrial dysfunction often associated with ROS generation.

I appreciate the reviewer’s insight here. We have not examined ROS levels but agree that changes in ROS levels could potentially contribute to the bone phenotype in diabetes.

All of the experiments used male mice (because STZ use and ease of T2D establishment in males). It would be better if this were made clear in the title.

The title has been revised to specify male mice.

Is the T2D model presented really represent what is observed in humans? Some experiments to test the other factors implicated in T2D and whether those are modulated in the rescue experiments might help address this.

Our T2D model exhibited all typical features of T2D patients, those including obesity, glucose intolerance and insulin resistance. We have shown that metformin modestly improved glucose tolerance and insulin sensitivity in the T2D mice (Fig. 6C, E). We have not examined whether those global metabolic features were modulated in the genetic rescue experiments which targeted only osteoblasts.

Author Response

Reviewer #1 (Public Review):

This paper establishes a strong case for the post-translational modification of C/EBPalpha to play a strong role in its effects, in this case, to promote macrophage differentiation in collaboration with PU.1. The cellular system being used for most of the experiments here takes advantage of the dual roles of PU.1 in B cells, which normally do not express C/EBP family factors, and in myeloid cells, which normally do express C/EBP family factors. The authors and others have previously shown that PU.1 and C/EBPalpha are very powerful collaborators, both needed to establish a macrophage identity. Thus, the title of the paper provocatively implies that the C/EBP modification that keeps it from being methylated on Arg35 works by increasing the re-distribution of PU.1 from B cells to myeloid gene sites in combination with C/EBP. Indeed, the authors show proximity ligation data to show that PU.1-C/EBPalpha juxtaposition is more frequent in the nucleus if C/EBPalpha cannot be Arg-methylated. The paper also shows careful and thorough characterization of the B to myeloid lineage conversion gene expression changes and the mapping of the Arg residues in C/EBPalpha that are most important to keep demethylation. Similarly, the paper provides strong evidence that it is Carm1, and not another protein arginine methyltransferase, that is responsible for the regulatory modification. This is a valuable and well-characterized demonstration of a mechanism that should be considered more generally as a regulator of transcription factor action.

The mechanism proposed by the authors is that C/EBPalpha relocates PU.1 to macrophage sites and that C/EBPalpha R35A binds and relocates PU.1 more efficiently than wildtype, and this seems likely and appealing. However, it is not as strongly supported by data within the paper itself as the other points in the paper are. There is a puzzling gap in the data: no direct evidence is shown that C/EBPalpha is really relocating PU.1 from B cell to macrophage regulatory elements at all. Despite the figure titles (Fig. 4 and Fig. S4), there is no ChIP-seq data to show PU.1 binding sites before and after interaction with either wildtype or R35A mutant C/EBPalpha, just accessibility data. There is also a question of whether such a redistribution would occur fast enough to account for the impressive speed of the R35A mutant's other effects. These questions seem fairly straightforward to address. If relevant data could be added, it would greatly increase the impact and generality of the paper. The paper could be published with this claim converted to a suggestion, based on the current data, or it could be published in a higher-impact form if additional data could be provided to demonstrate the relocation more directly. The authors would be more expert about the logistics of the experiment, but it seems that a direct ChIP-seq-based comparison should be feasible and powerful for the argument of the paper.

We have now included PU.1 and C/EBPa ChIP-seq experiments, using C/EBPaWT and C/EBPaR35A- induced cells, replacing the virtual ChIP-seq experiments. Integrating the data obtained with our dynamic ATACseq data, the new findings largely support the previously proposed PU.1 redistribution (‘theft’) model. To make the data easier to understand, we now first show the PU.1 and C/EBPa binding to distinct B cell- and macrophage- restricted GREs contained in a single genomic fragment (new Fig. 5). The findings nicely visualize how PU.1 becomes redistributed from B-GREs to M-GREs, in a C/EBPa mutant-accelerated manner. We were also happy to see that a genome-wide analysis of the data again shows the accelerated redistribution of PU.1 by C/EBPaR35A (new Fig. 6). Finally, the comparison of the ChIP-seq and ATAC-seq data also added more mechanistic detail, such as by revealing that chromatin remodeling of lineage restricted GREs can be uncoupled from the regulation of associated genes.

Finally, the effect of the mutation is assumed to be only on the interface for interaction between C/EBPalpha and PU.1 (or other co-factors). However, C/EBPalpha is such a short-lived protein that any modification that slightly increased its half-life could increase its potency. It seems important to present some quantitative protein staining evidence to clarify whether the steady-state level of C/EBPalpha in C/EBPalpha R35A-expressing cells is really unchanged from C/EBPalpha wild-type-expressing cells.

We agree that this is an important issue and have therefore now performed a cycloheximide experiment with 3T3 cells expressing inducible forms of the two proteins. The data in Figure S4C show that C/EBPaR35A exhibits a similar stability than wild type protein and is expressed at 20-30% lower levels under steady-state conditions in uninduced cells. They also show that C/EBPa is surprisingly stable. These new findings are in line with the comparison of the two proteins by Western blots of mutant and wild type transfected 293T cells and of infected B cells, which also show similar levels of the two proteins (Fig. 7C and D). Therefore, the finding that expression of C/EBPaR35A is similar or slightly lower than that of the wild type argues against the possibility that an elevated expression level of the mutant could explain the effects observed.

Finally, although not requested by the reviewer, we have now addressed the possibility that that the effect of the alanine replacement of R35 is mostly due to a change from a charged to a non-charged hydrophobic residue. This is not the case, as a replacement of arginine 35 by the charged amino acid lysine still leads to an accelerated BMT induction (Figure S7).

Author Response

Reviewer #1 (Public Review):

This paper is based on the premise that ketamine exerts antidepressant effects that are rapid by increasing glutamatergic transmission. However, the authors note that how this effect occurs is unclear because ketamine antagonizes the NMDA receptor, a glutamatergic receptor. Others have suggested a compensatory change in the glutamatergic transmission and the authors suggest how this might occur. The authors should clarify if prior studies suggested a mechanism different from theirs and if so, which might be correct.

There are also other mechanisms, such as the block of NMDA receptors on interneurons and the disinhibition of principal cells. It is important to clarify if this has already been addressed in the literature. Also, if their cultures are primarily glutamatergic neurons or they include interneurons and glia.

The authors show calcineurin is reduced after ketamine exposure and this increases AMPA receptor GluA1 phosphorylation. They also show that Calcium permeable AMPA receptors (CP-AMPARs) increase.

They also use suggest that the CP-AMPARs and other changes lead to enhanced synaptic plasticity, which could lead to antidepressant effects.

Although a lot of work is done in cultured hippocampal neurons, 14 days in vitro, they show effects in vivo that are consistent with the data from cultures. For example, ketamine increases GluA1 phosphorylation. Also, blocking CPAMPARs in vivo reduces anxiety/depressive behaviors such as the open field and tail suspension tests.

Overall the study appears to be done well and the presentation, writing, and references are good. There are important concerns regarding statistics, behavior, and pharmacology and several minor concerns.

Major concerns

1) Statistics.

What was the stat test if the control was always 1? Often the control group is 1.00 with no SD but in other tests, the control group is 1.000 with an SD.

In the previous submission, we neglected to include this information. Immunoblotting data have variable raw values; hence, the control group was used to normalize each group and was compared to the experimental groups. Thus, the control value for immunoblotting was always 1.000 without SD. Similarly, for imaging data, the average peak amplitude in control cells was used to normalize the peak amplitude in each cell and was compared to the experimental groups' average; thus, the control group is 1.000 with SD. The Franklin A. Graybill Statistical Laboratory at Colorado State University has been consulted for statistical analysis in the current study, including sample size determination, randomization, experiment conception and design, data analysis, and interpretation. Grouped results of single comparisons were tested for normality with the Shapiro-Wilk normality or Kolmogorov-Smirnov test and analyzed using the unpaired two-tailed Student’s t-test when data are normally distributed. Differences between multiple groups with normalized data were assessed by nonparametric Kruskal-Wallis test with the Dunn’s test.

2) Behavior.

It is not clear that the open field and tail suspension tests measure antidepressant actions. Why were more standard tests such as forced swim or sucrose preference, novelty-suppressed feeding, etc not used?

We agree with the Reviewer’s concern. However, both the open field test and tail suspension test have long been used to determine animals’ anxiety-like and depression-like behaviors, respectively, in rodents (Seibenhener and Wooten, 2015; Ueno et al., 2022). Specifically, the open field test has been widely used to measure the ketamine effects on anxiety-like behavior in rodents (Guarraci et al., 2018; Pitsikas et al., 2019; Shin et al., 2019; Akillioglu and Karadepe, 2021; Yang et al., 2022; Acevedo et al., 2023). The tail suspension test has also been used to examine the ketamine effects on depression-like behavior in animals (Fukumoto et al., 2017; Yang et al., 2018; Ouyang et al., 2021; Rawat et al., 2022; Viktorov et al., 2022). Studies suggest that the forced swim test and the tail suspension test are based on the same principle: measurement of immobility duration while rodents are exposed to an inescapable situation (Castagne et al., 2011). Importantly, it has been suggested that the tail suspension test is more sensitive to antidepressant agents than the forced swim test because the animal will remain immobile longer in the tail suspension test than the forced swim test (Cryan et al., 2005). For this reason, we chose to use the tail suspension test instead of the forced swim test. This information has now been included in the revised manuscript. Additionally, because ketamine produces antidepressant effects within one hour after administration in humans (Berman et al., 2000; Zarate et al., 2006; Liebrenz et al., 2009), our study aims to understand the mechanism underlying ketamine's rapid (less than an hour) antidepressant effects. Given that sucrose preference test and the novelty suppressed feeding test need multiple days, it would not be suitable to achieve our goals.

3) Pharmacology.

The conclusions rest on the specificity of drugs.

Is 5 uM FK506 specific?

20 μM 1-naphthyl acetyl spermine (NASPM)?

10 mg/kg IEM-1460?

We neglected to add the rationale for the drug concentrations in the previous submission. Previous research, including our own, has employed FK506 at a variety of different concentrations to inhibit neuronal calcineurin activity (1 - 50 μM) (Hsieh et al., 2006; Schwartz et al., 2009; Kim and Ziff, 2014). Specifically, we have shown that 5 μM FK506 treatment for 12 hours significantly reduces neuronal calcineurin activity to increase GluA1 phosphorylation, which induces the expression of CP-AMPARs to elevate AMPAR-mediated synaptic activity (Kim and Ziff, 2014). Moreover, previous studies, including our own, have used NASPM at a variety of different concentrations to inhibit CP-AMPARs (3 - 250 μM) (Tsubokawa et al., 1995; Koike et al., 1997; Noh et al., 2005; Nilsen and England, 2007; Hou et al., 2008; Kim and Ziff, 2014). In fact, we have shown that 20 μM NASPM significantly reduces CP-AMPAR-mediated synaptic and Ca2+ activity (Kim and Ziff, 2014; Kim et al., 2015b). Finally, multiple reports demonstrate that 10 mg/kg IEM-1460 significantly reduces in vivo CP-AMPAR activity (Wiltgen et al., 2010; Szczurowska and Mares, 2015; Adotevi et al., 2020). This information has now been included in the revised manuscript.

Reviewer #3 (Public Review):

Ketamine has been shown to be effective at producing a rapid-antidepressant effect at low doses, but the underlying molecular mechanism of this effect is still not clear. Previous studies have suggested that the effect of low-dose ketamine may occur by promoting neuronal plasticity in the hippocampus. However, this goes against the findings that ketamine acts as a noncompetitive NMDA receptor antagonist, which should prevent NMDAR-dependent plasticity. Furthermore, a therapeutic dose of ketamine has been shown to increase neuronal Ca2+ signaling, which again does not conform to its antagonistic action on NMDA receptors. In this paper, the authors provide evidence that therapeutic low-dose ketamine increases the expression of Ca2+-permeable AMPA receptors (CP-AMPARs) by increasing phosphorylation of GluA1 subunit of AMPARs and surface expression of GluA1-containing CP-AMPARs. They further provide evidence that this is likely mediated by a decrease in calcineurin activity and that blocking CP-AMPARs prevent the antidepressant effect of ketamine in mice. One interesting finding of this study is that the authors see heightened sensitivity of ketamine in female mice, both at the level of behavioral readout and for molecular correlates. This finding is interesting in light of the different pharmacokinetics of ketamine reported in females and that ketamine metabolites can bind estrogen receptors.

Based on their data and previous findings, the authors outline a plausible molecular signaling mechanism for the antidepressant effect of ketamine. Specifically, the authors propose that reduced neuronal activity, which could be triggered by ketamine-induced NMDAR antagonism, causes homeostatic plasticity to upregulate GluA1-containing CP-AMPARs. Their data would support this idea, as phosphorylation of GluA1 as well as increased surface expression and functional incorporation of CP-AMPARs at synapses have been shown before in models of homeostatic plasticity.

1) Overall, the study is well-done and the data presented support the main conclusions. One main question is whether the current finding provides a conceptual advancement in our understanding of the molecular signaling involved in ketamine's antidepressant effects.

We thank the reviewer's critique. In fact, research suggests multiple potential mechanisms of ketamine-induced neural plasticity. The main mechanism by which ketamine produce their therapeutic benefits on mood recovery is the enhancement of neural plasticity in the hippocampus (Miller et al., 2016; Aleksandrova et al., 2020; Kavalali and Monteggia, 2020; Grieco et al., 2022). However, ketamine is a noncompetitive NMDAR antagonist that inhibits excitatory synaptic transmission (Anis et al., 1983). A hypothesis to explain these paradoxical effects is that ketamine acts via direct inhibition of NMDARs localized on inhibitory interneurons, leading to disinhibition of excitatory neurons and a resultant rapid increase in glutamatergic synaptic activity to activate Ca2+ signaling pathway (Deyama and Duman, 2020; Gerhard et al., 2020). This stimulates the brain-derived neurotrophic factor (BDNF) signal pathway, which subsequently increases the translation and synthesis of synaptic proteins to enhance AMPAR-mediated synaptic plasticity (Deyama and Duman, 2020). Another potential explanation is that ketamine inhibits NMDARs on excitatory neurons, which induces a cell-autonomous form of homeostatic synaptic plasticity resulting in increased excitatory synaptic drive onto these neurons (Miller et al., 2016; Kavalali and Monteggia, 2020). Homeostatic synaptic plasticity is a negative-feedback response employed to compensate for functional disturbances in neurons and expressed via the regulation of AMPAR trafficking and synaptic expression (Wang et al., 2012). According to this hypothesis, ketamine disrupts basal activation of NMDARs on excitatory neurons, which engages a mechanism of homeostatic synaptic plasticity that results in a rapid compensatory increase in synaptic AMPAR expression in these neurons in a protein-synthesis dependent manner (Kavalali and Monteggia, 2023). Additionally, there is a NMDAR inhibition-independent mechanism mediated by hydroxynorketamine (HNK), the ketamine metabolite that lacks NMDAR inhibition properties (Carrier and Kabbaj, 2013; Franceschelli et al., 2015; Zanos et al., 2016). The current study offers a new neurobiological basis for ketamine’s actions that depend on the NMDAR inhibition-mediated elevation of GluA1-containing AMPAR trafficking, which is likely independent from the previous described mechanisms including the BDNF-induced protein synthesis-dependent (Deyama and Duman, 2020) or the NMDAR inhibition-independent pathway (Carrier and Kabbaj, 2013; Franceschelli et al., 2015; Zanos et al., 2016). Nonetheless, there are still many important questions surrounding the molecular mechanisms of ketamine's actions. This new information has now been included in the revised manuscript.

2) There are previous studies that showed an increase in CP-AMPARs in the nucleus accumbens and an increase in the expression of GluA1 in the hippocampus with low-dose ketamine. In addition, ketamine's antidepressant effect has been shown to require GluA1 phosphorylation. The main contribution of this paper might be that it provides the potential molecular signaling within the same preparation (i.e. hippocampal neurons) and provides a causal link of CP-AMPARs in mediating the behaviorally measured antidepressant effect of ketamine.

The study showing that ketamine induces the insertion of CP-AMPARs in the nucleus accumbens did not examine whether this change resulted in antidepressant behaviors (Skiteva et al., 2021). Therefore, it is difficult to conclude that the ketamine-induced expression of CP-AMPARs in the nucleus accumbens plays a role in behaviors. Moreover, as described above, a recent study shows that the hippocampus is selectively targeted by ketamine (Davoudian et al., 2023). We thus chose the hippocampus as our experimental model to test our hypothesis. However, we are unable to rule out the potential role of nucleus accumbens in ketamine’s antidepressant actions.

3) Another question is whether the behavioral effect of ketamine is due to molecular changes in the hippocampus as outlined in this paper. A more targeted inhibition of CP-AMPAR function could resolve this issue. With the systemic application of CP-AMPAR antagonist as done in this study, it would be hard to know the role of CP-AMPAR upregulation in the hippocampus in mediating ketamine's effect. Especially, considering that low-dose ketamine has been shown to upregulate CP-AMPARs in the nucleus accumbens. While it would have been nice to know the site of action, this does not alter the conclusion that CP-AMPARs are involved in mediating the antidepressant effect of ketamine on behavioral readouts.

We agree with this point. We have thus removed “the hippocampus” in the title and have further made equivalent revisions in the other parts of the revised manuscript.

Author Response

Reviewer #1 (Public Review):

The authors have used computational models and protein design to enhance antibody binding, which should have broad applications pending a few additional controls. The authors' new method could have a broad and immediate impact on a variety of diagnostic procedures that use antibodies as sensitivity is often an issue in these kinds of experiments and the sensitivity enhancement achieved in the two test cases is substantial. Affinity maturation is a viable approach, but it is laborious and expensive. If the catenation method is generalizable, it will open up opportunities for antibody optimization for cases where affinity maturation is either not feasible or otherwise impractical. Less clear is how this method might enhance therapeutic potency. Issues that arise when using therapeutic antibodies are often multifactorial and vary depending on the target and disease state. Many issues that occur with antibody-based therapies will not be rectified with affinity enhancement.

We agree with the limitation.

Reviewer #2 (Public Review):

The paper presents an interesting design approach to having homodimeric IgGs with higher binding affinity to the antigens on a surface by fusing a weakly homodimerizing protein (a catenator) to the C-terminus of IgG. Considering the homodimeric IgGs with likely enhanced antigen binding ability and their stabilization with a reversible catenation when bound to the surface is an interesting idea. With agent-based modeling - the simulations based on Markov Chain Monte Carlo (MCMC) sampling - and proof of concept experiments, it has been possible to show the enhanced antigen binding ability of the homodimer Igs for many folds, where the weakly homodimerizing ability of the catenator is indicated to have a central role, enabling proximity effect driven catenation on the antigen bound surfaces. While the results render the enhanced binding affinity of the catenated homodimeric IgGs, the study would benefit from a more elaborated interpretation and discussions of the results.

The following discussion is now stated in the revision (pages 19-20, in the revision); “While we demonstrated that dual catenator-fused heterodimeric IgGs can enhance binding avidity, the oligomer formation or potential intramolecular homodimerization of the catenator necessitates the development of a more robust catenator for application to conventional homodimeric IgGs. Specifically, the ideal catenator should geometrically disallow intramolecular homodimerization, exhibit fast association kinetics, and be able to withstand the standard low pH purification step. On the other hand, our demonstration indicates that this approach can be applied to bispecific antibodies employing a heterodimeric Fc.”

One interesting base of the discussion may include how the fusion of the catenator may likely affect the binding behavior, the intrinsic binding behavior, and/or on the global structural changes, of IgGs (monomeric and homodimeric (catenated) per se beyond its proximity-driven contribution. Would it lead to a more restricted structure in the mobility in the unbound states so as to decrease the entropic cost for the binding and thus increase the binding avidity/affinity (in addition to external proximity-driven association). In other words, what would be the role of entropy in the free energy of binding, given that the enthalpic contributions remain the same? Possible effects of the length of the catenator should also in parts be related to the entropy. For example, if a longer and more flexible catenator is considered, what would the resulting observation experimentally and computationally be?

The binding site occupancy depends on [catAb]/KD. Figure 4-figure supplement 2 shows the binding site occupancy and (KD)eff as a function of (KD)catenator. In this simulation, [catAb] was fixed (10-9 M) while KD was varied (from 10-8 to 10-6). In the figure legend and in the main text, we now explicitly state that KD was varied from 10-8 to 10-6 (page 30, in the revision). To address this comment, we set KD = 10 nM (as used for simulation in Figures 3 and 4), and varied [catAb] from 0.1 to 10 nM. The binding site occupancy and (KD)eff as a function of [catAb] are plotted for three different set values of (KD)catenator (1 μM, 10 μM and 100 μM). The new figures are now presented as Figure 4-figure supplement 3. This simulation shows that the enhancement of (KD)eff by increasing the concentration of catAb is much less dramatic than that by increasing the affinity for catenator homodimerization at [catAb] > 10 nM.

On the other side, simple simulation approaches have a high value with a level of abstraction while still keeping the physical and biological relevance. In the simulations, i.e. in the sampling of various states, three main terms/rules to govern the behavior are implemented. One is a term favoring an increase in the ability to bind (preventing to unbinding) to the surface upon the catenation of IgGs. This may need to be substantiated for the simulations not imposing a preassumed ability to increase the binding (or decrease the unbinding) ability upon the catenation.

We agree with the review in that the third rule favors the binding ability of catenated IgGs, because it assumes that catenated antibodies are not allowed to dissociate from the binding site. While this assumption is not exactly correct, we think that it is valid, considering the behavior of a multivalent ligand. When the IgG portion dissociates completely from the binding site, it is still anchored by the catenation arm, and thus it will rebind the same binding site immediately. This postulation agrees with the quantitative analysis showing that multivalent ligand exhibits orders of magnitude binding likelihood increase when the ligand size is comparable to the stretch length of a conjugating linker [Liese, S. & Netz, R. R., ACS Nano, 12, 4140 (2018)].

The weakly homodimerizing state of the catenator appears as one of the important aspects of the proposed design strategy. Would it also be possible that the experimental observations may readily also imply the higher binding ability of the catenator fused IfgG without the homodimerization on the surface (due to the reduced entropic cost for the binding)? The presentation of the evidence of the homodimerization of the catenator and the catenated IgGs on the surface would strengthen the findings and discussions.

To fully address this comment, we would need to consider the detailed molecular behavior of the IgG part, the catenator and the linker, probably using molecular dynamics simulation, which we think is outside the scope of the current work. We like to qualitatively describe what we think about the raised issues. Fused to the C-terminus of Fc, the catenator won’t affect the complementary determining region (CDR) of Fab which is located on the opposite side of the C-terminus of Fc. This notion is supported by the observation that the SDF-1α-fused antibodies exhibited association kinetics similar to those of the mother antibodies (Figure 5).

Regarding the mobility of the structure, we presume that the fused catenator would not interact with the antibody portion and thus it would not affect the intrinsic structural mobility of the antibody.

Since the catenator is fused to the C-terminus of Fc by a flexible linker, the homodimerization of catenator would decrease the entropy upon catenation. However, the enthalpic contribution would overcome the entropic loss, and result in negative free energy of the catenator homodimerization.

Figure 2-figure supplement 1 (in the revision) shows the simulation for five different values of the reach length (R), which is the sum of the linker length and half of the catenator length. The simulation results show that the likelihood of catenation decreases as the linker length increases over the distance (d) between the two adjacent catAb-2Ag complexes, while it is maximum when the reach length equals d. Since the catenator length is fixed, increasing the linker length (such that R > d) will lower the catenation effect.

Reviewer #3 (Public Review):

The authors proposed an antibody catenation strategy by fusing a homodimeric protein (catenator) to the C-terminus of IgG heavy chain and hypothesized that the catenated IgGs would enhance their overall antigen-binding strength (avidity) compared to individual IgGs. The thermodynamic simulations supported the hypothesis and indicated that the fold enhancement in antibody-antigen binding depended on the density of the antigen. The authors tested a catenator candidate, stromal cell-derived factor 1α (SDF-1α), on two purposely weakened antibodies, Trastuzumab(N30A/H91A), a weakened variant of the clinically used anti-HER2 antibody Trastuzumab, and glCV30, the germline version of a neutralizing antibody CV30 against SARS-CoV-2. Measured by a binding assay, the catenator-fused antibodies enhanced the two weak antibody-antigen binding by hundreds and thousands of folds, largely through slowing down the dissociation of the antibody-antigen interaction. Thus, the experimental data supported the catenation strategy and provided proof-of-concept for the enhanced overall antibody-antigen binding strength. Depending on specific applications, an enhanced antibody-antigen binding strength may improve an antibody's diagnostic sensitivity or therapeutic efficacy, thus holding clinical potential.

Thanks for the favorable comments.

Author Response

Reviewer #1 (Public Review):

The introduction does not clearly set up the background for the key questions that the manuscript addresses. One of the key parts of the manuscript is to attempt to determine whether locomotory behaviour evolves because of direct or indirect selection of the traits. However, the authors don't provide an argument for why a salty environment would select for locomotory traits. Indeed, in the discussion, the authors point out that it is likely an unmeasured trait (body size) correlated with locomotory traits that are under selection. They present arguments for why this might be the case and point to un-included data that show body size significantly genetically covaries with all of the traits studied. Since the authors appear to have these data, and one of their key questions is comparing direct vs. indirect responses to selection, it would be more powerful to include the body size data and estimate selection on all traits together.

We now include body size in all of our phenotypic and genetic analyses. We also include estimates of selection gradients from the ancestral selection differentials and the Gmatrix. We detail in the Introduction the biological significance of locomotion traits and their potential relationship with body size, in low and high salt environments. The experimental results show that divergence in locomotion traits (Figure 6) correlates with adaptation (Figure 5), because of direct and indirect selection (Figure 9).

Phenotypic plasticity was estimated from a series of univariate models, with estimates arranged in a vector. As the authors point out in the manuscript, traits that are not included in a model but covary with traits that are can largely bias estimates of the traits that are included. For this reason, it would make sense to estimate phenotypic plasticity using a multivariate model, as has been done for G matrices.

We analyze the ancestral phenotypic plasticity and the phenotypic divergence during evolution using a multivariate approach (MANOVA). This approach simplifies the text as from the eigen decomposition of the SSCP matrices we can estimate canonical traits of ancestral phenotypic plasticity (pmax; see Table 1 with notation definitions) and phenotypic divergence in the new target high salt environment (dmax). We continue to do the univariate analysis as it allows us to estimate BLUPs for each inbred line (used for visual representation), as well as the significance of phenotypic divergence at each replicate population relative to the ancestral population (delta_q). Both multivariate and univariate approaches led to similar results (shown as supplementary figures).

The estimation and interpretation of G matrices are a critical part of the manuscript. The authors state that broad sense estimates of G are a good proxy for additive genetic variation in this system, but in the Discussion they also state that overdominance was likely important during evolution to the salt environment, leading to some lack of clarity on whether dominance is important or not.

We are sorry for the lack of clarity. We have eliminated the discussion on overdominance as it was peripheral to our results. Broad-sense genetic variances should be a good proxy for additive genetic variances when there is no inbreeding depression and no directional dominance or dominance epistasis; cf. Lynch and Walsh 1998. We previously showed that there is no inbreeding depression for the trait we use as surrogate for relative fitness (self-fertility) and also that there is no directional dominance for locomotion behavior traits. We now explain our use of broad-sense genetic (co)variances as a proxy for additive genetic (co)variances in the Introduction and Methods.

It is also unclear how uncertainty in estimated G matrices was assessed. Showing that G differs from noise is critical to the majority of the results presented. The authors cite Morrissey and Bonnet (2019) as providing the method for generating the null distribution of G, however, this paper does not appear to propose or describe a method to do this.

Thanks for this comment. Morrissey and Bonnet (J Heredity, 2019) was incorrectly cited and the explanation for finding the expected noise distributions was misleading. In brief, we produced a set of 1000 G-matrices each computed after shuffling the line ID and the block ID from the phenotypic dataset. This was done to produce random expectations of the genetic variances as the MCMC estimates are positive-definite. We computed the posterior mode for each of these 1000 G-matrices to obtain a null distribution (shown in orange). To infer significance, we compared the posterior mode of the empirical estimate with the 95% CI of the posterior mode distribution obtained from the randomized G-matrices. When determining which eigenvectors explain standing genetic variation we also used the distribution of posterior modes of the randomized G-matrices. However, as pointed out by Sztepanacz and Blows (Genetics, 2017), the eigenvalues of the eigenvectors do not follow a uniform distribution, as would be expected by chance. Because of this we asked the question of whether the amount of variance in the eigenvectors of the empirical G-matrix (gmax, g2, etc.) was expected, by projecting the random G-matrices onto these eigenvectors. This is a null that is conditional on the observed data. We show these results in Figure 2 - supplement figure 3. Both approaches are similar, particularly for the first 2 eigenvectors. There is now a paragraph in the Discussion about finding potential consequences for adaptation of traits with little genetic variance.

Although the figure captions state that they are showing estimates of genetic variances, it appears to be heritability (bounded between 0 and 1). Whether the authors are studying heritability or genetic variance is an important difference, particularly in the context of a changing environment and phenotypic plasticity, where environmental variation is important and expected to change. For example, the result that G is smaller in evolved populations could simply be due to their being larger environmental variance in the salt environment (as you would expect). This is unrelated to an evolutionary response.

There might have been some confusion because transition rates are positive and not normally distributed. To achieve normality they were log transformed. We have not reported estimates of heritability, all estimates presented are of genetic variances, unscaled. The only exception is body size where the raw data was multiplied by 50 in order to have a similar phenotypic scale as the transition rates when estimating genetic (co)variances, not heritability. We agree that the evolution of environmental stochastic variance is interesting but not immediately relevant to the questions we address.

It seems that comparisons to the ancestral population were done for A160, not the founding population for each evolved line at G0. It is not clear whether the founder effects of each replicate are important and if this is the most appropriate comparison (the Discussion suggests that founder effects are important).

We have better detailed in the Methods, and also with an introductory section in the Results section, the derivation of the experimental populations. The population acronyms might have been misleading. The A6140 is a population that was domesticated to the lab conditions for 140 generations (replicate #6 of the domestication process). We report the evolution of 3 GA populations, which were all derived from A6140 with minimal sampling problems for the estimated effective population sizes (sampled 10^4 individuals from A6140 for each GA, for Ne of 1000 during domestication - Chelo and Teotónio Evolution 2013 -). Therefore, GA populations after 50 generations of evolution are appropriately compared with their (unique) ancestor population. We no longer discuss potential founder effects.

Overall, there is much interesting data collected and analysed in this manuscript, addressing a valuable question. However, it is not obvious whether the estimates of G matrices are different from noise, and heritability may not be the most appropriate scale to ask questions about phenotypic plasticity and evolution in a novel stressful environment that may affect levels of environmental variation.

Please see previous replies. Our ancestral G-matrix estimates indicate that at least 3 eigentraits are different from random expectation in both environments (Figure 2, supplement figure 3), and in high salt evolved populations continue to have more than expected genetic variance at 3-5 eigentraits (Figure 7, supplement 2). We are conservative in these estimates as depending on the null we could consider more eigentraits. In the previous version of the manuscript we concluded that only 2 ancestral eigentraits were orthogonal due to an error in the code (we did not divide by 2 the null expectations). But even presuming that only one eigentrait (gmax) has genetic variance in the ancestral population, we previously reported that mutational variance is not in the same trait (see Mallard et al., G3, 2023; and mmax in Table 3), and further that the trait under selection is neither gmax or mmax (compared in Table 3 the selection gradients with gmax or mmax). At a minimum there are 3 genetically or environmentally independent traits. As noted in previous replies, we estimate and present genetic variances throughout. We do not present estimates of environmental variances and feel that doing so would make the manuscript overly complicated.

Reviewer #2 (Public Review):

Response to selection: It was not clear to me that it was appropriate to interpret locomotor behavior as having evolved in response to the salinity environment. Specifically, where is the evidence that any change in trait means is a (direct or indirect) response to selection imposed by increased salinity rather than the neutral drift of a trait due to the reduction in population size caused by the salinity? Strong evidence of adaptive evolution would be provided by all 3 replicates significantly diverging from the ancestor in the same direction. Model 2 seems to aim to test the null hypothesis that the three replicates diverged from one another via a random effects model - but with only three replicates, there is very low power, and variance is likely to be estimated as zero. I'm not sure what is shown in Tables 3 & 4, or how these results relate to models 2 & 3, so my interpretation of the information may be incorrect. Nonetheless, and noting that the errors around estimates are not presented, there seems to be considerable heterogeneity in size and direction of divergence between replicates for most of the traits. Is this study really dissecting responses to directional selection, or is it dissecting drift?

We have modified the statistical modelling of the phenotypic data. Model 2 is no longer presented. We provide a MANOVA multivariate analysis equivalent to model 2 (with replicate populations as fixed effects) but now including both environments, together with the univariate models. MANOVA results show that all traits are significantly different across populations (i.e., at least two populations differ from one another). The fitted estimates from the MANOVA are not reported with errors in R but it is obvious that not all traits evolved in each replicate GA population (Figure 6). We therefore tested the difference between each of the evolved populations and the ancestral population using a univariate approach (Figure 6, supplemental source data table 2). In this univariate analysis, block was modeled as having random effects (which we could not model with MANOVA). In the high salt environment, the replicates GA 1,2,4 differed significantly for respectively 4, 6 and 4 transitions rates (out of 6). The traits are all evolving in the same direction, and this even when the trait difference between evolved and ancestral populations is not significant. We provide compelling evidence of parallel evolution and thus selection (see review about how to infer selection in evolution experiments in Teotónio et al. Genetics 2017). We tried to be exhaustive in our statistical reporting but would happily provide additional details if requested.

What are the traits, and what is the confidence in G? My outsider's interpretation of these results is that defining 6 transition states is a way of getting at a single behavioral trait, and I was not convinced that these data were suitable for addressing questions about multivariate evolution. Genetic parameters were estimated using MCMCglmm, which imposes boundaries on estimates. The authors state that they followed Morrissey and Bonnet 2019, but I was unable to infer what this means with respect to accounting for the contribution of sampling error to covariances (or how they accounted for the positive variance constraint). Because I was unsure how sampling error was being assessed for G, I was not confident about the interpretation of statistical support for individual parameters, or for eigenvalues of G. Following this forward, if the measured characteristics constitute a single trait, with an entirely shared genetic basis, then the results of strong alignment of everything with gmax makes complete sense - there is a single trait, that is heritable and plastic, and for which the mean evolved.

Our initial draft was misleading and we now provide more detailed description (see also replies #5 and #12 above). We computed 1000 randomized G matrices to account for the constraints imposed by the MCMCglmm algorithm. This should account for the bias inherent with variance estimation and the eigen decomposition we did given our sample sizes. You will find that all 6 transition rates show genetic variance (Figure 2, supplement figure 2) and that up to three eigentraits have more genetic variance than the randomized G-matrices (Figure 2, supplement figure 3).

The 6 transition rates are the mathematical description of changing movement states in 1-dimensional space (under memoryless assumptions). A priori we do not know how many relevant traits there are, if they are genetically or environmentally independent. To help the reader, we provide a Table 3 with the trait loadings for the several canonical traits of phenotypic plasticity, divergence and selection. The first canonical trait of standing genetic variation, gmax, is indeed aligned with phenotypic divergence (dmax; Figure 8, panels A and B) and with the axis of genetic variance reduction during evolution (emax; Figure 8, panels C and D), but not with ancestral plasticity (pmax; Figure 3) or mutational variance (mmax, from Mallard et al. G3 2023). pmax, for example, is aligned with g3, the third eigenvector of the ancestral G matrix. Note, however, that we do not have any power to detect the influence of g2 or g3 on phenotypic divergence or genetic divergence (Figure 8), though they together explain about 15% of the genetic variance. This is because performing such a test would require an alignment of the deviations in divergence not explained by gmax with g2 or g3. We now mention this issue in the Discussion. Overall, however, there are clearly several behavioral traits.

Author Response

Reviewer #1 (Public Review):

The manuscript by Zheng et al. examined the disease-causing mechanisms of two missense mutations within the homeodomain (HD) of CRX protein. Both mutations were found in humans and can produce severe dominant retinopathy. The authors investigated the two CRX HD mutants via in vitro DNA-binding assay (Spec-seq), in vivo chromatin-binding assay (ChIP-seq), in vivo expression assay of downstream target genes (RNA-seq), and retinal histological and functional assays. They concluded that p.E80A increased the transactivation activity of CRX and resulted in precocious photoreceptor differentiation, whereas p.K88N significantly changed the binding specificity of CRX and led to defects in photoreceptor differentiation and maintenance. The authors performed a significant amount of analyses. The claims are sufficiently supported by the data. The results not only uncovered the underlying disease-causing mechanisms, but also can significantly improve our understanding of the interaction between HD-TF and DNA during development.

Thank you for summarizing the key findings and strengths of our manuscript.

Minor concerns:

1) The E80A, K88N and R90W (previously reported by the same group) mutations are located very close to each other in the homeodomain (Figure 1A), but had distinct effects on the activity of CRX. Has the structure of the homeodomain (of CRX) been resolved? If so, could the authors discuss this phenomenon (mutations close to each other but have distinct effects) based on the HD-DNA structure?

In paragraphs 2, 4, 5 of the discussion section, we have added explanations on how each mutation could affect CRX HD-DNA interactions differently based on published structural studies. And we further explain how these biochemical changes relate to the molecular perturbations and cellular phenotypes seen in vivo.

In addition, has this phenomenon been observed in other homeodomain TFs?

Disease associated missense mutations at residues HD50 (K88) and HD52 (R90) have also been reported in other HD TFs implicated in CNS development (see discussion paragraph 7). Distinctively, different substitutions at CRX E80 residue have been reported in multiple CoRD cases, suggesting its essential role in HD-DNA-mediated regulation during retinal development. These new points are now included in the discussion section.

2) The authors should briefly summarize the effects/disease-causing-mechanisms of all the reported CRX mutations in the discussion part. The readers can then have a better overview of the topic.

We have added a concise summary of previously proposed CRX mutation classification scheme, all characterized Crx mutant mouse models and their pathogenic mechanisms. Please see paragraph 9 in the discussion section.

3) CRX can also function as a pioneer factor (reported by the same group). Would these HD mutations distinctively affect chromatin accessibility (which then leads to ectopic binding on the genome)?

Prior evidence has demonstrated that regulatory regions for many photoreceptor genes failed to stay accessible upon loss of CRX in the Crx-/- model (PMID: 30068366). It is unclear with the existing data whether CRX could initiate the chromatin remodeling (true pioneering function) of these regions, or it simply maintains the accessibility once these regions became accessible. Future studies comparing epigenomic landscape changes in mutant Crx KI models at various ages can be informative, particularly for the CRX K88N ectopic binding events. Determining how the CRX K88N mutant protein alters chromatin landscape important for photoreceptor fate and/or differentiation during development would shed light on the nature of these ectopic binding events.

4) The discussion part can be shortened and simplified.

We have re-written the discussion section to make it concise and to incorporate discussions on mutant CRX HD structures. Please see the revised manuscript.

Reviewer #2 (Public Review):

Zheng et al., investigated the molecular and functional mechanisms of two homeodomain missense mutations causing human retinal photoreceptor degeneration diseases in photoreceptor development regulated by the CRX transcription factor. They analyzed the E80A mutation associated with dominant cone-rod dystrophy (CRD) and the K88N mutation associated with dominant Leber Congenital Amaurosis (LCA). The authors found that E80A CRX binds to the same target DNA sites as WT CRX, but the binding specificity of K88N CRX is altered from that of WT in an in vitro assay. They generated Crx(E80A) and Crx(K88N) KI mice and performed ChIP assay and observed that K88N CRX binds to novel genomic regions from the WT-binding sites, while E80A binds to the WT sites. In addition, using the KI mice, they found that E80A and K88N differently affect the expression of Crx target genes. This study is well executed with proper and solid methodologies, and the manuscript is clearly written. This study gives us the insights how single missense CRX mutations lead to different types of human retinal photoreceptor degeneration diseases.

We greatly appreciate the reviewer’s summary and positive comments.

While the study has strengths in principle, it has a couple of weaknesses. One is how well E80A KI mice function as a pathological model of dominant CRD, in which cones are mainly first affected, is not clearly shown in this study. More data investigating how cones are affected by performing histological, molecular, and physiological analyses will be helpful and useful. For example, in the Discussion, the authors describe that E80A associates with S-cone opsin promoter results is "data now shown". This data must be presented for the readers. In addition, more molecular insights as to how E80A affects cones will strengthen this study.

The mouse retina is rod dominant and contains only a small number of cones (3% of all photoreceptors) that are born prenatally. This poses technical challenges to appropriately assess cone-specific changes during disease initiation/progression. We are in the process of developing cellular/molecular tools to investigate how cones are being affected in Crx E80A KI model, but this is beyond the scope of the current study.

At the same time, we have added a supplemental panel showing that, based on P0 retinal immunostaining of the early cone marker RXRγ, cones were initially born, and fate specified in CrxE80A retinas (see Figure S7A). Since the E80A protein also hyper-activated S-cone opsin promoter-luciferase (Sop-luc) reporter in HEK293 cells (see Figure S7B), we predict that CRX E80A affects cone photoreceptor differentiation in a similar manner as rod photoreceptors. Furthermore, the cone transcriptional program might be more prone to perturbations by abnormal CRX activities. These possibilities require future investigations. For this manuscript, we have included all these points in the discussion section.

Another point is that it will be very valuable if the authors could show how E80A and K88N differently affect the 3D structure of the CRX homeodomain. Even a simulation model would be valuable.

Please see our answer to Point 1 of Reviewer #1. In short, we have added in the discussion section our explanations on how each mutation could affect CRX HD-DNA interactions differently based on structural studies. We further explain how these biochemical changes relate to the molecular perturbations and cellular phenotypes seen in vivo. Additionally, since TF-DNA interactions are diverse and dynamic across binding sites with different sequence features and genomic environments, future studies that systematically and quantitatively evaluate CRX transcriptional activity at different regulatory sequences would be important.

Author Response:

We thank the reviewers for their insightful comments and will resubmit a revised version where we address most of the issues raised. At this time, our immediate responses are as follows.

1. We have data to confirm the presence of the merodiploid strain by PCR but did not show the data in the original version for brevity. We will show that data in the revised version.

2. We also have, of course, a no ATC control in our CRISPRi experiments and will also show that data in the resubmission.

3. As a loading control for the SecA2 strains, we will show PknG blots (a protein secreted by SecA2;PMID: 29709019) that we have with us.

4. In the nanoluc assays, the construct we made that was fused to CFP10 was generated so that there was a long linker between the C-terminus of CFP10 and nanoluc. We also have other controls in that experiment to show that the CFP10-nanoluc protein was secreted in the ΔRD10 strain and not in the ΔSecA2 strain. We will attempt to show fusion protein secretion using CFP10 antibodies in the revised version of the manuscript.

5. We will perform experiments with the inhibitor using the merodiploid strain and in partial knockdown strains to confirm that the inhibitor does indeed specifically act on Rv1636.

6. We will modify the discussion to talk more about the role and processes of cAMP synthesis and degradation in the revised version of the paper. Further, the manuscript will be checked for spelling and grammatical errors before resubmission, and the arrangement of data modified as suggested by the reviewers.

Author Response

Reviewer #1 (Public Review):

This manuscript describes the differences in the plasma proteome and metabolome in healthy Tanzanian and healthy Dutch adults. The inflammatory plasma proteome was measured using the Olink 92 Inflammation panel, while the plasma metabolome was analyzed using a mass spectrometry-based untargeted approach. The plasma metabolome was measured only in the Tanzanian cohort. This study aimed to link the pro-inflammatory proteome of Tanzanian and Dutch healthy individuals with environmental factors and dietary lifestyles.

The correlation between the plasma proteome and food-derived metabolome profiles can shed light on the development of non-communicable diseases. This observation stresses the importance of dietary transition and lifestyle changes in expressing inflammation-related molecules. Moreover, this study describes the inflammatory proteome profile in healthy Tanzanian individuals covering a cohort with limited studies. The molecular differences in circulating biomolecules between healthy individuals living in East Africa and individuals living in Western Europe and the correlations with intrinsic and environmental features are novel.

This study lacks a robust and solid validation of some of the differentially regulated circulating proteins and correlations between food-derived metabolites and proteins in a selected cohort. The discovery-driven approach in this manuscript highlights potential findings that need to be supported by a validation phase. According to this reviewer, the lack of such validation impacts the robustness of the results and the hypotheses generated. Due to that, the manuscript should incorporate validation experiments.

We acknowledge that our study was limited by the lack of a validation phase. To address this issue, we have undertaken additional analyses to validate our key findings related to the proteins associated with mTOR and Wnt/β-catenin pathways. These analyses involved data from a proof-of-concept intervention study conducted at the same site. Our response below provides more information on these validations.

Author Response

Reviewer #1 (Public Review):

For PRLR, the question being asked is whether and how the intracellular domain (ICD) interacts with the cellular membrane or how the disordered ICD can relay and transmit information. The authors show that PI(4,5)P2 in the membrane localizes around the transmembrane domain (TMD) due to charge interactions and facilitates binding of the ICD to the membrane, even in the absence of the TMD. Furthermore, the ICD and PI(4,5)P2 form a co-structure with JAK2 which locks a disordered part of the ICD into an extended conformation, allowing for signal relay and, through multiple complex conformations, may enable switching signalling on and off.

Strengths:

• NMR paired with MD is a powerful way to probe an interaction especially when peaks disappear and become difficult to probe by NMR.

• Using NMR and MD to formulate hypotheses which are then tested by cell studies is quite informative. The combination of MD, NMR, and cell biology is a strength.

• The authors are diligent in testing MD simulations on systems with and without PIP2.

• The use of Pep1 and Pep2 to differentiate the KxK region that interacts with PIP2 is helpful.

• The four utilized mutants help illustrate the co-dependence of the respective regions in the formation of the co-structure.

Weaknesses:

• In Figure 2G, there is a big change in CSP between 280 and 290, which the authors do not comment about.

The region referred to contributes to binding but is on the edge of the main binding site and where the local affinities are weaker. Therefore, the exchange rate is high and allows for following the chemical shift changes. In support of this, we see an almost inverse correlation between the CSPs and the changes in intensities. For the main binding site, the exchange rate between bound and free states is slower because the affinity is stronger. Therefore, we cannot follow the chemical shifts to extract the CSPs to the bound state, as the peaks disappear. We have commented on this in the main text (p.8) as follows:

“In the region from D285-E292 we observed an almost inverse correlation between the CSPs and the intensities. This suggests that in contrast to the preceding region, a faster local exchange rate allows us to follow the resonances from the bound state in this region, giving rise to the large CSPs.”

• The data in Figure 2 are summarized as indicating the formation of extended structure in the ICD upon binding. It is not clear to me what data show an extended structure.

The information on the extended structure comes from the analyses of the peptide Pep1 titrated with C8-PI(4,5)P2. The CD signature that develops in the bound state has a minimum ellipticity at 218 nm, which is a strong indicator of extended structure. We find this information adequately described in the main text (p.8), but have emphasized this further as follows:

“In contrast, for Pep1, large spectral changes were seen, which were unrelated to helix formation. Subtracting the spectra in the presence and absence of C8-PI(4,5)P2, revealed a negative ellipticity minimum at 218 nm, a strong indication of B-strands, showing that when bound to C8-PI(4,5)P2, a distinct extended (strand-like structure) signature was seen (Figure 2G).”

• No modelling or experiments were done with PIP3 despite conclusions and models which rely on the phosphorylation of PIP2 to PIP3. At the very least, these would be useful as negative controls.

We have in a previous work addressed the affinity for phosphoinositides using lipid dot blots where we observed a preference for certain species, including PI(4,5)P2 (Haxholm et al., BJ, 2015). In this study, we also observed that there was no affinity for PI(3,4,5)P3, but may not have highlight this sufficiently in the introduction. This can have caused some confusion in understanding our choices. We have now more explicitly described these data, both in the introduction (p.4), in the result section (p. 8) and later in the discussion (p. 21). We thank the reviewer for bringing this up.

• Only R2 experiments were done when the authors mention investigating dynamics. R1 and -HetNOE dynamics would be useful for creating a complete picture.

Our aim with recording the R2 values was not to map the detailed dynamics of the disordered regions, but to explain the changes in the peak intensities we see for the variants when adding C8PI(4,5)P2. In this case, the R2 values supported our suggestion of internal contacts and, although we agree with the reviewer that R1s and HetNOEs would be important and relevant for a more in-depth and complete analyses of the dynamics, we find that in this case, the R2 values suffice.

• Some of the exciting results are under-emphasized including Fig 3H and 3I.

A new version of Figure 3 has been generated to consider the reviewers’ comments and suggestions. This figure has been restructured to further emphasize some of the major conclusions obtained from the simulations. We have moved the former Figure 3 A, B, C and D to the supplemental information to increase this focus.

Reviewer #2 (Public Review):

The authors combine NMR experiments, cell experiments, and molecular simulations to address the question of how lipid interactions of the prolactin receptor contribute to signalling. They assess the interactions of the disordered cytoplasmic tail of the receptor with phosphoinositides among others by chemical shift perturbations from NMR for different PIP2-containing membranes, by coarse-grained simulations, as well as site-directed mutagenesis and subsequent cell signalling experiments to monitor the activation of the mutants. A major result is that PIP2 interactions are functionally important, which so far has not been known for this receptor. Their results are likely relevant for other non-receptor tyrosine kinases.

The hypothesis that the protein complex is regulated by IDR-membrane interactions is very novel. A major strength is the close connection of and feedback between state-of-the-art experiments and simulations.

We thank the reviewer for the positive comments on our work and on the novelty and importance of the work

This is where I see weaknesses:

1) The motivation of focusing on LID1 is limited.

We have now provided our rationale for selectively focusing on the LID1 in the PRLR. The selection was done to address the conundrum on how structural disorder in the juxtamembrane regions would be able to transmit the knowledge of extracellular hormone binding to the bound JAK2. This constitute the first step of signaling on the intracellular side and given the distance to the other two LIDs (LID2 and LID3) and their disconnect to the TMD by long disordered regions, they were disregarded, focusing on LID1 in this work. We have emphasized this choice in the introduction and in more detail in the result section (p. 5-6).

2) The data and analysis for the JAK2-PRLR complex appear somewhat superficial, and a connection between conformational states to their functional relevance is lacking. In fact, the majority of the simulation part of the paper is about suggesting different states of the PRLR-JAK2 complex but the states and their hypothesized functional relevance are not further taken up, e.g. by experiments, and yet presented as major results, e.g. in the abstract.

In the original manuscript we already provided a detailed analysis of the different states, highlighting accessible residues and lipid interacting residues and compare these across the states. From our experiment, including those performed in cellular assay, we cannot with certainty link the two major state to active and/or inactive states. We have therefore no intention or support from the data to claim this. However, what we do put forward as a major result, in the presence of more than one major state as also stated in the abstract and in the conclusion of the result section as follows:

“Another key observation is the existence of different states in which different regions of both JAK2 FERM-SH2 domain and LID1 of PRLR are exposed to the solvent or hidden below the bilayer.”

In the discussion we do speculate as to which state may be the active and/or inactive dimer/monomer but make no firm claims. We have now made the major find of more states clearer in the text, and further compare the two major states, the Y and the Flat state, to the resent cryo-EM structures of JAK1 bound to IFNAR1, which lend some support to our speculations. The abstract now reads:

“We find that the co-structure exists in different states which we speculate could be relevant for turning signalling on and off.”

To discern the functional relevance of these state, if possible, will require experiments also in cells that by themselves would be a new study. We have to the best of our ability clarified that the functional relevance of the states has not been elucidated by the current work.

3) The connection between simulations and mutational study is not very direct. An open question is if the mutants can distinguish between the effects of PRLR-PIP2 interaction or PRLR-JAK2 interaction, even though this conclusion is still drawn from the data.

We have now explained in much more detail by which arguments the different mutations were selected (see also answer above), which property of the co-structure they are most likely to engage in and affect, and we have emphasized that the separation of function by mutation may be complicated by the intimate structure formation among the three components of the co-structure. The conclusion has therefore also been softened.

4) The conclusions drawn from the mutagenesis study (lines 547-555) are not directly supported by data. Only a partial correlation between PRLR membrane localisation and STAT5 activation is no reason to attribute the unexplained part of the STAT5 activation to PRLR-JAK2 interactions without further studies.

We have now explained in much more detail by which arguments the different mutations were selected (see also answer above), which property of the co-structure they are most likely to affect and emphasized that the separation of function by mutation may be complicated by the intimate structure formation among the three components of the co-structure. The conclusion has therefore also been softened.

5) PIP2 is identified as an important regulator, with very solid support from the presented data. PIP3 is part of the model but not discussed before or as part of the results. The analysis could be similarly applied or the data directly relevant to the understanding of PIP3 plays a similar role, as interactions are likely primarily electrostatically driven.

We have in a previous work addressed the affinity for phosphoinositides using lipid dot blots where we observed a preference for certain species, including PI(4,5)P2 (Haxholm et al., BJ, 2015). In this study, we also observed that there was no affinity for PI(3,4,5)P3, but we agree that we did not highlight this sufficiently in the introduction. This have caused some confusion in understanding our choices. We have not more explicitly described these data, both in the introduction (p.4), in the result section (p. 8) and later in the discussion (p. 21). We thank the reviewer for bringing this up.

Reviewer #3 (Public Review):

Araya-Secchi and coauthors present a very interesting study on the role of PIP2 lipids in the potential modulation of prolactin receptor signaling. The study is well-conducted and employs an integrated approach that combines NMR spectroscopy, modeling (primarily coarse-grain MD simulations), and cell biology. This combination of methods is crucial for gaining a deeper understanding of cell receptors, from their biophysical properties to their cellular functions.

The modelling work is mainly based on both coarse grain forcefield versions Martini2.2 and Martini3. These two versions of the forcefield may produce different results. Therefore, depending on the system being modeled, the results presented here should be considered in light of the limitations inherent to each version of the forcefield.

We thank the reviewer for the positive appraisal of our work and the approach we employed. It is true that one must be aware of the limitations of the tools and models employed in this type of work. We agree that perhaps we were not too explicit about limitations of our methods in the presentation of the results. However, we have addressed and discussed such limitations in the revised version of the manuscript.

Author Response

Reviewer #1 (Public Review):

This study demonstrates that Chinmo promotes larval development as part of the metamorphic gene network (MGN), in part by regulating Br-C expression in some tissues (exemplified in the wing disc) and in a Br-C independent manner in other tissues such as the salivary gland. I have included below the following comments on the submitted version of this manuscript:

1) The authors have shown experimentally that Chinmo regulates Br-C expression in the wing disc but not the larval salivary gland. Based on this, they posit that Chinmo promotes larval development in a Br-C-dependent manner in imaginal tissues and a Br-C-independent manner in other larval tissues. This generalization of Chinmo's role in development would be more compelling if the relationship between Chinmo and Br-C were explored in other examples of imaginal/larval tissues.

We agree with the referee that confirmation of our observations in other tissues might help to generalize Chinmo’s role. To this aim, we have analyzed the role of chinmo in an additional larval, the larval tracheal system, and imaginal tissue, the eye disc. Consistent with the results reported in the manuscript, we found that the mode of action of Chinmo is conserved, as depletion of Br-C in the eye disc is able to rescue the lack of chinmo, whereas in the tracheal system it is not. We included this new information in the main text and in new SFigures 1 and 3.

2) Chinmo, Br-C, and E93 have all been shown to be EcR-regulated in larval tissues, including the brain and wing disc (as in Zhou et al. 2006, Dev Cell; Narbonne-Reveau and Maurange 2019, PLOS Biology; Uyeharu et al. 2017, ). It would be interesting (and I believe relevant to this study) to know whether the roles of these factors in their respective developmental stages are EcR-dependent and whether their regulation by EcR (or lack thereof) depends on whether the tissue is larval or imaginal.

Although the relevance of EcR on the regulation of the genes that conform the metamorphic gene network has been already established, a different response of EcR-mediated signalling of these genes in larval and imaginal tissues is still not properly addressed. Finding this possible different output of the EcR signalling would be very interesting. However, we think that this is out of scope of this report as the main aim of this study was to determine the main role of the temporal genes during development and their repressive interactions.

3) In the chinmo qPCR analysis shown in Fig1A, whether animals were sex-matched or controlled was not indicated. Since Chinmo has a published role in regulating sexual identity (Ma et al. 2014, Dev Cell; Grmai et al. 2018, PLOS Genetics), and since growth/body size is known to be a sexually dimorphic trait (Rideout et al. 2015, PLOS Genetics), it seems important to establish whether the requirement of Chinmo for larval development and/or growth. I recommend either 1) controlling for sex by repeating qPCRs in Fig 1A in either males or females, or 2) reporting male/female chinmo levels at each stage side-by-side.

As the referee pointed out, chinmo has been related to sexual identity raising the possibility of a different effect of chinmo in growth of males and females during development. However, several observations discard this option. First of all, the role of chinmo in sexual identity has been only reported in adult testis and specifically in cyst stem cells. In fact, specific mutations of chinmo that only affects the expression of chinmo in testis, do not affect testis formation but its maturation, suggesting a role of chinmo in sex determination specifically in the testis cyst stem cells (Ma et al. 2014, Dev Cell; Grmai et al. 2018, PLOS Genetics). Second, it has been described a sex dependent growth rate during larval development (Rideout et al. 2015, PLOS Genetics; Sawala A. and Gould AP, PLoS Biol, 2017). However, the main difference in growth rate between males and females is found in L3 larvae (Sawala A. and Gould AP, PLoS Biol, 2017), when the expression of chinmo strongly declines in both males and females, indicating that chinmo impact on sex dimorphism during larval development might be at least, limited.

Thus, considering that, based on our results, chinmo exerts its main role in larval tissue growth during L1 and L2 stages and that body growth is practically identical in male and female during these stages (Sawala A. and Gould AP, PLoS Biol, 2017), we can assume that chinmo might not contribute to sexual body size dimorphism.

Nevertheless, we would like to clarify that we have performed the measurements of chinmo expression always in females, when sex identification was possible, namely in L3 larvae. L1 and L2 larvae qPCRs were not sex-discriminated as sex identification was not possible in our conditions.

4) In Fig2E, the authors show that salivary gland secretion (sgs) genes are repressed in salivary glands lacking chinmo. Sgs genes are expressed during late larval stages as the animal prepares to pupate. Thus, based on the proposed model where Chinmo promotes larval development and represses the larval-to-pupal transition, one might expect that larval salivary glands lacking chinmo would express higher than normal levels of sgs genes. This expectation directly opposes the observed result - it would be helpful to speculate on this in the interpretation of results.

This is an interesting observation. As Sgs genes are regulated by Br-C (Duan et al. Cell Reports 2020), precocious expression of this transcription factor in chinmo depleted animals might result in an early activation of those genes. Interestingly, we were not able to detect any Sgs genes expression in chinmo depleted salivary glands. We think that this is due to the fact that in absence of chinmo, this organ does not properly develop and mature, and therefore it is unable to express Sgs genes. Proof of that is that the double knockdown of Br-C and chinmo shows the same dramatically low levels of those genes. Altogether, these results strongly suggest that SGs lacking chinmo expression are unable to grow and synthesise Sgs proteins, even in the premature presence of Br-C. We discussed this point in the main text of the edited Ms. Please also see the response to referee 2.

Reviewer #2 (Public Review):

The evolution and control of the three-part life history of holometabolous insects have been controversial issues for over a century. While the functioning of broad as a master gene controlling the pupal stage and of E93 as a master gene for the adult stage has been known for about a decade or more, chinmo has only recently been proposed as being the master gene responsible for maintaining the larval stage (Truman & Riddiford, 2022). While the former paper focused on the embryonic and early larval function of Chinmo, this paper explores its metamorphic effects and defines the roles of Broad and E93 in the phenotypes produced by manipulations of Chinmo expression.

Overall, the paper is well presented but in places, readers would be helped if the authors were more explicit about the logic and details of their manipulations. There are a couple of conceptual issues that the authors should address.

The role of Broad in larval tissues:

One intriguing issue relates to the relationship of Chinmo to Broad and E93 in larval versus imaginal tissues prior to metamorphosis. The knock-down of chinmo in imaginal discs results in severe suppression of growth and the lack of metamorphic patterning genes such as cut and wingless. Normal growth and patterning are reestablished though, if broad is also knocked-down, supporting the notion that the effects of the lack of Chinmo are mediated through the premature expression of Broad.

In the salivary glands, by contrast, chinmo knock-down suppresses growth, and this growth suppression is not reversed by simultaneous broad knockdown. They properly conclude that the role of Chinmo in supporting the growth of larval tissues does not involve Broad, but their data on the expression of salivary gland proteins suggest that Broad still plays some role in Chinmo function in salivary glands. Fig. 5E shows the levels of various salivary glue proteins in the glands of Chinmo knock-down larvae. The levels are reduced, as expected by the lack of salivary gland growth, but a significant finding is that they are there at all! The Costantino et al. (2008) paper shows that these genes are only induced in the mid-L3. Ecdysone, acting through Broad isoforms, is necessary for their appearance and these SGS genes can be induced in the L1 and L2 stages by ectopic expression of some Broad isoforms. Their low levels in Fig 5, would be due to the small size of the gland, but the gland's premature expression of Broad likely causes their induction. In larval cells, then, Chinmo may feed into two parallel pathways, one that does not involve broad and regulates growth and the other, utilizing Broad, regulating premetamorphic changes.

It would be useful to look at early larval salivary gland proteins such as ng-1 to -3 that are expressed in salivary glands before the critical weight. Also, it would be interesting if the appearance of the SGS proteins after chinmo knock-down (Fig 5E) is abolished by simultaneous knock-down of broad.

This is an interesting observation. We think that the main problem has derived from the way we presented the data. Our results showed that depletion of chinmo in the SGs dramatically impairs the induction of Sgs gene expression, even with the premature presence of Br-C, which has been shown to be responsible for Sgs expression (Duan et al. Cell Reports 2020). The confusion might come from the way we presented the level of expression of those genes. In fact, the levels of Sgs in both chinmoRNAi and chinmoRNAi/Br-CRNAi SGs were virtually undetectable, suggesting that chinmo in the SG is not only required for Br-C repression but also for proper development of the gland. We believe that based on the fact that the very low levels of expression of Sgs genes in chinmo depleted SGs are still detected in the double knockdown chinmoRNAi/Br-CRNAi. Dramatically reduced expression of the early larval SGs ng1-3 genes in chinmoRNAi and double knockdown chinmoRNAi/Br-CRNAi supports this statement. Altogether these results suggest that Br-C is necessary but not sufficient for the expression of those specific SGs genes. We have changed the plots in Figure 2 and 3 to clarify this point and added the levels of expression of ng1-3.

Role of Chinmo and Broad in Hemimetabolous insects:

In the conclusion of their comparative studies on the cockroach (line 342), the authors state that Broad exerts no role in the development of hemimetabolous insects. However, this conclusion is not consistent with the literature. The first study of broad knockdown in a hemimetabolous insect was in the milkweed bug Oncopeltus fasciatus by Erezyilmaz et al. (2006). Surprisingly to Erezyilmaz et al., broad knock-down in early-stage nymphs did not cause premature metamorphosis. However, Broad expression was essential for tissues of the wing pads and dorsal thorax to undergo morphogenetic growth (rather than simple isomorphic growth), and for stage-specific changes in coloration through the nymphal series (but not for the nymph to adult color change). A similar function for Broad on wing growth during the later nymphal stages was later shown in Blattella (Fernandez-Nicolas et al., 2022; Huang et al., 2013). The wing- and genital pads represent "imaginal" tissues in the nymph and the need for Broad in these tissues are the same as seen in imaginal discs as the latter shift from isomorphic growth to morphogenesis at the critical weight checkpoint in the L3. This would suggest that important roles for Broad and E93 are already established in the hemimetabolous insects with E93 controlling the shift from immature (nymphal) to adult phenotypes and Broad controlling the premetamorphic growth of imaginal tissues in early-stage nymphs. Chinmo might then be needed to keep both in check.

We are sorry for not having dealt with these observations in the submitted manuscript. We have taken them into consideration in the new version to discuss about the role of Br-C in the transition from hemimetabolous to holometabolous.

Author Response

Reviewer #1 (Public Review):

The authors study single and pairs of MDCK cells adherent to an H-shaped geometry on a flat surface. In this pattern, the cells form strong peripheral stress fibers. To a lesser extent, these cells also exhibit stress fibers in the cell interior, which otherwise has a rather homogenous actin distribution. Using a combination of traction force microscopy, from which they infer the stress distribution by monolayer stress microscopy, and "contour analysis" the authors quantify the 'bulk' and the 'surface' stress in these cells. This analysis shows that single cells are mechanically polarized whereas pairs are not.

The authors then go on to optogenetically activate the actomyosin contractility of either one half of a single cell or one cell of a pair. Combining their stress measurements in these situations and using a finite element mechanical model, the authors convincingly show that the mechanical response in the non-activated part is active. By varying the aspect ratio of the adhesion patterns, they also find that the efficacy of active stress propagation depends on the mechanical and structural polarity of the cell. Furthermore, they provide evidence that their results on cell pairs generalize to tissues.

Strengths:

This study uses a nice combination of physical tools to address an important question in tissue mechanics. The data is compelling and fully supports the authors' conclusions.

Weaknesses:

There are no major weaknesses.

In summary, although the fact that mechanical stress propagation in tissues is an active process might not come as a surprise, the study makes substantial contributions to a quantitative contribution of this process. As such it is of fundamental significance in the field. It will be interesting to explore the consequences of this mechanism for mechanical stress propagation in the context of developmental processes. It will be also of great interest to study how this local process can be accounted for in large-scale theories.

We thank reviewer #1 for this very positive assessment. We agree that in the future, our results should be used on the theory side to upscale them to tissue level. One way to do this would be the discontinuous Galerkin method, but it will take time to work this out. We also note that we would have loved to experimentally study intermediate cases between two and many cells, but it turned out to be very difficult to position few cells on micropattern and to repeat the force propagation analysis which we present here for two cells and for small tissues. In fact, it might be more rewarding to use optogenetics early in a developmental process with clearly defined cell positioning. In the revised manuscript, we now have added a comment on the challenge to work with three or four cells with the micropatterning approach, and that therefore we turned to small monolayers.

Author Response

Reviewer #3 (Public Review):

In this study, the authors probe the molecular changes that occur in a neural circuit for learned behavior that depends on sensory input to maintain stereotypy. Songbirds, as the Bengalese finches used here are, are premier systems in which to ask these questions because they produce a highly stereotyped song that emerges after sensory learning becomes integrated into the function of a sensorimotor neural circuit responsible for singing. By deafening a group of birds (who show a shift in their song structure) and comparing them to hearing birds, clues as to how plasticity in motor output may emerge from genomic changes that alter the function of cells within the various components of the neural circuit.

There are multiple strengths of the paper:

1) The results may have broad implications because the type of sensorimotor neural circuit (cortico-basal ganglia-thalamic-cortical loop) used for singing is generally necessary for learned behaviors.

2) The methods and analyses are generally rigorous, including the parsing of song elements, and the type of detailed RNA sequencing and analysis that demonstrates the power of a genomic view of neural plasticity as it relates to behavioral plasticity.

3) Because the authors assayed the pallial (cortical) areas, as well as the basal ganglia component, of the sensorimotor circuit they were able to creatively compare how different facets of the network contributed to a) unmodified brain properties, b) properties perturbed after the loss of the auditory input that is required to stabilize song structure. As a result, they have added to the known molecular profiles for each of these brain areas, the accounting of how they may be specialized in comparison to the surrounding non-song brain, and what changes occur after deafening. Utilizing some existing single-cell sequencing data, the authors present for the first time some insight into what cell types may be showing the most robust changes, and therefore which may be driving the shift in song structure. The analysis further pushes in new ways to suggest how the molecular properties of a given brain area may relate to those of directly-connected areas. Together, these findings provide valuable clues as to the specific cell types and signaling properties that may be central to the production of stabilized, learned behavior.

4) One of the cortical brain areas, LMAN, was lesioned in a subset of the hearing subjects because it projects to the area that showed the greatest molecular difference between deafened and hearing birds (RA). The idea was to compare how this affected molecular properties with properties after the loss of auditory input; because RA is the output motor area for the song, its properties may be most directly tied to song structure. Using unilateral lesions was a strong choice of experimental design that allowed for rigorous analysis of this idea, and was interpretable because birds do not have a direct inter-hemispheric callosum.

The foundation of the paper is solid, though the results shown raise several questions that are not fully addressed, and limit some of the power of the implications.

The biggest questions arise from the finding that RA shows the largest number of molecular changes after deafening. The analysis and interpretations do not fully incorporate what we know of this circuit, at least from another well-studied songbird, the zebra finch, from which the authors derive other types of information. For example, it is not yet clear if RA is most changed because it is most directly involved in song output or because it receives projections from two areas within the sensorimotor circuit (LMAN and HVC). How do we consider the fact that by adulthood, LMAN and HVC cells project onto the same RA neurons? Are those the cell types being identified here? Would HVC lesions be expected to have the same effect as LMAN lesions? Are the cell types showing the greatest change those that are most involved in song output (e.g. are they projecting to nXIIts)? How do these results relate to the findings of changes in RA after HVC and LMAN lesions reported decades ago? How do these findings compare to an earlier study that also performed sequencing on areas from the sensorimotor circuit in deafened juveniles? Further, RA also receives information from the auditory processing regions of the brain, via the immediate structure RA-cup. It is not yet explicitly addressed how some effects may be from the loss of this more direct access to auditory information, rather than from information and projections originating within the sensorimotor circuit, and reinforces the question of whether or not the number of inputs to a particular brain area is a driving factor in the general pattern of changed RNAs after perturbation.

We thank the reviewer for their review and for their excellent suggestions on how to improve its impact. The reviewer raises several important points, which we have expanded on in the Discussion of the revised manuscript, and will address here:

First, there is the general consideration of how the structure of inputs to RA influences the interpretation of our results. There is the question about whether we can consider RA expression alterations as due to its direct projections to song motoneurons (‘output’) or the convergence of two important song nuclei, HVC and LMAN, onto RA (‘input’). This is a difficult question to untangle. We could interpret ‘output’ only effects as local perturbations that do not depend on song circuit afferent activity, such as hormonal fluctuations associated with the loss of hearing. ‘Input’ effects would occur through changes in afferent activity, such as those that elicit plasticity associated with song destabilization or more general alterations to the amount of afferent neural activity (a point addressed in the revised manuscript, lines 842-848). By focusing on a measure of song destabilization in our differential expression analysis, we are specifically seeking to identify gene expression responses that are associated with changes to behavioral output. Yet these behavioral changes are certainly driven by alterations in upstream regions or the manner in which they converge onto RA. The reviewer also notes inputs from RA-cup as a potential avenue through which the loss of auditory information could more directly influence expression in RA. It is certainly possible that the loss of auditory information itself could influence gene expression in different components of the song system, a point we note in the revised Discussion (lines 848-853). We also note there that future experiments leveraging different plasticity induction techniques (TS cut, delayed auditory feedback) will be important to resolve the influence of this input.

Our lesion experiments aimed to characterize how input from LMAN influences expression in RA, due to LMAN’s important role in mediating song plasticity. We would expect HVC lesions to elicit different expression responses because of its distinct mode of transmission onto RA projection neurons (primarily AMPAR in contrast to primarily NMDAR for LMAN), the distinct activity patterns of HVC and LMAN, and likely distinct neuromodulatory signaling from the two afferents (e.g. LMAN acts as source of BDNF). We discuss how HVC lesions would be useful to further disentangle the influence of afferent input on RA gene expression in the Discussion of the revised manuscript (lines 926-946). In the revised manuscript, we also cite previous work that examined the influence of HVC and LMAN on RA neural activity, morphology, and cell survival (lines 928-932).

As to the cell types in RA that show expression changes following deafening, we show in Figure 5 that both glutamatergic projection neurons (‘RA_Glut’), i.e. the neurons that project to subcerebral structures such as nXIIts, as well as GABAergic interneurons (‘GABA’) show substantial expression alterations. In the Discussion, we highlight the functional roles of several genes that have enriched expression in each class (lines 864-873 and 887-893).

In the revised manuscript, we have added a paragraph in the Discussion (lines 854-862) that references results from Mori, C. & Wada, K. Audition-independent vocal crystallization associated with intrinsic developmental gene expression dynamics. J. Neurosci. 35, 878–889 (2015). This work examined the influence of early deafening on gene expression in the song motor pathway and identified a strong developmental and audition-independent expression response. It identified an important separation between developmentally-driven and experience-dependent molecular responses in the song system. We note that the aims were distinct from the present study, which sought to identify gene expression responses to deafening-induced song plasticity.

Importantly, since the LMAN lesions did not create significant changes in the song structure, it is difficult to know how to interpret the meaning of these molecular changes in RA, alone and in combination with the comparison to the RA profiles from deafened birds. Of importance is the question of whether or which molecular profiles are thus signatures of behavioral plasticity or not.

The reviewer raises an important set of followup experiments that assess the extent to which the transcriptional state of the song system tracks with song plasticity state. Coupling LMAN lesions with deafening, a manipulation that prevents song degradation, would be a strong approach to identify genes whose expression is closely tied to song destabilization, a possibility that we now discuss (lines 936-946).

Author Response

Reviewer #1 (Public Review):

1) There are two main 'weaknesses'. The first is the limited power that comes from only using measuring the phenotype of 387 strains. Whether this is because of the expense/ difficulty of the inToxSa is not discussed, leaving open the question of how much this assay could be scaled up in the future.

A previous study investigating the toxicity of S. aureus culture supernatants assessed 217 clinical strains (https://doi.org/10.1371/journal.pbio.1002229).That study had sufficient power to uncover important genetic determinants of S. aureus virulence. Here, we significantly increased the throughput to 387 clinical strains combined with a sophisticated cell toxicity assay that measures the kinetics of cytotoxicity caused by intracellular S. aureus. We have investigated the S. aureus genetic associations using this rich dataset (each of the 387 strains were assessed in 3 to 15 replicates, accruing 655,005 measurements corresponding to kinetic cytotoxicity assessments of intracellular S. aureus). This rich dataset enabled the accurate identification of genomic signatures that modulate cytotoxicity; genomic signatures that we then validated by reconstructing the mutations, thus demonstrating the power of our approach. The upscaling of this method (4-fold, with adequate technical adjustments) should be possible with the adoption of a 384-well plate format instead of a 96-well plate. We will continue to investigate additional clinical isolates and explore the use of 384-well plates, but the analysis we present of data from the 96-well format is already a substantial advance for the field.

Across this study, and as presented in the current manuscript, the maximum throughput of the InToxSa assay was of 7x 96 well plates per week, thus corresponding to 98 distinct clinical strains testable per week (encompassing 6 individual replicates, each tested across 2 different days/plates). Following the reviewer suggestion, we have added this information to the discussion (Lines: 406-409).

2) The second is that the main output of the assay is actually reduced intracellular toxicity (PI uptake AUC), which is inferred to be strongly linked to increased intracellular persistence. The linkage between the phenotypes comes primarily from microscopic studies on a limited number of strains. It may be true of all cases, but the possibility exists that for some of the strains, reduced cytotoxicity may be associated with intracellular elimination, which would presumably be a negative outcome for systemic infection.

Whilst the reviewer’s comment is pertinent, we note that none of the least cytotoxic S. aureus isolates identified by the InToxSa assay have resulted from bacterial clearance, intracellular bacterial growth defects or evasion from their cellular niche, as we have assessed intracellular bacterial loads at 3h and 24h (post-bacterial uptake) in experimental conditions using cell-impermeant antibiotics (which would kill extracellular bacteria and prevent over-infection of non-infected bystander HeLa cells), as shown in figures 5F and 5H and also in Figure 5 Supplementary figure 5, highlighting an inverse correlation between cytotoxicity and intracellular persistence.

Reviewer #2 (Public Review):

1) …Thus, my concerns are focused on further understanding the practical utility of the approach and whether or not the HeLa cell model recapitulates what happens in professional phagocytes.

HeLa cells have proven a useful cellular model in infection and in pathogen biology to assess the ability of bacterial pathogens to invade, persist and replicate within host cells. Several studies have convincingly used HeLa cells to assess S. aureus phenotypes at the bacteria-host cell interface, as exemplified by the following recent research (DOIs: 10.1128/mBio.02250-20, 10.1371/journal.ppat.1009874, 0.138/s41598-019-51894-3, and 10.1128/mSphere.00374-18). We do also acknowledge the limitations of cell line models in the discussion (Lines 494-510).

2) …it is not clear to me that this system has the statistical power to find novel, biologically relevant rare mutations without first being very mindful in selecting strains that are extremely genetically similar.

As described, this is a S. aureus bacteraemia study, wherein the strains composing the collection are, by definition, closely related. We articulated this in the manuscript “We used InToxSa to identify S. aureus pathoadaptive mutations, enriched in bacterial populations that are associated with human disease (e.g., upon transit from colonising to invasive”. “We hypothesised that these mutations would support an intracellular persistence for S. aureus.”) We see no foreseeable reasons preventing this type of study of being replicated elsewhere.

3) It is also not clear to me that the toxicity assay captures the important features of the intracellular persistence that occurs in vivo within professional phagocytic cells.

Response: Indeed, it is possible that InToxSa using HeLa cells may not capture the features of intracellular S. aureus persistence within professional phagocytes. However, our data shows that it remains possible to uncover genomic features related to intracellular cytotoxicity and persistence, both traits relevant S. aureus-host cell biology. The cells forming physical barriers, such as the epithelial cells and endothelial cells play major roles in staphylococcal pathobiology. Whilst HeLa cells are a model cell line, their tractability makes them ideal for high throughput studies tested over longer infection times.

Author Response

Reviewer #1 (Public Review):

Mermithid nematodes are ecologically important parasitoids of arthropods, annelids and mollusks today. Their fossil record in amber reaches back into the Early Cretaceous, some 135 million years ago. Luo et al. more than triple this record by presenting, with ample illustrations, exceptionally well preserved new specimens from the beginning of the Late Cretaceous (99 Ma ago) of Myanmar. Their most important finding is that mermithids parasitized a number of insect clades in the Cretaceous that they are not known to infect today or in Cenozoic amber; further, the proportion of holometabolous insects among the hosts is found to be lower in the Cretaceous than in the Cenozoic. The strengths of the paper lie in the specimens, the illustrations of the specimens, and the documentation of when, where and how the specimens were acquired. Certain nomenclatural aspects of the paper require improvement. A potential weakness of the paper could be collection bias: it is not tested whether the collections used to show the shift toward holometabolous hosts from the mid-Cretaceous to the Cenozoic are representative of the fossil record as it is preserved and accessible today.

Thank you very much for pointing out these issues. We have added a new Figure 10 and Table 1 to our paper. Indeed, collection bias is almost present in all amber biotas. However, we believe we have robust reasons to argue that the shift to holometabolous hosts does exist. Although Kachin amber has only been studied extensively in the last two decades (compared with centuries of study in Baltic amber or Dominican amber), it has become by far the most intensively studied amber biota since its Cretaceous age was appreciated, now comprising an exceptional 700 families (Ross, 2023). Also, the fossil record of holometabolous insects is clearly much better than heterometabolous insects in Kachin amber (1296 spp. vs 465 spp. respectively). But as shown in our paper, the nematodes we found in Kachin amber are mainly associated with heterometabolous insects. Therefore, even if collection bias might exist, such as the presence of some unreported nematode-Holometabola associations, we believe our conclusion about the shift is robust. We also add some explanation in our paper.

Reviewer #2 (Public Review):

This manuscript reports on mermithid nematode fossils from amber which dates from the Cretaceous period. The specimens described in the manuscript consist of insects and associated nematodes which have been trapped in amber and fossilised. The nematodes have been identified as belonging to the Mermithidae family, a family of nematode worm that infect insects. The findings of this manuscript provide an insight into the evolution history of nematodes and parasitism. Despite the ubiquity of both nematodes and parasites in extant ecosystems, fossil records of both are very rare. This is because nematodes and many parasites are soft bodied, and many are located inside their hosts' bodies, thus they rarely become fossilised. Thus, most of what is known about the evolutionary history of nematodes, and evolution of parasitism are based on what could be inferred from extant examples.

The specimens described in this manuscript provides a valuable contribution to our understanding of parasitism in the geological past. These amber specimens are a snapshot of parasite-host interactions - interactions which are commonly found in nature but are rarely captured in fossils. The identification of the specimens as mermithid nematodes are based on sound scientific reasoning. The worms' morphology and position in relation to the insects are consistent with what have been observed with extant mermithid nematodes.

Additionally, one of the values of such parasite fossils is that they provide us with insight into parasite-host combinations or interactions which may have existed throughout the geological past, but no longer exist today or cannot be inferred from extant taxa. It helps fill in major gaps in our understanding of parasitism. This was the case with the amber fossil that contained a bristletail with its nematode parasite.

We are very grateful for the positive and encouraging comments.

Reviewer #3 (Public Review):

The authors provide a timely description of new mermithid nematodes from Cretaceous amber and use it to argue an important shift in insect host exploitation. The descriptions are state-of-the-art and will become valid once the appropriate zoobank numbers are used after publication. The authors also compiled crucial and detailed new information on the host exploitation in amber nematodes in the supplementary material. This data is also depicted in pie diagrams and seems at first glance to support their interpretations of a shifts in host exploitation in fossil amber deposits when analysed appropriately and statistically but such an true analysis and depiction should be part of the main manuscript to do the compilation and interpretation justice. For the sake of reproducibility and the field, such fundamental statistical analysis as well as a statistical comparison with modern hosts would make this broad-sweeping claim of a major host shift and importance of amber deposits containing such nematode-insect interactions since the Cretaceous (even) more robust and fundamental.

Thanks. We realized this drawback and now we calculated the 95% CI using the Agresti-Coull method of the “binom.confint” function from the binom R package (https://cran.r-project.org/package=binom) of R 4.2.2. We also added a new Figure 10 and Table 1 in our paper. But, since we compiled the “occurrence” of invertebrate–nematode associations from these amber localities, it is impossible to compare with modern mermithids. For example, the parasite of Cretacimermis chironomae occurs five times in Kachin amber, but an extant dipteran-parasitized mermithid species can occur many times just in a single pond. However, it is evident that mermithids and all invertebrate-parasitized nematodes prefer to infect holometabolous insects rather than other invertebrates (Poinar, 1975; Poinar, personal observation). We have also added some explanation to our paper.

Author Response

Reviewer #2 (Public Review):

Yamaguchi et al. studied the roles of two proteins, Calaxin and Armc4, in the assembly of the outer arm dynein (OAD) docking complex (DC). By combination of the improved cryo-ET analysis and gene knockout zebrafish lacking each of these proteins, they found that Armc4 plays a critical role in the docking of OAD and that Calaxin stabilizes the molecular interaction in the docking.They further showed an evidence that Calaxin changes the conformation of another compartment of DC comprising CCDC151/114. This new information provides an important basis for understanding how the DC is assembled and regulates docking of OAD. The authors' conclusion is well supported by the data but some data presentation and discussion need to be completed.

Gui et al. (2021) already reported on a cryo-EM observation in bovine tracheal cilia, with the conclusion similar to this paper in the structure of OAD/DC on DMT. Using knockout zebrafish strain, the authors present detailed interaction of calaxin with other DC components. They show that the binding of calaxin induces the changes of conformation in N-terminal region of CCDC151/114. The conformation further changes in the presence of Ca2+; specific conformation of N-terminal region of CCDC151/114 becomes undetectable, instead additional structure appears in the vicinity of calaxin.

1) The authors conclude that the Ca2+-dependent conformational change of DC is subtle and not dynamic. This result is eventually valuable information but may be somewhat unexpected from the point of view that calaxin plays an important role in the regulation of flagellar motility in Ciona sperm. The authors found that calaxin changes the conformation of N-terminal CCDC151/114 region but the core dynein structure shows no dynamic change. What about the changes in the interaction between calaxin, core dynein, and DMT? Is this beyond the resolution of cryo-ET analysis?

Since Mizuno et al., 2009 reported that Ciona Calaxin switches its interactor depending on Ca2+ concentration, it is highly expected that zebrafish Calaxin also changes its interactor in 1 mM Ca2+ buffer conditions. However, the resolution of our cryo-ET data was insufficient to detect the change of Calaxin interactors. More detailed structural analyses are required to understand the OAD structures in the Ca2+ buffer conditions. We discussed this point as follows:

(line 389-395)

Regarding the Calaxin conformation, a previous biochemical analysis reported that Ciona Calaxin switches its interactor depending on Ca2+: β-tubulin at lower Ca2+ concentration and OAD γ-HC at higher Ca2+ concentration (Mizuno et al., 2009). Moreover, a crystal structure analysis revealed the conformational transition of Ciona Calaxin toward the closed state by Ca2+-binding (Shojima et al., 2018). In this study, however, such conformation change of Calaxin was not detected, probably due to insufficient resolution of our cryo-ET analysis. More detailed structural analyses in the Ca2+ condition are required to understand the mechanism of the Ca2+-dependent OAD regulation.

2) It would be very helpful if the authors could add the cryo-ET images of calaxin-/- axoneme in the presence of 1 mM EGTA in Figure 7. Although these images are thought to be similar or identical to Figure 4F, it would help to confirm that the conformational changes in CCDC151/114 and additional part of DC are induced in a Ca2+-dependent manner.

We added the cryo-ET images of calaxin-/- OAD-DC (1 mM EGTA) in Figure 7D.

3) To clarify the molecular interaction of calaxin with other components, it would also be helpful if the authors add the images rotated 80 degree to Figure 4F and G, in similar way in Figure 7.

We added the images of OADs rotated 80 degrees in Figure 4F and G.

4) Despite the molecular phylogenetic difference, there are several similarities between calaxin and Chlamydomonas DC3, not only in the in situ structure and configuration but in the phenotype of mutants; Chlamydomonas mutant lacking DC3 shows OAD loss in the distal part of a flagellum (Casey et al, MBC, 2003). It may be a good reference if the authors add the position of DC3 in Figure 4. A', B', and C.

To answer this comment, we created Figure 4—figure supplement 1, which shows the cryo-ET structures and models of OAD-DCs in vertebrates and Chlamydomonas.

5) There is a significant difference in sperm motility between WT and calaxin-/- or WT and armc4-/- (Figure 2E). However, it is not clear whether immotile sperm were included in the data for VAP (Figure 2F) or BCF (Figure 2G). For example, WT and calaxin-/- show similar VAP, although both are significantly different in the percent of motile sperm.

In our CASA study, spermatozoa with less than 20 μm/s velocities were considered immotile and excluded from the data for VAP (Figure 2F) and BCF (Figure 2G). To clarify this point, we revised the manuscript as follows:

before

Swimming velocity and beating frequency were calculated from the trajectories of the motile spermatozoa (Figure 2F-G; Figure 2—figure supplement 1; Video3).

after (line 139-141)

Swimming velocity (VAP) and beating frequency (BCF) were calculated from the trajectories of the motile spermatozoa, which have 20 μm/s or more velocities (Figure 2F-G; Figure 2—figure supplement 2; Video3).

6) In calaxin-/- zebrafish, OAD was clearly detected from the base to two-thirds of a flagellum with unclear border (Figure 2A). Typical distribution of OAD+class and OAD-class are shown in Figure 5 in the ~3 micrometer tomograms. Were these taken from around this unclear border? Are proximal most region of a flagellum occupied with OAD+class only? The authors should clearly indicate the region of a flagellum where the tomograms in Figure 5C and D were selected.

7) Line 229~: It is not clear what the authors meant by "probably reflecting the different distance from the sperm head". In relation to this and the comment 6, does the "proximal" in the sentence "OAD loss occurred even in the proximal part of the flagella" (line 232) indicate the region near the base of a flagellum?

In general, axonemes are tangled on the cryo-TEM grids, which makes it difficult to identify the ends of all axonemes, especially for the long zebrafish sperm flagella. Thus, we could not clarify the region of a flagellum about the tomograms shown in Figure 5D.

However, to answer comments (6) and (7), we created Figure 5—figure supplement 1. In this experiment, we newly generated cryo-TEM grids with sparse sperm axonemes and succeeded in finding two areas containing clear axonemal ends with suitable ice conditions for cryo-ET observations (Figure 5—figure supplement 1B). The polarity of the axonemes was judged from the 3D-reconstructed structures of the axonemes (Figure 5—figure supplement 1B, red dotted lines). By the structural classification of OAD+ class and OAD- class in the tomograms, we confirmed the OAD loss in calaxin-/- even in the proximal part of the flagella, which is near the base of a flagellum (Figure 5—figure supplement 1D, (a) and (c)). To clarify these points, we revised the manuscript as follows:

before

In calaxin-/-, the ratio of OAD+ class to OAD- class varied among tomograms (Figure 5D), probably reflecting the different distance from the sperm head. However, all calaxin-/- tomograms showed multiple clusters of OAD- class, indicating that the OAD loss occurred even in the proximal part of the flagella.

after (line 236-239)

In calaxin-/-, the ratio of OAD+ class to OAD- class varied among tomograms (Figure 5D), reflecting the different distances from the sperm head. Analysis of detailed OAD distributions along calaxin-/- axoneme revealed that OAD loss occurred even in the proximal part of the flagella (Figure 5—figure supplement 1D).

8) In conjugation with comment 7, it would be appreciated to show an authors' idea on why distal region of flagella tends to lack calaxin, if they do not discuss anywhere in the text.

We discussed this point as follows:

(line 316-323)

calaxin-/- spermatozoa exhibited a unique OAD distribution, with OAD-missing clusters at various regions of the flagella. Interestingly, OADs decreased gradually toward the distal end, by which the mechanism is unclear. The axoneme is elongated by adding flagellar components to its distal end during ciliogenesis (Johnson & Rosenbaum, 1992). IFT88, a component of the IFT machinery, disappears as the spermatozoa mature (San Agustin et al., 2015). Thus, we speculate that the OAD supply at the distal sperm axoneme is insufficient to compensate for the OAD dissociation in the calaxin-/-. Consistent with this idea, distal OAD loss is the sperm-specific phenotype, as olfactory epithelial cells in calaxin-/- have Dnah8 along the entire length of the cilia (Figure 6B).

9) Immunofluorescence in twister-/- epithelial cilia showed that the localization of calaxin is independent of OAD (line 271-274). Based on the authors' finding, the localization of calaxin requires Armc4, which is preassembled with calaxin in the cytoplasm. If this is true and the localization of calaxin is NOT resulting from diffusion, Armc4 must be localized with calaxin along the entire length of cilia in twister-/- epithelial cilia (Figure 6D). Although Armc4 is shown localized in cryo-ET images (e.g. Figure 1, Figure 7), authors may provide the immunofluorescence of Armc4 along the entire length of sperm flagella and epithelial cilia.

To answer this comment, we obtained a commercially available anti-ARMC4 (human) antibody and checked the cross-reactivity of the antibody against zebrafish Armc4, but no signal was detected in our western blot analysis. Thus, we could not assess the localization of zebrafish Armc4 in twister-/- epithelial cilia.

In our study, we found an ectopic accumulation of Calaxin at the ciliary base in armc4-/- cells (Figure 6C, white arrowheads). The small molecular weight of Calaxin (~25 kDa) suggests the possible diffusional entry of Calaxin into the ciliary compartment. However, in armc4-/- cells, Calaxin accumulated at the ciliary base, strongly suggesting that Calaxin requires Armc4 to be localized to cilia.

Reviewer #3 (Public Review):

ODA-DC anchors ODA, the main force generator of ciliary beating, onto the doublet microtubules. Vertebrate ODA-DC contains 5 proteins, including Calaxin and Armc4, whose mutations are associated with defective ciliary motility in animals and human. By generating calaxin-/- and armc4-/- knockout zebrafish lines, this manuscript examined the Kupffer's vesicle cilia and spermatozoa. They showed that calaxin-/- and armc4-/- knockouts both affect ciliary motility but to different degrees. The authors conducted careful structural analyses using cryo-ET and subtomo averaging on both mutants, revealing a partial loss of ODA in calaxin-/- and a complete loss of ODA in armc4-/-. I really like the distribution analysis of calaxin-/- OADs (Figure 5), which emphasizes the strength of cryo-ET in uncovering the molecule distribution of distinct conformational states in situ. Fitting of the atomic models of ODA and ODA-DC into the cryo-ET density maps and Calaxin rescue experiments showed how Calaxin stabilizes ODA at a molecular detail. By using olfactory epithelium, the authors also presented the possible assembly mechanism of ODA-DC proteins, which is also a beautiful experiment. Finally, the authors also investigated how Ca2+ regulate the ODA-DC using cryo-ET.

The thorough structural and functional analyses of Calaxin and Armc4 in WT and gene KO animals could serve as a reference for future study of the detailed function of other ciliary proteins. The experiments are overall well designed and conducted, but some aspects need to be clarified and improved.

The authors interpret the vertebrate ODC-DC to include four linkers (line 193). However, the authors also said that loss of one linker (Calaxin) makes ODA to attach on the DMT through two linkers (line 199 and 246). These descriptions are confusing. It would make more sense to interpret the vertebrate ODC-DC as containing three linkers (CCDC151/114, Armc4/TTC25, Calaxin).

This comment is reasonable because vertebrate OAD is tethered to DMT through three linker structures (the distal CCDC151/114, Armc4/TTC25, and Calaxin). However, vertebrate DC is composed of four parts (a) Calaxin, (b) the Armc4-TTC25 complex, (c) the proximal CCDC151/114, and (d) the distal CCDC151/114 (Figure 4E). The (c) part is embedded in the cleft between protofilaments A07 and A08. To clarify this point, we revised the manuscript as follows:

before

The bovine DC model shows that vertebrate DC is composed of four linker structures: (a) Calaxin, (b) the Armc4-TTC25 complex, (c) the proximal CCDC151/114, and (d) the distal CCDC151/114 (Figure 4E).

after (line 196-200)

The bovine DC model shows that vertebrate DC is composed of four parts: (a) Calaxin, (b) the Armc4-TTC25 complex, (c) the proximal CCDC151/114, and (d) the distal CCDC151/114 (Figure 4E). Among the four parts, three (a, b, and d) work as linkers between OAD and DMT, while (c) the proximal CCDC151/114 is embedded in the cleft between protofilaments of the DMT.

To confirm whether Calaxin directly interacts with β-tubulin (line 213), a control experiment could be needed by incubating WT axoneme with mEGFP-Calaxin followed by IF imaging.

In our manuscript, we wrote as follows:

(line 218-224)

To assess the specificity of Calaxin binding, we also performed a rescue experiment with mEGFP-Calaxin (Figure 4H-I; Figure 4—figure supplement 2). Ciona Calaxin was reported to interact with β-tubulin (Mizuno et al., 2009), suggesting the possible binding of Calaxin along the entire length of the axoneme. However, the rescued axonemes showed partial loss of EGFP signal (Figure 4H, white arrowheads). This pattern resembled the OAD localization of calaxin-/- in immunofluorescence microscopy, suggesting the preferential binding of Calaxin to the remaining OAD-DC. mEGFP alone showed no interaction with the axoneme (Figure 4H, asterisk).

Therefore, our manuscript is NOT intended to support or deny the interaction between Calaxin and β-tubulin, which was reported by Mizuno et al., 2009. Instead, we focused on the interaction between Calaxin and OAD-DC, revealing that Calaxin binds to Calaxin-deficient OAD-DC (Figure 4G, H, and I). Thus, we assume this comment refers to the interaction between Calaxin and OAD-DC.

To further discuss the interaction between Calaxin and OAD-DC, we created Figure 4—figure supplement 2. We tested Calaxin’s interaction by incubating recombinant mEGFP-Calaxin with sperm axonemes of calaxin-/-, armc4-/- (representing OAD-missing DMT), and WT (representing DMT with Calaxin and OAD). The localization of mEGFP-Calaxin was assessed by fluorescence microscopy of mEGFP signals. In calaxin-/-, mEGFP-Calaxin was bound to the limited region of the axoneme, with the partial loss of EGFP signals (Figure 4—figure supplement 2A, white arrowheads), consistent with Figure 4H. On the other hand, mEGFP-Calaxin showed no significant interaction with armc4-/- axoneme (Figure 4—figure supplement 2B) or WT axoneme (Figure 4—figure supplement 2C). These data show the preferential binding of Calaxin to the Calaxin-deficient OAD-DC than OAD-missing DMT or WT OAD. Although Mizuno et al., 2009 reported the interaction between Calaxin and β-tubulin, our analysis could not detect the signals for such interaction, probably due to the different binding affinity of Calaxin against OAD-DC and β-tubulin.

The Immunoblotting experiment should be improved in Figure 5E. Could the authors get the same results in repeating experiments? Why is the Dnah8 signal higher in 50 mM NaCl of the (+)Calaxin group compared to that in 0 NaCl? This makes me doubt if the difference between (-)Calaxin and (+)Calaxin groups are significant.

This comment is reasonable because NaCl concentration-dependent detachment of OAD-DMT suggests the highest Dnah8 signal in 0 mM NaCl of the (+)Calaxin group. To discuss this point, we created Figure 5—figure supplement 2, which shows the experimental replication of the immunoblot analysis in Figure 5E. In this experiment, we used calaxin-/- sperm axonemes collected independently of the Figure 5E data.

However, again, the Dnah8 signal was higher in 50 mM NaCl of the (+)Calaxin group than that in 0 mM NaCl, confirming the result in Figure 5E. One possible explanation for this result is that the NaCl concentration affects the rescue efficiency of the Calaxin protein. We speculate that the Calaxin protein requires NaCl for efficient binding to OAD-DC, which caused the lower amount of OAD in 0 mM NaCl of the (+)Calaxin group compared to that in 50 mM NaCl.

The authors have covered several important points in the Discussion section. Now that the function of Calaxin in both mouse and zebrafish have been reported, the authors could discuss the similarity and difference of Calaxin function in different species and tissues.

To discuss this point, we inserted the following paragraph:

(line 324-333)

In mouse Calaxin-/- mutant, motile cilia in various organs (sperm flagella, tracheal cilia, and brain cilia) showed abnormal motilities, although OADs in the mutant cilia/flagella seemed mostly intact when observed by conventional transmission electron microscopy (Sasaki et al., 2019). In our study, however, we revealed that mutation of zebrafish calaxin caused OAD-missing clusters at various regions of the flagella, by using detailed cryo-ET analysis and immunofluorescence microscopy. Thus, we speculate that the same OAD defects to zebrafish calaxin-/- caused abnormal ciliary motilities in mouse Calaxin-/- mutant. One exception is the mouse nodal cilia. In mouse Calaxin-/- mutant, the formation of nodal cilia was significantly disrupted (Sasaki et al., 2019). On the other hand, zebrafish calaxin-/- mutant showed the normal formation of Kupffer’s vesicle cilia (orthologous to the mouse nodal cilia), suggesting the tissue-specific function of Calaxin on the ciliary formation.

Because of the limited resolution, the authors should be more careful when interpreting the small densities in the difference map, for example, in Figure 4F-G black arrows. Considering that the CCDC151/114 coiled coil is overall poorly resolved both in the WT and mutant cryo-ET maps, the different densities could be due to different map quality or data processing. This makes the following statement suspicious "This structure corresponds to the N-terminus region of CCDC151/114, suggesting that Calaxin affects the conformation of neighboring DC components".

This comment is reasonable because the resolution of our cryo-ET data was insufficient to identify each molecule in the cryo-ET map. To be more careful about the interpretation of our cryo-ET structures, we revised the manuscript as follows:

before

However, the difference map also showed an additional missing structure adjacent to Calaxin (Figure 4F’, black arrowhead). This structure corresponds to the N-terminus region of CCDC151/114, suggesting that Calaxin affects the conformation of neighboring DC components.

after (line 207-210)

However, the difference map also showed an additional missing structure adjacent to Calaxin (Figure 4F’, black arrowhead). When fitting the bovine DC model, this structure overlapped the N-terminus region of CCDC151/114, indicating that Calaxin can affect the conformation of neighboring DC components.

To discuss the map quality and data processing of our cryo-ET analysis, we summarized the following points that can support the confidence of our data:

(1) Two independent experiments showed the same results of OAD-DC structures, suggesting that the small changes in DC conformations were not due to different map quality or data processing:

(a) For OAD structures in 1 mM EGTA condition, we analyzed the WT OAD (Figure 4D) and the calaxin-/- OAD rescued with recombinant Calaxin (Figure 4G). These samples were prepared in completely independent processes. However, in both cases, the small densities overlapping the N-terminus region of CCDC151/114 were visualized adjacent to Calaxin (Figure 4D and G, black arrowhead).

(b) For OAD structures in 1 mM Ca2+ condition, we analyzed the WT OAD (Figure 7B) and the calaxin-/- OAD rescued with recombinant Calaxin (Figure 7C). These samples were prepared in completely independent processes. However, in both cases, the small densities overlapping the N-terminus region of CCDC151/114 were not observed. Instead, the additional densities appeared around DC (Figure 7B and C, white arrowheads).

(2) We assessed the statistical significance of the changes in DC conformations. We applied Student’s t-test for WT and calaxin-/- OAD-DC structures and created Figure 7—figure supplement 1. p-values of each voxel were calculated as described in Oda & Kikkawa, 2013. The isosurface threshold of p-values corresponds to 0.05% probability in one-tailed test. p-value maps indicate not only Calaxin structures but also the adjacent small density (Figure 7—figure supplement 1A, black arrowhead) and the additional density around DC (Figure 7—figure supplement 1B, white arrowheads) as the statistically significant difference between WT and calaxin-/- OAD-DC.

Author Response

Reviewer #1 (Public Review):

This project aimed to understand if decision making impairments commonly observed in older adults arise from working memory (WM) or reinforcement learning (RL) deficits. Evidence in the paper suggests it is the former; they observe poorer task accuracy in older adults that is accompanied by a faster memory decay in older adults using a novel hierarchical instantiation of a previously validated computational model. There were no similar changes in RL in this model. These results are extended using Magnetic Resonance Spectroscopy (MRS) to measure glutamate and GABA levels in striatum, prefrontal and parietal regions. They found that impairments in working memory were linked to reductions of glutamate in PFC, particularly in the older adult group.

The task employed is elegant and has been studied extensively in different populations and is well-validated (though here a hierarchical Bayesian extension is developed and validated). The results however may not be definitive in some respects; the paper did not replicate previously observed RL deficits. It therefore, remains possible that this is due to the sensitivity of the task to this RL component in ageing and future work is needed to fully bridge the gap in the literature.

Thank you for the comment. If our understanding of the comment is correct, our results suggesting no impairments in the RL system conflict with previously observed RL deficits in older adults. In the introduction section, we discuss previous literature on RL deficits in old adults which yields largely mixed conclusions, wherein some experiments show RL impairments (Frank and Kong, 2008; Hämmerer et al., 2011; Samanez-Larkin et. al, 2014) and some do not (Grogan et al., 2019; Radulescu et al., 2016). Placing our experiment in the context of these mixed results, we aimed to use a task that addresses these inconsistencies, by reasoning that commonly used RL tasks and models do not account for additional processes that may contribute to learning (e.g. executive function/WM/attention), hence explaining why sometimes the deficits are observed and sometimes they are not. We can also point to our model parameter recovery (Appendix 1 - Figure 9), where we show that RL model parameters (e.g. learning rate) are successfully recovered - indicating that our model is sensitive to RL variability in participants, but we observe no differences split across age groups.

Although the study is well-executed, there is an obvious limitation in the use of a cross-sectional design to address this question. The authors acknowledge this limitation in the discussion but could go further to highlight the potential confound of cohort effects on gaming, RL and WM tasks more generally. Without within-person change data, the evidence can only be suggestive of potential age-related decline. For this reason, it may be more appropriate to use the terminology "age-related differences' rather than "age-related declines" given the study design.

Thank you for the comment. We have attempted to address the cohort effects by administering RBANS to old and young participants. Age-normed total RBANS (Randolph et al., 1998) scores were similar in both age groups (described in the first paragraph of the results section), which we took to suggest that our cohorts reflected comparable samples of the population with respect to overall cognitive ability. In addition, we show that certain aspects of performance (e.g. accuracy) decline within the group of older adults, and not just between the two groups, which would constitute an argument against cohort-based effects. We now elaborate further on the point of cross-sectional design in the discussion section on lines 410-417. As suggested by the reviewer, we have also adjusted the language throughout the manuscript to imply age-related differences instead of age-related decline.

Reviewer #2 (Public Review):

In this study, Rmus and colleagues contribute to the important open question of whether reinforcement learning deficits observed in older adults are due to impairments in basic learning processes, or can be attributed to a decline in working memory function. The authors present cross-sectional behavioral data from a task designed to assess the role of working memory in reinforcement learning. And they use computational modeling in conjunction with MR spectroscopy to demonstrate a relationship between prefrontal glutamate and age-related impairments in learning specific to working memory decay. I found the overall story compelling, the data novel, and the analysis carefully executed. Below I outline some areas in which the claims of the manuscript could be strengthened.

1) I may have missed this, but does glutamate correlate with other model parameters? Or did the authors only focus on the WM parameters because of the age difference? In support of the specificity argument, it would be important to show that glutamate only predicts WM related parameters regardless of whether there was an age difference or not.

Thank you for your suggestion. In Appendix 1-figure 7, we show correlations between glutamate and all model parameters. If glutamate captured impairments in RL computational processes, we would expect to see a correlation between glutamate and the learning rate. Below we show that glutamate does positively correlate with RL learning rate. However, there are parameter correlations within the model itself – making the direct correlations hard to interpret.To better understand the relationships between learning rate, working memory, and glutamate, we ran a model predicting MFG glutamate using all parameters that significantly correlated with MFG glutamate (MFG glutamate ~ 1 + learning rate + decay + omega3 + negative learning rate), and found that only WM decay predicted MFG glutamate when controlling for other factors (learning rate: t = -0.42, β = -.03, p =0.67; WM decay: t = -3.14, β = -0.30, p = .002; omega3: t = 1.84, β = .16, p = .07; negative learning rate: t = .56, β = .03, p = .57). Thus, while glutamate measures correlate with RL learning rate, these correlations seem to be driven by the fact that both glutamate and RL learning rate correlate with WM Decay. Note that negative learning rate influences both RL and WM processes’ updating (see computational modeling section), and thus cannot help us make claims about specificity of RL or WM mechanisms alone being related to glutamate.

2) As it is somewhat common with these tasks, it seems like the model does not fully capture the performance deficit in OA (Fig. 2B), even when all the individual difference parameters in WM are allowed to vary. Can the authors say more about the discrepancy? This is an interesting datapoint which may give clues to mechanism.

Thank you for your comment. We elaborated on this in detail in the Appendix 1 (Posterior predictive checks section). We have observed that in some blocks (particularly in ns=6 blocks), older adults only learned a correct response for a subset of the presented stimuli, and neglected to learn responses to other stimuli altogether. We have interpreted this as a possible strategy older adults used to reduce the difficulty of the ns=6 condition. This would explain the discrepancy between the data and the model predictions, as the model has no way of accounting for stimulus identity effects on learning (since the model predicts similar performance for all the stimuli). To test our reasoning, we have fit the model to a subset of data - excluding participants who have implemented this strategy, and predicted that this should reduce the model misfit. We found that this is indeed the case (Appendix 1 - Figure 4). This confirms that strategic prioritization of stimuli in some older adults negatively affected the fit of the model. While we believe that a better understanding of these contaminant response patterns in the RL-WM model is worthy of further investigation, we feel that it is beyond the scope of this paper, and might require task designs with even higher set sizes to elicit the strategic stimulus prioritization more robustly. We have now added a paragraph in the discussion to discuss this issue.

3) Relatedly, it may not be possible with these data alone, but can authors discuss what the WM decay parameter captures? In particular for OA, the distinction between generating and maintaining a "task set" has been extensively written about. Older adults tend to have difficulty internally generating and flexibly deploying task sets, but somewhat paradoxically can perform better than YA in certain decision situations (e.g. when reward is dependent on previous choices, see Worthy et. Al. 2011). The task in this study necessarily pushes OA in a regime in which relying on familiar decision strategies is sub-optimal, and task sets must be continuously generated. Is there a type of intervention do authors expect would reverse the observed deficit in WM?

In the RLWM model, WM stores stimulus-action-outcome weights. Using WM decay we can gradually reduce the stimulus-dependent weights on each trial where the stimulus is not observed (e.g. forgetting). These weights, therefore, get reduced with the rate of decay, by being pulled towards the uniform/uninformative values (1/nA, where nA is the number of actions) they were initialized to. It effectively captures forgetting of information with increased time delays (here time = number of intervening trials between successive stimulus presentations where the stimulus is not observed). It is possible that older adults might be prioritizing storage of different types of (irrelevant) task information (e.g. category of stimuli, or relationships between the stimuli), resulting in a tradeoff that might lead to faster decay in older adults, and that the younger adults neglect such information. This could also explain discrepancies between our model and older adults described above, as the model does not hold any assumptions about how stimulus identities might impact task performance strategy. If this was the case, if probed about such task-irrelevant prioritized information older adults could potentially perform better than younger adults (in a way that in the Worthy et al. (2011) paper the older adults perform better on a choice dependent task compared to younger adults). We are unable to test this idea in our dataset, but we believe that it could be a promising avenue for future research.

4) There is a wealth of evidence suggesting striatal DA loss in older adults, which served as the basis for many of the original investigations and hypotheses regarding a simple RL deficit in OA (e.g. work by Shu-Chen Li and others). While the authors do not directly measure DA in this study, it would be helpful to place the results in the context of that literature.

Thank you for pointing this out. In the introduction, we have discussed the mixed results from research on RL/dopamine deficits in older adults. Some of the literature suggests no impairments in striatal dopamine in older adults (Samanez-Larkin et. al, 2014; Bäckman et al., 2006), while some suggests absence of impairments (Grogan et al., 2019). Furthermore, while DA is important for RL updating, it is also potentially important for WM updating (O’Reilly and Frank, 2006), therefore a potential DA loss could affect both RL and WM, and not RL exclusively. Prior research also suggests that although correlative relationship between DA and cognitive functions has been recorded, the extent of generality/specificity of the effects of DA on cognition in aging (Bäckman et al., 2006), compared to resulting noise that impairs cognition (Li et al.,2001) should be studied more extensively in the future. We have not focused on dopamine in the study, but have now added a paragraph in the discussion section to address this on lines 402-407.

5) Finally, the main argument of the paper as I read it is that PFC glutamate mediates the performance deficits observed in RL because it reflects a compromised WM system. Sample size permitting, it would be helpful to see a formal test of this mediation relationship.

As highlighted in the response to the mediation point in essential revisions, we observe that glutamate mediates effect of WM on task performance, but that this mediation approach might be difficult to justify, due to WM decay and task performance having shared signal and noise (since WM decay is estimated from task performance). We have now included the mediation analysis in our Appendix 1 information and provided a conservative interpretation of it in the results section.

Reviewer #3 (Public Review):

Aging impacts many cognitive functions, and how these changes affect performance in different tasks is an important question. By testing 42 older and 36 younger healthy adults with a novel learning task and MR spectroscopy, Rmus et al addressed the important question whether age-related declines in learning are driven by WM, or by deficiencies of the RL system. The task varied the role of working memory in learning by asking participants to learn about either 3 or 6 stimulus response associations from feedback (set sizes 3 and 6). The paper combines a detailed computational account of participants behaviour and striatal and prefrontal/parietal MR spectroscopy in order to assess individual glutamate and GABA levels.

The authors report an effect of set-size on learning in both are groups, and show that participant age is associated with (1) worse accuracy, (2) a larger set size performance difference, and (3) a heightened sensitivity to reward. Computational modeling showed that working memory decay differed between age groups, but that reliance on WM to perform the task at hand was similar in both age groups (similarly differing between conditions in both groups). Turning to the MRS results, the paper shows that an aggregate measure of glutamate relates to aggregate task performance, that prefrontal glutamate specifically relates to WM decay observed in the task, and that age was negatively associated with glutamate levels.

While the paper is well worth reading and offers many interesting data points, the title's suggestion that "Age-related decline in prefrontal glutamate predicts failure to efficiently deploy working memory in working memory" is, in my opinion, not fully supported by the evidence. First, the authors don't report clear evidence for any age-related differences in WM reliance in the task overall. Second, the authors find that MFG glutamate relates significantly only to WM decay, not the parameter that captures WM deployment. Third, correlations don't imply predictive relations.

We apologize for the lack of clarity in our wording. We agree that the title of the paper implies that the reliance on WM parameter differentiates older and young adults, while the results show that the difference is mostly captured by the WM decay parameter. We meant to communicate that the age-difference seems to be particularly rooted in the WM, but have chosen misleading/confusing words. We have proposed changing the title of the manuscript to “Age-related differences in prefrontal glutamate are associated with increased working memory decay that gives appearance of learning deficits” to minimize confusion. With regards to your last point, as outlined in our response to essential revisions, we agree that we should modify the language used in our manuscript to be more consistent with the associative rather than predictive nature of our results.

Another important open question relates to the relatively large age difference in the effect of set-size on performance. The authors write that working memory will contribute less to performance in higher set size conditions. Yet, age differences are largest in the set size 6 condition, suggesting that RL-dependent learning is most severely impaired in learning (set size 6 performance), rather than WM dependent learning (set size 3 performance). Finally, a statistically significant age difference in reward sensitivity seems to be hardly integrated into the authors' overall interpretation.

Working memory does contribute less in higher set-size condition; however, given the higher number of items, the delays between successive presentations of the stimuli in the high set-size condition are on average longer - which makes the effect of WM forgetting more pronounced. Furthermore, a WM impairment can have an indirect effect in RL, in that frequent failure to select correct action through WM leads to reduced ability to train RL on encoding correct responses (especially earlier in training, when the incremental RL hasn’t ‘caught up’ yet), and thus worse performance overall. As such, a larger effect of set size could potentially be indicative of either or both WM or RL process deficits. This most clearly underscores the importance of modeling - these complex interactions are difficult to intuit, but modeling allows us to establish cleaner mechanistic explanations of observed behavioral patterns/group performance deficits (e.g. while on the surface impairment might look to be RL driven, it is actually better explained by a WM parameter, such as WM decay in older adults - this can). With regards to reward sensitivity, the same explanation applies - there are multiple mechanisms through which differences in reward sensitivity could occur (e.g. slower learning rate, or increased RL recruitment due to failure of WM), which further emphasizes the need for modeling.

In short, in a complex task, there are often multiple ways to explain the same qualitative feature and here we have leaned on computational modeling to identify the computational elements that differed across groups. However we have now also simulated data from our computational models using posterior predictive checks to show that they can reproduce core descriptive features of the original data, including those noted above, and to examine the degree to which different features can be mapped onto the working memory decay parameter (Appendix 1 Figure 5).

Author Response

Reviewer #1 (Public Review):

This paper presents a thorough biochemical characterization of inferred ancestral versions of the Dicer helicase function. Probably the most significant finding is that the deepest ancestral protein reconstructed (AncD1D2) has significant double-stranded RNA-stimulated ATPase activity that was lost later, along the vertebrate lineage. These results strongly suggest that the previously known differences in ATPase activity between extant vertebrates and, for example, extant arthropods is due to loss of the ATPase activity over evolutionary time as opposed to gains in specific lineages. Based on their analysis, the authors also "restore" ATPase function in the vertebrate dicer, but they did so by making many (over 40) mutations in the vertebrate protein, and it is not clear which of these many mutations is required for the restoration of the activity. Thus, it is difficult to discern how the results of this experiment relate to the evolutionary history.

We completely agree with this reviewer's assessment of our paper. Our Michaelis-Menten analyses raised the intriguing idea that loss of ATPase activity in the helicase domain of the vertebrate ancestor may indicate loss of the ability to couple dsRNA binding to formation of the active conformation. Our rescue experiments support this idea, albeit in future studies we hope to create an active ancestor with fewer amino acid changes. While the rescue experiments validate what these analyses told us, as the reviewer suggests, they do not themselves inform on the evolutionary history.

A criticism of the paper is the authors' tendency (probably unconscious) to ascribe a purposefulness to evolution. For example, in the introduction, "We speculate that the unique role of the RLR's in the interferon signaling pathway in vertebrates...created an incentive to jettison an active helicase in vertebrates." Although this sentence is clearly labelled as speculation and "incentive" is clearly a metaphor, the implication is that evolution somehow has forethought. (There are other instances of this notion in the paper, for example, in the last line of the abstract). The author's statement also implies that the developing interferon system somehow caused the loss of active helicase, but it seems equally plausible that the helicase function was lost before the interferon system co-opted it.

We agree with the stated critiques and have rephrased language that suggests that evolution is an active force. In addition to changing the last line of the abstract (page 2, line 35), and removing the quoted sentence from the Introduction, we have included a more nuanced discussion of the order of evolutionary events that may have preceded or followed the loss of helicase function in Dicer (page 18, lines 418-430)

Reviewer #2 (Public Review):

The manuscript by Aderounmu presents an interesting attempt to reconstruct evolution of the function of the helicase domain in ancestral Dicers, RNase III enzymes producing siRNAs from long double-stranded RNA and microRNAs from small hairpin precursors. The helicase has a role in long dsRNA recognition and processing and this function could have an antiviral role. Authors show on reconstructed ancestral Dicer variants that the helicase was losing dsRNA binding affinity and ATPase activity during evolution of the lineage leading to vertebrates while an early divergent Dicer-2 variant in Arthropods retained high activity and seemed better adapted for blunt ended long dsRNA, which would be consistent with antiviral function.

The work is consistent with apparent adaptation of vertebrate Dicers for miRNA biogenesis and two known modes of substrate loading: "bottom up" dsRNA threading through the helicase domain where the helicase domain recognizes the end of dsRNA and feeds it into the enzyme and "top-down" where the substrate is first anchored in the PAZ domain before it locks into the enzyme. Some extant Dicer variants are known to be adapted for just one of these two modes while Dicer in C. elegans exemplifies an "ambidextrous" variant. The reconstruction of the helicase domain complex enabled authors to test how well would be ancestral helicases supporting the "bottom up" feeding of long dsRNA and whether the helicase would be distinguishing blunt-end dsRNA and 3' 2 nucleotide overhang. Although the reconstruction of an ancestral protein from highly divergent extant sequences yields just a hypothetical ancestor, which cannot be validated, the work provides remarkable data for interpreting evolutionary history of the helicase domain and RNA silencing in more general. While it is not surprising that the ancestral helicase was a functional ATPase stimulated by dsRNA, particularly new and interesting are data that the decline of the helicase function started already at the level of the common deuterostome ancestor and the helicase was essentially dead in the vertebrate ancestor. It has been reported two decades ago that human Dicer carries a helicase, which has highly conserved critical residues in the ATPase domain but it is non-functional (10.1093/emboj/cdf582). Recently published mouse mutants showed that these highly conserved residues are not important in vivo (10.1016/j.molcel.2022.10.010). Aderounmu et al. now suggest that Dicer carried this dead ATPase with conserved residues for over 500 million years of vertebrate evolution.

I do not have any major comments to the biochemical analyses and while I think that the ancestral protein reconstruction could yield hypothetical sequences, which did not exist, I think they represent reasonable reconstructions, which yielded data worth of interpretations. My major criticism of the work concerns clarity for the readership and interpretations of some results where I wish authors would clarify/revise the text. The following three examples are particularly significant:

1) It should be explained to which common ancestor during metazoan evolution belongs the ancestral helicase AncD1D2 or at least what that sequence might represent in terms of common ancestry during metazoan evolution.

We thank the reviewer for bringing this issue to our attention, and we have now included a brief discussion of the complexity in identifying AncD1D2’s exact position in metazoan evolution (page 6, lines 124-134). Our maximum likelihood phylogeny is constructed from Dicer’s helicase and DUF283 subdomains which evidently do not contain enough phylogenetic signal to resolve the finer details of early metazoan evolutionary events surrounding the divergence of non-bilaterians: Porifera, Ctenophora, Cnidaria and Placozoa. In our tree, Cnidaria even diverges later than the Nematode bilaterian branch reflecting the fact that our reported phylogeny does not match consensus species relationships, especially in the invertebrate clades. This means we cannot pinpoint AncD1D2’s exact position with certainty. While we do not intend to overinterpret the evolutionary trends from these hypothetical ancestral constructs, we believe the functional differences in biochemical activity are meaningful and correspond to big-picture changes over evolutionary time. AncD1D2 thus corresponds to some early metazoan ancestor that existed before the divergence of bilaterians from non-bilaterians. In support of this interpretation, when the phylogeny is constrained such that the bilaterian branches match the consensus species tree (Figure 1-figure supplement 2A) we observe that AncD1D2 is ancestral to the bilaterian ancestor, AncD1BILAT (now labeled on the figure), but retains 95% identity to the version of AncD1D2 constructed from the maximum likelihood phylogeny (Figure 1-figure supplement 3B).

2) This is linked to the first point - authors work with phylogenetic trees reconstructed from a single protein sequence, which are not well aligned with predicted early metazoan divergence (https://doi.org/10.1098/rstb.2015.0036). While their sequence-based trees show early branching of Dicer-2 as if the two Dicers existed in the common ancestor of almost all animals (except of Placozoa), I do not think there is sufficient support for such a statement, especially since antiviral RNAi-dedicated Dicers evolve faster and Dicer-2 is restricted to a few distant taxonomic group, which might be better explained by independent duplications of ambidextrous ancestral Dicers. I would appreciate if authors would discuss this issue in more detail and make readers more aware of the complexity of the problem.

We agree with the reviewer that in our initial submission we did not properly address the incongruence between our maximum likelihood phylogeny and the consensus species tree of life. We have now addressed this by revisions that discuss the difficulty in using a single gene or protein to accurately date ancient evolutionary events, especially in the case of Dicer, a protein whose evolutionary history is littered with multiple duplication events (page 6, lines 124-147, beginning with “Importantly, we observed multiple instances…”; page 16, lines 365-371, sentence beginning with “Uncertainty in the single gene or protein phylogeny…”). Our assumption that an early gene duplication produced the arthropod Dicer-2 clade is consistent with previous Dicer phylogenies that have been constructed with maximum likelihood algorithms with different parameters (https://doi.org/10.1371/journal.pone.0095350, https://doi.org/10.1093/molbev/msx187, https://doi.org/10.1093/molbev/mss263) using full length Dicer sequences with different taxon sampling depths and tree construction parameters. Removing other fast evolving taxa with long branch lengths from the sequence alignment still resulted in arthropod Dicer-2 branching out early in metazoan phylogeny (https://doi.org/10.1093/molbev/mss263).

In analyses not included in our manuscript, we also independently constructed trees using full-length metazoan Dicers, helicase and DUF-283 subdomains using both RAXML-NG and MrBayes. We tried different taxon sampling depths and tried rooting the tree using either a non-bilaterian outgroup or a fungal outgroup and also tried breaking up potential long-branch attraction with deep taxon sampling. In every iteration, the arthropod Dicer-2 clade diverged early in animal evolution at some point before or during non-bilaterian evolution. We recognize that all these efforts are still prone to long-branch attraction that may cause the rapidly evolving Dicer-2 clade to artificially cluster with distant outgroups, but so far, the only evidence to support an arthropod-specific duplication event is parsimony. This parsimony model is plausible and one might expect a recently duplicated arthropod Dicer-2 to cluster closely with nematode Dicer-1, another antiviral Dicer that would have descended from a common ecdysozoan ancestor but this is not the case. The nematode HEL-DUF clade does get attracted to non-bilaterian Cnidaria clade in our ML tree, but unlike the arthropod Dicer-2 clade, this position varied depending on the parameters of phylogenetic analysis, and so we cannot conclude that arthropod Dicer-2’s position is due to long branch attraction. More sophisticated phylogenetic and statistical tools are needed to answer this question definitively, so we decided to proceed with the highest scoring maximum-likelihood phylogeny generated by our analysis.

While we have now included a short discussion on the nature of this uncertainty in the revised manuscript (page 6, line 124., page 16, lines 365-371), we have excluded these additional details (paragraph above) from the main text in an attempt to prioritize readability for the generalist reader, and we hope that more specialized readers will find this discussion in the public comments helpful.

3) Authors should take more into the account existing literature and data when hypothesizing about sequences of events. Some decline of the helicase activity is apparent in AncD1DEUT suggesting that it initiated between AncD1D2 and AncD1DEUT. This implies that a) antiviral role of Dicer was becoming redundant with other cellular protein sensors by then and b) Dicer was already becoming adapted for miRNA biogenesis, which further progressed in the lineage leading to vertebrates to the unique top-down loading with the distinct pre-dicing state where the helicase forms a rigid arm. Authors even cite Qiao et al. (https://doi.org/10.1016/j.dci.2021.103997) who report primitive interferon-like system in molluscs - this places the ancestry of the interferon response upstream of AncD1DEUT and suggests that this ancestral protein-based system was taking over antiviral role of Dicer much earlier. In fact, a bit weaker performance of AncD1LOPH/DEUT combined with the aforementioned interferon-like system and massive miRNA expansion in extant molluscs (10.1126/sciadv.add9938) suggests that molluscs possibly followed a convergent path like mammals. While I am missing this kind of discussion in the manuscript, I think that the model where "interferon appears ..." in AncD1VERT (Fig. 6) is incorrect and misleading.

This comment is similar to others, including point 3 of Essential revisions, and we have revised our model in Figure 6 accordingly. We agree with the reviewer that we did not sufficiently explore the significance of the decline in Dicer helicase function between AncD1D2 and AncD1DEUT. In addition to the changes noted in point 3 of Essential revisions, we have corrected this by adding or modifying sentences in the Results (page 9, sentence beginning on line 197 “This reduction in ATP hydrolysis efficiency prior to deuterostome divergence may have coincided with…”, and page 11, sentence beginning on line 247 “One possibility is that between AncD1D2 and the deuterostome ancestor…”).

We did not intend to suggest that this loss of Dicer helicase function was unique to vertebrates, but we focused on the deuterostome-to-vertebrate transition for the following reasons:

a) The mollusk clade in our analysis is incongruent with its expected species position as a protostome. In our tree it clusters with deuterostomes instead. On one hand, this is probably an artefact of incomplete lineage sorting or long branch attraction. On the other hand, it is possible that this clade’s position is an underlying signal of the convergent evolution proposed by the reviewer. In support of the latter, some extant mollusk Dicer helicases (ACCESSION: XP_014781474, ACCESSION: XP_022331683) show a loss of amino acid conservation in Dicer’s ATPase motifs implying that extant mollusks have also lost Dicer helicase function like vertebrates. However, this is in contrast to vertebrate Dicer helicase where loss of function exists, but ATPase motifs remain conserved. We do not discuss this in the paper because the evidence remains inconclusive until extant mollusk Dicers can be functionally characterized, similar to Human Dicer and Drosophila Dicer-1, to determine that they are truly specialized for miRNA processing to the detriment of helicase function.

b) Caenorhabditis elegans Dicer is an example of an ambidextrous Dicer, that processes both miRNAs, with the top-down mechanism, and viral dsRNAs, with the bottom-up mechanism. Recently, work has been published that suggests that C. elegans also possesses a protein-based innate immune defense mechanism, but instead of competing with the RNA interference mechanism, both mechanisms seem to work in concert and even share a protein in both pathways: DRH-1, a RIG-I-Like receptor homolog (https://doi.org/10.1128/JVI.01173-19). Furthermore, a protein-based pathway has also been reported in Drosophila and in this scenario Drosophila Dicer-2 is the dsRNA sensor that is common to both pathways (https://doi.org/10.1371/journal.pntd.0002823). This collaboration observed in ecdysozoan invertebrates is different from the competition that has been well established in vertebrates. More data is needed to understand whether a model of competition or collaboration exists in lophotrochozoan invertebrates like mollusks.

Author Response

Reviewer #1 (Public Review):

VO2max is one of the most important gross criteria of peak performance ability and a plethora of studies focused on VO2max prediction. This manuscript provides huge and comprehensive data from male runners and male cyclists. The endurance-trained athletes performed cardiopulmonary exercise testing on a treadmill (n= 3330) or cycle ergometer (n=1094). In contrast to former studies, the authors used machine learning for algorithms and VO2max prediction. Models were derived and internally validated with multiple linear regression. The present study substantially expands current research.

Sadly, the manuscript has an important and relevant main shortcoming as the limitations of the study had not been addressed properly:

• The authors paid no attention to the fact that their results are strongly influenced by the exercise protocol used. It is obvious e.g. that maximal performance attainable in protocols with 2-minute exercise steps will be higher compared to an identical protocol with 3- or 4-minute steps.

• The exercise intensity was kept constant for only 2 minutes before the workload was increased (by 1km/h treadmill or by 20-30 W cycle ergometer). Due to the kinetics of lactate, VO2, etc., it is evident that the short 2-min intervals aggravate the correct determination of aerobic and anaerobic threshold. It is well-known that longer-lasting constant exercise steps (e.g. 4 minutes) are better when the focus is centered on threshold determinations.

The quality of this manuscript will be substantially improved when the authors could implement a comprehensive and blunt paragraph showing the limitations of their study.

We have completed our manuscript by indicating its limits as recommended. It is reasonable to suspect that the type of protocol used matters in the cardiorespiratory indices obtained. Interestingly, according to available studies, this effect is more pronounced for the determination of cyclists' threshold power output or runners' treadmill running speed than for threshold and maximum cardiorespiratory indices such as VO2max or Hrmax (Silva et al. 2021; Weston et al. 2002; Vucetić et al. 2014).

In the regression models presented, the main explanatory variables with the largest effect on the prediction value are the AT/RCP threshold VO2 values (rVO2RCP; rVO2AT). The coefficients for the other explanatory variables are relatively low and differences in their values due to the use of potentially different protocols appear to be marginal. Nevertheless, we see the possibility of worsening the prediction when using less suitable testing protocols for athletes such as ramp tests or typically clinical tests such as the Bruce test.

Author Response

Reviewer #1 (Public Review):

This study represents an important work in the field of (CAR)T-cell immunotherapy by analyzing the effect of different oxygen tension on the function and differentiation of T-cells (especially CD8+). Although it has been described that low oxygen levels can influence effector function/differentiation of T-cells, as nicely acknowledged by the authors in the introduction, a comprehensive analysis in the context of immunotherapy has been missing so far and this study adds significant findings that will be relevant for patient care in all fields applying (CAR)T-cell immunotherapy.

The strength of the evidence is generally solid although there are some discrepancies between the different ways to induce HIF-1α (i.e. low O2, pharmacological inhibition, shRNA knockdown) that need to be clearly stated and/or discussed.

1) The first section of the results determines the impact of low oxygen and pharmacological HIF-1α stabilization on CD8+ T-cell activation/differentiation. Low oxygen diminishes cell growth but induces T-cell activation and effector cytokines, while HIF-1a stabilization mimics the effects on activation without alterations in expansion. Unfortunately, it remains unclear why effects upon low O2 are more pronounced although pharmacological HIF-1a stabilization is more efficient.

2) As a next step, in vitro conditioned T-cells are transferred into a subcutaneous B16-OVA model. Although only the low O2 levels increase T-cell numbers in vivo after the transfer, the initial tumor burden was nicely decreased by both low O2 and HIF-1a stabilization. However, only the latter significantly improved survival and it remains unclear and uncommented why.

3) Next, the authors address whether pre-conditioning of human CART-cells to induce HIF-1α either by pharmacological stabilization or by silencing of VHL shows similar effects. Surprisingly, both ways of HIF-1a stabilization resulted in different effects concerning differential gene expression and cytotoxic capacity of CART-cells. Accordingly, pharmacologically pre-conditioned CART-cells did not have a significant impact on survival in an in vivo model, while the VHL-silenced ones did significantly improve animal survival. This discrepancy between the two modes of HIF-1a stabilization remains uncommented. Unfortunately, it also remains unclear why the pharmacological HIF-1a stabilization significantly improved the survival in animals of the B16-OVA model and not in the human CART-cell model.

4) After this, the researchers determine how the timing of hypoxic conditioning affects the (CAR)T-cells. Here it is convincingly shown that already a short period of hypoxic conditioning (1 day) with a subsequent expansion phase (additional 6 days) is sufficient to induce HIF-1a mediated alterations (e.g. metabolic changes, calcium flux, intracellular signaling). Although this section is coherent in itself, the switch between different times of hypoxic conditioning, expansion, and analysis is difficult to follow and might lead to confusion. The expression pattern of e.g. HIF-1a on day 1 and day 7 together with the nuclear amounts of NFAT and c-Myc might be misunderstood, like the other presented data as well.

5) Last, short-term hypoxic conditioning of CART cells is tested in a solid tumour mouse model. The previously identified conditioning protocol also increases CART-cell function against solid tumours (as shown by enhanced cytotoxicity, reduced tumour burden, and prolonged survival). Unfortunately, although both HER2-CART-cells and CD19-CART-cells are shown to have superior cytotoxicity in vitro after the pre-conditioning, only HER2-CART-cells are demonstrated to be superior upon low O2 conditioning in an in vivo adoptive transfer mouse model and CD19-CART-cells remain an open question.

Generally spoken, the limitations of the manuscript are:

1) The occurring discrepancies of determining effects caused by the different modes of Hif-1a stabilization which certainly are caused by the complex nature of Hif-1a regulatory network, and;

We now extend our observations and discuss these concerns more extensively in the manuscript.

2) The limitation of detected effects primarily on CD8+ T cells while CART-cells products usually are a mixture of CD4+ and CD8+ ones.

Figure S6H now shows that the effects of shorter periods of low oxygen conditioning obtained with CAR-T cells generated from isolated CD8+ T cells are reproducible in CAR-T cells generated from PBMCs. We have found that a 24h incubation of PBMC-derived CAR-T cells in 1 %O2 increases cytotoxicity against target cell effector differentiation at day 7, when compared to the cytotoxic effects of cells cultured at 21% oxygen levels.

Reviewer #3 (Public Review):

In this study, Cunha et al. examined the role of different oxygen tensions (21%, 5%, and 1% O2) and HIF-1α stabilisation in regulating murine and human CD8+ T cell proliferation and function. The authors find that hypoxia (1% O2) and pharmacological PHD inhibition with FG-4592, enhance murine T cell activation but impair proliferation. Furthermore, adoptive cell transfer (ACT) therapy of CD8+ T cells from both conditions reduced tumour burden in a B16-OVA melanoma model. Short hypoxic conditioning (1% O2) of human CD8+ T cells for 1 day increased HIF-1α stabilisation, with increased activation, glycolysis, and mitochondrial function still observed following 6 days of normoxic cell culture. Short hypoxic conditioning of HER2 and CD19 CAR-T cells improved their activation and cytotoxicity in vitro, while HER2 CAR-T cell counts were increased in vivo, reducing tumour burden, and increasing survival when compared to 21% O2.

Strengths:

The paper convincingly demonstrates that short hypoxic conditioning in a defined window improves CAR-T cell function through in vitro cytotoxicity assays and following adoptive transfer in a preclinical HER2+-SKOV3+ positive tumour model. Thus, the major conclusion of the paper is mostly well supported by the data and could represent a novel strategy to improve CAR-T cell immunotherapy for solid tumours in the future.

Weaknesses:

The extent to which hypoxic conditioning-mediated improvement in CAR-T cell function is dependent on HIF-1-driven metabolic reprogramming is unclear and other potential mechanisms are not explored. 5FG-4592 and VHL silencing in HER2 CAR-T cells did not phenocopy each other faithfully. In addition, neither approach was as effective as short hypoxic conditioning with 1% O2 in improving CAR-T cell function in vitro or in vivo. Although the authors suggest the temporal dynamics of HIF-1α stabilisation is the key point, this is not convincingly proven, and no metabolic characterisation of these CAR-T cells was performed.

The revised manuscript now includes live metabolic analyses in a Seahorse set up, using T cells following FG-4592 treatment or VHL silencing. We found exposure of human CD8+ T cells to FG-4592 leads to a suppression of their oxygen consumption rates, both at basal and maximal levels. This can underpin the observed reduced expression of effector molecules (PMID: 33398183). Treatment of human T cells with FG-4592 resulted in a dose-dependent reduction of in vitro cytotoxicity, similar to that observed with exposure to low oxygen (e.g., 7 day OT-I expansion in 1%O2 impairs antitumour function [Figure supplement 6L]).

Regarding VHL silencing, we did not observe metabolic differences compared to controls. This might arise from the fact that shVHL vectors only caused an overall 30% reduction in VHL protein expression, and that the silencing occurred after T cells had been activated. As we show, the moment of activation is key for T cell differentiation and function, and this could explain the lack of metabolic differences between shNCT and shVHL-expressing cells. These points are now added to 5th paragraph of the Discussion section.

It is unclear how changes elicited during short hypoxic conditioning are maintained following continued normoxic cell culture. Hypoxia is known to rapidly regulate histone methylation and chromatin structure in a HIF-independent manner (PMID: 30872525; PMID: 30872526). Are similar epigenetic changes observed in T cells, and if so, could these epigenetic changes underlie improved T cell activation?

We thank the reviewer for the insightful comment on potential epigenetic changes observed in T cells cultured in hypoxia. We have now carried out an extensive analysis of histone methylation and acetylation (Figure 4H). Human CD8+ T cells cultured for 1 day in 1% and 6 days in 21% showed decreased acetylation of H3K9 and H3K27, reduced trimethylation of H3K4 and H3K27 and increased methylation of H3K9me2, as compared to the levels of cells continuously grown in ambient oxygen. These differences might underpin the altered differentiation and metabolic shifts of 1% cultured T cells and further indicate that the oxygen tensions during the first 24 hours of activation elicit permanent alterations in T cells. Future work will be dedicated to understanding the link between the observed alteration in histone post-translational modifications and T cell function in response to hypoxia.

Complications may also arise when comparing different oxygen tensions given recent data that suggests standard cell culture conditions can lead to local hypoxia through a combination (https://www.biorxiv.org/content/10.1101/2022.11.29.516437v1) of cellular respiration and poor O2 diffusion. Although it is unclear how this will impact suspension T cells it does beg the question as to whether HIF-1α stability following T cell activation is (at least in part) mediated by pericellular O2 limitations in cell culture over time, even in presumed hyperoxic (21% O2) conditions? Or if T cells subsequently cultured at 21% O2 following short hypoxic conditioning (1% O2) still experience local hypoxia during the 6-day culturing protocol? It would be important to assess this in future work and at least discuss these potential weaknesses.

Upon analysing HIF-1α accumulation on day 7, we only found substantial HIF levels in cells that had been in low oxygen tensions for the last 3 days of culture (Figure S4G). This suggests that cells were not experiencing hypoxia at the time of analysis on day 7, given that we did not observe substantial HIF accumulation. We have additionally designed an experiment where 21% and 1% 1 day T cells were cultured for 7 days with a single media change on day 4 (standard) or with 5 media changes (each media change performed on separate days to minimize local hypoxia in ambient oxygen). Regardless of the number of media changes, 1% 1d cultures showed increased effector differentiation and expression of effector molecules, relative to 21% cells (Figure S4H). We also did not observe any differences between control cells cultured with 1 or 5 media changes. As hypoxia elicits changes in T cell differentiation, this suggests cells do not experience local hypoxia during the phase of ambient oxygen expansion. Nevertheless, we very much agree that it is important to accurately assess oxygen concentrations in cell culture media.

Author Response

Reviewer #1 (Public Review):

The authors provide evidence for chromatin, which in Drosophila muscle cells is peripherally localized in the nucleus, whereas the central region is depleted of chromatin, and is organised such that RNA polymerase II (RNAp) is surrounding dense regions of chromatin. The authors theoretically study the formation of these regions by describing chromatin as a multi-block copolymer, where the blocks correspond to active and inactive chromatin regions. These regions are assumed to phase separately and to have different solvability. The solvability of the active region is regulated by binding RNAp. The authors study the core-shell organization in a layered geometry by analyzing the various contributions to free energy. In this way, they in particular obtain the dependence of the shell-layer thickness, which is described as a polymer brush. From these results, they infer chromatin organization in spherical coreshell chromatin domains and compare these results to Brownian dynamics simulations.

The work is well done and even though it uses standard methods for studying block copolymers and polymer brushes obtains interesting information about local chromatin organization. These findings should be of great interest to researchers in the field of chromatin organization and in general to everybody interested in understanding the physical principles of biological organization.

The work has two main weaknesses: The experimental evidence for RNAp and chromatin microorganization is weak as only one example is shown. It remains unclear whether the observed organization pattern is common or not. Also, no data is shown concerning the dependence of the extensions of the active and inactive phases on parameters, for example, solvent properties or transcriptional activity. Second, some parts could prove difficult for biologists to assess. For example, the expression for the brush-free energy should be explained in more detail and notions like that of 'mushrooms' need to be introduced. As a second example, biologists might benefit from a better explanation of the concept of a theta solvent and its relevance.

We thank Reviewer #1 for the positive review and critical feedback. Below we answer the points raised in the last paragraph of its review.

In the original version of the manuscript we only showed a representative image of nuclei of muscle cells in an intact, live Drosophila larvae. Notably, this organization is representative of many nuclei analyzed in muscle tissue. In the revised version we show that in a distinct tissue, e.g. salivary gland epithelium of live Drosophila larvae, RNA Pol II distribution is similarly facing the nucleoplasm, although chromatin condensation differs due to higher DNA ploidy. The new images were added as Supplement information (Fig A1). Since these representative images are the main motivation behind our theoretical analysis, we think that including them will help the reader in understanding the relevance of our minimal model. The effect of different biological perturbations, such as changes in the repressive marks and how these change the core-shell structure require extensive experiments that are outside the scope of the present paper. We also note, that in live organisms (not just live cells) such as those studied here, one can only reliably use genetic perturbations; solvent quality is regulated by the organism and cannot be controlled as in synthetic polymer experiments. Our main focus in the present paper is to highlight an area that has been relatively unexplored by the chromatin organization community, which is how changes in concentrations binding-partners of chromatin may have a strong effect in nuclear architecture.

We have also improved the explanation of the physical concepts for biologists. We added a more thorough explanation of the concept of a polymer brush and explained more clearly what the concept of theta solvent in terms of the scaling properties of a polymer in solution. We quote these revisions below.

Reviewer #2 (Public Review):

This work formulates a detailed theoretical polymer physics model intended to explain the observed morphology of chromatin in the Drosophila cell nucleus. The model is examined in detail by both analytical calculation and computer simulation. The central premise of the suggested theory is that it is again based on equilibrium statistical mechanics. Within this paradigm, authors explore the model that views chromatin fiber as a block copolymer and, most importantly, describes the role of RNA polymerase as it interacts with one of the copolymer blocks and regulates its effective solvent quality. Blocks are assumed to be fixed on the time scale of interest by, e.g., different levels of acetylation or methylation. RNA polymerase is supposed to interact only with one of the chromatin blocks, called active, and assumed interaction is quite peculiar. Namely, RNA polymerase complex may absorb on chromatin fiber and, the model assumes, the fiber decorated with absorbed RNA polymerase molecules is less sticky to itself, or more repulsive than the fiber itself. This peculiar assumption allows authors to make interesting predictions about how proteins can regulate the genome folding architecture.

We thank the reviewer for the positive and critical feedback. We agree that our assumption of changes in the effective solvent stemming from protein complexes binding to chromatin is at the core of our analysis and we justify it further below.

STRENGTH

The work includes a rather detailed theoretical description of the model and its equilibrium statistical mechanics. As both analytical theory and accompanying simulation indicate, the assumptions put forward in formulating the model do indeed produce the desired morphology, with isolated regions ("micelles") of core inactive chromatin surrounded by the less dense shell region in which RNA polymerization may potentially take place. Having such a detailed theory is potentially beneficial for the field and opens up avenues for further exploration.

We thank the referee for appreciating the potential benefit of our minimal theory of solvent-quality regulation by binding processes.

WEAKNESS

The underlying assumption about the interaction of RNA polymerase complex with the fiber, although important and organic for the model, does not seem easy to justify from a molecular standpoint, especially thinking of the charges and electrostatic interactions.

We visualize that the binding of RNA Pol II (mediated by different transcription factors) to chromatin is also associated with larger protein complexes that may contain hydrophobic and hydrophilic components, such as pre-initiation complexes. Some regions of these complexes might associate directly with chromatin due to positive charges on the surface of the Pol II complex , whereas the hydrophilic negative regions may be directed towards the solvent. Our theory is typical of the approach used in polymer physics where coarse-grained interactions are considered. While the origin of hydrophilic interactions lies in electrostatics, such interactions are highly screened in cells (typically 200 mM concentration of salts) and can be considered as short-ranged and competitive with hydrophobic interactions. Chromatin in solution is known to condense (see Gibson, et. al., Cell 2019 and Strickfaden, et. al., Cell 2020) and even phase separate from the nucleoplasm (see Amiad-Pavlov, et. al., Science Advances, 2021); this can arise either from hydrophobic interactions of the histone tails or from opposite charge attraction of the histones and linker DNA. In our model, this competes with the binding of protein complexes which then disrupt the self-attraction of chromatin. Previous work has shown that RNA Pol II associating with chromatin (in the absence of transcription) prevents the coarsening of dense chromatin domains (see Hilbert, et. al. Nat. Comm. 2021), which agrees with our modeling of protein complexes that bind to chromatin and interfere with its condensation; in addition, the binding of Pol-II and all its binding partners that form the pre-initiation complex (see Hahn, Nat. Struct. & Mol. Biol. 2004, 11) will result in effective, steric repulsion between different active and Pol II bound chromatin domains. Another interesting observation is that most of the surface of RNA Polymerase II is negatively charged with a few positively charged patches with which it specifically interacts with DNA while others serve as exit paths of RNA (see Cramer, et. al., Science, 2001.). We agree that a more thorough analysis of the molecular interactions between what we name protein complexes and chromatin is interesting, but it is out of the scope of our paper that uses a coarsegrained, polymer physics approach. This approach also allows our model to be to be predictive as to the physical organization and growth of the domains, independent of those molecular details that are as yet unknown.

Reviewer #3 (Public Review):

This theoretical study provides a theoretical explanation for a puzzling question arising from recent experiments: How can chromosomes behave like polymers collapsed in a poor solvent but also contain "open" active chromatin sections? The authors propose that the binding of proteins (e.g. RNAP's) to the active sections can effectively change the solvent quality for these sections and thus open them. They suggest further that chromosomes show micellar structures with inactive blocks forming the cores of the micelles. Protein binding causes swelling of the micellar shells which affects the whole chromosome structure by changing the total number of micelles. This theory fits well to live imaging data of chromatin in Drosophila larvae, like the one shown in the striking Figure 1.

The manuscript is written very clearly.

My only suggestion is that the authors, in both the theory and simulation parts, are more explicit about how the interactions between the various components are modeled. From what I could see, in the theory part, one needs to look closely at Eq. 5 to understand how the influence of the binding of proteins affects the interaction between active monomers, and in the simulation part, one needs to go to the appendix to learn that interaction strengths between monomers within the active blocks and monomers within the inactive blocks have different values. The latter is crucial to understand the micellar structure shown at the top of Fig. 5A.

We thank the reviewer for his positive response. We have explained Eq. 5 more carefully now and included other explanatory remarks throughout the text. We also explained more clearly the interactions considered in the simulations. Below we answer point by point and add quotes from the revised manuscript.

Author Response

Reviewer #1 (Public Review):

Marchal-Duval et al studied the role of Prrx1 in lung fibroblasts. Prrx1 is a transcription factor expressed in lung fibroblasts but not in other cell types. The authors showed that Prrx1 gene expression was enhanced in IPF patients. Immunohistochemistry in IPF tissue suggested that Prrx1 was expressed in fibroblasts in fibroblastic foci. The authors then showed that Prrx1 expression was regulated by TGF-b1 stimulation or stiffness of substrate by in vitro experiments using primary human lung fibroblasts from either normal or IPF lungs. The authors also showed that Prrx1 regulated fibroblast proliferation and TGF-b signaling by regulating PPM1A and Tgfbr2 expression. Finally, the authors revealed that Prrx1 knockdown suppressed fibrosis in bleomycin-induced fibrosis or PCLS. This manuscript identified novel molecular roles of Prrx1 in fibroblast activation, which is expressed in not only lung fibroblasts but also in other injured or developing organs. To support the idea that Prrx1 plays a critical role in lung fibrosis, however, some discrepancies between in vitro and in vivo data need to be clarified.

Comment #1. Although the authors showed that Prrx1 knockdown in primary fibroblasts reduced Smad2/3 phosphorylation, the reduction of Acta2 or Col1a1 after Prrx1 knockdown and TGF-b1 stimulation was not impressive (Fig. S6), suggesting that the inhibition of TGF-b signaling by Prrx1 knockdown is only partial. In contrast, Prrx1 knockdown by ASO in bleomycin-induced fibrosis showed remarkable fibrosis suppression (Fig. 6, 7). Admittedly there are differences in models and nucleotides used, but this discrepancy needs to be addressed.

We agree with the reviewer that Prrx1 inhibition only partially affects the upregulation of ACTA2, but this effect was significant (around 50% inhibition at the protein level). As stated in the discussion (lines 569-572), our data show that key ECM proteins such as Collagen 1 and Fibronectin were still upregulated in TGF-1 stimulated lung fibroblasts transfected with PRRX1 siRNA, whereas TNC and ELN mRNA expression levels were perturbed. These findings suggest that broader phenotypical changes are associated with Prrx1 knockdown. Notably, we also observed that Prrx1 inhibition impacted cell proliferation in vitro. We believe that the observed suppression of fibrosis in bleomycin treated mice following Prrx1 knockdown by ASO is the result of both the partial inhibition of TGF-β1 effect and the decrease in mesenchymal cell proliferation. Supporting this hypothesis, we observed a decrease in PDGFR-positive cell proliferation in Prrx1 ASO-treated animals (see comment #4 hereafter).

Comment #2. Fig.6 and 7 lack control groups, where mice are treated with PBS instead of bleomycin and treated with either control ASO or Prrx1 ASO.

As stated in the revised version of the material and method (line 683-686), the knockdown efficiency of Prrx1 ASO and lack of effects of control ASO were first validated in naive mice, which were treated with either Prrx1 ASO or control ASO, compared to PBS-treated mice (see Figure R2 in the answer to comment #11 of reviewer 2). Those groups were not repeated / included in the first set of bleomycin experiments in order to comply with institutional regulation to limit animal usage. In the first set of experiments (Prrx1 ASO treatment between day 7 and day 13 after bleomycin insult), the saline + PBS was just used to confirm fibrosis development while the bleomycin + Control ASO was the proper control of the bleomycin + Prrx1 ASO group. In the new second set of experiment (ASO treatment between day 21 and day28 suggested by reviewer #2), we were authorized by our local animal ethical committee to include a control ASO group in the saline treated group to confirm that the lack of effect of these control ASO compared to the PBS group (see new Figure 7-figure supplement 1).

Comment #3. In Fig. 6F, the hydroxyproline content is shown with ug collagen/ug protein. Total protein in the lung is influenced by infiltration of hematopoietic cells, which are the major population in injured lungs by cell count. Fibrosis should be ideally assessed as ug hydroxyproline/lung (or lobe).

We completely agree with the reviewer that hydroxyproline content should ideally be assessed by lobe/lung. As stated in the revised material and methods (lines 882-885), hydroxyproline and protein contents were measured using paraffin lung sections (15 sections of 10µm per sample) with the Quickzyme Biosciences hydroxyproline assay and total protein assay kits; due to limited material access and to refine its use to limit animal usage. Furthermore, the infiltration of hematopoietic cells would rather undermine the effect of Prrx1 ASO (less fibrosis and inflammation) since the contribution of those cells would be higher in control ASO-treated bleomycin mice. Considering the reviewer’s concern, a complete lobe was used to measure hydroxyproline content in the new set of experiments generated during the revision of the manuscript (see new Figure 7-figure supplement 1).

Comment #4. Major proliferating populations in bleomycin-treated lungs are not mesenchymal cells but epithelial/endothelial/hematopoietic cells. Mki67+ cells (Fig. 7D) need to be identified by co-staining with mesenchymal markers if the authors claim that Prrx1 knockdown suppresses fibroblast proliferation in vivo.

We agree with the reviewer that epithelial/endothelial/hematopoietic cells are the main proliferating populations in bleomycin treated animals at day 14. As suggested by the reviewer, we performed a MKI67 / PDGFR co-staining to identify proliferating mesenchymal cells and confirmed a decrease in proliferation in these cells after Prrx1 knock down in bleomycin treated mice (see lines 448-451 and Figure 6-figure supplement 3).

Comment #5 Bleomycin-injured lungs or IPF tissue are patchy and mixed with normal and abnormal areas. Therefore, how areas of interest are chosen for histological quantifications (Fig. 6C, S14D) need to be described in the methods section.

As now stated in the revised material section (lines 864-866), areas of interest were chosen according to the presence of major alveolar thickening as well as fibrous changes and masses (confirmed by picrosirius staining on serial section).

Reviewer #2 (Public Review):

The paper from Marchal-Duval et al reports for the first time the important role played by the transcription factor PRRX1, expressed specifically in the mesenchyme of the lung, in the context of fibrosis. The authors used a combination of human (Donor and IPF) and mouse lungs (saline and bleomycin treated) as well as associated fibroblasts and PCLS to test the functional role of PRRX1 in the context of proliferation and differentiation induced by TGFb1. The work is supported by an impressive amount of data (7 main figures and 14 supplementary figures).

Comment #1: A main weakness in this work is the counterintuitive result that PRRX1 is downregulated in human lung fibroblasts (from both IPF and Donor) treated with TGFb1.

We agree with reviewer that PRRX1 downregulation upon TGFb1 treatment may appear counterintuitive. First, as stated in the manuscript, this inhibitory effect is partial. Secondly, we performed additional experiments in the revised manuscript to better understand (timewise) the downregulation of PRRX1 in response to TGF-b1 in lung fibroblast as suggested by the reviewer. Time course analysis of PRRX1 isoform expression levels showed that PRRX1 was downregulated only after 48h. This late downregulation of PRRX1 in response to TGF-b1, could be the signature of a negative feedback loop to limit cell-responsiveness to TGF-b1 when lung fibroblasts are fully differentiated into myofibroblasts at 48h as discussed in the revised manuscript (see lines 175-180 and lines 589-594).

Comment #2: Another smaller weakness is the inactivation of Prrx1 in vivo using ASO starting at d7 post bleomycin treatment.

In our study of Prrx1 inhibition in vivo, we followed a therapeutic/interventional protocol consistent with current literature on the bleomycin model of lung fibrosis (Moeller A. et al, Int J Biochem Cell Biol 2008 and Kolb M. et al., Eur Resp J. 2020), treating the animals with either control or Prrx1 ASO every other day between day 7 and day 14 during the active fibrotic phase. In the revised manuscript, we extended our investigation to assess the potential effect of Prrx1 inhibition during the late fibrosis phase after bleomycin treatment at day 28, treating the animals with either control or Prrx1 ASO every other day between day 21 and day 27. Interestingly, we found that the effects of Prrx1 inhibition during the late fibrosis phase were less (but still) potent compared to the active fibrotic phase (see Figure 7-figure supplement 1).

Author Response

Reviewer #2 (Public Review):

We thank the reviewer for their assessment that our work “supports the idea that epithelial-endothelial crosstalk is important for lung regeneration and proposes a potential candidate for this process” and their helpful suggestions for strengthening and clarifying our work.

1) The scRNA-seq analysis is performed in two separate objects ("control lung" and "H1N1 infected lung 14dpi"). For these two sets of data to be comparable, the authors should have integrated the objects and analyzed them together. This is not only important for deciding the clusters' identities and making sure that the same clusters are compared between control and infected, but also necessary to compare gene expression.

We have integrated the control and H1N1-infected scRNA-seq datasets and reanalyzed the integrated data. We then analyzed CAP1_A and CAP1_B populations, comparing their gene expression between control and influenza conditions. Unbiased clustering of the integrated dataset reveals the same clusters we identified in the individual datasets, with cells from control and flu contributing to each cluster (with the exception of proliferating endothelial cells, which are found only in the H1N1-infected lung). We have added a supplemental figure outlining these data (Figure 1 – Figure Supplement 3).

2) ATF3 is not only present in Cap1_B, in the infected lung there seems like Cap1_A express less ATF3. The authors should comment on this difference.

We have added violin plots to Figure 1, which we feel will better represent the greater Atf3 expression in CAP1_Bs relative to other endothelial cell subtypes. The reviewer is correct that Atf3-expressing cells are found in large vessels, but they are also numerous in the alveolar capillary space and increase with influenza in these regions. We have added lower-magnification, higher-resolution images of Atf3CreER; ROSA26tdTomato animals, both control and influenza-infected, to illustrate this expansion in a new Figure 2 – Figure Supplement 3. This increase is also quantified in Figure 2C. We have also clarified this in the text.

3) It is unclear how the clusters Cap1_A and Cap1_B were decided. The manuscript would benefit from clarification.

We have added text to the Materials and Methods section to clarify this.

4) It would be beneficial to see via immunofluorescence the morphological and spatial differences between ATF3-expressing and non-expressing endothelial cells since this transcription factor is expressed in multiple endothelial cell types.

We have added lower-magnification, higher-resolution images of Atf3CreER; ROSA26tdTomato animals, both control and influenza-infected, to illustrate the spatial distribution of Atf3-expressing endothelial cells. This data is now shown in the new Figure 2 – Figure Supplement 3. We have also added further data to the new Figure 5 – Figure Supplement 1 to include the cytoplasmic endothelial marker Endomucin-1 (EndoM1) in an analysis of the spatial distribution of endothelial cells in wild-type and Atf3-knockout animals at 21 dpi.

5) The authors mention ATF3 is not endothelial-specific. Expression of ATF3 in other cell types should be evaluated via immunofluorescence.

This data is present in Figure 2 – Figure Supplement 2.

6) The authors should have shown evidence of the deletion in their Atf3EC-KO mouse and addressed whether they had residual ATF3. If there is no antibody available, RNAscope could be used, or Western Blot or RT-PCR on sorted endothelial cells.

We agree that this is an important quantification to make. We have performed qRT-PCR for Atf3 in both the animals used to perform the RNA sequencing experiment as well as a new cohort of animals to confirm Atf3 deletion. We have added these results to a new supplemental figure accompanying Figure 4 (Figure 4 – Figure Supplement 1).

7) The authors only show the epithelium as evidence that the alveolar region is altered in their mutant after infection. The endothelium should have also been investigated, especially since their mutant is an endothelial-specific deletion. Within this, the different endothelial cells should have been assessed by a method other than RNAscope such as immunofluorescence, given that this method is unable to show morphology and there are antibodies available.

This data is present in Figure 5. We have also added additional data to the new Figure 5 – Figure Supplement 1 to extend our analysis to 21 dpi and to incorporate a cytoplasmic marker of endothelial cells, Endomucin (EndoM1).

8) Bulk RNA-seq from endothelial cells is used in the manuscript. However, because ATF3 is not specific to Cap1_B cells or even capillaries alone, the downstream gene expression analysis of bulk RNA should be placed into the context of lung endothelial heterogeneity.

We have added qRT-PCR analysis of several downstream genes to address the comments of Reviewer #3, point #3. To place this into the context of endothelial heterogeneity, we have added dot plots to show the expression of selected genes from the RNA-seq analysis in each endothelial subtype from the H1N1 scRNA-seq dataset. These data can be found in the new Figure 4 – Figure Supplement 1. However, because of the relatively low sequencing depth of scRNA-seq compared to bulk RNA-seq, many of the transcripts examined were only present in a small percentage of endothelial cells in the scRNA-seq dataset, so the differences seen are more striking in the RNA-seq data.

9) Although the authors mentioned that the infection with H1N1 influenza can have regional differences, they do not show how they picked regions for their analysis and quantification, and whether ATF3 upregulation was found in more severely affected regions. Furthermore, since they quantified via FACS, this heterogeneity in the infection itself could have affected their observations.

We agree that it is essential both to define the extent of H1N1-mediated inflammation in Atf3 wild-type and knockout mice and to compare this factor between genotypes. We have therefore used a previously published method for quantifying regions of severe, damaged, and normal tissue structure (Liberti et al., Cell Reports 2021) in both Atf3 wild-type and knockout animals. Our results show that Atf3 wild-type and knockout mice have similar levels of tissue damage, and we have added a supplemental figure demonstrating these data (new Figure 3 – Figure Supplement 2). We have also clarified how regions were selected for quantification of alveolar area.

H1N1 influenza injury in mice is heterogeneous, with regions of severe alveolar destruction marked by densely packed immune cells, adjacent regions of damaged tissue, and regions of tissue that appear to have normal tissue structure, as we and others have previously described (Zacharias, Frank et al., Nature 2018; Liberti et al., Cell Reports 2021; Niethamer et al., eLife 2020). However, it has become increasingly apparent that these regions where tissue structure appears normal are actually regions of active regeneration, and endothelial cell proliferation is increased in these regions (Niethamer et al., eLife 2020). We therefore selected 20X fields in these areas to use for quantifying alveolar area, as these are actively regenerating regions where alveolar structures are present for quantification. Because of the changes to tissue structure seen in damaged or destroyed tissue areas, we did not select these regions for quantification, although they were present at similar frequency in Atf3 wild-type and knockout animals.

Author Response

Joint Public Review:

These RNAs come from a screen which is not well described and the descriptions of the sequence analyses are unclear, so it is difficult to know exactly what they are analyzing in the manuscript.

We apologize for not including the required details in the manuscript. The cell cycle lncRNA screen where we identified the initial SNUL-1 probe was published in an earlier paper 6. By performing RNA-seq in cell cycle synchronized samples, we identified several hundreds of lncRNAs that differentially expressed in a particular stage of the cell cycle. We performed a large-scale RNA-FISH-based screen to characterize the localization of these cell cycle-regulated lncRNAs. One of the probes in this screen hybridized to SNUL-1 RNA in the nucleolus. The original double-stranded DNA probe that detected the SNUL-1 RNA cloud(s) was mapped to hg38-Chr17: 39549507-39550130 genomic region, encoding a lncRNA. However, other unique non-overlapping probes generated from the Chr17-encoded lncRNA failed to detect the SNUL-1 RNA cloud. Furthermore, BLAST-based analyses failed to align the SNUL-1 hybridized sequence to any other genomic loci. Since a large proportion of the p-arms of nucleolus-associated NOR-containing acrocentric chromosomes is not yet annotated, we speculated that SNUL-1 could be transcribed from an unannotated genomic region from the acrocentric p-arms.

We have now provided the information in the revised manuscript. Specifically, we have provided the details of the PacBio iso-seq, nanopore seq analyses as well as the bioinformatic approaches that were conducted to determine the identity of the full-length SNUL-1 ncRNA.

If these are RNAs with reasonable abundance, then they should be findable without the extensive PCR amplification they appear to have done for the PacBio sequencing (the methods section is not clear on exactly how many rounds of PCR were performed).

We apologize for not providing the essential details. In the PacBio-iso-seq analyses, we utilized the standard protocol (recommended by the scientists from PacBio, who are authors in the manuscript), which included 13 PCR cycles. However, as described in the manuscript, in parallel to PacBio-seq, we also performed nano-pore sequencing of the nucleolus-enriched RNA without any amplification. The SNUL-1 full-length candidate sequence (CS) that we described in the manuscript is the ncRNA that showed 100% sequence similarity in both independent PacBio Iso-seq as well as nanopore seq analyses. We argue that if the SNUL-1 candidate transcripts would have been an artifact of PCR amplification in PacBio-seq, then we would not have obtained the full-length sequence with 100% match in the nanopore-seq reads. We have now included the detailed bioinformatic analyses in the methods section of the ms.

Moreover, given the acknowledged sequence similarities of the SNULs with other RNAs, the possibility of chimaera formation during PCR amplification is high. They are clearly detecting RNAs associated with nucleoli but exactly what they are examining is unclear.

Please see our response above (public Reviewer comment 2). In addition, we performed detailed bioinformatic analyses to test whether the SNUL-1 full-length sequence obtained in the PacBio-seq is not an artifact of PCR amplification. This analysis is described in detail in the methods section under the sub-title “sequencing analyses”.

It is possible that a clear determination of the genomic origin of these RNAs will be complicated by the repetitive sequences in the regions of the genome where they reside.

We thank this reviewer for acknowledging the technical limitation of mapping the genomic locus of SNUL1 genes. We have pointed out this as the limitation of the present manuscript. Mapping the SNUL-1 genomic locus and characterizing the regulatory sequence elements and factors that control the monoallelic expression of SNULs will be part of future research plans.

Note also that the idea of monoallelic expression from rRNA encoding loci is interesting, but has been established in 2009. Title: Allelic inactivation of rDNA loci. Genes Dev. 2009 Oct 15;23(20):2437-47. doi: 10.1101/gad.544509.

We thank the reviewer for pointing out the study from Cedar lab published in 2009. To test the idea that SNULs contribute to allele specific expression of rRNA, which was previously reported by Cedar lab in their 2009 G&D paper, we performed the same set of experiments described in their paper in three different cell lines in the presence or absence of SNULs (please see the response to Editorial comment-2). However, we could not reproduce any of the data presented in the G&D manuscript. Also, we have not seen any other follow up study, where mono-allelic expression of rDNA genes was observed. Currently, no concrete data supports monoallelic expression of rRNA 5. We, therefore, argue that our current study is the first one, demonstrating the mono-allelic association of a ncRNA from the p-arm containing rDNA cluster.

Author Response

Reviewer #1 (Public Review):

The shift from outcrossing to selfing is one of the most prevalent evolutionary events in flowering plants. The ecological and genetic backgrounds of these transitions have been of major interest for decades, and one of the key questions was the dating of this transition. Timing of pseudogenization of the self-incompatibility (SI) genes has been used as a proxy for this transition because loss-of-function mutations of SI genes are often responsible for the evolution of predominant selfing. However, SI genes are identified only in a limited number of taxa, and in some cases, the evolution of selfing is not necessarily associated with loss of SI. Therefore, an independent time estimate of the evolution of selfing by genome-wide polymorphism data has been considered important in this field.

This study provides two statistical methods: SMC-based and ABC-based methods. Both methods intend to detect the genome-wide signatures of the outcrossing-to-selfing transition that alters the ratio of population recombination rate and mutation rate. Authors validated these methods by using the simulated data, confirming that both methods can generally infer the timing of the outcrossing-to-selfing transition jointly with population size changes, although its precision depends on several population history settings.

This study would be an important contribution to the field of mating system evolution. By applying the proposed methods to many other selfing organisms, we may be able to see a general picture of the timescale of the outcrossing-to-selfing transition combined with population size dynamics. At the same time, this is one of the extensions of the SMC method, which has already been well utilized for various inferences, including population size and recombination rate heterogeneity.

We thank the reviewer for his positive comments and acknowledging the novelty and relevance of our study for the field.

I do not find a major weakness in the methodologies of this study, but I have a few comments on their applications to the data of Arabidopsis thaliana. It is important that these estimates largely depend on what input data is used, especially the mutation rate and recombination rate. While the authors claim that their estimate is older than Bechsgaard's estimate (<413 kyrs), these two studies used different mutation rates: the authors used Ossowski's mutation rate, and Bechsgaard used Koch's mutation rate (Koch et al. MBE 2010). To compare these two estimates, it is important to use the same mutation rate. Shimizu & Tsuchimatsu (2015; Ann Rev Eco Evo Syst) in detail discussed this point and showed that Bechsgaard's estimate becomes <1.48 myrs when Ossowski's mutation rate was used (see Figure 4). Then it happens to overlap with the estimate of this study.

Thank you very much for identifying this important problem. It is indeed critical to re-scale Bechsgaard’s age of the transition using the same mutation rate as used in our analysis (Ossowski et al 2010). We now use the rescaled estimate published in your review (Shimizu and Tsuchimatsu 2015, figure 4). We note that Bechsgaard et al did not publish a measure of uncertainty around their estimate of the transition; making it difficult to compare it with our posterior distributions. However, Bechsgaard’s estimate is not contained within the credibility intervals of our posteriors for t_sigma and therefore we consider both results significantly different. We have modified the text accordingly, at page 4 l. 8-10; and p.12 l. 27 to p.13 l. 5

I am also concerned about the genomic regions of Arabidopsis thaliana used for this study. Authors chose specific five regions based on homogeneity of recombination rates and diversity, but how does the estimated change when randomly chosen genomic regions are used? If it is important to choose "preferable" regions according to the homogeneity of recombination rates and diversity, it may be useful to describe how these regions should be chosen for future applications of this method to other organisms.

The genomic intervals used for the application to A. thaliana are indeed not random. They were defined such as to avoid, on each chromosome, the increased diversity observed at and surrounding pericentromeric regions. This effect has already been described by Clarck et al (2007, Science) but however, no explanation for this pattern has been published yet. We have updated the text, including a recommendation for future application to other species, at lines p. 13 l. 8-15 and p. 18 l.25-30, and Figure S15. We have also replicated our analysis of the A. thaliana data using a different set of genomic intervals located outside pericentromeric regions (Figure S15 and S16)

Reviewer #2 (Public Review):

This submission seeks to detect changes in the rate of selfing through pairwise comparison of haplotypes sampled from a population. It begins, as did a previous paper by a subset of the authors (Sellinger et al. 2020), with the well-known theoretical finding that partial selfing increases the rate of coalescence and decreases the rate of crossing-over events in genealogical histories.

I am supportive of pitching this contribution as primarily theoretical, with the very short discussion of the Arabidopsis data provided as a worked example. This perspective increases my enthusiasm, compared to an initial reading. My comments are intended to encourage development.

Some thematic characteristics reduce the impact of the submission. Among these are:

(1) a rather less than a scholarly perspective on previous literature;

(2) tendency to avoid theoretical development in favor of computation;

(3) little interpretation of results of their only analysis of real data.

We have now revised the manuscript along the lines suggested by reviewer 2. We provide more references when needed, have emphasized in the abstract and in the theoretical part of the manuscript that it is primarily a new theoretical/methodological development with an application to A. thaliana data, and have improved the interpretation of the A. thaliana data (see reply to reviewer 1).

Author Response

Reviewer #1 (Public Review):

The authors of this study sought to test whether the optogenetic induction of context-related freezing behavior could be enhanced by synchronizing light pulses to the ongoing hippocampal theta rhythm. Theta is a hippocampus-wide oscillation that strongly modulates almost every cell in this structure, which suggests that causal interventions locked to theta could have a more pronounced impact than open-loop ones. Indeed, the authors found that activating engram-associated dentate gyrus (DG) neurons at the trough of theta resulted in an increase in freezing relative to baseline when averaging across all stimulation epochs. In contrast, open-loop stimulation and peak-locked stimulation had weaker effects. Analysis of local field potentials showed that only the theta-locked stimulation facilitated coupling between theta and mid-gamma, indicating that this manipulation likely enhances the flow of activity from DG to CA1 via CA3 (as opposed to promoting transmission from entorhinal cortex to CA1). Previous results from mice, rats, and humans support the hypothesis that memory encoding and recall occur at distinct phases of theta. This work further strengthens the case for phase-specific segregation of memory-related functions and opens up a path toward more precise clinical interventions that take advantage of intrinsic theta rhythm.

Strengths:

This study recognizes that, when artificially reactivating a context-specific memory, the brain's internal context matters. In contrast to previous attempts at optogenetically inducing recall, this work adds an additional layer of precision by synchronizing the light stimulus to the ongoing theta rhythm. This approach is more challenging, because, in addition to viral expression and bilateral optical fibers, it also requires a recording electrode and real-time signal processing. The results indicate that this additional effort is worth it, as it results in a more effective intervention.

The findings on theta-gamma cross-frequency coupling suggest a possible mechanism underlying the observed behavioral effects: trough stimulation enhances DG to CA1 interactions via CA3. LFP recordings showed that stimulation increases the coupling between theta and mid-gamma (though not in all mice), and the percentage of freezing during reactivation is correlated with the gamma modulation index.

Weaknesses:

Given the precision of the intervention being performed, one might expect to see a stronger behavioral impact. Instead, the overall effect is subtle, and quite variable across mice. Looking at individual data points, the biggest overall increase in freezing actually occurred in 2 mice during the 6 Hz stimulation condition. Furthermore, trough stimulation decreased freezing in 3 mice. This is not a weakness in itself; rather, the weakness lies in the lack of an attempt to make sense of this variability. There are a number of factors that could explain these differences, such as viral expression levels, electrode/fiber placement, and behavior during baseline. There is of course a risk of over-interpreting results from a few mice, but there is also a chance that the results will appear more consistent after accounting for these additional sources of variation.

Although two mice that had negative light induced freezing for trough stimulation, the other 15 mice showed a positive result. Stringent inclusion criteria were used to ensure that mice had adequate viral expression levels and behavior during baseline. Mice without at least 5% light induced freezing in at least two of the four epochs were not included in the study. The negative behavior from some mice is further explained through the correlation between MI and light induced freezing (Figure 5D). 6 Hz showed mixed behavioral results across the different behavioral measures quantified. Additionally, 6 Hz did not show the physiological hallmarks of memory reactivation through the theta-gamma modulation index so having an increased number of negative light induced freezing samples is expected.

Finally, the elevated baseline freezing rate relative to previous literature could have masked some of the behavioral effect.

In the revised manuscript, we discuss the effects of exclusion criteria more clearly.

While trough-locked optogenetic stimulation significantly increases freezing, the effects are much weaker than placing the mouse in the actual fear-conditioned context (average time freezing of 15% vs. 50%). The discussion would benefit from additional treatment of ways to further increase the specificity and effectiveness of artificial memory reactivation.

We have content on future directions for artificial memory reactivation to further approach the behavioral response of natural recall. We believe that incorporating time varying stimulation to different cells or parts of the hippocampus could improve the induced recall value as all current methods stimulate the entire sub-region simultaneously.

Using an open-source platform (RTXI) for real-time signal processing is commendable; however, more work could be done to make it easier to adopt these methods and make them compatible with other tools. The RTXI plugin used for closed-loop stimulation should be fully documented and publicly available, to allow others to replicate these results.

The RTXI plugin can be found here: (https://github.com/ndlBU/phase_specific_stim). The URL has been added in the description of Figure 1.

Author Response

Reviewer #1 (Public Review):

The screening effort has revealed a number of interesting and novel suggestions of new modulators of nuclear appearance that are exciting and have the potential to be of value to the field.

We appreciate the reviewer’s view that identification of new modulators of nuclear morphology is exciting and of value to the field.

Major Points:

1) The discussion of the screen hits and prior knowledge key to their interpretation is lacking. For example, the authors only report on the purported localization of the hits without an unbiased analysis of their function(s). As a sole example, multiple members of the condensin complex are hits in Fig.1 while multiple members of the cohesin complex are hits in Fig. 2 - but there are many more factors worthy of further discussion. Moreover, the authors need to provide more information on the data used to assign the localization of the hits and how rigorous these assignments may be. For example, multiple CHMP proteins (ESCRTs) are listed - indeed CHMP4B is the highest scoring hit in Fig.1 - but this protein does not reside at the nuclear envelope at steady-state; rather, it is specifically recruited at mitotic exit to drive nuclear envelope sealing. Moreover, there are many hits for which there is prior published evidence of a connection to nuclear shape or size that are ignored: examples include BANF1, CHMP7, Nup155 (and likely far more that I am not aware of). This is a missed opportunity to put the findings into context and to provide a more mechanistic interpretation of the type of effects that lead to the observed changes in nuclear appearance. For example, there is already hints as to whether the effects occur as a mitotic exit defect versus an interphase defect, but conceptually this is not addressed.

We appreciate this important point. We find that one of the major challenges in presentation of screening results is to provide detailed information on all interesting hits within the length limits of a manuscript! To provide a more comprehensive picture, we have now performed pathway analysis using STRING to display protein interaction networks to more comprehensively classify hits and groups of hits (Figures S7 and S8). We find highly connected regions in the network corresponding to condensin and histone modifiers in fibroblast hits altering nuclear shape. In contrast, MCF10AT hits showed increased connectivity with nucleoporin proteins. Fibroblast hits displaying an increase in nuclear size identified multiple nucleoporins and MCF10AT hit analysis identified components of DNA replication. We have added these findings to Supplementary Figures 7 and 8 and discuss them on page 16. Also, as requested, we added more than 20 new references and additional information on previously identified functions of some hits discussed in the text on p. 22-24.

2) Validation of the screen is lacking. There appears to be no evidence that the authors validated the initial screen hits by addition siRNA experiments in which the levels of the knock-down could be assessed. As an example: do nucleoporin hits decrease in their abundance at the nuclear envelope in these conditions? This validation is absolutely essential.

As requested, we now include in Tables S6A-C, data from independent validation experiments in which we selected the primary hits and validated them using an independent set of siRNAs with distinct chemistry and target sequences. Additionally, we demonstrate efficient knockdown capabilities for 8 targets in Supplementary Figure 9 with knockdown levels for most siRNAs of at least 60%. We find no strong relationship between knockdown efficiency and the extent of the observed phenotype (compare Figure S9 and Figure S10).

3) Differences in cell type - the authors' interpretation that a lack of overlap in the hits across cell types reveals that there are fundamentally cell type-specific mechanisms at play is a stretch. This could also reflect a lack of robustness in the screen, which should be addressed by directly testing the knock-down of the hits from one cell line in the other. Even if this approach reinforces the cell type specificity, the differences in the biology beyond the nucleus itself - an obvious example being the mechanical state of the cell - organization of the cytoskeleton, adhesions, etc that influence forces exerted on the nucleus are different rather than the nuclear response is different. These caveats needs to be explicitly acknowledged.

As requested, we have now performed side by side experiments between both cell lines to directly compare a subset of nuclear morphology hits in parallel. They are shown in Supplementary Figure 10. We find a number of hits display strong nuclear shape abnormalities in either fibroblasts or MCF10AT cells but not both, with the exception of LMNA, which confirms our screen data. In addition, we compared the hits from our screen with previously published reports of other factors which regulate nuclear morphology to further strengthen our findings. We mention these results on p. 16. Despite these results, we have now toned down our statements regarding cell-type specificity of individual hits considering the small number of cell lines analyzed and the possible cellular factors which could contribute to cell-type specific differences.

4) There are major issues with the interpretation of the presented biochemistry. For example, the basis for the supposed effect of monomer/dimer state of lamin is confusing and likely misinterpreted. It is well established that GST imposes dimerization on proteins expressed as GST fusions independent of cysteines. Any effect of DDT would have to manifest through some other mechanism (disulfides between the lamin domains - assumedly what the authors are thinking). Further, GST will impose dimerization of lamin A and lamin C in the co-incubation experiments. It is therefore entirely expected that if lamin A binds H3 and lamin C does not that the mixed dimers will bind H3 with lower affinity. Critically, this does not, however, address how full-length lamin C influences binding of lamin A to H3 in vivo. Last, how an effect of lamin C on lamin A would manifest through a disulfide bond in the nucleus, which has a reducing environment, is entirely unclear.

We directly tested the possibility that GST causes artifactual dimerization of lamins by mutating cysteines to alanine in GST-lamin and assessing their effect on histone binding experiments. We show the results in Supplementary Figure 14E. If the observed binding were artifactually due to GST-mediated dimerization, we should not expect an effect of the cystine mutants on histone binding. We find, however, that the C522A mutation in lamin A results in increased binding of H3 in the presence of lamin C, demonstrating that the observed effects are not due to GST dimerization. We discuss these results on p. 18 and p. 19.

We agree with the referee that it will be exceptionally challenging to determine the in-vivo relevance of disulfide bonds, not knowing what the precise environment of the nucleus is. Given these caveats, we have now toned down this point and discuss the limitations of these findings in more detail on p. 19, 23, 24, and 25.

5) It is important for the authors to address the concept of nuclear size changes versus changes in the nuclear to cell volume ratio – biologically these could be quite different conditions, but obviously these cannot be distinguished by measuring nuclear volume alone. Addressing this experimentally would be best (to provide more depth to the size measurements).

This is an important point. As requested, we now clearly indicate on p. 23 that we are measuring nuclear area using nuclear cross-sections as a proxy for nuclear size rather than nuclear to cell volume ratio. We have found in our imaging studies over the past two decades that measuring cell volumes is exquisitely challenging and often highly inaccurate. A major challenge in these approaches is the correct identification of cell boundaries and this is particularly challenging in a high-throughput setting since cell volume measurements require z-stacks that greatly complicates the imaging and quantitative analysis and increases the complexity of this kind of analysis of the millions of cells analyzed in a screen. Ultimately, measurements of cell volume for adherent cells will only be estimates (see for example PMID 28622449). We now clearly indicate this limitation of our approach and discuss on p. 15 and 23 previous studies measuring nuclear size and cell volume ratio measurements and how it compares to measuring nuclear area alone. We have also added several references on this topic on p. 15 and 23.

6) There are important caveats to the approach of using the nuclear area as proxy measurement for nuclear size, most prominently that it is highly responsive to changes in nuclear height that can occur for a multitude of reasons (increased height = small radius and decreased height = larger radius), particularly given the different cell types. This needs to be acknowledged directly.

Along the lines of point 5 and as requested, we now more clearly acknowledge on p. 23 these caveats due to our screening method of measuring nuclear area as a proxy for nuclear size. Nuclear cross-sectional area has been experimentally shown to be a good proxy for nuclear size in many systems (see PMID 31085625). For this reason, and because quantifying nuclear size from z-stacks would have greatly complicated the imaging and quantitative analysis, we chose to use nuclear cross-sectional area as our metric for nuclear size. In looking through our data, we did not find any significant differences in nuclear height between the two cell lines used or amongst hits and non-hits. With respect to the issue of different cell types, our analysis focused on RNAi knockdowns that altered nuclear morphology in a given cell line and we did not compare cell lines against each other. Separate analyses were performed for each cell line, so possible differences in nuclear height between the different cell lines used should not affect our analysis. We now discuss these issues on p. 23.

7) What is the evidence that the H3 effects manifest through lamins rather than directly?

We apologize for not being clear. We did not mean to intend to state that H3 acts via lamins. We do find that H3 physically interacts with lamins and that H3.3 mutants (K9M, K27M, and K36M) result in nuclear morphology defects. We now also show in the new Figure S17 that H3.3 mutants slightly affect lamin levels. However, as pointed out by the reviewer, these observations do not categorically rule out non-lamin related mechanisms and we now make it clear in our discussion on p. 20 that the effect of H3 may either be mediated via lamins or independently.

8) Context is needed for the "methyl-methyl" histone states described as being the highest binders in the peptide array experiments. Are these states commonly found? Where in the genome? Does this match any DamID data? Again - more depth of investigation is required.

This is a good point. Unfortunately, to our knowledge there is currently no ChIP-seq human genome map of di-methyl modifications on histone tails available. We were unable to generate or procure the individual dually methylated peptides and methyl-methyl H3 antibodies are not available and we are thus not able to perform quantitative binding assays. However, to begin to address this issue, we now provide in a new Supplementary Table 8 quantitative data of binding intensities. Given these limitations, we have now toned the claims regarding the methyl-binding sites.

9) That oncohistones induce changes in nuclear shape or size does not mean that this is related to the mechanism in cancer. Also - how over-expression of H3 without its obligate partner H4 could disrupt the cell or an assessment of the extent of the oncohistone incorporation into chromatin achieved in these experiments makes it challenging to interpret.

We agree and did not intend to imply that the oncogenic function of the histone mutants involves changes in nuclear morphology. We now clearly state so on p. 25 and we also mention the caveat of the overexpression experiment.

10) Throughout the manuscript it would be helpful to the reader if the author would provide at minimum a brief statement on the previously identified functions of the hits that are explicitly discussed beyond their localization (membrane versus chromatin). References would also be helpful (for example, again - what is the evidence that SLC27A3 resides at the nuclear envelope?).

As requested, we added more than 20 new references and now provide additional information and previously identified functions of many of the hits mentioned in the text.

Author Response

Reviewer #1 (Public Review):

The manuscript by Huang et al. examines the potential "self-policing" of Bacillus cells within a biofilm. The authors first discover the co-regulation of lethal extracellular toxins (BAs) and the self-immunity mechanisms; the global regulator Spo0A controls both. The authors further show that a subpopulation of cells co-express these genes and speculate that these cells engage in preferential cooperation for biofilm formation (over cells that produce neither). Based on previous literature, the authors then evaluate the relative fitness of the wild-type strain compared to mutants locked into either constantly exporting the toxins or permanently immune to these poisons. The wild-type exhibited increased fitness (compared to the mutants) for the tested biofilm conditions. The manuscript raises interesting ideas and provides a potential model to probe questions of cooperatively in Bacillus biofilms.

Strengths:

• The authors use fluorescence-producing reporter strains to discern the spatial expression patterns within biofilms. This real-time imaging provides striking confirmation of their conclusions about shared co-regulation.

• The authors also nicely deploy genetic constructs in microbiological assays to show how toxin production and immunity can influence biofilm phenotypes, including resilience to stress.

Thank you very much for your positive comments. The detailed response to your comments and suggestions are as follows.

Concerns:

• My biggest concern is that the claim of policing on a single-cell level needs more quantitive microscopy, particularly of the xylose-induced strain. The data support a more tempered consideration of self-policing via BAs and self-resistance in this Bacillus species. It seems sufficient that this manuscript opens the door for a novel and readily examinable system for examining potential cooperation and its molecular controls (without making broader claims).

Thank you very much for your comments. We demonstrated the policing system on a single-cell level by re-filming the progress of individual nonproducers from alive to death and even disappearance in a biofilm population (please see the pictures in Figure 2 and the statistical data in Figure 2-figure supplement 1 of the revised manuscript, as well as revised Figure 2-video 1-4). Alternatively, the xylose-induced strain (SQR9-Pxyl-accDA) was constructed to assess the involvement of AccDA expression (controlled by Spo0A~P in wild-type while induced by exogenous xylose here) in regulating BAs synthesis and immunity. The expression of AccDA is likely to be homogeneous in the colony with xylose addition, instead of a heterogeneous expression in the wild-type population.

• The discussion is more speculative than the presented data warrants. For example, the speculation in lines 289 - 310 is not anchored in the results. It is hard for this reviewer to imagine how one would use the genetic framework and tools developed in this manuscript to address the ideas proposed in lines 289 - 310.

Thank you for your comments. We have revised the discussion to ensure it is more related to data warrants than speculation. As a complement to the molecular mechanism of the policing system in the discussion, the hypothesis of the evolution of this system (Lines 289-310 in the original version) was included to give a possibility that how it raised, which is based on a couple of ecological theories with regards to division of labour and kin selection4-6; we have shortened this discussion in the revised manuscript.

• Some conclusions (in the results section) are more decisive than the data supports. For example, the microscopy of the PI staining (as presented in Figure 2 and the supplemental movies) does not prove that only non-expressing cells die. Yet the conclusion in line 143 states that "ECM and BAs producers selectively punish the nonproducing siblings." Also, the presented data shows many non-labeled cells without PI; why do some nearby non-gfp-expressing cells remain alive?

Thank you for your constructive comments. According to the reviewer's suggestion, an observation covering more complete biofilm forming process, as well as a more convinced data statistics, should be performed. We then re-conducted microscope observation lasting for 3 h during biofilm formation, and assess the source and location of dead cells for statistical analysis. The results showed that all dead cells were originated from the subpopulation that didn't express the gfp (the nonproducers), and the number of dead cells adjacent to the producers was significantly higher than that closed to the non-producers (please see the pictures in revised Figure 2 and Figure 2-figure supplement 1).

In addition, regarding the survival of some non-gfp-expressing cells near the producers, based on several relevant literatures1-3 and the observation in the present study, we speculate that the coordination system for optimizing the division of labor is relatively temperate, thus only a part of the nonproducers (relative sensitive cells or facing higher concentrations of the toxin) are eliminated. We think this scene is a balance between restraining the cheater-like subpopulation and retaining the advantages of cell differentiation.

Author Response

Reviewer #1 (Public Review):

The work in this study builds on previous studies by some of the same authors and aims to test whether the heartbeat evoked response was modulated by the local/global auditory regularities and whether this differed in post-comatose patients with different contagiousness diagnosis. The authors report that during the global effect there were differences between the MCS and UWS patients.

The study is well constructed and analysed and has data from 148 participants (although the maximum in anyone group was 59). The reporting of the results is excellent and the conclusions are supported by the results presented. This study and the results presented are discussed as evidence that EEG based techniques maybe a low cost diagnostic tool for consciousness in post-comatose patients, although it should be stressed that here no classification of diagnostics was performed on the EEG data.

One potential weakness was the relationship between the design of the experiment and the analysis pathway for the results. If I have understood correctly the experimental design the auditory regularity changed on whether the local/global regularity was standard/deviant. In the analysis the differences between all conditions in which the local or global regularity were compared between the standard and deviant trials. This difference was then compared between MCS and UWS patient groups. For these analyses the results for the health and emerging MCS were not included. If this is correct it would be interesting to understand the motivation for this. Relatedly, it would be good to clarify if the effects reported were corrected for the multiple planned contrasts and if not why they should not be corrected.

Thanks for the appreciation and constructive comments to our work. The misdiagnosis of MCS/UWS patients in the clinical practice occurs because of misdetection of covert consciousness given the absence of overt behavioral signs of consciousness. Therefore, the main motivation of our study is to contribute to a better distinction between those two patients’ groups.

We have modified the introduction to clarify that the objective of the paper is to show in major detail the group differences between MCS and UWS patients:

"In this study, we analyze HERs following the presentation of auditory irregularities, with special regard on distinguishing UWS (n=40) and MCS (n=46) patients. Note that the automated classification of this cohort was previously performed in another study (Raimondo et al., 2017). Therefore, our aim is to characterize the group-wise differences between UWS and MCS patients that may allow a multi-dimensional cognitive evaluation to infer the presence of consciousness (Sergent et al., 2017), but also complement the bedside diagnosis performed with neuroimaging methods that capture neural correlates of covert consciousness (Sanz et al., 2021)."

Reviewer #2 (Public Review):

The goal of this study was to determine whether heartbeat-evoked responses measured at the scalp level with EEG, which followed regularity violations, could potential help inform the diagnosis of patients with altered states of consciousness.

The authors use high density EEG and an oddball paradigm that probes violations of both local and global regularities. Four groups were considered including unresponsive wakefulness syndrome patents, minimally consciousness patients, emerging minimally consciousness patients and healthy controls. A difference was found between unresponsive and minimally conscious patients in the amplitude of the heartbeat evoked responses measure with EEG following a sound that violated a global regularity. Similarly, differences were found between the variance of these responses between the two above mentioned groups (N=58 and N=59), but no differences were found in relation to the healthy control group, which appear to be "in between" the two other groups (at least for global effect of HER). I thought this was a little counterintuitive and raises some questions about what this neural signature can tell us about the state of consciousness. Having said that, the healthy control sample was very small, more than 5 times smaller (only N=11).

Thanks to the reviewer for their comments. As described above, distinguishing between MCS/UWS patients is one of the main challenges in the clinical practice. We have modified the manuscript to show the differences between these two patients’ groups. Further data on EMCS and healthy participants is not included in this revision because of the new inclusion criteria.

In general, I thought the Discussion section was a little light on the implications of the findings, what they tell us about the brain mechanisms of consciousness and their different levels/states. A question is raised about whether it is necessary to lock EEG to heartbeats to find differences between patients. The data appeared to say that this is not the case but the discussion does not appear to reflect that very clearly.

We have enriched the discussion to comment on the relation of HERs in perception:

"Our results contribute to the extensive experimental evidence showing that brain-heart interactions, as measured with HERs, are related to perceptual awareness (Azzalini et al., 2019; Skora et al., 2022). For instance, neural responses to heartbeats correlate with perception in a visual detection task (Park et al., 2014). Further evidence exists on somatosensory perception, where a higher detection of somatosensory stimuli occurs when the cardiac cycle is in diastole and it is reflected in HERs (Al et al., 2020). Evidence on heart transplanted patients shows that the ability of heartbeats sensation is reduced after surgery and recovered after one year, with the evolution of the heartbeats sensation recovery reflected in the neural responses to heartbeats as well (Salamone et al., 2020). The responses to heartbeats also covary with self-perception: bodily-self-identification of the full body (Park et al., 2016), and face (Sel et al., 2017), and the self-relatedness of spontaneous thoughts (Babo-Rebelo et al., 2016) and imagination (Babo-Rebelo et al., 2019). Moreover, brain-heart interactions measured from heart rate variability correlate with conscious auditory perception as well (Banellis and Cruse, 2020; Pérez et al., 2021; Pfeiffer and Lucia, 2017)."

Reviewer #3 (Public Review):

I found the results very interesting but wondered why the ERP results for the global vs. local effects are not reported. This analysis is mentioned in the methods section, but I do not find it in the results. Is this what is shown in the mid row in panel D? If yes, it should be made clearer. Is there a significant local and global deviant response in each patient group?

We thank the reviewer for their appreciation of our work and their comments.

We have reported the new results showing clustered effects in both ERPs and HERs.

Additionally, eyeballing Figure 1, there are a few potential issues that may be affecting the conclusion re HER:

(1) Panel D top: it seems that the orange trace (MCS) is largely the same in both the "Local" and "global" condition. But the blue trace (UWS) shows a larger negative going deflection in the "global" case. Put differently, the UWS, but not MCS patients appear to generate a different response to the Global effect relative to the local effect. Is this the case?

We have separated the Figure 1 into 3 new figures to clarify on the results. And we also provide a more detailed description of our results.

In brief, our results show that MCS may have a distinctive response to global and local effects. We have included new correlation analysis in which we show that the responses to global and local effects are uncorrelated (Table 2):

With respect to the “negative” responses in UWS. Note that the measured effect correspond to a linear combination of evoked potentials, e.g.: global effect = mean(global deviants) – mean(global standard). Therefore, the negative group-wise response may imply that global standard responses are larger than global deviants. We have included in Table 1 the statistical tests to show whether the responses to local and global effects are different from zero:

(2) There are some MCS subjects that appear to show a global effect that is larger than that observed in EMCS and healthy controls. How do you interpret these data?

We have included in the discussion a paragraph in which we discuss on the outliers:

"Note that outliers are expected in disorders of consciousness and exact physiological characterization of the different levels of consciousness remains challenging. First, the standard assessment of consciousness based on behavioral measures has shown a high rate of misdiagnosis in MCS and UWS (Stender et al., 2014). The cause of the misdiagnosis of consciousness arises because consciousness does not necessarily translate into overt behavior (Hermann et al., 2021). Unresponsive and minimally conscious patients, namely non-behavioral MCS (Thibaut et al., 2021), represent the main diagnostic challenge in clinical practice. Second, some of these patients suffer from conditions that may translate to no response to stimuli, even in presence of consciousness. For instance, when they suffer from constant pain, fluctuations in arousal levels, or sensory impairments caused by brain damage (Chennu et al., 2013). Third, these patients were recorded in clinical setups, which may lead to a lower signal-to-noise ratio, and lead to biased measurements in evoked potentials (Clayson et al., 2013)."

(3) How do you interpret the negative average HER data shown by many UWS patients?

As mentioned above, the negative HER is a result of a linear combination of different HER-based markers (deviants minus standard).

Author Response

Reviewer #3 (Public Review):

1) While the data are generally very convincing, the authors overstated the conclusions in several instances. For example, the authors state that EPAC and PKCε are "required" or "essential" for vesicle docking and release. However, the author's own data show that both vesicle docking and release are clearly present (though reduced) in the absence of EPAC and PKCε, demonstrating they are not absolutely required. The language could be toned down without diminishing the impact of the excellent work.

We thank you for these important comments. We have double-checked the manuscript and modified the language of our statements. In particular, we have changed the unnecessary words “required” and “essential” to “regulate” or “important”.

2) The authors used analysis of cumulative EPSCs to estimate release probability (Pr) and the readily releasable pool (RRP) size. Unfortunately, this approach is likely not suited for low release probability synapses such as parallel fibers (the authors estimate Pr to be 0.04-0.06). Thanawala and Regehr (2016) extensively investigated the validity of cumulative EPSC analysis under a variety of conditions. They found that this analysis produces large errors in Pr and RRP at synapses with a Pr below ~0.2. In addition, 20 Hz EPSC stimulation (as was used here) produces much larger errors compared to the more commonly used 100 Hz stimulation. Between the low Pr at parallel fiber synapses and the low stimulus frequency used, it is likely that the cumulative EPSC analysis provides a poor estimate of Pr and RRP in this case.

Thanks for the very insightful comment. In the previous experiments, we measured RRP and Pr based on parameter taken from the work in the hippocampal CA1 neurons (He et al., 2019), which, in our opinion, is similar to PF-PC synapses concerning low release probability. We have carefully read Thanawala and Regher (2016) paper and compared different methods. While the performance of the EQ method is in general more reliable to estimate small RRP and low Pr, it relies on p to be constant throughout a stimulus train (Thanawala and Regher, 2016). Although p may be constant for the calyx of Held synapses they studied, it cannot be case for PF-PC synapses. Therefore, we decided to redo the estimations of RRP and Pr using 100-Hz train (previously 20-Hz train). This method does not require constant p and allows us to have a better estimation on RRP and Pr at PF-PC synapses (Thanawala and Regher, 2016).

The new results have been presented in new Fig. 2E and 2F. The PF-PC synapses were stimulated at the frequency of 100 Hz, and the artifacts were truncated and the EPSCs were aligned (Fig. 2E and 2F). Note that the aim of this experiment was to investigate whether there is difference between control and cKO mice. Indeed, we found that the amplitudes of both EPSC0 and follow-up EPSCs were smaller in cKO mice, indicating that both the initial release and the replenishment are reduced by the conditional knockout o EPACs or PKCε. Compared to 20-Hz train, the 100-Hz train resulted in steady-state EPSCs brought EPSCs into steady state faster. We created linear fit from normalized steady-state EPSCs and back-extrapolated the curve to the y-axis to calculate Pr. Indeed, we found that the Pr value estimated from the 100-Hz train stimulus was significantly larger than that from the 20-Hz train, showing 0.17 (Math1-cre) and 0.19 (PKCεf/f) with 100-Hz, but 0.07 (Math1-cre) and 0.08 (PKCεf/f) in previous submission. This result was similar to Thanawala and Regher (2016), in which they claimed that the accuracy of estimation from a 100-Hz train is about three times of that from a 20-Hz train. Moreover, we found that the conditional knockout of either EPACs or PKCε produced significant decrease on Pr (Math1-cre 0.17 vs Math1-cre;EPAC1cKOEPAC2cKO 0.11; PKCεf/f 0.19 vs PKCεcKO 0.12). These results have been added in the text and figure legend (Fig. 2E and 2F), and corresponding methods have also been updated.

3) Using a combination of genetic knockouts and pharmacology, this paper convincingly shows that presynaptic EPAC/PCKε are necessary for presynaptic LTP, but do not alter postsynaptic LTP/LTD. However, given the experimental conditions in the slice experiments, it is difficult to extrapolate from the slice data to in vivo plasticity during motor learning. Synaptic plasticity in the cerebellar cortex is quite complex and can depend significantly on age, temperature, location, and ionic conditions. Unfortunately, these were not well matched between slice and in vivo experiments. Slice experiments used P21 mice, while in vivo experiments were performed at P60. Slice experiments were performed in the vermis, while VOR expression/adaptation generally requires the vestibulo-cerebellum/flocculus. Slice experiments were performed at room temperature, not physiological temperature. Lastly, slice experiments used 2 mM Ca2+ in the ACSF, somewhat high compared to the physiological extracellular fluid. Each of these factors can significantly affect the induction and expression of plasticity. These differences leave one wondering how well the slice data translate into understanding plasticity in the in vivo context.

This is a great question. To date, almost all PC plasticity in published work were recorded in young adult mice (< 1 month) and at room temperature, and most behavioral experiments were conducted around 2-3 months of age. To better answer the reviewer’s comment, we tried our best to redo the LTP experiments under the requested, alternative conditions (in 2-month-old mice, low Ca2+ or high recording temperature). Our new data show that, under these conditions, EPACs and PKCε are still needed for the induction of presynaptic PC-LTP (Figure 3–figure supplement 2-4). In addition, we have tried to record PC EPSCs in the flocculus. Unfortunately, we found PC EPSCs there were quite unstable, which might be due to the more complex orientation of PCs and their innervations. We have discussed the reviewer’s comment in the revised manuscript “Second, presynaptic PF-PC LTP was performed in the cerebellar vermis in the present work, whereas VOR learning generally requires PC activity in the flocculus. Unfortunately, we found that PC-EPSCs in the flocculus were not suitable to record PC plasticity because they were unstable” (Line 557).

4) Many experiments use synaptosomal preparation. The authors identify excitatory synapses by VGLUT labelling, but it is unclear how, or if, the authors distinguish between parallel fiber, climbing fiber, and mossy fiber synaptosomes. These synapses likely have very different properties and molecular composition, some quantification or estimation of how many synaptosomes are derived from each type of synapse would be helpful.

We have performed synaptosome staining vGluT1/vGluT2, EAAT4 and bassoon to identify PF-PC synapses (vGluT1+EAAT4+) or CF-PC (vGluT2+EAAT4+) synapses. Our staining results showed that PF-PC synapses covered 88.8% of the total and CF-PC synapses covered 7.5% of the total. Thus, we estimated the number of mossy fiber synapses to be less than 3.7%, which would not affect our conclusion. These results have been presented in Figure 1–figure supplement 1.

5) The math1-cre mouse line is used to selectively knockout EPAC or PKCε expression in cerebellar granule cells. This line also expresses Cre in unipolar brush cells (UBCs) of the cerebellum (Wang et al., 2021). This is likely not a factor in the molecular/slice studies of EPAC/PKC signaling, but UBC dysfunction could play a role in motor/learning deficits observed in vivo. This possibility is not considered in the text.

There is indeed evidence that UBCs are involved in cerebellar ataxias (Kreko-Pierce et al., 2020). How UBCs precisely participate in motor learning or VOR learning is unclear, but they are suggested be involved in motor performance (Mugnaini et al., 2011; Guo et al., 2021). So, we agree with the reviewer that this option cannot be excluded. Therefore, we have revised the discussion about the potential role of UBCs “Two caveats should be considered in the present studies. First, Math1-Cre-induced deletion of EPAC or PKCε might affect the function of unipolar brush cells (UBCs), which are involved in cerebellar ataxias (Kreko-Pierce et al., 2020). However, we believe that the EPAC-PKCε module regulates VOR learning through presynaptic plasticity mechanism at PF-PC synapses rather than UBCs, in line with the observations in other granule cell-specific mutations (Galliano et al., 2013; Schonewille et al., 2021).” (Line 552).

References:

Mugnaini E, Sekerková G, Martina M. The unipolar brush cell: a remarkable neuron finally receiving deserved attention. Brain Res Rev. 2011;66(1-2):220-45.

Guo C, Rudolph S, Neuwirth ME, Regehr WG. Purkinje cell outputs selectively inhibit a subset of unipolar brush cells in the input layer of the cerebellar cortex. Elife. 2021;10:e68802.

Kreko-Pierce T, Boiko N, Harbidge DG, Marcus DC, Stockand JD, Pugh JR. Cerebellar ataxia caused by Type II unipolar brush cell dysfunction in the Asic5 knockout mouse. Sci Rep. 2020;10:2168.

Author Response

Reviewer #1 (Public Review):

This paper uses light field microscopy to measure calcium signals across the fly brain while it is walking and turning, and also while the fly is externally driven to walk and turn, using a treadmill. The authors drive calcium indicator expression using pan-neuronal drivers, as well as drivers specific to individual neurotransmitters and neuromodulators. From their experiments, the authors show that inhibitory and excitatory neurons in the brain are activated in similar patterns by walking and that neurons expressing machinery for different neuromodulatory amines tend to show differentially strong calcium signals during walking. By examining spontaneous and forced walking and turning, the authors identify brain regions that activate before spontaneous turning and that activate asymmetrically in concert with spontaneous or forced turning.

Strengths: Overall, the strength of this paper is in its careful descriptions and analyses of whole brain activation patterns that correlate with spontaneous and forced behaviors. Showing how the pattern of activity relates to broad classes of cells is also useful for understanding brain activation. Especially in brain regions identified as preceding spontaneous walking and in being asymmetrically involved in spontaneous and forced turning, it provides a wealth of potential hypotheses for new experiments. Overall, it contributes to a coarse-grained understanding of broad changes in brain activity during behavior.

Weaknesses: The primary weakness of this paper is that it presents some speculative interpretations and conclusions too strongly. Most importantly, average activity in a neuropil can represent the calcium activity of hundreds or thousands of neurons, and it is hard to know what fraction is active, for instance, or how expression pattern differences might play into calcium signals. Calcium signals also do not reliably indicate hyperpolarization, so a net increase in the average Ca++ indicator signal does not necessarily reflect that the average neuron is becoming more active, just that some labeled neurons are becoming more active, while others may be inactive or hyperpolarized. The conclusions about regions triggering walk (rather than just preceding it) are too strong for the manipulations in this paper, as are some of the links with individual neuron types. Thus, more presenting substantial caveats is required for the conclusions being drawn from the data presented here.

We thank the reviewer for their assessment and the positive comments on our manuscript. We have made these caveats clear throughout the manuscript by adding text and removing overly strong conclusions and speculations.

Reviewer #2 (Public Review):

Aimon et al. used fast whole-brain imaging to investigate the relationship between walking and neural activity in adult fruit flies. They find that increases in brain-wide activity are tightly correlated with walking behavior, and not with grooming or flailing, and are independent of visual input. They reveal that excitatory, inhibitory, and neuromodulatory neurons all contribute to brain-wide increases in neural activity during walk. Aimon et al. extend their observations of brain-wide activity to reveal that activity in some inferior brain regions is more correlated with walk than in other brain regions. The authors further analyzed their imaging dataset to identify candidate brain regions and cell types that may be important for walking behavior, which will be useful in hypothesis generation in future studies. Finally, the authors show that brain-wide activity is similar between spontaneous and forced walk and that severing the connection between the ventral nerve cord and central brain abolishes walk-related increases in brain activity. These results suggest that increases in brain-wide activity during walking may be largely attributed to sensory and proprioceptive feedback ascending to the central brain from the ventral nerve cord rather than to top-down executive and motor control programs. The observations presented in this study suggest hypotheses that may be tested in future studies.

Strengths: This paper presents a rich imaging dataset that is well-analyzed and cataloged, which will be valuable for researchers who use this paper for future hypothesis generation. The comparison of many different reagents, imaging speeds, and behavioral conditions suggests that the observed increases in brain-wide activity during walking are quite robust to imaging methods in adult fruit flies.

Weaknesses: This study is largely observational, and the few experimental manipulations presented are insufficient to support the author's broad claims about the generation of brain-wide neural activity.

We thank the reviewer for their assessment and have toned down claims throughout the paper accordingly.

Notably, the authors suggest that their image analysis can reveal individual cell types that are important for walking by matching their morphologies to registered components from whole-brain imaging experiments. While these predictions are a useful starting point for future experiments, they have not convincingly shown that their method can identify individual cell types in genetic reagents with more restricted expression patterns. Adding further validation to show that genetically subtracting the candidate neurons from the overall expression pattern of the calcium indicator abolishes that component from the response would strengthen this claim. Furthermore, imaging the matched candidate neuronal cell type to show that it recapitulates the activity dynamics of the proposed component would add additional evidence.

We agree that the correspondence to specific neuron types is often very speculative. We have clarified this throughout the manuscript. There are a few exceptions where the neurons we discuss are the only known neurons in a specific GAL4 expression pattern in a given region, and where we find the exact anatomical pattern matching these neurons’ anatomy. Together, this makes us quite confident that the activity results indeed from these neurons. However, the experiments proposed by the reviewer would be interesting complementary approaches. We believe, however, that abolishing activity in one neuron will be difficult to interpret regarding the neuron type as it would affect the activity of other neurons in the network (which is, in our opinion, an interesting point and research direction). Nevertheless, we plan to perform such experiments and experiments looking at the activity in more restricted drivers in the future.

In addition, increases in neural activity prior to walk onset in specific brain regions are intriguing but insufficient to demonstrate the neurons in these regions trigger walking. This claim should await further studies that employ targeted and acute manipulation of neural activity, as noted by the authors. Furthermore, that activity in these brain regions is significantly increased prior to walk onset awaits more rigorous statistical testing, as do the authors' claims that spontaneous versus forced walking alters these dynamics. The suggestion that walking increases brain-wide activity via feedback from the ventral nerve cord is an interesting possibility and would also benefit from additional experimental validation. Activating and silencing neurons that provide proprioceptive feedback from the legs and determining the effect of this manipulation on brain-wide neural activity would be a good starting point.

We have removed claims of causality in the result section. We have also added a statistical test for activation before walk onset. Activating and silencing proprioceptive neurons from the legs would be interesting follow up experiments although it is likely to affect walking. Nevertheless, we are planning to carry out such experiments in the future. We have added this point in the discussion.

Reviewer #3 (Public Review):

Aimon and colleagues investigated brain activity in flies during spontaneous and forced walking. They used light-field microscopy to image calcium activity in the brain at high temporal resolution as the animal walked on a ball and they used the statistical inference methods PCA and ICA to tease out subregions of the brain that had distinct patterns of activity. They then sought to relate those patterns to walking. Most interesting are the experiments they performed comparing forced walking to spontaneous walking because this provides a framework to generate hypotheses about which aspects of neural activity are reporting the animal's movements versus generating those movements. The authors identify subregions and neuron types that may be involved in generating vs reporting walking. Their analysis is reasonable but could be further strengthened with a more powerful statistical framework that explicitly considered the multiple hypotheses being tested. More broadly, the work serves as a starting point to investigate the role of different regions in the brain and should spur follow-up investigations that involve more perturbative approaches in addition to the correlative approaches presented here.

We thank the reviewer for their overall positive assessment of our work and fully agree with the conclusion of its current limitations.

Author Response:

Reviewer #1 (Public Review):

Tomasi et al. performed a combination of bioinformatic, next-generation tRNA sequencing experiments to predict the set of tRNA modifications and their corresponding genes in the tRNAs of the pathogenic bacteria Mycobacterium tuberculosis. Long known to be important for translation accuracy and efficiency, tRNA modifications are now emerging as having regulatory roles. However, the basic knowledge of the position and nature of the modifications present in a given organism is very sparse beyond a handful of model organisms. Studies that can generate the tRNA modification maps in different organisms along the tree of life are good starting points for further studies. The focus here on a major human pathogen that is studied by a large community raises the general interest of the study. Finally, deletion of the gene mnmA responsible for the insertion of s2U at position 34 revealed defects in in growth in macrophage but in test tubes suggesting regulatory roles that will warrant further studies. The conclusions of the paper are mostly supported by the data but the partial nature of the bioinformatic analysis and absence of Mass-Spectrometry data make it incomplete. The authors do not take advantage of the Mass spec data that is published for Mycobacterium bovis (PMID: 27834374) to discuss what they find.

Important points to be considered:

1) The authors say they took a list of proteins involved in tRNA modifications from Modomics and added manually a few but we do not know the exact set of proteins that were used to search the M. mycobacterium genome.

Thank you for pointing out this issue. We will add the complete list of proteins used for the BLAST query.

2) The absence of mnmGE genes in TB suggested that the xcm5U derivatives are absent. These are present in M. bovis (PMID: 27834374). Are the MnmEG gene found in M. bovis? If yes, then the authors should perform a phylogenetic distribution analysis in the Mycobacterial clade to see when they disappeared. If they are not present in M. bovis then maybe a non-orthologous set of enzymes do the same reaction and then the authors really do not know what modification is present or not at U34 without LC-MS. The exact same argument can be given for the xmo5U derivatives that are also found in M.bovis but not predicted by the authors in M. tuberculosis.

The reviewer raises a valid point. In M. bovis mnm5U and cmo5U derivatives were observed in LC-MS analysis. However, we did not identify candidate genes known to be involved in the biogenesis of mnm5U and cmo5U in the Mycobacteriaceae, including M. bovis and Mtb, suggesting that if these modifications are indeed present, they are not synthesized through a canonical biogenesis pathways in this family. There are several examples where the same modification is generated by distinct modification enzymes (Kimura, 2021). These observations raise the interesting possibility that in the Mycobacteriaceae and most species in actinomycetota (except for Bifidobacterium, Corynebacterium and Rhodococcus species), major wobble modifications are generated by biosynthesis pathways that are distinct from those employed by well-characterized organisms. Future studies will examine this hypothesis.

3) Why is the Psi32 predicted by the authors because of the presence of the Rv3300c/Psu9 gene not detected by CMC-treated tRNA seq while the other Psi residues are? Members of this family can modify both rRNA and tRNA. So the presence of the gene does not guarantee the presence of the modification in tRNAs

Thank you very much for the careful read. We did not include RluA in the list of query proteins because it is not classified as a tRNA modification enzyme in Modomics. Additionally, the CMC-coupled tRNA-seq is imperfect for detection of all pseudouridylated positions. Due to this limitation, we only assigned modifications that are both predicted by the presence of putative biosynthetic enzymes and RT-derived signatures. As the reviewer points out, we cannot rule out that this homolog targets only rRNAs. We will clarify this possibility in the revised manuscript. Also, RluA will be added to the query and the name of Rv3300c will be changed to RluA in the text and related figures.   

4) What are tsaBED not essential but tsaC (called sua5 by the authors) essential?

Thank you for pointing out this interesting observation. We are also curious about differences in the essentiality among t6A biogenesis genes. We speculate that TsaC potentially has critical roles in cell viability other than t6A synthesis. TsaC synthesizes a compound, threonylcarbamoyl-AMP, as an intermediate for t6A biogenesis. Thus, it is possible that this intermediate has a role in other essential cellular activities besides t6A biogenesis. Further study of these factors in Mtb could reveal interesting crosstalk between modification synthesis and other cellular activities.

Reviewer #2 (Public Review):

In this study, Tomasi et al identify a series of tRNA modifying enzymes from Mtb, show their function in the relevant tRNA modifications and by using at least one deleted strain for MnmA, they show the relevance of tRNA modification in intra-host survival and postulate their potential role in pathogenesis.

Conceptually it is a wonderful study, given that tRNA modifications are so fundamental to all life forms, showing their role in Mtb growth in the host is significant. However, the authors have not thoroughly analyzed the phenotype. The growth defect aspect or impact on pathogenesis needs to be adequately addressed.

- The authors show that ΔmnmA grows equally well in the in vitro cultures as the WT. However, they show attenuated growth in the macrophages. Is it because Glu1_TTC and Gln1-TTG tRNAs are not the preferred tRNAs for incorporation of Glu and Gln, respectively? And for some reason, they get preferred over the alternate tRNAs during infection? What dictates this selectivity?

Thank you very much for raising this excellent point. As the reviewer suggests, the attenuation of DmnmA Mtb growth inside of macrophages could be caused by disparate codon usage between genes required for in vitro growth and intracellular growth. Among multiple codons encoding Glu, Gln, or Lys, s2U modification-dependent codons might be preferentially distributed in genes associated with intracellular growth. For example, Mtb has two tRNA isoacceptors, Glu1_TTC and Glu2_CTC, to decipher two Glu codons, i.e., GAA and GAG. According to the wobble pairing rule, GAA is only decoded by Glu1_TTC, whereas GAG is decoded by both Glu1_TTC and Glu2_CTC; i.e., GAG can be deciphered by an s2U-independent tRNA. Thus, genes required for intracellular growth might be enriched with GAA, an s2U-dependent codon. The same thing can happen to other Gln and Lys codons deciphered by s2U-containing tRNAs. In the revised manuscript, we will include the perspective of codon usage for explaining the intracellular fitness defect of the ΔmnmA Mtb mutant.

- As such the growth defect shown in macrophages would be more convincing if the authors also show the phenotype of complementation with WT mnmA.

The reviewer raises a valid point. We note however, that Rv3023c, a putative transposase, is downstream of MnmA and unlike MnmA, Rv3023c appears to be dispensable for in vivo growth, according to the Tn-seq database. Therefore, it is likely that the intracellular growth defect is caused by loss of mnmA.

An important consideration here is the universal nature of these modifications across the life forms. Any strategy to utilize these enzymes as the potential therapeutic candidate would have to factor in this important aspect.

This is a valid point. Targeting a pathogen-specific system enables avoidance of the adverse side effects caused by many therapeutic reagents. There are a couple of Mtb modification enzymes that are specific to bacteria and critical for Mtb fitness (e.g., TilS). These enzymes represent ideal potential therapeutic targets to suppress Mtb intracellular growth.

Reviewer #3 (Public Review):

The work presented in the manuscript tries to identify tRNA modifications present in Mycobacterium tuberculosis (Mtb) using reverse transcription-derived error signatures with tRNA-seq. The study identified enzyme homologs and correlates them with presence of respective tRNA modifications in Mtb. The study used several chemical treatments (IAA and alkali treatment) to further enhance the reverse transcription signals and confirms the presence of modifications in the bases. tRNA modifications by two enzymes TruB and MnmA were established by doing tRNA-seq of respective deletion mutants. Ultimately, authors show that MnmA-dependent tRNA modification is important for intracellular growth of Mtb. Overall, this report identifies multiple tRNA modifications and discuss their implication in Mtb infection.

Important points to be considered:

- The presence of tRNA-based modifications is well characterised across life forms including genus Mycobacterium (Mycobacterium tuberculosis: Varshney et al, NAR, 2004; Mycobacterium bovis: Chionh et al, Nat Commun, 2016; Mycobacterium abscessus: Thomas et al, NAR, 2020). These modifications are shown to be essential for pathogenesis of multiple organisms. A comparison of tRNA modification and their respective enzymes with host organism as well as other mycobacterium strains is required. This can be discussed in detail to understand the role of common as well as specific tRNA modifications implicated in pathogenesis.

The reviewer raises a fair point. However, with the exception of Chionh et al., the other studies cited here are not genome-wide characterization of tRNA modification. We will add a discussion of the distribution of tRNA modification enzymes across multiple mycobacterium species and the implications of this distribution for pathogenesis to the revised manuscript.

- Authors state in line 293 "Several strong signatures were detected in Mtb tRNAs but not in E. coli". Authors can elaborate more on the unique features identified and their relevance in Mtb infection in the discussion or result section.

Thank you for the suggestion. We will lengthen the discussion of the RT-derived signatures observed in Mtb but not in E. coli but the relevance of these modifications for Mtb pathogenicity remains speculative at this point.

- Deletion of MnmA is shown to be essential for E. coli growth under oxidative stress (Zhao et al, NAR, 2021). In similar lines, MnmA deleted Mtb suffers to grow in macrophage. Is oxidative stress in macrophage responsible for slow Mtb growth?

This is an excellent hypothesis which we will raise in the revised manuscript.

- Authors state in line 311-312 "Mtb does not contain apparent homologs of the tRNA modifying enzymes that introduce the additional modifications to s2U". This can be characterised further to rule out the possibility of other enzyme specifically employed by Mtb to introduce additional modification.

The reviewer raises a valid point. As discussed above (Reviewer #1, pt 2), Mtb may employ distinct enzymes to generate certain tRNA modifications. Future mass spec-based analyses of Mtb tRNAs will be carried out to identify the precise chemical structure of the sulfurated uridine, and subsequent studies will attempt to determine the enzymes that account for the biogenesis of these modifications.

Author Response

Reviewer #1 (Public Review):

This refinement of their model, coupled with the demonstration that the Sis1 J protein chaperone does not appear to play a direct role in the inactivation phase of the HSR, provide a significant advance over their earlier work.

We are pleased that the reviewer is satisfied that our new results represent a significant advance.

A main weakness is that while the evidence that Sis1 is important for fitness of heat-stressed yeast cells is reasonable, exactly how Sis1 achieves this is not clear. In a single sentence the authors suggest that Sis1 might be an orphan ribosome chaperone, partly based on its nucleolar localization, but provide no evidence for this. If this were true, then one might expect a reduction in ribosome content under stress conditions (because there are more ORPS to take care of because of translation stalling?) and a decreased rate of protein synthesis (yes, this happens, how much this is due to overall translation suppression vs there being less ribosomes to translation things, is unknown and hard to test), which could be tested. Some further insights into this more general role of Sis1 would strengthen the authors' conclusions.

We would like to make a distinction between the important biochemical roles for Sis1 in the cellular response to heat shock – which we explore elsewhere – and the role we are investigating here for the regulation of Sis1 expression by Hsf1. For new insights into the functional role of Sis1 as a chaperone for orphan ribosomal proteins, please see our recent preprint (Ali et al., https://www.biorxiv.org/content/10.1101/2022.11.09.515856v1). Here, we have focused on how Sis1 transcriptional regulation promotes fitness. Please see above for the description of the new mechanistic insight we have into the role of Sis1 expression tuning in controlling stress granules.

Moreover, whether Sis1 plays a general role in the fitness of cells under stress has not been firmly established, i.e., is its mechanistic role the same in heat shock conditions and under nutrient stress conditions? Without knowing the mechanistic basis for how Sis1 maintains the fitness of heat-stressed cells, it is not possible to conclude that the same mechanism is at play in cells grown on a non-preferred carbon source.

As described above, we have now provided evidence that the inability to properly tune Sis1 expression levels in the 2xSUP35-SIS1 strain results in disrupted stress granule homeostasis, linking a known function of Sis1 to a known process driven by nutrient stress.

Figure 4: This is an ingenious experiment to study the subcellular localization of newly synthesized Sis1 in response to heat shock, compared to that of the heat-shock inducible Hsp70 Ssa1. However, based on the images presented in panel B it is hard to know how discrete the subnuclear distributions of Sis1 and Ssa1 really are, and ideally what is needed is to be able to analyze their localizations when both tagged proteins are expressed in the same cell, although this would obviously not be possible using the halo-tagged protein system. In addition, one would like to know the localization of Hsf1 in the cell at the same time. As it stands, these data seem overinterpreted, and it remains possible that some other event such as an inactivating post-translational modification of Sis1 under heat shock conditions might be involved in inactivating its function.

To address this concern, we constructed two new imaging strains expressing Hsf1-mVenus/Halo-Sis1 and Hsf1-mVenus/Halo-Ssa1 (Hsp70) and used pulse-labeling followed by live lattice light sheet 3D imaging to resolve the subcellar localization of newly synthesized Sis1 and Hsp70 with respect to Hsf1 over a heat shock time course. Unfortunately, we cannot monitor newly induced Sis1 and newly induced Hsp70 simultaneously in the same cells with the HaloTag pulse labeling system. We found that a significantly greater fraction of newly synthesized Hsp70 colocalizes with Hsf1 than new Sis1. Thus, while we cannot directly image new Sis1 and Hsp70 in the same cell, we clearly observe a differential localization pattern with respect to Hsf1. These data are included in the revised Figure 4.

One way to establish whether Sis1 nucleolar sequestration prevents it from acting on Hsf1 during the inactivation phase of the HSR would be to selectively disrupt its nucleolar localization signal eliminated while retaining its nuclear localization and determine how expression of such a mutant perturbed the inactivation kinetics of the HSR.

Unfortunately, there is no known Sis1 nucleolar localization signal that we could use in the experiment you propose. In the preprint described above, we show that direct interactions with oRPs recruit Sis1 to the nucleolar periphery, but we do not yet know binding to oRPs is competitive with binding to Hsf1.

Reviewer #2 (Public Review):

This study aims to provide a needed update and validation of a previously outlined mathematical model that describes HSR/Hsf1 regulation. The purpose of the update is to incorporate the impact of newly translated proteins as negative regulators of Hsf1 following heat shock. A requirement for ongoing translation to mount the HSR and activate Hsf1 has been described in several recent studies. Moreover, the study addresses the role of the Hsp70 cochaperone Sis1 in HSR regulation, including its potential function in negative feedback regulation following heat-shock.

The main strength of the study is that it combines quantitative modeling with a well-defined experimental system to generate data. Overall, the model appears to accurately reflect the behavior of HSR under the employed experimental conditions and provides and elegant example of a formalized model for this simple regulatory circuit. Another strength of the study is that it addresses the functional involvement of Sis1 in HSR/Hsf1 regulatory mechanisms and rules out Sis1 involvement in negative feedback regulation of Hsf1 following heat shock. This finding is of importance in light of the complexity of Sis1 involvement in HSR/Hsf1 regulation suggested by the literature. The authors also document a need for endogenous SIS1 promoter regulation during growth on non-fermentable carbon sources.

The study is important for the advancement of Hsf1 research and it may provide inspiration for the study of other chaperone-titrated transcriptional mechanisms such as the UPR or bacterial stress sigma factors.

We thank the reviewer for the generous evaluation.

Reviewer #3 (Public Review):

This paper follows other excellent work from the Pincus laboratory detailing the molecular mechanisms of Hsf1 regulation and extending experimental observations into predictive mathematical models. Overall, the work is top-quality, however, the findings are incremental in nature with respect to our understanding of the HSR and refine existing models rather than break new experimental or conceptual ground. Additionally, the relevance of the non-fermentable carbon source growth phenotype for the 2XSUP35pr-SIS1 strain is unclear with respect to HSR regulation.

We thank the reviewer for this fair assessment of the work.

Author Response

Reviewer #1 (Public Review):

Pelentritou and colleagues investigated the brain’s ability to infer temporal regularities in sleep. To do so, they measured the effect on brain and cardiac activity to the omission of an expected sound. Participants were presented with three different categories of sounds: fixed sound-to-sound intervals (isochronous), fixed heartbeat-to-sound intervals (synchronous), and a control condition without any regularity (asynchronous). When omitting a sound, they observed a difference in the isochronous and synchronous conditions compared to the control condition, in both wakefulness and sleep (NREM stage 2). Furthermore, in the synchronous condition, sounds were temporally associated with sleep slow waves suggesting that temporal predictions could influence ongoing brain dynamics in sleep. Finally, at the level of cardiac activity, the synchronous condition was associated with a deceleration of cardiac frequency across vigilance states. Overall, this work suggests that the sleeping brain can learn temporal expectations and responds to their violation.

We thank the reviewer for the very useful and informed comments, to which we carefully reply below.

Major strengths and weaknesses:

The paradigm is elegant and robust. It represents a clever way to investigate an important question: whether the sleeping brain can form and maintain predictions during sleep. Previous studies have so far highlighted the lack of evidence for predictive processes during sleep (e.g. (Makov et al., 2017; Strauss et al., 2015; Wilf et al., 2016)). This work shows that at least a certain type of prediction still takes place during sleep.

However, there are some important aspects of the methodology and interpretations that appear problematic.

(1) The methodology and how it compares to previous articles would need to be clarified. For example, the Methods section indicates that the authors used a right earlobe electrode as a reference. This is quite different from the nose reference used by SanMiguel et al. (2013) or in Dercksen et al. (2022). This could affect the polarity and topographies of the OEP or AEP and thus represents a very significant difference. Likewise, SOs are typically detected in a montage reference to the mastoids. Perhaps the left/right asymmetries present in many plots (e.g. Figure 3) could be due to the right earlobe reference used.

We thank the reviewer for raising this important point which has prompted us to clarify the reference choice in the manuscript both for completing the information about data recordings in our experiment and for emphasizing the influence of the reference on the EEG results and how they compare to previous reports.

First, we would like to clarify that although EEG data is referenced to the right earlobe online, electrophysiological data from both earlobes were acquired and offline re-referencing to paired earlobes was performed. This is now clarified in the Methods section on page 26, lines 648-651 as follows:

‘Continuous EEG (g.HIamp, g.tec medical engineering, Graz, Austria) was acquired at 1200 Hz from 63 active ring electrodes (g.LADYbird, g.tec medical engineering) arranged according to the international 10–10 system and referenced online to the right earlobe and offline to the left and right ear lobes.’

Additionally, after preprocessing, we performed common average re-referencing, as is common practise and recommended in the literature (see e.g. Niso et al., 2022), and hence the initial online referencing is no longer of relevance. Nonetheless, we agree with the reviewer that different online and offline referencing schemes could explain why some results in the literature are not optimally reproducible. We have clarified this point in the discussion on page 17, lines 408-411 as follows:

‘Finally, while we used largely similar pre-processing (i.e. filters) and experiment implementation (i.e. online and offline reference) as in Chennu et al. (2016), this was not the case for other studies with which direct comparisons are unwarranted.’

For the SO analysis chosen reference (linked earlobes online and common average offline in our case) we acknowledge that - as the reviewer mentioned - many groups indeed employ mastoid re-referencing for SO detection (e.g. Siclari et al., 2018; Schneider et al., 2020; Ameen et al., 2022). However, to the best of our knowledge, this is not a standard choice, as many other groups choose a linked earlobe reference for online SO detection and the mastoids only for offline SO detection (Ngo et al., 2013; Besedovsky et al., 2017; Ngo and Staresina, 2022). In addition, other recent studies used linked earlobe referencing (Bouchard et al., 2021) or common average re-referencing (Züst et al., 2019) for offline SO detection. In our study we opted for using the same average reference for SO detection and evoked potential analysis in order to be able to relate the results of the omission evoked response comparison to that of the SO analysis.

Also, the authors did not use the same filters in wakefulness and sleep, which could introduce an important bias when comparing sleep and wake results or sleep results with previous wake papers.

We fully agree with the reviewer and thank him/her for this suggestion. We have now re-analysed the wakefulness data using a bandpass filter of 0.5-30 Hz as used for the sleep data. The chosen filtering range is commonly used in sleep research. Moreover, Chennu et al. (2016) employed a very similar filtering range (0.5-25 Hz) in an omission EEG study, whose results are similar to ours (Chennu et al., 2016). This new preprocessing resulted in a higher number of valid trials (average trial number: before N=245, now N=286) in wakefulness. Hence, the data from more participants could be used (before N=21, now N=23) and the statistical power of observed differences in our comparisons was improved. The Methods section has been updated accordingly on page 31, lines 763-764 as follows:

‘Continuous raw EEG data were band-pass filtered using second-order Butterworth filters between 0.5 and 30 Hz for the wakefulness and sleep session.’

(2) The ERP to sound omission shows significant differences between the isochronous and asynchronous conditions in wakefulness (Figure 3A and Supp. Fig.) but this difference is very different from previous reports in wakefulness. Topographies are also markedly different, which questions whether the same phenomenon is observed. For example, SanMiguel and colleagues observed an N1 in response to omitted but expected sounds. The authors argue that they observe a similar phenomenon in the iso vs baseline contrast, but the timing and topography of their effect are very different from the typical N1. The authors also mention that, within their study, wake and N2 OEPs were "largely similar" but they differ in terms of latencies and topographies (Figure 3A-B). It would be better to have a more objective way to explore differences and similarities across the different analyses of the paper or with the literature.

We concur with the reviewer and reviewing editor, who both pointed that the way we previously analysed (see our reply to the reviewer’s previous comment) and reported our data was sub-optimal. The new analysis of the wake data reveals more similarities with the MMN and to some extend with the omission literature (Figure 4). As requested, we also improved the description of the comparison of our results to those from the literature, in the Discussion section (pages 17-19, lines 391-458).

(3) The authors applied a cluster permutation to identify clusters of significant time points. However, some aspects of this analysis are puzzling. Indeed, the authors restricted the cluster permutation to a temporal window of 0 to 350ms in wake (vs. -100 to 500ms in sleep). This can be misleading since the graphs show a larger temporal window (-100 to 500ms). Consequently, portions of this time window could show no cluster because the analysis revealed an absence of significant clusters but because the cluster permutation was not applied there. Besides, some of the reported clusters are extremely brief (e.g. l. 195, cluster's duration: 62ms), which could question their physiological relevance or raise the possibility that some of these clusters could be false positives (there was no correction for multiple comparisons across the many cluster permutations performed). Finally, there seems to be a duplication of the bar graphs showing the number of significant electrodes in the positive and first negative cluster for Figure 2 Supp. Fig. 1.

We thank the reviewer for raising this point. We have now performed cluster permutation statistical analysis over the entire -100 to 500 ms window in wakefulness, thus matching the temporal window used for the sleep data (Methods, page 34, lines 843-846). Please note that this modified temporal window was applied to the wake data for which the pre-processing had also been modified (see our reply to comment #1 above). With matching analysis for wakefulness and sleep, we now identify clusters of higher or similar significance compared to our earlier results (Cohen’s d for isoch vs asynch = 0.92 now and 0.67 before; for synch vs asynch = 0.91 now and 1.06 before). In addition, for the isoch vs asynch omission response comparisons, overlapping cluster periods are identified in wakefulness (114-159 ms) and sleep (85-223 ms). The relevant results are thoroughly described on pages 9-10, lines 202-210; page 11, lines 238-251, pages 38-39, lines 970-985.

We would like to also mention that while multiple comparisons correction is performed across channels and electrodes in the EEG using cluster permutation statistics, it is true that we do not perform multiple comparisons correction across the many comparisons. We now explicitly mention the lack of this correction for multiple comparisons in the Methods section page 34, lines 840-843 as follows:

‘Of note, the cluster permutation based multiple comparisons correction only applied across channels and latencies when comparing two experimental conditions, however no multiple comparisons correction was applied across the number of comparisons made in this study.’

(4) More generally, regarding statistics, the absence of exact p-values can render the interpretation of statistical outputs difficult. For example, the authors report a significant modulation of the sound-to-SO latency across conditions (p<0.05) but no significant effect of heartbeat peak-to-SO latency (p>0.05). They interpret this pattern of results rather strongly as evidence that the "readjustment of SOs was specific to auditory regularities and not to cardiac input". Yet, examining the reported chi-square values show very close values between the two analyses (7.9 vs. 7.4). It seems thus difficult to argue for a real dissociation between the two effects. Providing exact p-values for all statistical tests could help avoid this pitfall.

To assist the interpretation of statistical analysis results, we have now included exact p-values.

Specifically, for SOs, we agree with the reviewer on the highly similar chi-squared values for the two analyses of Sound onset to SO peak and R peak onset to SO peak and have now included a comment in the discussion to reflect this on page 20, lines 478-480 as follows:

‘However, it should be noted that although not significant, we observed a trend of lower R peak to SO peak latencies during cardio-audio regularity compared to the other auditory conditions, possibly driven by the fixed relationship between heartbeat and sound in the synch condition.’

Reviewer #2 (Public Review):

This study was designed to study the cortical response to violations in auditory temporal sequences during wakefulness and sleep. To this end, the study had three levels of temporal sequence, a regular temporal sequence, an auditory tone that was yoked to the cardiac signal, and an irregular tone. The authors show significant EEG differences to an omitted tone when the auditory tone was predictable both during wakefulness and sleep.

The authors analyze the ERP to the omitted tone as well as when aligned to the R-peak of the HEP. The analysis was comprehensive and the effects reported align with the interpretation given. Of particular interest was the fact that a deceleration of the heart rate was present for omissions when the auditory tone was yoked to the R-peak (synch) in all stages of wakefulness and sleep.

We thank the reviewer for his/her positive judgment.

However, one weakness was the rationale for the current study and how the results link to current theoretical frameworks for the role of interoception in perception and cognition. This was in contrast to the clear background and explanation to study the response to omissions for a predictable auditory sequence in wakefulness and sleep. It was unclear why the authors selected the cardiac signal to yoke their auditory stimuli. What is the specific motivation for the cardiac signal rather than the respiratory signal? This was not clear.

In the revised Introduction section, we improved our description of these aspects, including the interaction between interoception and external stimulus processing. We hypothesized that cardiac signals would be more relevant than respiratory signals in coordinating temporal expectation because of existing prior experimental evidence thereof, as well as data showing a modulation of the neural response to heartbeat by levels of vigilance/consciousness, and the sharp cardiac R peak offering an ideal candidate for online temporal locking to administered sounds (see our detailed reply to the reviewer’s comment #2 below). However, we cannot exclude that respiratory signals could also be used by the brain to assist temporal regularities detection.

Future studies may test for this possibility.

Author Response

Reviewer #1 (Public Review):

Kozol et al adapt an important tool, in the form of the atlas, to the Astyanax research community. While broadly the atlas appears to correctly identify large brain regions, it is unclear what is the significance of the finer divisions. The external confirmations are restricted to just a few large brain regions (by independent human observer: e.g., optic tectum, hypothalamus. By molecular marker: hypothalamus only.). As such, interpretations of results from as many as 180 small subregions should be interpreted sceptically.

The authors also suggest that some brain regions have increased in size during cavefish evolution (e.g., hypothalamus, subpallium). The analysis of progeny from a genetic cross of cave and surface morphs suggest a complex genetic program has evolved to control this variant set of brain structures. With the development of genetic manipulation tools in this species, an exciting series of experiments may link causal variants with brain development differences.

MAJOR ISSUES

Line 85+. Segmentation accuracy is not well established by the authors. For example, Figure S2 states that the pixel correlation is high between Astyanax populations. But the details of how this cross-correlation was done are sparse. Is the Y- axis here showing the fraction of pixels that are shared in the morphs? While the annotation appears to function similarly across morphs, the 80% machine:human correlation is difficult to put into context. On the one hand, this seems low. For what values should one strive? Are there common "mistakes" or differences in human & machine annotations that lead to certain regions being excluded? A discussion of these is warranted and will be useful to others who wish to use this approach.

Line 87. "such as" is misleading since these were the only two antibodies used to confirm molecular definitions of regions.

But more to the point, additional markers should be used to confirm more than just the ISL+ hypothalamic divisions.

This is particularly warranted, as Fig 1d is not convincing. I believe that the yellow label is ISL; this is difficult to see in the figures. ISL is not ideal since this is widespread in the hypothalamus. There are no ISL-negative regions depicted, which would be necessary to demonstrate that the resolution of this subregion labeling tool is high. A complementary approach would be to find molecular markers that are more restricted than ISL which label only subsets of hypothalamic regions.

Finally, do the mid/hindbrain ISL labeled regions correspond to known ISL+ subregions?

We agree with the reviewer that the Islet1/2 assessment was insufficient for demonstrating automated segmentation accuracy and that the labeling was difficult to visualize in the previous version of the figure. We have addressed this reviewers concern by adding new molecular markers for verification of segment accuracy and through a modified presentation of the original data. The first, and in our opinion most convincing, is the addition of more markers of known neuroanatomical regions. This required not only adding extra antibody stains to our brain atlas, but also optimizing Hybridization Chain Reaction (HCR) in situ protocol that could be coupled with immunohistochemistry, permitting automated segmentation via total ERK registration and brain atlas inverse registration. This novel protocol showed corresponding localization of markers, such as 5-hydroxytryptamine (5-HT), gastrulation brain homeobox 1 (gbx1), and oxytocin (oxt), in the expected neuroanatomical areas. It should be noted that these markers included both large neuroanatomical areas as well as small, well-defined areas such as the superior, and also labeled disparate neuroanatomical loci throughout the brain. We also modified our original figure to better illustrate the regions that islet+ staining labeled. These markers show that islet1/2 labels precise regions of the hypothalamus which correspond to known expression patterns. The updated methodology can be found in lines 422- 440, while the results can be found in lines 105-118 of the text, Figure 1 and Figure 1 – Figure Supplement 1a.

We believe these two changes address the reviewers concerns, and suggest that the neuroanatomical labels generated in this study faithfully label the Astyanax brain.

The molecular and human-observed confirmations of brain regions suggests that the annotated borders of gross anatomical regions are correctly identified by the algorithm. However, data is not presented that indicates whether the smaller regions correspond to biologically meaningful compartments.

We agree with the reviewer that our assessment of regional accuracy for automated segmentation necessitated additional markers, which labeled smaller, more refined compartments. To address this, we developed an HCR in situ hybridization strategy that was compatible with our brain atlas, and used several markers that label smaller regions, such as the 5-HT positive neurons of the dorsal raphe and oxytocin positive neurons of the medial preoptic region. Together, these results were consistent with our previous finding that anatomical regions confirmed by human- observation and molecular staining did faithfully label the correct regions of the brain. These findings can be found in lines 105-118 in the text, along with Figure 1 and Figure 1 – Figure Supplement 1a-d. Together, we hope this shows that not only large neuroanatomical areas, but also finer areas are correctly labeled by CobraZ.

Parameters used in CobraZ to perform the segmentation are not defined. More transparency is required here for others to replicate.

We agree with the reviewer that parameters used for CobraZ and Advanced Normalization Tools (ANT) are necessary for reproducibility of our results. We have since added sentences to clarify that we did not change the original ANTs or CobraZ parameters from Gupta et al. 2018. (line 474- 475) and have added the CobraZ parameter file and ANTs bash scripts to our dryad depository.

Reviewer #3 (Public Review):

In this manuscript the authors use novel techniques and analytical methods on an up and coming animal model for brain evolution. The manuscript utilizes the cavefish Astyanax mexicanus, which can provide future important insights into the field of neurobiology and in evolution in general.

The authors however, only argue that Astyanax is a powerful system for functionally determining basic principles of brain evolution (which clearly it will be), but fail to actually describe what brain evolution insights Astyanax gives. The data is in the paper, but the interpretation needs refinement. This would be a much more valuable paper with a thorough evolutionary context based on the already existing, extensive literature. I believe this manuscript has the potential to be extremely impactful.

We thank the reviewer for her positive critique of our manuscript, and more broadly for the thoughtful comments, the challenge to re-evaluate the way we have thought about our own data, and for hinting us in a direction of scientific direction that is more impactful. We have spent a lot of time re-thinking this work to address this reviewers critique, and believe that it is a far better study for it.

Author Response

Reviewer #1 (Public Review):

The authors of this manuscript aimed to systematically evaluate the pleiotropic effects of MCR-1-mediated colistin resistance. They evaluated the effect of MCR-1 and MCR-3 carried on different plasmids on antimicrobial peptides (AMPs) and assessed their ultimate effect on virulence. The authors find that MCR-1-mediated colistin resistance correlates with increased resistance against some host AMPs, but also increased sensitivity to others. The authors also find that MCR-1 alone is associated with resistance to human serum and to elements of the complement system. This highlights a potential selective advantage for MCR-1-mediated resistance to host immune factors and a potential for enhanced virulence.

The methods have been well established before and adequately support their main findings. While determining the role of MCR-1 in a single genetic background is important to better understand its potential pleiotropic effects against a diversity of AMPs and in a variety of scenarios, the impact and significance of the results are partially ameliorated because different genetic backgrounds, particularly those most relevant to a clinical (or agricultural) context were not considered. The results depicted here are still a necessary and important step towards a more comprehensive understanding of the pleiotropic effects of MCR-1. But, interactions between plasmids and host genomes and their co-evolution can have important effects more generally. The authors do mention this in the discussion and suggest it to be an important avenue for future work. However, given the objective of the study and the clinical and agricultural context in which the authors have framed their work, it seems more relevant to include those distinct genetic backgrounds already here.

The conclusions stemming from the results found in Figure 3, and Figures 4c and d seem too overreaching to me. The associated resistance to AMPs from pigs seems to be only strong enough against one of the five tested AMPs and hence concluding that these impose a strong selective pressure in the pig's gut seems unsubstantiated. Similarly, the difference in survival probability within their in vivo system, though statistically significant, seems to be very ild between their MCR-1 and empty vector control.

Thank you for the comment. We agree on the effect of MCR-MOR on AMP susceptibility and have edited the paragraph by removing the lines on strong selective pressure in the pig gut. As regards the 4c and 4d results (4e and 4f in the revised version), it is interesting and statistically convincing that MCR increases bacterial virulence despite the cost of MCR expression. And importantly, this effect is even stronger in the case of LPS treatment where the immune system is stimulated, expressing diverse host AMPs (PMID: 19897755). This shows MCR-mediated advantages to bacteria in the complex host environment.

Reviewer #2 (Public Review):

Jangir et al test the hypothesis that resistance to the antimicrobial peptide (AMP) colistin can simultaneously increase resistance to other AMPS with related modes of action. Because AMPS comprise part of innate immunity, their central concern is that colistin resistance may compromise host defenses and thereby increase bacterial virulence. Their results show that MCR-1, whether expressed from naturally circulating or synthetic plasmids, can increase the MIC to AMPS from humans, pigs, and chickens, and impart fitness benefits at sub-MIC concentrations. In addition, they find that MCR-1-containing strains have increased survival in human plasma and are more lethal in an insect infection model.

The conclusions of the paper are generally well supported by the results, but some aspects could be clearer and better defended with a few small additional experiments.

Strengths:

Using both synthetic and natural plasmids makes it possible to cleanly separate the effects of MCR-1 from the effects of other plasmid-borne genes or plasmid copy numbers. This helps confirm the causal role of MCR-1 on altered AMP susceptibility.

Testing the survival of transformed isolates in human serum and in insects points to relevance in the more immunologically complex host environment where cells are exposed to a suite of factors that reduce bacterial survival.

Thank you!

Weaknesses/suggestions:

Although increases in MIC are evident for different AMPS, the effects are generally modest. To address this, it might be helpful to use pairwise competition assays, as in Figure 1, to establish that even small changes to MIC are associated with clear selective benefits.

Thank you for the suggestion. We agree that in some cases the change in MIC is modest, however, we would like to highlight that small-level changes in resistance have important clinical implications. For example, resistance mutations conferring a small change in MIC can ensure the survival of pathogenic bacteria in antibiotic-treated hosts (PMID: 30131514). Additionally, a comparison between competition assays (Fig 1) and MICs (Fig 2) clearly shows that small changes in MIC are associated with substantial fitness benefits. For example, for pSEVA:MCR-1, the fold change in MIC of CATH2 (chicken), PMAP23 (pig), and LL37 (human) ranges between 1.05 and 1.5, however, the competitive fitness ranges from 10% to 17%. This issue is discussed in the revised manuscript (lines 306-317, page 13)

….This would be especially helpful in assays with human serum and in Galleria where the concentrations of AMPS or other immune components are unknown.

It is clear that MCR-1 increases resistance to serum and virulence (Figure 4). However, we agree with the reviewer that the selective benefits of MCR-1 in complex host environments are not known (i.e., serum or Galleria). We have revised the final paragraph of the discussion to reflect this limitation of our study (lines 370-382, page 15).

Assays using human serum are interesting but challenging to interpret given the diverse causes of bacterial killing, including complement. Although this was partly addressed in Supplementary Figure 6, I found the predictions of these experiments unclear. First, I think these experiments are too central to be relegated to the supplemental materials; they belong in the main text. Secondly, it is important to explicitly spell out the expectations of using heat-killed serum (which will degrade any heat-labile components) or complement-deficient serum. It should be clearer under which conditions MCR-1-containing strains are predicted to do better or worse than controls.

We have addressed this in the revised version. We have moved Supplementary Fig 6 to the main text, and have edited the text, clarifying the model prediction (lines 245-257, page 10).

Galleria is a useful infection model for virulence, but it is unclear what drives differences between strains. First, bacterial numbers aren't measured in this assay, so it isn't known if increased virulence is due to increased bacterial growth or decreased bacterial clearance. As above, I think these assays would be stronger using the competition-based approach in Figure 1. This would indicate bacterial numbers through time and directly show the selective benefit associated with MCR-1. Second, it would be useful to elaborate on why MCR-1 increases virulence, especially any known similarities between Galleria AMPS and those tested in Figures 1 and 2. Overall, it would help if Galleria were less of a black box.

We agree that the mechanism underlying increased virulence remains to be explored and thus, we have already discussed this in the discussion as a limitation (lines, 370-382, page 15). However, elucidating the mechanisms by which MCR-1 increases virulence would clearly be an interesting line of research moving forward.

Author Response

Reviewer #1 (Public Review):

The adhesion of Leishmania promastigotes to the stomodeal valve in the anterior region of the sandfly vector midgut is thought to be important to facilitate the transmission of the parasites by bite. The promastigote form found in attachment is termed a 'haptomonad', although its adhesion mechanism and role in facilitating transmission have not been well studied. Using 3D EM techniques, the paper provides detailed new information pertaining to the adhesion mechanism. Electron tomography was especially useful to reveal the ultrastructure of the attachment plaque and the extensive remodelling of the flagellum that occurs. A few of the attached haptomonads were found to be in division, which is a novel observation. The attachment of cultured promastigotes to plastic and glass surfaces in vitro was found to involve a similar remodeling of the flagellum and was exploited to image the sequential steps in attachment, flagellar remodeling, and haptomonad differentiation. The in vitro attachment was found to be calcium2+ dependent. Based mainly on the in vitro observations, a sound model of the haptomonad attachment plaque and differentiation process is provided.

We thank the reviewer for highlighting the significant progress we have made in dissecting the adhesion mechanism and flagellum restructuring in the Leishmania haptomonad.

Reviewer #2 (Public Review):

The study by Yanase et al. investigated the details of the 3D architecture of Leishmania haptomonad promastigote's adhesion to the midgut of the insect vector. The authors generated a dataset of images that reveal intricate details of the formed adhesion plaque and expanded the study with in vitro alternatives for the exploration of how Leishmania promastigotes strong adhesion by hemidesmosomes to surfaces can happen and be maintained. They show with unprecedented detail the ultrastructure of the attachment plaque. The in vitro dataset of the paper adds to the specific literature important details on how to explore micro/nanostructures involved in an important attachment step for this eukaryotic parasite. However, the in vitro data should be reconsidered in its discussion and conclusions as it does not support direct comparison with in vivo Leishmania forms as pictured by the authors. In general, the dataset presented in this manuscript adds valuable data and resources for the study of Leishmania promastigotes to surfaces, especially to the thoracic midgut parts of its insect vector.

The dataset of this paper is well-collected and robust, but some aspects of image analysis need to be clarified and extended. Also, the in vitro data from the manuscript will benefit from an extensive adjustment in its discussion. Points to focus on:

We thank the reviewer for recognising the ultrastructural detail we have now provided of this cryptic parasite life cycle stage. Below we address each of your points in detail.

1) The haptomonad promastigote is indeed a possible critical form for transmission, but it lacks formal demonstration still in all literature available. This should not be claimed without proper formal demonstration.

We agree with the reviewer that any relationship between transmission and the haptomonad form has yet to be formally demonstrated. Hence, we revised the descriptions referring to the relationship between transmission and the haptomonad form (Line 22-23, 31 and 113-114).

2) Literature available and cited in this manuscript regarding in vitro adhesion of culture Leishmania promastigotes does not provide direct evidence for haptomonad differentiation. Haptomonads are still a largely unknown promastigote form with no defined ontogeny. With that, to propose an in vitro haptomonad differentiation protocol, more detailed direct evidence of in vivo haptomonads will be necessary. The in vitro experiments available show how cultured promastigotes attach to surfaces. Detailed studies in vivo will be needed still to attribute the findings in vitro to haptomonads.

We would like to highlight that promastigotes and haptomonads have morphological definitions within the literature and our cells are definitely more like haptomonads than promastigotes. As the reviewer highlights, the haptomonad-like cells we generate in vitro have an almost identical morphology and attachment plaque structure to those haptomonads we observed attached to the stomodeal valve. In addition, we have been able to watch individual cells that had a promastigote morphology acquire a haptomonad morphology and we believe this will provide future insights to the ontogeny of these forms. However, as there are currently no published molecular markers for haptomonads we have not been able to provide direct evidence other than the morphology and ultrastructure that in vitro attachment replicates in vivo haptomonad differentiation. Therefore, we have revised our nomenclature and now refer to the in vitro haptomonad-like cell. In the discussion, we have been careful to highlight that certain aspects of our model rely on in vitro data and therefore may not accurately reflect the situation in the sand fly.

3) This manuscript will benefit by having a detailed description of how to analyze and get to the 3D models presented. This has a strong potential for usage beyond the Leishmania/sand fly field. Statistics should be made available with ease across the manuscript and with a dedicated section on methods.

We added a detailed description of how to analyse the 3D models (Line 756-763), and added videos showing a rotated view of each 3D model (Figure 1—video 3 and 4, Figure 2—video 2, and Figure 3—video 2 and 4). We have deposited the SBF-SEM and tomography data on the Electron Microscopy Public Image Archive (EMPIAR; https://www.ebi.ac.uk/empiar/), enabling access to the raw data (Line 763-766). We have added a statistics section into the Materials and Methods (Line 864-868).

Author Response

Reviewer #1 (Public Review):

In this manuscript, Sampaio et al. tackle the role of fluid flow during left-right axis symmetry breaking. The left-right axis is broken in the left-right organiser (LRO) where cilia motility generates a directional flow that permit to dictate the left from the right embryonic side. By manipulating the fluid moved by cilia in zebrafish, the authors conclude that key symmetry breaking event occurs within 1 hour through a mechanosensory process.

Overall, while the study undeniably represents a huge amount of work, the conclusions are not sufficiently backed up by the experiments. Furthermore, the results provided present a limited advance to the field: the transient activity of the LRO is well established, and narrowing down this activity to 1 hour (even though unclear from the presented data that it is a valid conclusion) does not help to understand better the mechanism of symmetry breaking.

We thank the reviewer1 for acknowledging the hard experimental set up. However, we must argue that knowing the exact timing that is more sensitive to fluid flow manipulations is a very important advance we provide here. The reason is because this type of experiment is giving us the physiological timing in a WT embryo. It is one thing to know the system can respond to optical tweezers earlier than 5 ss and later than 5 ss, as Yuan lab did recently, but quite another to constrain the physiological timing at which the process occurs in an unperturbed manner (as much as possible). Our aim was the latter. Our rationale is that knowing the physiological time is important to provide clues, for example we had these types of questions at the time: is the physiological time before or after cell rearrangements occur? is it falling in a directional or non-directional flow regime? Is it governed by a mild flow or stronger one? Is it before or after dand5 becomes asymmetric? Some of these questions that we think we all know the answers for, could be challenged by our experiments… so it is indeed very important to not assume we know the answer, and ask the question again in an unbiased way with every new technique available! We wanted to be unbiased, and we think that is the beauty of our time-window experiment. Indeed, it shows the physiological time-window peaks at 5 ss which is later than Yuan’s lab calcium transient recording and before dand5 asymmetric expression. In our opinion this is compatible and makes perfect sense because although the system already shows calcium transients before and can respond to lack of Pkd2 or optical tweezer cilia manipulations at 1 ss – 3 ss, it is from 4 to 6 ss, peaking at 5 ss, that it is most responsive physiologically to the fluid extraction and therefore both mechanical and chemical perturbations.

We have made additional experiments and used smFISH on WT embryos for detecting dand5 expression with cellular resolution, and we have quantified asymmetries in dand5 number of transcripts as early as 6 ss (new Figure 7 and new author: Catarina Bota) that further support our time-window claim. Degradation of dand5 mRNA has been the mechanism suggested to be at the base of the asymmetric dand5 expression, which is usually a very fast mechanism. This new piece of evidence supports that the physiological breaking of symmetry is stronger around 5 ss. (see new discussion on this subject on page 27).

Regarding the symmetry breaking. The fact that anterior angular velocity was the major difference between embryos that recovered without LR defects versus those that did not, reveals that angular velocity must be tightly regulated by cilia motility and CFTR activity to bring back fluid and flow directionality, which together confer the robustness of flow. This is now better explained in the manuscript. We agree that the novelty regarding angular velocity may seem incremental compared to our work from 2014, where we only analyzed speed (Sampaio et al, 2014). However, here we provided more resolution and detailed parameters of angular velocity per sections of the LRO as well as tangential and radial velocities, the components of angular velocity. The Radial component shows a trend towards left anterior that is now discussed in the text as evidence for a left difference. The present work shows that anterior angular velocity has a major role in the successful recovery of the symmetry breaking process, which was not claimed before. Here we challenged the embryo to bring to light the most important parameters.

Importantly, the authors do not provide any convincing experiments to back up the mechanosensory hypothesis because the fluid extraction experiments affect both the chemical and physical features of the LRO, so it is impossible to disentangle the two with this approach.

We agree the first extraction experiment (Figures 1-3 and Table 1) affects both mechanisms and does not disentangle them, and that was, in fact, our goal for the first experiment - the finding of the exact time-window for symmetry breaking. However, in the second part of the work (Figures 4-5 and Table 2) we provide a 20,000 times dilution experiment, this dilution experiment is very different than the extraction one. We apologize if this was not clear and hope to have made it clear this time.

We must agree with the reviewer that chemosensing is not excluded, in fact we had provided a paragraph in the discussion about EV secretion rates to tone down our claim and did acknowledge that secretion could still overcome the dilution we are causing. We think we had already addressed this problem in the previous eLife manuscript but now we have discussed the possibilities and the experimental evidence that supports each of them (see page 28, last paragraph). The key experiment that does not fit with secretion is pointed out in the end, and we ask the reviewer to read it in the context of wildtype animals. We agree both scenarios must be discussed and leave space for future data on mmp21 and CIROP. However, so far, in zebrafish we cannot favor chemosensing as much as mechanosensing, we can only wait for more discoveries and be open.

Author Response

Reviewer #1 (Public Review):

In this manuscript, Lee and colleagues address the participation of NBR1 in chloroplast clearance after treatment with high light intensity. Authors use NBR1 fused to reporter proteins (GFP, mCherry), with the aid of nbr1, atg7, and nbr1-atg7 mutants, in combination with immunogold labelling to show localization of NBR1 to surface and interior of photodamaged chloroplasts, which follows with their engulfment in the vacuole, a process which is independent of ATG7. The combined use of ATG8 fused to GFP further shows that NBR1 and ATG8 are recruited independently to photodamaged chloroplasts. In addition, the use of mutant versions of NBR1 in combination with mutants lacking E3 ligases PUB4 and SP1 and mutant toc132-2 and tic40-4 lacking members of the TIC-TOC complex of protein translocation to the chloroplast, authors show that chloroplast localization of NBR1 requires the ubiquitin ligase domain (UBA2) of the protein, whereas, the PB1 domain exerts a negative effect on NBR1 chloroplast association, yet neither the PUB4 and SP1 E3 ligases nor the TOC-TIC are required for NBR1 association to photodamaged chloroplasts. All these approaches are well described and strongly support the authors' conclusions that the loss of chloroplast envelope integrity allows the entrance of cytosolic ubiquitin ligases and the participation of NBR1 in photodamaged chloroplast clearance by a process of microautophagy. All these findings add valuable information to our knowledge of chloroplast homeostasis in response to light stress.

To further support these conclusions, authors perform a chloroplast proteomic analysis of the WT, nbr1, atg7, and nbr1-atg7 mutants. However, in contrast with the above results, the description of the proteomic data is rather confusing. The paragraph on Page 17 (lines 393-406) is hard to follow. The term "over-representation of less abundant chloroplast protein" is also quite confusing, like the data in Fig. 6 and supplementary to this figure (what does show the PCA analysis in Fig. 6-suppl. 1?). I wonder whether it would be possible to show all these data as supplementary and try to present the data supporting the major conclusion of these analyses (if I understood correctly, that nbr1, atg7, and the double mutant have lower contents of chloroplast proteins), in a more simple and clear format.

Following the reviewer’s comments, we have re-written the result section describing the proteomic data to make it more concise and clearer. We have also made modified Figure 6 to make it more concise and generated new graphs for Figure 6 supplemental figures 1 and 2.

Reviewer #2 (Public Review):

The authors conducted a wide-ranging series of experiments which lead to the conclusion that NBR1 is involved in the clearance of photodamaged chloroplasts. It is a novel finding because the role of NBR1 in this process was never documented. Notably, the NBR1-mediated clearance is only one of the several possible mechanisms responsible for chloroplast turnover. It is not surprising, considering that the nbr1 mutants are viable. The work is arranged very well. The rationale of the subsequent experiments is logically justified and the outcomes and followed by clear conclusions. In consequence, the authors managed not only to observe the association of NBR1 with the chloroplasts but they threw some light on the corresponding mechanisms. The manuscript contains numerous high-quality images from a confocal microscope and from a transmission electron microscope. All images are accompanied by statistical analysis of the respective microscopic observations, which greatly improves the credibility of the conclusions. Shortly, the authors demonstrated that NBR1 decorates not only the exterior but also the interior of damaged chloroplasts in an ATG7-independent way. Next, they establish that NBR1 and ATG8 are recruited to different populations of damaged chloroplasts, and they document differences in chloroplasts turnover, differences in chlorophyll abundance and chlorophyll photochemical properties, as well as differences in the total proteome of the nbr1 mutant in comparison to the wild type and atg7 mutant in two light regimes (low light and high light). Finally, they exclude the requirement for the known E3 ligases PUB4 and SP1 for NBR1mediated degradation and show that the NBR1 internalization relies rather on the chloroplastic membrane rupture than on the TIC-TOC-dependent processes. In summary, the authors postulate that NBR1-mediated chloroplast clearance is a novel, not yet described mechanism and summarize it in a clear diagram.

The work is interesting, the figures are convincing and the conclusions are justified by the results. It provides novel data on the function of selective autophagy receptors NBR1 in plant cells, however, it also leaves the reader with some unanswered questions. The most important is the relative contribution of each of the chloroplast's degradation routes to the turnover of these organelles in different stresses, light regimes, plant growth stages, etc. This is a difficult problem because the mutations in relevant genes have pleiotropic effects and it is difficult to separate the functions of the individual turnover routes. For example, the defects in core autophagy genes (like the atg7 mutant used in this study) result in an increased level of NBR1. These issues are not sufficiently addressed in the discussion.

The reviewer is correct and indeed, we also detected higher levels of NBR1 in the atg7 mutant (Fig 2G). This could be, for example, the underlying reason why there are more chloroplasts decorated with NBR1 in that atg7 mutants than in complemented nbr1 plants, 24h after high light treatment (Fig 1F). However, the higher frequency of photodamaged chloroplasts observed in atg7 (Fig 2D), supports a different scenario: the higher number of photodamaged chloroplasts that are not successfully repaired or degraded by canonical autophagy in atg7, become substrates of NBR1. The increased levels of NBR1 in the agt7 mutant and how this could influence the effects seen in the mutants studied in this manuscript is now discussed in lines 670-673.

Reviewer #3 (Public Review):

The authors use an impressive array of techniques to determine the role of the NBR1 autophagy receptor protein specifically in the clearing of photodamaged chloroplasts. The authors describe the mechanism(s) by which this receptor operates in this context and demonstrate that this NBR1-mediated process occurs independently of SP1 and PUB4 (whose own roles in other aspects of chloroplast autophagy have previously been shown). The authors further dissect the functional domains of NBR1 to identify which are important in this process.

The major strength of this work is the myriad techniques used to approach the problem. The data are of high quality, and on the whole, well replicated and statistically analysed. In the main, these data substantiate the findings of the authors, although some findings are quite correlative/descriptive. However, the authors show good circumspection in their conclusions and discussion. One potential weakness is that the genetic data (use of mutants) rely on single mutant alleles, therefore whilst genetic linkage to the mutations is assumed, it cannot strictly be guaranteed. The authors performed effective genetic complementation to analyse the domain structure of NBR1 shown in Figure 7. It would have been good if complementation of nbr1 and atg1 mutants and/or alternative mutant alleles had been used for experiments described in Figures 1 to 6. Without this, I think even more circumspection regarding the data obtained from these single-allele mutants would be advised.

We agree with the reviewer that more mutant alleles would have provided stronger support to our conclusions, but we would also like to highlight that the atg7-2 (Chung et al 2010), nbr1-2, and atg7-2 nbr1-2 mutants (Jung et al 2020) have been well characterized previously and the nbr1-2 mutant, shown to be rescued by the expression of fluorescently tagged NBR1 (Jung et al 2020). We are confident about the results on the localization of NBR1 in chloroplasts, not only because the fluorescently tagged NBR1 proteins are functional but also because we were able to corroborate the localization of NBR1 by using antibodies against the native proteins (Fig 2). That said, the reviewer does raise an important point and therefore, we have acknowledged more explicitly the limitation of our conclusions based on the analysis of single mutant alleles in lines 630-631 of the discussion.

Author Response

Reviewer #1 (Public Review):

The model put forward by the authors in this manuscript is a simple and exciting one, explaining the function of AGS3 as a negative regulator of LGN, acting as a 'dominant-negative' version of LGN. Overall, the results support the model very well, and the results shown in Fig 6, which clearly reveal the functional relevance of AGS3, add strength to the paper.

We thank the reviewer for their enthusiasm regarding our finding that AGS3 acts as an endogenous dominant-negative to inhibit LGN. We appreciate their assertion that the results support the model and that the functional relevance to epidermal stratification is a strength.

In Figures 3A and B, the authors claim that AGS3 overexpression leads to depolarization of LGN in epidermal stem cells. However, in the example provided in Figure 3A, the LGN signal appears to be stronger than the control, with more LGN still on the apical side (many would categorize this as 'apically polarized'). In the scoring shown in Figure 3B, I am not sure if 'eyeballing' is the right way to decide whether it is polarized/depolarized/absent. The authors should come up with a bit more quantitative method to quantify the localization/amount of LGN and explain the method well in the manuscript. A similar concern regarding the determination of the LGN localization pattern applies to the rest of figure 3 as well.

We agree with this important critique about the methodology used to assess LGN expression patterns. While we have historically included categorical analyses like those used in Fig. 3A,B in past publications (Williams et al, NCB 2014; Lough et al eLife, 2019), we have also now performed additional, unbiased, quantitative measures of LGN fluorescent intensity, as described in greater detail above. We added these new data in Fig. 4C-J, while the data previously in Fig. 3A,B have now been redistributed between Fig. 3E,F (overexpression) and Fig. 4A,B (knockdown).

Reviewer #2 (Public Review):

To date, only a handful of studies have addressed the importance of AGS3, a paralog of the relatively well-characterized spindle orientation factor LGN. The authors now show that AGS3 acts as a negative regulator of LGN and propose that this activity could work through competition for binding partner(s). Remarkably, regulation is temporally restricted in such a way that the conserved role played by LGN in metaphase spindle orientation is unaffected. Instead, AGS3 regulates a post-metaphase function for LGN, namely Telophase Correction. The article is well-written, the experiments are performed at a high level, and the claims are generally supported by the data. Two main points of confusion are raised in the current version. 1) The authors show that AGS3 regulates cortical localization of LGN, but would need to clarify how LGN is being affected. 2) The authors propose in the discussion that AGS3 might exert its regulatory effect through competition for NuMA, an important binding partner for LGN, but would need to clarify how and why NuMA would be involved in Telophase Correction.

We thank the reviewer for appreciating the novelty of our findings regarding the understudied LGN/pins paralog AGS3. In regards to the first point, as described earlier, we have added additional quantitative analyses of how AGS3 affects cortical LGN fluorescent intensity in Fig. 4C-J. We now show that AGS3 loss leads to broader and higher expression levels throughout mitosis, and therefore we have amended our model to soften the claim that AGS3 primarily operates during telophase correction. This renders the second point somewhat moot, but we nonetheless have expanded our Discussion to note that NuMA can be cortically recruited to the anaphase cortex independent of LGN (lines 531-542). We also contextualize our findings with the Reviewer’s own recent study which proposes a “threshold model” of cortical Insc as a determinant of spindle orientation (Neville et al, 2023), and speculate that a similar model could apply in our system, perhaps with AGS3 binding and sequesting Insc rather than NuMA (lines 543-556).

Reviewer #3 (Public Review):

This paper examines the mechanisms that control division orientation in the basal layers of the epidermis. Previous work established LGN as a key promoter of divisions where one of the siblings populates the differentiated layers (perpendicular). This work addresses two important, related issues - the mechanisms that determine whether a particular division is planar vs perpendicular, and the function of AGS3, and LGN paralog that has been enigmatic. A central finding is that AGS3 is required for the normal distribution of planar and perpendicular divisions (roughly equal) such that in its absence the distribution is skewed towards the perpendicular. Interestingly, however, the authors find that AGS3 has no detectable effect on orientation if the orientation is measured at anaphase. This timing aspect builds upon previous work from this group demonstrating a phenomenon they term "telophase correction" in which the orientation changes at the latest phases of division (and possibly post division?). Thus AGS3 seems to exert its effect using these later mechanisms and this is supported by further analysis by the authors. Importantly, the authors show that AGS3 acts through LGN, based on localization data and an epistasis analysis. The function of AGS3 has been highly enigmatic so resolving this issue while providing a useful step towards understanding how the division orientation decision is made, makes for exciting progress towards an important problem. I found the overall narrative and presentation to be quite good and especially appreciated the thoughtful discussion section that did an excellent job of putting the results in context and speculating how unknown aspects of the mechanism might work based on current clues. With that said, I think there are some important issues that should be resolved.

We thank the Reviewer for this excellent summary of our findings and appreciation of the significance of the issues that our study addresses.

Regarding the orientation measurements, the authors should specify how the midbody marker was used to mark sibling cells, especially given the midbody can move following division. For example, how can the authors be confident that the siblings in the middle panel of 1A are correct and not an adjacent cell? Regarding quantification, it would be useful for the authors to comment on how the following would influence their measurements: 1) movements along the z-axis, and 2) movement of the nucleus within the cell

We have used this methodology for over a decade, and while it is not flawless, we have included several safeguards to ensure that sibling cells are correctly identified. We have added additional details to the Methods section (lines 867-869, 873-879).

A similar question is how much telophase correction really happens in telophase. How confident are the authors that the process actually occurs during division and not subsequent to it? What is drawn in their previous paper and in Figure 7A implies that post-division movements may be important. It would be useful for the authors to comment on whether they can make the distinction and whether or not it might be important.

Our intent in coining the term “telophase correction” was to imply that this process initiates, rather than completes, during telophase. We apologize for this confusion and have clarified this in the text (lines 80-82). Since most mammalian cells complete M phase in ~1h, with the longest time spent in prophase, in the absence of direct evidence to the contrary, it may be prudent to assume that telophase, like metaphase and anaphase, is relatively short, on the order of minutes. Since we cannot directly observe reformation of the nuclear membrane in our movies, we cannot be sure when telophase ends. Likewise, we do not currently have a suitable marker of the spindle midbody for live-imaging, so cannot be sure when cytokinesis completes. That said, we feel confident that most of the reorientation is occurring prior to cytokinesis, because we have previously reported that the greatest changes in daughter cell positioning occur within the first 10-15 minutes of anaphase onset, when a gap in membrane-GFP/TdTomato is still visible (Lough et al, eLife, 2019). However, while we feel that there are many interesting questions that our work raises about the timing or reorientation relative to specific mitotic stages—e.g. is the midbody asymmetrically positioned, inherited, or ejected?—these questions are beyond the scope of the present study.

Does the division angle in the AGS3 OE experiment (Figure 1D) correlate with AGS3 levels within the cell?

This is an interesting question, and indeed, we our hypothesis would predict that it would. However, it is not straightforward to quantify AGS3 or mRFP1 levels, and as we explain in a new section of the Results (lines 212-237), we have some concerns that N-terminally tagged AGS3 may not be fully functional. We have added new data with C-terminally tagged AGS3-mKate2, which we feel provides even stronger evidence that mKate2+ cells show a planar shift compared to mKate2- cells (Fig. 3C,D). In the future, we could test this hypothesis at the population level by comparing division orientation profiles for AGS3-mKate2+ cells carrying either a non-targeting scramble or Gpsm11147 shRNA. We would predict that knocking down endogenous AGS3 while overexpressing AGS3-mKate2 should give an intermediate phenotype.

I found the localization data to be the weakest part of the paper and feel that some reconsideration and reanalysis are warranted. First, the quantifications in Figures 2C, 3B, and 3F are unnecessarily vague scoring-based metrics. In 2C, "Localization pattern" should be replaced with membrane/cytoplasm ratio or an equivalent quantification. In 3B "LGN localization" should be replaced with apical/cytoplasmic and apical/basal ratios or equivalents. In 3F, "Polarized LGN frequency" should be replaced with apical/basal ratio or equivalent. It seems to me that non-AI processed data would be most appropriate for these quantifications unless such processing can be justified.

This issue was raised by the previous two Reviewers and has been addressed by new data added to Figure 4.

Second, it is important to note that the cytoplasmic localization of AGS3 does not allow one to conclude that AGS3 is not on the membrane. Unfortunately, high cytoplasmic signal can preclude the determination of membrane-bound signal.

We agree with the Reviewer and have softened our language throughout the text.

Finally, I had difficulty reconciling the images of LGN shown in Figure 3 with the conclusions made by the authors.

We have added additional, representative images of LGN expression in control and AGS3 KD cells in Figure 4C-E.

The challenge of the localization data is troubling because an important conclusion of the paper is that AGS3 acts via LGN. The localization data provided one leg of support for this conclusion and the other is provided by an epistasis analysis. Unfortunately, this data seems to be right on the edge because it is based on the difference between the solid and dashed blue lines in Figure 5B not being significant. However, we can see how close this is by comparing the solid and dashed red lines in the adjacent 5C, which are significantly different. Between the localization data, which doesn't seem clear cut, and the epistasis experiment, which is on the razor's edge, I'm concerned that the conclusion that AGS3 acts through LGN may be going beyond what the data allows.

We appreciate the Reviewer’s comments about the importance of these two lines of experimentation: 1) AGS3’s effect on LGN localization, and 2) epistasis experiments between AGS3/Gpsm1 and LGN/Gpsm2. We feel we have significantly strengthened this first pillar with the additional data presented in Fig. 4C-J. Regarding the second point, we would like to emphasize that we present three lines of evidence for the existence of an epistatic relationship between LGN and AGS3: 1) the static division orientation data comparing LGN single KOs to both LGN KO + AGS3 KD and AGS3+LGN dKOs (Fig. 6B); 2) live imaging division orientation/telophase correction comparing LGN KOs to AGS3+LGN dKOs (Fig. 6C-E); 3) lineage tracing data comparing LGN KOs to AGS3+LGN dKOs (Fig. 7H,I). Further, we think the reviewer may have misconstrued the data presented in Fig. 5C (now Fig. 6C). The dashed lines indicate orientation at anaphase and solid lines 1h after anaphase, so the shift between dashed and solid lines indicates telophase correction, which occurs to similar (and statiscially significant) degrees in both LGN single mutants and AGS3+LGN dKOs. Comparisons between the single and double mutant would be between red and magenta solid lines or red and magenta dashed lines, and neither of these are statistically significant. We realize that our use of dashed lines in Fig. 5B (now Fig. 6B), which we normally only use to refer to anaphase entry in live imaging data, may have caused this confusion. Therefore, we have changed all plots to solid lines¬ in Fig. 6B, and use light and dark magenta, respectively, to differentiate between LGN KO + AGS3 KD and AGS3+LGN dKOs.

Author Response

Reviewer #3 (Public Review):

The authors took a comprehensive set of analyses to examine the relationship between pupil diameter / derivative and BOLD-signal during rest in the ascending arousal system nuclei in 72 young participants. Focus is on the locus coeruleus, ventral tegmental area, substantia nigra, dorsal and median raphe nuclei and the basal forebrain. Analyses were performed using various processing pipelines: canonical versus custom hemodynamic response functions, with/without smoothing, time to peak analyses and cross spectral power density analyses to define the time lag between both measurements. The authors could not replicate previous correlations between locus coeruleus BOLD and pupil measurements using standard analytic approaches, and also found no relationship between locus coeruleus BOLD and pupil measurements when using custom hemodynamic response functions. When using time to peak and cross-correlation analyses, the authors found that coupling between pupil size and AAS BOLD patterns increases with decreasing time to peak, when the two signals were close in time. The authors conclude that these findings suggest that pupil size could be used as a noninvasive readout of AAS activity under passive conditions.

These authors did a thorough assessment, and described the methods and results well and in a balanced manner.

Outstanding questions:

• the reliability of these observations? would we see the same findings in a different cohort or using a different sequence/field strength?

• What is the independent association of each assessed nucleus with pupil dilation? That could be informative to understand their shared or unique role.

We are grateful to the reviewer for their expert advice in helping us strengthen our manuscript. We agree with the reviewer that these two outstanding questions are important and we have done our best to answer these questions below. We believe that our manuscript has greatly improved, thanks to the reviewer’s suggestions for running these additional analyses.

Author Response

Reviewer #2 (Public Review):

The availability of large collections of Mycobacterium tuberculosis (Mtb) isolates has enabled many important studies looking to identify mycobacterial genetic polymorphisms associated with anti-tuberculosis (TB) drug resistance, including both classical "resistance-conferring" mutations and novel "resistance-enabling" mutations. Importantly, these studies have expanded our understanding of mycobacterial genetic adaptations undermining chemotherapy, in many cases allowing for improved diagnostic tests and predictions of treatment failure. In this submission, Gao and colleagues adopt a different approach to the problem: although also applying a GWAS-type analysis, they instead attempt to elucidate polymorphisms implicated in poor outcomes of TB patients undergoing treatment for the drug-susceptible disease. Starting with a large dataset comprising 3496 samples with corresponding clinical (host) metadata, the authors generate Mtb whole-genome sequence data for 91 samples obtained from patients with "poor" outcomes and 3105 patients with "good" outcomes. These are used to identify 14 fixed and >230 unfixed mutations that might be associated with "poor" treatment outcomes, a conclusion which they argue is plausible given transcriptional evidence implicating many of the identified genes in the mycobacterial response in vitro to first-line drug exposure and/or hypoxia, both of which are considered relevant to clinical disease. Notably, they also identify a tendency for a greater proportion of "ROS mutational signatures" in unfixed mutations from "poor" outcome samples. Finally, incorporating these observations in a prediction model, the authors observe that the mycobacterial factors aren't adequate on their own but, when combined with key host factors - including patient age, sex, and duration of diagnostic delay (which have stronger predictive value) - they enhance predictive capacity. In summary, this paper reports a novel approach yielding observations that offer tantalizing insight into the mycobacterial factors which might influence TB treatment outcomes independent of drug resistance, however, the following must be considered:

(i) The manuscript provides little to no detail about how the samples were obtained, other than the fact that they comprise "pre-treatment" samples: are they all sputum samples? Were they induced? Similarly, no information is provided about sample propagation: were the samples cultured to achieve sufficient biomass for whole-genome sequencing? If so, in what growth media, for how long, and how many passages? Were all samples treated identically? And were they plated to single colonies - or are the "isolates" referred to throughout the manuscript actually heterogenous populations of potentially different Mtb clones obtained - and propagated - as a mixed sample? This information is critical given the potential that the identified polymorphisms - both fixed and (perhaps even more so) unfixed - might have arisen as a consequence of in vitro (laboratory) manipulation under standard aerobic conditions.

Thanks for your encouraging comments. The requested information about sample propagation has been added to the methods section in the new version. For details, please see our response, above, to the essential revisions (Q1).

(ii) A key question that arises from this study (and others like it) is whether causation has been adequately established. Ideally, the Mtb genotypes contained within samples obtained pre-treatment should be compared with samples obtained from the same patients following treatment - that is, when the "poor" outcome was manifest. The expectation is that the polymorphisms identified prior to initiation of therapy - especially the 14 fixed mutations - should be evident (even dominant) at the later stage when therapy failed (or at the subsequent presentation in cases of relapse). Recognizing that this is not easily accomplished, though, it seems fair to suggest that the perceived relevance of the identified mutations would be strengthened if the authors were able to provide any other evidence - perhaps from studies of drug-resistant Mtb isolates - supporting their inferred role in undermining frontline treatment.

Thank you for these insightful questions. We sequenced the isolates obtained at the time of relapse for all 47 relapse cases and found that the 14 GWAS-identified fixed mutations were only detected in relapse isolates from the 13 patients whose first samples also contained the GWAS-identified mutations. None of the 14 mutations we identified were found in isolates from the other relapsed patients. We also searched for the presence or absence of theses 14 mutations in published studies seeking noncanonical mutations associated with drug-resistant Mtb isolates [5-7]. None of the 14 mutations we identified were reported in any of these studies, but two of the genes (ctpB & metA) in which our mutations were found had been previously identified as potentially associated with first-line drug resistance.

(iii) Related to the above, the authors make the valid point that their intention here was different from other studies which have deliberately utilized drug-resistant Mtb isolates to identify resistance-conferring and resistance-enabling mutations (such as in the study they cite by Hicks et al). It would be interesting to know, however, if any of the mutations identified in those other studies were also picked up in this work - and, if not, why that might be the case.

As mentioned in our response to the previous question, none of our mutations were mentioned in prior studies. Our inference is that the 14 fixed mutations we identified had only limited effects on outcomes, which would explain why: they were not identified in previous studies; isolates from only 24.2% (22/91) of patients carried any of these 14 mutations; and none of the mutations were shared amongst all 22 patients.

(iv) Finally, the analyses presented in this study are heavily dependent on the use of appropriate statistical methods to identify potentially rare genetic polymorphisms. However, as noted for sample processing (see my earlier comment above), there is very little detail provided about the methodology applied. This omission detracts from the interpretation, especially given that the predominance of lineage 2 (which contributes >75% of the isolates, with sublineage 2.3 constituting >50%) risks a lineage-specific association, rather than a more generalizable pathogenicity phenotype. Similarly, the heavy skew in the numbers of "good" (3105 samples) versus "poor" (91 samples) collections (approximately 34x difference in sample size) raises the possibility that mutations identified in the "poor" category might be artificially over-represented. More clarity in detailing the statistical methods is required to allay any concerns about the identification of candidate polymorphisms.

Thank you for pointing this out. We have added details of our statistical methods to the methods section, and in the results section we have indicated the specific statistical methods used and the meaning of the statistical metrics.

Author Response

Reviewer #1 (Public Review):

Lammer et al. examined the effects of social loneliness, and longitudinal change in social loneliness, on cognitive and brain aging. In a large sample longitudinal dataset, the authors found that both baseline loneliness and an increase in loneliness at follow-up were significantly associated with smaller hippocampal volume, reduced cortical thickness, and worse cognition in healthy older adults. In addition, those older adults with high loneliness at baseline showed even smaller hippocampal volume at follow-up. These results are interesting in identifying the importance of social support to cognitive and brain health in old age. With a longitudinal design, they were able to show that increased loneliness was related to reduced brain structural measures. Such results could help guide clinicians and policymakers in designing social support systems that would benefit the growing aging population.

The strength of the current study lies in the large sample size and longitudinal follow-up design. The multilevel models used to separate within and between subject effects are well constructed. Combining neuroimaging data with behavioral changes provided further evidence that social loneliness may be related to accelerated brain aging. Stringent FDR correction, Bayes factor comparison, and the additional analyses for sensitivity showed the robustness and credibility of the results.

Thank you for a thorough and overall positive evaluation of our manuscript and the constructive feedback. We considered all of your comments valuable, please see point-by-point responses below for more details.

Weaknesses of the study were related to the interpretation and discussion of their findings.

1a) Social loneliness is a relatively little-studied factor in cognitive ageing, and the authors should consider expanding the discussion, with some additional analyses, as to how their results could be used by clinicians and older adults to monitor social behaviors.

We agree with the reviewer and are thankful for these suggestions. We have run additional analyses following the clinical cut-off of the questionnaire on social isolation and added those and their interpretations to the results and discussion section. Please see below response to questions 2a) and 3a) as well as to those in section b) to this reviewer how we implemented the reviewer’s advice in detail.

2a) The authors examined the interaction between baseline and age change to see if higher baseline loneliness was associated with accelerated decline. The interaction was significant, but the authors did not further explore the interaction effect, which may have clinical significance. The authors should consider identifying a cut-off point in LSNS that suggests persons scoring less than this score on the LSNS may be at greater risk of accelerated brain decline than others. Such a cut-off point is important for clinicians, as well as for future researchers to compare their results.

2a) Thanks to your recommendation, we decided to explore differences between handling LSNS as a categorical (using the standard threshold of 12) and continuous variable and recalculated all LMEs on HCV and cognitive functions with LSNS coded dichotomously. We found the results to be similarly good in detecting adverse effects of social isolation (see new Tables S16-18). The interaction of categorical LSNS with change in age on HCV tends towards showing an effect but does not reach significance even before FDR-correction.

As cut-off points are central to clinical work, we are convinced that this expansion improved our study greatly, contributed to its benefit to our readers and we are thus very grateful for this valuable question.

Our analyses indicate that the cut-off can be employed in clinical settings to detect social isolation that might harm patients’ brain health.

However, this does not answer another important question, namely which public health strategy is most suitable to target social isolation for preventive purposes. Should it focus on the most isolated individuals (i.e. those categorised as socially isolated) or pursue a population strategy (Rose et al., 2008)? This actually is the topic of ongoing research in our group and we hope to answer it in future work. For now, we ran additional models testing an interaction effect of dichotomous LSNS with continuous LSNS. Finding evidence for such an interaction effect would suggest that having less social contact has stronger negative effects for those that are categorised as socially isolated. Roughly speaking, is it worse to have one instead of two reliable friends than it is to have four instead of five? If this were the case, this would point public health towards a high-risk rather than population strategy. We did not find any evidence for such an interaction effect and thus can not say that we have found that more social contact ceases to be beneficial beyond the threshold score of 12. In addition to the new results, we have expanded on this in the discussion section where it now reads: „We showed that the established LSNS cut-off can be employed by clinicians to identify subjects likely to suffer adverse effects due to social isolation. However, the absence of evidence for more pronounced negative effects of less social contact amongst those that are deemed socially isolated by the cut-off renders a public health strategy focused on high-risk individuals questionable.”

3a) Although it was not directly tested in the paper, LSNS scores did not seem to change with increasing age (Table 1). This general stability of LSNS scores in older adults should be discussed further. The authors should consider how their relatively healthy and high SES sample may be less vulnerable to loss of family or friends in old age, making this sample sub-optimal for the question they have. The significance of the subject effect suggests that some individuals still experience a loss of social connectedness. The authors may want to elaborate on this and give some explanations of such subject differences in the ageing effect on social loneliness. Although stress was not a significant mediating factor, is it related to baseline loneliness or changes in loneliness in the current sample?

Concerning the link between change in age and LSNS we indeed found a statistically significant effect of age change on higher social isolation in an ancillary LME. However, as the reviewer noticed, the per year effect is very small, meaning that it would need getting more than 20 years older to score one point higher on the LSNS sum score (see new Table S2, see also answer below to questions 4a and 3b). We therefore tend to agree that in our sample, higher age does not affect social isolation substantially.

Furthermore, we very much appreciated your recommendation to further discuss how our relatively high SES-sample might be less vulnerable to loss of social contact during the aging process. As a foundation for this discussion, we investigated the link between SES and LSNS using an LME and found the association to be highly significant (see new Table S2). Furthermore, we added a table showing which percentage of our participants fell into the SES quintiles that would be observed in a fully representative German sample to help our readers to interpret our findings (see new Table S3). Following your advice, we have added a comment highlighting how the relatively high SES of our sample might have contributed to this in the limitations section: “As we found higher SES to be associated with lower LSNS scores, this relatively high SES sample might have led to underestimation of the detrimental effects of social isolation and increases in social isolation in the aging process.”

Regarding the importance of chronic stress to social isolation, we did not only find no mediating effect of stress, we also did not find a significant simple association between TICS and LSNS scores (see new Table S2). We are hesitant to attribute this finding to the incorrectness of the stress-buffering hypothesis as the missingness in stress data makes all interpretations of analyses involving TICS scores problematic. We have expanded on this in the discussion section and added emphasis to the importance of also pursuing other mechanistic theories in our discussion, where it reads: “we could not find evidence that social isolation affected hippocampal volume through higher chronic stress measured with questionnaires, a hypothesis put forward by the stress buffering theory (Kawachi & Berkman, 2001). These latter analyses suffered from small sample sizes and a limited number of timepoints. Nonetheless, the lack of any significant link between chronic stress and social isolation (see Table S2) is hard to align with the stress-buffering hypothesis in spite of the missingness in the TICS.”.

4a) The presentation of longitudinal data (Figure 1) lacks dimensionality. The scatter plots presented here are more suitable for cross-sectional studies and could cause confusion regarding the interpretation of the results. The authors should consider individual growth curves or spaghetti plots in visualizing change within subjects.

We are grateful for your advice to visualise individual developments in social isolation and outcome measures over time in spaghetti plots and have done so to give our readers insight into these developments (see new Fig. S1). As you had assumed, there is no unequivocal pattern of increasing social isolation over time (see also answer to 3a). In addition, we decided to stick with presenting results of the statistical modeling of linear mixed effect using scatterplots in Figure 1, as this is regarded the most appropriate visualization of the tested effectors. Please see also response to 5b.

Reviewer #2 (Public Review):

The paper by Laurenz Lammer and colleagues used cohort data to investigate the cross-sectional and longitudinal association between loneliness and brain structure and cognitive function. The main finding was that baseline social isolation and change in social isolation were associated with smaller hippocampus volumes, reduced cortical thickness, and poorer cognitive function. Given that more and more people feel lonely nowadays (e.g., due to the pandemic), the study by Lammer and colleagues addresses a highly relevant health concern of our time.

Significant strengths of the study:

• large cohort;

• the cross-sectional and longitudinal analyses confirmed the findings;

• the study was preregistered;

• the study included men and women;

• analyses were sound and controlled for essential confounders.

Thank you for your time to thoroughly review the manuscript and for the encouraging comments. Please see below how we implemented your advice.

The major weaknesses of the study:

1a) it is unclear whether loneliness causally contributes to brain structure and cognitive function;

Indeed, based on structural equation analyses of the available data from this cohort, we could not find strong evidence for neither causality (social isolation causes brain/cognitive decline) nor reverse causality (brain/cognitive decline causes social isolation). This could be due to a lack of power to detect such effects due to the drop in sample size for these analyses. Overall, regarding these two competing hypotheses, we see some minor indication of support for causality of social isolation in our data due to the presence of robust and significant associations in our very healthy sample, the absence of clear increases in effect size when including cognitively less healthy participants and the absence of clear decreases in effect sizes when only including participants with high MMST scores. Accordingly, we added this concluding synopsis to our paragraph on causality in our discussion: “Still, overall these results only add a modicum of corroboration to the case for a causal role of social isolation.” and pointed towards the key role of RCTs in understanding causality in this regard: ”Intervention studies will be the gold standard to provide evidence with regards to the causal role and effect size of social isolation.”

2a) the factors that may cause loneliness are unclear.

Thank you very much for encouraging us to shed some light on participant characteristics of potential relevance to social isolation. Starting from the impulse to look into marital status and employment, we also investigated links to socioeconomic status, migration background, age at baseline, change in age, gender, living alone and the number of persons living in the participants dwelling. We found all of these factors except for gender and migration background to be significantly linked to social isolation. Results are presented in Table S2 and briefly referred to in the results section: “In our sample, social isolation was positively correlated with not living alone, being married, the number of persons living in the participants’ dwelling, being gainfully employed, younger baseline age and less change in age and being married but no to gender or having a migration background. See Tables S1-2 for descriptive statistics and details of the associations. To contextualise the observed link to SES, a comparison of SES category frequencies in LIFE-Adult and a fully representative sample (Lampert et al., 2013) is provided in Table S3.” And added to the discussion: “Existing and future research on reasons for and the role of social isolation in health and disease should provide guidance for the urgently needed development and evaluation of tailored strategies against social isolation and its detrimental effects.”

Author Response

Reviewer #1 (Public Review):

Weakness of the study include:

1) There are no data supporting a role for insulin regulation of microtubule-dependent GLUT4-containg vesicle movement. The data in Fig.2B do not support a differences in the number of "moving" GLUT4 vesicles between basal and insulin-stimulated fibers. The statement on line 103 that they "observed a ~16% but insignificant increase" to be confusing. These data do not support an effect of insulin on the number of moving GLUT4 vesicles that can be detected in an individual experiment. There is also effect of insulin on GLUT4 vesicles in the data reported in Fig.S2D, Fig.S5B, and Fig.S5F. However, the data in Fig. 2C suggest there was a consistent increase in "moving" vesicles in insulin-stimulated conditions in 4 independent experiments (how are these data normalized?). Because the basis of insulin-regulation of glucose uptake is the control of GLUT4 translocation to the plasma membrane, the authors need to clarify their thinking on why they do not detect insulin robust effects on GLUT4 dynamics in the individual experiments. Is it that they are not measuring the correct parameter? That the assay is not sensitive to the changes?

The small (or no effect) of insulin distracts a bit from the findings that there is microtubule-dependent GLUT4 movement in basal and stimulated muscle fibers, and that disruption of this movement by depolymerization of microtubules or Kif5b knockdown blunts GLUT4 translocation. As noted above, the data strongly support microtubule-dependent GLUT4 dynamics as permissive for insulin-stimulated GLUT4 translocation even if this dynamics might not be a target of insulin action.

In light of the reviewer´s comment and to avoid confusing/distracting readers we have removed figure 2C showing the effect of insulin based on pooled data across all our independent experiments. We discuss several possibilities for the lack of significant insulin effect on GLUT4 movement in individual experiments in the discussion section (lines 342 to 361 in TC version of MS). The discussion has been updated to reflect the points raised by the reviewer. More sensitive techniques than currently available in our lab are required to firmly conclude whether microtubule-based GLUT4 trafficking is directly regulated by insulin.

2) The analyses of GLUT4-containing structures are not particularly informative. Co-localization with other markers (beyond syntaxin6) are needed to understand these structures. Defining structures as small, medium or large is incomplete. In particular, it is important to probe the microtubule nucleation site clusters for other membrane markers. Transferrin receptor? IRAP?

While our analysis based on structure-segmentation clearly demonstrate a microtubule-dependent effect on GLUT4 localization, we completely agree that additional work including co-labelling of GLUT4 and various compartment markers is required to fully understand the localization changes observed for GLUT4-containing structures upon microtubule disruption. However, for practical reasons, it is not currently feasible for us to complete these analyses within a reasonable time-frame so we will reserve this for future studies.

3) The Kinesore data do not support the authors hypothesis. The data show that Kinesore increases the amount of GLUT4 in the plasma membrane of basal cells and that insulin further increases plasma membrane GLUT4 to the same extent as it does in control cells. How does that provide insight into the role microtubules (or kif5b) in GLUT4 biology? Why does Kinesore increase plasma membrane GLUT4? Is it an effect of Kinesin 1 on GLUT4 vesicles? Kinesore is reported to remodel the microtubule cytoskeleton by a mechanism dependent on Kinesin 1. Is that the reason for the change in GLUT4?

To better understand the effect of kinesore on GLUT4-dependent glucose uptake, we have now incubated EDL and Soleus muscles ± kinesore and ± insulin and measured 2-DG uptake (GLUT4 translocation and glucose transport is considered the rate-limiting step for 2-DG uptake in incubated muscles due to the lack of muscle perfusion in this model) and proximal insulin signaling. In contrast to the enhancing effect on membrane GLUT4 observed following kinesore treatment in basal and insulin stimulated L6 cells, kinesore did not stimulate basal 2-DG uptake in EDL and Soleus. Furthermore, kinesore markedly impaired insulin-stimulated 2-DG uptake (figure 4B). We also tested the effect of 2h kinesore treatment in differentiated primary human myotubes. In this model, kinesore reduced basal glucose uptake and blocked the insulin effect (figure 4C). Together, this suggests that kinesore inhibits GLUT4-dependent glucose uptake in adult muscle and primary human muscle cells, presumably by inhibiting the binding of GLUT4 containing cargo, despite kinesore also having an activating effect on Kinesin-1 motor function. This possibility is discussed in the current version of the manuscript (line 177-180, 203-211). These data are consistent with the KIF5B knockdown data in L6 and support a necessary role of this motor protein in skeletal muscle GLUT4 trafficking.

To better understand, why kinesore led to increased rather than decreased GLUT4 translocation in L6 cells, we also disrupted the microtubule network using nocodazole and colchicine prior to kinesore stimulation. Surprisingly, kinesore stimulation enhanced membrane GLUT4 even in microtubule-disrupted L6 cells, indicating that the effect of kinesore on GLUT4 translocation is microtubule-independent in L6 cells. With three of four data sets supporting a necessary role of Kinesin-1 motor proteins in GLUT4 trafficking, including the adult muscle data, we end up concluding:

…our shRNA data in L6 myoblasts and kinesore data in adult muscle support the requirement of KIF5B-containing Kinesin-1 motor proteins in insulin-stimulated GLUT4-dependent glucose uptake in skeletal muscle.

However, we would also like to include the discrepant effect of Kinesore in L6 myoblasts as this may be useful information to others using this compound and/or studying GLUT4 in cultured cells.

4) The analysis of Kif5b is a bit cursory. Depolymerization of microtubules in muscle fibers essentially blocks all GLUT4 movement (only the insulin condition is shown in Fig.2B but I assume basal would be equally inhibited), and fully inhibits insulin-stimulated glucose uptake in muscle fibers. What are the effects of nocodazole in L6 cells (cell used for kif5b studies) and is it similar in magnitude to kif5b knockdown? Those data would identify there are non-Kif5b microtubule-dependent effects.

To address the magnitude of reduced insulin-stimulated GLUT4 translocation in microtubule-disrupted L6 cells, we investigated the effect of nocodazole (13 µM) and colchicine (25 µM) on GLUT4 translocation in L6 cells.

Insulin stimulated GLUT4 translocation was reduced but not blocked by either nocodazole or colchicine. This is in accordance with previous in vitro studies in 3T3 adipocytes and muscle cells (PMID: 11085918, PMID: 11145966, PMID: 24705014). Overall, these data still support that Kif5b is a major microtubule motor protein regulating GLUT4 translocation across cell-types.

5) The authors need to show that the fibers isolated from the HFD mice remain insulin-resistant ex vivo by measuring glucose uptake. It is possible that once removed from the mice they "revert" to normal insulin-sensitivity, which might contribute to the differences reported in Fig5.

This is an important point. In figure 5 figure supplement 1E, we show that the fibers isolated from the diet-induced obese mice display impaired insulin-induced p-Akt Thr308 and p-TBC1D4 Thr642 after isolation and in vitro culture. This shows that the insulin resistance is present at the muscular level and is preserved after isolation and in vitro culturing.

6) Although it is interesting that the authors have included the insulin-resistance models/experiments, they are not well developed and therefore the conclusions are not particularly strong.

In this study, we induced insulin resistance by two different means (C2 ceramide treatment and diet-induced obesity) and demonstrated at the level of p-Akt and p-TBC1D4 in cultured muscle fibers that we successfully achieved insulin resistance in our models. In particular the high fat diet model is arguably the most common in vivo model of obesity-linked insulin resistance. Thus, we were able to study GLUT4 trafficking on microtubules in normal vs. insulin-resistant muscle fibers and found this to be impaired in insulin-resistant muscle. Although one could always have done more, we believe that our data on adult muscle GLUT4 movement in insulin-resistance are robust, novel and do support our conclusions and title.

7) The data do not support the title.

We respectfully disagree. See our reply to comment 6 above.

Authorr Response

Reviewer #1 (Public Review):

1) The study finds Lyn to be degraded more efficiently via the proteasome and to be more tightly controlled by phosphatases when compared to Lck. However, rather than interpreting the findings as distinct kinase-intrinsic properties, one could attribute the slower degradation and stricter PTP control of Lyn to the fact that Lyn is the principal and predominant SFK in B cells and thus a "standard target" of the B-lymphoid molecular machinery, to which it is better adapted to.

We respectfully disagree with the reviewer’s comment that our interpretation is limited to “kinase-intrinsic properties”. In many points within the manuscript we refer to the “B-lymphoid molecular machinery”. More specifically:

• Lines 62-64 in the original submission (lines 60-61 in the revised manuscript): “….enzymatic promiscuity of SFKs can be buffered by their differential susceptibility to regulatory control mechanisms designed for keeping global SFK activity levels under strict control….”

• Lines 113-114 in the original submission (lines 137-138 in the revised manuscript): “Lck and Lyn differ in the efficiency for signal ignition and in their susceptibility to regulatory mechanisms in B-cells”

• Lines 135-136 in the original submission (lines 159-160 in the revised manuscript): “Thus, the proteasomal degradation machinery constrains the abundance of Lyn, but not Lck, within B-cells.”

• Lines 162-163 in the original submission (lines 185-186 in the revised manuscript): “Collectively these data show that the BCR signaling machinery is more responsive to the action of Lyn, at the same time imposing stricter regulation on its expression and activity levels.”

• Lines 475-477 in the original submission (lines 527-528 in the revised manuscript): “…identified specialized control mechanisms designed to keep Lyn, but not Lck, activity levels under strict control.”

However, we cannot rule out, as a mutually inclusive scenario, that intrinsic SFK features contribute to their differential regulation by cellular mechanisms, a possibility that we also refer to in the manuscript. More specifically:

• Lines 335-337 in the original submission (modified text in the revised version, lines 372-374): “On one hand there is the total amount of SFK activity within the cell, and on the other the individuality of SFK family members, dictated by intrinsic molecular features.”

• Lines 477-478 in the original submission (lines 528-529 in the revised manuscript): “These data may signify that SFKs have been evolutionarily diversified to best suit the needs of the cellular environment they are expressed in…”

Based on the reviewer’s comment, and to clarify further, we have modified the revised version of the manuscript (lines 372-374) as follows:

“On one hand there is the total amount of SFK activity within the cell, and on the other the individuality of SFK family members, dictated by intrinsic molecular features and/or adaptation to cell-specific regulatory mechanisms.”

We hope that our clarifications, satisfy the reviewer.

2) Venn diagram depicting differentially regulated transcripts between Lck- and Lyn-expressing cells, it does not seem like Lck is able to regulate pathways which are not "canonically" regulated by Lyn.

and

As a distinct functional difference between Lck and Lyn is not established in this work, said SFKs' largely exclusive expression in T and B cells remains enigmatic.

We thank the reviewer for the comment. We address this issue on the discussion section of the revised manuscript (lines 514-519).

3) There is also the persisting problem of Lck being expressed to a much higher extent and the effect of the endogenously expressed Lyn since the model systems are not based on a Lyn-deficient cell line.

For the purpose of the analysis, we tried to circumvent the discrepancies between Lck and Lyn expression levels by our equal GFP gating strategy (explained in Figure 1-figure supplement 3E/Fig.S3E in the original submission). Nevertheless, as shown in Figure 1C there is a physiological reason for the two SFKs not being equally expressed, and we refer to the biological implications of these individualities in the Discussion.

The effect of endogenously expressed Lyn is represented by the phenotype of -Dox cells which we use as background in all our studies, especially since we show that there are no alterations on Lyn or any other SFK activation status resulting from Lck overexpression (Figure 1-figure supplement 2B/ Fig.S2B in the original submission), so we do not believe this is a problem. Additionally, a Lyn-deficient environment would also not be perfect, since very plausibly it could have undergone further signaling and survival adaptations that we could not account for.

4) Lastly, the authors follow up their finding of deregulated transcripts belonging to the ER/UPR ontology cluster. Flow cytometric analysis indeed shows an influence of Lck and Lyn expression on ER homeostasis, which can be reverted with SFK inhibitors. Alas, additional follow-up experiments to functionally investigate the deregulated pathways suggested by the RNAseq analysis are not included in this study.

We thank the reviewer for the comment, and we agree. However, its beyond of our capabilities and manpower and the scope of the present work to perform numerous functional or semi-functional studies for every GO analysis pathway that emerged from the transcriptomics studies. Although follow up work from our group will focus on comprehensive and meticulous analyses of gene expression profiles, currently such an effort would require long-lasting studies which would also significantly extend the size of the manuscript but also distort the focus from the effects we wish to pinpoint with the present work i.e. the unique adaptation of SFKs within the lymphocyte environment and gene expression profile tendencies exclusively controlled by SFK-generated signals.

In an effort to satisfy the reviewer, we performed focused follow up studies specifically on the ER effect of SFK-transduced signals, since it appears to be a so-far unknown aspect of their function. The new data are presented in the revised version of Figure 4 (panels C and D) and Supplementary Figure 4-figure supplement 1. Corresponding text can be found in lines 323-345 of the revised manuscript (results section) and lines 499-512 and line 531 of the discussion. In brief, we show an SFK kinase-activity dependent activation of the ER-phagy receptor FAM134B, which is not accompanied by recruitment of LC3B, as dictated by the currently known canonical ER-phagy pathway. This is the first report of SFKs’ involvement in ER-phagy process and first time FAM134B activation is described in B-cells. Since this field is relatively new, and the role and regulation of ER-phagy is almost unexplored in B-cells, we hope that the reviewers will appreciate the novelty of the finding and its sufficiency for the current manuscript. We do realize that these initial data prompts for more detailed mechanistic investigation, which we are pursuing in the form of a more complete and comprehensive future study.

Reviewer #2 (Public Review):

1) Studies reveal no qualitative functional differences in Lck and Lyn that are likely to explain its unique ectopic expression of Lck in CLL

and

If Lck promotes pathophysiology by transduction of a qualitatively unique signal, one would expect that transcriptome analysis should reveal this difference.

We thank the reviewer for the comment. We address this issue on the discussion section of the revised manuscript (lines 514-519).

2) It is unclear from the material and methods whether the overexpressed Lyn is LynA or Lyn B. It appears in the text (lines 130-133) that they overexpress LynB specifically. A recent paper from Tania Freedman (Sci Adv 2022 PMID:35452291) suggests that LynA is more activating whereas LynB is more balanced with an inhibitory bias. The point is that it is important to discuss this because they may not be making a relevant comparison.

We thank the reviewer for the comment, to clarify this, we added in the Materials and Methods section of the revised manuscript (under “Cloning and Plasmids”) the use of Lyn isoform B.

We initially attempted to produce BJAB lines overexpressing LynA, however expression levels of this isoform was particularly low and we could not proceed with further analyses, so we cannot comment on how LynA might behave in an overexpression model in B-cells, especially given the absence of relevant information in the existing literature.

The recent Sci Adv 2022 PMID:35452291 study deals with germline LynA and LynB isoform-specific knockouts and their propensity towards autoimmunity in mice. The authors compared the single isoform (LynA or LynB) and total Lyn knockouts by performing systemic phenotypic analyses of autoimmunity features (splenomegaly, myeloid cell profiles, proinflammatory markers on myeloid cells, B cell development, expansion of activated and autoimmunity-associated B cell subsets, autoimmunity scores). Differences they pinpoint between LynA and LynB are summarized as follows:

1. “It was found that LynB has the dominant regulatory role in mice of both sexes, but that LynA expression is uniquely required to prevent autoimmunity in female mice”. The etiology of which is unclear.

2. “LynB generally appears to be the dominant immunosuppressive isoform, with LynB deletion causing severe autoimmune disease in male and female mice. For some indicators (splenomegaly, glomerular IgG and C3 deposition, and kidney fibrosis), LynBKO and total LynKO mice developed equally severe phenotypes. In other cases (serum IgM and BAFF, glomerular immune infiltration, myeloid cell polarization, and monocyte/granulocyte expansion), LynBKO mice had less severe phenotypes than total LynKO mice, suggesting an additive effect with LynA”.

3. “LynA and LynB seemed equally capable of promoting B cell development, regulating myeloid cell polarization and restraining myeloid-driven inflammation. Given the increased number of activated/inflammatory B cell types in LynAKO and LynBKO mice, future studies will be aimed at determining whether the single-isoform knockouts have a more B cell–initiated than myeloid cell–initiated form of autoimmune disease”.

After careful reading of the manuscript, we could not find any functional analyses on the activation status of the distinct isoforms, or signaling events they elicit. Furthermore, the authors do not report any conclusions that LynA is more activating at the molecular level. Based on the above, we cannot connect the data published in PMID:35452291 paper and our results for discussing “LynA being more activating” and implications this might have on our studies.

To comply with the reviewer’s suggestion, in our revised manuscript we cite this study (ref number 29) in the following sentence appearing in lines 380-383:

“Lyn exists as two alternatively spliced variants LynA and LynB. Distinct biological functions between the two isoforms still remain poorly understood. A recent study (29) documented that LynB provides an advantage in protecting against autoimmunity compared to LynA; however, the underlying mechanisms for this phenotype are unclear.”

Author Response

Reviewer #2 (Public Review):

The authors use data from 3 cross-sectional age-stratified serosurveys on Enterovirus D68 from England between 2006 and 2017 to examine the transmission dynamics of this pathogen in this setting. A key public health challenge on EV-D68 has been its implication in outbreaks of acute flaccid myelitis over the past decade, and past circulation patterns and population immunity to this pathogen are not yet well-understood. Towards this end, the authors develop and compare a suite of catalytic models as fitted to this dataset and incorporate different assumptions on how the force of infection varies over time and age. They find high overall EV-D68 seroprevalence as measured by neutralizing antibodies, and detect increased transmission during this time period as measured by the annual probability of infection and basic reproduction number. Interestingly, their data indicate very high seroprevalence in the youngest children (1 year-olds), and to accommodate this observation, the authors separate the force of infection in this age class from the other groups. They then reconstruct the historical patterns of EV-D68 circulation using their models and conclude that, while the serologic data suggest that transmissibility has increased between serosurvey rounds, additional factors not accounted for here (e.g., changes in pathogenicity) are likely necessary to explain the recent emergence of AFM outbreaks, particularly given the broader age-profile of reported AFM cases. The Discussion mentions important current unknowns on the biological interpretation of EV-D68 neutralizing antibody titers for protection against infection and disease. The analysis is rigorous and the conclusions are well-supported, but a few aspects of the work need to be clarified and extended, detailed below:

1) Due to the lack of a clear single cut-point for seropositivity on this assay, the authors sensibly present results for two cut-points in the main text (1:16 and 1:64). While some differences that stem from using different cut-points are fully expected (i.e., seroprevalence being higher using the less stringent cut-point), differences that are less expected should be further discussed. For instance, it was not clear in Figure 2 why the annual probability of infection decreased after 2010 using the 1:64 cut-point, while it continued to increase using the 1:16 cut-point. It would also be helpful to explain why overall seroprevalence and R0 continue to increase over this time period using the 1:64 cut-point. Lastly, it would be useful to see the x-axis in Figure 4 extended to the start of the time period that FOI is estimated, with accompanying credible intervals.

For the discussion on differences between the two cut-offs, please see response to essential comment 1.

Extending the x-axis before 2006 in Figure 4 is not possible. Estimates of the overall seroprevalence at a year y require FOI estimates up until y-40. This implies the first estimates we can provide are for 2006.

Credible intervals have been added to Figure 4.

2) Additional context of EV-D68 in the study setting of England would be useful. While the Introduction does mention AFM cases "in the UK and elsewhere in Europe" (line 53), a summary of reported data on EV-D68/AFM in England prior to this study would provide important context. The Methods refers to "whether transmission had increased over time (before the first reported big outbreak of EV-D68 in the US in 2014)" (lines 133-134), rather than in this setting. It would be useful to summarize the viral genomic data from the region for additional context - particularly since the emergence of a viral clade is highlighted as a co-occurrence with the increased transmissibility detected in this analysis.

We have added a figure (new Figure 1 – figure supplement 1) showing the annual number of EV-D68 detections reported by Public Health England from 2004 to 2020.

We have also added the following text to the introduction: “Similarly, in the UK, reported EV-D68 virus detections also show a biennial pattern between 2014 and 2018 (Figure 1 – figure supplement 1).”

We have also amended the sentence in the Methods.

Finally, below is a screenshot of the nexstrain tree for EV-D68 based on the VP1 region and with tips representing sequences from the UK (light blue) and European countries in colour. There is a lot of mixing between sequences from different regions, indicating widespread transmission and small regional clustering. We have added the following text to the Discussion: “Reported EV-D68 outbreaks in 2014 and 2016 were due to clade B viruses, while the 2018 outbreaks were reported to be linked to both B3 and A2 clade viruses in the UK (10), France (32) and elsewhere.”

Reviewer #3 (Public Review):

In the proposed manuscript, the authors use cross-sectional seroprevalence data from blood samples that were tested for evidence of antibodies against D68 for the UK. Samples were collected at 3 time points from individuals of all ages. The authors then fit a suite of serocatalytic models to explain the changing level of seropositivity by age. From each model they estimate the force of infection and assess whether there have been changes in transmissibility over the study period. D68 is an important pathogen, especially due to its links with acute flaccid myelitis, and its transmission intensity remains poorly understood.

Serocatalytic models appear to be appropriate here. I have a few comments.

The biggest challenge to this project is the difficulty in assigning individuals as seronegative or seropositive. There is no clear bimodal distribution in titers that would allow obvious discrimination and apparently no good validation data with controls with known serostatus. The authors tackle this problem by presenting results to four different cut-points (1:16 to 1:128) - resulting in seropositivity ranging from around 50% to around 80%. They then run the serocatalytic models with two of these (1:16 and 1:64) - leading to a range of FoI values of 0.25-0.90 for the 1 year olds and 0.05-0.25 for older age groups (depending on model and cutpoint). This represents a substantial amount of variability. While I certainly see the benefit of attacking this uncertainty head on, it does ultimately limit the inferences that can be made about the underlying risk of infection in UK communities, except that it's very uncertain and possibly quite high.

I find the force of infection in 1 year olds very high (with a suggestion that up to 75% get infected within a year) and difficult to believe, especially as the force of infection is assumed much lower for all other ages.

The authors exclude all <1s due to maternal antibodies, which seems sensible, however, does this mean that it is impossible for <1s to become infected in the model? We know for other pathogens (e.g., dengue virus) with protection from maternal antibodies that the protection from infection is gone after a few months. Maybe allowing for infections in the first year of life too would reduce the very large, and difficult to believe, difference in risk between 1 year olds and older age groups. I suspect you wouldn't need to rely on <1 serodata - just allow for infections in this time period.

Relatedly, would it be possible to break the age data into months rather than years in these infants to help tease apart what happens in the critical early stages of life.

Yes. We have added two figures (new Figures 1C and 1D) showing the prevalence of antibodies in children <1 yo. We show these data for the three serosurveys combined, because the number of individuals per month of age is very small.

One of the major findings of the paper is that there is a steadily increasing R0. This again is difficult to understand. It would suggest there are either year on year increases in inherent transmissibility of the virus through fitness changes, or year on year increases in the mixing of the population. It would be useful for the authors to discuss potential explanations for an inferred gradual increase in R0.

We have removed the estimates of R0 from the manuscript.

On a similar note, I struggle to reconcile evidence of a stable or even small drop in FoI in the 1:64 models 4 and 5 from 2010/11 (Figure 3) with steadily increasing R0 in this period (Figure 4). Is this due to changes in the susceptibility proportion. It would be good to understand if there are important assumptions in the Farrington approach that may also contribute to this discrepancy.

We have removed the estimates of R0 from the manuscript and only present the reconstruction of the annual number of new infections per age class and year (new Figure 5). We think this measure is more adapted to the discussion of the results.

In addition, when using the classical expression R{0t}=1/(1-S(t)), with S(t) the annual proportion seropositive, the high seroprevalence estimates (new Figure 4) result in extremely high estimates of the basic reproduction number (median ranges: 11.6 – 29.7 for 1:16 and 3.3 – 7.6 for 1:64 during the period 2006 to 2017).

We had previously used the Farrington approach as it is adapted to cases when the force of infections is different for different age classes.

The R0 estimates (Figure 4) should also be presented with uncertainty.

R0 no longer presented, but estimates of overall seroprevalence now presented with uncertainty.

Finally, given the substantial uncertainty in the assay, it seems optimistic to attempt to fit annual force of infections in the 30 year period prior to the start of the sampling periods. I would be tempted to include a constant lambda prior to the dates of the first study across the models considered.

We thank the reviewers for the suggestion.

We implemented this change (constant FOI before 2006) in the previous models without maternal antibodies and the result for the random-walk-based models was that the variance of the random walk was estimated over a very short period, thus resulting in a rather non- smoothed FOI.

Implementing this change with the new models with maternal antibodies and random-walk on the FOI was technically a bit complex. We therefore kept the simple random-walk over the whole period and added the following paragraph to the Discussion:

“It is important to interpret well the results for the estimates of the FOI over time from our analysis under the assumptions of the models. First, as the best model uses a random walk on the FOI, the change in transmission that we infer happens continuously over several years. In reality, this may have occurred differently (e.g. in a shorter period of time). Our ability to recover more complex changes in transmission is limited by the data available. It would not be surprising if EV-D68 has exhibited biennial (or longer) cycles of transmission in England over the last few years, as it has been shown in the US (7) and is common for other enteroviruses (30). However, it is difficult to recover changes at this finer time scale with serology data unless sampling is very frequent (at least annual). Therefore, our study can only reveal broader long-term secular changes. Second, interpretation of the results before 2006 must be avoided for two resasons. On the one hand, as we go backwards in time, there is more uncertaintly about the time of seroconversion of the individuals informing the estimates of the FOI. On the other hand, because age and time are confounded in cross-sectional seroprevalence measurements, the random walk on time may account for possible differences in the FOI through age (possibly higher in the youngest age classes, and lowest in the oldest), which are note explicitly accounted for here. This may explain the decline in FOI when going backwards in time before the first cross-sectional study in 2006.”

Author Response

Reviewer #3 (Public Review):

A large body of work in the literature has established that the diversity in cells of identical genetic background occurs due to two components: 1) intrinsic noise - such as stochastic fluctuations in gene expression - as well as 2) extrinsic noise - variability that arises from sources that are external to the biochemical process of gene expression, such as abundances of ribosomes or stage in the cell cycle. Note that this widely-accepted definition does not separate intrinsic and extrinsic from intracellular and extracellular. The authors cite a few of these seminal papers (which focus on noise introduced to gene expression) but then define their interpretation of intrinsic noise much more broadly "... intrinsic noise as phenotype(s) fluctuations across isogenic cell populations cultured under the same conditions. Measurement noise in some cases can also be thought of as intrinsic noise. Fluctuations in cellular phenotype(s) driven by the global environment will be referred to as extrinsic noise." This misuse of widely accepted terminology creates significant confusion in the interpretation of the results.

A point of contention with redefining noise as the authors have done is that they are lumping all processes unique to the cell as intrinsic and all environmental factors as extrinsic. Thus, when statements are made such as "external factors that contribute to noise are principally manifest through convection" (line 40-41, page 2) the veracity of these assumptions must be established. For example, when a ligand binds and unbinds from a receptor due to thermal energy, that "noise" in cellular stimulation is not convection-based, yet an example of how extrinsic noise can influence cellular responses. The definition is important because the underlying premise for the pipeline presented is that "While intrinsic cell variability can be significant, we believe that it is the extrinsic factor(s) that drive sample variability in most experimental cellular systems" (lines 42-43, page 4).

We thank the referee for this very important critical comment. The referee correctly points out that the terminology (intrinsic vs. extrinsic noise) used in the cited papers has to be adapted and more clearly stated.

We wish to point out that the autonomous system in Michael Elowitz and colleagues’ original paper was a single protein within a single cell. The noise that was measured in these experiments was driven by temporal fluctuations. An example of extrinsic noise for this system is, indeed, as pointed out by the referee, ligand binding and unbinding from a receptor.

By contrast, our autonomous system is an ensemble of cells isolated from other samples but still subject to fluctuations in the external environment. We did not continuously measure temporal fluctuations in individual cells, but recorded snapshot(s) of cellular phenotype(s) within a single sample. The source of noise in these measurements is variability between individual cells, and we referred to this type of noise as intrinsic because it driven by the processes within the sample. We denoted as extrinsic noise that which is driven by external factors to this autonomous system (a particular sample), such as variability between different samples due to temperature, humidity, etc.

All of these external factors (to the best of our knowledge) are related to movement and gradient formation of fluid or gas and, hence, from a physicochemical perspective, driven by convection process(es). The initial cell seeding that eventually leads to unique microenvironment formation can also be thought as an example of extrinsic noise using this terminology. The process of cell sedimentation and attachment is driven by advection, as the referee correctly points out. We have, therefore, adjusted the text accordingly.

We hope that clarifying the intrinsic/extrinsic terminology in the "Introduction" section of the manuscript (line 37) should be sufficient to avoid the confusion the referee discusses. We are open (very reluctantly) to switching terminology to terms internal and external noise.

Throughout, figures lack labels and sufficient explanation for interpretation, as well as the number of experiments used to generate the data that is processed through the pipeline for each condition. For a study designed to eliminate replicate culture conditions, the onus is on the authors to show that replicates are in fact fully recapitulated in the population variance after statistical binning/processing.

To address this comment, we modified the figure legends and labels of most of the figures.

We wish to emphasize that each point-injection experiment we performed is unique due to randomness in the local delivery method. This is due to the variability in the manual micro-injection release rate and direction of the initial flow. Several experiments (3+) were performed to improve the width of the label(s) distribution(s) and their mixing condition, and the results of the better optimized local delivery were selected as representative for the manuscript. Sample selection was independent of the outcome of drugs action and based on initial label distribution only. An experimental improvement of our method, similar to initialization of the pseudo-random number generator in numerical experiments, is required to achieve systematic reproducibility of drug(s) distribution(s). One way to do so is robotically, but certainly the best is to design a system that utilizes a predictably constant drug gradient within a sample that contains large enough cells, a topic that will be the subject of future experiments.

Ultimately, when the paper presents results such as Figure 9 as the culmination of the pipeline as applied to cell viability studies, it is unclear how useful insight is extracted from this methodology. Four drugs are applied in combination to adherent HeLa cells and time-dependent local cell density is provided as a proxy for cell viability. While it is stated that "The absolute drug concentration can be determined using the homogeneous delivery method discussed above" (line 421-422, page 19), this analysis is not performed, and I am left unsure of whether extrinsic factors are truly driving sample variability under this context. It is unclear to the reader how the point injections were administered, and no discussion of how the confounding factors of synergy or antagonism will be addressed through this methodology.

We attempted to explain that data shown in Figure 9 were not meant to be the climactic point of the entire pipeline (rather, the data shown in Figure 6 represent our key achievement). In this four-drug experiment, we exhausted the fluorescent spectrum bandwidth necessary to distinguish drug labels (i.e., using commonly available microscopy tools). In order to estimate local cell density, we had to rely on bright field imaging data which is not the most accurate possible implementation (see further response to your comment below). More importantly, we had to wash samples between the measurements to remove detached (dead) cells and cell debris. This step can (and usually does) influence local cell density in a non-uniform fashion, since both media removal and deposition are performed locally by pipetting (cells in the vicinity of aspiration/media deposit sites can be washed off regardless of the drug treatment.)

To clarify how point injections were administered, we added a detailed description in the Methods section. Please see section Drug labeling and delivery, pages 11-12.

In this manuscript, we wished to establish possible applications of our method and avoid in depth analysis or biological interpretation of a specific drug combination that is dependent on the cell line or on a particular experimental condition. We added a paragraph in the "Discussion" section suggesting the necessity of future research dedicated to methodology and analytical interpretation of high-dimensional context-dependent drug interaction data.

Author Response

Reviewer #2 (Public Review):

The authors unexpectedly found that the protein Grb2, an adaptor protein that mediates the recruitment of the Ras guanine-nucleotide exchange factor, SOS, to the EGF receptor, can be recruited to membranes by the immune cell tyrosine kinase Btk. The authors show, using total internal reflection fluorescence (TIRF) microscopy that the interaction with Grb2 is reversible, dependent on the proline-rich region of Btk, and independent of PIP3. These experiments are well performed and unambiguous.

The authors next asked whether Grb2 binding to Btk influences its kinase activity, by evaluating (i) Btk autophosphorylation and (ii) the phosphorylation of a peptide from the endogenous substrate PLCy1. The readout relies on non-specific antibody-mediated detection of phosphotyrosine but nevertheless reveals a concentration-dependent increase in both Btk autophosphorylation and PLCy1 phosphorylation. The experiments, however, have only been performed in duplicate and, particularly in the case of PLCy1 phosphorylation, exhibit enormous variability which is not reflected in the example blot the authors have chosen to display in Figure 3C. Comparison of the same, duplicate experiment presented in Figure 3 Supplement 2 paints a very different picture.

We added an experiment wherein we measure phosphorylation of the PLC𝛾2-peptide fusion by Btk in the presence of different concentrations of Grb2, and we have carried out LC-MS/MS to probe which Tyr are phosphorylated in these experiments. We have also modified our presentation of the Western blot data to allow readers to view each replicate separately. We believe this makes it easier to evaluate the trends observed in each replicate, and because the intensity measured here is only semi-quantitative, due to limitations of the technique, we believe this is a more accurate way to present our results. Both Tyr of the PLC𝛾2-peptide are phosphorylated, as well as one Tyr at the very C-terminus of GFP (Figure 3 – Supplements 3-5).

The authors next sought to determine which domains of Grb2 are required for activation of Btk. Again, these experiments were only performed in duplicates, and the authors’ claims that Grb2 can moderately stimulate the SH3-SH2-kinase module of Grb2 are not well supported by their data (Figure 4C-D).

We have opted to remove the data for the activation of the SH3-SH2-kinase construct (Src module) from the revised manuscript. Upon further inspection, we agree that these experiments only showed a weak trend and believe that much more experimentation is needed to draw firm conclusions regarding this construct. We do still speculate that SH2 linker displacement may contribute to our observations of enhanced catalytic activity of Btk in the presence of Grb2, however this speculation is based solely on previous work with Btk and other kinases (Aryal et al., 2022; Moarefi et al., 1997).

The authors next asked whether Grb2 stimulates Btk by promoting its dimerization and trans- autophosphorylation. The authors measured the diffusion coefficient of Btk on PIP3- containing supported lipid bilayers in the presence and absence of Grb2. They noted that the diffusion coefficient of individual Btk particles decreases with increasing unlabeled Btk, which they interpret as Btk dimerization. Grb2 does not appear to influence the diffusion of Btk on the membrane (Figure 5A). Presumably, the diffusion coefficient reported here is the average of a number of single-molecule tracks, which should result in error bars. It is unclear why these have not been reported. Next, the authors assessed the ability of Grb2 to stimulate a mutant of Btk that is impaired in its ability to dimerize on PIP3-containing membranes. In contrast to wild-type Btk, autophosphorylation of dimerization-deficient Btk is not enhanced by Grb2. Whilst the data are consistent with this conclusion, again, the experiment has only been repeated once and the western blot presented in Figure 5 Supplement 2 is unreadable. It is also puzzling why Grb2 gets phosphorylated in this experiment, but not in the same experiment reported in Figure 3 Supplement 2.

The diffusion coefficient reported here is determined from a large number of single molecule tracks. We have expanded our explanation of how this is done in the Materials and Methods, as well as providing an example of the data and fits from one of the conditions in Figure 4 – Supplement 3. We are now including standard deviation for each diffusion coefficient determined from the fit of the step size distribution.

We have opted to remove the data involving the dimerization-deficient Btk construct. We agree that these results are difficult to interpret.

We have investigated the Grb2 phosphorylation signal and concluded that this is an off-target effect of the antibody. MS/MS cannot detect and phosphorylation on Grb2. We now comment on this in the figure legend of Figure 3 – Supplement 2.

Finally, the authors argue that Grb2 facilitates the recruitment of Btk to molecular condensates of adaptor and scaffold proteins immobilized on a supported lipid bilayer (SLB) (Figure 6). This is a highly complex series of experiments in which various components are added to supported lipid bilayers and the diffusion of labelled Btk is measured. When Btk is added to SLBs containing the LAT adaptor protein (phosphorylated in situ by Hck immobilized on the membrane via its His tag), it exhibits similar mobility to LAT alone, and its mobility is decreased by the addition of Grb2. The addition of the proline-rich region (PRR) of SOS further decreases this mobility. In this final condition, the authors incubate the reactions for 1 h until LAT undergoes a phase transition, forming gel-like, protein-rich domains on the membrane, shown in Figure 6B. The authors’ conclusion that Btk is recruited into these phase-separated domains based on a slow-down in its diffusion is not well supported by the data, which rather indicates that Btk is excluded from these domains (Figure 6B – Btk punctae (green) are almost exclusively found in between the LAT condensates (red)). As such, the restricted mobility of Btk that the authors report may simply reflect the influence of barriers to diffusion on the membrane that result from LAT condensation into phase- separated domains. The authors also present data in Figure 6 Supplement 1 indicating that Grb2 recruitment to Btk is out-competed by SOS-PRR and that Btk does not support the co- recruitment of Grb2 and SOS-PRR to the membrane. These data would appear to suggest that the authors’ interpretation of the decreased mobility of Btk on the membrane may not be correct.

We have now included an example of one of the single molecule videos, overlayed with the surrounding LAT phase, to more directly display the data that was recorded for this experiment. In this video, it is possible to see that the LAT dense phase occupies only some of the observed window, and although it is possible that these dense “islands” function as barriers to Btk diffusion, Btk would be expected to diffuse freely outside of the LAT dense areas of the bilayer. This property can be clearly seen in the video we have now included. This is reminiscent of what was observed previously during the LAT phase transition for tracking of LAT itself (Sun et al., 2022). Given the extensive previous analysis of LAT diffusion on supported lipid bilayers (Lin et al., 2022; Sun et al., 2022), we believe the necessary controls have been included to support our conclusions. However, we agree there is much to be learned about this interaction and we hope that future studies will further investigate the relationship between cytoplasmic kinases and plasma membrane associated signaling clusters.

Reviewer #3 (Public Review):

The study of Nocka and colleagues examines the role of membrane scaffolding in Btk kinase activation by the Grb2 adaptor protein. The studies appear to make a case for a reinterpretation of the "Saraste dimer" of Btk as a signaling entity and assigns roles to the component domains in the Src module in Btk activation. The point of distinction from earlier studies is that this work ascribes a function to an adaptor protein as promoting the kinase activation, rather than vice versa, and also illustrates why Btk can be activated via modes distinct from its close relative, such as Itk. Importantly, these studies address these key questions through membrane tethering of Btk, which is a successful, reductionist way to mimic cellular scenarios. The writing could be improved and can absolutely be more economical in word choice and use; currently, there is a good deal of background to each section that is not always comprehensive or crucial to contextualise the findings, while key information is often omitted. The results are currently not described in a detailed manner so there is an imbalance between the findings, which should be the focus, relative to background and interpretations or models.

We have assessed the manuscript and made many improvements to shift the focus to the findings, while providing only the necessary background for readers unfamiliar with the specifics of Btk and Grb2 signaling and structure.

Author Response

Reviewer #1 (Public Review):

Ge et. al., examined sodium-glucose cotransporter-2 inhibitors (SGLT2i) in Alport syndrome (AS), and demonstrate that it was beneficial in AS through reduced lipotoxicity in podocytes as a key mechanism of action. The SGLT2i empagliflozin has been previously shown to have positive effects on hyperglycemia control, as well as on cardiovascular and renal outcomes of type II diabetes mellitus through tubuloglomerular feedback, but its effect on glomerular diseases such as AS are unknown to date. The authors have previously identified that cholesterol efflux in podocytes plays a critical pathogenic role in a diabetic kidney disease setting. The evidence that authors provide in favor of their hypothesis in a disease of non-metabolic origin such as AS, was supported as the SGLT2i was effective in reducing the deleterious effects of lipotoxicity in podocytes, ameliorated glomerular injury and proteinuria, and extending the life span of Col4a3 knockout mice. They further show that empagliflozin treatment mitigated AS podocytes from cell death through apoptosis, but did not impact the cell's cytotoxicity. These results support the notion that empagliflozin affects the regulation of important metabolic switch in mouse kidneys, perhaps through decreasing lipid accumulation in podocytes.

However, the authors solely rely on one IHC staining image of a human biopsy to demonstrate SGLT2 expression in podocytes in vivo. Although the authors have done several experiments which greatly increase the confidence in their findings that empagliflozin is beneficial in AS and would have clinical significance, their data does not rule out the possibility that empagliflozin has beneficial effects through the other glomerular cells in AS, or limited to impacting lipids in podocytes in AS.

We thank the reviewer for recognizing the significance of our findings and for pointing out some additional concerns with our study. In this revised version, we have added experiments that focus on investigating the specific effect of empagliflozin on AS podocytes. We added immunofluorescence staining of AS mouse kidney sections which supports the idea that SGLT2 is expressed in podocytes. We investigated the effect of SGLT2 knockdown in AS podocyte using siRNA and compared the anti-lipotoxic effects of siSGLT2 to SGLT2i.

Reviewer #3 (Public Review):

Using cultured human podocytes the expression of SGLT2 is established using immunostaining and western blotting. An analysis of podocyte RNA wasn't performed, but the expression in cultured podocytes was comparable to that seen in human cultured proximal tubular cells. This work then paved the way for treatment of immortalized cells obtained from an Alport syndrome mouse model (Col4A3-/-), representing an autosomal recessive form of Alport syndrome. Podocytes from Alport syndrome mice showed a lipid droplet accumulation which was reduced to some extent by SGLT2 inhibition. In a series of metabolic experiments, it was shown that SGLT2 inhibition reduced the formation of pyruvate as a metabolic substrate in Alport podocytes. In vivo experiments showed an improvement in survival of Col4a3-/- mice treated with SGLT2 inhibition. When compared to ace inhibitor, SGLT2 inhibition has a similar effect on renal function and no additive effect was seen with SGLT2 inhibitor plus ace inhibitor. Like the cell assays, the in vivo treatment seemed to prevent the podocyte lipid accumulation in Alport syndrome mice.

This data in cells and animals generally supports the findings in SGLT2 inhibitor human studies, where Alport syndrome patients with proteinuria and progressive CKD seem to benefit. The work paves the way for a dedicated trial of SGLT2i in Alport patients and a reassessment of the human podocyte disease phenotype in this condition, before and after treatment. There are patients with mutations in SGLT2 with familial renal glycosuria - it would be interesting to test via urine derived podocytes whether a similar metabolic switch was occurring and its consequences to pave the way for long term treatment regimes.

We thank the reviewer for recognizing the significance of our findings. We appreciate the reviewer’s concern that podocyte SGLT2 RNA levels should be studied. In this revised version, we added the results of SGLT2 mRNA expression analysis in immortalized podocytes and tubular cells. These results were added in Figure 1E. We agree with the insightful suggestions to study the metabolic switch in familial renal glucosuria in patients with SGLT2 mutations, as well as to evaluate Col4a5 AS model. We have included these insights in our discussion.

Author Response

Reviewer #1 (Public Review):

The authors address the origin of the macrophage increase in sensory ganglia after peripheral nerve injury, showing that there is no major influx by blood-derived monocytes into ganglia after injury and that resident macrophages proliferate, which is dependent on CX3CR1 signaling.

• Interesting and relevant question, mainly addressed with adequate experimental approaches.

• Most conclusions are supported by the data, however, some important controls and experiments are missing.

• The authors should demarcate their results from the study of Iwai et al, 2021 which addresses similar questions.

Thank you for the positive comments, we hope that our point-by-point responses below and the important changes/inclusions in the MS satisfactorily addressed your concerns. We agree that some important controls were missing, and we have included additional data in the revised manuscript. Regarding the Iwai et al. paper, it is in line with our hypothesis. In fact, they suggest that in trigeminal ganglia (TG), resident macrophages proliferate after peripheral injury, although they detected few blood monocytes infiltrating the TG. Our paper, besides to confirm Iwai et al. results, by using different and complementary approaches are more specific compared to BM transfer in irradiated mice, we also advanced in terms of the mechanisms that these cells proliferate (CX3CR1 signalling) and the impact of these proliferation for neuropathic pain development. We discussed these points in the new version of the MS. Please see page 4 lines 88-93.

Reviewer #2 (Public Review):

The investigators looked at mφs in lumbar DRG after a spared nerve injury in which two of the three branches of the sciatic nerve are transected and the third left intact. This is a classical preparation for studying neuropathic pain. This paper demonstrates that the increase of mφs is an increase in the number of CX3CR1+ (resident) mφs and not CCR2+ (infiltrating mφs) by using CX3CR1 and CCR2 individual reporter mice. Using a CX3CR1 conditional knockout (KO) mouse, they found that this receptor must be present on the mφs for the increase in number to occur. Next, they did a parabiosis experiment with GFP+ mice and found that neither of these mφ subtypes infiltrated into the DRG. To examine proliferation, they injected animals with Ki67 and found this label, which is an indication of proliferation, was present in the CX3CR1+ mφs (but not the CCR2+ mφs). Finally, they identified the CX3CR1 mφs to be the cells that express TNFα and IL-1β but not IL-6.

An experiment that would be useful would be to determine if there is an increase or a decrease in the availability to mφs of the ligand CXC3L1 after the spared nerve injury. The authors state from the work of others that membrane-bound CX3CL1 is constitutively expressed and that it is decreased after nerve injury. They hypothesize that this indicates a release of the chemokine, but such a decrease could also indicate a decrease in expression. A few sentences on what is known in other systems on the importance and mode of action of membrane-bound and non-membrane-bound CX3CL1 would be useful.

Thanks to the reviewer for a great summary of our manuscript. We have now performed a time course of Cx3cl1 expression in the DRG after the spared nerve injury and it was included in figure 7A. We also apologise for the lack of information regarding the importance and mode of action of membrane-bound and non-membrane-bound CX3CL1, which is now included in the discussion section (Page 16).

The main weakness of the manuscript is that many highly relevant previous findings, in some cases reporting nearly identical experiments sometimes with the same and sometimes with somewhat different results, are not mentioned. Kalinski et al. (which is cited but not in this context) reported a very similar parabiosis experiment. While they did not identify subtypes of mφs, they found an increase in infiltration of mφs, which was small (though statistically significant) compared to the larger increase that occurred in the distal nerve. In 2013 and 2018, Niemi et al. and Lindborg et al (J Neurosci and J

Neuroinflammation respectively) reported that mφs in the DRG are somewhat decreased in a CCR2 KO mouse, suggesting again that there is some infiltration of mφs into the DRG after axotomy. They also showed that the mφ chemokine CCL2 increases in the DRG after sciatic nerve injury. With regard to proliferation, Yu et al. in 2020 (which again is cited but not in this context) also used a spared nerve paradigm stained DRGs for CX3CR1+ mφs and found an increase. They then stained DRG sections for Ki67 and demonstrated proliferation in this population. An earlier reference by Krishnan et al in 2018 published in J Neuropathol Exp Neurol is entitled "An Intimate Role for Adult Dorsal Root Ganglia Resident Cycling Cells in the Generation of Local Macrophages and Satellite Glial Cells". With regard to cytokine expression, in 1995, Murphy et al published a paper in J Neurosci demonstrating induction of interleukin-6 in axotomized sensory neurons.

Thank you for the comment. These papers, you have indicated, are the main reason we have idealised our MS. The controversy regarding the possible infiltration of peripheral blood monocytes for the increase in the number of macrophages in the sensory ganglia after peripheral nerve injury. Furthermore, some of these papers you also indicated, came out during the execution of this manuscript, and they also brought controversies or did not explore some points. Therefore, we believe that our work by using different and complementary approaches strongly support the hypothesis that after peripheral nerve injury, peripheral blood monocytes did not infiltrate the DRGs significantly, but that the increase in the macrophages population is due to the proliferation of resident macrophages. Furthermore, we provided novel mechanistic evidence of the role of CX3CR1 signalling for the proliferation of these cells (figures 7 and S6). In addition, our new experiments suggested by the referees and editor suggest that CX3CR1-dependent proliferation of DRG macrophages is involved in the development of neuropathic pain (Figures 6D and 7E). We will make these points clear in the new version of the MS. Please see pages 11, 12, 14 and 17 (discussion and introduction section).

Reviewer #3 (Public Review):

This paper addresses the mechanism underlying a well-documented finding whereby the numbers of resident macrophages increase in dorsal root ganglia following peripheral nerve injury. It delineates the relative contribution of monocyte recruitment via circulation and local proliferation. The paper is clearly structured and written, and the data overall support the main conclusion that the increase in nerve-associated macrophages is primarily driven by proliferation, not monocyte recruitment. Its main weakness is that the question that is being asked is rather restricted, so the additional insight gained for the field will be incremental. It would be particularly interesting in the future to address whether the existence of a protective barrier indeed is the reason peripheral cells are not recruited to the nerve injury lesion and to assess e.g. whether forced breaching of this barrier results in monocyte influx and altered injury response.

We appreciate your comments and suggestions. In the new version of the MS, we are presenting a series of novel experiments that confirm and support our initial hypothesis. Furthermore, novel experiments also explore the importance of the phenomenon we have explored in the context of neuropathic pain development. Regarding your suggestion about the next steps, we are working now in an attempt to understand why these cells are not able to infiltrate the DRGs after injury. Interestingly, one paper that came out during the revision of this work, showed that CD8+ T cells that are not able to infiltrate the DRGs after nerve injury in adult mice, start to infiltrate the DRGs of old mice (Zhou et al. 2022), indicating that ageing process may promote changes in this protective barrier. In addition, we have published a recent paper indicating that immune cells infiltrate the dorsal root leptomeninges after SNI (Maganin et al. 2022). We included these references and discussed these points in the new version of our MS. Please see page 15 lines 366 and 370.

References:

Zhou, L., G. Kong, I. Palmisano, M. T. Cencioni, M. Danzi, F. De Virgiliis, J. S. Chadwick, G. Crawford, Z. Yu, F. De Winter, V. Lemmon, J. Bixby, R. Puttagunta, J. Verhaagen, C. Pospori, C. Lo Celso, J. Strid, M. Botto, and S. Di Giovanni. 2022. "Reversible CD8 T cell-neuron cross-talk causes aging-dependent neuronal regenerative decline." Science 376 (6594): eabd5926. https://doi.org/10.1126/science.abd5926.

Maganin, A. G., G. R. Souza, M. D. Fonseca, A. H. Lopes, R. M. Guimarães, A. Dagostin, N. T. Cecilio, A. S. Mendes, W. A. Gonçalves, C. E. Silva, F. I. Fernandes Gomes, L. M. Mauriz Marques, R. L. Silva, L. M. Arruda, D. A. Santana, H. Lemos, L. Huang, M. Davoli-Ferreira, D. Santana-Coelho, M. B. Sant'Anna, R. Kusuda, J. Talbot, G. Pacholczyk, G. A. Buqui, N. P. Lopes, J. C. Alves-Filho, R. M. Leão, J. C. O'Connor, F. Q. Cunha, A. Mellor, and T. M. Cunha. 2022. "Meningeal dendritic cells drive neuropathic pain through elevation of the kynurenine metabolic pathway in mice." J Clin Invest 132 (23). https://doi.org/10.1172/JCI153805.

Author Response

Reviewer #1 (Public Review):

This study focuses on the role of polo like kinase 1 (PLK-1) during oocyte meiosis. In mammalian oocytes, Plk1 localizes to chromosomes and spindle poles, and there is evidence that it is required for nuclear envelope breakdown, spindle formation, chromosome segregation, and polar body extrusion. However, how Plk1 is targeted to its various locations and how it performs these functions is not well understood. This study uses C. elegans oocytes as a model to explore PLK-1 function during meiosis. They take advantage of an analogue-sensitive allele of plk-1, which enabled them to bypass nuclear envelope breakdown defects that occur following PLK-1 RNAi. This allowed them to dissect later roles of PLK-1 in oocytes, demonstrating that depletion causes defects in spindle organization, chromosome congression, segregation, and polar body extrusion. Moreover, the authors defined mechanisms by which PLK-1 is targeted to chromosomes, showing that CENP-C (HCP-4) is required for localization to chromosome arms and that BUB-1 is required for targeting to the midbivalent region. Finally, they demonstrate that upon removal of PLK-1 from both domains, there are severe meiotic defects. These findings are interesting. However, there is a need for additional analysis to better support some of their conclusions, and to aid in interpretation of particular phenotypes. Specific comments are below.

• For many important claims of the paper, a single representative image is shown but the n is not noted. This is an issue throughout the paper for much of the localization analysis (e.g. Figure 1B, 1C, 1D, 2A, 2B, 3A, 3B, 3C, etc.); in cases like this, numbers should be included to increase the rigor of the presented data. How many images or movies were analyzed that looked like the one shown? For linescans, were they done only on one image? How many independent experiments were done, etc?

We had initially chosen a representative image. Localisation was the same in all images that allowed ‘proper’ assessment of PLK-1 localisation. In our case, this means that we can only analyse bivalents that are perpendicular to the light path to distinguish between bivalent, chromosome arms, and kinetochore. We now report the number of oocytes (N) and bivalents (n) analysed for each condition. The line scans were done in one representative image.

• In the abstract, it is stated that PLK-1 plays a role in spindle assembly/stability (this is also stated elsewhere, e.g. line 101). This phrasing implies that the authors have demonstrated roles in both spindle assembly and stability. However, to distinguish between these roles, they would have to show that removal of PLK-1 before spindle assembly causes defects, and also that removal of PLK-1 from pre-formed spindles causes collapse. I don't think it is necessary to do this, as the spindle roles of PLK-1 are not a focus of the paper. However, the language should be altered so that it does not imply that the paper has demonstrated roles in both. A good place to do this would be in the section from lines 144-147, where they first discuss the spindle defects. It would be straightforward to explain that their approach does not distinguish between spindle assembly and stability, and that PLK-1 could have a role in either or both.

We fully agree with this comment. We cannot distinguish between spindle assembly and stability, and it is also not the focus of our current work. We have changed the text accordingly.

• It is stated that there is kinetochore localization of PLK-1 (and I do see some dim cup-like localization in images after PLK-1 is removed from the chromosome arms via HCP-4 RNAi). However, this cup-like localization is not clear in most wild-type images (e.g. Figure 1B, 1D, 2A, 3A, etc.). Although I recognize that the chromatin staining might be obscuring kinetochore localization, if PLK-1 was truly a kinetochore protein I would also expect it to localize to filaments within the spindle (as many other kinetochore proteins do), especially since the authors state that BUB-1 targets PLK-1 to the kinetochore (and BUB-1 is in the filaments). In fact, the only images where it looks like PLK-1 may be localized to filaments are in Figure 4C and 6A, when HCP-4 has been depleted (though I don't know if this generally true across all HCP-4 RNAi images). For me, this calls into question the conclusion that PLK-1 truly is on the kinetochore in wild type conditions - could it be that PLK-1 only localizes to the kinetochore (and to the filaments) when HCP-4 is depleted? The authors need to resolve this issue and provide better evidence that PLK-1 normally localizes to the kinetochore, if they want to make this claim. Additionally, the observation that PLK-1 is not on the kinetochore filaments (in wild type conditions) should be addressed in the text somewhere - do the authors think that this is a special type of kinetochore protein that does not localize to the filaments?

While our initial claim of PLK-1 kinetochore localisation was based on its cup-like localisation, we have now performed additional analysis and experiments to confirm this claim. First, we corroborated that PLK-1 cup-like pattern co-localises with the Mis12 complex component KNL-3 (New Figure 5-figure supplement 1). Second, we show that PLK-1 is present in the so called ‘linear elements’ (filaments) both within the spindle and in the cortex. Since PLK-1 presence in these filaments is seen in wild type as well as hcp-4 mutant oocytes, we conclude that PLK-1 likely localises in kinetochore in normal conditions.

• The authors should provide a control experiment, treating wild-type worms with 10uM 3-IB-PP1. This would be important to ensure that the spindle defects seen at this concentration in the plk-1as strain are not non-specific effects of the inhibitor. There is a control in Figure 1 - figure supplement 3 using 1uM 3-IB-PP1 but didn't see a control for 10uM (the concentration at which spindle defects are observed).

This control has now been included in Figure 1-figure supplement 3.

• In Figure 2F, the gels for BUB-1+PLK-1 look different in the presence and absence of phosphorylation by Cdk1 - for these data, I agree with the authors that it looks as if the complex elutes at a higher volume if BUB-1 is not phosphorylated (lines 200-204). However, Figure 2G has a repeat of the condition with phosphorylated BUB-1, and in this panel, the complex appears to elute at a higher volume than it did on the gel in panel F. The gel in panel G looks much more similar to the unphosphorylated condition in panel F. The authors need to explain this discrepancy (i.e., Is there a reason why the gels cannot be compared between panels? How reproducible are these data?). Ideally, the authors would include a repeat of the unphosphorylated BUB-1 + PLK-1 condition in panel G, done at the same time as the conditions shown in that panel, to avoid the impression that their results may not be reproducible.

The specific elution volume cannot be compared in different experiments as the column has proven to “drift” over time – with proteins eluting at a later volume than they did previously despite extensive washing. What is reproducible under the experimental conditions is that the unphosphorylated wild type proteins, or the phosphorylated T527A/T163A mutant proteins A) elute at a later volume than the phosphorylated wild type proteins and B) bind to a lower proportion of the MBP-PLK1PBD (as you can see in the relative absorbance profiles and Coomassie gels).

• The authors would need to provide convincing evidence that co-depletion of BUB-1 and HCP-4 delocalizes PLK-1 from the chromosomes entirely, and that this co-depletion condition is more severe than either single depletion alone.

We now provide a quantitation on the total PLK-1 levels to go along the images (New Figure 8-figure supplement 1).

Additionally, the bub-1T527A and hcp-4T163A alleles are nice tools to, in theory, more specifically delocalize PLK-1 from the midbivalent and chromosome arms, respectively, to explore the functions of chromosome-associated PLK-1. However, I think the authors cannot rule out the possibility that other proteins are also being depleted from the midbivalent and/or chromosome arms in their conditions, and that this delocalization may contribute to the phenotypes observed. For example, hcp-4 depletion was recently shown to delocalize KLP-19 from the chromosome arms (Horton et.al. 2022), so in the experiment shown in Figure 6E (HCP-4 RNAi in the bub-1 mutant), PLK-1 was likely not the only protein missing from the chromosome arms. Therefore, understanding if other proteins are absent from these domains (in the bub-1T527A and hcp-4T16A3 mutants) would help the reader understand and interpret the presented phenotypes (and how specific they are to PLK-1 loss). Consequently, I think that to better understand the co-depletion analysis presented in Figure 6 (and Figure 6 supplement 1), the authors should analyze other midbivalent and chromosome arm proteins, to determine if any are also delocalized (e.g. SUMO, KLP-19, MCAK, etc.).

As stated above, this paper focuses on identifying the specific meiotic events PLK-1 plays a role in and characterising its targeting mechanism. We are following on this work to understand what proteins are regulated by PLK-1 in different chromosome domains and how this relates to the observed phenotypes.

For the current, we should emphasise that mutating a single Thr residue within an STP motif in a largely disordered region is far more specific than depleting HCP-4 or BUB-1, making it likely that the observed effects are mediated through PLK-1 targeting. It should be noted that the finding presented in Horton et.al. 2022 is in contradiction with another study in which hcp-4 depletion did not impact KLP-19 localisation (Hattersley et al 2022).

Additionally, instead of performing a combination of mutant and RNAi analysis (i.e. HCP-4 RNAi in the bub-1 mutant (Figure 6) and BUB-1 RNAi in the hcp-4 mutant (Figure 6 figure supplement 1)), it would be more powerful to generate a double mutant - this has a higher chance of being a more specific depletion condition.

We have performed these experiments, which are now presented in Figure 9.

Author Response

Reviewer #1 (Public Review):

Sorkac et al. devised a genetically encoded retrograde synaptic tracing method they call retro-Tango based on their previously developed anterograde synaptic tracing method trans-Tango. The development of genetically encoded trans-synaptic tracers has long been a difficult stumbling block in the field, and the development of trans-Tango a few years back was a breakthrough that was immediately, widely, and successfully applied. The recent development of the retrograde tracer method BActrace was also exciting for the field, but requires lexA driver lines and required by its design the test of candidate presynaptic neurons instead of an unbiased test for connectivity.

Retro-Tango now provides an unbiased retrograde tracer. They cleverly used the same reporter system as for trans-Tango by reversing the signaling modules to be placed in pre-synaptic neurons instead of post-synaptic neurons. Therefore, synaptic tracing leads to the labeling of pre-synaptic neurons under the regulation of the QUAS system. Using visual, olfactory as well sexually dimorphic circuits authors went about providing examples of specificity, efficiency, and usefulness of the retro-Tango method. The authors successfully demonstrated that many of the known pre-synaptic neurons can be successfully and specifically labelled using the retro-Tango method.

Most importantly, because it is based on the most used, very well tested and widely adopted trans-Tango method, retro-Tango promises to not just be a clever development, but a really widely and well-used technique as well. This is an outstanding contribution.

We would like to thank Dr. Hiesinger for his very kind words and for the overall appreciation of the contribution of the development of retro-Tango to the field. We are also grateful for the suggestions below aimed at improving the clarity of our manuscript. We individually address the points raised by Dr. Hiesinger below.

Reviewer #2 (Public Review):

Tools that enable labeling and genetic manipulations of synaptic partners are important to reveal the structure and function of neural circuits. In a previous study, Barnea and colleagues developed an anterograde tracing method in Drosophila, trans-TANGO, which targets a synthetic ligand to presynaptic terminals to activate a postsynaptic receptor and trigger nuclear translocation of a transcription factor. This allows the labeling and genetic manipulation of cells postsynaptic to the ligand-expressing starter cells. Here, the same group modified trans-TANGO by targeting the ligand to the dendrites of starter cells to genetically access pre-synaptic partners of the starter cells; they call this method retro-TANGO. The authors applied retro-TANGO to various neural circuits, including those involved in escape response, navigation, and sensory circuits for sex peptides and odorants. They also compared their retro-TANGO data with synaptic connectivity derived from connectivity obtained from serial electron microscopy (EM) reconstruction and concluded that retro-TANGO can allow trans-synaptic labeling of presynaptic neurons that make ~ 17 synapses or more with the starter cells.

Overall, this study has generated and characterized a valuable retrograde transsynaptic tracing tool in Drosophila. It's simpler to use than the recently described BAcTrace (Cachero et al., 2020) and can also be adapted to other species. However, the manuscript can be substantially strengthened by providing more quantitative data and more evidence supporting retrograde specificity.

We thank Dr. Luo for his kind words and his assessment of the value of retro-Tango as a new tool in the transsynaptic labeling toolkit in Drosophila. We followed the suggestions of Dr. Luo for providing more quantitative data and addressing the specificity and directionality of retro-Tango. We strongly believe that the implementation of his suggestions did enhance the quality of our manuscript.

Reviewer #3 (Public Review):

This is a valuable addition to the currently available arsenal of methods to study the Drosophila brain.

There are many positives to the present manuscript as it is:

(i) The introduction makes a clear and fair comparison with other available tracing methods.

(ii) The authors do a systematic analysis of the factors that influence the labeling by retro-tango (age, temperature, male versus female, etc...)

(iii) The authors acknowledge that there are some limitations to retro-TANGo. For example, the fact that retro-T does not label all the expected neurons as indicated by the EM connectome. This is fine because no technique is perfect, and it is very laudable that the authors did a serious study of what one should expect from retro-tango (for example, a threshold determined by the number of synapses between the connected neurons).

We would like to thank the reviewer for the kind words and the positive assessment of our manuscript. In addition, we would like to acknowledge the reviewer for the recommendations below, which we followed and we think made our manuscript stronger.

Author Response

Reviewer #1 (Public Review):

Bustion and colleagues outline the creation and testing of an in-silicon method to query gut microbiome databases for genes encoding enzymes predicted to catalyze a reaction of interest, which is provided by the user. Strengths of the tool include attempts to examine nearly 9,000 MetaCyc reactions in a pre-calculated fashion and to rank order enzymes based on their likelihood of catalyzing a reaction. Substrates, products, and even cofactors, if known, are employed to strengthen the power of the search algorithm, which also employs a hidden Markov model to improve the selection of putative hit enzymes. The authors outline high success rates with examples presented and compare those results with other extant methods, which are reported to perform in a less robust manner. Weaknesses include lack of evidence of success on a more difficult "real world" example. However, the tool outlined is a clear advance over existing methods and will be useful to explore the diversity of chemical transformation performed by commensal microbiota.

We thank Reviewer 1 for their positive feedback and constructive summary. We agree that a real-world example would add confidence to our findings. We previously demonstrated SIMMER’s utility using published datasets. To expand upon these findings, we added another evaluation on an external dataset (Artacho et al., 2020) and performed new experiments to test SIMMER predictions for methotrexate metabolism into DAMPA and glutamate, a reaction known to be performed by the human microbiome but for which human gut strains and specific gut enzymes were not previously known. Both the new external dataset and our experimental findings validate SIMMER’s predictions of bacteria capable of metabolizing methotrexate, the mainline therapeutic for rheumatoid arthritis patients.

Reviewer #2 (Public Review):

This work provides a new computational tool for the systematic characterization of biotransformation reactions in the human gut microbiome: given a biotransformation reaction of interest, it predicts a list of candidate bacterial species, enzymes, and EC identifiers putatively capable of performing the queried reaction. The method is innovative and clearly presented.

The pipeline that relies on both chemical and protein similarity algorithms, is in principle applicable to any biotransformation reaction that can be formulated as linked substrates and products (possibly including co-factors). This contrasts with other approaches that, for example, only rely on smaller databases and solely rely on substrates and chemical similarity. Moreover, SIMMER outperformed two other recently developed methods, against which it was benchmarked for its prediction accuracy when tested on a control test set derived from literature.

The work interestingly focuses on predicting bacterial enzymes responsible for drug biotransformation, therefore showcasing its potential as a hypothesis generator for characterizing and validating novel bacterial enzymes in vitro.

The authors correctly describe the relevance of an accurate input (in terms of reaction completeness, including cofactors and reaction products) as paramount for the quality of the prediction.

The conclusions of this paper are mostly well supported by data, but some aspects of performance evaluation and its generality might benefit from additional elaborations and clarifications.

1) Great emphasis has been dedicated to the prediction performance of SIMMER over a positive control set derived from the available literature. However, a more extensive description and analysis of false positive results are needed to better understand the possible impact of the (potentially many) false positive predictions listed for each reaction.

We agree that our analysis would benefit from an assessment of false positives. Unfortunately, current literature usually reports which reactions an enzyme is capable, rather than incapable, of performing. For this reason, we took a conservative approach and decided to define all reactions preceding that which yielded a positive control enzyme sequence as false positives. This is now described above in Essential Revisions Response 1.3.

2) The authors imply that the current method is superior to two other methods based on accuracy. However, a more extensive description of the benchmarking results would strengthen these benchmarking efforts.

We have addressed this concern in Essential Revisions Response 3.

3) The authors only showcase SIMMER in the context of drug metabolism but claim its applicability to be general enough to also describe other biotransformation in the human gut microbiota. Although in principle believable, the authors could improve the credibility and generalizability of their method by demonstrating another use case, e.g., food compounds, for which extensive metagenomic and metabolomic data are already available from previous gut microbiome studies.

We agree that assessments of SIMMER’s predictions on food metabolism would improve the generalizability of the method. We have edited the text to focus on drug metabolism, as we believe SIMMER’s application to food metabolism merits a more thorough, future investigation.

4) Showcasing experimental in vitro validation of SIMMER predicted enzyme(s) could greatly strengthen the relevance of this work.

We have addressed this in Essential Revisions Response 2.

5) Throughout the text and the title, a more careful and precise phrasing of the tool's scope (characterization of microbiome-encoded enzymatic reactions and not the identification of novel chemical transformations) would improve the reader's understanding of the work.

We agree, and have reworded many key phrases in the text, including the title.

Reviewer #3 (Public Review):

This manuscript presents a new tool, SIMMER, to predict bacterial enzymemediated transformations of compounds, an important and incompletely understood aspect of microbiome drug metabolism. The authors compare their resource to existing resources that allow users to generate hypotheses related to compound toxicity and putative routes of compound metabolism. The authors identify the key innovations of their resource as including full chemical representations of reactions and a novel method to predict an enzyme's EC number (a description of function) from its reaction.

Strengths

Generating user-friendly tools to explore existing knowledge of bacterial enzymes and their reactions is important.

SIMMER is a novel resource where the user provides the substrates and products as input and receives a list of potential microbiome enzymes as output.

SIMMER includes a novel EC predictor based on reaction rather than based on sequence.

Weaknesses

Validation claims are not well supported by the results.

We have extensively edited the manuscript to better describe our previous computational validations, and we have added new analyses to further evaluate SIMMER. We added an additional validation on an external dataset, an in vitro experimental assessment of SIMMER’s predictions for methotrexate metabolism, two new reactions to the positive control analysis, a false positive rate, and additional comparisons to the two competing methods.

Need for the user to know both the substrate and the product for a reaction of interest limits the utility of the resource.

We agree that this is a limitation for the user, but as we show in our Results, relying on substrates alone does not yield appropriate representations of reactions and therefore does not allow for accurate predictions of responsible species/strains and enzymes (i.e., finding True Positives, and confirming associations from previously collected data). We agree that tools requiring only substrates are convenient, but our results show that they are less helpful in finding appropriate metabolism and enzyme predictions. Many studies of biotransformation in the human gut identify the product information or product structure via HPLC, LC-MS, and NMR techniques. In cases where such data was not gathered, or not gathered with enough structural resolution, researchers can use tools such as Biotransformer to make product template predictions before inputting a query to SIMMER. This recommendation is included in the present manuscript’s lines 376–391:

In instances when DrugBug and MicrobeFDT did make predictions, they suffered from low accuracy (Table 1), which we hypothesized was due to both methods’ reliance on substrate rather than reaction chemistry. Biotransformations involve the relationship between substrate(s), cofactor(s), and an enzyme to yield a particular product(s). As one substrate can exhibit affinity for multiple enzymes, resulting in multiple unique products, sole employment of substrates in a chemical fingerprint does not achieve the resolution necessary to make relevant predictions. To test if SIMMER’s better performance could be attributed to including cofactors and products, we modified our code to run with a chemical representation that includes only the substrate of each positive control reaction. Enzyme prediction accuracy dropped from 88% down to 33%, and EC prediction accuracy dropped from 93% down to 48% (Table 1—source data), supporting the hypothesis that SIMMER’s better performance when compared to DrugBug and MicrobeFDT is due in large part to our using chemical representations that include the full reaction. These results are in line with our previous demonstration that SIMMER clusters enzymatic reaction chemistry only when a full reaction is employed (Figure 2, Figure 2—figure supplement 4).

Reliance on homology transfer annotation to predict enzyme function; this approach has important, microbiome-relevant, limitations.

Please refer to our separate Common_Questions.pdf document, Common question 1: Are EC codes sufficient to select enzyme orthologs within an overall class?

Author Response:

The authors would like to thank the Editors and reviewers for their careful consideration of our article and we express our appreciation for the work required by both Editors and reviewers to study and produce the detailed reviewer reports. We are pleased at the general consensus that our paper is of interest and highlights an important region of the channel for drug-protein interaction. We are also cognizant that the reviewer reports highlight areas where important revisions need to be made to our work before it can be considered fully complete. We will revise the paper according to the comments of the reviewers and submit a new version in the near future which we hope will become the version of record.

Author Response

We thank the editors and reviewers for their support of our work, as well as their constructive feedback and useful suggestions, which have improved the readability and presentation of the manuscript for a broader audience.

Author Response

Reviewer 1 (Public Review):

Fox, Birman, and Gardner use a previously proposed convolutional neural network of the ventral visual pathway to test the behavioral and physiological impact of an attentional gain spotlight operating on the inputs to the network. They show that a gain modulation that matches the behavioral benefit of attentional cueing in a matching behavioral task, induces changes in the receptive fields (RFs) of the model units, which are consistent with previous neurophysiological reports: RF scaling, RF shift towards the attentional focus, and RF shrinkage around the focus of attention. Ingenious simulations then allow them to isolate the specific impact of these RF modulations in achieving performance improvements. The simulations show that RF scaling is primarily responsible for the improvement in performance in this computational model, whereas RF shift does not induce any significant change in decoding performance. This is significant because many previous studies have hypothesized a leading role of RF shifts in attentional selection. With their elegant approach, the authors show in this manuscript that this is questionable and argue that changes in the shape of RFs are epiphenomena of the truly relevant modulation, which is the multiplicative scaling of neural responses.

Strengths:

The use of a multi-layer network that accomplishes visual processing, with an approximate correspondence with the visual system, is a strength of this manuscript that allows it to address in a principled way the behavioral advantage contributed by various attentional neural modulations.

The simulations designed to isolate the contributions of the various RF modulations are very ingenious and convincingly demonstrate a superior role of gain modulation over RF shifts in improving detection performance in the model.

We thank the reviewer for these supportive comments.

Weaknesses:

There is no mention of a possible specificity of the manuscript conclusions in relation to the type of task to be performed. It is conceivable that mechanisms that are not important for detection tasks are instead crucial for a reproduction task, as in Vo et al. (2017).

We agree that other behavioral tasks may rely on different attentional mechanisms then the ones we have studied here for detection and discrimination and now specifically point this out in the discussion [379-395].

The manuscript puts emphasis on the biological plausibility of the model, and some quantitative agreements. But at some important points these comparisons do not appear very consistent:

1) It is unclear what output of the model at each cortical area is to be compared with neurophysiological data. On the one hand, the manuscript argues that a 1.25 attentional factor is consistent with single-neuron results, but here this factor is applied to the inputs into V1 units. When this modulation goes through normalization in area V1, the output of V1 has a 2x gain. Intuitively, one would think that recordings in V1 neurons would correspond to layer V1 outputs in the model, but this is not the approach taken in the manuscript. This needs clarification. Also, note that the 20-40% gain reported in line 287 corresponds to high-order visual areas (V4 or MT), but not to V1, in the cited references. The quantitative correspondence between gain factors at various processing steps in the model and in the data is confusing and should be clearer.

We agree that making a one-to-one mapping of gain effects measured in neurophysiology and different layers of the CNN is problematic. We therefore have clarified that the introduction of gain at the earliest stages of processing is meant to study how gain propagates through a complex CNN and has downstream effects [49-52 and 410-447] and we have also also clarified the various uncertainties in making one-to-one mapping from the CNN to neurophysiological measurements of gain [410-447].

2) The model assumes a gain modulation in the inputs to V1. This would correspond to an attentional gain modulation in LGN unit outputs. There is little evidence of such strong modulation of LGN activity by attention. Also in V1 attentional modulation is small. As stated in Discussion (line 295), there is no reason to favor the current model as opposed to a model where the attentional gain is imposed later on in the visual hierarchy (for example V4). If anything, neurophysiology would be more consistent with this last scenario, given the evidence for direct V4 gain control from frontal eye fields (Moore and Armstrong, Nature 2003). The rationale for focusing on a model that incorporates the attentional spotlight on the inputs to V1 should be disclosed.

We agree that measurements of gain changes with attention appear larger in later stages of visual processing and do not wish to explicitly link the gain changes imposed at the earliest stages of processing in our CNN observer model with changes in input from LGN as we agree this would be unrealistic. Instead, our goal was to examine how gain changes can propagate through complex neural networks and cause downstream effects on spatial tuning properties and the efficacy of readout. We have substantially re-written the manuscript, in particular the introduction [24-38, 49-52] and discussion [441-447] to better describe this rationale. We also now explicitly discuss how our propagated gain test demonstrates exactly the reviewer’s point - that gain can be injected late in the system, rather than at the earliest stages [274-276, 441-447].

3) The model chosen is the CORnet-z model, but this model does not include recurrent dynamics within each layer. Recurrent dynamics is a prominent feature in the cortex, and there is evidence indicating that attentional modulations operate differently in feedforward and in recurrent architectures (Compte and Wang, Cerebral Cortex 2006). A specific feature of recurrent models is that the attentional spotlight need not be a multiplicative factor (which is biologically complicated) but an additive term before the ReLU non-linearity, which achieves the expected RF modulations (Compte and Wang, 2006). A model with recurrence thus represents another architecture that links gain and shift in a way that has not been explored in this manuscript, and this may limit the generalization of the conclusions (line 205).

We appreciate the reviewer pointing us toward the Compte paper and we’ve added a discussion of recurrence as an alternate model [410-423].

Reviewer 2 (Public Review):

This manuscript by Fox, Birman, and Gardner combines human behavioral experiments with spatial attention manipulation and computational modeling (image-computable convolutional neural network models) to investigate the computational mechanisms that may underlie improvements in behavioral performance when deploying spatial attention.

Strengths:

• The manuscript is clear and the analyses, modeling, and exposition are executed well.

• The behavioral experiments are carefully conducted and of high quality.

• The manuscript takes a creative approach to constructing a ”neural network observer model”, that is, coupling an image-computable model to a potential readout mechanism that specifies how the representations might be used for the purposes of behavior. The focused analyses of the model innards (architecture, parameters) provide insight into how different model components lead to the final behavior of the model.

Thank you for these supportive comments.

Weaknesses:

• The overall conclusions and insights gained seem heavily dependent on particular choices and design decisions made in this specific model. In particular, the readout mechanism lacks some critical descriptive details, and it is not clear whether the readout mechanism (512-dimensional representation that reflects summing over visual space) is a reasonable choice. As such, while the computational analyses and results may be correct for this model, it is not clear whether the strong general conclusions are justified. Thus, the results in their current form feel more like exploratory work showing proof of concept of how the issue of attention and underlying computational mechanisms can be studied in a rigorous and concrete computational modeling context, rather than definitive results concerning how attention operates in the visual system.

Please see below for our response to the issue with readout and conclusions.

Overall, the work is solidly constructed, but the overall generality and strength of the conclusions require substantial dampening.

Author Response:

We would like to thank the reviewers for their time, insights, and constructive feedback. We appreciate the recognition by the reviewers of the value and importance of our study. The reviewers also highlighted: the importance of carefully using and interpreting data from small molecule inhibitors due to possible off-target effects, considering inter-study differences in the cardiomyocyte cell trajectories, examining a possible role of PI3K signaling in proliferation and the intriguing yet not fully elucidated role of membrane protrusions in cardiac fusion. We agree with this important feedback. We plan to address these comments and others directly, in detail.

Author Response:

We thank the reviewers and editors for their careful reading and reviews of our work. We are grateful that they appreciate the value in our experimental approach and results. We acknowledge what we interpret as the major criticism, that in our original manuscript we focused too heavily on the hypothesized role of GABAergic neurons in driving habituation. This hypothesis will remain only indirectly supported until we can identify a GABAergic population of neurons that drives habituation. Therefore, we will revise our manuscript, decreasing the focus on GABA, and rather emphasizing the following three points:

1. By performing the first Ca2+ imaging experiments during dark flash habituation, we identify multiple distinct functional classes of neurons which have different adaptation profiles, including non-adapting and potentiating classes. These neurons are spread throughout the brain, indicating that habituation is a complex and distributed process. 

2. By performing a pharmacological screen for dark flash habituation modifiers, we confirm habituation behaviour manifests from multiple distinct molecular mechanisms that independently modulate different behavioural outputs. We also implicate multiple novel pathways in habituation plasticity, some of which we have validated through dose-response studies.

3. By combining pharmacology and Ca2+ imaging, we did not observe a simple relationship between the behavioural effects of a drug treatment and functional alterations in neurons. This observation further supports our model that habituation is a multidimensional process, for which a simple circuit model will be insufficient. 

We would like to point out that, in our opinion, there appears to be a factual error in the final sentence of the eLife assessment: “However, the data presented are incomplete and do not show a convincing causative link between pharmacological manipulations, neural activity patterns, and behavioral outcomes”. We believe that a “convincing causative link” between pharmacological manipulations and behavioural outcomes has been clearly demonstrated for PTX, Melatonin, Estradiol and Hexestrol through our dose response experiments. Similarly a link between pharmacology and neural activity patterns has also been directly demonstrated. As mentioned in (3), we acknowledge that our data linking neural activity and behaviour is more tenuous, as will be more explicitly reflected in our revised manuscript. Nevertheless, we maintain that one of the primary strengths of our study is our attempt to integrate analyses that span the behavioural, pharmacological, and neural activity-levels.

Author Response

Reviewer #1 (Public Review):

Rosas et al studied the mechanism/s that enabled carbapenems resistance of a Klebsiella isolate, FK688, which was isolated from an infected patient. To identify and characterize this mechanism, they used a combination of multiple methods. They started by sequencing the genome of this strain by a combination of short and long read sequencing. They show that Klebsiella FK688 does not encode a carbapenemase, and thus looked for other mechanisms that can explain this resistance. They discover that both DHA-1 (located on the mega-plasmid) and an inactivation of the porin OmpK36, are required for carbapenem resistance in this strain. By using experimental evolution, it was shown that resistance is lost rapidly in the absence of antibiotics selection, by a deletion in pNAR1 that removed blaDHA-1. Moreover, their results suggested that it is likely that exposure to other antibiotics selected for the acquisition of the mega-plasmid that carries DHA-1, which then enabled this strain to gain resistance to carbapenemase by a single deletion.

The major strength of this study is the use of various approaches, to tackle an important and interesting problem.

The conclusions of this paper are mostly well supported by data, but one aspect is not clear enough. The description of the evolutionary experiment is not clear. I could not find a clear description of the names of the evolved populations. However, the authors describe strains B3 and A2, but their source is not clear. The legends of the relevant figure (Figure 5) are confusing. For example, the text describing panel B is not related to the image shown in this panel. Moreover, it is shown in panel C (and written in the main text) that the OmpK36+ evolved populations had only translucent colonies, so what is the source of B3(o)?

We appreciate the point and in response have added a panel to Figure 5 (in the revised paper this is now Fig. 5A) to illustrate the evolutionary experiment and specify that there are two lineages (A and B) with 20 replicates each that, after 200 generations of evolution, give rise to populations of which A2 and B3 are the exemplars characterized.

We have corrected the legends in Figure 5.

We now explain (sentence starting on Line 197) that the B3 (o) is the single isolate of an opaque colony from lineage B3, it is the only colony that we identified from out of 595 colonies observed in the B3 population. B3(o) was sequenced and analysed as a comparator and has some value in that regard, despite being an anomaly.

Reviewer #2 (Public Review):

The authors sequenced a clinical pathogen, Klebsiella FK688, and definitively establish the genetic basis of the carbapenem-resistance phenotype of this strain. They also show that the causal mutations confer reduced fitness under laboratory conditions, and that carbapenem sensitivity readily re-evolves in the lab due to the fitness costs associated with the resistance mutations in the clinical isolate. They also establish that subinhibitory concentrations of ceftazidime select for the otherwise deleterious blaDHA-1 gene. Based on this finding the authors speculate that prior beta-lactam selection faced by the ancestors of Klebsiella FK688 potentiated the evolution of the carbapenem-resistance phenotype of this strain. If this hypothesis is true, then prior history of beta-lactam exposure may generally potentiate the evolution of carbapenem resistance.

Strengths:

From a technical perspective, the findings in this paper are solid. In addition, the authors establish a simple genetic basis for carbapenem resistance in a clinical strain, which is a valuable and non-trivial finding (i.e. they show that the CRE phenotype in this strain is not an omnigenic trait distributed over hundreds of loci).

Weaknesses:

The main weakness of this paper is that the authors draw overly broad conclusions of a conceptual nature from narrow experimental findings. This could be addressed by drawing more modest and narrow implications from the findings.

1) The title of this paper is "Treatment history shapes the evolution of complex carbapenem-resistant phenotypes in Klebsiella spp." But they provide no data on the treatment history of the patient from whom this strain was isolated from. Therefore, the authors have no evidence to support their central claim. Indeed, it is completely possible that this strain never faced beta-lactam selection in the past, or that the patient's hypothetical history of betalactamase was irrelevant for the evolution of FK688. First, it is completely possible that this is a hospital-acquired infection, such that the history of this strain is due to selection in other contexts in the hospital that have little to do with the patient's treatment history. Second, it is completely possible that this strain (the chromosome anyway) has no prior history of beta-lactamase selection, and that it acquired the megaplasmid containing blaDHA-1 via conjugation from some other strain. In this second hypothetical scenario, it is possible that the fitness cost of the blaDHA-1 gene is not particularly high in a different source strain, but that it has some cost in the FK688 strain that it was isolated from. And of course, fitness costs in the human host could be very different than fitness costs in the laboratory, where strains are evolving under strong selection for fast growth. And given the benefit of resistance, it's clear that this strain clearly has a strong fitness advantage over faster-growing sensitive strains in the context of the source patient under antibiotic treatment.

My general point here is that the broad claims made about patient history or prior history shaping the evolution of this strain are largely indefensible because there is no data here to make solid inferences about how prior history shaped the evolution of this strain.

We appreciate the point and have changed our title and scaled back the strength of our conclusions regarding patient treatment history.

2) Historical contingency. The authors claim that their work shows how historical contingency shapes the evolution of resistance. One problem with this claim is that it is trivial- this is only a significant claim if the reader believes that prior history is not important in the evolution of antibiotic resistance, which is a straw-man null hypothesis, to mix a couple metaphors. To be more concrete, clearly strain background (prior history) matters-eliminating the plasmid with the resistance gene eliminates resistance. But that is not particularly surprising, given the past 50 years of evolutionary microbiology literature on plasmids and resistance. By contrast to this work, the major contribution of papers that examine the role of historical contingency in evolution (i.e. various Lenski papers) is that those works quantitatively measure the role of history in comparison to other factors (chance, adaptation). Since this work is a deep dive into a single clinical isolate, the data presented here do not and cannot shed light on the role of historical contingency in the emergence of this strain. The authors' claims about the prior history that led to the CRE phenotype are reasonable- but are fundamentally speculative. I have nothing against speculation, as long as it is clear what claims are speculative, and what are concrete implications. But the authors frame these speculative claims as concrete implications of their findings.

This is a fair point. We have reframed the study to not focus on historical contingency.

As the reviewer points out, any discussion about historical contingency in the context of evolution is trivial in one sense. One of the reasons that the studies of Lenski and Blount provide new insights into the role of historical evolution because they knew the history of their populations (at, least for the number of generations since the LTEE began), and had a high degree of control and understanding of the growth conditions where the trait evolved. As such, they could go back to time points before the trait evolved, and then repeat the evolution experiment many times, in the exact same environment where the trait originally evolved, and then count how often they observed the evolution of that trait.

Here we study a clinical isolate, and have less understanding of the evolutionary history of our strain. While we cannot re-evolve carbapenem resistant in the exact same environment experienced by the FK688 strain, we did test the capacity for the wild type, and two possible intermediate genotypes genotypes, to evolve carbapenem resistance in growth media with carbapenem.

Altogether- we have comprehensive evidence for the genetic cause of carbapenem resistance: the BLA1 plasmid + OmpK36. We showed, by experiment, that it is much more likely for carbapenem resistance to evolve in a FK688 strain that carries the BLA1 plasmid, than in an FK688 strain that did not carry the plasmid even if it had acquired the OmpK36 mutation. We think this not trivial because a significant proportion of all of the carbapenem resistant Klebsiella that have been isolated are non-carbapenemase CRE. Our reconstruction provides a plausible explanation for why non-carbapenemase CRE evolve – because they are evolving from strains that have already been treated with a non-carbapenem beta-lactam drug and have thereby selected for the presence of a beta-lactamase (that is not a carbapenemase).

So, while we have scaled back the strength of our claims, we do think that our results can provide some insight into how the evolutionary history of a pathogen can shape the molecular path to antibiotic resistance.

3) The authors claim that "[This work] suggests that the strategic combinations of antibiotics could direct the evolution of low-fitness, drug-resistant genotypes". I suppose this is true, but I also think this is a stretch of an implication given these findings. To be blunt, while I suppose it's better to have costly resistance variants that re-evolve sensitivity than to have low-cost high-resistance strains circulating, I think the patient's family would probably disagree that the evolution of a low-fitness drug-resistant genotype was good or strategic in the clinical context, even if better from a public health perspective. Low-fitness drug-resistant strains are just as lethal under clinical antibiotic concentrations!

Thank you for the comment, we see how this sentence could be seen as too strong a conclusion and have rewritten the last sentence of the DISCUSSION (line 351):

“These results show how an individual’s treatment history might shape the evolution of AMR, and should be taken into consideration in order to explain the evolution of non-carbapenemase CRE”

The authors do show the plausibility of their hypothesis/model that prior beta-lactam selection is sufficient to potentiate the evolution of carbapenem-resistance (by the additional ompK loss-of-function mutation). I think those findings are very nice. But the authors undermine their results by extrapolating too far from their data. Hence, I think narrowing the scope of the implications would improve this paper.

In addition to narrowing the scope of the implications as written, I also would like to add that there may be other ways of framing this paper (other than historical contingency) that may make the significance of this work more apparent to a broader audience. This may be worth considering during the revision process.

We have taken these suggestions on board and have re-framed the final sentences of the ABSTRACT, INTRODUCTION and DISCUSSION accordingly. Specifically, we have removed reference to historical contingency and instead have reframed our experiments as providing a genetic and evolutionary explanation for an interesting and concerning cause of antibiotic resistance – non-carbapenemase CRE.

Author Response

Reviewer #1 (Public Review):

During the height of the Covid19-pandemic, there was great and widely spread concern about the lowered protection the screening programs within the cancer area could offer. Not only were programs halted for some periods because of a lack of staff or concern about the spreading of SARS CoV2. When screening activities were upheld, participation decreased, and follow-up of positive test results was delayed. Mariam El-Zein and coworkers have addressed this concern in the context of cervical screening in Canada, one of the rather few countries in the world with well organized, population-based, although regionalized, cervical screening program.

Comment 1: Despite the existence of screening registries, they choose to do this in form of a survey on the internet, to different professional groups within the chain of care in cervical screening and colposcopy. The reason for taking this "soft data" approach is somewhat diffuse.

We are happy to provide a counterargument to the reviewer’s concern about the “soft data” approach. Our unit – McGill’s Division of Cancer Epidemiology – is a major stakeholder in policymaking and cervical screening guideline development in Canada. It is one of the components in a McGill Task Force on COVID-19 and Cancer that has been widely engaged in assessing the pandemic’s impact on the entire spectrum of cancer control and care (examples: PMID: 33669102, PMID: 34843106). Canada is a country of continental size, and during the pandemic even travel between provinces was interrupted. It is only via a web-based survey that one could have captured the required information. We took advantage of our unit’s credibility and stature to secure a substantial response to the survey, which elicited a high level of detail.

The survey questionnaire instrument was thoughtfully developed with input from Canadian experts who are active in the field of cervical cancer prevention and involved in clinical care to comprehensively formulate informative questions (and practical, reasonable responses) underpinning each of the themes covered. Of note, some of these coinvestigators, having executive roles in relevant clinical professional bodies, advised our team on the logistics of circulating the survey to members. The administration of the survey was coordinated with the pertinent societies. Our aim was to provide an overall portrait across Canada of the extent of the harms to cervical cancer screening and treatment processes at the beginning of the COVID-19 pandemic (specifically a snapshot from mid-March to mid-August 2020), as perceived by professional groups in multiple health disciplines.

Indeed, as the reviewer mentioned, there are fully (i.e., for Saskatchewan) and partially (i.e., for British Columbia, Alberta, Manitoba, Ontario) organized cervical cancer screening programs in Canada in addition to opportunistic programs (i.e., for North West Territories, Yukon, Nunavut, Quebec). The Canadian Partnership Against Cancer also collects information on cervical cancer screening programs and/or strategies across Canada. Using data from these different sources enables a quantitative assessment of the impact of the pandemic on cervical cancer screening, but this was not the research methodology used; the survey approach was our research strategy as we attempted to collect responses from all provinces and territories, regardless of the different screening programs and modalities implemented across the country, and including regions that do not have an official screening program.

Since the effects of the COVID-19 pandemic will stay with us for years to come, our research team is also examining – using a “hard data” approach via administrative healthcare datasets – the long-term effects that will accrue on cervical cancer morbidity and mortality from the interruptions and delays in screening processes and other activities in the process of care. A discussion of this is, however, beyond the scope and objectives of our manuscript.

No modifications were made in the manuscript to address this comment.

Comment 2: The authors claim they want to "capture modifications". However, the suggestions that come from this study are limited and are submitted for publication 2 years after the survey when the height of the pandemic has passed long since, and its burden on the screening program has largely disappeared. The value of the study had been larger if either the conclusions had been communicated almost directly, or if the survey had been done later, to sum up the total effect of the pandemic on the Canadian cervical screening program.

We appreciate this comment. As part of our commitment to transparency, we now plainly acknowledge that considerable time (1.5 years) has elapsed between the time the survey data were available (March 2021) and manuscript submission (September 2022) for publication in the special issue, curated by eLife, on the impact of the COVID-19 pandemic on cancer prevention, control, care and survivorship. However, we also argue that this lag time is reasonable given the undertaking of data management, analysis, and reporting of a large amount of data, including the synthesis of replies to open-ended questions. We also took this opportunity to expose two graduate students to the research process.

Changes made: Page 15, Lines 437-440.

In terms of assessing the total effect of the pandemic on the Canadian cervical screening program, this work is in progress, but not within the current manuscript. The PubMed references mentioned above show examples of directions we are taking. Also, as mentioned in our response 1 to comment 1, we will use data from administrative healthcare datasets (medical and drug claims, hospitalization data, death registry data) and hospital cancer registries (clinical characteristics such as cancer stage, grade, and biomarkers) on cancer patients diagnosed in Quebec between 2010 and 2026. Using these datasets, we intend to compare the pre- and post-pandemic eras in order to analyze changes in patterns of cancer care, cancer prognosis, and survival, including shifts at stage at diagnosis.

Comment 3: Another major problem with this study is the coverage. The results of persistent activities to get a large uptake is somewhat depressing although this is not expressed by the authors. 510 professionals filled out the survey partially or in total. 10 professions were targeted. The authors make no attempt to assess the coverage or the validity of the sample. They state the method used does not make that possible. But the number of family practicians, colposcopists, cytotechnicians, etc. involved in the program should roughly be known and the proportion of those who answered the survey could have been calculated. My guess is that it is far below 10%.

There were no extensive additional efforts to increase participation rate, apart from follow-up reminder emails to complete the survey, which is standard practice followed by the societies that administered the survey to their constituents. We respectfully disagree with the reviewer concerning coverage being a major limitation, particularly in view of the difficulty in general to secure a high response rate in a survey such as ours, at a time like the middle of the pandemic. Although it appears to be a seemingly easy to compute classic non-response rate, information on the “population of interest” (i.e., number of professionals approached in addition to the advertisement of the survey on social media platform”) is not available to estimate the extent of non-response. Even if the response rate is below 10% as suggested by the Reviewer, our survey and findings should be considered on their merits; the target population was involved in the survey design to ensure the validity of coverage of the questions along the continuum of care in cervical cancer screening and treatment. In addition, we followed the Checklist for Reporting Results of Internet E-surveys to inform the design, conduct, and reporting of our survey research.

Changes made: Page 14, Lines 421-425.

Comment 4: The national distribution seems shewed despite the authors boosting its pan-Canadian character. I am just faintly familiar with the Canadian regions, but, as an example, only 2 replies from Quebec must question the national validity of this survey.

We apologize for this typo error in Table 1; many cells were accidently shifted down (the last couple of provinces had the wrong numbers). There were actually 21 survey respondents from the province of Quebec. This has now been corrected.

Changes made: Page 19.

Comment 5: The result section is dominated by quantitative data from the responses to the 61 questions. All questions and their answers are tabulated. As there is no way to assess the selection bias of the answers these quantitative results have no real value from an epidemiological standpoint.

Indeed, we opted to provide the reader with descriptive results on all the questions and sub-questions that were asked, with explicit annotation to each question number and clear reference to the formulated question by appending the full survey instrument to the manuscript. We designed the survey as a descriptive and not an analytical study, contrary to traditional epidemiology studies that investigate a specific exposure-outcome relationship.

Changes made: Page 12, Lines 366-368.

In the spirit of other papers in the special issue on COVID-19 and cancer, curated by eLife, we measured the impact of the pandemic on the process of care like many other eLife articles did. The eLife collection is a snapshot of a period when not only was cancer control disrupted, but the ability to conduct valid research was also severely curtailed. The reviewer will likely agree that our paper is not the only one to suffer from these methodological shortcomings. Yet, taken together, the gestalt value of the eLife collection will inform epidemiologic modellers for the next long while on how this period affected cancer control. We are happy to contribute with this paper a few more pieces of the puzzle, adding to that which eLife published for many other jurisdictions.

Comment 6: The replies to the open-ended questions are summarized in a table and in the text. The main conclusion of the content analysis of the answers to the direct questions, and one of the main conclusions of the study, is that the majority favors HPV self-sampling in light of the pandemic. However, this not-surprising view is taken by only 80 responders while almost as many (n=60) had no knowledge about HPV self-sampling.

Another aim of our survey was to identify the windows of opportunity that were created by the pandemic and pinpoint positive aspects that could enable the transformation of cervical cancer screening (i.e., HPV primary based screening and HPV self-sampling). We found that 33% of respondents were of the opinion that the pandemic context could facilitate the implementation of self-sampling and that 50.1% were in favor of the implementation of this new screening practice (described in Results Theme 1: Screening Practice and Stable 5).

Changes made: Page 4, Lines 93-97.

The reviewer is correct that in the open-ended sub-question of Question 23 “Are you in favor of the implementation of HPV self-sampling as an alternative screening method in your clinical practice?”, 60 respondents justified their answer to the nominal question by their lack of familiarity with HPV self-sampling, compared to 80 who shared positive comments. However, we would like to draw the reviewer’s attention to the responses to the nominal part of the question in Stable 5. Of those who answered “Maybe”, 47.1% said that they were not familiar enough to express a favorable or unfavorable opinion. We would also like to draw the reviewer’s attention to the results of our cross-tabulation of profession and the question of relevance (described in Results Theme 1: Screening Practice). The lack of familiarity with novel screening practices such as self-sampling can be explained by the fact that most (75.0%) of those who expressed these views were primary healthcare professionals, and not secondary and tertiary specialists.

Changes made: Page 12, Lines 344-346

Comment 7: The authors conclude that their study identified the need for recommendations and strategies and building resilience in the screening system. No one would dispute the need, but the additional weight this study adds, unfortunately, is low, from a scientific standpoint.

Although no one would dispute the need as the reviewer is suggesting, but as epidemiologists we needed to collect this empirical evidence. We urge the reviewer to consider that this article is to contribute to a more complete picture of the collective process of discovery of the impact of the pandemic initiated by eLife’s special issue.

No modifications were made in the manuscript to address this comment.

Comment 8: The conclusion I draw from this study is that the authors have done a good job in identifying some possible areas within the Canadian screening programs where the SARS-Cov2 pandemic had negative effects and received some support for that in a survey. Furthermore, they listed a few actions that could be taken to alleviate the vulnerability of the program in a future similar situation, and received limited support for that. No more, no less.

We thank the Reviewer for the positive feedback provided in the first part of the comment. As for the rest, we believe we have addressed above the reviewer’s concerns.

Reviewer #2 (Public Review):

The study aimed to provide information on the extent to which the COVID-19 pandemic impacted cervical cancer (CC) screening and treatment in 3 Canadian provinces. The survey methodology is appropriate, and the results provide detailed descriptive statistics by province and type of practice. The results support the authors' conclusions. This evidence together with data gathered from other national surveys may provide baseline data on the impact of the pandemic on CC outcomes such as late-stage diagnoses and CC treatment outcomes due to these delays.

We are flattered by the Reviewer’s overall assessment of our manuscript.

Comment: This study relies mostly on descriptive statistics and open-ended questions that provide details about what CC screening and treatment procedures were delayed. It is unclear how the reader would use the results to affect current or future practice.

As mentioned in our reply above to a similar comment raised by reviewer 1, our overarching aim was to portray in a purely descriptive manner the negative and positive impacts of the COVID-19 pandemic on cervical cancer screening-related activities, as perceived by healthcare professionals. Please refer to arguments above.

Changes made: Page 12, Lines 366-368; Page 15, Lines 437-440.

Author Response

Reviewer #1 (Public Review):

In this study, the authors set out to determine the degree to which early language experience affects neural representations of concepts. To do so, they use fMRI to measure responses to 90 words in adults who are deaf. One group of deaf adults (n=16) were native signers (and thus had early language exposure); a second group (n=21) was exposed to sign language later on. The groups were relatively well-matched in other respects. The primary finding was that the high dimensional representations of concepts in the left lateral anterior temporal lobe (ATL) differed between native and delayed signers, suggesting a role for early language experience in concept representation.

The analyses are carefully conducted and reflect a number of thoughtful choices. These include the "inverted MDS" method for constructing semantic RDMs, a normal hearing comparison group for both behavioral and fMRI data, and care taken to avoid bias in defining functional ROIs. And, comparing early and delayed signing groups is a clever way to study the role of early language experience on adult language representations.

We greatly appreciate the reviewer’s positive evaluation and constructive comments on our study.

One interesting result that I struggled to put in a broader context relates to the disconnect between behavioral and neural results. Specifically, the behavioral semantic RDMs (Figure 1a) did not differ between any of the groups of participants. This suggests that the representations of the 90 concepts are represented similarly in all of the participants. However, the similarity of the neural RDMs in left lateral ATL differs between the native and delayed signing groups (but not in other regions). Given the similarity of the behavioral semantic RDMs, it is unclear how to interpret the difference in left lateral ATL representations. In other words, the neural differences in left ATL do not affect behavior (semantic representation). The importance of the differences in neural RDMs is therefore questionable.

Thank you for this comment. In the Revision we have added explicit discussions about this important issue of the relationship between the behavioral and neural profiles for semantics:

Introduction (pages 4-5): “(previous) studies have reported little effects on semantics behaviors, including semantic interference effects in the picture-sign paradigm (Baus et al., 2008), scalar implicature (Davidson and Mayberry, 2015), or accuracy scores of several written word semantic tasks (e.g., synonym judgment) (Choubsaz and Gheitury, 2017). However, as shown by the color knowledge in the congenitally blind studies (e.g., Wang et al., 2020), similar semantic behaviors may arise from (partly) different neural representations. Semantic processing is supported by a multifaceted cognitive system and a complex neural network entailing distributed semantic regions (Bi, 2021; Binder and Desai, 2011; Lambon Ralph et al., 2017; Martin, 2016), and thus focal neural changes may not necessarily lead to semantic behavioral changes. Neurally, neurophysiological signatures assumed to reflect semantic processes showed incongruent effects across studies: N400 effects in the semantic violation of written sentences were not affected (Skotara et al., 2012), whereas M400 in the picture-sign matching task showed atypical activation patterns (reduced recruitment of left fronto-temporal regions and involvement of right parietal and occipital regions) (Ferjan Ramirez et al., 2016, 2014; Mayberry et al., 2018). It remains to be tested whether and where delayed L1 acquisition affects how semantics are neurally represented, using imaging techniques with higher spatial resolutions.”

Discussion (pages 17-18): “Notably, different from phonological and syntactic processes, where both visible behavioral underdevelopment (e.g., Caselli et al., 2021; Cheng and Mayberry, 2021; Mayberry et al., 2002) and brain functional changes (Mayberry et al., 2011; Richardson et al., 2020; Twomey et al., 2020) were observed, for semantics we only observed brain functional changes in dATL but no visible behavioral effects. Consistent with the literature where deaf delayed signers did not show differences to controls in semantic interference effects in the picture-sign paradigm (Baus et al., 2008), scalar implicature (Davidson and Mayberry, 2015), or N400 measures (Skotara et al., 2012), we did not observe visible differences in terms of semantic distance structures (Figure 1a) or reaction time of lexical decision and word-triplet semantic judgment (Supplementary file 1). As reasoned in the Introduction, this seeming neuro-behavior discrepancy might be related to the multifaceted, distributed nature of the cognitive and neural basis of semantics more broadly. The general semantic behavioral tasks we employed could be achieved with representations derived from multiple types of experiences, supported by highly distributed neural systems (e.g., (Bi, 2021; Binder and Desai, 2011; Lambon Ralph et al., 2017; Martin, 2016), including those not affected by the delayed L1 acquisition in regions beyond the dATL. This finding invites future studies to specify the exact developmental mechanisms in the left dATL (Fu et al., 2022; Unger and Fisher, 2021) and to uncover semantic behavioral consequences related to the functionality of this area.”

An important point is that, if I understand correctly, the semantic space is defined by the 90 experimental items. That is, behavioral RDMs were created by having normal hearing participants arrange 90 items spatially, and neural RDMs were created by comparing patterns of responses to these 90 experimental items. This 90-dimensional space is thus both (a) lower dimensional than many semantic space models that include hundreds of directions and (b) constrained by the specific 90 experimental items chosen. On the one hand, this seems to limit the generalizability of the findings for semantic representations more broadly.

Indeed, for the RDM the spaces were constructed by the relations among the 90 items, as is the standard practice for current RSA analyses. Regarding the dimensionality issue, we would like to clarify that although the space is a 90 x 90 matrix, the semantic distance for each pair was obtained by the subjects’ ratings, i.e., the psychological space, which is likely to be high-dimensional in nature. That is, we compressed the potentially high-dimensional psychological construct into one measure to construct the 90 x 90 matrix. If we understood correctly, semantic space models with hundreds of directions the reviewer referred to are various types of embedding and/or distributional models. There although each word is projected onto a high-dimensional vector, the distance for each pair is still extracted (e.g., by cosine similarity) to construct the cross-item similarity matrix for RSA. Regarding the generalization of the findings across items, we greatly appreciate this concern and indeed that was one of the reasons why we extracted the categorical structure based on the clustering of the items (see also response to the next Comment). We also examined the univariate abstractness contrast, which looked at the broad categorical effects rather than specific items. We have made clarifications accordingly in the Revision to address these concerns (page 8).

The logic behind using a categorical semantic RDM (e.g., Figure 2a) was not clear. The behavioral semantic RDMs (Figure 1a) clearly show gradations in dissimilarity, particularly for the abstract categories. It would seem that using the behavioral semantic RDM would capture a more accurate representation of the semantic space than the categorical one.

Thank you for this suggestion. We opted for the categorical structural similarity based on the clustering analyses to boost signal and to allow for better generalization across items (i.e., along the categorical structure). Agreeing with the reviewer that such an approach may lose the important graded space especially for the abstract items, we added an analysis using continuous semantic distances specifically focused on the abstract items (page 10):

“1) Types of semantic distance measures: While semantic categories for concrete/object words are robust and well-documented, the semantic categorization within the abstract/nonobject words is much fuzzier and remains controversial (Catricalà et al., 2014; Wang et al., 2021). The behavioral semantic RDM in Figure 1a indeed shows gradations in dissimilarity for abstract/nonobject words. We thus checked the two groups’ semantic RDMs using the continuous behavioral measures and further examined whether group differences in the left dATL were affected by the types of semantic distance (categorical vs. continuous) being used for abstract/nonobject words. The two deaf groups showed comparable similarities to the hearing benchmark (by correlating each deaf subject’s RDM with the group-averaged RDM of hearing subjects, Welch’s t23.0 = -0.12, two-tailed p = .90). RSA was performed by correlating each deaf subject’s neural RDM in the left dATL with these two types of semantic RDMs. Significant group differences were observed (Figure 3), for both the categorical RDM (Welch’s t31.0 = 3.06, two-tailed p = .005, Hedges’ g = 0.98) and the continuous behavioral semantic RDM (Welch’s t36.7 = 2.47, two-tailed p = .018, Hedges’ g = 0.76), with significant semantic encoding in dATL observed in both analyses for native signers (one-tailed ps < .003) and neither for delay signers (one-tailed ps > .42). These results indicate that the reduced dATL encoding of abstract/nonobject word meanings induced by delayed L1 acquisition was reliable across semantic distance measures.”

As the reviewer suggested, we could also carry out RSA using the 90-word behavioral semantic RDM. We did observe similar group differences with this RDM, with delayed signers showing a trend of semantic encoding reduction in the left dATL relative to native signers (native signers, mean (SD): 0.019 (0.023); delayed signers, mean (SD): 0.006 (0.022), Welch’s t31.5 = 1.78, two-tailed p = .085; a delayed signer was excluded from this analysis for being an outlier beyond 3 standard deviations). It appears that the behavioral semantic RDM yielded smaller effect sizes in group differences than the categorical RDM, but the ANOVA (the within-subject factor - RDM-type: categorical, behavioral; the between-subject factor – group: native, delayed) revealed no significant effects of RDM-type or its interaction with the group (ps > .71), but a significant main effect of group (F(1,36) = 9.19, p = .004). The seemingly weaker group differences using the behavioral semantic RDM should not be over-interpreted.

Reviewer #2 (Public Review):

The authors investigated patterns of fMRI activation for familiar words in two groups of deaf people. One "language rich" group received exposure to sign from birth, whereas the "language poor" group included kids born to hearing parents who had limited exposure to language during the first few years of life. The primary findings involved group differences in BOLD activation patterns across different areas of interest within the semantic network when participants made intermittent 1-back category judgments for words appearing in succession.

There was much to be liked about this study, including the rigor of the methods and the novel contrasts of two deaf samples. These strengths were balanced by a number of questions about the assumptions and theoretical interpretations underlying the data. I will elaborate on the major points in the paragraphs to follow, but briefly, the ways in which the authors are framing critical period constraints in language fundamentally differ from the standard nativist perspectives (e.g., Chomsky, Lenneberg). The assumptions of what constitutes a deprivation model require further justification and perhaps recasting to avoid unnecessary stigma (i.e., this reviewer was uncomfortable with the assertion that being born deaf to hearing parents by default constitutes deprivation). The introduction lacked principled hypotheses that motivated the choice of comparing abstract and concrete words, and potential accounts of group differences were underdeveloped (e.g., how do parents in China typically react to having a deaf child, and what supports are in place for preventing language deprivation? Are newborn infants universally screened for hearing loss in China? The answers to these questions might help the readers to understand why/how deaf children in this circumstance might experience deprivation).

We appreciate the reviewer’s positive evaluations and constructive comments on our study. We have revised the manuscript substantially in light of these comments (see below).

References to critical periods require a bit more elaboration with respect to lexical-semantic vs. semantic acquisition. The nature of the critical period in language acquisition remains controversial with respect to its constraints. Lenneberg and Chomsky speculated that the limit of the critical period for language acquisition was about puberty (13ish years of age). This is much older than the deaf sample tested here so arguments about aging out of the critical period at least for language acquisition need more nuance. Another issue relates to learning semantic mappings vs. learning language as falling under the same critical period umbrella. This seems highly unlikely as semantic acquisition in early childhood is aided by linguistic labeling but would likely occur in parallel even in the context of language deprivation. Much of the prior literature on critical periods and nativist approaches to language development has focused on syntactic acquisition and elements such as recursion rather than a mapping of symbols to conceptual referents. This makes the critical period group comparison somewhat tenuous because what you are really interested in is a critical period for word meaning acquisition not the more general case of syntactic competency.

The point above is highlighted in the following statement underlying one of the primary assumptions of the study:

Pg. 3, "Here, we take advantage of a special early-life language-deprivation human model: individuals who were born profoundly deaf in hearing families and thus had very limited natural language exposure (speech or sign) during the critical period of language acquisition in early childhood"

"hypofunction of the language system as a result of missing the critical period of language acquisition" (pg 3), same critique as previous - the critical period window is thought to be 13ish years old.

There are a couple of problems with this assertion/assumption. Although it is true that most children who are born deaf have hearing parents, it is not justifiable to label this condition an early-life deprivation model. Hearing parents who are extremely motivated to learn sign language and pursue related language enrichment strategies can successfully offset many of these effects. Similarly, it is not inconceivable that a deaf child born to a deaf parent might be neglected or abandoned without the benefit of early sign exposure. My argument here is that classifying deaf children born to hearing parents as automatically 'language deprived' is potentially both stigmatizing and scientifically unjustified.

We originally used the term “language deprivation” because it has been recently advocated in the deaf field mainly to increase society’s awareness of the risks of language deprivation and the lifelong impact that deaf and hard-of-hearing children face (e.g., Hall, 2017, Maternal and Child Health Journal; Lillo-Martin & Henner, 2020, Annual Review of Linguistics). In the current context, we agree with the reviewer that “early-life deprivation” model may not precisely describe the language acquisition condition of delayed signers. Indeed, for some of the delayed subjects in our study, their hearing parents actively tried to provide additional aids of exposure to signs (via preschool special education programs; learning signs by themselves) or speech (via hearing aids). In the revision, we avoided the term “language deprivation” and used the terms “subjects with varying amounts and qualities of early language exposure” or “delayed L1 acquisition” to more precisely describe our experimental manipulation throughout the revised manuscript.

We fully agree with the reviewer that the “critical period” of language acquisition is too much an umbrella term, which may be taken to refer to critical period for different, specific cognitive and/or neural development in the literature. In the Revision we avoided using this term to reduce ambiguity. Instead, we now made explicit throughout the specific processes being discussed (phonology, syntax, semantics). The effects of early language experience (reduced in delayed L1 acquisition) on the behavioral and neural patterns relating to phonology, syntax, and semantics are now elaborated, discussed separately and explicitly in both the Introduction and Discussion (pages 3-4, 17-18).

Regarding the potential nonlinguistic socio-environmental differences (e.g., coping strategies after deafness awareness), we have added further clarifications (page 15): “Notably, routine nation-wide neonate hearing screening in China did not start until 2009, years after the early childhood of our participants (born before 2000), and some hearing parents may nonetheless try to give deaf children additional aids of exposure to signs (via preschool special education programs) or speech (via hearing aids). Critically, our positive results of the robust group differences in dATL suggest that early homesign/aid measures and later formal education for sign and written language experiences are insufficient for typical dATL neurodevelopment; the full-fledged language experience during early infancy and childhood (before school age) plays a necessary role in this process.” Relevant information has also been added in the Method/Result sections.

Pg. 6 "It should be noted that the neural semantic abstractness effect does not equate with language-derived semantic knowledge, as it might arise from some nonverbal cognitive processes that are more engaged in abstract word processing (Binder et al., 2016)." - I had great difficulty understanding what this meant.

We have revised this sentence as follows: “While the abstractness effect has often been used to reflect linguistic processes (e.g., (Wang et al., 2010)), “abstractness” is not a single dimension and instead relates to both linguistic and nonlinguistic (e.g., emotion) cognitive processes (Binder et al., 2016; Troche et al., 2014; Wang et al., 2018).” (page 11)

Author Response

Reviewer #1 (Public Review):

In this paper, the authors present a method for discovering response properties of neurons, which often have complex relationships with other experimentally measured variables, like stimuli and animal behaviors. To find these relationships, the authors fit neural data with artificial neural networks, which are chosen to have an architecture that is tractable and interpretable. To interpret the results, they examine the first- and second-order approximations of the fitted artificial neural network models. They apply their method profitably to two datasets.

The strength of this paper is in the problem it is attempting to solve: it is important for the field to develop more useful ways to analyze and understand the massive neural datasets collected with modern imaging techniques.

The weaknesses of this paper lie in its claims (1) to be model free and (2) to distinguish the method from prior methods for systems identification, including spike triggered averaging and covariance (or rather their continuous response equivalents). On the first claim, the systems identification methods are arguably substantially more model free approach. On the second claim, this reviewer would require more evidence that the presented approach is substantially different from or an improvement on systems identification methods in common use applied directly to the data.

We thank the reviewer for carefully engaging with the manuscript and believe that our revisions address these points of critique both through novel analysis and through clarifications.

First claim: We fully agree that systems identification approaches are in theory truly model-free while MINE imposes constraints through the chosen architecture. However, our new analysis comparing MINE to direct fitting of the kernels of a Volterra expansion highlights that this is not really the case in practice. In order to obtain good fits, the model-free-ness has to be substantially reduced by imposing constraints on the degrees of freedom. We quantify this reduction in Figure S3 and directly compare it to the effective degrees of freedom of the CNN. Reducing degrees of freedom is also a theme that can be found throughout the literature on systems-identification, especially when the analysis does not involve Gaussian white noise as input stimuli. We therefore stand by our claim that MINE is “essentially model-free” in the sense that it does not rely on defining a model a-priori much like systems identification. And we also clarify our choice of calling the method “model-free” in the introduction where we state: “While the architecture and hyper-parameters of the CNN used by MINE do impose constraints on which relationships can be modeled, we consider the convolutional network ``model-free’’ because it does not make any explicit assumptions about the underlying probability distributions or functional forms of the data.”

Second claim: We believe that our new analysis for the comparison with the Volterra expansion approach of systems identification addresses this point. By directly fitting Volterra kernels instead of relying on spike-triggered analysis we put the comparison on a more equal footing than our previous STA/STC exposition. We can show that while the methods are equivalent for Gaussian white noise stimuli, MINE is superior for highly correlated input stimuli. We show that imposing constraints on the regression used to identify the Volterra kernels can overcome this gap to a large extent, but MINE still produces a model that has higher predictive power and MINE also does more than extracting receptive fields. We are also not entirely sure to what extent Wiener/Volterra analysis has been applied to calcium imaging data. While there is a vast body of literature on systems identification, there is little evidence that it has been widely applied to data in which both inputs and outputs are highly correlated across time, such as calcium imaging experiments using naturalistic stimuli. While this doesn’t have to mean anything in and of itself it might point to the fact that this analysis is not easily accessible and requires ample tuning. These are precisely two problems that MINE aims to overcome. We now more explicitly state in the manuscript that we believe this accessibility to be one of the core strengths of MINE.

Reviewer #2 (Public Review):

This paper describes a relatively unbiased and sensitive method for identifying the contributions of different behavioral parameters to neural activity. Their approach addresses, in an elegant way, several difficulties that arise in modeling of neuronal responses in population imaging data, namely variations in temporal filtering and latency, the effects of calcium indicator kinetics, interactions between different variables, and non-linear computations. Typical approaches to solving these problems require the introduction of prior knowledge or assumptions that bias the output, or involve a trade-off between model complexity and interpretability. The authors fit individual neuron's responses using neural network models that allow for complex non-linear relationships between behavioral variables and outputs, but combine this with analysis, based on Taylor series approximations of the network function, that gives insight into how different variables are contributing to the model.

The authors have thoroughly validated their method using simulated data as well as showing its applicability to example state of the art data sets from mouse and zebrafish. They provide evidence that it can outperform current approaches based on linear regression for the identification of neurons carrying behaviorally relevant signals. They also demonstrate use cases showing how their approach can be used to classify neurons based on computational features. They have provided Python code for the implementation and have explained the methods well, so it will be easy for other groups to replicate their work. The method could be applied productively to many types of experiments in behavioral and systems neuroscience across different model systems. Overall, the paper is clearly written and the experiments are well designed and analysed, and represent a useful contribution to the neuroscience field.

We thank the reviewer for their favorable assessment of our work.

Reviewer #3 (Public Review):

In the current study, the authors present a novel and original approach (termed MINE) to analyze neuronal recordings in terms of task features. The method proposed combines the interpretability of regressor-based methods with the flexibility of convolutional neural networks and the aim is to provide an unbiased, "model-free" approach to this very important problem.

In my opinion, the authors succeed in most of these aspects. They use three datasets: an artificially-generated one that provides a ground-truth, a published dataset from wide-scale cortical mouse recordings and a novel one that studies thermosensation in larval zebrafish. MINE compares favorably in all three cases.

I believe that the paper would mostly benefit from an increased effort in clear exposition of the Taylor expansion approach, which is at the core of the method. The methods section describes the mathematics, but I wonder whether it would be possible to illustrate or schematize this in a main Figure, e.g. as an addition to Figure 1 or as a new figure. Around line 185, the manuscript reads: "We therefore perform local Taylor expansions of the network at different experimental timepoints. In other words, we differentiate the network's learned transfer function that transforms predictors into neural activity."

It would help to explicitly state with respect to what the derivative is being computed (i.e. time) and maybe a diagram (which I had to draw to understand the paper) in which a neuronal activity trace is shown and from time t onwards a prediction is computed using terms in the Taylor expansion would be very instructive (showing on an actual trace how disregarding certain terms changes the prediction and hence the conclusions about the actual dependence of the trace on the behavioral features). The formulation in terms of Jacobians and Hessians can then be restricted to the Methods section and the paper will be easier to read for a wider audience.

We agree with the reviewer that readability is key. We hope that our re-write and re-organization of the manuscript makes it easier to follow. We now start with a unified description of complexity and non-linearity both derived from a Taylor decomposition around the data-average. We use this section (starting Line 91) to lay out the logic of the Taylor expansion and explicitly state that the derivatives describe the expected change in output given any change in predictors. We did not want to remove the math entirely from the paper, simply because we found it hard to explain the concept entirely without it. We have provided an annotation to the formula parts in the new Figure 2 and a small schematic to illustrate the pointwise expansion of the Taylor metric in the new Figure 4.

The method is presented as a "model-free" approach (title and introduction). I think it would help to discuss this with some precision. The Taylor expansion approach does imply certain beliefs on the structure of the data (which are well founded in most cases). Do the authors agree that MINE would encapsulate any regression model where both linear and interaction terms are allowed to include an arbitrary non-linearity (in the case of the interaction terms, different non-linearities for both variables)? If this is the case, maybe an explicit statement would allow the reader to quickly identify the versatility of MINE.

We are now attempting to make the statement of model-free more precise through quantifications in our rewritten section on deriving receptive fields. We now provide an explanation in the introduction for why we believe that “model-free” is justified. We state: “While the architecture and hyper-parameters of the CNN used by MINE do impose constraints on which relationships can be modeled, we consider the convolutional network ``model-free’’ because it does not make any explicit assumptions about the underlying probability distributions or functional forms of the data.”

In principle, MINE can accommodate higher-order interactions as well (say of the form xyz or x*y^2) and it certainly has flexibility in applying nonlinear transformations. However, we did not find a satisfying way to quantify the space of possible models MINE can represent exactly and therefore do not feel comfortable to make a precise statement about this.

I find the section relating to non-linearities interesting, but was slightly disappointed to find that the authors do not propose a single method. In Figure 3E, the authors show that a logistic regression model that combines the curvature and NLC apporaches outperforms either, but the model is not described in any sort of detail. I appreciate the attempt made by the authors to apply this to the zebrafish imaging dataset in Figure 7, but it was still unclear to me how non-linearities and complexity are related.

We fully agree with the reviewer. We have now merged non-linearity and complexity determination. We hope that this a) simplifies the paper and b) creates a metric that likely generalizes better and in which specific values are more interpretable. In brief, we now define both the nonlinearity and complexity based on truncations of the Taylor expansion around the data average. This new result section (Lines 90-142) also gives us a chance to (hopefully) better introduce the Taylor expansion approach.

Author Response

Reviewer #1 (Public Review):

Li et al investigated the behavioral response and fMRI activations associated with deep brain stimulation (DBS) of the lateral habenula (LHb) in 2 distinct rodent models of depression. They found that a) LHb DBS reduces depressive and anxiety behaviors using multiple behavioral tests: sucrose preference, forced swim, and open field. These results held across multiple models of depression and multiple tests, and generally restored results of these behavioral tests to parity with controls. Furthermore, fMRI activations of brain regions with known connectivity to LHb strongly correlated with behavioral responses to LHb DBS, particularly in limbic regions. These behavioral responses clearly depended on electrode location, with more medial placements within the LHb producing a more robust behavioral effect.

The conclusions of this paper are generally well supported by the data, with the primary weaknesses of the study being 1) limited novelty due to LHb already being a well-established target for DBS in depression, and 2) the questionable validity of rodent models of depression in general. The authors deal with the first point (novelty) by extending their study to electrode localization and fMRI correlates with the behavioral response, leading to insight into surgical targeting as well as mechanism of effect, respectively. They also partially mitigate fundamental problems with rodent models of depression by using 2 different models and showing consistent responses to LHb DBS across both. The methods used in this study were sound, with high-quality techniques used for electrode implantation, confirmation of electrode placement, fMRI acquisition, anesthesia and physiological monitoring, as well as an appropriate statistical analytic approach.

We thank the reviewer deeply for the positive assessment on our work.

Reviewer #2 (Public Review):

This important paper is a real tour de force and combines functional MRI, behaviour, and brain stimulation to characterise the effect of stimulation of the lateral habenula in a rodent model for depression. The results are stunning and the data presented seems compelling.

My only comment is I would like more discussion on the relevance of these results for the treatment of depression in humans, both in terms of the rodent model and in terms of the results shown in this study.

We thank the reviewer deeply for the positive assessment on our work. We have added discussion on the relevance of our finding for the treatment of depression in humans on Page 17 of the revised manuscript as follows:

“The WKY and LPS-treated depressive rat models share similar characteristics, including abnormalities in various neurotransmitter and endocrine systems and emotional changes resulting from inflammatory stimuli. These models are widely used in pharmacological and nonpharmacological depression treatment studies(Caldarone et al., 2015; Aleksandrova et al., 2019; Lasselin et al., 2020). Previous research indicates that classic antidepressants used in humans, such as selective serotonin reuptake inhibitors, also cause an antidepressant reaction in WKY rats. Ketamine, a rapid-acting antidepressant in clinical practice, has been shown to be effective in both WKY and LPS-treated rats(Aleksandrova et al., 2019; J. Zhao et al., 2020). In WKY rats, DBS of the NAc increased exploratory activity and exerted anxiolytic effects, and NAc-DBS was found to be effective for TRD treatment in humans(Dandekar et al., 2018; Aleksandrova et al., 2019). These results suggest that the depression rat models can provide valuable information about the efficacy of various pharmacological and nonpharmacological therapies. In a recent case report, researchers observed acute stimulation effects in addition to long-term clinical improvements in depression, anxiety, and sleep in a patient with TRD upon administering LHb-DBS (Wang et al., 2020). This finding supports the clinical relevance of our observations. However, no animal model of depression can completely replicate human symptoms, and further research is necessary to validate our findings in human patients. Additionally, the long-term efficacy and side effects of LHb-DBS require further investigation. Nevertheless, we believe that our findings propose a promising addition to the rapid-acting therapeutic options for the most refractory depression patients.”

Authorr Response

Reviewer #2 (Public Review):

This manuscript is clear in that it shows no/minimal weight gain in a mouse model of trisomy 21 compared to the control mouse, even under a high-calorie diet. The difference is the clear demonstration of the increased expression of sarcolipin. It is important that the expression of SERCA was also shown not different between the genotypes. Additionally, an important result is that manipulating the skeletal muscle was sufficient to promote weight loss without the need for hypermetabolism in other tissues such as adipose tissue.

• A clear explanation of why the expression of sarcolipin/hypermetabolism is different between mouse and human under the same condition would be useful.

Overexpression of sarcolipin is only seen in this particular mouse model carrying the near complete human chromosome 21. In another widely used mouse model (Ts65Dn) of Down syndrome where all the triplicated genes (~40% of the human Chr21 orthologs) are of mouse origin, we did not observe the same overexpression of sarcolipin (PMID: 36587842). The reason for this is presently unknown. Human Chr21 contains a significant number of non-coding human genes (>400) with uncertain effects on the mouse transcriptome. Data in Figure 8 represents our efforts to understand what drives the overexpression of mouse sarcolipin (Sln) gene expression in the TcMAC21 mouse model. Although we narrowed it down and highlighted some potential candidate transcriptional drivers for Sln overexpression (Fig. 8), future work is clearly needed to confirm and establish if any of those candidates are the or one of the bona fide driver(s).

• p.12-13 and15. The language around 'futile' cycling is not correct because Ca movement through the sarcoplasmic reticulum of the resting fiber is essential to the function of the muscle. Firstly, the cycle of Ca through the SR is through the ryanodine receptor (RyR) as well as due to slippage through the SERCA (PMID: 11306667, PMID: 35311921). This is not made clear anywhere in the manuscript. Ca leak out of the SR through RyR is an essential component to the control/setting of the resting cytoplasmic [Ca2+] via the activation of store-operated Ca2+ entry, which is in a balance with the activation of the PMCA on the t-system membrane (PMID: 35218018). The SERCA resequesters the leaked Ca2+ from the SR. It is not possible that the resting [Ca2+] is set by the reduced efficiency of the SERCA, as indicated in the ms (PMID: 20709761). It is expected that the mito [Ca2+] steady state is set by the raised resting cyto [Ca2+] (PMID: 20709761). Ca2+ transients during EC coupling will promote transient increases in mito Ca2+ (PMID: 21795684, PMID: 36121378), but not steady-state increases. Some of these problems are highlighted by the errors in the diagram Fig 5D: please change/correct (i) the invagination of the sarcolemma is called the t-system; (ii) the cycle of Ca leak through the SR starts with RyR Ca leak, where the Ca is resequestered by the SERCA, in addition to Ca slippage through the pump. Draw a RyR opposite the t-system on the SR terminal cisternae. The heat generated by SERCA is absorbed in the cytoplasm, metabolites enter the mito and the OxPhos generates heat (PMID: 31346851). (iii) Ca does not enter mito because it cannot get into the SR (the resting cyto Ca is controlled by the t-system/plasma membrane, PMID: 20709761, PMID: 35218018). Please redraw.

We have redrawn Fig. 6D diagram as suggested by the reviewer. We have also clarified the information as presented in revised Fig. 6D in the text and figure legend. Heat is generated by mitochondrial oxidative activity. In addition, ATP hydrolysis by the Ca2+ ATPase (SERCA pump) also generates heat (PMID: 12512777; PMID: 34826239; PMID: 11342561; PMID: 17018526; PMID: 12887329). In resting muscle, for every ATP hydrolyzed by the SERCA pump, 2 Ca2+ molecules get transported into the sarcoplasmic reticulum (SR) (PMID: 15189143). In the presence of sarcolipin (SLN), a higher number of ATP needs to be hydrolyzed to move the same number of Ca2+ molecules into the SR, due to Ca2+ slippage (PMID: 34826239; PMID: 23341466). In essence, ATP hydrolysis and Ca2+ transport into the SR by SERCA becomes uncoupled in the presence of SLN. This uncoupling of the SERCA pump, in the context of Ca2+ cycling in and out of the SR (also involving Ryr1), represents the ATP-consuming futile cycle in the skeletal muscle (PMID: 34741717). Since SLN is persistently overexpressed, the ATP-consuming futile activity of the SERCA pump is presumably happening in resting muscle, as well as during EC coupling (since the TcMAC21 mice are also hyperactive).

• The changing of the properties of the muscle towards oxidative properties is consistent with the expression of sarcolipin in mouse muscle (all of it is in type II fibers). It is important to show whether the muscles have fiber-type shifts. Please report the fiber types of the muscles that have been surveyed in this project.

In the qPCR data as shown in Figure 6C, we have profiled many genes associated with slow- and fast-twitched muscle fibers in gastrocnemius, and little if any changes were noted. At least at the level of the transcript, there is no indication of fiber type switching in gastrocnemius muscle. However, we did not perform the same qPCR analyses for all the other muscle types isolated (i.e., EDL, quadriceps, plantaris, soleus, and tongue). The main reason for this is that we had used all of these muscle tissues in our respirometry analysis as shown in Figure 6O-Q and Figure 6-Figure Supplement 4-9. Unfortunately, we did not have any leftover muscle tissues to profile muscle fiber types.

• Non-shivering thermogenesis (NST) is mentioned in this manuscript as the means of hypermetabolism, as has the lengthened duration of the cyto Ca transients during EC coupling. It is not clear at all what the contribution of NST compared to the increased work of the SERCA to clear released Ca from the cyto to the hypermetabolism. What are the relative proportions? If sarcolipin is largely for NST, then hypermetabolism is about the resting muscle.

In our view, the hypermetabolism we observed in the TcMAC21 mice is primarily due to SLN-mediated uncoupling of the SERCA pump. Chronic effects of SLN overexpression elevates ATP consumption by the SERCA pump and drives the catabolic process (i.e., increased mitochondrial OXPHOS) to generate the ATP needed to meet the demand created by the persistent uncoupling of the SERCA pump. However, the TcMAC21 mice are also hyperactive, and this can also contribute to increased metabolic rate. Since the mice are both hyperactive and hypermetabolic, we do not know the relative contribution of each to the overall phenotype of the mice.

• The link that SLN is causing more ATP use at the pump but the heat generated by OxPhos in mito is important and should be made, see Barclays' work (eg. PMID: 31346851). A direct link between the SERCA function and mito function is occurring but I currently don't see one being made in the ms. This could be made clear in Fig 5D diagram.

We have modified and clarified Figure 6D as suggested.

p.22. "The reprogramming of glycolytic...elevated Ca transients...". The language is wrong here. Oxidative fibers do not have elevated Ca transients compared to glycolytic. The amplitude of Ca release is greater in glycolytic and the duration of the transient is longer in the oxidative (eg. PMID: 12813151).

We have corrected this in the text and added the citation.

• p.22. "as less calcium is being transported into the SR due to uncoupling of the SERCA pumps". The same amount of Ca is being transported, just at the expense of more ATP than would be the case in the absence of SLN. Otherwise, the SR Ca2+ content would not be at a steady state while the SR continuously leaks Ca2+.

We have corrected this in the revised text. The incorrect statement has been deleted.

• p.23. Tavi & Westerblad (PMID: 21911615) show how Ca transient amplitude and frequency signal in slow and fast twitch fibres. Here, we are not concerned with what is happening in myotubes, where the SR is less developed than in adult fibres.

We did not use any myotubes in the present study. The myotube was mentioned in the context of discussing a published work (PMID: 30208317).

Reviewer #3 (Public Review):

Sarver et al., propose that TcMAC21 mice are hypermetabolic and that this is the cause of their reduced weight. Unfortunately, the developmental defects of TcMAC21 mice make this a challenging question to definitively answer. The authors claim that TcMAC21 mice are hypermetabolic due to a futile calcium cycling in skeletal muscle, which is caused by up-regulation of SLN. However, all of the data that would go into the energy balance equation (food intake, energy absorption, and energy expenditure) have been improperly analyzed. TcMAC21 pups are 8.5 g lighter than euploid littermates. The body weight data and images in Fig. 3A indicate that TcMAC21 mice runted. This difference is primarily a result of lower lean mass (FIG. 2B). This is important as it sets up many concerns that need to be addressed. Specific comments are noted below.

There is no overt developmental defect in the TcMAC21 mice as their birth weight are not different from the euploid controls (PMID: 32597754). A “runted” mouse is considered very small, poorly developed, and less competitive (PMID: 22822473). The lean phenotype of TcMAC21 mice is due to their hypermetabolism and not the result of developmental defects. The absolute lean mass of TcMAC21 mice is lower than the euploid controls. This is to be expected. A human being that weighs 150 pounds will have less lean mass compared to another person weighing 250 pounds. Lean mass scales with body weight. This does not mean that there is a muscle deficit in the person weighing 150 pounds. That is the reason why the lean mass is also generally presented as % lean mass (after normalizing to body weight). This normalization can tell us whether the amount of lean mass is appropriate (or normal) for a given weight. The % lean mass is either not different between TcMAC21 or euploid mice fed a control chow (Fig. 2B) or significantly higher in TcMAC21 mice fed a high-fat diet (Fig. 3B). This tell us that there is no developmental deficit in the skeletal muscle (biggest contributor to lean mass) of TcMAC21. The amount of lean mass seen in TcMAC21 mice scale appropriately with their lower body weight. Our food intake and energy absorption data were correctly done and analyzed (addressed below). In fact, TcMAC21 mice have the same or slighter higher food intake (absolute amount without normalization) despite weighing much less than the euploid controls (Fig. 2C and Fig. 3A, and Supplementary File 2 and Supplementary File 5). A sick or runted mouse generally consumes much less food and are physically much less active. The TcMAC21 mice are actually hyperactive (Fig. 2D-F and Fig. 4D-F). All our data argue against the notion of “runting” or “developmental defects” in TcMAC21 mice, and instead support our conclusion that TcMAC21 mice are lean due to elevated activity and hypermetabolism.

Specific comments:

1) It is incorrect to normalize EE to lean mass if this parameter is different between groups. Normalizing the EE data to lean mass makes it appear as though TcMAC21 mice exhibited increased EE when in fact this is a mathematical artefact. EE data should simply be plotted as ml/h (or kcal/h) per mouse. Alternatively, ANCOVA can be applied using lean mass as a covariate. Excellent reviews on this topic have been written (PMID: 20103710; PMID: 22205519).

Energy expenditure (EE) data should not be plotted as kcal/h per mouse, as indicated in the review article that the reviewer alluded to (PMID: 22205519). It is a given that EE increases as a function of body weight, as larger body mass requires greater energy to maintain. Plotting EE data per mouse (i.e., kcal/h) would lead to the erroneous conclusion that a fat mouse would have a higher EE compared to a lean mouse. Because lean mass is metabolically much more active than fat mass, normalizing EE data to lean mass is an acceptable way to plot EE data, although not ideal, as indicated by the review article the reviewer alluded to (PMID: 20103710). Often times, normalizing EE to lean mass gives similar results as the ANCOVA, as pointed out by the authors (PMID: 22205519). However, both review articles recommend ANCOVA (using body mass as a covariant of EE) as the preferred method to plot and evaluate EE data. Alongside the EE data (normalized to lean mass), we have now also included the ANCOVA data (Fig. 2D-F and Fig. 4D-F) where we used body weight as a covariate as recommended (PMID: 22205519). The results clearly indicate that the TcMAC21 mice have significantly higher EE compared to the euploid controls.

2) It makes no sense to normalize food intake to weight, as it makes no sense to divide metabolic rate by weight as well (see above). If food intake is not normalized, this will clearly show that TcMAC21 mice eat much less than controls, and if plotted as cumulative food intake will show that TcMAC21 are smaller and gain less weight on a high-fat diet because they simply eat less. This further indicates that the major tenet of this paper is not correct.

It is expected that a smaller mouse will eat less food compared to a bigger mouse. Normalizing food intake to body weight can tell you whether the amount of food intake is appropriate (or normal) for a given weight. Amazingly, despite a much lower body weight, ad libitum fed TcMAC21 mice consumed the same or a slightly higher absolute amount of food, without normalizing the data to body weight (Fig. 2C and Fig. 4A and Supplementary File 2 for the chow-fed group and Supplementary File 5 for the HFD-fed group). In fact, the absolute food intake (without normalization) in the refeeding period, after a fast, was significantly higher in the TcMAC21 mice relative to euploid controls (17.7 ± 0.082 vs. 13 ±0.87 kcal, P = 0.002; Supplementary File 5). Thus, relative to their body weight, ad libitum fed TcMAC21 consumed a significantly higher amount of calories (Fig. 2C and Fig. 4A). For transparency, we chose to show side-by-side both the absolute and relative food intake data. These results, along with the rest of the data, provide compelling evidence that hypermetabolism, and not reduced food intake, underlies the lean phenotype of the TcMAC21 mice.

3) The authors have tried to address the smaller weight of TcMAC21 mice by including weight-matched wild-type mice. However, they only focus on analyzing surface temperature, which is not an indicator of thermogenesis. Moreover, there is no information on whether these weight-matched wild-type mice are similar in age or body composition to the TcMAC21 mice. Nevertheless, the increased surface temperature can also indicate increased heat conservation, which is opposite to thermogenesis. It would make sense that TcMAC21 mice with massive reductions in lean mass would activate compensatory mechanisms of heat conservation to offset increased heat dissipation to the environment. This does seem to be the case, based on the data shown in Fig. 6D (see below).

Skin temperature has been widely and extensively used a proxy for thermogenesis, often in association with thermogenesis of brown adipose tissue (BAT), which is located just deep to the skin over the shoulder blades of the mouse. Mice fed a high-fat diet lose the “brownness” of their brown adipose tissue as excessive circulating lipid is stored in this depot. This is a well-known phenomenon. One can see this clearly in Figure 4K where the euploid BAT has accumulated a significant amount of lipid while the TcMAC21 BAT has not. The addition of weight-matched mice was solely to help indicate whether or not the BAT was a major contributor to the TcMAC21 hypermetabolic phenotype.

We did not conduct body composition analysis on the weight-matched mice. With a body weight of less than 30 grams, these wild-type mice represent a similarly lean and healthy adult mouse. They are not age-matched (the control mice are younger) because this is not possible. A wild-type mouse of the same age of TcMAC21 (already on high-fat diet for 12 weeks or longer) will weigh significantly more than the TcMAC21, just as the age-matched euploid littermates weighed significantly more than the TcMAC21 mice.

The idea of heat conservation is possible, but our data clearly indicate the TcMAC21 mice have elevated thermogenesis. The supporting data include: 1) increased deep colonic temperature; 2) activation of oxidative and thermogenic gene program in skeletal muscle; 3) overexpression of sarcolipin in the skeletal muscle, leading to futile SERCA pump activity and heat generation; 4) Increased skeletal muscle mitochondrial respiration; 5) elevated T3 levels; 6) increased physical activity level; 7) increased energy expenditure (EE normalized to lean mass or ANCOVA using body weight as a covariate). Taken together, these data provide compelling evidence to support our conclusion that the TcMAC21 mice are indeed hypermetabolic and have elevated thermogenesis.

4) A more optimal method of testing whether increased heat dissipation plays a role in the EE of TcMAC21 mice, is to measure EE at thermoneutrality, where energy dissipation to the environment will be minimized. Here the authors have attempted this in Fig. 6D. Unfortunately, the authors normalized EE to lean mass, artefactually elevating TcMAC21 EE. Despite this mistake, it now looks as though the large differences in EE that were seen at room temp have been attenuated, and only significantly limited to the dark phase. This indicates that in addition to the normalization artefact, higher heat dissipation from smaller TcMAC21 mice may also contribute to the elevated EE at 22C.

It is well known that at thermoneutrality mouse will markedly reduce their EE. Therefore, it is not surprising that the TcMAC21 mice, housed at thermoneutrality, will have lower EE compared to the TcMAC21 mice housed at room temperature. This also holds true for the euploid controls. This is to be expected. Yet, remarkably, the TcMAC21 mice still have significantly higher EE compared to the euploid controls when housed at thermoneutrality. The TcMAC21 mice never reduce their EE to the level of the euploid controls. We have now included the ANCOVA data for EE using body weight as a covariate as recommended (PMID: 22205519) (Fig. 7F). The results clearly indicate that the TcMAC21 mice have significantly higher EE compared to euploid controls even at thermoneutrality. The data obtained at thermoneutrality, as well as the body weight-matched control experiment as shown in Figure 4I, argue against heat dissipation as the driver of increased EE. Instead, our data support hyperactivity and hypermetabolism as the driver of increased EE.

5) In Fig. 6D, why is the hourly plot not shown here (like 2D and 4C)? The data clearly are not as striking as the EE data at 22C?

Because of space limitation in Figure 7, we did not include the hourly tracing data and instead showed the overall energy expenditure (EE) during the light and dark cycle as bar graphs. Per reviewer request, we have now included the hourly tracing data in Fig. 7F, along with the ANCOVA data. The data clearly indicates that TcMAC21 mice, housed at thermoneutrality, have higher EE, especially in the dark cycle when they are active. This is quite remarkable. We know from many published studies that mice significantly reduce their EE when house at thermoneutrality. And yet, the TcMAC21 mice never reduce their EE to the level of euploid controls when housed at thermoneutrality.

6) GTT was similar between TcMAC21 and controls (Fig. 3I). However, the smaller insulin response could be due to the fact that glucose was normalized to body weight. It would be better to normalize to lean mass, since that is different as well, or simply give all mice the same amount of glucose that the control group receives since this is how it is done in humans.

The dose of glucose injection in GTT based on mouse weight is widely and extensively practiced across the metabolic community. The TcMAC21 mice are markedly more insulin sensitive, supported by multiple independent lines of evidence: 1) Overnight fasting blood glucose and insulin levels are significantly lower in TcMAC21 mice relative to euploid controls (Figure 3G). 2) Insulin tolerance test clearly indicate a substantial improvement in insulin sensitivity in TcMAC21 mice even though the insulin dose injected was much smaller (i.e., insulin dose was based on body weight) (Figure 3K). 3) The insulin response during refeeding, after an overnight fast, is dramatically lower even though the refeeding blood glucose levels rise to the same levels as the euploid controls (Fig. 3L-M). This is similar to the GTT data where the rate of glucose clearance in TcMAC21 mice is the same as the euploid controls despite a dramatically lower insulin response (Fig. 3I-J). Taken together, these data clearly indicate a markedly heightened insulin sensitivity in TcMAC21 mice relative to euploid controls.

7) The fecal energy in Fig. 4B only measures the concentration of energy per gram of feces. However, this analysis has failed to take into account total fecal excretion, which should be used to multiply the energy density of the feces. Thus, these data are incomplete and not sufficient to exclude absorption differences between the groups. And it is now curious why if all other metabolic measurements (even though wrong), such as food intake and EE are normalized to body weight, why have the authors not normalized to body weight for the feces data? Is this because if this was done this would show massive elevating in fecal energy in TcMAC21 mice and thus falsify their hypothesis?

The fecal data the reviewer requested was originally in the supplemental figure section. We have now moved these data to the main figure to ensure that this will not be missed by any reader. As indicated in the text and in Fig. 4B, TcMAC21 mice fed a HFD show no difference in fecal frequency (movements/day), fecal weight (g/movement), fecal energy composition (cal/g) and total fecal energy (kcal/day). These data clearly indicate that the fecal energy content is not different between TcMAC21 and euploid mice. These results, along with the rest of the data in the paper, provide compelling evidence that hypermetabolism, and not reduced nutrient absorption in the gut, underlies the lean phenotype and resistance of TcMAC21 mice to weight gain when fed a high-fat diet.

8) I cannot find any indication of sample size in any of the EE experiments, aside from the bar graph in Fig. 6D. In any case, this experiment only an n=4 to 5 per group. This is an extremely small number for these types of experiments, so how can the authors be sure of reproducibility with such a low sample size? Are all of the other EE experiments also of similarly small sample sizes?

Sample size for all EE experiments were clearly indicated in the original text, figure legends, and figures themselves, as well as in all supplemental figures and Supplementary files. In addition, for transparency, we always include individual data points, whenever possible, for all our data figures. They were sufficiently powered (n = 8-9 per genotype) and the effect size was large. Sample size for all thermoneutral experiments were lower than both the chow-fed and HFD-fed experiments because these mice are hard to breed and in limited supply.

Author Response

Reviewer #1 (Public Review):

How morphogens spread within tissues remains an important question in developmental biology. Here the authors revisit the role of glypicans in the formation of the Dpp gradient in wing imaginal discs of Drosophila. They first use sophisticated genome engineering to demonstrate that the two glypicans of Drosophila are not equivalent despite being redundant for viability. They show that Dally is the relevant glypican for Dpp gradient formation. They then provide genetic evidence that, surprisingly, the core domain of Dally suffices to trap Dpp at the cell surface (suggesting a minor role for GAGs). They conclude with a model that Dally modulates the range of Dpp signaling by interfering with Dpp's degradation by Tkv. These are important conclusions, but more independent (biochemical/cell biological) evidence is needed.

As indicated above, the genetic evidence for the predominant role of Dally in Dpp protein/signalling gradient formation is strong. In passing, the authors could discuss why overexpressed Dlp has a negative effect on signaling, especially in the anterior compartment. The authors then move on to determine the role of GAG (=HS) chains of Dally. They find that in an overexpression assay, Dally lacking GAGs traps Dpp at the cell surface and, counterintuitively, suppresses signaling (fig 4 C, F). Both findings are unexpected and therefore require further validation and clarification, as outlined in a and b below.

a) In loss of function experiments (dallyDeltaHS replacing endogenous dally), Dpp protein is markedly reduced (fig 4R), as much as in the KO (panel Q), suggesting that GAG chains do contribute to trapping Dpp at the cell surface. This is all the more significant that, according to the overexpression essays, DallyDeltaHS seems more stable than WT Dally (by the way, this difference should also be assessed in the knock-ins, which is possible since they are YFP-tagged). The authors acknowledge that HS chains of Dally are critical for Dpp distribution (and signaling) under physiological conditions. If this is true, one can wonder why overexpressed dally core 'binds' Dpp and whether this is a physiologically relevant activity.

According to the overexpression assay, DallyDeltaHS seems more stable than WT Dally (Fig. 4B’, E’, 5H, I). As the reviewer suggested, we addressed the difference using the two knock-in alleles and found that DallyDeltaHS is more stable than WT Dally (Fig.4 L, M inset), further emphasizing the insufficient role of core protein of Dally for extracellular Dpp distribution.

(During the revising our figure, we found labeling mistake in Fig. 4M, N and Fig. 4Q, R and corrected the genotypes.)

In summary, we showed that, although Dally interacts with Dpp mainly through its core protein from the overexpression assay (Fig. 4E, I), HS chains are essential for extracellular Dpp distribution (Fig. 4R). Thus, the core protein of Dally alone is not sufficient for extracellular Dpp distribution under physiological conditions. These results raise a question about whether the interaction of core protein of Dally with Dpp is physiologically relevant. Since the increase of HS upon dally expression but not upon dlp expression resulted in the accumulation of extracellular Dpp (Fig. 2) and this accumulation was mainly through the core protein of Dally (Fig. 4E, I), we speculate that the interaction of the core protein of Dally with Dpp gives ligand specificity to Dally under physiological conditions.

To understand the importance of the interaction of core protein of Dally with Dpp under physiological conditions, it is important to identify a region responsible for the interaction. Our preliminary results overexpressing a dally mutant lacking the majority of core protein (but keeping the HS modified region intact) showed that HS chains modification was also lost. Although this is consistent with our results that enzymes adding HS chains also interact with the core protein of Dally (Fig. 4D), the dally mutant allele lacking the core protein would hamper us from distinguishing the role of core protein of Dally from HS chains.

Nevertheless, we can infer the importance of the interaction of core protein of Dally with Dpp using dally[3xHA-dlp, attP] allele, where dlp is expressed in dally expressing cells. Since Dally-like is modified by HS chains but does not interact with Dpp (Fig. 2, 4), dally[3xHA-dlp, attP] allele mimics a dally allele where HS chains are properly added but interaction of core protein with Dpp is lost. As we showed in Fig.3O, S, the allele could not rescue dallyKO phenotypes, consistent with the idea that interaction of core protein of Dally with Dpp is essential for Dpp distribution and signaling and HS chain alone is not sufficient for Dpp distribution.

b) Although the authors' inference that dallycore (at least if overexpressed) can bind Dpp. This assertion needs independent validation by a biochemical assay, ideally with surface plasmon resonance or similar so that an affinity can be estimated. I understand that this will require a method that is outside the authors' core expertise but there is no reason why they could not approach a collaborator for such a common technique. In vitro binding data is, in my view, essential.

We agree with the reviewer that a biochemical assay such as SPR helps us characterize the interaction of core protein of Dally and Dpp (if the interaction is direct), although the biochemical assay also would not demonstrate the interaction under the physiological conditions.

However, SPR has never been applied in the case of Dpp, probably because purifying functional refolded Dpp dimer from bacteria has previously been found to be stable only in low pH and be precipitated in normal pH buffer (Groppe J, et al., 1998)(Matsuda et al., 2021). As the reviewer suggests, collaborating with experts is an important step in the future.

Nevertheless, SPR was applied for the interaction between BMP4 and Dally (Kirkpatrick et al., 2006), probably because BMP4 is more stable in the normal buffer. Although the binding affinity was not calculated, SPR showed that BMP4 directly binds to Dally and this interaction was only partially inhibited by molar excess of exogenous HS, suggesting that BMP4 can interact with core protein of Dally as well as its HS chains. In addition, the same study applied Co-IP experiments using lysis of S2 cells and showed that Dpp and core protein of Dally are co-immunoprecipitated, although it does not demonstrate if the interaction is direct.

In a subsequent set of experiments, the authors assess the activity of a form of Dpp that is expected not to bind GAGs (DppDeltaN). Overexpression assays show that this protein is trapped by DallyWT but not dallyDeltaHS. This is a good first step validation of the deltaN mutation, although, as before, an invitro binding assay would be preferable.

Our overexpression assays actually showed that DppDeltaN is trapped by DallyWT and by dallyDeltaHS at similar levels (Fig. 5H-J), indicating that interaction of DppDeltaN and HS chains of Dally is largely lost but DppDeltaN can still interact with core protein of Dally.

(Related to this, we found typo in the sentence “In contrast, the relative DppΔN accumulation upon DallyΔHS expression in JAX;dppΔN was comparable to that upon DallyΔHS expression in JAX;dppΔN (Fig. 5H-J).” and corrected as follows, “In contrast, the relative DppΔN accumulation upon Dally expression in JAX;dppΔN was comparable to that upon DallyΔHS expression in JAX;dppΔN (Fig. 5H-J).”

We thank the reviewer for the suggesting the in vitro experiment. Although we decided not to develop biophysical experiments such as SPR for Dpp in this study due to the reasons discussed above, we would like to point out that our result is consistent with a previous Co-IP experiment using S2 cells showing that DppDeltaN loses interaction with heparin (Akiyama2008).

However, in contrast to our results, the same study also proposed by Co-IP experiments using S2 cells that DppDeltaN loses interaction with Dally (Akiyama2008). Although it is hard to conclude since western blotting was too saturated without loading controls and normalization (Fig. 1C in Akiyama 2008), and negative in vitro experiments do not necessarily demonstrate the lack of interaction in vivo. One explanation why the interaction was missed in the previous study is that some factors required for the interaction of DppDeltaN with core protein of Dally are missing in S2 cells. In this case, in vivo interaction assay we used in this study has an advantage to robustly detect the interaction.

Nevertheless, the authors show that DppDeltaN is surprisingly active in a knock-in strain. At face value (assuming that DeltaN fully abrogates binding to GAGs), this suggests that interaction of Dpp with the GAG chains of Dally is not required for signaling activity. This leads to authors to suggest (as shown in their final model) that GAG chains could be involved in mediating the interactions of Dally with Tkv (and not with Dpp. This is an interesting idea, which would need to be reconciled with the observation that the distribution of Dpp is affected in dallyDeltaHS knock-ins (item a above). It would also be strengthened by biochemical data (although more technically challenging than the experiments suggested above). In an attempt to determine the role of Dally (GAGs in particular) in the signaling gradient, the paper next addresses its relation to Tkv. They first show that reducing Tkv leads to Dpp accumulation at the cell surface, a clear indication that Tkv normally contributes to the degradation of Dpp. From this they suggest that Tkv could be required for Dpp internalisation although this is not shown directly. The authors then show that a Dpp gradient still forms upon double knockdown (Dally and Tkv). This intriguing observation shows that Dally is not strictly required for the spread of Dpp, an important conclusion that is compatible with early work by Lander suggesting that Dpp spreads by free diffusion. These result show that Dally is required for gradient formation only when Tkv is present. They suggest therefore that Dally prevents Tkv-mediated internalisation of Dpp. Although this is a reasonable inference, internalisation assays (e.g. with anti-Ollas or anti-HA Ab) would strengthen the authors' conclusions especially because they contradict a recent paper from the Gonzalez-Gaitan lab.

Thanks for suggesting the internalization assay. As we discussed in the discussion, our results suggest that extracellular Dpp distribution is severely reduced in dally mutants due to Tkv mediated internalization of Dpp (Fig. 6). Thus, extracellular Dpp available for labelling with nanobody is severely reduced in dally mutants, which can explain the reduced internalization of Dpp in dally mutants in the internalization assay. Therefore, we think that the nanobody internalization assay would not distinguish the two contradicting possibilities.

The paper ends with a model suggesting that HS chains have a dual function of suppressing Tkv internalisation and stimulating signaling. This constitutes a novel view of a glypican's mode of action and possibly an important contribution of this paper. As indicated above, further experiments could considerably strengthen the conclusion. Speculation on how the authors imagine that GAG chains have these activities would also be warranted.

Thank you very much!

Reviewer #2 (Public Review):

The authors are trying to distinguish between four models of the role of glypicans (HSPGs) on the Dpp/BMP gradient in the Drosophila wing, schematized in Fig. 1: (1) "Restricted diffusion" (HSPGs transport Dpp via repetitive interaction of HS chains with Dpp); (2) "Hindered diffusion" (HSPGs hinder Dpp spreading via reversible interaction of HS chains with Dpp); (3) "Stabilization" (HSPGs stabilize Dpp on the cell surface via reversible interaction of HS chains with Dpp that antagonizes Tkv-mediated Dpp internalization); and (4) "Recycling" (HSPGs internalize and recycle Dpp).

To distinguish between these models, the authors generate new alleles for the glypicans Dally and Dally-like protein (Dlp) and for Dpp: a Dally knock-out allele, a Dally YFP-tagged allele, a Dally knock-out allele with 3HA-Dlp, a Dlp knock-out allele, a Dlp allele containing 3-HA tags, and a Dpp lacking the HS-interacting domain. Additionally, they use an OLLAS-tag Dpp (OLLAS being an epitope tag against which extremely high affinity antibodies exist). They examine OLLAS-Dpp or HA-Dpp distribution, phospho-Mad staining, adult wing size.

They find that over-expressed Dally - but not Dlp - expands Dpp distribution in the larval wing disc. They find that the Dally[KO] allele behaves like a Dally strong hypomorph Dally[MH32]. The Dally[KO] - but not the Dlp[KO] - caused reduced pMad in both anterior and posterior domains and reduced adult wing size (particularly in the Anterior-Posterior axis). These defects can be substantially corrected by supplying an endogenously tagged YFP-tagged Dally. By contrast, they were not rescued when a 3xHA Dlp was inserted in the Dally locus. These results support their conclusion that Dpp interacts with Dally but not Dlp.

They next wanted to determine the relative contributions of the Dally core or the HS chains to the Dpp distribution. To test this, they over-expressed UAS-Dally or UAS-Dally[deltaHS] (lacking the HS chains) in the dorsal wing. Dally[deltaHS] over-expression increased the distribution of OLLAS-Dpp but caused a reduction in pMad. Then they write that after they normalize for expression levels, they find that Dally[deltaHS] only mildly reduces pMad and this result indicates a major contribution of the Dally core protein to Dpp stability.

Thanks for the comments. We actually showed that compared with Dally overexpression, Dally[deltaHS] overexpression only mildly reduces extracellular Dpp accumulation (Fig. 4I). This indicates a major contribution of the Dally core protein to interaction with Dpp, although the interaction is not sufficient to sustain extracellular Dpp distribution and signaling gradient.

The "normalization" is a key part of this model and is not mentioned how the normalization was done. When they do the critical experiment, making the Dally[deltaHS] allele, they find that loss of the HS chains is nearly as severe as total loss of Dally (i.e., Dally[KO]). Additionally, experimental approaches are needed here to prove the role of the Dally core.

Since the expression level of Dally[deltaHS] is higher than Dally when overexpressed, we normalized extracellular Dpp distribution (a-Ollas staining) against GFP fluorescent signal (Dally or Dally[deltaHS]). To do this, we first extracted both signal along the A-P axis from the same ROI. The ratio was calculated by dividing the intensity of a-Ollas staining with the intensity of GFP fluorescent signal at a given position x. The average profile from each normalized profile was generated and plotted using the script described in the method (wingdisc_comparison.py) as other pMad or extracellular staining profiles.

Although this analysis provides normalized extracellular Dpp accumulation at different positions along the A-P axis, we are more interested in the total amount of Dpp or DppDeltaN accumulation upon Dally or dallyDeltaHS expression. Therefore, we plan to analyze the normalized total amount of Dpp against GFP fluorescent signal (Dally or Dally[deltaHS]) in the revised ms. In this case, normalization will be performed by dividing total signal intensity of extracellular Dpp staining (ExOllas staining) divided by GFP fluorescent signal (Dally or Dally[deltaHS]) in ROI in each wing disc.

We agree with the reviewer that additional experimental approaches are needed to address the role of the core protein of Dally. As we discussed in the response to the reviewer1, to understand the importance of the interaction of core protein of Dally with Dpp, it is important to identify a region responsible for the interaction. Our preliminary results overexpressing a dally mutant lacking the majority of core protein (but keeping the HS modified region intact) showed that HS chains modification was also lost. Although this is consistent with our results that enzymes adding HS chains also interact with the core protein of Dally (Fig. 4D), the dally mutant allele lacking the core protein would hamper us from distinguishing the role of the core protein of Dally from HS chains.

Nevertheless, we can infer the importance of the interaction of core protein of Dally with Dpp using dally[3xHA-dlp, attP] allele, where dlp is expressed in dally expressing cells. Since Dally-like is modified by HS chains but does not interact with Dpp (Fig. 2, 4), dally[3xHA-dlp, attP] allele mimics a dally allele where HS chains are properly added but interaction of core protein with Dpp is lost. As we showed in Fig.3O, S, the allele could not rescue dallyKO phenotypes, consistent with the idea that interaction of core protein of Dally with Dpp is essential for Dpp distribution and signaling.

Prior work has shown that a stretch of 7 amino acids in the Dpp N-terminal domain is required to interact with heparin but not with Dpp receptors (Akiyama, 2008). The authors generated an HA-tagged Dpp allele lacking these residues (HA-dpp[deltaN]). It is an embryonic lethal allele, but they can get some animals to survive to larval stages if they also supply a transgene called “JAX” containing dpp regulatory sequences. In the JAX; HA-dpp[deltaN] mutant background, they find that the distribution and signaling of this Dpp molecule is largely normal. While over-expressed Dally can increase the distribution of HA-dpp[deltaN], over-expression of Dally[deltaHS] cannot. These latter results support the model that the HS chains in Dally are required for Dpp function but not because of a direct interaction with Dpp.

Our overexpression assays actually showed that both Dally and Dally[deltaHS] can accumulate Dpp upon overexpression and the accumulation of Dpp is comparable after normalization (Fig. 5H-J), consistent with the idea that interaction of DppdeltaN and HS chains are largely lost. As the reviewer pointed out, these results support the model that the HS chains in Dally are required for Dpp function but not because of a direct interaction with Dpp.

In the last part of the results, they attempt to determine if the Dpp receptor Thickveins (Tkv) is required for Dally-HS chains interaction. The 2008 (Akiyama) model posits that Tkv activates pMad downstream of Dpp and also internalizes and degrades Dpp. A 2022 (Romanova-Michaelides) model proposes that Dally (not Tkv) internalizes Dpp.

To distinguish between these models, the authors deplete Tkv from the dorsal compartment of the wing disc and found that extracellular Dpp increased and expanded in that domain. These results support the model that Tkv is required to internalize Dpp.

They then tested the model that Dally antagonizes Tkv-mediated Dpp internalization by determining whether the defective extracellular Dpp distribution in Dally[KO] mutants could be rescued by depleting Tkv. Extracellular Dpp did increase in the D vs V compartment, potentially providing some support for their model. However, there are no statistics performed, which is needed for full confidence in the results. The lack of statistics is particularly problematic (1) when they state that extracellular Dpp does not rise in ap>tkv RNAi vs ap>tkv RNAi, dally[KO] wing discs (Fig. 6E) or (2) when they state that extracellular Dpp gradient expanded in the dorsal compartment when tkv was dorsally depleted in dally[deltaHS] mutants (Fig. 6I). These last two experiments are important for their model but the differences are assessed only visually. In fact, extracellular Dpp in ap>tkv RNAi, dally[KO] (Fig. 6B) appears to be lower than extracellular Dpp in ap>tkv RNAi (Fig. 6A) and the histogram of Dpp in ap>tkv RNAi, dally[KO] is actually a bit lower than Dpp in ap>tkv RNAi, But the author claim that there is no difference between the two. Their conclusion would be strengthened by statistical analyses of the two lines.

We will provide the statistical analyses in the revised ms.

Strengths:

1) New genomically-engineered alleles

A considerable strength of the study is the generation and characterization of new Dally, Dlp and Dpp alleles. These reagents will be of great use to the field.

Thanks. We hope that these resources are indeed useful to the field.

2) Surveying multiple phenotypes

The authors survey numerous parameters (Dpp distribution, Dpp signaling (pMad) and adult wing phenotypes) which provides many points of analysis.

Thanks!

Weaknesses:

1) Confusing discussion regarding the Dally core vs HS in Dpp stability. They don't provide any measurements or information on how they "normalize" for the level of Dally vs Dally[deltaHS]? This is important part of their model that currently is not supported by any measurements.

We explained how we normalized in the above section. We will update the analysis in the revised ms.

2) Lacking quantifications and statistical analyses:

a) Why are statistical significance for histograms (pMad and Dpp distribution) not supplied? These histograms provide the key results supporting the authors' conclusions but no statistical tests/results are presented. This is a pervasive shortcoming in the current study.

Thanks. We will provide statistics in the revised ms.

b) dpp[deltaN] with JAX transgene - it would strengthen the study to supply quantitative data on the percent survival/lethal stage of dpp[deltaN] mutants with or without the JAK transgene

In this study, we are interested in the role of dpp[deltaN] during the wing disc development. Therefore, we decided not to perform the detailed analysis on the percent survival/lethal stage of dpp[deltaN] mutants with or without the JAX transgene in the current study. Nevertheless, the fact that dpp[deltaN] allele is maintained with a balanced stock and JAX;dpp[deltaN] allele can be maintained as homozygous stock indicates that the lethality of dpp[deltaN] allele comes from the early stages. Indeed, our preliminary results showed that pMad signal is severely lost in the dpp[deltaN] embryo without JAX (data not shown), indicating that the allele is lethal at early embryonic stages.

c) The graphs on wing size etc should start at zero.

Thanks. We corrected this in the current ms.

d) The sizes of histograms and graphs in each figure should be increased so that the reader can properly assess them. Currently, they are very small.

Thanks. We changed the sizes in the current ms.

The authors' model is that Dally (not Dlp) is required for Dpp distribution and signaling but that this is not due to a direct interaction with Dpp. Rather, they posit that Dally-HS antagonize Tkv-mediated Dpp internalization. Currently the results of the experiments could be considered consistent with their model, but as noted above, the lack of statistical analyses of some parameters is a weakness.

Thanks. We will perform the statistical analyses in the revised ms.

One problematic part of their result for me is the role of the Dally core protein (Fig. 7B). There is a mis-match between the over-expression results and Dally allele lacking HS (but containing the core). Finally, their results support the idea that one or more as-yet unidentified proteins interact with Dally-HS chains to control Dpp distribution and signaling in the wing disc.

Our results simply suggest that Dpp can interact with Dally mainly through core protein but this interaction is not sufficient to sustain extracellular Dpp gradient formation under physiological conditions (dallyDeltaHS) (Fig. 4Q). We find that the mis-match is not problematic if the role of Dally is not simply mediated through interaction with Dpp. We speculate that interaction of Dpp and core protein of Dally is transient and not sufficient to sustain the Dpp gradient without HS chains of Dally stabilizing extracellular Dpp distribution by blocking Tkv-mediated Dpp internalization.

There is much debate and controversy in the Dpp morphogen field. The generation of new, high quality alleles in this study will be useful to Drosophila community, and the results of this study support the concept that Tkv but not Dally regulate Dpp internalization. Thus the work could be impactful and fuel new debates among morphogen researchers.

Thanks.

The manuscript is currently written in a manner that really is only accessible to researchers who work on the Dpp gradient. It would be very helpful for the authors to re-write the manuscript and carefully explain in each section of the results (1) the exact question that will be asked, (2) the prior work on the topic, (3) the precise experiment that will be done, and (4) the predicted results. This would make the study more accessible to developmental biologists outside of the morphogen gradient and Drosophila communities.

Thanks. We will modify our texts to help non-experts understand our story in the revised ms.

Author Response

Reviewer #2 (Public Review):

Major points:

1). This study does not provide any evidence about the cell death of the transplanted cells. The immunostaining of the Caspase-3 or TUNEL staining should be used to address this issue.

We have conducted immunostaining of Caspase-3 at 7 days after transplantation using the human-specific STEM121 antibody to demonstrate the transplanted cells. We have added the results to Figure 3A and modified the text accordingly (Page 8, Line 156-165).

2). The authors showed that the neurological functions (evaluated by balance beam, ladder lung, rotarod test and Modified Neurological Severity Score (mNSS) up to 8 weeks after treatment (Figure 1C)) were significantly improved in the NES+Exo group compared to their control groups. However, these cells (transplanted cells) are progenitors (Nestin+) or undifferentiated cells (Tuj1+) at this stage (Figure 3). Thus, I was curious about that how can the immature neurons play neurological functions? This point should be explained.

We agree with the reviewer’s insightful comments. We have performed immunostaining using antibodies against the post-mitotic mature neuron marker RBFOX3/NeuN, post-synaptic marker PSD-95 and human-specific STEM121 at 4 weeks after transplantation. The results confirmed that NeuN+/STEM121+ and PSD-95+/STEM121+ mature neurons appeared in NSC group and increased in NSC+Exo group (Figure 3B and Figure 3 - supplement 1D). Furthermore, our additional data showed that the expression of presynaptic marker SYN1 was increased in both NSC and NSC+Exo groups at 8 weeks after treatment. Therefore, we believe that there are mature neurons and newly formed synapses involved in neurological functions.

3). The authors used the Golgi staining to show the NES+Exo can improve dendritic density and length. How do you know these neurons are transplanted cells?

Our data show that mature neurons and synapses are generated by the transplanted cells (please also see response to reviewer #2-major ponts #2). We believe that the newly generated neurons partly contribute to the improved dendritic density and length. However, we agree that the neurons with increased dendritic density and length may be both survived local neurons and those generated by the transplanted cells.

4). The cell morphology of tdTomato+ cells is fuzzy and it is difficult to distinguish the cell body. It looks like that these cells out of whack.

We have immunostaining using the human-specific STEM121 antibody to demonstrate the transplanted cells and more neuronal markers such as RBFOX3/NeuN to identify NSC differentiation (Figure 3A and 3B; Figure 3 - supplement 1C and 1D).