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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

In this study, Lebdy et al propose a new mechanism to regulate the resection of nascent DNA at stalled replication forks. The central element of this mechanism is nucleolar protein GNL3, whose downregulation with siRNA stimulates DNA resection in the presence of stress induced by HU (Figure 1). Resection depends on the activity of nucleases MRE11 and CtIP, and can be rescued by reintroducing exogenous GNL3 protein in the cells (Figure 1G). GNL3 downregulation decreases fork speed and increases origin activity, without any strong effect on replication timing (Figure 2). Inhibition of Dbf4-dependent kinase CDC7 (a known origin-activating factor) also restricts fork resection (Figure 3). GNL3 interacts with ORC2, one of the subunits of the origin recognition complex, preferentially in nucleolar structures (Figure 4). A mutant version of GNL3 (GNL3-dB) that is not sufficiently retained in the nucleoli fails to prevent fork resection as the WT protein (Figure 5). In the final model, the authors propose that GNL3 controls the levels of origin activity (and indirectly, stalled fork resection) by maintaining a fraction of ORC2 in the nucleoli (Figure 6).

This model is interesting and provocative, but it also relies on a significant degree of speculation. The authors are not trying to "oversell" their observations, because the Discussion section entertains different interpretations and possibilities, and the model itself contains several interrogative statements (e.g. "ORC2-dependent?"; "exhaustion of factors?").

While the article is honest about its own limitations, the major concern remains about its highly speculative nature. I have some questions and suggestions for the authors to consider that could contribute to test (and hopefully support) their model.

1. If GNL3 downregulation induces an excess of licensed origins and mild replicative stress resulting in some G2/M accumulation (Figure 2), what is the consequence of longer-term GNL3 ablation? Do the cells adapt, or do they accumulate signs of chromosomal instability? (micronuclei, chromosome breaks and fusions, etc)
2. The model relies on the link between origin activity and stalled fork resection that is almost exclusively based on the results obtained with CDC7i (Figure 3). But CDC7 has other targets besides pre-RC components at the origins, such as Exo1 (from the Weinreich lab, cited in the study), MERIT40 and PDS5B (from the Jallepalli lab, also cited). The effect of CDC7i could be exerted through these factors, which are linked to fork stability and DNA resection. The loss of BRCA1 (Figure 3F) could somehow entail the loss of control over these factors. Could the authors check the possible participation of these proteins?
3. The model also relies on the fact that GNL3-dB mutant (not retained in the nucleoli) is not sufficient to counteract fork resection induced by HU (Figure 5G). The authors should test directly whether GNL3-dB induces extra origin activation, using their available DNA fibers-based technique.
4. Finally, the model implies an exquisite regulation of the amount of ORC2 protein, which could influence the number of active origins and the extent of fork resection in case of stress. In this scenario, one could predict that ORC2 ectopic expression would have similar, or even stronger effects, than GNL3 downregulation. Is this the case?
5. Even if the connection between origins and fork resection could be firmly established, the molecular link between them remains enigmatic. The authors hint (as "data not shown") that it is neither mediated by RPA nor RAD51. Unfortunately, the reader is left without a clear hypothesis about this point.

Referees cross-commenting

In addition to each reviewer's more specific comments, the three reviews share a main criticism: the lack of mechanistic information about the proposed link between origin activity and resection of nascent DNA at stalled forks.

Significance

In principle, this study would appeal to the readership interested in fundamental mechanisms of DNA replication and the cellular responses to replicative stress.

For the reasons outlined in the previous section, I believe that in its current version the study is not strong enough to provide a new paradigm about origins being regulated by partial ORC2 sequestering at nucleoli. The other potentially interesting advance is the connection between frequency of origin activity and the extent of nascent DNA resection at stalled forks, but the molecular link between both remains unknown.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

This manuscript explores the role of GNL3/nucleostemin in DNA replication and specifically in the response of DNA replication to DNA damage. GNL3 is a predominantly nucleolar protein, previously characterised as a GTP-binding protein and shown to be necessary for effective recruitment of the RAD51 recombinase to DNA breaks. The entry point for this report is a mini screen, based on proteins identified previously by the authors to associate with replication forks by iPOND, for factors that increase gamma-H2Ax (an indicator of DNA damage) after treatment with the Top1 inhibitor camptothecin (CPT). In this mini-screen GNL3 emerged as the top hit.

The authors put forward the hypothesis that GNL3 is able to sequester the replication licensing factor ORC2 in the nucleolus and that failure of this mechanism leads to excessive origin firing and DNA resection following CPT treatment.

The model put forward is interesting, but currently rather confusing. However, for the reasons upon which I expand below, I do not believe that the data provide a compelling mechanistic explanation for the effects that are reported and I am left not being certain about some of the links that are made between the various parts of the study, even though individual observations appear to be of good quality.

Specific points:

The knockdown of GNL3 is very incomplete. In this regard, the complementation experiments are welcome and important. However, is it an essential protein? Can it be simply deleted with CRISPR-Cas9?

Global instant fork density is not quite the same as actually measuring origin firing. Ideally, it would be good to see some more direct evidence of addition origin firing e.g. by EdU-seq (Macheret & Halazonetis Nature 2018) but this would be quite a significant additional undertaking. However, given the authors have performed DNA combing with DNA counterstain, they should be able to provide accurate measurements of origin density and inter-origin distance.

'Replication stress' is induced with CPT. This term is frequently used to describe events that lead to helicase-polymerase uncoupling (e.g. O'Connor Mol Cell 2015) but that is not the case with CPT, which causes fork collapse and breaks. Are similar effects seen with e.g. UV or cisplatin? Additionally, a clear statement of the authors definition of replication stress would be welcome.

It is really not clear how the authors explain the link between potential changes in origin firing and resection. i.e. What is the relationship between global origin firing and resection at a particular fork, presumably broken by encounter with a CPT-arrested TOP1 complex. What is the link mechanistically? This link needs elaborating experimentally or clearly explaining based on prior literature.

Related to this, I remain unconvinced that the experiments in Figure 3 show that the effects of ATRi and Wee1i on origin firing and on resection are contingent on each other. I do not believe that the authors have adequately supported the statement (end of pg 9) 'We conclude that the enhanced resection observed upon GNL3 depletion is a consequence of increased origin firing.' The link between origin firing and resection needs really needs further substantiation and / or explanation.

It is not clear whether the binding of ORC2 to GNL3 also sequesters other components of the origin recognition complex? Does loss of the ability of GNL3 to bind ORC2 actually lead to more ORC bound to chromatin? How does GNL3 contribute to regulation of origin firing under normal conditions? Is it a quantitatively significant sink for ORC2 and what regulates ORC2 release?

Minor points:

All blots should include size markers

Some use of language is not sufficiently precise. For instance:

• the meaning of 'DNA lesions' at the end of the first paragraph of the introduction needs to be more explicit.
• the approach to measurement of these 'lesions' (monitoring gamma-H2Ax) needs to be spelled out explicitly, e.g. line 4 of the last paragraph of the introduction.
• 'we observed that the interaction between GNL3-dB and ORC2 was stronger' ... I do not see how number of foci indicates necessarily the strength of an interaction.
• in many places throughout 'replication origins firing' should be 'replication origin firing' (or 'firing of replication origins').

Significance

The model put forward here has the potential to shed light on an important facet of the cellular response to DNA damage, namely the control of origin firing in response to replication stress that will certainly be of interest to the DNA repair / replication community and possibly more widely. The roles of GNL3 are poorly understood and this study could improve this state of affairs. However, the gaps in the mechanism outlined above and somewhat confusing conclusions do limit the ability of the paper to achieve this at present.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In this manuscript Lebdy et al. describe a new role of GNL3 in DNA replication. They show that GNL3 controls replication fork stability in response to replication stress and they propose this is due to the regulation of ORC2 and the licensing of origins of replication. Their data suggest that GNL3 regulates the sub nuclear localization of ORC2 to limit the number of licensed origins of replication and to prevent resection of DNA at stalled forks in the presence of replication stress.

While many of the points of the manuscript are proven and well supported by the results, there are some experiments that could improve the quality and impact of the manuscript. The main issue is that the connection between the role of GNL3 in controlling ORC2, the firing of new origins and the protection of replication forks is not clearly established. At the moment the model relies on mainly correlative data. In order to further substantiate the model, we propose to address some of the following issues:

1. The authors indicate that RPA and RAD51 accumulation at stalled forks is not affected by GNL3 depletion. These data should be included and other proteins should be analysed. In addition, the role of helicases could be explored through the depletion of the main helicases involved in the remodelling of the forks.
2. The proposed model implies that GNL3 depletion leads to increased origin licensing. FThe authors should address if the primary effect of GNL3 depletion is on origin firing by using CDC7 inhibition in the absence of stress (Rodríguez-Acebes et al., JBC 2018).
3. A way to prove that origin firing mediates the effect of GNL3 on fork protection would be to reduce the number of available origins. The depletion of MCM complexes has been shown to limit the number of back-up origins that are licensed and leads to sensitivity to replication stress (Ibarra et al., PNAS 2008). If GNL3 depletion results in increased number of origins, this effect should be prevented by the partial depletion of MCM complexes.
4. Alternatively, the authors could try to modulate the depletion of GNL3. Origin licensing takes place in the G1 phase and thus the depletion of GNL3 by siRNA could affect the following S phase. Using an inducible degron for GNL3 depletion would allow to deplete GNL3 in G1 or S phase specifically. If the model is correct, the removal of GNL3 in S phase should not affect fork protection but removing GNL3 in the previous G2/M phase should reduce the number of licensed origins and lead to impaired fork protection. In addition to the connection GNL3-origin firing-fork protection, it is unclear how the lack of GNL3 in the nucleolus and the change in the sub nuclear localization of ORC2 controls origin firing and resection. The strong interaction observed between GNL3-dB and ORC2, and the subsequent change in ORC2 localization does not explain how origin licensing can be affected. In this sense, the authors could address:
   1. Does the depletion of GNL3 and the expression of GNL3-dB affect the formation of the ORC complex, its subnuclear localization or its binding to chromatin? The authors have not explored if the interaction of GNL3 with ORC2 is established in the context of the ORC complex. An IF showing NOP1 with PLA data from GNL3-dB and ORC2 is needed to analyse how the expression of increasing amounts of GNL3-dB affects ORC2.
   2. In order to confirm if the mislocalization of ORC2 by the expression of GNL3-dB increases origin firing and mediates the effects on fork protection the authors could check DNA resection levels inhibiting CDC7 in high GNL3-dB conditions. Also, the levels of MCM2, phosphor-MCM2, CDC45, have not been analysed upon expression of GNL3-dB.
   3. The data in the paper suggest that GNL3 may affect the role of ORC2 in centromeres. Since depletion of GNL3 leads to increased levels of H2AX, it would be interesting to address if this damage is due to incomplete replication in centromeres by analysing the co-localization of H2AX and centromeric markers both in unstressed conditions and upon the induction of replication stress.

Minor points:

1. In the initial esiRNA screen the basal levels of H2AX should also be shown.
2. Figure EV1B: I think the rank needs another RS mark to see better the effect of each esiRNA on DNA lesions (high variability in all the conditions showed).
3. Figure 1C and Figure EV1D/E: the quantification of the pCHK1/CHK1 levels could be included to show that there are no changes in phosphorylation upon GNL3 depletion.
4. In the first section of the results, at the end Figure 4B is incorrectly called for.
5. The levels of GLN3 expression in 293 cells should be already included in section GNL3 interacts with ORC2.
6. The full MS data needs to be included for both GNL3 and ORC2.
7. Figure 4B should be improved, since there is a faint band in the IgG mouse control.

Significance

The work is nicely written, the figures are well presented and the experiments have the necessary controls. It provides relevant information to understand how replication stress is controlled and linked to replication fork protection through origin firing. These results are relevant to the field, linking GNL3 to origin firing and with potential to help understand the role of GNL3 in cancer. They provide new information and can give rise to new studies in the future. Many of the conclusions of the manuscript are well supported. Additional support for some of the main claims would strengthen the results and also increase the impact providing a bigger conceptual advance by performing some of the suggested experiments.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements [optional]

We are grateful for the thoughtful comments and suggestions from the reviewers which we feel have resulted in a manuscript that is both clearer to the reader and more rigorous. We have addressed the suggested revisions point by point below:

2. Point-by-point description of the revisions

Reviewer #1 Major comments:

Figure 2E. This shows the residence time of Cse4 without Ndc10 association. How does this compare to the residence time on mutant CEN3 (Supplemental Figure 1). It looks like Cse4 still binds to CEN3 with some specificity even in the absence of Ndc10. Does this suggest that Cse4 has some intrinsic ability to recognize CEN3? Alternatively, Ndc10 is still required for Cse4 binding but is below detection in the Cse4-alone residence lifetimes. Ideally, the authors would compare this with Cse4 binding in an Ndc10 mutant.

We thank the reviewer for this interesting question. As suggested, we analyzed Cse4 behavior on the mutant CDEIIImut CEN3 DNA, which does not stably recruit Ndc10 (Figure 1C), using the real-time colocalization assay. Although overall Cse4 associations were reduced, we still observed transient interactions, consistent with the possibility that Cse4 has some intrinsic ability to recognize CEN3. Kaplan-Meier analysis of the lifetimes of Cse4 colocalizations on CDEIIImut CEN3 DNA were significantly reduced when compared to CEN DNA (Figure EV1G, H). We have added these points to the text (p. 10, lines 29-31 and p. 11, lines 1-8).

Figure 3 explores the very interesting relationship between Scm3 dynamics and Cse4 binding but I feel that the data is not fully flushed out. A key finding is that Cse4 can potentially bind to CEN DNA prior to engaging with Scm3 to be incorporated. This predicts that Cse4 should bind first and then colocalize with Scm3. Can this be detected in the timing of the traces? How often does Scm3 bind to already prebound Cse4 and does this correlate with Cse4 residing longer?

We performed new and more rigorous analyses of the data to address these questions in the revised manuscript. After our analysis to calculate ternaryScm3 off-rates, we analyzed the relative timing of these ternary residences and found that indeed Cse4 can bind to CEN DNA prior to Scm3 and these events do correlate with Cse4 residing longer. Complete analysis of the binding order of Cse4 and Scm3 ternary residences revealed that Scm3 bound to CEN3 DNA prior to Cse4 more often than Cse4 preceding Scm3 (46% vs 34% of ternary residences) with the remaining 20% arriving simultaneously (Figure EV2E). Despite a difference in prevalence, the median lifetimes of both Scm3-Firstand Scm3-Last contexts were similar to each other (Figure EV2F) and both were significantly stabilized when compared to non-ternary residences. These results highlight a potential mechanism where Scm3 catalyzes stable Cse4 incorporation at centromeric DNA instead of delivering it to the centromere regardless of the order of arrival. These data are now reported and discussed in the revision (p. 12, line 11-p.13, line 10).

A related and perhaps even more interesting point is whether Scm3 is involved in "loading" of Cse4. If so, then one would expect that once Cse4 is assembled into nucleosomes it should be stable, even if Scm3 leaves. Can the authors extract this from the data? Alternatively, it is possible that Scm3 remains associated to Cse4 to maintain the nucleosome which would imply a more extended role for Scm3 apart from assembly alone. It would be interesting if information on this can be extracted from the data.

Using our updated analysis of ternaryScm3 Cse4 residences, feel we have addressed this possibility indirectly in a couple of ways. First, we found that in instances where ternary Scm3/Cse4 complexes are formed, Scm3 co-occupied the CEN DNA an average of 56.0% of the total Cse4 residence time, which is distinctly shorter than the 84% of the total Cse4 residence that was occupied by Ndc10 in Cse4/Ndc10 ternary residences. These findings are consistent with a Scm3-turnover mechanism that occurs post ternary complex formation with Cse4 as we found that Cse4 off-rates were significantly reduced after Scm3 association (Figure 3D).

Second, further analysis of Scm3 residence behavior revealed that there was no significant stabilization of Scm3 after ternary Cse4/Scm3 complex formation vs non-ternary Scm3 residences found in either off-rates (33 hr-1 vs 32 hr-1, Figure EV2C) or median lifetimes (45 s vs 52 s, Figure EV2D). These results indicate that Scm3 association is not stabilized like Cse4 after ternary complex formation and points to a potential catalytic role in Cse4 nucleosome formation, leaving a stable Cse4 nucleosome after turnover. We reported these findings in the revision results section (p. 12, lines 11-16) as well as briefly within the discussion.

Even in the presence of Scm3 and CCAN components, Cse4 appears to have a limited lifetime in the in vitro assay compared to in vivo stability. The authors should speculate on whether activities exist in their extract that actively disassembles nucleosomes. Perhaps ATP could be depleted to inactivate remodellers?

This is an excellent suggestion that we addressed with a new experiment. We performed an endpoint localization experiment with lysate containing fluorescently labeled Ndc10 and Cse4 and then removed the lysate and incubated the slide for 24hr in imaging buffer at RT. Strikingly, the proteins were maintained at the CEN DNA with a high protein total colocalization (~75% retention) (shown in Figure EV1B, C). These data suggest that the lysates may contain negative regulatory factors and we have added this point in the revised text (p. 9 lines 6-25).

We were not able to address whether the removal of ATP stabilized the proteins because we previously found that ATP depletion of the lysates completely abrogates kinetochore assembly in extracts. We will need to eventually dissect the role of remodelers in future work using a different approach.

For Figure 6, it is not clear why AT-track mutants of CDEII are labeled as genetically stable and genetically unstable. This is confusing as the "genetically stable" show a more than 10-fold increase in chromosome loss rates. Perhaps these can be renamed into "unstable" and "very unstable" or "weak" and "strong" mutants, just to make clear that these classes are both poorer than wild type.

We had deferred to the nomenclature used in the previous study (Baker and Rogers, 2005) which initially characterized these mutants. To avoid this confusion, we have renamed these mutants “unstable” and “very unstable” as suggested to make it clearer that none of these synthetic mutants fully recapitulate WT CEN3 behavior.

Finally, it would be wonderful to include data to assess whether a full Cse4 nucleosome is assembled or a partial nucleosome e.g. just Cse4/H4 tetrasomes. This could be done by tracking the accumulation of H2A or H2B at the CEN3. This would give further insight into what step Scm3 catalyses.

This is a very interesting suggestion that we were not able to directly address. Epitope tagging of these histone proteins in Saccharomyces cerevisiae with endogenous fluorophores has proved challenging due to gene duplication, overall sensitivity to histone levels within the cells and disruption of histone function by epitope tagging. We hope to find a workable method to address this in the future to address this question directly.

Minor comments:

Typo on page 5, line 1 "nucleosom" missing an e.

We have corrected this in the revised text.

Kaplan-Meyer should be spelled Kaplan-Meier

We have corrected this in the revised text.

The term "censored" is mentioned across many figures but comes up just ones in the methods where it is not clearly explained. Perhaps this could be clarified in the legend.

We have now provided a clear explanation of the term “censored” in the text on p. 28, lines 25-27. It has also been added to the figure legends and reported in the Statistical tests section of the methods section to address this point.

The abstract states that Cse4 can arrive at the centromere without its chaperone. More likely, Cse4 is in complex with other chaperones that may allow it to bind. Perhaps the abstract can be modified to read "Cse4 can arrive at the centromere without its dedicated centromere-specific chaperone Scm3..."

We updated the abstract to reflect this point in the revised text.

Related to this point, the discussion states the possibility that Cse4 can initially bind to CEN3 via other more general chaperones. However, it should be acknowledged that transient Cse4 binding in their assay may simply occur through mass action due to high concentrations of CEN3 DNA. In vivo, this transient binding may not be that relevant.

We acknowledge this potential caveat in the discussion section (p. 20 lines 15-18), although we feel this is somewhat unlikely due to our observation of significantly reduced Cse4 binding on CDEIIImut DNA despite DNA concentrations being similar to previous assays (Figure EV3A). We speculate that some of this transient behavior is at least in part driven by two major factors: negative regulatory factors within our cellular extracts that counter nucleosome formation (as explored in Figure EV1B-C) and photostability of the endogenous fluorophores used within the study (Figure EV1D-E). These points were highlighted within the second paragraph of page 9.

Reviewer #2 Major comments:

1. Figure S1A-D seem like some of the most compelling data in the paper to bolster the rigor of their experimental setup. There appears to be plenty of space to include these data in the main figure set in Figure 1 after panel D. The authors would be well served to consider moving S1A-D somewhere in the main figure set.

We appreciate the helpful feedback on the importance of the date found in Supplemental Figure 1 and have now incorporated it into Figure 1 within the main text as suggested.

The authors conclude that Ndc10 recruits HJURP(Smc3) to the yeast point centromeres. If this is the case, can the TIRFM assay measure ternary residence lifetimes complexes between Ndc10/HJURP(Smc3)/CenDNA?

We made the conclusion that Ndc10 recruits Scm3 based on previous publications showing this requirement in vivo. We have now attempted to address this in our assay indirectly by monitoring Scm3 behavior on the CDEIIImut CEN3 DNA that lacks Ndc10. Surprisingly, we found that Scm3 interacted similarly with CDEIIImut CEN3 DNA and actually showed an increase in median lifetimes vs. CEN3 DNA (Figure EV2B), suggesting its intrinsic DNA-binding activity may be the primary driver of its CEN DNA binding and that stable Cse4 association is required for its turnover. These data suggest that Ndc10 is not driving Scm3 interaction (or targeting) to CEN3. We are grateful to the reviewer for pointing this out and have adjusted our conclusions in the revised manuscript (p. 12, line 26-p.13 line 10).

Throughout the manuscript short- and long- term residence lifespans are mentioned, referencing the figures containing lines with lengths depicting residence times. This is a qualitative reference to short and long residences. Can the authors provide a graph for short-term ( 300 s) residence life-spans for, CENPA alone, CENPA/Ndc10, and CENPA/HJURP on CEN3 DNA? Or some figure similar to Figure 3C, but reporting the proportion of short-term vs long-term residence?

We typically used Kaplan-Meier survival function estimates to compare binding behavior but agree that quantification of residences within these contexts may be easier for the reader to follow. We have therefore quantified short-term ( 300 s) as suggested and added them as a panel (F) to Figure EV1 and as panel (E) in Figure 3.

The choice of CCAN components for analysis in Fig. 5 is interesting, but many readers may be curious why Mif2 wasn't selected for disruption, since it has such a cozy placement with CENP-A and CEN DNA. Can choice to not include Mif2 mutants/degrons be mentioned/justified in the text (unless, even better yet, they choose to address Mif2 role directly in new experimentation)?

We relied on structural models to choose CCAN proteins that are in close proximity to the DNA. Because Mif2 is not in these structures, we did not include it in our studies. We have explained this in the revised text (p. 16 lines 14-17) and agree it is an interesting future area of study.

Minor comments:

1. Are these whole cell extracts (WCE) DNA-free? I'm curious if there is any competition from endogenous DNA from the yeast cellular extract.

The extracts are not DNA-free so it is likely there is some competition from endogenous DNA. We have avoided enzymatically removing the DNA since the TIRF assay depends on the integrity of DNA.

In relation to Mif2 and comment #4 above, do the authors make any connection to their results with synthetic nucleosome sequences not being conducive to yeast centromere formation with the prior observation (Allu et al 2019) using recombinant components that the human version of Mif2 more easily saturates its binding sites on CENP-A nucleosomes when they are assembled with natural centromere DNA rather than the Widom 601 sequence?

We did not speculate on the role of Mif2 and stability of synthetic nucleosome sequences. This is an interesting point but the differences between the yeast and human systems combined with the fact we have not yet started to study Mif2 made it seem too premature to include in this manuscript.

Providing a gel (or other measure) of the DNA templates (750, 250, and 80 bp) used in TIRFM assay would be nice to show to confirm the designed size of the pre-tethered DNA.

We agree this is a helpful control and we have now included it as a panel in Figure EV5B.

1. Some of the references to figures/figure panels in the main text do not match the figures. (discussion, pg 16, paragraph 1 & 2;pg 18, paragraph 1).

We have updated references mentioned to reflect to the correct figure in both sections of the discussion.

Reviewer #3 Major comments: -->Statistics are highly recommended for all the data in the ms.

We have included log-rank analysis to instances where two survival functions were plotted together where appropriate. P-values for all these analyses were reported in the appropriate figure legends.

• At what rate is data collected in the TIRFM setup. For clarity for the reader, it is important to provide imaging details for time-lapse. What is the impact of photobleaching on the stability of the fluorophore signal? Please clarify.

This is a helpful suggestion and we have now included imaging details (like time intervals for each channel) when the real-time assay is introduced in the results section (p. 8, lines 10-12). We have also provided additional details for the photobleaching estimates and how these might censor data in turn (p. 9, lines 14-25). Although photobleaching is a primary limitation of the time-lapse assays, we point out that it is appropriate to compare protein behavior under identical imaging parameters within differing contexts. We also noted that we typically compared time-lapse behavior (which is affected by photobleaching) with endpoint assays to ensure consistent behavior.

• The power of single-molecule technique is precisely that such data can be made quantitative. Indeed, the Kaplan-Meyer graphs do show nice quantitative results. Unfortunately, in the text few quantitative measurements are reported. In fact, the Kaplan-Meyer graphs can be interpreted in a quantitative manner such as probability of residency time. Although in most cases the statistical significance between two conditions can be expected, this is not formally calculated and shown. What is the 50% survival time of Cse4 alone or with Ndc10, for instance? This manuscript would greatly benefit from a quantitative approach to the data, including a summary table of the results of the various conditions tested. Please add this important table.

We initially put the quantitative data in the figure legends but omitted it from the main text for simplicity but appreciate the Reviewer’s point. We note that we performed log-rank tests on all Kaplan-Meier analyses that are plotted on the same graph to provide statistical differences where applicable and have included all P-values in the figure legends. In response to the suggestion, we have now also included a table (Table 1) that contains the median survival time for all proteins tested as well as the median survival times for the differing contexts tested for quick reference and easier comparison for the reader.

• This reviewer would expect that the endpoint (90 min) would roughly reflect the occupancy results from time-lapse (45 min) experiments. Based on the data presented in Figures 1, 2, S1-3, this does not appear to be the case. 50% of Cse4-GFP and 70% Ndc10-mCherry colocalized with CEN3 DNA in the endpoint experiment, whereas in Fig 2C, ~30 and ~52 traces were positive for Cse4-GFP and Ndc10-mCherry, resp. with the former having drastically lower residency survival compared to Ndc10-mCherry. If indeed, 50% of Cse4-GFP makes it to the endpoint, about 50% of all traces should reach the end of the 45 minutes time-lapse window. Only about 1/3 of all positive Cse4-GFP traces can be seen at the end of the 45 min window. Could this be due to technical issues of photostability of GFP? Why does the colocalization of Cse4 signal with the DNA signal disappear so readily? Are Cse4 so unstable? Is the same true for canonical H3 nucleosomes? This unlikely true for nucleosomes in cells.

This is a valid concern, and we appreciate the thoughtful critique. The inconsistency noted between the initial endpoint colocalization and those reported later in the paper is mainly due to the difference between yeast strains carrying Cse4 tagged alone in comparison to multiple kinetochore proteins with tags that exhibit mild genetic interactions. This point is now explained in the revised text (p. 8, line 29-p. 9. line 3).

Photostability is also a factor in the live imaging experiments compared to the endpoint localization assays. However, our photobleaching estimates suggest that the Cse4 lifetimes are not limited by photobleaching (Figure EV1D, E) so we do not believe that accounts for the differences between experiments and it is mainly the presence of multiple epitope tags.

In regard to why Cse4 is not more stable, Reviewer 1 had the same question so we performed an experiment to address whether the lysate contains negative regulatory factors. We found that Cse4 is stable once the lysate is removed (Figure EV1B, C), consistent with the idea that there are factors that disrupt it in the lysate. We discuss these potential reasons for transient association in the revised text (p. 9, lines 4-25).

It should also be noted that there are clear differences in nucleosome formation in reconstitutions and within our extracts, as evident by the Widom-601 DNA data (Figure 6D). This was not necessarily unexpected, as extracts are a much more complex medium, but we are encouraged by the fact that at least once formed, these Cse4-containing particles on CEN DNA are perhaps more stable than their reconstituted counterparts that seem to be so far unsuitable for structural studies.

Along the same lines, in Suppl Fig 3 there is a disconnect between residency survival and endpoint colocalization on either CEN3, CEN7, or CEN9. What could be the underlying mechanism between the discordance of endpoint results and time-lapse results? Could this be the result of experimental differences?

We are grateful that this discrepancy was highlighted to us, as upon closer examination we discovered that endpoint colocalization analysis had not been correctly updated in the figure to include data from equivalent genetic backgrounds as the CEN3 and CEN9 assays. Updating the figure in Appendix Figure S2 to include this data remedied this discrepancy.

• What fraction of particles show colocalization between Cse4-GFP and Ndc10-mCherry? What fraction of occupancy time show colocalization between Cse4-GFP and Ndc10-mCherry? Altogether, understanding the limitation and benefits of endpoint analysis and time-lapse analysis in this particular experimental set-up is important to be able to interpret the results. Please clarify these points.

We have now added particle numbers to all survival estimate plots which makes it much easier for the reader to interpret the proportion of Cse4 residences that are ternary vs. non-ternary in instances where off-rates were quantified and Kaplan-Meier analysis was performed on the resulting lifetimes. We determined that for ternary Cse4-Ndc10 residences, Cse4 and Ndc10 co-occupied the CEN DNA an average of ~84% of the total Cse4 residence.

• Page 9, third sentence of third paragraph it is stated that the "results suggests that Scm3 helps promote more stable binding of Cse4 ...". This is indeed one possible explanation of the results, and this possibility is tested by overexpressing Psh1 or Scm3 by endpoint colocalization analysis. 1) Taking the concerns regarding the endpoint vs time-lapse results into account, wouldn't it be more accurate to compare either time-lapse results against each other or endpoint results? 2) Alternatively, more stable Cse4 particles are able to recruit Scm3, simply because of the increased binding opportunity of a more stable particle. In this scenario, just the residency time of Cse4 alone is the predicting factor in likelihood to associate with Scm3. To test the latter possibility, Cse4 stability would need to be altered. Please consider this experiment- should be relatively easy with the right mutant of either CSE4 or CDEII (see Luger or Wu papers).

We appreciate the points raised here and addressed both as follows. For point (1) we altered the text in the third paragraph of the section, The conserved chaperone Scm3HJURP is a limiting cofactor that promotes stable association Cse4CENP-A with the centromere, to make it more clear to the reader that in the experiments presented in Figure EV4, endpoint analysis results were only compared to each other, and likewise time-lapse experiments were only compared to each other for each genetic background. While the results were consistent between experiments, we did not directly compare results from one to the results of another, but instead we used both assays to characterize Cse4CENP-A behavior more completely in differing contexts.

To test the alternative hypothesis proposed in point (2), we sought to avoid potential selection bias by utilizing off-rate analysis, which enabled us to separate the portions of Cse4 residences that occurred prior to ternary association with either Ndc10 or Scm3. This unbiased approach allowed us to compare Cse4 residence lifetimes pre and post ternary association and we found that there were still significant differences in off-rates and median lifetimes of the associated ternary and non-ternary residences using this updated analysis. We thank the reviewer for helping to guide us towards this more robust analysis.

Based on the recommendation in point (2), we also sought to directly compare the behavior of Cse4 and Scm3 on the “Very Unstable” CDEII mutants described in the section, DNA-composition of centromeres contributes to genetic stability through Cse4CENP-A recruitment. In this case, equivalent extracts were used and Cse4 stability was altered directly via the DNA template. When the off rates of ternaryScm3 Cse4 residences were compared, we found a significant increase in off-rates of Cse4 on the “Very Unstable” CDEII mutant CEN DNA (Appendix Figure S3B) compared to WT CEN DNA. If the alternative hypothesis proposed in point (2) were true, we would expect this reduction in median lifetime to correlate with a lower proportion of Cse4-Scm3 ternary association but quantification yielded proportions that, while varied, were not on average lower than the proportion of Cse4-Scm4 association on CEN3 DNA (.23 vs .31, Appendix Figure S3A). This finding, taken together with the fact that it would be difficult for us to propose an alternative hypothesis that explains the results outlined in Figure EV4, supports our hypothesis that Scm3 helps promote more stable binding of Cse4 and that this stabilization is directly influenced by DNA sequence composition.

• In Figure 1C and Supplemental Figure 5B, there appears to be foci that CEN3-ATTO-647 positive, but Cse4-GFP negative and visa verse. It seems logical that there are DNA molecules that didn't reconstitute Cse4 nucleosomes. But how can there be Cse4-GFP positive foci without a positive DNA signal? Is it possible that unlabeled DNA make it onto the flow chamber? If so, can these unlabeled DNA be visualized by Sytox Orange for instance to confirm that no spurious Cse4 deposition occurred? Please clarify.

Because it is unlikely that random associations will colocalize with the labeled DNA based on control assays (Supp. Figure 1C) that show this occurs rarely (

• On page 10, at the end of the first paragraph, growth phenotype of pGAL-SCM3 and pGAL-PSH1 mutants were tested. On GAL plates, pGAL-PSH1 shows reduced growth, but not pGAL-SCM3. Wouldn't a more accurate conclusion be that Psh1 is limiting stable centromeric nucleosome formation, instead of Scm3?

The growth defects on galactose don’t necessarily mean that a factor is limiting in cells. Instead, they report on whether changing the relative amounts of the complex lead to phenotypes in cells that could be the result of many causes that would require characterization of the phenotypes to understand. In this case, we presume that Psh1 titrates Scm3 away from Cse4 to prevent nucleosome formation in vivo. However, we have not directly tested this so we just concluded that the relative levels of the complexes are important for cell growth.

• In the section where DNA was tethered at either one or both ends, an important control is missing. How does such a set-up impact nucleosome formation in general. Can H3 nucleosomes form on random DNA that is either tethered at one or both ends? Does this potentially affect the unwrapping potential/topology of AT-tract DNA? Please comment.

This is an interesting point and one that we hope to explore further in the future but was beyond the scope of this paper. We suspect that restrictions via tethering would also limit canonical nucleosome formation on random DNA. We envision that unwrapping may be affected as well and hope to explore this via other, potentially better suited techniques like optical tweezers.

Minor comments + Censored data points are not explained in the text.

A brief explanation of censorship was added to the figure legends and we have now provided a clear explanation of the term “censored” in the text on p. 28, lines 25-27. It has also been reported in the Statistical tests section of the methods section to address this point.

• Number of particles tested should be reported in the main and supplemental figures, not just the legends for those readers who first skim the manuscript before deciding to read it.

We add these values to all Kaplan-Meier plots in all figures (Main, Expanded View, and Appendix)

• Typo on page 5, first line: "nucleosom" should be "nucleosome".

We fixed this in the text.

• Typo on page 9, second line: sentence is missing something "... is required for Scm3-dependent ..."

We fixed this in the text.

• It is unclear how the difference in Supplemental figure 5D was calculated.

We included log-rank test generated P-values as well as description in the figure legend of EV4.

• Figure 4C: why are there more Ndc10-mCherry foci observed in double tethered constructs vs single tethered constructs?

There can be variances in DNA density between slides, particularly with non-dye labeled DNA template. We updated figure panel C to include a representative image with similar Ndc10 density.

• For the display of individual traces as shown in Fig 2B, 3A, 4E, and 5E, it might be more informative if the z-normalized signal and the binary read-out are shown in separate windows to better appreciate how the z-normalized signal was interpretated.

Due to spacing limits within figures we attempted to accommodate this by reducing the thickness of the binary read-out and ensured that the raw data traces were overlaid for easier interpretation by the reader.

• Page 17, fifth line of the second paragraph, it is stated that a conserved feature of centromeres is their AT-richness. This is most certainly true for the majority of species studied thus far, but bovine centromeres for instance are about 54% GC rich. Indeed, Melters et al 2013 Genome Biol showed that in certain clades centromeres can be comprised of GC-rich sequences. It might be worthwhile to nuance this statement.

We have updated the text to reflect that AT-rich DNA is widely conserved but not a universal feature of centromeres.

• Page 17, last paragraph. Work by Karolin Luger and Carl Wu is cited in relationship to AT-rich DNA being unfavorable for canonical nucleosome deposition. A citation is missing here: Stormberg & Lyubchenko 2022 IJMS 23(19): 11385. Also, the first person to show that AT-tracts affect nucleosome positioning are Andrew Travers and Drew. This landmark work should be cited.

We thank the Reviewer for noticing this and have added the appropriate citations.

• Page 18, 9th line from the top, it is stated that yeast centromeres are sensitive to negative genetic drift. This reviewer is of the understanding that genetic drift is a statistical fluctuation of allele frequency, which can result in either gain or loss of specific alleles. Population size is a major factor in the potential power of genetic drift. The smaller a population, the greater the effect. Budding yeast is found large numbers, which would mean that drift only has limited predicted impact. Maybe the authors meant to use the term purifying selection?

We appreciate this clarification and agree with the reviewer, we have updated the manuscript to cite purifying selection and not genetic drift as at centromeres.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

The manuscript "Single molecule visualization of native centromeric nucleosome formation reveals coordinated deposition by kinetochore proteins and centromere DNA sequence" by Popchock and colleagues describes a new high-throughput single-molecule technique that combines both in vitro and in vivo sample sources. Budding yeast centromeres are genetically defined centromeres, which makes them ideal for studying short DNA segments at the single-molecule level. By flowing in whole cell lysates, Cse4 nucleosomes can form under near physiological conditions. Two analytical experiments were performed: endpoint and time-lapse. In the former case, nucleosomes were allowed to form within 90 minutes and the latter case, nucleosome formation was tracked for up to 45 minutes. In addition, well described genetic mutants were used to assess the stability of Cse4 nucleosomes, as well as different DNA sequences (this we particularly liked- well done). Overall, this is a very interesting technique with potential to be useful for studying any DNA-based effect, ranging from DNA repair to kinetochore assembly. This is strong and impactful work, and the potential this kind of microscopy has for solving kinetic problems in the field. We think it's worthy of publication after revising technical and experimental concerns that would elevate the ms significantly.

Major comments:

Statistics are highly recommended for all the data in the ms.

• At what rate is data collected in the TIRFM setup. For clarity for the reader, it is important to provide imaging details for time-lapse. What is the impact of photobleaching on the stability of the fluorophore signal? Please clarify.
• The power of single-molecule technique is precisely that such data can be made quantitative. Indeed, the Kaplan-Meyer graphs do show nice quantitative results. Unfortunately, in the text few quantitative measurements are reported. In fact, the Kaplan-Meyer graphs can be interpreted in a quantitative manner such as probability of residency time. Although in most cases the statistical significance between two conditions can be expected, this is not formally calculated and shown. What is the 50% survival time of Cse4 alone or with Ndc10, for instance? This manuscript would greatly benefit from a quantitative approach to the data, including a summary table of the results of the various conditions tested. Please add this important table.
• This reviewer would expect that the endpoint (90 min) would roughly reflect the occupancy results from time-lapse (45 min) experiments. Based on the data presented in Figures 1, 2, S1-3, this does not appear to be the case. 50% of Cse4-GFP and 70% Ndc10-mCherry colocalized with CEN3 DNA in the endpoint experiment, whereas in Fig 2C, ~30 and ~52 traces were positive for Cse4-GFP and Ndc10-mCherry, resp. with the former having drastically lower residency survival compared to Ndc10-mCherry. If indeed, 50% of Cse4-GFP makes it to the endpoint, about 50% of all traces should reach the end of the 45 minutes time-lapse window. Only about 1/3 of all positive Cse4-GFP traces can be seen at the end of the 45 min window. Could this be due to technical issues of photostability of GFP? Why does the colocalization of Cse4 signal with the DNA signal disappear so readily? Are Cse4 so unstable? Is the same true for canonical H3 nucleosomes? This unlikely true for nucleosomes in cells. Along the same lines, in Suppl Fig 3 there is a disconnect between residency survival and endpoint colocalization on either CEN3, CEN7, or CEN9. What could be the underlying mechanism between the discordance of endpoint results and time-lapse results? Could this be the result of experimental differences?
• What fraction of particles show colocalization between Cse4-GFP and Ndc10-mCherry? What fraction of occupancy time show colocalization between Cse4-GFP and Ndc10-mCherry? Altogether, understanding the limitation and benefits of endpoint analysis and time-lapse analysis in this particular experimental set-up is important to be able to interpret the results. Please clarify these points.
• Page 9, third sentence of third paragraph it is stated that the "results suggests that Scm3 helps promote more stable binding of Cse4 ...". This is indeed one possible explanation of the results, and this possibility is tested by overexpressing Psh1 or Scm3 by endpoint colocalization analysis. 1) Taking the concerns regarding the endpoint vs time-lapse results into account, wouldn't it be more accurate to compare either time-lapse results against each other or endpoint results? 2) Alternatively, more stable Cse4 particles are able to recruit Scm3, simply because of the increased binding opportunity of a more stable particle. In this scenario, just the residency time of Cse4 alone is the predicting factor in likelihood to associate with Scm3. To test the latter possibility, Cse4 stability would need to be altered. Please consider this experiment- should be relatively easy with the right mutant of either CSE4 or CDEII (see Luger or Wu papers).
• In Figure 1C and Supplemental Figure 5B, there appears to be foci that CEN3-ATTO-647 positive, but Cse4-GFP negative and visa verse. It seems logical that there are DNA molecules that didn't reconstitute Cse4 nucleosomes. But how can there be Cse4-GFP positive foci without a positive DNA signal? Is it possible that unlabeled DNA make it onto the flow chamber? If so, can these unlabeled DNA be visualized by Sytox Orange for instance to confirm that no spurious Cse4 deposition occurred? Please clarify.
• On page 10, at the end of the first paragraph, growth phenotype of pGAL-SCM3 and pGAL-PSH1 mutants were tested. On GAL plates, pGAL-PSH1 shows reduced growth, but not pGAL-SCM3. Wouldn't a more accurate conclusion be that Psh1 is limiting stable centromeric nucleosome formation, instead of Scm3?
• In the section where DNA was tethered at either one or both ends, an important control is missing. How does such a set-up impact nucleosome formation in general. Can H3 nucleosomes form on random DNA that is either tethered at one or both ends? Does this potentially affect the unwrapping potential/topology of AT-tract DNA? Please comment.

Minor comments

• Censored data points are not explained in the text.
• Number of particles tested should be reported in the main and supplemental figures, not just the legends for those readers who first skim the manuscript before deciding to read it.
• Typo on page 5, first line: "nucleosom" should be "nucleosome".
• Typo on page 9, second line: sentence is missing something "... is required for Scm3-dependent ..."
• It is unclear how the difference in Supplemental figure 5D was calculated.
• Figure 4C: why are there more Ndc10-mCherry foci observed in double tethered constructs vs single tethered constructs?
• For the display of individual traces as shown in Fig 2B, 3A, 4E, and 5E, it might be more informative if the z-normalized signal and the binary read-out are shown in separate windows to better appreciate how the z-normalized signal was interpretated.
• Page 17, fifth line of the second paragraph, it is stated that a conserved feature of centromeres is their AT-richness. This is most certainly true for the majority of species studied thus far, but bovine centromeres for instance are about 54% GC rich. Indeed, Melters et al 2013 Genome Biol showed that in certain clades centromeres can be comprised of GC-rich sequences. It might be worthwhile to nuance this statement.
• Page 17, last paragraph. Work by Karolin Luger and Carl Wu is cited in relationship to AT-rich DNA being unfavorable for canonical nucleosome deposition. A citation is missing here: Stormberg & Lyubchenko 2022 IJMS 23(19): 11385. Also, the first person to show that AT-tracts affect nucleosome positioning are Andrew Travers and Drew. This landmark work should be cited.
• Page 18, 9th line from the top, it is stated that yeast centromeres are sensitive to negative genetic drift. This reviewer is of the understanding that genetic drift is a statistical fluctuation of allele frequency, which can result in either gain or loss of specific alleles. Population size is a major factor in the potential power of genetic drift. The smaller a population, the greater the effect. Budding yeast is found large numbers, which would mean that drift only has limited predicted impact. Maybe the authors meant to use the term purifying selection?

Significance

This study developed an in vitro imaging technique that allows native proteins from whole cell lysates to associate with a specific DNA sequence that is fixed to a surface. By labeling proteins with specific fluorophore-tags colocalization provides insightful proximity data. By creating mutants, the assembly or disassembly of protein complexes on native or mutated DNAs can therefore be tracked in real time. In a way, this is a huge leap forward from gel shift EMSA assays that have traditionally been used to do comparative kinetics in biochemistry.

This makes this technique ideal for studying DNA binding complexes, and potentially, even RNA-binding complexes. This study shows both the importance of using genetic mutants, as well as testing the effects of the fixed DNA sequence. As many individual fixed DNA molecules can be tracked at one, it allows for high-throughput analysis, similar to powerful DNA curtain work from Eric Greene's lab. Overall, this is a promising new single-molecule technique that combines in vitro and ex vivo sample sources, and will likely appeal to a broad range of molecular and biophysics scientists.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Centromeres drive chromosome inheritance from one cell generation to the next, so understanding their nature is of utmost importance in biology. The assembly of centromeric chromatin is of outstanding interest since defects in this step impair faithful genetic inheritance. Popchock et al investigate the molecular mechanisms of Cse4(CENPA) deposition and stabilization on a native budding yeast centromeric DNA. Using TIRMF (Total Internal Reflection Fluorescent Microscopy) enabled single molecule visualization of de novo Cse4(CENPA) nucleosome formation from a yeast cellular extract. The centromeric DNA used in this study was derived from PCR amplification of plasmids containing the native yeast point centromere from chromosome 3 (CEN3 DNA) sequences containing the CDEI, CDEII, CDEII, flanking pericentromeric DNA and linker DNA totaling ~750bp. In the PCR preparation, the CEN3 DNA contains a 5'-fluorescent dye and a biotinylated 3'-end, and was tethered to a functionalized (streptavidin) slide for TIRFM. The yeast extract was extracted from cells arrested in mitosis. This system is a novel application of a single molecule to study Cse4(CENPA) formation on centromeric DNA.

Using this system, the authors observed coordinated Cse4(CENPA) deposition on the CEN3 DNA, reporting inherent, transient colocalization of Cse4(CENPA) with CEN3 DNA. Stable Cse4(CENPA) colocalization on the CEN3 DNA is correlated with the Cse4(CENPA) chaperone Smc3(HJURP), and an ability for nucleosome formation on the CEN3 DNA. Further stabilization of Cse4(CENPA) was shown to depend on the DNA binding CCAN protein chl4(CENPN) and okp1(CENPQ) which dimerizes with the DNA/CENPA binding Ame1(CENPU).

Using this single molecule system they also demonstrated a role for the A/T run (>4) content in the CDEII as specifically important for Cse4(CENPA) deposition on CEN3 DNA. Cse4(CENPA) colocalization was preferred on the native CDEII sequence, relative to mutant CDEII sequences with similar A/T content but variable homopolymeric runs.

Major comments:

1. Figure S1A-D seem like some of the most compelling data in the paper to bolster the rigor of their experimental setup. There appears to be plenty of space to include these data in the main figure set in Figure 1 after panel D. The authors would be well served to consider moving S1A-D somewhere in the main figure set.
2. The authors conclude that Ndc10 recruits HJURP(Smc3) to the yeast point centromeres. If this is the case, can the TIRFM assay measure ternary residence lifetimes complexes between Ndc10/HJURP(Smc3)/CenDNA?
3. Throughout the manuscript short- and long- term residence lifespans are mentioned, referencing the figures containing lines with lengths depicting residence times. This is a qualitative reference to short and long residences. Can the authors provide a graph for short-term (<120 s) and long-term (> 300 s) residence life-spans for, CENPA alone, CENPA/Ndc10, and CENPA/HJURP on CEN3 DNA? Or some figure similar to Figure 3C, but reporting the proportion of short-term vs long-term residence?
4. The choice of CCAN components for analysis in Fig. 5 is interesting, but many readers may be curious why Mif2 wasn't selected for disruption, since it has such a cozy placement with CENP-A and CEN DNA. Can choice to not include Mif2 mutants/degrons be mentioned/justified in the text (unless, even better yet, they choose to address Mif2 role directly in new experimentation)?

Minor comments:

1. Are these whole cell extracts (WCE) DNA-free? I'm curious if there is any competition from endogenous DNA from the yeast cellular extract.
2. In relation to Mif2 and comment #4 above, do the authors make any connection to their results with synthetic nucleosome sequences not being conducive to yeast centromere formation with the prior observation (Allu et al 2019) using recombinant components that the human version of Mif2 more easily saturates its binding sites on CENP-A nucleosomes when they are assembled with natural centromere DNA rather than the Widom 601 sequence?
3. Providing a gel (or other measure) of the DNA templates (750, 250, and 80 bp) used in TIRFM assay would be nice to show to confirm the designed size of the pre-tethered DNA.
4. Some of the references to figures/figure panels in the main text do not match the figures. (discussion, pg 16, paragraph 1 & 2;pg 18, paragraph 1).

Significance

This work demonstrates the dynamics of Cse4(CENPA) coordination with Smc3(HJURP) to form nucleosomes on point centromeric DNA, and the necessity for homopolymeric A/T runs. It is a truly impressive piece of work that makes sense of findings from prior genetics experiments. Then it extends the understanding and clarifies the role of both centromere proteins and DNA sequence. The quantitative and powerful single molecule-based experimentation, the high importance of the subject matter, and its connection to studies using yeast genetics, will make this work, upon modest improvements (see section A of this review), of outstanding interest to an extremely broad audience of biologist.

My relevant expertise keywords: centromeres, nucleosomes, biochemical reconstitution, chromosome engineering

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary

This work presents a novel experimental setup to explore native centromeric nucleosome formation at a single molecule level with a high degree of temporal resolution. It combines TIRF microscopy with an immobilized fluorescent yeast CEN3 DNA sequence. Binding of centromere proteins was probed by using whole-cell lysates produced from various mutant yeast strains expressing fluorescently tagged centromere proteins. This system was used to investigate the recruitment of Cse4 (yeast CENP-A), the histone H3 variant which defines centromeres, to the native yeast CDE3 sequence. Consistent with previous studies, the authors identified a requirement for Ndc10, which binds the CDE3 sequence with high efficiency, in Cse4 recruitment. Colocalization of Ndc10 with Cse4 significantly increases the stability of the Cse4-CDE3 interaction, and Ndc10 binding predominantly precedes Cse4 recruitment. Secondly, they investigated the impact of Scm3 (yeast equivalent of HJURP) on Cse4 stability and found that Scm3 is not required for transient short-term associations between the DNA and Cse4 but is needed for longer more stable interactions between Cse4 & the CDE3 sequence. Overexpression of Scm3 & its negative regulator Psh1 also showed that Scm3 availability is the rate-limiting factor in stable Cse4 associations with DNA. However, these interactions were abolished when the CDE3 DNA sequence was topologically restricted by tethering both ends to the streptavidin-coated coverslip suggesting that the formation of a nucleosome is essential for stable binding of Cse4 to CEN3. The sequence-specific DNA binding kinetochore proteins ChI4 & Okp1 were also found to improve the lifetime of Cse4-CDE3 interactions suggesting a stabilizing interaction. Finally, the impact of A/T runs on nucleosome formation was examined using a library of scrambled CDEII sequences, with approximately native levels As & Ts but shorter runs of the 2 nucleotides. Previous studies found a correlation between the length of AT runs and the incidence of chromosome missegregation. They now extend this correlation to a decreased ability of DNA to stably recruit Cse4. Finally, experiments with other sequences including the Wisdom601 sequence suggest that A/T runs along with CBF3 binding inhibit the recruitment of histone H3 to the CDEII sequence which is overcome by the interactions between Cse4 and Scm3.

Assessment

This is an insightful, well-executed study of a high technical standard. While several observations confirm or further validate previous findings in vivo (such as the requirement for Ndc10 and Scm3 for Cse4 assembly), this work adds a better understanding of the dynamics of Cse4 requirement and the role of Scm3 in assembly and the role of CCAN components in stabilizing Cse4. Figures are well laid out and methods are clearly described. In general, the claims are supported by the data. The development of their single-molecule setup to assay Cse4 nucleosome assembly is a promising tool for future work and for this reason work reporting. In short, I believe this paper contains exciting developments in the understanding of the specific mechanism and temporal dynamics of the formation of a Cse4 nucleosome on its native DNA and the interactions which underpin its stability in vitro and in vivo.

Major comments

I have several comments and suggestions for further analysis of the data. I also have suggestions for additional experiments, but I would stress that none of these are essential for publication.

Figure 2E. This shows the residence time of Cse4 without Ndc10 association. How does this compare to the residence time on mutant CEN3 (Supplemental Figure 1). It looks like Cse4 still binds to CEN3 with some specificity even in the absence of Ndc10. Does this suggest that Cse4 has some intrinsic ability to recognize CEN3? Alternatively, Ndc10 is still required for Cse4 binding but is below detection in the Cse4-alone residence lifetimes. Ideally, the authors would compare this with Cse4 binding in an Ndc10 mutant.

Figure 3 explores the very interesting relationship between Scm3 dynamics and Cse4 binding but I feel that the data is not fully flushed out. A key finding is that Cse4 can potentially bind to CEN DNA prior to engaging with Scm3 to be incorporated. This predicts that Cse4 should bind first and then colocalize with Scm3. Can this be detected in the timing of the traces? How often does Scm3 bind to already prebound Cse4 and does this correlate with Cse4 residing longer?

A related and perhaps even more interesting point is whether Scm3 is involved in "loading" of Cse4. If so, then one would expect that once Cse4 is assembled into nucleosomes it should be stable, even if Scm3 leaves. Can the authors extract this from the data? Alternatively, it is possible that Scm3 remains associated to Cse4 to maintain the nucleosome which would imply a more extended role for Scm3 apart from assembly alone. It would be interesting if information on this can be extracted from the data.

Even in the presence of Scm3 and CCAN components, Cse4 appears to have a limited lifetime in the in vitro assay compared to in vivo stability. The authors should speculate on whether activities exist in their extract that actively disassembles nucleosomes. Perhaps ATP could be depleted to inactivate remodellers?

For Figure 6, it is not clear why AT-track mutants of CDEII are labeled as genetically stable and genetically unstable. This is confusing as the "genetically stable" show a more than 10-fold increase in chromosome loss rates. Perhaps these can be renamed into "unstable" and "very unstable" or "weak" and "strong" mutants, just to make clear that these classes are both poorer than wild type.

Finally, it would be wonderful to include data to assess whether a full Cse4 nucleosome is assembled or a partial nucleosome e.g. just Cse4/H4 tetrasomes. This could be done by tracking the accumulation of H2A or H2B at the CEN3. This would give further insight into what step Scm3 catalyses.

Minor comments:

Typo on page 5, line 1 "nucleosom" missing an e.

Kaplan-Meyer should be spelled Kaplan-Meier

The term "censored" is mentioned across many figures but comes up just ones in the methods where it is not clearly explained. Perhaps this could be clarified in the legend.

The abstract states that Cse4 can arrive at the centromere without its chaperone. More likely, Cse4 is in complex with other chaperones that may allow it to bind. Perhaps the abstract can be modified to read "Cse4 can arrive at the centromere without its dedicated centromere-specific chaperone Scm3..."

Related to this point, the discussion states the possibility that Cse4 can initially bind to CEN3 via other more general chaperones. However, it should be acknowledged that transient Cse4 binding in their assay may simply occur through mass action due to high concentrations of CEN3 DNA. In vivo, this transient binding may not be that relevant.

Significance

This paper offers new insight into the assembly of yeast centromeres with a focus on the role of the Chaperone Scm3 in the assembly of the centromere-specific histone Cse4. This is still a poorly understood process and the authors offer an elegant in vitro system to study this and have presented new insights. For this reason, this is study is of interest to a broad readership in the area of mitosis and chromosome structure. The advance is strong at the technical level but also new insight is provided particularly in the role of Scm3 and the nature of centromeric DNA in centromeric chromatin assembly. Overall a strong, high-quality paper.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

The authors do not wish to provide a response at this time.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary

In this manuscript, Wang and colleagues demonstrate that a single systemic injection of a high dose of Akkermansia muciniphila (A.m.) lysate drives a rapid pancytopenia followed by prolonged anaemia and hepatosplenomegaly with late-onset extramedullary hematopoiesis (EMH). The latter, as well as the splenomegaly, were likely mediated through activation of pattern recognition receptors and IL-1R signalling pathways. This was demonstrated through the partial and full phenotype reversal in Tlr2;4-/- and MyD88;Trif-/- mice, respectively. Moreover, the phenotype was partially reversed following IL-1R antagonism. After performing multiplex protein assays and flow cytometry, the authors conclude that EMH, in this model, is mediated by IL-1a produced within the spleen by local monocytes and DC.

Overall, the manuscript by Wang et al. is quite well presented, the experiments are mostly well controlled, the methods are well reported, and the data fit a clearly defined story with clinical relevance. Nevertheless, there are several major concerns that if addressed would greatly increase the strength of the authors conclusions.

Major comments

1. The "two wave" hypothesis of hematopoiesis - first in the bone marrow (BM) and then in the spleen - is interesting. However, although an early wave of BM hematopoiesis would make sense, under these circumstances, I don't think the data are strong enough to support this hypothesis as they stand. For example, although the frequency of LSK cells increase, the numbers of most LSK subsets decrease. Given the decrease in the absolute number of BM cells 1d after A.m. injection, isn't it possible that the LSK cells are only proportionally increased relative to the remaining Lin- cells? What happens to the absolute number of LSK cells following A.m. injection?

Also, describing "two distinct waves of HSPC increase in the A.m.-injected spleen" (Fig 2 & S2 titles) and describing a "first wave" of HSPC expansion in the BM (lines 396, 399, 402 etc.) is misleading for the following reasons: (i) the data strongly support a single wave of increasing HSPC in the spleen, peaking at d14, and (ii) there is no evidence HSPC are increased in the BM until d56, although there does appear to be an early increase in MPP. The language should be changed accordingly. 2. The flow cytometry panel is not comprehensive enough to fully characterize the mature hematopoietic cell populations to the levels that are claimed here. For example, it is a stretch to assume that all B220- CD3- CD11c- cells are DC (splenic NK cells, eosinophils, monocytes and red pulp macrophages, for example, can express CD11c, particularly following inflammatory insult), or that CD11b+ F4/80+ SSC-hi cells are eosinophils, especially when eosinophils should be F4/80-lo are not known to express Ly6C in the spleen (For reference, see Immgen). These gating issues may explain the conspicuous absence of macrophages (should be F4/80+CD11b+Ly6C- and would also have a higher SSC than monocytes) in the plots. The B cell gate will also contain PDC, which express B220 (but can be easily excluded using Ly6C and CD11c). With respect to assessing the mature leukocyte populations in the spleen, relabelling the gates (CD11c+ cells instead of DC, F4/80+ myeloid cells instead of eosinophils) would suffice, however, these issues become a problem when trying to identify which cell populations express IL-1a.

Due to the limited antibody panel used here, there is not enough evidence to suggest that DC and monocytes are producing IL-1a. Moreover, the histograms showing the changes in expression of IL-1a on the "DC" and "Mo" are not very convincing. How does the IL-1a staining look on a dot plot? Is there good separation between positive and negative? These plots need to be included. What happens if you gate on the IL-1a+ cells first, then phenotype them?

Macrophages and splenic stromal cells are also likely candidates for IL-1a production. To assess which cell types are the true source of IL-1a, the authors need to repeat these experiments (namely, injecting A.m. and assessing IL-1a expression by leukocytes (and ideally also mesenchymal cells)) at d1 and d14, using a more comprehensive panel. Consider adding MHCII, CD64, Siglec F and CD24 to help differentiate between DC, MF, eosinophils and monocytes. CD45+ vs CD45- could be used as a minimum to assess the expression of IL-1a on leukocytes vs. stroma.

OPTIONAL: The mechanism could be better defined using bone marrow chimeras to assess the different contribution of TLR2/4 signalling and IL-1R signalling on the hematopoietic vs. mesenchymal cell compartments. 3. From these experiments, it is difficult to fully rule out a contribution from the adaptive immune system to the splenomegaly phenotype due to the marked difference in the size of BALB/c and MSTRG spleens at steady state. The authors should show the differences in spleen weight and total cell number as a % increase from control. The no of HSPC should also be normalized per gram of tissue weight or represented as a fold change compared to the relevant control groups. 4. When using fluorescent imaging to compare the abundance of HSPC and other cell populations in the spleen, the authors should provide absolute quantification from multiple FOV and multiple mice. 5. Finally, although the experiments are adequately replicated, the stats are not always appropriate. For example, a t-test shouldn't be used when there are >2 groups, or for a time course. This needs to be amended.

Minor comments

• Line 82-83: I'm fairly certain monocytes and inflammatory Ly6Chi cells are the same thing.
• Line 83-84: "IL-1a is crucial for sustaining inflammatory responses, recruiting myeloid cells to infected tissue and inducing hematopoietic stem and progenitor cell (HSPC) mobilization and expansion both in vitro and in vivo" - I don't believe IL-1a has been shown to be crucial for either, even if it has been shown to play a role. If I am mistaken, please reference with a manuscript showing relevant phenotypes using KO mice.
• Line 214: "Thus, we decided to use 200ug of lysate for the rest of all experiments." - is this what was usen for Figures 1A-C? This is not mentioned anywhere.
• Line 227: "containing both non-hematopoietic cells and immature HSPCs" Please reference Fig. 1H here. Otherwise, it is unclear how you identified the "HSPC and other cell types" in Fig. 1G.
• Figure S2A is described in text before Supp 1I-O and Fig S1H is not referenced in text at all.
• It would be interesting to include what happens to hepatomegaly in MSTRG, Tlr2;4-/- and MyD88;Trif-/- mice.
• Please define WBM. Presumably whole bone marrow?
• Notably, CCL2 is increased in spleen lysate, BM lysate and serum. Given is role in myeloid cell mobilization from the BM, I would expect its role in the phenotype described here to at least be discussed.
• HSPC LT gate includes MPP1, and should be labelled as such.

Significance

General assessment: The manuscript provided by Wang et al. describes, for the first time, a prolonged anaemia and hepatosplenomegaly with late-onset extramedullary hematopoiesis following a single systemic injection of A.m. lysate. The EMH phenotype appears robust and the data implicating TLR-signalling and IL-1a production are compelling. The work has clinical relevance as it increases our understanding of the factors driving EMH.

There are two key limitations that let this study down. Firstly, the lack of depth in the flow cytometry panel used for immunophenotyping means it is not at all clear which cell types are producing IL-1a. Secondly, the authors use an enormous dose of bacterial lysate that is well above physiological levels, even following a loss of barrier integrity (e.g., in patients with IBD). This makes me question the biological relevance of the study, particularly with respect to Akkermansia translocation.

Advance: With some improvement, this study will advance the field, in general. Previous work has looked at EMH following LPS injection, or live E. coli infection, however; the authors are able to demonstrate a distinct Akkermansia-specific effect that differs to that of LPS, membrane components of a different gram-negative bacteria, B. theta. The advancements implicate IL-1a in the modulation of EMH, for the first time, providing some mechanistic insight into this phenomenon.

Audience: This work will likely be of interest to basic researchers interested in EMH. It may also be of interest to clinical researchers of pathologies where EMH is a known complication, such as rheumatoid arthritis and cancer. The impact of the work will depend on whether or not EMH contributes to pathogenesis, or is an epiphenomenon. To my knowledge, this has not been fully established, although this is not my area of research.

I am a basic researcher with expertise in immunology focused on host-microbe interactions, both within the intestine and at distal tissues. I have knowledge of BM hematopoiesis and the microbial factors that influence if although my knowledge on extramedullary hematopoiesis is limited.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

In this study, Wang et al reported single injection of Akkermansia muciniphila (A.m.) induces two waves of extramedullary hematopoiesis (EMH), and demonstrated the mechanism of "second wave" was contributed by IL-1α secretion in spleen DCs and Ly6C+ monocytes. It is an important work on understanding infection-induced EMH. However, several major concerns about claims in this manuscript need to be addressed.

1. The authors demonstrated that A.m.-induced EMH were alleviated by knockout of Tlr2;4 or Myd88-Trif, or even IL-1R inhibition. EMH in the spleen is a mechanism by which the hematopoietic system responds to stresses. Therefore, whether inhibition of EMH by these ways can affect normal hematopoiesis in mice? Do mice have pancytopenia? Will the function of HSC in bone marrow be affected?
2. In Fig.S1E, why did the WBC and PLT recover quickly after the first day, while the RBC took 14 days to recover? Are WBC and RBC regulated by two waves of EMH, respectively?
3. The authors should show absolute numbers of each cell type, not just the percentage of immunophenotypically defined cells. For example, in Fig.1G, I, 2B.
4. In transplantation assays, whole BM cells or splenocytes was use. However, the proportion of HSCs in BM and spleen were all changed post A.m injection, which could affect the outcome of chimerism rate after transplantation. Transplantations should be done on by transplanting sorted fresh immunophenotypic HSCs.
5. The concentration of IFN-γ in both spleen and serum were increased continuously from D1 to 14. Would IFN-γ cause the second wave of EMH? Relevant assays for exclusion are necessary.

Significance

It is an important work on understanding infection-induced EMH.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Wang et al. examined the mechanisms of splenic extramedullary hematopoiesis upon systemic injection of Akkermansia muciniphila in mice. They showed that components of this mucin-degrading bacterium mobilize bone marrow hematopoietic cells and induce splenomegaly by MYD88/TRIF-dependent innate immune signaling pathways. Activation of TLRs and release of interleukin 1-alpha from splenic cells were then responsible for the expansion and differentiation of functional hematopoietic progenitors in the spleen. Genetic deletion of TLR2 and 4 restrained splenomegaly, while, pharmacological inhibition of IL1 receptor abrogated splenomegaly and extramedullary hematopoiesis suggesting their cooperation in the observed phenotype. It has been widely accepted that splenomegaly arises as a consequence of inflammation and that TLRs are major drivers of this process in the context of bacterial or viral infections. Here, the novelty relies on the potential circuit with the IL1alpha-IL1R axis as an additional driver of splenic extramedullary hematopoiesis. Although the results summarized above indicate that both TLRs and IL1 individually participate to some extent in splenomegaly after Akkermansia muciniphila administration, they fail to demonstrate that they concertedly do so in the spleen. In fact, the blockade of IL1R has a more profound impact.

Significance

My concerns are the following: - The authors mentioned that a specific lipid from Akkermansia muciniphila is able to trigger a non-canonical TLR2-TLR1 heterodimer to release inflammatory cytokines and regulate human immune response. Why TLR1 was not considered in the experimental strategy? - IL1alpha is up-regulated in several splenic cells (in particular in macrophages, Fig. S4F). To demonstrate a critical involvement of dendritic cells or monocytes, depletion studies or conditional mice models should be evaluated. What about megakaryocytes? Why were excluded from the analyses? - Interleukin 1alpha KO mice model should be also evaluated. - The paper is very dense and not easy to read and follow. English editing is required.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

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

The manuscript presents a detailed numerical model of blood flow in a region of the zebrafish vasculature.

The results section is quite intense and detailed. it is difficult to understand what the authors are after. I think a rewrite would beneficial. The authors present simulations for a wild type and a couple of phenotypes. For each of these they speculate on the possible adaptation mechanism leading to the discussed phenotype, as preservation of constant wall shear stress. However, the comparison between experiments and numerical simulations is really elusive as the conclusions on those mechanisms. Overall we suggest a rewrite with clearer organisation in a way that the reader is not overflown with useless details.

We thank the reviewer for the advice on the general writing standard and data organization. We have reanalyzed experiment data and interpreted the findings more conservatively for application into the simulation models. As a result, some conclusions to the results sections have changed. Accordingly, we have done a major revision of the entire Results, Discussion and Models and Methods sections in the paper to articulate these reinterpretations while removing superfluous details that may obfuscate the data.

It is not always clear what info of the experiments are used in the simulations on top of the anatomy. Our understanding is that the pressure boundary conditions are set to match the red blood cel velocity observed in experiments. Is this always the case for the three phenotypes and which vessels ?

We thank the reviewer for the question. Only WT and Marcksl1 KO have been matched for peak velocities in the CA, CV and ISVs between experiments and simulations. WT results were compared to both the experimental reference of 27 embryos in Table 3 and also to the current experiment pool of WT (5 embryos) in Table 6. Marcksl1 KO simulation models 1, 2 and 3 were compared against the average level seen in the low and moderate perfusion Marcksl1 KO phenotypes (8 embryos) from the experiment (Table 5 and Table 6). Additionally, we also have represented the similar level of RBC hematocrit in the CA for WT model to WT experiment data from the reference cited in Table 3.

In addition to the velocity comparisons, we now use the experimentally observed trend of decreased flow rate in the CA of Marcksl1 KO experiment data to assess the model boundary conditions amongst Marcksl1 KO models 1, 2 and 3 that best reflect the experimental observations:

Page 11, lines 1 to 20

The Marcksl1 OE cannot be matched because we do not have the experiment data for that, the same goes for PlxnD1 where we have no experiment flow data. These two networks represent more conceptual discussions, particularly in PlxnD1 case where we have explicitly stated in the new discussion section:

Page 15, lines 24 to 34

There are about 7 inlets and outlets where to impose pressure boundary conditions. Can the author comment on the uniqueness of this problem?

Can different combination of pressure boundary condition leading to the same result ? In how many points/vessels is the measured velocity matched ?

We thank the reviewer on this insightful concern. Indeed, the uniqueness of flow and pressure field can be a problem without careful consideration. We have tried to address this to some extent, because CA, CV are connected by the ISV and DLAV network, to match flow velocity in all regions, the pressure distribution ought to be unique to the particular setting we employed.

As shown in table 3, average systolic peak flow velocities in the entire CA and CV encompassing the 5 ISV segment domain is matched between the simulation and the population-averaged experimental data from the experimental reference (27 fish sampled in the cited reference) for the same regions in WT network. Average systolic peak flow velocities for the 10 ISVs in the simulation were matched against WT experiment population-averaged systolic peak flow velocities in arterial and venous ISVs in the same caudal region.

Additionally, we also compared the flow velocities to the experiment conducted within this study (5 WT, and embryos). This comparison data is shown in Table 6. Admittedly the discrepancy was large for CV and ISVs regions likely due to a smaller data set sampled in this study and biological variations that happen from one experiment to another. We have acknowledged this deficiency in the revised discussion section:

Page 15, lines 3 to 9

The argument that similar beating frequency in the WT and GATA1 MO suggest pressure does not change is not clear. If the heart was a volumetric pump it would impose the same flow rate, not the same pressure. It would be more useful to measure the cardiac output in terms of flow rate in the Dorsal Aorta. Previous measurements by Vermot suggested the latter would not change much in gata1 MO. It could be that the cardiac output is the same but the vasculature network is different in a way that the shear stress remain the same. It does not look like this was checked by the authors.

We thank the reviewer for this insight. In accordance with the reviewer’s suspicion, we have estimated the flow rates in the CA of gata1 MO injected embryos and found the level to be similar to WT. This supports the reviewer’s opinion that the heart rate similarity indicates cardiac output similarity and not arterial pressure similarity as we previously put forward. Furthermore, we have checked that the gata1 morphants do in fact present reduced ISV diameters. In light of this reinterpretation, we performed an additional zero hematocrit model (model 3 in section 2.1). We have consequently rewritten the entire section on how RBC hematocrit modulates hemodynamics in a microvascular network:

Page 6, line 18 to page 8 line 10.

Additionaly, it would be useful to provide an effective viscosity for the different vessels, and an effective hydraulic impedance relating DP and Q to interpret the results.

We have followed the reviewer’s advice and have analyzed for vessel hydraulic impedance and effective viscosity in all the network models presented. This is included in the main figures and discussion. The vessel impedances are discussed for the various models in these following parts of the manuscript:

Page 9, lines 20 to 29

Page 11, lines 28 to 30

Page 12, line 1 to page 13 line 10

Is the hydraulic impedance of the vessels kept constant in the smooth-geometry model? This needs clarification

The SGM diameters have been determined based on geometric averages and not impedance equivalency. The reason why we did this is because the impedance will not be known until the CFD is performed for the WT network. This is because without a pressure distribution (which cannot be determined experimentally) we cannot calculate vessel impedance since only flow can be measured and both flow and pressure are requirements to impedance calculation. Our intention with the SGM is to highlight how geometric averaging of morphological characteristics lead to incorrect flow and stress predictions. However, we understand the reviewer’s sensibility and have revised the entire section of the SGM results. We have now discussed three SGM models with varying degrees of geometry simplification. The SGM1 in the revised manuscript matches WT network impedance in the ISVs by including both axial variation in lumen diameter of the WT network and the elliptical fit representation of cross-sectional skewness seen in WT ISV lumens. SGM 2 has representation of axial variation but not luminal skewness and SGM3 has only geometric average similarity to WT ISVs. The new findings and discussion can be found in the revised manuscript here:

Page 8, line 19 to page 9 line 36.

As mentioned by the authors they propose a very complex and time expensive simulation. However the results they report are kind of intuitive. Given the availability of the experimental results, would it be useful to use a simpler red blood cell model in the future, to make their simulation more practical? Or clarify when such demanding simulations can add something new?

We agree that the intuition feedback depends on the expertise of the investigator. The boundary condition selection is intuitive from the experimental findings and key data like pressures in the network cannot be measured. Furthermore, population-averaged flow data does not always match the flow-to-geometry situations that vary from sample to sample, thus demonstrated by the high margin of prediction discrepancy for flow velocities in table 6. We have discussed these challenges and our recommendations for improvement in the Discussion section:

Page 15, lines 3 to 9

Page 15, lines 35 to 40

Page 16, lines 12 to 15

On the topic of RBC model simplification, we agree with the reviewer that our work suggests the methodology would benefit from a further coarse-graining approach to the RBC phase. Accordingly, we discussed the possibility of using a low-dimensional RBC model already published in literature:

Page 14, lines 13 to 17

The authors should check their references as this is not the first time work has been done on the topic. Would be good to have a check in the work of Freund JB and colleagues, as well as Dickinson and colleagues and Franco and colleagues to discuss how the work compares. There may be interesting work in modelling cardiac flow forces in the embryo too.

Thank you for referring us to other publications that are related to our study. To our knowledge and after performing publication search on these authors, we find that although Dickinson and colleagues performed experiments to examine the effects of perturbed blood flow on vessel remodelling (Udan et al., 2013), they did not perform any numerical modelling to calculate hemodynamic forces such as WSS and luminal pressure. Instead, changes in vessel morphogenetic process were only correlated with blood flow velocity. In our study, we attempt to quantitatively correlate WSS and pressure distributions within a vascular network. Franco and colleagues (Bernabeu et al., 2014) developed PoINet to model haemodynamic forces in mouse retina model of angiogenesis. From what we understand, PoINet is different from our 3D CFD model by 1) not having red blood cells incorporated in their model and as such, the blood viscosity prediction is modelled using shear-rate dependent formulation and not through red blood cell hematocrit, 2) cross sections of blood vessels are assumed to be circular and therefore have no irregularity and 3) live imaging of blood flow is difficult in mouse retina therefore preventing accurate boundary conditions for the model.

We have included the reference to work of Franco and colleagues:

Page 14, line 28 to line 31

Page 9, lines 12 to 14

Freund JB indeed has had extensive work on RBC and cellular flow in microvessels. We have included a reference of his work in:

Page 14, lines 22 to 25.

Reviewer #1 (Significance (Required)):

The authors discuss the applicability of a detailed numerical model of blood flow in a region of the zebrafish vasculature.

We are not expert in the lattice boltzmann method used here, but the results are what it would be expected from a physical stand point, and together with the information from the method section, we do not have major concerns about the numerics.

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

Summary: The authors report corroborating numerical-experimental studies on the relationship between morphological alterations (e.g. vessel lumen dilation/constriction, network mispatterning) and hemodynamical changes (e.g. variation in flow rate, pressure, wall shear stress) in the vascular network of zebrafish trunk circulation. Various physiological or pathological adaptation scenarios were proposed and tested, with a range of simulation and experiment models. Where I found it a solid piece of work supported by abundant data, certain aspects need to be clarified/enhanced to improve the scientific rigor and potential impact of the manuscript. Below are my detailed comments in the hope of helping the authors improve the manuscript's quality.

Major comments:

1. Cellular blood flow in vascular networks has been extensively studied in recent years by existing computational models (some of which were published open-source) with similar methods and features to the one proposed by the present work. Can the authors be more explicit about the original contributions of the current model, and provide evidence accordingly (e.g. Github repository or code resources)

The RBC model is essentially the model developed by Fedosov and colleagues (Fedosov, et al., 2010). Likewise, the LBM solver for fluid flow calculation is not. Following the reviewer’s advice, we have removed the details of these non-novel aspects of the methodology and placed them in sections E and F of supplementary material instead. The new Models and methods now show condensed descriptions of the three numerical solvers used and the addition of a grid independence matrix discussion section:

Page 17, line 8 to page 20, line 33.

Crucial details for the simulation setup and model configuration are missing. What were the exact boundary conditions (e.g. inlet and outlet pressures) and initial conditions (e.g. feeding hematocrit of RBCs), and how the numerical-experimental validation process of "to match the velocities of various segments of the network by iteratively altering the pressure inputs ..." as stated on page 13 (lines 1-2) was performed for simulations in this work?

We apologize for the vagueness of our description on how numerical to experimental validations were performed. As replied to reviewer 1 for a similar clarification, we have indicated in Table 3 how average systolic peak flow velocities in the entire CA and CV encompassing the 5 ISV segment domain were matched between the simulation and the population-averaged experimental data for the same regions in WT network. Average systolic peak flow velocities for the 10 ISVs in the simulation were matched against WT experiment population-averaged systolic peak flow velocities in arterial and venous ISVs in the same caudal region.

With regards to what iterative alterations of pressure inputs mean, we monitored the average systolic peak velocities and hematocrit levels in CA, CV and ISVs in intervals of 5 cardiac cycle intervals before manually correcting the pressure input levels to better match average systolic peak velocities in these vessels from the experiment averages. Since we are using population averaged flow data, we do not expect their levels to match the levels in a particular fish-specific geometry, the degree of discrepancy between experiment averages and the model predictions of systolic velocities can be large (Table 6). Admittedly, this is one of the weaknesses of our approach and this limitation is stated in the Discussion section:.

Page 15, lines 3 to 9

As RBC flow typically requires roughly 5 cardiac cycles of flow to reach flow development this process of iterative correction typically takes place over 10 to 20 cardiac cycles. We understand that validation may be a subject of keen interest to readers, hence we have now briefly described the solution initialization and flow development protocol in our modeling approach here:

Page 6, lines 5 to 8

What lattice resolution was used for the flow solver and was the RBC membrane mesh chosen accordingly? Were there any sensitivity analysis (regarding pressure input) or grid-independence study (regarding lattice resolution)

We originally decided on the grid (∆X) and time (∆T) discretization resolutions (0.5 µm and 0.5 µs) based on the acceptable computing turnaround time for each model within our scale of resources. We have now included a section on the grid independence matrix in Models and Methods:

Page 19, line 20 to page 20, line 33

Details of the statistical tests (type of tests used, assessment of data normality, sample size etc.) should be given in the figure caption where applicable (e.g. Fig. 3C, Figs. 7-9).

We apologize for the lack of clarity. All statistical tests used have now been mentioned at least once in each section of results and also in Figure captions wherever significance bars are displayed.

The regression models should also be used with caution, e.g. in Fig. 4B, why should data from two different fish types, namely Gata1 MO and WT, be grouped to fit a linear regression model?

We understand the reviewer’s concern that two population data sets should not be carelessly pooled together for regression analysis without adequate justification. In this case we are utilizing gata1 morpholino injection as a means to alter hematocrit level. There is no reported side-effect as to the best of our knowledge, only hematocrit and possibly hemodynamics and morphological response related to hematocrit level should be affected. Moreover, we have mislabelled the companion set to the gata1 morpholino as WT, the data is in fact data from control morphants and not WT. This change has been applied to Fig. 3 graphs and Table 4 and results section:

Page 7, lines 3 to 16

Finally, as we want to generate a continuum range of varying hematocrit for embryos of the same developmental age. In this regard, we think that within the scope of our intentions and well-accepted usage of gata1 morpholino as a hematocrit reduction protocol it is reasonable to pool the two data sets together for regression analysis.

4.I found the data presented in Fig. 7 insufficient to confidently exclude the numerical models 2, 3 but favor model 1 as the adaptation scenario for the Marcksl1KO case. The first question is, how are the threshold RBC perfusion levels determined to categorize the experimented Marcksl1KO fishes into four groups, i.e. "high", "moderate", "low", "zero"? The authors also need to justify why the "high", "moderate", "low" groups can be mapped to the three modelling scenarios (namely models 1, 2, 3) is it just because "a qualitative match with the experimental trend of ascending CA blood velocity" (Fig. 7F)?

We thank the reviewer for his interpretation of our results. Firstly, we apologize for generating the confusion but we are not trying to map simulation models 1, 2 and 3 to high moderate and low groups respectively in Fig. 7. The high, moderate and low categorizations of experimental Marcksl1 KO phenotypes are based on RBC flux levels observed experimentally. We are trying to ascertain which Marcksl KO phenotype the models 1, 2 and 3 fit, if they do fit the experiment trend at all.

Second, in Fig. 7C, it is shown that no significant difference exists between the "high" group and the WT in their average ISV diameter, then what is defining that group as Marcksl1KO type ?

We apologize for the confusion generated. High flow phenotype is similar to WT flow, the diameter is also similar to WT. In Marcksl1 KO mutants we don’t always see clear phenotyping and often a range is presented from mutant to mutant. Hence the high group is essentially morphometrically and hemodynamically similar to WT, the only reason we know it is a mutant because we have genotyped the zebrafish (marcksl1a-/-;marcksl1b1-/-).

Third, a central assumption here is using heart rate as a measure of the pressure drop in different fish individuals (Fig. 7D). Can't two fishes with similar heart rate have distinct pressure drops in the trunk due to difference in network architecture and topology, vice versa?

We agree with the reviewer’s opinion and now feel that our initial proposition was naïve. After addressing the interpretation of heart rate similarity in the gata1 morphants with more convincing CA flow rate estimations, we now believe that heart rates might not be useful indicators of flow or pressure levels in the network. Instead, cardiac output in the form of CA flow rate as reviewer 1 has suggested might be a better indicator. As the reanalysis has dismantled the earlier interpretation, and found that based on the flow rate estimation for the CA, Marcksl1 KO networks have reduced blood flow rates in the CA.

Page 11, lines 9 to 20

This finding has been incorporated into the consideration of flow adaptation scenarios predicted by the simulation models accordingly in the revised manuscript:

Page 12, line 1 to page 13, line 10

Fourth, the authors should explain why a power-law fit (note that it is not "exponential" as stated on page 10, line 3) should be adopted for the regression analysis in Figs. 7E-v,vi (a useful reference may be Joseph et al. eLife 2019: 10.7554/eLife.45077).

We thank the reviewer for the useful reference and the careless mislabeling of regression curve used. This figure has been redone and a linear regression is instead used that does not attempt to imply any physical law for a power or exponential fitting.

Change made: Fig. 7C

Minor comments:

1. The state of art of cell-resolved blood flow models employed to simulate microcirculatory hemodynamics is not accurately described in the introduction (page 4). More recent works should be cited and critically reviewed to present a fair view on the novelty of the computational model developed herein.

We apologize that the models were mentioned in a passing manner. However ,the need for brevity in introduction somewhat limits their expansion. We have instead gave more direct discussion on similar studies and their relevance to our present work in the Discussion section:

Page 14, lines 13 to 31

It is unclear what "realistic representation of local topologies in the network" (page 7, lines 28-31) means as a claim of novelty. If it means vessel "diameter variation", this geometric feature has been modeled by the works the author referenced (namely Roustaei et al. 2022, Zhou et al. 2021). If it means something else, for example, unsmooth or non-circular vessel surface (or "irregularity of the local endothelium surface" as mentioned on page 5, line 2), then strangely the effects of such features are actually not described in the manuscript.

We apologize for not meeting the expectation of novelty as claimed. We see value in the SGM study matrix have now generated data on three SGM scenarios. The SGM1 in the revised manuscript matches WT network impedance in the ISVs by including both axial variation in lumen diameter of the WT network and the elliptical fit representation of cross-sectional skewness seen in WT ISV lumens. SGM 2 has representation of axial variation but not luminal skewness and SGM3 has only geometric average similarity to WT ISVs. Essentially the comparison between SGM1 and SGM2 highlights the role of luminal cross-sectional shape skewness while SGM2 to SGM3 highlights the role of axial variation in luminal diameter. With this new SGM data set, we think we can better qualify the aspiration of demonstrating how vessel shape “irregularities” can alter network hemodynamics. The new findings and discussion can be found in the revised manuscript here:

Page 8, line 19 to page 9 line 36.

Why should Fig. 8 contain data from Marcksl1KO model 2? The scenario underlying model 2 was rejected earlier in the manuscript (see point 6 above), and the Marcksl1KO model 2 data are not mentioned in the text when describing the results of Fig. 8, either.

We have reanalyzed the experiment trend and rewritten the outcome of this results section. In summary, both models 1 and model 2 meet the trend of flow rate reduction (with respect to WT levels) in the CA observed in the experiment. Hence, model 2 inclusion is relevant to the WSS analysis. The changes pertaining to this can be found here:

Page 11, line 9 to page 13 line 10.

It is a dense article with loads of data, which is an advantage but only if appropriately streamlined. More subheadings should be considered, especially for section 2.3 (for which the current subsections appear mistaken, 2.3.1 followed by 2.4.2) The manuscript could also benefit from restructuring through optimal combination of simulation visualizations and quantitative analyses. For example, in Fig. 6, not all simulation snapshots are needed here (it is difficult to visually compare the changes between different cases), whereas some quantification in the form of histograms or boxplots will be handy for the readers to note the variation of WSS magnitudes and ranges.

Thank you for the advice, we removed the unnecessary graphical plots and refer to simulation videos in supplementary data instead for such cases. The bad indexing of results subsections has been fixed, while new subsections have been made for better directional narrative to the paper. These changes are colored in red throughout the revised results section:

Page 4, line 37 to page 13 line 39

Related to point 8, the authors could also consider integrating or synthesizing the analyses for individual aISVs and vISVs presented in various figures. Current descriptions for the ISV data appear scattered with frequent exceptions to the summarized trends or relationships. Some minor formatting issues should also be addressed, e.g. the confusing color codes in Figs. 9D-i, E-i.

Thank you for the advice, we have now pooled aISVs together into one group and vISVs into another, instead of discussing data trends on each of the 10 ISVs.

The mispattening case presented in the end of the results section (section "2.4.2") is interesting but appears loosely connected to the preceding contents. Also, it seems not even mentioned in the discussion section.

We agree that the mispatterning case has been only tangentially relevant to the rest of the manuscript. We have linked the topic thematically by network alterations transforming network flows. It is also now included in the discussion section here:

Page 15, lines 30 to 34

Finally, apart from the effect of topological features on local blood flow, the authors should consider the global flow redistribution arising from the network structure (useful refs. Include Chang et al. PLOS Computational Biology 2017: 10.1371/journal.pcbi.1005892; Meigel et al. Physical Review Letters 2019: 10.1103/PhysRevLett.123.228103; Schmid et al. eLife 2021: 10.7554/eLife.60208).

Thank you for the additional references. These are solid pieces of work that have been added to the discussion here:

Page 16, lines 3 to 10

**Referees cross-commenting**

This review report resonates with mine from an experimental perspective and I agree with all points made regarding issues of the current manuscript that the authors need to address with a revised version.

Reviewer #2 (Significance (Required)):

Significance: The particular merit of the work lies in its comprehensiveness of design and abundance of data, which will be of great interest to both the computational and experimental communities in this research field. However, some crucial details (especially with respect to the modelling aspects) are missing, thus hampering the scientific rigor and potential impact of the work. Furthermore, certain justifying statements appear speculative and inconclusive to explain the obtained data, especially regarding the effect of boundary conditions and systemic parameters. The citation of references (some not cited, some cited already but not properly discussed) also needs to be enhanced with engaging discussions to better bridge the findings of the current work (e.g. RBC partitioning in vascular network, effect of WSS on vasculature morphogenesis) with recent works on this research topic.

References

Fedosov DA, Caswell B, Karniadakis GE. 2010. A Multiscale Red Blood Cell Model with Accurate Mechanics, Rheology, and Dynamics. Biophys J 98:2215–2225. doi:10.1016/j.bpj.2010.02.002

Freund JB, Goetz JG, Hill KL, Vermot J. 2012. Fluid flows and forces in development: functions, features and biophysical principles. Dev Camb Engl 139:1229–45. doi:10.1242/dev.073593

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary: The authors report corroborating numerical-experimental studies on the relationship between morphological alterations (e.g. vessel lumen dilation/constriction, network mispatterning) and hemodynamical changes (e.g. variation in flow rate, pressure, wall shear stress) in the vascular network of zebrafish trunk circulation. Various physiological or pathological adaptation scenarios were proposed and tested, with a range of simulation and experiment models. Where I found it a solid piece of work supported by abundant data, certain aspects need to be clarified/enhanced to improve the scientific rigor and potential impact of the manuscript. Below are my detailed comments in the hope of helping the authors improve the manuscript's quality.

Major comments:

1. Cellular blood flow in vascular networks has been extensively studied in recent years by existing computational models (some of which were published open-source) with similar methods and features to the one proposed by the present work. Can the authors be more explicit about the original contributions of the current model, and provide evidence accordingly (e.g. Github repository or code resources)?
2. Crucial details for the simulation setup and model configuration are missing. What were the exact boundary conditions (e.g. inlet and outlet pressures) and initial conditions (e.g. feeding hematocrit of RBCs), and how the numerical-experimental validation process of "to match the velocities of various segments of the network by iteratively altering the pressure inputs ..." as stated on page 13 (lines 1-2) was performed for simulations in this work? What lattice resolution was used for the flow solver and was the RBC membrane mesh chosen accordingly? Were there any sensitivity analysis (regarding pressure input) or grid-independence study (regarding lattice resolution)?
3. Details of the statistical tests (type of tests used, assessment of data normality, sample size etc.) should be given in the figure caption where applicable (e.g. Fig. 3C, Figs. 7-9). The regression models should also be used with caution, e.g. in Fig. 4B, why should data from two different fish types, namely Gata1 MO and WT, be grouped to fit a linear regression model? 4.I found the data presented in Fig. 7 insufficient to confidently exclude the numerical models 2, 3 but favor model 1 as the adaptation scenario for the Marcksl1KO case. The first question is, how are the threshold RBC perfusion levels determined to categorize the experimented Marcksl1KO fishes into four groups, i.e. "high", "moderate", "low", "zero"? The authors also need to justify why the "high", "moderate", "low" groups can be mapped to the three modelling scenarios (namely models 1, 2, 3); is it just because "a qualitative match with the experimental trend of ascending CA blood velocity" (Fig. 7F)? Second, in Fig. 7C, it is shown that no significant difference exists between the "high" group and the WT in their average ISV diameter, then what is defining that group as Marcksl1KO type? Third, a central assumption here is using heart rate as a measure of the pressure drop in different fish individuals (Fig. 7D). Can't two fishes with similar heart rate have distinct pressure drops in the trunk due to difference in network architecture and topology, vice versa? Fourth, the authors should explain why a power-law fit (note that it is not "exponential" as stated on page 10, line 3) should be adopted for the regression analysis in Figs. 7E-v,vi (a useful reference may be Joseph et al. eLife 2019: 10.7554/eLife.45077).

Minor comments:

1. The state of art of cell-resolved blood flow models employed to simulate microcirculatory hemodynamics is not accurately described in the introduction (page 4). More recent works should be cited and critically reviewed to present a fair view on the novelty of the computational model developed herein.
2. It is unclear what "realistic representation of local topologies in the network" (page 7, lines 28-31) means as a claim of novelty. If it means vessel "diameter variation", this geometric feature has been modeled by the works the author referenced (namely Roustaei et al. 2022, Zhou et al. 2021). If it means something else, for example, unsmooth or non-circular vessel surface (or "irregularity of the local endothelium surface" as mentioned on page 5, line 2), then strangely the effects of such features are actually not described in the manuscript.
3. Why should Fig. 8 contain data from Marcksl1KO model 2? The scenario underlying model 2 was rejected earlier in the manuscript (see point 6 above), and the Marcksl1KO model 2 data are not mentioned in the text when describing the results of Fig. 8, either.
4. It is a dense article with loads of data, which is an advantage but only if appropriately streamlined. More subheadings should be considered, especially for section 2.3 (for which the current subsections appear mistaken, 2.3.1 followed by 2.4.2). The manuscript could also benefit from restructuring through optimal combination of simulation visualizations and quantitative analyses. For example, in Fig. 6, not all simulation snapshots are needed here (it is difficult to visually compare the changes between different cases), whereas some quantification in the form of histograms or boxplots will be handy for the readers to note the variation of WSS magnitudes and ranges.
5. Related to point 8, the authors could also consider integrating or synthesizing the analyses for individual aISVs and vISVs presented in various figures. Current descriptions for the ISV data appear scattered with frequent exceptions to the summarized trends or relationships. Some minor formatting issues should also be addressed, e.g. the confusing color codes in Figs. 9D-i, E-i.
6. The mispattening case presented in the end of the results section (section "2.4.2") is interesting but appears loosely connected to the preceding contents. Also, it seems not even mentioned in the discussion section.
7. Finally, apart from the effect of topological features on local blood flow, the authors should consider the global flow redistribution arising from the network structure (useful refs. Include Chang et al. PLOS Computational Biology 2017: 10.1371/journal.pcbi.1005892; Meigel et al. Physical Review Letters 2019: 10.1103/PhysRevLett.123.228103; Schmid et al. eLife 2021: 10.7554/eLife.60208).

Referees cross-commenting

This review report resonates with mine from an experimental perspective and I agree with all points made regarding issues of the current manuscript that the authors need to address with a revised version.

Significance

The particular merit of the work lies in its comprehensiveness of design and abundance of data, which will be of great interest to both the computational and experimental communities in this research field. However, some crucial details (especially with respect to the modelling aspects) are missing, thus hampering the scientific rigor and potential impact of the work. Furthermore, certain justifying statements appear speculative and inconclusive to explain the obtained data, especially regarding the effect of boundary conditions and systemic parameters. The citation of references (some not cited, some cited already but not properly discussed) also needs to be enhanced with engaging discussions to better bridge the findings of the current work (e.g. RBC partitioning in vascular network, effect of WSS on vasculature morphogenesis) with recent works on this research topic.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

The manuscript presents a detailed numerical model of blood flow in a region of the zebrafish vasculature.

The results section is quite intense and detailed. it is difficult to understand what the authors are after. I think a rewrite would beneficial. The authors present simulations for a wild type and a couple of phenotypes. For each of these they speculate on the possible adaptation mechanism leading to the discussed phenotype, as preservation of constant wall shear stress. However, the comparison between experiments and numerical simulations is really elusive as the conclusions on those mechanisms. Overall we suggest a rewrite with clearer organisation in a way that the reader is not overflown with useless details.

It is not always clear what info of the experiments are used in the simulations on top of the anatomy. Our understanding is that the pressure boundary conditions are set to match the red blood cel velocity observed in experiments. Is this always the case for the three phenotypes and which vessels ? There are about 7 inlets and outlets where to impose pressure boundary conditions. Can the author comment on the uniqueness of this problem? Can different combination of pressure boundary condition leading to the same result ? In how many points/vessels is the measured velocity matched ?

The argument that similar beating frequency in the WT and GATA1 MO suggest pressure does not change is not clear. If the heart was a volumetric pump it would impose the same flow rate, not the same pressure. It would be more useful to measure the cardiac output in terms of flow rate in the Dorsal Aorta. Previous measurements by Vermot suggested the latter would not change much in gata1 MO. It could be that the cardiac output is the same but the vasculature network is different in a way that the shear stress remain the same. It does not look like this was checked by the authors.

Additionaly, it would be useful to provide an effective viscosity for the different vessels, and an effective hydraulic impedance relating DP and Q to interpret the results.

Is the hydraulic impedance of the vessels kept constant in the smooth-geometry model? This needs clarification

As mentioned by the authors they propose a very complex and time expensive simulation. However the results they report are kind of intuitive. Given the availability of the experimental results, would it be useful to use a simpler red blood cell model in the future, to make their simulation more practical? Or clarify when such demanding simulations can add something new?

The authors should check their references as this is not the first time work has been done on the topic. Would be good to have a check in the work of Freund JB and colleagues, as well as Dickinson and colleagues and Franco and colleagues to discuss how the work compares. There may be interesting work in modelling cardiac flow forces in the embryo too.

Significance

The authors discuss the applicability of a detailed numerical model of blood flow in a region of the zebrafish vasculature.

We are not expert in the lattice boltzmann method used here, but the results are what it would be expected from a physical stand point, and together with the information from the method section, we do not have major concerns about the numerics.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

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

The current study uses 3D organotypic rafts to culture primary keratinocytes from Foreskin, Tonsil and Cervix. Further the authors looked at the transcriptomic profiles of each tissue types to study similarities and differences depending on the tissue of origin as well as show the similarity in the tissue specific gene signatures and the ex-vivo samples (data from GTEx). As mentioned by authors Skin and Cervix keratinocytes have been previously cultured on collagen rafts however extending it to Tonsil provides resource and possibility of growing more tissue specific epithelial cells in 3D.

Major comments 1. As the papers focus is to culture epithelial/ epidermal cells on 3D rafts, methods section needs more details about the raft composition, preparation, fibroblast embedding what was the plate size used for raft preparation and culturing of cells on those rafts. What culture media was used for epithelial raft cultures?

We have a detailed published protocol that highlight these details. However, we will expand on some of these details in the manuscript

Results: Figure 1, authors show IF staining's for COL17A1 as marker for basal cells and cornulin for differentiated layers. However, it is important to show how many cells in the basal layer are proliferative? (or how many layers of proliferative cells are present in different epithelia analysed here?) after 14 days majority of cells might already start losing their stemness potential (maybe staining for at least ki67 if staining for basal stem cell marker not possible? Along with loricrin or Involucrin might be good idea).

We will stain for ki67 as suggested. However, based on published data using these raft cultures, we do not expect that many cells will be positive.

This is also important as from supp fig 3 you can see F1 has higher expression of Loricrin, filaggrin etc as compared to all other samples indicating higher diff in this sample. Also, if authors can comment on what was the passage of cells used? And have they observed any difference in the re-epithelization in early passage versus late passage of keratinocytes?

We will expand on this is the updated manuscript. Importantly, we grow these cells in a rho-kinase inhibitor that ‘conditionally’ immortalizes these cells as described (DOI: 10.1172/JCI42297).

It is interesting to see Tonsillar 3D epithelia recapitulate the crypt and surface epithelia and authors also show this with gene expression profile, if possible (Optional), can authors show staining for crypt specific and surface specific markers.

We agree that this is an important control. This will be included.

For all the Supplementary tables where only Ensembl ids are represented, please add gene Id column alongside (it is easier to get biological context from gene id for the reader rather than looking up Ensembl ids). Rename the file names to include the Supplementary file 1, 2, 3?

Since there is 1-to-1 conversion for Ensembl to Gene Id, we elected to not include these. The online app does try tp accommodate this as much as possible. We propose to include two versions of each table. 1 with Ensemble ids only and one with both IDs.

Its excellent to see that in vitro tissue signature matched the in vivo tissue samples (Figure 8) but it will be interesting to show the gene expression differences if found any between the in vitro and in vivo samples that will give insight on the changes as result of in vitro system.

Since the in vivo data will be a mixture of epithelial cells and stroma, these comparisons are not straightforward. However, we are currently examining the use of existing scRNA-Seq data to begin addressing these concerns. This data will be included in the next revision.

Minor comments

1. Abstract: Give sample number (n?) and brief results about the genes that had tissue specific expression pattens.
2. Gene names needs to be in Italics throughout.
3. Introduction: page 5 line 9, authors claim that they based on comparisons they can "identify potential therapeutic targets for various disease" I think this statement either needs experimental evidence or statement / claim needs to be modified.
4. Data submission to GEO???
5. Typo (page 15, line 16 should be "HFK-down", same on page 23 "ectocervix", "endocervix", "uterus", so on, please correct, comma needs to be placed after "
6. Page 24 last line is the heatmap referred here Fig 9B?
7. Fig. 1 legends please indicate what F1, F2, F3, C1--- T1--- represent. Fig 1C Please add axis range/ values for protein atlas data as well.
8. Can authors comment in discussion how was current 3D cervix cells on raft method different from Meyers, C., 1996 3D system?

All these ‘minor’ comments will be addressed.

Reviewer #1 (Significance (Required)):

This article does extend and validate the 3D raft culture method to different epithelial tissues in addition to Skin and cervix. This will be useful for the researchers using co culture systems and interested in understanding epithelial cell and immune cell interactions or host pathogen interacts etc

Describe your expertise: establishing and maintaining primary skin and oral keratinocyte cultures on feeders and 3D cultures on DEDs, Organoid cultures from oral keratinocytes, Oral cancer biology, Histopathology, transcriptomics study, Immuno-oncology.

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

Summary Jackson et al.'s manuscript describes an experiment that directly compares 3D organotypic assays created with primary human epithelial cells from foreskin, cervix and tonsil using histological and bulk RNA sequencing approaches. The authors convincingly show the retention of site-specific histological and transcriptomic differences between the stratified epithelial tissues in culture. Differentially expressed genes are identified and pathway analyses suggest genes that might be involved in the different differentiation processes between these tissue sites and differential regulation of ECM and immune pathways. Differentially expressed genes are used to develop a classifier for tissue identification, which is tested using GTEx data.

Major Comments • The interferon stimulated genes of B cells and macrophages (from Mostafavi et al., 2016) are likely to be very different from those in epithelial cells, so the analysis presented in Figure 9 seems like a stretch to me.

We will include caveats to this interpretation. We are planning stimulation experiments of each tissue to compare IFN responses. However, depending on the outcomes, these may end up being outside of the scope of the current manuscript.

• OPTIONAL: Further data comparing the nature and magnitude of the interferon responses of the three epithelia would improve interest in the manuscript but are not necessary for publication of the current dataset.

See above

Minor Comments • Details of n numbers and what each point represents should be added to Figure 1C. Are these points measurements from 25 um intervals of just one raft per donor? What are 'fields of view' here? Are measurements from one section or from multiple sections per raft? • Page 12 - provide a figure/panel citation for the "micrograph derived from a tonsillectomy" that is suggested for comparison. • In Figure 1 - Figure Supplement 1, how representative of the whole raft are these images? Does the extent of stratification change near to the edge of the collagen gel, for example? How well matched for location within a raft are the images shown? • Page 24 - clarify uses of the phrase "down-regulated in tonsils". Presumably this section refers to tonsil epithelium in 3D organotypic rafts.

Typos • Page 3 - "the cervix is lined with stratified squamous epithelia", should be epithelium. • "J.G. Rheinwald" in in text references. • Page 6 - 'or' not 'and' in first sentence of primary cell culture section.

All these ‘minor’ comments and typos will be addressed.

Reviewer #2 (Significance (Required)):

This highly descriptive study provides a detailed analysis of a bulk RNA sequencing experiment comparing foreskin, cervix and tonsil 3D organotypic rafts. Retained histological and transcriptional differences between epithelial tissues of different origins in organotypic assays are well documented in the literature (e.g., parmoplantar vs non-parmoplantar skin, PMID: 36732947; airway tract, PMID: 32526206) so the observed differences between these three very distinct anatomical tissues are unsurprising overall. The data have been made available via SRA and a shiny web app and are likely to be of interest and use to other researchers working on these tissues in culture. The experiment was performed in matched cell culture conditions so replicates are well-controlled, if limited in number (n=3).

We appreciate this feedback. We agree this is a descriptive study. Nonetheless, we believe there is value in formally demonstrating differences and similarities between these tissues. The provided references will be included to expand our discussion.

I am an epithelial cell biologist specializing in human cell culture models. I do not have sufficient computational background to comment in detail on the RNA sequencing methods or analysis within the manuscript.

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

This is very carefully analysed and written study describing the transcriptional differences between in vitro models of epithelia derived from cervix, foreskin and tonsil tissues. Importantly, they compare the findings to in vivo samples using publicly available data. The findings are significant and will be of interest to the scientific community. I cannot fault the analysis pathways or the conclusions, and the manuscript is a pleasure to read. I recommend it is accepted for publication as is.

Reviewer #3 (Significance (Required)):

This is an important study that is highly significant for researchers interested in epithelia tissue and infection. The data are clearly presented and the analysis is thorough. The authors state that they will make the data publicly available. This will be an important resource for the community.

We appreciate the kind words

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

This is very carefully analysed and written study describing the transcriptional differences between in vitro models of epithelia derived from cervix, foreskin and tonsil tissues. Importantly, they compare the findings to in vivo samples using publicly available data. The findings are significant and will be of interest to the scientific community. I cannot fault the analysis pathways or the conclusions, and the manuscript is a pleasure to read. I recommend it is accepted for publication as is.

Significance

This is an important study that is highly significant for researchers interested in epithelia tissue and infection. The data are clearly presented and the analysis is thorough. The authors state that they will make the data publicly available. This will be an important resource for the community.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary

Jackson et al.'s manuscript describes an experiment that directly compares 3D organotypic assays created with primary human epithelial cells from foreskin, cervix and tonsil using histological and bulk RNA sequencing approaches. The authors convincingly show the retention of site-specific histological and transcriptomic differences between the stratified epithelial tissues in culture. Differentially expressed genes are identified and pathway analyses suggest genes that might be involved in the different differentiation processes between these tissue sites and differential regulation of ECM and immune pathways. Differentially expressed genes are used to develop a classifier for tissue identification, which is tested using GTEx data.

Major Comments

• The interferon stimulated genes of B cells and macrophages (from Mostafavi et al., 2016) are likely to be very different from those in epithelial cells, so the analysis presented in Figure 9 seems like a stretch to me.
• OPTIONAL: Further data comparing the nature and magnitude of the interferon responses of the three epithelia would improve interest in the manuscript but are not necessary for publication of the current dataset.

Minor Comments

• Details of n numbers and what each point represents should be added to Figure 1C. Are these points measurements from 25 um intervals of just one raft per donor? What are 'fields of view' here? Are measurements from one section or from multiple sections per raft?
• Page 12 - provide a figure/panel citation for the "micrograph derived from a tonsillectomy" that is suggested for comparison.
• In Figure 1 - Figure Supplement 1, how representative of the whole raft are these images? Does the extent of stratification change near to the edge of the collagen gel, for example? How well matched for location within a raft are the images shown?
• Page 24 - clarify uses of the phrase "down-regulated in tonsils". Presumably this section refers to tonsil epithelium in 3D organotypic rafts.

Typos

• Page 3 - "the cervix is lined with stratified squamous epithelia", should be epithelium.
• "J.G. Rheinwald" in in text references.
• Page 6 - 'or' not 'and' in first sentence of primary cell culture section.

Significance

This highly descriptive study provides a detailed analysis of a bulk RNA sequencing experiment comparing foreskin, cervix and tonsil 3D organotypic rafts. Retained histological and transcriptional differences between epithelial tissues of different origins in organotypic assays are well documented in the literature (e.g., parmoplantar vs non-parmoplantar skin, PMID: 36732947; airway tract, PMID: 32526206) so the observed differences between these three very distinct anatomical tissues are unsurprising overall. The data have been made available via SRA and a shiny web app and are likely to be of interest and use to other researchers working on these tissues in culture. The experiment was performed in matched cell culture conditions so replicates are well-controlled, if limited in number (n=3).

I am an epithelial cell biologist specializing in human cell culture models. I do not have sufficient computational background to comment in detail on the RNA sequencing methods or analysis within the manuscript.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

The current study uses 3D organotypic rafts to culture primary keratinocytes from Foreskin, Tonsil and Cervix. Further the authors looked at the transcriptomic profiles of each tissue types to study similarities and differences depending on the tissue of origin as well as show the similarity in the tissue specific gene signatures and the ex-vivo samples (data from GTEx). As mentioned by authors Skin and Cervix keratinocytes have been previously cultured on collagen rafts however extending it to Tonsil provides resource and possibility of growing more tissue specific epithelial cells in 3D.

Major comments

1. As the papers focus is to culture epithelial/ epidermal cells on 3D rafts, methods section needs more details about the raft composition, preparation, fibroblast embedding what was the plate size used for raft preparation and culturing of cells on those rafts. What culture media was used for epithelial raft cultures?
2. Results: Figure 1, authors show IF staining's for COL17A1 as marker for basal cells and cornulin for differentiated layers. However, it is important to show how many cells in the basal layer are proliferative? (or how many layers of proliferative cells are present in different epithelia analysed here?) after 14 days majority of cells might already start losing their stemness potential (maybe staining for at least ki67 if staining for basal stem cell marker not possible? Along with loricrin or Involucrin might be good idea). This is also important as from supp fig 3 you can see F1 has higher expression of Loricrin, filaggrin etc as compared to all other samples indicating higher diff in this sample. Also, if authors can comment on what was the passage of cells used? And have they observed any difference in the re-epithelization in early passage versus late passage of keratinocytes?
3. It is interesting to see Tonsillar 3D epithelia recapitulate the crypt and surface epithelia and authors also show this with gene expression profile, if possible (Optional), can authors show staining for crypt specific and surface specific markers.
4. For all the Supplementary tables where only Ensembl ids are represented, please add gene Id column alongside (it is easier to get biological context from gene id for the reader rather than looking up Ensembl ids). Rename the file names to include the Supplementary file 1, 2, 3?
5. Its excellent to see that in vitro tissue signature matched the in vivo tissue samples (Figure 8) but it will be interesting to show the gene expression differences if found any between the in vitro and in vivo samples that will give insight on the changes as result of in vitro system.

Minor comments

1. Abstract: Give sample number (n?) and brief results about the genes that had tissue specific expression pattens.
2. Gene names needs to be in Italics throughout.
3. Introduction: page 5 line 9, authors claim that they based on comparisons they can "identify potential therapeutic targets for various disease" I think this statement either needs experimental evidence or statement / claim needs to be modified.
4. Data submission to GEO???
5. Typo (page 15, line 16 should be "HFK-down", same on page 23 "ectocervix", "endocervix", "uterus", so on, please correct, comma needs to be placed after "
6. Page 24 last line is the heatmap referred here Fig 9B?
7. Fig. 1 legends please indicate what F1, F2, F3, C1--- T1--- represent. Fig 1C Please add axis range/ values for protein atlas data as well.
8. Can authors comment in discussion how was current 3D cervix cells on raft method different from Meyers, C., 1996 3D system?

Significance

This article does extend and validate the 3D raft culture method to different epithelial tissues in addition to Skin and cervix. This will be useful for the researchers using co culture systems and interested in understanding epithelial cell and immune cell interactions or host pathogen interacts etc

Describe your expertise: establishing and maintaining primary skin and oral keratinocyte cultures on feeders and 3D cultures on DEDs, Organoid cultures from oral keratinocytes, Oral cancer biology, Histopathology, transcriptomics study, Immuno-oncology.

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

Learn more at Review Commons

---

Reply to the reviewers

We thank the three reviewers for their thorough evaluation of our work and for their positive and constructive feedback. Below we provide an initial response to these comments, including descriptions of several changes incorporated in our preliminary revision. Please note that introductory and significance comments are not repeated here:

Reviewer 1

In my opinion, the main strength of this work is in the development and use of the original assay for adapter identification. As I already indicated, this is a biologically very important problem for cytoplasmic dynein. Another important strength of the paper is the extension of the work to Drosophila. Demonstration of the fact that Heatr5B is an essential gene, and that the product of this gene is involved in dynein-dependent trafficking in fly embryos makes the results significantly more important.I do not think there are many problems with the results in this manuscript. Generally speaking, the data on biochemical interactions are not as strong as I would like them to be. This is explained mainly by the fact that the authors do not have an expressed recombinant Heatr5B that they can use in biochemical experiments, and they limit their biochemistry by pulling down the protein from cell extracts.

Whilst we are very grateful to the reviewer for their thorough evaluation of our work, we do not understand this particular comment. We did include data with recombinant HEATR5B showing binding in vitro to purified dynein and dynactin complexes. The results are shown in Fig. 2B. We have now made it clearer (from line 153 of the preliminarily revised manuscript) that these experiments used recombinant HEATR5B. We hope in the future to determine the biochemical and structural basis of HEATR5B’s interaction with dynein and dynactin but feel that this goes well beyond the scope of this initial study (which already covers a lot of ground), especially as we have not yet found a way to express HEATR5B fragments (line 151).

This creates one of the few experimental problems with the paper. The authors claim that dynein and dynactin do not compete for Heatr5B binding, and therefore they can bind to both components of the complex at the same time. Unfortunately, I do not think that this claim is justified because concentrations of dynein and dynactin in their pull-down assay are much higher than the concentration of GFP-HEATR5B, and likely that HEATR5B does not saturate the binding sites on the motor complex. Therefore, it is unclear whether dynein and dynactin compete for Heatr5B binding. In any case, the conclusion about the competition cannot be seriously made without analysis of saturation curves.

The purified dynein and dynactin were not in excess to recombinant HEATR5B in this assay (80 pmol HEATR5B, 20 pmol dynein tail and 10 pmol dynactin). Nonetheless, the reviewer makes a very good point that we cannot draw strong conclusions about competition unless we generate saturation curves. We have therefore toned down the interpretation of this experiment and included the caveat raised by the reviewer (from line 157):

‘Compatible with this notion, we did not observe competition between the purified dynein tail and dynactin for HEATR5B binding in our in vitro binding assay when both complexes were added simultaneously to the beads (Figure 2B). However, we cannot rule out the possibility that a competitive interaction was masked by binding sites on one of the components not being saturated. Nonetheless, we can conclude from this set of experiments that HEATR5B complexes with endogenous dynactin and dynein in cell extracts and can interact with both complexes directly.’

Please note that this was only a minor point in our manuscript.

My second concern with this paper is the quality of imaging in mammalian cells. Unfortunately, not much can be done with live cell imaging because GFP-HEATR5B is expressed in cells at a low level (see, for example, Fig. 3A). However, in fixed cells GFP-HEATR5B signal could be easily amplified using anti-GFP antibodies.

The fixed cell images of GFP-HEATR5B cells are stained with anti-GFP antibodies and are the result of extensive optimisation of staining and imaging conditions. Due to its low expression and presence in both cytoplasm and membrane-bound pools, the signal for GFP-HEATR5B is not as striking as, for example, RAB11A and AP1γ. Nonetheless, the punctate signals are sufficiently strong to confidently evaluate co-localisation with membrane markers. We have now added to the relevant legends that the GFP signal is obtained via GFP antibody staining. Please note that the association of GFP-HEATR5B with AP1γ (Fig. 3A, B) was also confirmed by immunoprecipitation (Fig. 2A).

A minor problem with movie presentations is that the authors should include both a timer and a scale bar for all their live cell sequences, especially because the movies are looped. The authors did it for Movie 5, and they should do it for the rest of their live cell sequences.

Although information on the duration of movies (including loops) was included in the legends, we agree that it would be helpful to incorporate timers and scale bars in the movies. We have not been able to include this change in the preliminary revision as we have to co-ordinate with our visual aids team to reapply labels and arrows to the edited movies. This change will be made to the full revision.

In my opinion, the main novelty of this paper is in its pull-down assay, I would like to have it discussed more extensively. The authors state that they "were particularly drawn to Heatr5B". Is there an objective reason for this choice? If so, it should be specified.

We included our two reasons for focusing on HEATR5B in the previous submission, namely that it was the only protein to be enriched on the tail by dynactin in both the N- and C-terminal tethering configuration and that a previous study found it was one of a number of proteins present on dynactin-associated vesicles. We have modified the language in this section (which starts on line 120 of the preliminary revision) by using the connective ‘because’. This change makes it clearer that there were objective reasons for focusing on HEATR5B in the first instance.

Furthermore, I would like to see the authors discuss the other hits. Their list of hits includes a large number of ribosomal proteins. Do ribosomes really interact with dynein? Can the authors speculate on the number of true hits? Finally, it is likely that dynein interacts with some of the cargoes only transiently. How can the assay be modified to capture these transient interactions?

This is another very good suggestion. As requested, we have added a comment to the Discussion (from line 441) about how transient interactions might be captured using a variation of our strategy. We have now added a comment in the Discussion (from line 435) about the capture of ribosomes and other RNA-associated proteins in our screen, as well as the potential significance of this observation. We have also highlighted in this section another dynactin-stimulated hit, Wdr91, which we are following up. We also discussed the STRIPAK complex, which warrants further study, in the Results (line 106). We do not have space to discuss other hits but their identities are listed in Tables S1 and S2 together with a summary of their known functions for easy reference.

Reviewer 2:

Major points:

As a view of non-expert of light microscopy cellular imaging, some confocal images are difficult to accept as proofs of their conclusion that mutation to decrease HEAT5B/AP1 interaction results in diffusion from perinucleolar surface. For example, fluorescent signals in Control of Fig.4A seem more diffused than HR5B KO, which have fluorescence clearly localized on the surfaces of nuclei. Can they have explanation how it ends up with their statistical analysis in Fig.4E?

Fig. 4A is representative of what we typically see in mutant cells, with dispersion of the dimmer AP1γ signal in the cytoplasm and less disturbed localisation of the brighter AP1γ signal at the TGN (see Fig. S6B for quantification of AP1γ signal at the TGN in control and mutant cells). We should have made it clear in the Results that the unbiased image analysis pipeline used to produce Fig. 4E detects the total AP1γ signal not just bright signals (this feature of the pipeline is important given the differences in fluorescent intensity of puncta in the two genotypes). We have now clarified this issue in the Results (line 257) and the Fig. 4E legend. We have also added arrowheads to Fig. 4A to highlight the dispersed dim signal in the mutant cells. We thank the reviewer for leading us to improve the description of this experiment__.__

When they mention statistically more distance between target molecules and the perinuleolar surface, are dynein/dynactin connected to AP1 via HEAT5B stalled on the microtubule before reaching the minus end, or dissociate from the microtubule? Clarifying this will improve impact of this work. If the current data is not enough to answer, this reviewer will propose another confocal microscopy with also tubulin labeled. With this, the location of HEAT5B, AP1 etc. with respect to both nuclei and microtubule cytoskeleton will be clarified.

We would love to know the answer to the question of whether HEATR5B disruption reduces the association of AP1γ with microtubules. We have looked into co-localisation of microtubules and dynein’s cargoes previously using advanced light microscopy and found that it is not possible to draw conclusions about meaningful versus coincidental associations because of the density of the microtubule network. In the case of our current study, this approach would be further confounded by the difference in size in fluorescent AP1 puncta in control and HEATR5B mutant cells. We have also in the past attempted to purify recycling endosomal membranes from cells to determine how loss of HEATR5B influences dynein-dynactin association. However, even after extensive efforts we could not reproduce selective purification of recycling endosomes using the published protocol, or indeed variants of it. What is more, we find in general that there is rapid dissociation of motors during purification of membranes from cells, which would confound our results even if we could purify the recycling compartment. We therefore feel that the only way to address the question of how HEATR5B modulates dynein function at the molecular level is to reconstitute the transport machinery with pure proteins (including the as-of-yet unidentified activating adaptor) and microtubules in vitro, which is beyond the scope of this study. We have discussed the future aim of in vitro reconstitution to dissect mechanism in the Discussion (from line 494).

In Line168-169, they concluded AP1gamma associated with TGN rarely overlapped with HEATR5B, based on Fig.3A (where HR5B and AP1 seem overlapped in HeLa cells), Fig.S2A (where AP1gamma and TGN46 seem overlapped in U2OS cells) and Fig.S2B (where HR5B and TGN46 are not overlapped in HeLa cells). Is Fig.3A not contradictory to their conclusion (AP1gamma and HEATR5B not overlapped)? Why did they not directly check the overlap between AP1gamma and HR5B in the same cell in U2OS cells?

We don’t understand why our co-localisation data might be contradictory to our conclusions. Fig. 3A, together with the associated insets and quantification in Fig. 3B, show overlap of HEATR5B with AP1γ puncta in the cytoplasm of HeLa cells but not the AP1γ that is strong enriched in the perinuclear region, as we stated in the results. Absence of enrichment of HEATR5B with the TGN is additionally shown in Fig. S2B. These observations are commented on further in the Discussion (from line 517). We do, however, agree with the reviewer that is was not ideal that we did not show AP1 and TGN association in a HeLa cell (even though it has been documented in the literature). We have now corrected this oversight by showing HeLa cell data in Fig. S2A. We could not check the overlap of AP1 and HEATR5B in U2OS cells as we do not have a GFP-HEATR5B stable U2OS cell line.

Minor points:

Line100-105 and Fig.1EF are not clear. Is it correct that proteins in red bold letters and in blue letters in Fig.1EF are 28 proteins enriched on the dynactin tail?

We should have been clearer here and thank the reviewer for spotting this. We have modified the figure call outs in the text to include the labelling scheme, which we think helps significantly. We have also clarified the labelling system in the legend. To summarise, bold labelling indicates interactors of the dynein tail that are not core components of the dynein-dynactin machinery (such proteins are labelled in non-bold and italics); the blue bold text shows those ‘none core’ interactors that were only enriched on the dynein tail when exogenous dynactin was spiked into the lysates.

Do authors have any idea why the "dynactin-stimulated" ones (in blue) are localized at left end of this group (relatively less significance of dynactin tail binding, if this reviewer understands correctly)?

This question appears to indicate some confusion about whether we are capturing the dynein tail or dynactin. We believe the changes made in response to the previous comment about the labelling scheme should help clear this up. Being positioned to the left of this grouping shows a lower degree on enrichment vs the control (although still greater than 10 fold), rather than a difference in statistical significance. The observation that core dynein-dynactin subunits are more enriched on the dynein tail indicates that these interactions are the most stable or the most frequent.

Fig.S7: More explanation how to conclude that HR5B KO is dimmer than Ctrl based on this plot would be helpful.

We have added a line to the legend to Figure S7C to clarify this matter.

Reviewer 3:

Minor comments

HEATR5B overexpression in U2OS cells increased perinuclear clustering of Rab11A/AP1/dynactin-associated membrane. To which compartment are these vesicles directed and associated, the Golgi apparatus? Could the authors show which compartment it is?

We plan to perform new experiments to address this minor comment in the full revision. However, given their location near the microtubule organising centre, it is likely that the relocalised membranes will be in the vicinity of the Golgi apparatus.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary:

The goal of the authors is to identify dynein regulators which control how dynein and dynactin complexes orchestrate trafficking of diverse cargoes. To do so, the authors have performed a well thought proteomic screen for novel interacting proteins of the dynein tail potentially enhanced by dynactin. These pull-down experiments identified about 50 new dynein tail-interacting proteins, many of which were enhanced by dynactin.

The authors focused on one candidate, HEATR5B, because it was robustly isolated from the screens and its association with the dynein tail was stimulated by exogenous dynactin. HEATR5B is known to interact with AP1 complex, as adaptors that orchestrate cargo loading of clathrin-coated vesicles from intracellular membranes.

The authors further show that HEATR5B complexes with endogenous dynactin and dynein as reveal by immuno precipitation from human cells extracts and can interact with both complexes directly. Then by using Hela cell line stably expressing GFP-HEATR5B, they show that HEATR5B is selectively enriched on the AP1 structure, some of which can be subjected to long-distance transport. They provide evidences that a large proportion of the HEATR5B-positive structures are associated with endosomal recycling membranes, as revealed by colocalization with RAB11A. They further show that the HEATR5B/ AP1 and HEATR5B/ RAB11 membrane structures show similar dynamics, indicating that HEATR5B associate with endosomal membranes that are capable of directed movement. SiRNA depletion of DYNC1H1 reveals that dynein promotes retrograde trafficking of AP-1 associated endosomal membranes.

The authors then investigate the contribution of HEATR5B to AP1-associated membrane trafficking by CRIPR/cas9-mediated mutagenesis in human U2OS cells that disrupt HEATR5B protein expression. They provide evidences that in HEATR5B mutant cells, there is a reduction in the amount of AP1 signal associated with RAB11A-positive structures indicating that disrupting HEATR5B reduces the association of AP1with endosomal membranes. This indicates that HEATR5B promotes AP1 recruitment to endosomal membranes. HEATR5B overexpression in U2OS cells increased perinuclear clustering of Rab11A/AP1/dynactin-associated membrane, suggesting that HEATR5B can stimulate retrograde trafficking of AP1-associated endosomal membranes by dynein- dynactin.

To assess the importance of HEATR5B function at the organismal level, as well as in polarized cell type the authors investigate its function in Drosophila in which there is a single HEATR5B homologue (Heatr5). They generated via crisper an Heatr5 mutant strain. Heatr5 homozygous mutants are zygotic lethal that died in second larval instar stage. They further provide evidence by investigating nos-cas9 gRNA-Hr51+2 mothers, that Heatr5 plays maternal function essential for embryogenesis. They further show that in early embryos from nos-cas9 gRNA-Hr5 females AP1 puncta are strongly reduced and dimmer.

Next, to understand the effect of Heatr5 disruption on AP1-based trafficking in Drosophila, they used the syncytial blastoderm embryo in which the microtubule cytoskeleton is highly polarized with less apically nucleated ends above the nuclei and more basally extended ends. In this system, the activity of minus end-directed motor, such as dynein, and minus end-directed motor, such as kinesin, can be distinguished by the direction of cargo movement.

By injecting AP1 antibodies into wild-type and Heatr5 mutant embryos, they provide evidence that AP1 undergoes net apical transport in the Drosophila embryo and that this process is strongly promoted by Heatr5. They further show that this process is microtubule and dynein dependent and that Heatr5 selectively promotes dynein-mediated transport of AP1 structures in the embryo. They then show that Heatr5-dependent AP1 trafficking pathways in the embryo involves the endosomal and Golgi membranes and that Heatr5 is also required for Golgi organization.

Major Comments

This study is very comprehensive and multi-scale. It ranges from the identification of a dynein motor adaptor for membrane trafficking by a proteomic screen, to its functional characterization in human cells and then during development with Drosophila embryo as model organism. The data are of high quality and are supported by very convincing quantitative analyses. The results are conclusive and the experiments have been carried out and presented in a very constructive way. This combination makes the manuscript very interesting.

Minor comments

HEATR5B overexpression in U2OS cells increased perinuclear clustering of Rab11A/AP1/dynactin-associated membrane. To which compartment are these vesicles directed and associated, the Golgi apparatus? Could the authors show which compartment it is?

Significance

This study is important in two aspects. Firstly, it has identified HEATR5B as a new adaptor of the dynein motor for intracellular membrane trafficking. It is important to mention that this motor is involved in many transport processes and it is still unclear how a single motor orchestrates the traffic of so many cargoes. Second, this work shed new light on the retrograde trafficking from endosomal material to the Golgi apparatus, in particular with HEATR5B, a known interactor of the AP1 clathrin adapter complex. This study highlights a role of HEATR5B in a novel dynein-based process for retrograde trafficking of AP1-associated endosomal vesicle to the Golgi apparatus. It also indicates that HEATR5B promotes association of AP1 with endosomal membrane in a dynein-independent manner.

This work is particularly important for the cell biology field.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Major points:

As a view of non-expert of light microscopy cellular imaging, some confocal images are difficult to accept as proofs of their conclusion that mutation to decrease HEAT5B/AP1 interaction results in diffusion from perinucleolar surface. For example, fluorescent signals in Control of Fig.4A seem more diffused than HR5B KO, which have fluorescence clearly localized on the surfaces of nuclei. Can they have explanation how it ends up with their statistical analysis in Fig.4E?

When they mention statistically more distance between target molecules and the perinuleolar surface, are dynein/dynactin connected to AP1 via HEAT5B stalled on the microtubule before reaching the minus end, or dissociate from the microtubule? Clarifying this will improve impact of this work. If the current data is not enough to answer, this reviewer will propose another confocal microscopy with also tubulin labeled. With this, the location of HEAT5B, AP1 etc. with respect to both nuclei and microtubule cytoskeleton will be clarified.<br /> In Line168-169, they concluded AP1gamma associated with TGN rarely overlapped with HEATR5B, based on Fig.3A (where HR5B and AP1 seem overlapped in HeLa cells), Fig.S2A (where AP1gamma and TGN46 seem overlapped in U2OS cells) and Fig.S2B (where HR5B and TGN46 are not overlapped in HeLa cells). Is Fig.3A not contradictory to their conclusion (AP1gamma and HEATR5B not overlapped)? Why did they not directly check the overlap between AP1gamma and HR5B in the same cell in U2OS cells?

Minor points:

Line100-105 and Fig.1EF are not clear. Is it correct that proteins in red bold letters and in blue letters in Fig.1EF are 28 proteins enriched on the dynactin tail? Do authors have any idea why the "dynactin-stimulated" ones (in blue) are localized at left end of this group (relatively less significance of dynactin tail binding, if this reviewer understands correctly)?

Fig.S7: More explanation how to conclude that HR5B KO is dimmer than Ctrl based on this plot would be helpful.

Significance

In this work, Madan and colleagues studied dynein adaptor proteins, which are stimulated by dynactin, using proteomics, fluorescent microscopy, live cell imaging techniques for U2OS and fly embryo cells. They especially focused on HEATR5B and proved its role to bind AP1 membrane associate protein for intracellular transport. They first conducted proteomic studies and presented novel lists of dynein-associated proteins and proteins stimulated by dynactin. Among them they decided to prioritize HEATR5B protein (it would be interesting to know their motivation to choose this protein) and carried on fluorescent microscopy studies to characterize roles of HEATR5B in microtubule-based motility. Their approach using U2OS cells is to correlate distribution of HEATR5B and such proteins as AP1gamma, TGN46, RAB11A, which they expect interaction with HEATR5B, between WT and mutants. They remarkably demonstrated distance from perinucleolar membrane is heavily influenced by defect of adaptor function of HEATR5B, by fluorescent microscopy and statistical analysis. Next they made HEATR5B depleted Drosophila embryo by CRSPR-CAS9. They proved its influence on AP1 trafficking to Golgi, which is another novel finding of this study, consistent with the case of U2OS cells.

In general the whole study proved importance of HEATR5 proteins on AP1 trafficking. Many data are presented in convincing way and carefully statistically analyzed. This work will attract attention of wide audience from the field of cytoskeleton, motor proteins and membrane trafficking. After addressing a few points, the manuscript will be ready for publication.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

This work is the first systematic attempt to identify and characterize a diverse set of adapters that attach cytoplasmic dynein to its different cargoes and thus activate the motor. It is an important work because in animal cells dynein is the only efficient motor that can perform processive transport toward the minus ends of microtubules, and therefore the specificity of transport for multiple cargoes along microtubules is determined by these adapters.

The authors use the recombinant tail of dynein for pulling down interacting proteins from the cell extract. This is a straightforward approach, but its main problem is the large number of non-specific proteins that bind to the column. To solve the problem, the authors use a very smart approach. It is based on the fact that in all known cases so far dynein does not transport cargoes without dynactin, and, therefore, potential adaptors are unlikely to bind to the affinity column very efficiently. They compare pull-downs in the presence and absence of dynactin paying specific attention to proteins that bind in the presence of both dynein and dynactin but not dynein alone.

Among the proteins that have been identified by this assay is Heatr5B, the protein known to associate with AP1 clathrin adaptor. Functional characterization of the protein can be divided into two parts, work with mammalian Heatr5B in tissue culture cells and analysis of its function in Drosophila.

In my opinion, the main strength of this work is in the development and use of the original assay for adapter identification. As I already indicated, this is a biologically very important problem for cytoplasmic dynein. Another important strength of the paper is the extension of the work to Drosophila. Demonstration of the fact that Heatr5B is an essential gene, and that the product of this gene is involved in dynein-dependent trafficking in fly embryos makes the results significantly more important.

I do not think there are many problems with the results in this manuscript. Generally speaking, the data on biochemical interactions are not as strong as I would like them to be. This is explained mainly by the fact that the authors do not have an expressed recombinant Heatr5B that they can use in biochemical experiments, and they limit their biochemistry by pulling down the protein from cell extracts. This creates one of the few experimental problems with the paper. The authors claim that dynein and dynactin do not compete for Heatr5B binding, and therefore they can bind to both components of the complex at the same time. Unfortunately, I do not think that this claim is justified because concentrations of dynein and dynactin in their pull-down assay are much higher than the concentration of GFP-HEATR5B, and likely that HEATR5B does not saturate the binding sites on the motor complex. Therefore, it is unclear whether dynein and dynactin compete for Heatr5B binding. In any case, the conclusion about the competition cannot be seriously made without analysis of saturation curves.

My second concern with this paper is the quality of imaging in mammalian cells. Unfortunately, not much can be done with live cell imaging because GFP-HEATR5B is expressed in cells at a low level (see, for example, Fig. 3A). However, in fixed cells GFP-HEATR5B signal could be easily amplified using anti-GFP antibodies.

A minor problem with movie presentations is that the authors should include both a timer and a scale bar for all their live cell sequences, especially because the movies are looped. The authors did it for Movie 5, and they should do it for the rest of their live cell sequences.

In my opinion, the main novelty of this paper is in its pull-down assay, I would like to have it discussed more extensively. The authors state that they "were particularly drawn to Heatr5B". Is there an objective reason for this choice? If so, it should be specified. Furthermore, I would like to see the authors discuss the other hits. Their list of hits includes a large number of ribosomal proteins. Do ribosomes really interact with dynein? Can the authors speculate on the number of true hits? Finally, it is likely that dynein interacts with some of the cargoes only transiently. How can the assay be modified to capture these transient interactions?

Significance

The bottom line is very clear. For me, it is an excellent technical paper with biological results that clearly demonstrate the validity of the technique. As such, it can and should be published.

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

Learn more at Review Commons

---

Reply to the reviewers

Manuscript number: RC-2023-01861

Corresponding author(s): Manuela, Baccarini

[The “revision plan” should delineate the revisions that authors intend to carry out in response to the points raised by the referees. It also provides the authors with the opportunity to explain their view of the paper and of the referee reports.

The document is important for the editors of affiliate journals when they make a first decision on the transferred manuscript. It will also be useful to readers of the reprint and help them to obtain a balanced view of the paper.

If you wish to submit a full revision, please use our "Full Revision" template. It is important to use the appropriate template to clearly inform the editors of your intentions.]

1. General Statements [optional]

We were delighted to learn that all three reviewers found the paper novel and of interest for a cell biology audience. They especially highlighted the carefully conducted screen, whose results will be integrally published with this paper and will be of use for scientists interested in lysosome biology.

2. Description of the planned revisions

Insert here a point-by-point reply that explains what revisions, additional experimentations and analyses are planned to address the points raised by the referees.

Reviewer 1

• The claim that "peripheral accumulation of lysosomes inhibits protrusion formation and limits cell motility" should be tested more rigorously using the RAMP method, preferably in living cells. Other approaches, such as overexpression/siRNA of Arl8b and other motor adaptors, such as SKIP/PLEKHM2, can be used to alter lysosome positioning and confirm this central finding of the manuscript. The authors could also consider including additional mechanistic data in order to comprehend how lysosome positioning controls cell motility. For instance, the RAMP approach could be employed to investigate cortical actin dynamics upon repositioning of lysosomes to the peripheral/perinuclear region.

Answer: We have purchased the RAMP system from Addgene and are adapting it to our color setup to use it in HeLa cells expressing GFP-PLEKHG3 and (hopefully) in PLEKHG3 KO cells, adding LiveAct to investigate cortical actin dynamics upon lysosomal repositioning as well as PLEKHG3 KO.

Reviewer 2

5 - It is not clear if in cells KO for PLEKHG3, the overexpression of KIF1A leads to more lysosomes localizing close to the PM, as well as more protrusions and more cell motility, as the authors only compare cell overexpressing GFP or GFP-PLEKHGL3.

Answer: Currently, we do not have a PLEKHG3 KO. We have, however, redoubled our efforts, so far unsuccessful, to generate a PLEKHG3 CRISPR-Cas KO in HeLa, going up to 10 sg-guides, and hope that we will be successful in the next future. In this case, we will be able to easily address this interesting question.

Reviewer 3

• Data presented in Figure 6 showing cell motility analysis is interesting and has potential to make the manuscript impactful. Similarly, data in Figure 4F (live cell imaging) looks attractive but is not informative in the absence of relevant genetic perturbations as comparisons. These types of experiments would benefit greatly from PLEKHG3 loss of function analysis, as well as mutational analysis in the over-expression setting.*

Answer: We have redoubled our efforts to generate a PLEKHG3 CRISPR-Cas KO in HeLa, going up to 10 sg-guides, and hope that we will be successful in the next future. This cell line will be helpful in answering the Reviewer’s question.

Mutational analysis cannot be performed, because of the lack of binding between LAMTOR3 and PLEKHG3, which leaves us without a read-out.

3. Description of the revisions that have already been incorporated in the transferred manuscript

Please insert a point-by-point reply describing the revisions that were already carried out and included in the transferred manuscript. If no revisions have been carried out yet, please leave this section empty.

Reviewer 1:

• The authors should also confirm the specificity of the PLEKHG3 antibody in immunofluorescence using control and PLEKHG3 siRNA in untransfected cells that have not been transfected with GFP-PLEKHG3 (as is shown in Fig. S2C). Numerous antibodies recognize the overexpressed protein but do not recognize the same protein at endogenous expression levels.

Answer: To assess the specificity of the antibody for endogenous PLEKHG3 we have used HEK293T cells, which based on the fact that PLEKHG3 is most highly expressed in neuronal cells (https://www.proteinatlas.org/ENSG00000126822-PLEKHG3/tissue#expression_summary) should yield a clearer endogenous signal. The results of this experiment are shown in the revised Figure S2B-C. The pattern of PLEKHG3-positive bands is similar to that observed in HeLa cells, and only the band around 250 kD is clearly reduced by the siPLEKHG3. The IF images show a selective loss of the PLEKHG3 signal in correspondence of actin filaments close to the plasma membrane, whereas the nuclear signal is preserved, and therefore to be considered non-specific (Figure only shown in attached revision plan and revised Figure S2B-C). In addition, we have redoubled our efforts, so far unsuccessful, to generate a PLEKHG3 CRISPR-Cas KO in HeLa cells, going up to 10 sg-guides, and hope that we will be successful in the next future.

Extract from revised Figure S2B-C: ____PLEKHG3 KD test in HEK293T cells: B) Western blot of HEK293T cells showing downregulation of PLEKHG3 expression upon siPLEKHG3 treatment compared to siScr. Bar plot shows quantification of PLEKHG3 bands from immunoblot above. Error bars = SEM, n=3. * = p values according to student's t-test. C) Immunofluorescence images of HEK293T cells. siPLEKHG3 shows drop in PLEKHG3 intensity in the periphery of the cell and less colocalization with Phalloidin. Scale bar = 50 µm. Line plots show intensity profiles of Phalloidin (green) and PLEKHG3 (red) along the white lines in the merged inset images. Scale bar = 10 µm.

• *The colocalization of endogenous PLEKHG3 and LAMP1 as depicted in figures 3B and 3C (data from fixed cells) is not convincing. PLEKHG3 appears to be present on cortical actin structures as opposed to being colocalized with LAMP1 on lysosomes. *

And related to this point:

• There is no apparent colocalization of PLEKHG3 and lysotracker in the movie S5.

Answer:

We do not claim that the two structures always colocalize. The images in Figure 3C are a schematic representation of the colocalization analysis shown in the plot and were included to explain how we define PLEKHG3 high/low regions or LAMP1 high/low regions, respectively. We agree with the Reviewer and with the previous literature that PLEKHG3 main localization is to cortical actin structures, as shown in Figure 3F of the original version and in Figure S2C (HEK293T cells) and in Figure S3A in HeLa cells in the revised version. The claim is rather that PLEKHG3 has been identified as a vicinal protein of LAMTOR3, seen by a fraction of lysosomes when they traffic into protrusions. We have clarified the text referring to Figure 3F on page 13, line 7-10 as follows:

"Immunofluorescence experiments showed the reported colocalization of endogenous PLEKHG3 (Figure S2C in HEK293T cells, Figure S3A in HeLa cells) and GFP-PLEKHG3 with cortical actin structures and the partial localization of LAMP1-positive vesicles to these structures in correspondence with vinculin-positive focal adhesions."

This specific claim is also based on the observation made in GFP-PLEKHG3-expressing cells (including movie S5, and particularly the stills of the leading edge in Figure 4F). In the text describing Figure 4F, we now clearly state on page 14, line 15-17: “Following a single cell over time, we could observe that __a subset of __lysosomes appears to travel to PLEKHG3 accumulation sites and specifically move into developing protrusions.”

• It is not clear how the authors conclude that Figure 4E graph shows "the LAMP1 signal was stronger in paxillin-labeled FA compared to control regions". The 4E graph shows LAMP1 signal in GFP versus GFP-PLEKHG3 and shows a modest enrichment of LAMP1 in FAs in GFP-PLEKHG3 overexpression. LAMP1 enrichment in FAs is also not obvious in the image shown in Figure 4B.

Answer: We stand corrected – the Figure we referred to was actually not in the manuscript. It has been inserted now, as a plot next to Figure 4B on page 16 Figure 4B (schematic representation of colocalization analysis) was designed to explain how we define focal adhesions (paxillin positive) and adjacent control regions (same size and shape, but paxillin-negative). The actual analysis was missing and has now been inserted. We apologize for this mistake.

We do not claim that PLEKHG3 brings lysosomes to FAs. The enrichment of lysosomes in FA regions of cells expressing GFP-PLEKHG3 compared to GFP-expressing cells shown in 4E, as the Reviewer correctly notes, is marginal and is not highlighted anywhere in the text exactly for this reason.

• In Fig. 2B, there appears to be a labeling error. The lanes 2,4 and 7 appear to be transfected with L3-T-V5 but labeled as GFP-V5-cyto. Here the PLEKHG3 band should be indicated.

• AND -Fig. 2C is an IP experiment as per the manuscript text but it is labeled as pulldown.

Answer: We stand corrected, and the necessary changes have been made in the revised version in Figure 2B on page 11.

Reviewer 2

*1 - Specificity of PLEKHG3 antibody: In Fig. S2, authors show that PLEKHG3 antibody recognizes 3 bands (above 100 kDa, above 130 kDa and 250 kDa) and all of them are reduced by the silencing of PLEKHG3. Then, in Fig. 2A and C, authors only show the band above 130 kDa, despite implying that the specific band should be "much higher than the 134 kDa calculated from the aminoacid sequence of the protein". *

In Fig. 2 B, they show all the bands shown in Fig. S2 and presumably favor that the specific and is the 250 kDa one. Finally, in Fig. 2D, they show all bands and note that the band above 130 kDa is not specific. Therefore, authors need to conclude what is the specific band and always analyze the same one, and, possibly, use a different antibody or purify this one to remove non-specific binding. Without this, the main result of the paper, cannot be substantiated.

Answer: We apologize for this misunderstanding. The antibody recognizes three bands, all reduced by siRNA treatment. These three bands are only resolved in the gels in Figure S2A and B, and in Figure 2B. The reason for this is the high molecular weight of the isoforms, that are resolved in these 8% gels, but collapse into one band in the 15% gels shown in Figure 2A and C. Therefore, the high molecular weight bands are not resolved under these conditions. 8% gels such as the ones in Figure 2B are needed to resolve the high molecular weight bands.

Figure 2D shows an 8% gel, and therefore all bands are visible. The band marked by an arrow is only present in the streptavidin pulldowns but not in the input or in the supernatant and is therefore considered unspecific. This has been clarified in the revised figure legend on page 11. In addition, to assess the specificity of the antibody for endogenous PLEKHG3 we have used HEK293T cells, which based on the fact that PLEKHG3 is most highly expressed in neuronal cells (https://www.proteinatlas.org/ENSG00000126822-PLEKHG3/tissue#expression_summary) should yield a clearer endogenous signal. The results of this experiment are shown in the Figure S2B-C of the revised manuscript. The pattern of PLEKHG3-positive bands is similar to that observed in HeLa cells, and only the band around 250 kD is clearly reduced by the siPLEKHG3. The IF images show a selective loss of the PLEKHG3 signal in correspondence of actin filaments close to the plasma membrane, whereas the nuclear signal is preserved, and therefore to be considered non-specific.

2 - In page 12, authors state that "These results indicated that PLEKHG3 is a transient interactor, or a proximal, not directly binding protein, of L3" and in page 14 that "... PLEKHG3 is a proximal L3 protein rather than a transient physical interactor". It is not clear at all how did the authors reach such conclusions, nor they have data to conclude this. Indeed, they would have to express the proteins in vitro and test their interaction to conclude about a direct binding. They also do not know what is the stability of the interaction.

Answer: This is also a misunderstanding. Unfortunately, we mislabeled Figure 2C as “pulldown”, rather than “IP”, as it is characterized in the text. The fact that we cannot co-ip PLEKHG3 by immunoprecipitating L3 using a V5 antibody led us to conclude that the interaction between the proteins is not direct or stable enough to survive a co-ip. Therefore, the most likely conclusion is that PLEKHG3 is a vicinal protein rather than an interactor of L3 – we changed the labeling of Figure 2C to clarify the issue on page 11.

Based on these negative data, we did not proceed to test the possibility of complex formation in vitro.

3 - Still in page 12, authors state that "... two different membrane structures, protrusions and ruffles". What do the authors mean exactly by "protrusions", as there are several different ones (e.g., lamellipodia, filopodia, pseudopods)? And how can they distinguish between ruffles and, for example, lamellipodia? They need to use markers and more carefully analyze their morphology to be able to distinguish these. Like this, it is too preliminary.

Answer: It was our intention to indicate with the arrows the trajectories in the figure along which we measured the MFI of LAMP1 and PLEKHG3. Although this is indicated in the figure legend, it had apparently given the impression that the arrows indicated specific membrane structures. Since we are focusing on different types of membrane protrusions rather than ruffles, we replaced the terms "ruffles" and "protrusions" with the terms "elongated protrusions" (Figure 3D upper panel) and then compared these with "non" elongated protrusions” (Figure 3D lower panel). Indeed, we note that PLEKHG3 accumulation is possible below and along the plasma membrane, but colocalization with lysosomes occurs preferentially in elongated protrusions. We therefore amended the text on page 12, line 24 – page 13, line 5 as follows:

„More specifically, we found that PLEKHG3 colocalized more strongly with LAMP1-positive vesicles in elongated membrane structures (Figure 3D-E). Focal adhesion sites, which anchor the intracellular cortical actin network to the extracellular matrix and are remodeled with the help of late endosomes/lysosomes during protrusion formation and cell motility, can also be found in such elongated membrane protrusions (reviewed in Burridge and Burridge, 2017; Schiefermeier et al., 2014).”

6 - Regarding the statistical analysis, authors assert that it was done using Student's t tests, unless otherwise stated. However, they never refer in figure legends other statistical analysis methods. If so, they cannot use such test, for example, in cases where more than two groups are compared.

Answer: in all our experiments we performed two-group comparisons. We have now deleted “unless otherwise stated” from the Materials and Methods section on page 41, lines 1-2.

*Minor comments: *

*1 - In the abstract, authors refer that cytosolic proteins are recruited to platforms on the limiting membrane of lysosomes. What do they mean by "platforms"? Is it microdomains? *

Answer: We apologize for this unclarity and have now changed the first sentence in the abstract on page 1 to “Lysosomes are key organelles involved in metabolic signaling pathways through their ability to recruit cytosolic molecules to protein platforms bound to the lysosomal membrane”. We refer to protein platforms as multifunctional protein complexes that can recruit and assemble signaling components (e.g., the recruitment of mTORC1 activating proteins by the LAMTOR complex).

*2 - In the Introduction, there is a period before the reference at the end of the first paragraph. *

Answer: We stand corrected. See changes on page 3, line 8.

3 - In the results, Fig. 1E is mentioned before Fig. 1D and Figure S1F before Fig S1E, which can be confusing.

Answer: Figure S1E on page 6 was mislabeled as Figure 1E and Figure S1K on page 9 was mislabeled as Figure 1K. We stand corrected. See changes on page 6, line 20 and page 9, line 4.

4 - All the immunofluorescence images need to be bigger, in general, and have zoom-ins, except Fig. 3A, 4B, 4F, and S2C. Also, in Fig. S1F, the green channel has different intensities and the V5-lyso signal is clearly saturated. Finally, Fig. S1D, S1I and S3F must be enlarged, too.

Answer: We appreciate the Reviewer's suggestion, but enlarging all the immunofluorescence images and including zoom-ins would make the manuscript overly crowded and could distract from the main findings. Regarding the expression levels of the baits, as mentioned in the manuscript, we aimed to express them at near-endogenous levels. However, TMEM192 is expressed at higher levels than LAMTOR3 in these cells, which may have resulted in the observed discrepancy. We hope the Reviewer will understand our decision and find the current presentation of the data clear and informative.

5 - In page 9, where it reads "Figure 1K", should read "Figure S1K".

Answer: See answer to minor point 3.

6 - The observation that PLEKHG3 silencing leads to loss of the perinuclear clustering of LAMP1-positive vesicles, and increase in their accumulation at the cell tips, is not referred in the text.

Answer: While this might seem the case in part of the cells shown in the representative image in Figure S2C, quantification of lysosome distribution did not show a significant difference throughout the population as displayed in the figure below (Figure only shown in attached revision plan).

__Figure 1 for Reviewer 2: __Lysosomal distribution in HeLa cells transfected with either siScr or siPLEKHG3. X-axis is relative distance from the nucleus and Y-axis the normalized intensity of the LAMP1 channel. Results are averages of >30 cells from one experiment.

7 - Fig. 2C is not referred in the legend.

Answer: We stand corrected and have changed the legend of Figure 2 accordingly on page 11.

8 - Figure S3A and B: authors should show the colocalization of endogenous PLEKHG3 with phalloidin and not only the GFP-tagged protein.

Answer: We thank the Reviewer for this comment and have performed this experiment showing the colocalization of endogenous PLEKHG3 with F-actin structures stained by Phalloidin. Even though the endogenous PLEKHG3 staining in HeLa cells is rather weak, sites where membrane protrusions are formed are clearly marked with PLEKHG3 staining below the plasma membrane. These data confirm the specific colocalization of PLEKHG3 with Phalloidin shown in the revised Figure S3A. See also the extract from Figure S3A below (Figure only shown in attached revision plan and revised Figure S3A).

Extract from revised Figure S3A: Immunofluorescence images of HeLa cells. A) HeLa cells stained with PLEKHG3 (red) and Phalloidin (green). The nucleus is indicated by DAPI staining (blue). Scale bar = 50 µm. Insets on the right as indicated by white box in image on the left. Scale bar = 10 µm. Line plot corresponds to white line in merged inset.

*9 - In page 14, authors refer to Fig. 3G, which does not exist. *

Answer: We stand corrected, the sentence on page 14, line 9 refers to Figure S3G.

10 - In page 30 and page 32, different antibodies for LAMP1 and PLEKHG3 are mentioned, but in the figure legends authors do not refer which one they used.

Answer: We tried different PLEKHG3 antibodies but ended up using only one. The other antibody has been excluded from the list on page 32, line 18. We have specified which LAMP1 antibodies were used in which Figure in the Material and Methods on page 30, lines 17.20-23.

11 - In page 33, where it reads "300 µm protein", it should probably read "300 µg protein".

Answer: We stand corrected and thank the Reviewer. See changes on page 33, line 17.

Reviewer 3

• A key issue … is that the authors focus solely on peripheral lysosomes as target compartments for PLEKHG3. This is not self-evident, particularly in light of images presented in Figures 2 and 3, where colocalization of PLEKHG3 with perinulcear lysosomes appears very likely. The authors should make differences/similarities they observe between effects on perinuclear versus peripheral lysosomes explicit both with data and in the text, if such differences exist.*

Answer: The Reviewer is likely addressing the images in Figure 3, which have been obtained by staining endogenous PLEKHG3 and do show diffuse staining around the nucleus. This perinuclear staining is resistant to siPLEHG3 (revised Figure S2C) and is not observed with the GFP-PLEKHG3 fusion protein (revised Figure S2E-F; including PLEKHG3 knockdown), which gives a less diffuse signal. This is why we are confident about the colocalization of PLEKHG3 with peripheral lysosomes. This said, we have redoubled our efforts to generate a PLEKHG3 CRISPR-Cas KO in HeLa, going up to 10 sg-guides, and hope that we will be successful in the next future. This cell line will be helpful in answering the Reviewer’s question.

Minor point: 1. Multicolor overlays with one of the channels in white is in my view not reader-friendly. Appreciating colocalization between endosomes/lysosomes, actin and G is very important for this study, and while is typically reserved to show overlay between green and magenta or green (standard for 2 channels), red and blue (standard for 3-channels). I therefore advise the authors to choose a different color combination throughout the figures when presenting microscopy images.

Answer: White as a channel color has been substituted for with red (in the 2- and 3-color images) or with blue (in the 4-color images) throughout the images of the revised manuscript, except for the stills from the videos that have not been changed because no colocalization analysis has been performed in this case.

4. Description of analyses that authors prefer not to carry out

Please include a point-by-point response explaining why some of the requested data or additional analyses might not be necessary or cannot be provided within the scope of a revision. This can be due to time or resource limitations or in case of disagreement about the necessity of such additional data given the scope of the study. Please leave empty if not applicable.

Reviewer 2

4 - At least Fig. 2F and 3A need quantification. Regarding cell motility, there is no quantification and the authors must perform a quantitative assay (despite stating that "As another measure of cell motility, analysis of the number of forming protrusions and retracting membranes..."). Not only this is not a measure of cell motility, but there the issue of what are "protrusions" referred above. Therefore, authors need to quantify the distance that the cells move and/or perform quantitative motility/migration assays.

Answer: We appreciate the Reviewer’s attention to detail and agree that the quantification of these figures is essential to understand the results. We believe that the Reviewer is referring to Figure 3F and Figure 4A, as there is no Figure 2F, and Figure 3A only confirms the localization of endogenous PLEKHG3, as previously reported in (Nguyen et al., PNAS 2016). If our assumption is correct, then the salient aspects of Figure 3F, which is a representative image, are quantified in Figure 3C-E (endogenous PLEKHG3 colocalization colocalization with LAMP1/lysosomes) and Figure 4E and 5F-G (FA with LAMP1/lysosomes). Figure 4A is quantified in Figure 4C-E (GFP-PLEKHG3 colocalization with FAs, this time labeled with paxillin, and LAMP1 colocalization with FAs).

In response to the Reviewer's comment regarding the absence of quantification for cell movement/migration in our study, we apologize for any confusion that may have arisen from our use of the term "cell motility." We did not use this term to describe exclusively directed cell movement, but rather in a broader sense, to encompass the entirety of membrane remodeling processes involved in migration. In this context, our statement that lysosomal subcellular localization plays a role in cell motility remains valid. The relationship between membrane protrusive activity and motility is evident from our observations in cells overexpressing KIF1A-mCherry, where both membrane remodeling/protrusive activity and movement are significantly impaired compared to control cells (refer to Movie S7 vs. S6 and S10 vs. S9).

To address the Reviewer's concern, we have clarified our definition of motility in the introduction by stating on page 5, lines 1-4: "We demonstrate that PLEKHG3 colocalizes with lysosomes at focal adhesion and protrusion sites, and that the localization and function of this protein, as well as cell motility – which we define as the sum of membrane remodeling processes involved in migration – depend on lysosomal dynamics." This revision ensures that our results are accurately described and minimizes any potential confusion. Additionally, we have removed the statement on page 23, line 1 of the original manuscript. We apologize for any confusion our original wording may have caused and appreciate the opportunity to clarify our intentions.

Reviewer 3

1. • The mechanism of PLEKHG3 action on lysosomes/late endosomes is underdeveloped in my view. In the absence of for instance mutational analyses to examine what drives the interaction of PLEKHG3 with LAMTOR3, as well as delineation of at least some molecular consequences of this binding, the study remains incomplete.*

Answer: We are grateful for the Reviewer's feedback and concur that gaining insight into the mechanistic details of PLEKHG3's interaction with LAMTOR3 would be beneficial. However, our investigation revealed that PLEKHG3 is a transient interactor/proximal protein of LAMTOR3, and due to the absence of direct binding between LAMTOR3 and PLEKHG3 (Figure 2C on page 11), we are unable to perform a mutational analysis of this interaction, as it lacks a direct read-out.

Furthermore, as demonstrated in Figure S3H-L, LAMTOR3 ablation does not affect PLEKHG3 subcellular localization, suggesting that delving deeper into the molecular consequences of their interaction may be highly complex and beyond the scope of this study. We kindly ask the Reviewer to bear with us on this point, considering the novelty of our findings, which illuminate the interplay between lysosomes and actin dynamics as well as the role of PLEKHG3 in regulating cell protrusions—findings not previously reported in the literature.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

The manuscript by Ettelt et al describes the identification of PLEKHG3 as a collaborator of the LAMTOR complex on lysosomes using proximity-based biotinylation. The biotinylation screen is well executed and controlled. The authors choose to follow up on PLEKHG3, a protein involved in actin dynamics, which they refer to as understudied (I let the validity of the latter statement to be evaluated by the editor). Generally speaking, the data are of good quality, and the manuscript is clear and well written. However, much of the evidence on the role of PLEKHG3 on lysosomes is suggestive at best and further investigation into its mechanisms of action is warranted. Some important points to address prior to publication are detailed below.

Major Points:

1. The mechanism of PLEKHG3 action on lysosomes/late endosomes is underdeveloped in my view. In the absence of for instance mutational analyses to examine what drives the interaction of PLEKHG3 with LAMTOR3, as well as delineation of at least some molecular consequences of this binding, the study remains incomplete.
2. A key issue possibly (but not necessarily) related to the point above is that the authors focus solely on peripheral lysosomes as target compartments for PLEKHG3. This is not self-evident, particularly in light of images presented in Figures 2 and 3, where colocalization of PLEKHG3 with perinulcear lysosomes appears very likely. The authors should make differences/similarities they observe between effects on perinuclear versus peripheral lysosomes explicit both with data and in the text, if such differences exist.
3. Data presented in Figure 6 showing cell motility analysis is interesting and has potential to make the manuscript impactful. Similarly, data in Figure 4F (live cell imaging) looks attractive but is not informative in the absence of relevant genetic perturbations as comparisons. These types of experiments would benefit greatly from PLEKHG3 loss of function analysis, as well as mutational analysis in the over-expression setting.

Minor point

1. Multicolor overlays with one of the channels in white is in my view not reader-friendly. Appreciating colocalization between endosomes/lysosomes, actin and G is very important for this study, and while is typically reserved to show overlay between green and magenta or green (standard for 2 channels), red and blue (standard for 3-channels). I therefore advise the authors to choose a different color combination throughout the figures when presenting microscopy images.

Significance

In principle, I consider this study to be of interest to the community of cell biologists working on the endolysosomal system and/or the actin cytoskeleton and its relationship to intracellular membranes. However, the authors find themselves in a rather crowded field. I feel that developing the mechanism of action of PLEKHG3 on lysosomes beyond this first submission could help with boosting the impact of the study. There is clearly something interesting going on, but what that is exactly, remains unclear in my view.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary: The authors use proximity-dependent labelling and mass spectrometry to identify cytoplasmic proteins that interact with lysosomes. They show that PLEKHG3 interacts with the LAMTOR complex; that PLEKH3 accumulates in focal adhesion sites, where it colocalizes with peripheral lysosomes; and that the increased translocation of lysosomes to the periphery leads to less "protrusions", as well as rounder cells and less motile cells.

Major comments: While the study is generally carefully performed and thorough, there are major shortcomings that affect the conclusions taken, namely the specificity of the PLEKHG3 antibody, the identification of "protrusions" and ruffles, several quantifications missing, and the data used to conclude about cell motility. There are also conclusions for which there is no concrete or solid evidence.

Specific issues:

1. Specificity of PLEKHG3 antibody: In Fig. S2, authors show that PLEKHG3 antibody recognizes 3 bands (above 100 kDa, above 130 kDa and 250 kDa) and all of them are reduced by the silencing of PLEKH3. Then, in Fig. 2A and C, authors only show the band above 130 kDa, despite implying that the specific band should be "much higher than the 134 kDa calculated from the aminoacid sequence of the protein". In Fig. 2 B, they show all the bands shown in Fig. S2 and presumably favor that the specific and is the 250 kDa one. Finally, in Fig. 2D, they show all bands and note that the band above 130 kDa is not specific. Therefore, authors need to conclude what is the specific band and always analyze the same one, and, possibly, use a different antibody or purify this one to remove non-specific binding. Without this, the main result of the paper, cannot be substantiated.
2. In page 12, authors state that "These results indicated that PLEKHG3 is a transient interactor, or a proximal, not directly binding protein, of L3" and in page 14 that "... PLEKHG3 is a proximal L3 protein rather than a transient physical interactor". It is not clear at all how did the authors reach such conclusions, nor they have data to conclude this. Indeed, they would have to express the proteins in vitro and test their interaction to conclude about a direct binding. They also do not know what is the stability of the interaction.
3. Still in page 12, authors state that "... two different membrane structures, protrusions and ruffles". What do the authors mean exactly by "protrusions", as there are several different ones (e.g., lamellipodia, filopodia, pseudopods)? And how can they distinguish between ruffles and, for example, lamellipodia? They need to use markers and more carefully analyze their morphology to be able to distinguish these. Like this, it is too preliminary.
4. At least Fig. 2F and 3A need quantification. Regarding cell motility, there is no quantification and the authors must perform a quantitative assay (despite stating that "As another measure of cell motility, analysis of the number of forming protrusions and retracting membranes..."). Not only this is not a measure of cell motility, but there the issue of what are "protrusions" referred above. Therefore, authors need to quantify the distance that the cells move and/or perform quantitative motility/migration assays.
5. It is not clear if in cells KO for PLEKHG3, the overexpression of KIF1A leads to more lysosomes localizing close to the PM, as well as more protrusions and more cell motility, as the authors only compare cell overexpressing GFP or GFP-PLEKHGL3.
6. Regarding the statistical analysis, authors assert that it was done using Student's t tests, unless otherwise stated. However, they never refer in figure legends other statistical analysis methods. If so, they cannot use such test, for example, in cases where more than two groups are compared.

Minor comments:

1. In the abstract, authors refer that cytosolic proteins are recruited to platforms on the limiting membrane of lysosomes. What do they mean by "platforms"? Is it microdomains?
2. In the Introduction, there is a period before the reference at the end of the first paragraph.
3. In the results, Fig. 1E is mentioned before Fig. 1D and Figure S1F before Fig S1E, which can be confusing.
4. All the immunofluorescence images need to be bigger, in general, and have zoom-ins, except Fig. 3A, 4B, 4F, and S2C. Also, in Fig. S1F, the green channel has different intensities and the V5-lyso signal is clearly saturated. Finally, Fig. S1D, S1I and S3F must be enlarged, too.
5. In page 9, where it reads "Figure 1K", should read "Figure S1K".
6. The observation that PLEKHG3 silencing leads to loss of the perinuclear clustering of LAMP1-positive vesicles, and increase in their accumulation at the cell tips, is not referred in the text.
7. Fig. 2C is not referred in the legend.
8. Figure S3A and B: authors should show the colocalization of endogenous PLEKHG3 with phalloidin and not only the GFP-tagged protein.
9. In page 14, authors refer to Fig. 3G, which does not exist.
10. In page 30 and page 32, different antibodies for LAMP1 and PLEKHG3 are mentioned, but in the figure legends authors do not refer which one they used.
11. In page 33, where it reads "300 µm protein", it should probably read "300 µg protein".

Significance

The study provides evidence that lysosome positioning can affect cortical actin cytoskeleton dynamics, as well as cell shape and motility. Experiments are in general thorough and data subjected to quantification. However, there are fundamental conclusions that are preliminary at this stage, as some of the data is not yet solid enough. Therefore, it needs to be further strengthened to be considered for publication. In general, it reads well but the amount of abbreviations (e.g. in the case of the constructs) makes it somehow difficult to follow. The study will be interesting for the cell biology, membrane trafficking and cytoskeleton dynamics communities.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

The manuscript by Ettelt et al., describes identification of Rho guanine nucleotide exchange factor- PLEKHG3 as one of the positive hits from a TurboID proximity-dependent labeling screen using LAMTOR3 (one of the subunits of the pentameric LAMTOR complex/Ragulator) as a bait protein. The authors find that PLEKHG3 colocalizes with lysosomes at focal adhesions and that peripheral clustering of lysosomes promotes PLEKHG3 localization near the plasma membrane, and also inhibits protrusion formation and cell motility. The experiments, particularly the Turbo ID proximity-dependent labeling screen, are well-executed, and the imaging data is aptly quantified. The manuscript explores an exciting question of how lysosome positioning regulates cortical actin dynamics and thereby cell motility.

Major comments:

• The colocalization of endogenous PLEKHG3 and LAMP1 as depicted in figures 3B and 3C (data from fixed cells) is not convincing. PLEKHG3 appears to be present on cortical actin structures as opposed to being colocalized with LAMP1 on lysosomes. The authors should also confirm the specificity of the PLEKHG3 antibody in immunofluorescence using control and PLEKHG3 siRNA in untransfected cells that have not been transfected with GFP-PLEKHG3 (as is shown in Fig. S2C). Numerous antibodies recognize the overexpressed protein but do not recognize the same protein at endogenous expression levels.

Moreover, do the authors observe colocalization between GFP-PLEKHG3 and lysotracker in living cells? There is no apparent colocalization of PLEKHG3 and lysotracker in the movie S5. - The authors observe that GFP-PLEKHG3 is concentrated at the cell's periphery when KIF1A is overexpressed, whereas RUFY3 overexpression results in more cytosolic staining. To bolster their conclusion that a change in lysosomal positioning alters the subcellular localization of PLEKHG3, it is preferable to employ inducible techniques, such as the recently described "reversible association with motor proteins" (RAMP) (PMID: 31100061). The method is a rapid and reversible method for altering organelle positioning. It is still unknown whether PLEKHG3 is associated with lysosomes and mechanism of how positioning of lysosomes affects PLEKHG3 localization. - Similarly to the preceding point, the claim that "peripheral accumulation of lysosomes inhibits protrusion formation and limits cell motility" should be tested more rigorously using the RAMP method, preferably in living cells. Other approaches, such as overexpression/siRNA of Arl8b and other motor adaptors, such as SKIP/PLEKHM2, can be used to alter lysosome positioning and confirm this central findings of the manuscript. The authors could also consider including additional mechanistic data in order to comprehend how lysosome positioning controls cell motility. For instance, the RAMP approach could be employed to investigate cortical actin dynamics upon repositioning of lysosomes to the peripheral/perinuclear region. - It is not clear how the authors conclude that Figure 4E graph shows "the LAMP1 signal was stronger in paxillin-labeled FA compared to control regions". The 4E graph shows LAMP1 signal in GFP versus GFP-PLEKHG3 and shows a modest enrichment of LAMP1 in FAs in GFP-PLEKHG3 overexpression. LAMP1 enrichment in FAs is also not obvious in the image shown in Figure 4B. - In Fig. 2B, there appears to be a labeling error. The lanes 2,4 and 7 appear to be transfected with L3-T-V5 but labeled as GFP-V5-cyto. Here the PLEKHG3 band should be indicated. - Fig. 2C is an IP experiment as per the manuscript text but it is labeled as pulldown.

Significance

Using a TurboID proximity-dependent labelling screen, the authors identified an interesting subset of actin-remodeling proteins that interact with the lysosomal protein LAMTOR3. The authors further characterized one of these proteins, PLEKGH3, and found that lysosome positioning regulates PLEKGH3 localization, as well as plasma membrane protrusion formation and cell motility. This study suggests that lysosome peripheral accumulation could regulate cortical actin remodelling and consequently cell migration by regulating PLEKGH3 localization (although this is not tested in the manuscript). This study adds to the previous findings that microtubule-based transport of late endosomes regulate delivery of late endosomal LAMTOR proteins to the vicinity of focal adhesions, which in turn, regulate focal adhesion dynamics. The mechanism of how lysosomes can influence actin remodeling will be important in the context of cancer cell migration. My area of expertise is lysosome fusion and motility and I have limited expertise in regulation of actin dynamics and how Rho family members regulate actin remodeling.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

Manuscript number: RC-2023-01846

Corresponding author(s): Anastassis Perrakis

1. General Statements

We thank the reviewers for the feedback, highlighting the synergy between computational modeling approaches and the experimental techniques we used to study the interaction between JBP1 and J-DNA. We would like to re-iterate that this approach has led to new findings regarding JBP1 and J-DNA interactions, namely:

• We identified and validated an additional interface in DNA binding domain of JBP1 (JDBD), that contributes to J-DNA binding.
• Through analysis of the AlphaFold model of JBP1, we propose how the Thymidine Hydroxylase domain (THD) of JBP1 binds J-DNA, and how the JDBD and THD domains are connected flexibly but explicitly to each other.
• The AlphaFold model allowed to hypothesize that the N-terminus of JBP1 is contributing to J-DNA binding, which we confirmed experimentally. These findings collectively suggest a mechanistic and structural basis on the synergy between the JDBD, the THD and the N-terminus of JBP1, providing a possible explanation to the previously observed conformational changes of JBP1 upon J-DNA binding. Our findings on the conservation of the N-terminus region and the new interface of JDBD, could be offering an explanation on the differences on how essential base-J is for different Trypanosomidaespecies. They also offer a first glimpse of how these domains synergize to provide new insights in the semi-conservative replication mechanisms of the base-J epigenetic marker in kinetoplastids.

2. Point-by-point description of the revisions

Reviewer #1____: ____Evidence, reproducibility and clarity

Summary:

The paper presents a combined experimental (X-ray, SAXS, mutational analysis) and computational (MD, docking, AlphaFold) work that elucidates the mechanism of JDBD:JDNA complex formation.

Major comment:

• How did the authors decide the timescale of the production run? Wouldn't the loop motion (which can be necessary for this study) occur on a timescale of 300+ ns?

While our simulations showed overall stability of the simulated protein (as reflected by the RMSD time series), the RMSF provided clear indications for differences in flexibility in the loops and termini of JBP1. We believe that performing an MD simulation of 100 ns in duplo samples the flexibility and behavior of the JBP1 DNA binding domain (JDBD) sufficiently for obtaining templates for docking studies, which was the purpose of running the simulations. We emphasized this in the manuscript (page 10) by adding: “and obtain additional templates for further docking studies”.

Minor comments:

• Did I understand correctly that the LINCS algorithm constrained only hydrogen-involving bonds? It is not mentioned explicitly. Or am I missing something?

LINCS was indeed used to only constrain hydrogen-involving bonds. We made this more explicit in the MD protocol described in the method section of the manuscript: “The LINCS algorithm (40) was used to constrain hydrogen-involving bond lengths to their zero-energy value”.

• The authors should increase the resolution of figures S1 and S3. They look a bit blurry.

We apologize for that; we tried to improve that by adjusting the figure sizes, but we are constrained by the output from e.g., the Bitclust software. We sincerely hope that the current resolution does not present an obstacle to the reader.

Significance:

The J DNA base is critical for transcription termination at the ends of the polycistronic gene clusters in Leishmania and related species. Hence, understanding the formation mechanism of the JDBD:J-DNA complex can provide an opportunity to develop novel chemotherapeutic treatments against these diseases. This work provides the first crystal structure of JDBD with the disordered loop and suggests that R448 and N455, as well as the N-terminus, are involved in the J-DNA binding process. The article is well-written and can interest readers from biological and biochemical societies. However, my field of expertise is computational chemistry and biochemistry; therefore, I cannot adequately evaluate the accuracy of the experimental techniques used in this work.

We thank the reviewer, but would like to emphasize that our work goes well beyond the new interface of the JDBD, offering significant new insights on the synergy between the THD, the JDBD and the newly identified N-terminus binding to J-DNA, as also outlined in the general summary above.

__Reviewer #2_:_ Evidence, reproducibility and clarity __

The manuscript by de Vries et al. reported the crystal structure of the J-DNA binding domain of JBP1. Although, the structure was already solved, the new structure allows to observe a loop that was disordered in the previous structure. This structure was next used to propose models of DNA complex analyzed by MD. The proposed model was then validated by mutagenesis.

Overall, the findings are interesting and the technical quality of the work is high.

We believe, as we outlined in the summary, that our work goes well beyond showing the new structure and validate the new interface by mutagenesis. In our view, the major findings of the paper have to do with the AlphaFold modeling analysis and validation, and the finding that the N-terminus of JBP1 is involved in DNA binding, something that is not only new, but has also been totally unexpected.

Comments:

-For clarity, a figure showing the domain organization of JBP1 could help the reader in the introduction part.

The domain architecture of JBP1 was added to Figure 1 as panel B.

-In addition to Figure 2 showing the newly observed loop in 2Fo-Fc map, an omit map should be included in Supp data.

A figure of the omit map was added to the supplemental information as Supplemental Figure S1.

-Figure S6 legend should be more precise about the type of HDX MS analysis.

As the HDX-MS data and methods were described in detail in previous work (Heidebrecht et al. 2011), we left the details out in the current manuscript. The reference to this paper was added to Figure S6 for clarity.

-The authors performed MD simulations. But what about DNA curvature upon complex formation?

For the JDBD MD simulations, we did not add at all the (J-)DNA and the current simulations provide no information about its curvature. As mentioned in the discussion, we do expect conformational changes when the J-DNA:JBP1 complex forms, and this likely includes DNA curvature as well as conformational changes between the protein domains. We felt that the current data would not allow to extract new insights from such complicated simulations.

-p.12 Some mutants were characterized, notably their melting temperatures. One mutant shows decreased stability while R448A shows increased stability. What is the structural explanation?

Indeed, the E437A mutant (that showed lower expression compared to the other mutants) showed decreased thermal stability. The R448A shows an increase in stability (~3o C), and so does the H440 mutant (~2o C). While there is no specific structural explanation for these observations, in general mutation to alanine reduces the entropy-loss upon protein folding. The reason we comment about the stability is to point out that the dramatically decreased binding of the N455A and R448A mutants is not due to a decrease of the protein stability. This is now clarified in the manuscript: “the other mutants are as stable or slightly more stable compared to the wild-type, suggesting that the DNA-binding analysis is not affected significantly by altered protein stability.”

-Figure 6E: The chi2 value for the comparison of the experimental curve for THD domain and calculated curve is very high indicating a poor fit. What is the explanation?

The χ2 value indeed reflects differences between the JBP1 THD selected from the AlphaFold model and the structure used in the SAXS experiment. The experimental model is the so-called ΔDBD, which is the full length JBP1, where the JDBD is missing (it has been “spliced out”). Hence, the connecting loops and the N-terminus are present in this ΔDBD structure, whereas in the THD of the AlphaFold model, these parts of the structure were left out. In other words, while the full-length computational model refers to the exact same purified protein, the computationally truncated model and the purified protein for the experiments, have actual differences. Thus, the shape and fit of the experimental curve to the calculated curve can be considered pretty good. This is now clarified in the manuscript by adding: “The χ2 value of the fit is slightly elevated due to presence of the connecting loops between the THD and the JDBD and the N-terminus in the protein used for measuring the SAXS curve, which were removed from the computational model.”.

Significance:

The novelty of the manuscript mainly relies on the description of the crystal structure of JDBD protein without DNA and proposed models of DNA complex within full length protein, models that were validated by mutations or truncations. The current manuscript well suited for a specialized journal.

These findings are indeed novel, especially the discovery and validation of the new interface of JDBD to J-DNA. We want to iterate that we are most excited by the totally unexpected and mechanistically important discovery of the role of the N-terminus of JBP1, that brings together legacy data and raises interest for additional structural studies.

__Reviewer #3_: _Evidence, reproducibility and clarity __

Base J, also known as β-D-glucopyranosyloxymethyluracil, is a modified form of thymidine that has been identified in kinetoplastids and related organisms. It is worth noting that the distribution of Base J in the genome may vary depending on the organism and its life stage and influences its genome dynamics. The synthesis of this hypermodified nucleotide occurs in two steps, which involve the participation of two distinct thymidine hydroxylases, namely J-binding protein 1 and 2 (JBP1 and JBP2), along with a β-glucosyl transferase. In this study, the authors have presented a crystal structure of JBP1 J-DNA binding domain (J-DBD), which includes a previously reported disordered loop that might be involved in JBP1:J-DNA contact. Using this disordered structure as a starting point, the authors conducted Molecular Dynamics simulations and computational docking studies to propose models for the recognition of J-DNA by JBP1 J-DBD. The results from these studies were validated by punctual mutagenesis experiments, which provided additional data for docking and revealed a binding pattern for JBP1 J-DBD on J-DNA. By combining the crystallographic structure of the TET2 JBP1-homologue in complex with DNA and the AlphaFold model of full-length JBP1, the authors have hypothesized that the flexible N-terminus of JBP1 contributes to DNA-binding, which they have confirmed experimentally. However, according to the authors, to gain a deeper understanding of the unique molecular mechanism that underlies the replication of epigenetic information, an experimental determination of a high-resolution JBP1:J-DNA complex involving conformational changes would be necessary. Nevertheless, the present proposed objectives were fully contemplated by the authors.

Major comments:

In my opinion, the present article effectively achieved all the described objectives using appropriate and reproducible methodology, including protein expression and crystallization analysis, Molecular Dynamics analysis using GROMACS-2020.2 software, docking analysis, punctual mutations analyses, and modeling of JBP1:J-DNA complex using the AlphaFold tool. The authors presented the results in a logical and organized manner, making it easy for readers to extract the most important points. However, I believe that the section titled "Results and Discussion" contains more "results" than "discussion". While I understand that the literature on JBPs and base J is still in its early stages, other species of kinetoplastids have JBP1, in which mutations were not lethal as in L. tarentolae (e.g. T. brucei). Therefore, providing information about the structure of JBP1 and how the present results relate to what is known about JBP1 in other species in terms of structure and J-DNA interactions would significantly enrich the discussion of the findings and reinforce their significance and impact. Thus, the authors should have been clearer about the impact of their findings. When discussing the results, the authors should have answered questions such as how the identification of the new residues involved in JBP1 J-DNA binding impacts the current model of JBP1:J-DNA interactions, how this improved model contributes to the understanding of base J synthesis, and if the new model can be extrapolated to other species of kinetoplastids, according to the conservation of JBP1 among them.

Although the article is more focused on protein research rather than parasite general molecular biology and medical studies, the findings may have implications for the development of new treatments for leishmaniases. Therefore, the authors should have discussed the potential of their new improved model as a target for lacking treatments of leishmaniases or at least brought up the point at conclusion section.

We thank the reviewer for pointing out that the comparison between kinetoplastid species could be described more explicit to highlight the impact of the presented results with respect to the variety in JBP1 sequence, and possibly contribute to understanding the role of base-J in the differences in lethality and transcription regulation within these species. We now elaborate on our results in multiple places in the manuscript:

• In the introduction we introduce the differences in lethality and transcription regulation within Leishmania and Trypanosoma (see also minor comment 2 below).
• An alignment of full-length JBP1 sequences of different Trypanosomatidae species was added as Supplemental Figure S10 to support the discussion below.
• The section describing the docking model of JDBD:J-DNA has been ammended with a description regarding the conservation of the residues involved in the binding interface: “The residues described in the JDBD:J-DNA interface are in general highly conserved (Supplemental Figure S12). Asp525 is fully conserved in Leishmania, Trypanosoma, Leptomonas and Bodo saltans species, so are Lys522, Arg532 and ArgR448. Asn455, which we identify in this study, is not conserved in Leptomonas, and Arg517 is not conserved in Trypanosoma also.”

• We renamed the final section to “Conclusions and Outlook” and added some discussion focusing on the conservation of the residues in the N-terminus and in the JDBD between different Trypanosomatidae Specifically, we now discuss how these could contribute in understanding the differences in lethality and transcription termination between Leishmania and Trypanosoma in the absence of base-J.

  Minor Comments:

• Please, re-check the affirmation "99% of base-J is found in telomeric repeats, mainly in GGGTTA repeats wherein the second thymine is modified to base-J (2-4)" in the Introduction. I believe that the distribution of base-J varies among different species of trypanosomatids and, therefore, cannot be generalized. Moreover, among different life stages in some organisms such as T. brucei and Leishmania major, differences on base-J distribution are found. The 99% of telomeric base-J mentioned would be a feature of Leishmania genus. Please, re-check the references 3 and 4.

Indeed, the referee is right to mention that the 99% is a Leishmania-specific observation. We have modified the introduction accordingly. “Base-J is found mainly in telomeric repeats and other repetitive sequences. In Leishmania 99% of base-J is found in telomers, mainly in GGGTTA repeats, wherein the second thymine is modified to base-J (2–4).”

Please, enrich the introduction topic with information about the model species, such as importance as pathological agent, its genomic organisation (core, subtelomeres, telomeres, what is present in subtelomeres, including base j) and polycistronic transcription and base J relevance on this aspect. That way, the reader will have a broad and more complete overview of the relevance of the present study.

We have enriched the introduction with a paragraph (“Leishmania species are uni-cellular […] essentiality of base-J for the life circle of these parasites.”) outlining the issues raised by the referee. As suggested by the referee, this makes it easier to both understand the relevance of the present study and to enrich the discussion about our findings discussed earlier in this letter.

Please, inform the expression vector for Leishmania tarentolae JBP1 used to express the mentioned protein on BL21(DE3)T1R.

The expression vector for JDBD JBP1 used for the crystallization was mentioned in Heidebrecht et al. 2011, which we refer to in the text. For clarity we added the vector to the first sentence in the protein expression and purification section in the material and methods: “using the pETNKI-his-3C-JBP1-JDBD plasmid”.

Please, supply the picture of the gel containing the extracted protein.

The gel of the JDBD mutants and the JBP1 N-terminus truncations was added to the supplemental information as Figure S10.

Significance:

Overall, this study provides important insights into the JBP1 and DNA interactions, which were lacking in the literature. The use of techniques such as protein expression and crystallization analysis, molecular dynamics, and docking analysis is in line with the research objectives. However, the lack of some information about methodology needs to be addressed (minor comments 3 and 4). Personally, methodology such as molecular dynamics and docking analysis is not easy to critique but the results are clear and understandable.

Although the authors should have been clearer about the impact of their findings, as addressed on my major comments, I believe that protein focused molecular parasitologists would benefit from the finds and methodology presented on this manuscript, since the article is more focused on protein research rather than parasite general molecular biology and medical studies, as mentioned on my major comments.

In summary, this study provides new insights into JBP1 and DNA interactions and uses appropriate and reproducible techniques to achieve its objectives. However, the authors should provide more clarity on the impact of their findings and discuss the potential of their new improved model.

My area of expertise: Cell and molecular biology; stem cells and factor controlling their fate; DNA, RNA, and molecular biology related techniques; Trypanosomatids telomere and telomerase

We would like to thank the referee for his positive and constructive outlook, which we believe resulted in changes that put the impact of our findings in clearer perspective.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Base J, also known as β-D-glucopyranosyloxymethyluracil, is a modified form of thymidine that has been identified in kinetoplastids and related organisms. It is worth noting that the distribution of Base J in the genome may vary depending on the organism and its life stage and influences its genome dynamics. The synthesis of this hypermodified nucleotide occurs in two steps, which involve the participation of two distinct thymidine hydroxylases, namely J-binding protein 1 and 2 (JBP1 and JBP2), along with a β-glucosyl transferase. In this study, the authors have presented a crystal structure of JBP1 J-DNA binding domain (J-DBD), which includes a previously reported disordered loop that might be involved in JBP1:J-DNA contact. Using this disordered structure as a starting point, the authors conducted Molecular Dynamics simulations and computational docking studies to propose models for the recognition of J-DNA by JBP1 J-DBD. The results from these studies were validated by punctual mutagenesis experiments, which provided additional data for docking and revealed a binding pattern for JBP1 J-DBD on J-DNA. By combining the crystallographic structure of the TET2 JBP1-homologue in complex with DNA and the AlphaFold model of full-length JBP1, the authors have hypothesized that the flexible N-terminus of JBP1 contributes to DNA-binding, which they have confirmed experimentally. However, according to the authors, to gain a deeper understanding of the unique molecular mechanism that underlies the replication of epigenetic information, an experimental determination of a high-resolution JBP1:J-DNA complex involving conformational changes would be necessary. Nevertheless, the present proposed objectives were fully contemplated by the authors.

Major comments:

In my opinion, the present article effectively achieved all the described objectives using appropriate and reproducible methodology, including protein expression and crystallization analysis, Molecular Dynamics analysis using GROMACS-2020.2 software, docking analysis, punctual mutations analyses, and modeling of JBP1:J-DNA complex using the AlphaFold tool. The authors presented the results in a logical and organized manner, making it easy for readers to extract the most important points. However, I believe that the section titled "Results and Discussion" contains more "results" than "discussion". While I understand that the literature on JBPs and base J is still in its early stages, other species of kinetoplastids have JBP1, in which mutations were not lethal as in L. tarentolae (e.g. T. brucei). Therefore, providing information about the structure of JBP1 and how the present results relate to what is known about JBP1 in other species in terms of structure and J-DNA interactions would significantly enrich the discussion of the findings and reinforce their significance and impact. Thus, the authors should have been clearer about the impact of their findings. When discussing the results, the authors should have answered questions such as how the identification of the new residues involved in JBP1 J-DNA binding impacts the current model of JBP1:J-DNA interactions, how this improved model contributes to the understanding of base J synthesis, and if the new model can be extrapolated to other species of kinetoplastids, according to the conservation of JBP1 among them. Although the article is more focused on protein research rather than parasite general molecular biology and medical studies, the findings may have implications for the development of new treatments for leishmaniases. Therefore, the authors should have discussed the potential of their new improved model as a target for lacking treatments of leishmaniases or at least brought up the point at conclusion section.

Minor Comments:

I have some minor comments regarding the text:

1. Please, re-check the affirmation "99% of base-J is found in telomeric repeats, mainly in GGGTTA repeats wherein the second thymine is modified to base-J (2-4)" in the Introduction. I believe that the distribution of base-J varies among different species of trypanosomatids and, therefore, cannot be generalized. Moreover, among different life stages in some organisms such as T. brucei and Leishmania major, differences on base-J distribution are found. The 99% of telomeric base-J mentioned would be a feature of Leishmania genus. Please, re-check the references 3 and 4.
2. Please, enrich the introduction topic with information about the model species, such as importance as pathological agent, its genomic organisation (core, subtelomeres, telomeres, what is present in subtelomeres, including base j) and polycistronic transcription and base J relevance on this aspect. That way, the reader will have a broad and more complete overview of the relevance of the present study.
3. Please, inform the expression vector for Leishmania tarentolae JBP1 used to express the mentioned protein on BL21(DE3)T1R.
4. Please, supply the picture of the gel containing the extracted protein.

Significance

Overall, this study provides important insights into the JBP1 and DNA interactions, which were lacking in the literature. The use of techniques such as protein expression and crystallization analysis, molecular dynamics, and docking analysis is in line with the research objectives. However, the lack of some information about methodology needs to be addressed (minor comments 3 and 4). Personally, methodology such as molecular dynamics and docking analysis is not easy to critique but the results are clear and understandable.

Although the authors should have been clearer about the impact of their findings, as addressed on my major comments, I believe that protein focused molecular parasitologists would benefit from the finds and methodology presented on this manuscript, since the article is more focused on protein research rather than parasite general molecular biology and medical studies, as mentioned on my major comments.

In summary, this study provides new insights into JBP1 and DNA interactions and uses appropriate and reproducible techniques to achieve its objectives. However, the authors should provide more clarity on the impact of their findings and discuss the potential of their new improved model.

My area of expertise: Cell and molecular biology; stem cells and factor controlling their fate; DNA, RNA, and molecular biology related techniques; Trypanosomatids telomere and telomerase

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

The manuscript by de Vries et al. reported the crystal structure of the J-DNA binding domain of JBP1. Although, the structure was already solved, the new structure allows to observe a loop that was disordered in the previous structure. This structure was next used to propose models of DNA complex analyzed by MD. The proposed model was then validated by mutagenesis.

Overall, the findings are interesting and the technical quality of the work is high.

Comments:

• For clarity, a figure showing the domain organization of JBP1 could help the reader in the introduction part.
• In addition to Figure 2 showing the newly observed loop in 2Fo-Fc map, an omit map should be included in Supp data.
• Figure S6 legend should be more precise about the type of HDX MS analysis.
• The authors performed MD simulations. But what about DNA curvature upon complex formation?
• p.12 Some mutants were characterized, notably their melting temperatures. One mutant shows decreased stability while R448A shows increased stability. What is the structural explanation?
• Figure 6E: The chi2 value for the comparison of the experimental curve for THD domain and calculated curve is very high indicating a poor fit. What is the explanation?

Significance

The novelty of the manuscript mainly relies on the description of the crystal structure of JDBD protein without DNA and proposed models of DNA complex within full length protein, models that were validated by mutations or truncations. The current manuscript well suited for a specialized journal.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary:

The paper presents a combined experimental (X-ray, SAXS, mutational analysis) and computational (MD, docking, AlphaFold) work that elucidates the mechanism of JDBD:JDNA complex formation.

Major comment:

• How did the authors decide the timescale of the production run? Wouldn't the loop motion (which can be necessary for this study) occur on a timescale of 300+ ns?

Minor comments:

• Did I understand correctly that the LINCS algorithm constrained only hydrogen-involving bonds? It is not mentioned explicitly. Or am I missing something?
• The authors should increase the resolution of figures S1 and S3. They look a bit blurry.

Significance

The J DNA base is critical for transcription termination at the ends of the polycistronic gene clusters in Leishmania and related species. Hence, understanding the formation mechanism of the JDBD:J-DNA complex can provide an opportunity to develop novel chemotherapeutic treatments against these diseases. This work provides the first crystal structure of JDBD with the disordered loop and suggests that R448 and N455, as well as the N-terminus, are involved in the J-DNA binding process. The article is well-written and can interest readers from biological and biochemical societies.

However, my field of expertise is computational chemistry and biochemistry; therefore, I cannot adequately evaluate the accuracy of the experimental techniques used in this work.

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

Learn more at Review Commons

---

Reply to the reviewers

We thank the reviewers for their comments and constructive suggestions to improve the manuscript. We are encouraged to see that both reviewers acknowledge how the results from our manuscript uses state-of-art technologies to advance molecular underpinnings of centriole length, integrity and function regulation. Both reviewers also highlighted that the manuscript is well laid out and presents clear, rigorous, and convincing data. Reviewer#1 described our manuscript of highest experimental quality and broad interest to the field of centrosome and cell biology form a basic research and genetics/clinical point of view. Here, we explain the revisions, additional experimentations and analyses planned to address the points raised by the referees. We will perform most of the experimentations and corrections requested by the reviewers. We have already made several revisions and are currently working on additional experiments.

Our responses to each reviewer comment in bold are listed below. References mentioned here are listed in the references section included at the of this document.

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

Summary: __In this manuscript, Arslanhan and colleagues use proximity proteomics to identify CCDC15 as a new centriolar protein that co-localizes and interacts with known inner scaffold proteins in cell culture-based systems. Functional characterization using state-of-the-art expansion microscopy techniques reveals defects in centriole length and integrity. The authors further reveal intriguing aberrations in the recruitment of other centriole inner scaffold proteins, such as POC1B and the SFI1/centrin complex, in CCDC15-deficient cells, and observe defects in primary cilia. __

We thank the reviewer for the accurate summary of the major conclusions of our manuscript.

Major points:

1) The authors present a high-quality manuscript that identifies a novel centriolar protein by elegantly revealing and comparing the proximity proteomes of two known centriolar proteins, which represents an important component for the maintenance of centrioles.

We thank the reviewer for highlighting that our manuscript is of high quality and presents important advances for the field.

__2) Data are often presented from two independent experiments (n = 2), which is nice, but also the minimum for experiments in biology. It is strongly recommended to perform at least three independent experiments. __

We agree with the reviewer that analysis of data form three experimental replicates is ideal for statistical analysis. We performed three replicates for the majority of experiments in the manuscript. However, as the reviewer pointed out, we included analysis from two experiments for the following figures:

• Fig. 4H: quantification of CCDC15 total cellular levels throughout the cell cycle by western blotting
• Fig. 5A: CCDC15-positive centrioles in control and CCDC15 siRNA-transfected cells
• Fig. 6B: % centriolar coverage of POC5, FAM161A, POC1B and Centrin-2 in control and CCDC15 siRNA-transfected cells
• Fig. 6C, 6E: Centrin-2 or SFI1-positive centrioles in control and CCDC15 siRNA-transfected cells
• Fig. 6J, K: normalized tubulin length and percentage of defective centrioles in cells depleted for CCDC15 or co-depleted for CCDC15 and POC1B
• Fig. 7F, H: SMO-positive cilia and basal body IFT88 levels in control and CCDC15 siRNA-transfected cells
• Fig. S3H: centriole amplification in HU-treated control and CCDC15 siRNA-transfected cells (no)
• Fig. S3A: centrosomal levels upon CCDC15 depletion There are two reasons for why we performed two experimental replicates for these experiments: 1) results from the two experimental replicates were similar, 2) quantification of data by U-ExM is laborious. To address the reviewer’s comments, we will perform the third experimental replicate for the sets of data that led to major conclusions of our manuscript, which are Figures 4H, 6C, 6E, 6J, 6K, 7F, 7H and S3A.

3) The protein interaction studies presented in Fig. 3 could be of higher quality. While it is great that the authors compared interactions to the centriolar protein SAS6, which is not expected to interact with CCDC15, the presented data raise many questions.

__a) In most cases, co-expression of tagged CCDC15 stabilizes the tested interaction partners, such that the overall abundance seems to be higher. The increase in protein abundance is substantial for Flag-FAM161A (Fig. 3D) and GFP-Centrin-2 (Fig. 3E) and is even higher for the non-interactor SAS6 (Fig. 3G), while it cannot be assessed for GFP-POC1B (Fig. 3F). Hence, the higher expression levels under these conditions make it more likely that these proteins are "pulled down" and therefore do not represent appropriate controls. __

We agree with the reviewer that the differences in protein abundance of the prey proteins upon expression of CCDC15 relative to control might impact the interpretation of the interaction data. To address this concern, we will perform the following experiments:

• To account of the potential stabilizing effects of CCDC15 expression, we will change the relative ratio of plasmids expressing proteins of interest and assess the expression of bait and prey protein levels. We will then repeat the co-immunoprecipitation experiments in conditions where prey expression levels are similar.
• To avoid the potential stabilizing effects of CCDC15 overexpression, we will perform immunoprecipitation experiments in cells expressing GFP or V5-tagged inner scaffold proteins and assess their potential physical or proximity interaction by blotting for endogenous CCDC15. __b) All Co-IP experiments are lacking negative controls in the form of proteins that are not pulled down under the presented conditions. __

For the co-IP experiments, we only included a specificity control for the interaction of the bait protein with the tag of the prey protein (i.e. GBP pulldown of GFP or GFP-CCDC15-expressing cells). As the reviewer suggested, we will also include a specificity control for the interaction of bait with the tag of the prey protein for co-immunoprecipitation experiments (i.e. GFP pulldown of cells expressing GFP-CCDC15 with V5-BirA* or V5-BirA*-FAM161A).

__c) The amounts of co-precipitation of the tested proteins appears very different. Could this reflect strong or weak interactors, or does it reflect the abundance of the respective proteins in centrioles? __

We agree with the reviewer that the quantity of the co-precipitated prey proteins might be a proxy for the interaction strength if the abundance of the bait proteins is similar. However, the expression levels of bait and prey proteins in co-transfected cells are different and thus, cannot be used to derive a conclusion on the interaction strength. For the revised manuscript, we will repeat the IP experiments and comment on this in the discussion section.

__4) The observation that IFT88 is supposedly decreased at the base of cilia in CCDC15-depleted cells requires additional experiments/evidence. Fig. 7G shows the results of n = 2 and more importantly, a similar reduction of gamma-tubulin in siCCDC15. Could the observed reduction in IFT88 be explained by a decrease in accessibility to immunofluorescence microscopy? Would the reduction in IFT88 at the base also be apparent when the signals were normalized to gamma-tubulin signals? __

To address the reviewer’s concern, we quantified the basal body gamma-tubulin and IFT88 levels in control and CCDC15-depleted cells and plotted the basal body IFT88 levels normalized to gamma-tubulin levels in Fig. 7H. Similar to the reduction in IFT88 levels, gamma-tubulin-normalized IFT88 levels was significantly less relative to control cells. Moreover, the gamma-tubulin basal body levels were similar between control and CCDC15 cells. We revised the gamma-tubulin micrographs in Fig. 7G to represent this. These results indicate that the reduction in basal body IFT88 levels upon CCDC15 depletion in specific.

__5) The observed Hedgehog signaling defects are described as follows: "CCDC15 depletion significantly decreased the percentage of SMO-positive cells". It is similarly described in the figure legend. If this was true, the simplest explanation would be that it reflects the reduction in ciliation rate (which is in a similar range). If SMO-positive cilia (instead of "cells") were determined, the text needs to be changed accordingly. __

As the reviewer pointed out, we quantified SMO-positive cilia, but not cells. We are sorry for this typo. We corrected SMO-positive cells as SMO-positive cilia in the manuscript text, Fig. 7 and figure legends.

__6) OPTIONAL: While expansion microscopy is slowly becoming one of the standard super-resolution microscopy methods, which is particularly well validated for studying centrioles, the authors should consider confirming part of their findings (as a proof of principle, surely not in all instances) by more established techniques. This could serve to convince critical reviewers that may argue that the expansion process may induce architectural defects of destabilized centrioles, as observed after disruptions of components, such as in Fig. 6. Alternatively, the authors could cite additional work that make strong cases about the suitability of expansion microscopy for their studies, ideally with comparisons to other methods. __

• SIM imaging was previously successfully applied for nanoscale mapping of other centriole proteins including CEP44, MDM1 and PPP1R35 (Atorino et al., 2020; Sydor et al., 2018; Van de Mark et al., 2015). To complement the U-ExM analysis, we have started imaging cells stained for CCDC15 and different centriole markers (i.e. distal appendage, proximal linker, centriole wall) using a recently purchased 3D-SIM superresolution microscope. We already included the SIM imaging data for CCDC15 localization in centrosome fractions purified from HEK293T cells in Fig. S5B. In the revised manuscript, we will replace confocal imaging data in Fig. 3A and 3B with SIM imaging data.
• As the reviewer noted, expansion microscopy has been successfully used for the analysis of a wide range of cellular structures and scientific questions including nanoscale mapping of cellular structures across different organisms. In particular, U-ExM of previously characterized centrosome proteins various centriole proteins have significantly advanced our understanding of centriole ultrastructure. In our manuscript, we used the U-ExM protocol that was validated for centrioles by comparative analysis of U-ExM and cryo-ET imaging by our co-authors (Gambarotto et al., 2019; Hamel et al., 2017). To clarify these points, we included the following sentence along with the relevant references in the introduction: “Application of the U-ExM method to investigate known centrosome proteins has started to define the composition of the inner scaffold as well as other centriolar sub-compartments (Chen et al., 2015; Gambarotto et al., 2021; Gambarotto et al., 2019; Kong and Loncarek, 2021; Laporte et al., 2022; Mahen, 2022; Mercey et al., 2022; Odabasi et al., 2023; Sahabandu et al., 2019; Schweizer et al., 2021; Steib et al., 2022; Tiryaki et al., 2022; Tsekitsidou et al., 2023).”

Minor points:

1) Text, figures, and referencing are clear and accurate, apart from minor exceptions.

We clarified and corrected the points regarding text, figures and references as suggested by the two reviewers.

__ 2) The title suggests a regulator role for CCDC15 in centriole integrity and ciliogenesis, which has formally not been shown. __

We revised the title as “CCDC15 localizes to the centriole inner scaffold and functions in centriole length control and integrity”.

__3) As the authors observe changes in centriole lengths in the absence of CCDC15, it would be very insightful to compare these phenotypes to other components that affect centriolar length, such as C2CD3, human Augmin complex components (as HAUS6 is identified in Fig. 1) or others. These could be interesting aspects for discussion, additional experiments are OPTIONAL. __

We agree with the reviewer that comparative analysis of centriole length phenotypes for CCDC15 and other components that regulate centriole length will provide insight into how these components work together at the centriole inner core. To this end, we phenotypically compared CCDC15 loss-of-function phenotypes to that of other components of the inner scaffold (POC5, POC1B, FAM161A) that interact with CCDC15. In agreement with their previously reported functions in U2OS or RPE1 cells, we found that POC5 depletion resulted in a 4% slight but significant increase in centriole length and POC1B depletion resulted in a 15% significant decrease. In contrast, FAM161A depletion did not alter centriole length (siControl: 447.8±59.7 nm, siFAM161A 436.3±64 nm). Together, our analysis of their centriolar localization dependency and regulatory roles during centriole length suggest that CCDC15 and POC1B might form a functional complex as positive regulators of centriole length. In contrast, POC5 functions as a negative regulator and might be part of a different pathway for centriole length regulation. We integrated the following sub-paragraph in the results section and also included discussion of this data in the discussion section:

“Moreover, we quantified centriole length in control cells and cells depleted for POC5 or POC1B. While POC5 depletion resulted in longer centrioles, POC1B resulted in shorter centrioles (POC5: siControl: 414.1 nm±38.3, siPOC5: 432.7±44.8 nm, POC1B: siControl: 400.6±36.1 nm, siPOC1B: 341.5±44.39 nm,). FAMA161A depletion did not alter centriole length (siControl: 447.8±59.7 nm, siFAM161A 436.3±64 nm). Together, these results suggest that CCDC15 might cooperate with POC1B and compete with POC5 to establish and maintain proper centriole length.”

__ 4) While the reduced ciliation rate in the absence of CCDC15 is convincing, the authors did not investigate "ciliogenesis", i.e. the formation of cilia, and hence should re-phrase. The sentence in the discussion that "CCDC15 functions during assembly" should be removed. __

To clarify that we only investigated the role of CCDC15 in the ability of cells to form cilia, we replaced sentences that indicates “CCDC15 functions in cilium assembly” with “CCDC15 is required for the efficiency of cilia formation”.

__5) The existence of stably associated CCDC15 pools with centrosomes (Fig. 2) requires further evidence. The recovery of fluorescence after photobleaching in FRAP experiments is strongly dependent on experimental setups and is only semi-quantitative. A full recovery is unrealistic, hence, it is ideally compared to a known static or known mobile component. I personally think this experiment -as it is presented now- is of little value to the overall fantastic study. The authors may consider omitting this piece of data. __

We agree with the reviewer that FRAP data by itself does not prove the existence of stably associated CCDC15 pool. As controls in these experiments, we use FRAP analysis of GFP-CCDC66, which has a 100% immobile pool at the cilia and 50% immobile pool at the centrosomes as assessed by FRAP (Conkar et al., 2019). To address these points, we toned down the conclusions derived from this experiment by revising the sentence as follows:

Additionally, we note that the following data provides support for the stable association of CCDC15 at the centrioles:

• About 49.6% (± 3.96) of the centrioles still had CCDC15 fluorescence signal at one of the centrioles upon CCDC15 siRNA treatment (Fig. 5A, 5B). The inefficient depletion of the mature centriole pool of CCDC15 is analogous to what was observed upon depletion of other centriole lumen and inner scaffold proteins including WDR90 and HAUS6 (Schweizer et al., 2021; Steib et al., 2020). __6) The data that CCDC15 is a cell cycle-regulated protein is not very convincing (see Fig. 3H), as the signals area weak and the experiment has been performed only once (n= 1). This piece of data does not appear to be very critical for the main conclusions of the manuscript and may be omitted. Otherwise, this experiment should be repeated to allow for proper statistical analysis. __

We will perform these experiments two more times, quantify cellular abundance of CCDC15 in synchronized populations from three experimental replicates and plot it with proper statistical analysis.

__7) Experimental details on how "defective centrioles" are determined are missing. __

We included the following experimental details to the methods section:

“Centrioles were considered as defective when the roundness of the centriole was lost or the microtubule walls were broken or incomplete. In the longitudinal views of centrioles, defective centrioles were visualized as heterogenous acetylated signal along the centriole wall or irregularities in the cylindrical organization of the centriole wall (Fig. 5F). We clarified these points in the methods section.

__ 8) For figures, in which the focus should be on growing centrioles (see Fig. 4), it could be helpful to guide the reader and indicate the respective areas of the micrographs by arrows. __

We added arrows to point to the respective areas of the micrographs in Fig. 4F.

__ 9) Page18: "centriole length shortening" could be changed to "centriole shortening". __

We corrected this description as suggested.

__10) It is unclear how the authors determine distal from proximal ends of centrioles in presented micrographs (see Fig. 5D). __

We determined the proximal and distal ends of the centrioles by taking the centriole pairs as a proxy. Even though we only represent a micrograph containing a single centriole in some of the U-ExM figures including Fig. 5D, the uncropped micrographs contain two centrioles, which are oriented orthogonally and tethered to each other at their proximal ends in interphase cells. We added the following sentence to the methods section to clarify this point:

*“Since centrioles are oriented orthogonally and tethered to each other at their proximal ends in interphase cells, we also used the orientation of the centriole pairs as a proxy to determine the proximal and distal ends of the centrioles.” *

__11) Fig. 7A is missing scale bars and Fig.7 overall is lacking rectangle indicators of the areas that are shown at higher magnification in the insets. __

We added scale bar to Fig. 7A and rectangle indicators for zoomed in regions in Fig. A, E, G.

12) Fig. 7C displays cilia that appear very short, especially when comparing to the micrographs and bar graphs presented. The authors may want to explain this discrepancy.

We thank the reviewer for the comment. The zoomed in representative cilia is 4.1 µM in control cells and 1.4 µM in CCDC15-depleted cells. Therefore, the representative cilia is in agreement with the quantification of cilia in Fig. 7C.

Reviewer #1 (Significance (Required)): • From a technical point of view the authors use two state-of-the-art technologies, namely proximity labeling combined with proteomics and ultrastructure expansion microscopy, that are both challenging and very well suited to address the main questions of this study. ____ • General assessment: The presented study is of highest experimental quality. Despite being very challenging, the expansion microscopy and proximity proteomics experiments have been designed and performed very well to allow solid interpretation. The results of the central data are consistent and allow strong first conclusions about the putative function of the newly identified centriolar protein CCDC15. The study presents a solid foundation for future hypothesis-driven, mechanistic analysis of CCDC15 and inner scaffold proteins in centriole length control and maintaining centriole integrity. The only limitation of the study is that the technically simpler experiments should be repeated to allow proper statistical assessment, which can be addressed easily. • Advance: This is the first study that identifies CCDC15 as a centriolar protein and localizes it to the inner scaffold. It further describes a function for CCDC15 in centriole length control and shows its importance in maintaining centriole integrity with consequences for stable cilia formation in tissue culture. The study provides further functional insights into the interdependence of inner scaffold proteins and the role of CCDC15 in the recruitment of the SFI1/centrin distal complex. • Audience: The manuscript will be of broad interest to the fields of centrosome and cell biology, both from a basic research and genetics/clinical point of view due to the association with human disorders. The state-of-the-art technologies applied will be of interest to a broader cell and molecular biology readership that studies subcellular compartments and microtubules. • Reviewer's field of expertise: Genetics, imaging, and protein-protein interaction studies with a focus on centrosomes and cilia.

We thank the reviewer for recognizing the importance of our work and for supportive and insightful comments that will further strengthen the conclusions of our manuscript. Our planned revisions will address the only major technical limitation raised by the reviewer that requires adding one more experimental replicate for analysis of the data detailed in major point#1. Notably, we also thank the reviewer to specifying the experiments that are not essential or will be out of the scope of our manuscript as “optional”.

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

Summary:

__In this study, Arslanhan et al. propose CCDC15 as a novel component of the centriole inner scaffold structure with potential roles in centriole length control, stability and the primary cilium formation in cultured epithelial cells. Using proximity labelling they explore the common interactors of Poc5 and Centrin-2, two resident molecules of the centriole inner scaffold, to hunt for novel regulators of this structure. The authors leverage expansion microscopy-based localization and siRNA-dependent loss-of-function experiments to follow up on one such protein they identify, CCDC15, with the aforementioned roles in centriole and cilia biology.

This study is designed and laid out nicely; however, to be able to support some of the important claims regarding their proximity labelling results and exploration on the roles of CCDC15, there are several major technical and reproducibility concerns that deem major revision. Similarly, the introduction (perhaps inadvertently) omits much of the recent studies on centriole size control that have highlighted the complexity of this biological problem. As such, addressing the following major points will be essential in further considering this work for publication. __

__We thank the reviewer for recognizing the importance of our work and appreciate the positive reflections on our manuscript and the feedback comments that were well thought-out and articulated and will further strengthen the conclusions of our manuscript. Our planned revisions focus on addressing the reviewer’s comments especially in further supporting our conclusions for proximity-labeling, phenotypic characterization and immunoprecipitation experiments, examining CCDC15 centriole localization in an additional cell line and investigating how CCDC15 works together during centriole length control with known components of the inner scaffold. __

Major points:

__1a) The authors use Poc5 and Centrin-2 molecules as joint baits to reveal the interactome of the centriole inner scaffold, however the work lacks appropriate experimental and analytical controls to argue that this is a proximity mapping "at the centriole inner scaffold". In its current state, it is simply an interactome of total Poc5 and Centrin-2, and it might be misleading to call it an interactome at the centriole inner scaffold (the statistical identification of shared interactors cannot do full justice to their biology at the centrosome). Appropriate expression data needed to delineate how large the centrosomal vs. cytoplasmic (or nucleoplasmic) fraction is for either of these molecules, both without and upon the addition of biotin (to see whether the bulk of interaction data stem from the cytoplasm/nucleoplasm or the centrioles themselves). The authors can test this by selectively blotting a lysate fraction containing the centrosomes after centrifugation, and compare them with the simultaneous blot of the supernatant (which were readily used for the blots presented in Fig. 1B). This experiment also becomes very relevant for the case of Centrin-2, as it also heavily localizes to the nucleoplasm as the authors found out (see Fig. 1A and Fig. S1A). __

__ Additionally, an orthogonal approach should be taken to perform bio-image analysis on their biotin/streptavidin imaging data to demonstrate the exact ratios between the centrosomal vs. cytoplasmic/nucleoplasmic biotin activation with appropriate signal normalization between the biotin/streptavidin images. This is particularly important, as although the authors claim that these cells stably express the V5BirA*, it seems that there is partial clonality to the expression. Some cells in both the Poc5 and Centrin-2 fusion constructs appear to lack the V5/Streptavidin signals upon Biotin addition (such as the two cells in the centre right in Poc5, and again a cell in the centre right for Centrin-2 images). In its current form, Fig. 1A lacks signal quantification and does not report any information about the replicates and distributions of the data. I worry that this may raise concerns on the reproducibility if published in its current form. __a) We agree with the reviewer that the proximity maps of POC5 and

a) Centrin-2 are not specific to the centriole inner scaffold and thus, do not represent the inner scaffold interactome. The proximity maps identified interactions across different pools of POC5 and Centrin-2 in nucleus, cytoplasm and centrosomes (Fig. 1, S1). To highlight these important points, we already included extensive analysis of the different cellular compartments and biological processes identified by the POC5 and Centrin-2 proximity maps in the results section (pg. 9-10).

We think that there are two reasons that caused the misinterpretation of the use of these proximity maps as the “inner scaffold interactome”: 1) the way we introduced the motivation for proximity mapping studies, 2) proposing the use of the resulting interactomes as resources for identification of the full repertoire of the inner scaffold proteins. To clarify these points, we revised the manuscript in all relevant parts that might have led to misinterpretation. Following are the specific revisions:

• To clarify that the proximity maps are not specific to the inner scaffold pools of POC5 and Centrin-2, we revised the title of the results section for Fig. 1 and 2 as follows: “Proximity mapping of POC5 and Centrin-2 identifies new centriolar proteins”.

• To indicate that POC5 and Centrin-2 localizes to the cytoplasm and/or nucleus in addition to the centrosome, we added the following sentence to the result section: “In addition to centrosomes, both fusion proteins also localized to and induced biotinylation diffusely in the cytoplasm and/or nucleus (Fig. 1A).”

• In the introduction, we revised the following sentence “Here, we used the known inner scaffold proteins as probes to identify the molecular makeup of the inner scaffold in an unbiased way.” as follows: *“Here, we used the known inner scaffold proteins as probes to identify new components of the inner scaffold”. *

• To highlight the different cellular pools of POC5 and Centrin-2 and identification of their interactors in these pools, we included the following sentence in the results section: “As shown in Fig. S1, Centrin-2 and POC5 proximity interactomes were enriched for GO categories that are relevant for their published functions during centrosomal, cytoplasmic and/or nuclear biological processes and related cellular compartments (Azimzadeh et al., 2009; Dantas et al., 2013; Heydeck et al., 2020; Khouj et al., 2019; Resendes et al., 2008; Salisbury et al., 2002; Steib et al., 2020; Yang et al., 2010; Ying et al., 2019).”

• We replaced the “interactome” statement with “proximity interaction maps” or “proximity interactors” throughout the manuscript to prevent the conclusion that the proximity maps represent the inner scaffold interactome. b) As the reviewer noted, most centrosome proteins have multiple different cellular pools including the centrosome. For most proteins like gamma-tubulin and centrin, their cytoplasmic/nucleoplasmic pools are more abundant than their centrosomal pools (Moudjou et al., 1996; Paoletti et al., 1996). For the Firat-Karalar et al. Current Biology 2015 paper, I compared the biotinylation levels of centrosomal fractions versus cytoplasmic fractions and confirmed that this is also true in cells expressing myc-BirA* fusions of CDK5RAP2, CEP192, CEP152 and CEP63 (unpublished) (Firat-Karalar et al., 2014). For the revised manuscript, we will compare the biotinylation level of centrosomal, nuclear and cytoplasmic pools of V5Bir*-POC5 and V5BirA*-Centrin-2 using the stable lines. To this end, we will use published centrosome purification protocols. We will include this data in Fig. S1 to highlight that the proximity interactomes represent the different pools of the bait proteins and to show the relative levels of the baits across their different pools.

c) BioID approach has been successfully used to probe centrosome interactions by my lab and other labs in the field. In fact, proximity interaction maps of over 50 centrosome proteins were published as resource papers by Pelletier&Gingras labs (Gheiratmand et al., 2019; Gupta et al., 2015). Analogous to our strategy in this manuscript, these studies generated proximity maps of centrosome proteins by creating cell lines that stably express BioID-fusions of centrosome proteins followed by streptavidin pulldowns from whole cell extracts and mass spectrometry analysis. Since majority of centrosome proteins also have pools in multiple cellular locations, the published BioID proximity maps for centrosome proteins are not specific to centrosomes. However, the proximity maps included all known centrosome proteins and identified new proteins, which shows that centrosome interactions are represented in pulldowns form whole cell lysates. Moreover, maps form whole cell lysates are also advantageous as they are are unbiased and can be used in future studies as resources for studying the functions and interactions of the bait proteins in different contexts.

In the Firat-Karalar et al. Current Biology 2015 paper, I combined centrosome purifications with BioID pulldowns to enrich for the centrosomal interactions in the proximity maps of centriole duplication proteins(Firat-Karalar et al., 2014). However, I started the purification with cells transiently transfected with the BioID-fusion constructs, which resulted in high ectopic expression of the fusions in the cytoplasm and/or nucleus. Therefore, centrosome enrichments were useful as an additional step before mass spectrometry. Comparative analysis of the data for proximity maps of 4 centrosome proteins generated from stable lines or centrosome fractions of transiently transfected cells substantially overlap as compared in the Gupta et al. Cell 2015 study and were more comprehensive (Table S2) (Gupta et al., 2015). Therefore, we are confident that the proximity interactomes we generated for POC5 and Centrin-2 include their centrosomal interactions.

__1b) Similarly, it is not clear whether the expression of Poc5 and Centrin-2 fusion molecules somehow interfere with their endogenous interactions or function. At least some loss-of-function (e.g., RNAi) experiments should be performed where the depletion of endogenous proteins should be attempted to rescue by the fusion constructs. This will help evaluate whether the fusion proteins can rescue the depletion of their endogenous counterparts and behave as expected from a wild-type scenario. __

The reviewer raises an important concern regarding the physiological relevance of the POC5 and Centrin-2 proximity maps. In the manuscript, we showed and discussed the validation of their proximity interactomes by two lines of evidence, which are: 1) the interactomes identified the previously described cellular compartments, biological processes or interactors of POC5 and Centrin-2, 2) the interactomes led to the identification of CCDC15 as a new inner scaffold protein.

As the reviewer indicated, stable expression of POC5 and Centrin-2 in the presence of their endogenous pools might affect cellular physiology and thereby the landscape of the interactomes. We plan to address this using the following experiments:

a) We will perform a set of functional assays to assess whether stable V5BirA*-Centrin-2 and V5BirA*-POC5 cells behaves like control cells in terms of their centrosome number, cell cycle profiles and mitotic progression. We will specifically quantify:

• centrosome number (immunofluorescence analysis for gamma-tubulin and centrin)
• their mitotic index (immunofluorescence analysis by DAPI)
• spindle polarity and percentage of multinucleation (immunofluoerescence analysis for microtubules, gamma-tubulin and DAPI)
• cell cycle profiles (flow cytometry and immunofluorescence)
• apoptosis (immunoblotting for caspase 3) Together, results from these experiments indicate that the V5BirA*-POC5 or Centrin-2-expressing stable lines do not exhibit defects associated with their stable expression.

b) We will perform expansion microscopy in V5BirA*-Centrin-2 and V5BirA*-POC5 cells to assess whether the fusion protein specifically localizes to the centriole inner scaffold, which will provide support for the presence of inner scaffold proteins in their proximity maps. Specifically, we plan to stain the fusion proteins by V5 or BirA antibodies and include the data for the antibody that works for expansion microscopy. This experiment will address whether their stable expression results in specific localization of these proteins at the centriole inner scaffold.

1c) Overall, as the entire claim around the proximity mapping revolve around its assumption about the centriole inner scaffold, these controls seem imperative to substantiate the ground truth of the biology presented in the manuscript.

In the revised manuscript, we toned down and made it clear that Centrin-2 and POC5 proximity maps are not specific to the inner scaffold and do not represent the inner scaffold interactome. Since the maps were generated from the whole cell extract, they will provide a resource for future studies aimed at studying functions and mechanisms of POC5 and Centrin-2 across their different cellular pools including the centrosome.

We would like to also highlight that the proximity maps of POC5 and Centrin-2 are not the major advances of our manuscript. The major advance of our manuscript is the identification of CCDC15 as a new inner scaffold protein that is required for regulation of centriole size and architectural integrity and thereby, for maintaining the ability of centrioles to template the assembly of functional cilia. Importantly, our results identified CCDC15 as the first dual regulator of centriolar recruitment of inner scaffold protein POC1B and the distal end SFI1/Centrin complex and provided important insight into how inner scaffold proteins work together during centriole integrity and size regulation. The new set of experiments we will perform for the revisions of the paper will strengthen these conclusions.

__2) I am curious about the choices of the cell lines in this work. The proximity mapping to reveal CCDC15 as a candidate protein for centriole inner scaffold was performed in HEK293T cells (human embryonic kidney), however its immunostaining was performed using RPE1 and U2OS cells (human retinal and osteosarcoma epithelial cells respectively). This raises questions regarding the generality of CCDC15 as a centriole inner scaffold protein. Could CCDC15 be simply unique to the centriole inner scaffold of epithelial cells such as RPE1 and U2OS cells? Or could the authors demonstrate any information/data on whether it's similarly localized to the inner scaffold in embryonic kidney cells or other cell types? If not, the claims should be moderated to reflect this fine detail. __

To test whether CCDC15 localizes to the inner scaffold in other cell types, we performed U-ExM analysis of CCDC15 localization relative to the centriolar microtubules in differentiating multiciliated epithelial cultures (MTEC). As shown in Fig. S3A, CCDC15 localized to the inner scaffold in the centrioles in MTEC ALI+4 cells. Given that the inner scaffold proteins including CCDC15 and previously characterized ones have not been studied in multiciliated epithelia, this result is important and provides support for potential role of the inner scaffold in ensuring centriole integrity during ciliary beating. Additionally, we examined CCDC15 localization by 3D-SIM in centrosomes purified from HEK293T cells, which showed that CCDC15 localizes between the distal centriole markers CEP164 and Centrin-3 and proximal centriole markers gamma-tubulin and rootletin (Fig. S3B).

3) Discussions and data on the localization of CCDC15 to centriolar satellites appear anecdotal and not fully convincing (Fig. S2D). Given that the authors test the relevance of PCM1 for CCDC15's centriolar localization, it is key to have quantitative data supporting their claim that centriolar satellites can help recruit CCDC15 to the centriole. Could the authors quantify what proportion of CCDC15 localize to the centriolar satellites? One way to do this could be to quantify the colocalization coefficience of CCDC15 and PCM1 signals.

We only observed co-localization of CCDC15 with the centriolar satellite marker PCM1 in cells transiently transfected with mNG-CCDC15. In Fig. S2E, we included the quantification of the percentage of U2OS and RPE1 cells that exhibit co-localization of PCM1 (100% of U2OS cells, about 80% of RPE1 cells). Like CCDC15, ectopic expression of WDR90 revealed its centriolar satellite localization, suggesting a potential link between centriolar satellites and inner scaffold proteins that can be investigated in future studies (Steib et al., 2020). We now included these results in the discussion section as follows:

“As assessed by co-localization with the centriolar satellite marker PCM1, mNG-CCDC15 localized to centriolar satellites in all U2OS cells and in about 80% of RPE1 cells (Fig. S2C-E). Association of CCDC15 with centriolar satellites is further supported by its identification in the centriolar satellite proteomes(Gheiratmand et al., 2019; Quarantotti et al., 2019).”

Even though endogenous staining for CCDC15 did not reveal its localization to centriolar satellites, following lines of data support the presence of a dynamic and low abundance pool of CCDC15 at the centriolar satellites: 1) CCDC15 was identified in the centriolar satellite proteome and interactome (Gheiratmand et al., 2019; Quarantotti et al., 2019). 2) CCDC15 centrosomal targeting is in part regulated by PCM1 (Fig. S2F, S2G). For majority of the proteins identified in the centriolar satellite proteome, their satellite pool can only be observed upon ectopic expression. This might be because their centriolar satellite pool is of low abundance and transient as satellite interactions are extensively identified only in proximity mapping studies, but not in traditional pulldowns

__4) Similar to above (#3), there is no quantitative information on the co-localization or partial co-localization of the signal foci in Fig. 3A and 3B. The authors readily study CCDC15's localization in wonderful detail in their expansion microscopy data, so they could actually consider taking out Fig. 3A and 3B, as the data seem redundant without any quantification. __

To address the reviewer’s concern, we included plot intensity profile analysis of CCDC15 and different centriole markers along a line drawn at the centrioles in Fig. 3A and 3B, which shows the extent of their overlap. As part of our revision plan, we will replace the confocal imaging data in Fig. 3A and 3B with 3D-SIM imaging data of CCDC15 relative to different centriole markers together with plot profile analysis. We already included 3D-SIM imaging of centrosomes purified form HEK293T cells in Fig. S3B. 3D-SIM imaging data will complement the localization data revealed by U-ExM.

__5) Do the authors also feel that CCDC15 localize to the core lumen in a somehow helical manner (Fig. 1A, Fig. 1F top and bottom panels, Fig. 5A etc.)? Le Guennec et al. 2020's helical lattice proposal for the inner scaffold further reaffirms that CCDC15 is indeed a likely major component of the inner scaffold. In my view, authors should state this physical similarity explicitly to further support their findings on CCDC15. __

As the reviewer indicated, cryo–electron tomography and subtomogram averaging of centrioles from four evolutionarily distant species showed that centriolar microtubules are bound together by a helical inner scaffold covering ~70% of the centriole length (Le Guennec et al., 2020). Although U-ExM data do not have enough resolution to show that CCDC15 localizes in a helical manner, we agree with the reviewer that the discussion of this possibility is important and thus we included the following sentence in the results:

“Longitudinal views suggest potential helical organization of CCDC15 at the inner scaffold, which is consistent with its reported periodic, helical structure (Le Guennec et al., 2020).”

__6a) The data on the link between the CCDC15 recruitment and the centriole growth (Fig. 4F) or the G2 phase of the cell cycle (Fig. 4H) are not fully convincing without quantitative data. For Fig. 4F, the authors should consider plotting the daughter centriole length vs the daughter CCDC15 intensities against each another, to see whether more elongated daughters truly tend to have more CCDC15. __

To address the reviewer’s concern, we will plot the daughter centriole length versus CCDC15 intensity at different stages of centriole duplication. In asynchronous cultures that we analyzed with U-ExM, we were not able to find enough cells to perform such quantification. To overcome this limitation, we will perform U-ExM analysis of cells fixed at different points after mitotic shake-off and stained for CCDC15 and tubulin. We will include minimum 10 different representative U-ExM data for different stages of centriole duplication in the revised manuscript along with quantification of length versus signal.

As detailed in the results section, the goal of these experiments was to determine when CCDC15 is recruited to the procentrioles during centriole duplication, but not to suggest a role for CCDC15 in centriole growth. We clarified this by including the following sentence:

“To investigate the timing of CCDC15 centriolar recruitment during centriole biogenesis, we examined CCDC15 localization relative to the length of procentrioles that represent cells at different stages of centriole duplication (Fig. 4F).”

__6b) For Fig. 4H, the argument regarding the cell cycle regulation requires quantification of the bands from several WB repeats, normalized to the expression of GAPDH within each blot (this is particularly relevant, as the bands of CCDC15 do not look dramatically different enough to draw conclusions by eye). __

We will perform these experiments two more times, quantify cellular abundance of CCDC15 in synchronized populations from three experimental replicates and plot it with proper statistical analysis.

__7a) The authors find herein that CCDC15 depletion lead to centrioles that are ~10% shorter than the controls. With the depletion of Poc5 and Wdr90 (other proposed components of the inner scaffold), the centrioles end up larger however (Steib et al., 2020). If the role of inner scaffold in promoting centriole elongation is structural, why are these two results the opposite of each other? I realize there is a brief discussion about this at the end of the paper, however, this requires a detailed discussion and speculation on the relevance of these findings. It would be key to clarify whether the inner scaffold as a structure inhibits or promotes centriole growth - or somehow both? If so, how? __

We agree with the reviewer that comparative analysis of centriole length phenotypes for CCDC15 and other components that regulate centriole length will provide insight into how these components work together at the centriole inner core. To this end, we phenotypically compared CCDC15 loss-of-function phenotypes to that of other components of the inner scaffold (POC5, POC1B, FAM161A) that interact with CCDC15. In agreement with their previously reported functions in U2OS or RPE1 cells, we found that POC5 depletion resulted in a 4% slight but significant increase in centriole length and POC1B depletion resulted in a 15% significant decrease. In contrast, FAM161A depletion did not alter centriole length (siControl: 447.8±59.7 nm, siFAM161A 436.3±64 nm). Together, our analysis of their centriolar localization dependency and regulatory roles during centriole length suggest that CCDC15 and POC1B might form a functional complex as positive regulators of centriole length. In contrast, POC5 functions as a negative regulator and might be part of a different pathway for centriole length regulation. We integrated the following sub-paragraph in the results section in pg. 19 and also included discussion of this data in the discussion section in pg. 23:

“Moreover, we quantified centriole length in control cells and cells depleted for POC5 or POC1B. While POC5 depletion resulted in longer centrioles, POC1B resulted in shorter centrioles (POC5: siControl: 414.1 nm±38.3, siPOC5: 432.7±44.8 nm, POC1B: siControl: 400.6±36.1 nm, siPOC1B: 341.5±44.39 nm,). FAMA161A depletion did not alter centriole length (siControl: 447.8±59.7 nm, siFAM161A 436.3±64 nm). Together, these results suggest that CCDC15 might cooperate with POC1B and compete with POC5 to establish and maintain proper centriole length.”

__7b) There might be some intriguing opposing regulatory action of Poc5 and CCDC15 as demonstrated here, where CCDC15 depletion leads to slightly over-recruitment of Poc5, and vice versa. Does this suggest that a tug-of-war going on between different molecules that localize to the inner scaffold? Does this provide some dynamicity to this structure, which might in turn regulate centriole length both positively and negatively? This may be analogous to how opposing forces of dyneins and kinesins provide robust length control for mitotic spindles. I am speculating here, but hopefully these may provide some useful grounds for further discussion in the paper. If the authors deem it interesting experimentally, they can test whether the two molecules indeed regulate centriole length by opposing each other's action, by a double siRNA of CCDC15 and Poc5 to see if this retains the centriole length at its control siRNA size (like how they do a similar test for Poc1's potential co-operativity with CCDC15 in Fig. 6J). __

We thank the reviewer for proposing excellent ideas on how inner scaffold proteins work together to regulate centriole length. As proposed by the reviewer, different proteins oppose each other analogous to how dynein and kinesin regulate mitotic spindle length. Loss-of-function and localization dependency data support that CCDC15 cooperates with POC1B, which was supported by phenotypic characterization of co-depleted cells (Fig. 6I-K).

The increase in POC5 levels and coverage at the centrioles upon CCDC15 depletion and vice versa (Fig. 7B, 7G) suggest that CCDC15 and POC5 compete with each other in centriole length regulation. As suggested by the reviewer, we attempted to test this by comparing centriole length in cells co-depleted for CCDC15 and POC5 relative to their individual depletions. Although we tried different depletion workflows, we were not able to co-deplete CCDC15 and POC5. Specifically, we tried transfecting cells with CCDC15 and POC5 siRNAs at the same time or sequentially for 48 h or 96 h. The centrioles in cells that survived co-depletion were positive for both CCDC15 and POC5. This might be because co-depletion of both proteins is toxic to cells. Since CCDC15 and POC5 are likely part of two different pathway in regulation of centrioles and also have other cellular functions, this might have caused cell death. We included the following statement in the discussion to address the excellent model proposed by the reviewer:

“Taken together, our results suggest that CCDC15 cooperates with POC1B and competes with POC5 during centriole length regulation. Moreover, they also raise the exciting possibility that centriole length can be regulated by opposing activities of inner scaffold proteins. Future studies that explore the relationship among centriole core proteins are required to uncover the precise mechanisms by which they regulate centriole integrity and size.”

__8) In their introduction section, the authors discuss how relatively little is known about the size control of centrioles, however they fail to mention a series of recent primary literature that uncover striking, new mechanisms and novel molecular players that highlight the complexity of centriole size control. This complexity appears to arise from the existence of multitude of length control mechanisms that influence the cartwheel or the microtubule length individually, or simultaneously via yet-to-be further explored crosstalk mechanisms. a. As such, when the authors talk about the procentriole size control in the introduction, they should discuss and refer to the following studies, in terms of: • How theoretical and experimental work demonstrate that procentriole length may vary dependent on the levels of its building block Sas-6 in animals (Dias Louro et al., 2021 PMID: 33970906; Grzonka and Bazzi, 2022 bioRxiv). • How a homeostatic Polo-like kinase 4 clock regulates centriole size during the cell cycle (Aydogan et al., 2018 JCB PMID: 29500190), and how biochemistry and genetics coupled with mathematical modelling unravel a conserved negative feedback loop between Cep152 and Plk4 that constitutes the oscillations of this clock in flies (Boese et al., 2018 PMID: 30256714; Aydogan et al., 2020 PMID: 32531200) and human cells (Takao et al., 2019 PMID: 31533936). __

__b. Similarly, when the authors refer to centriole size control induced by microtubule-related proteins, they should highlight the further complexity of this process by referring to: • How a molecule located at the microtubule wall, Cep295/Ana1, can regulate centriole length in flies (Saurya et al., 2016 PMID:27206860) and human cells (Chang et al., 2016 PMID:27185865) - like all the other centriolar MT molecules that the authors discuss in the manuscript. • How a crosstalk between Cep97 and Cep152 influences centriole growth in fly spermatids (Galletta et al., 2016 PMID:27185836). • How a crosstalk between CP110-Cep97 and Plk4 influences centriole growth in flies (Aydogan et al., 2022 PMID:35707992), and this molecular crosstalk is conserved, at least biochemically, in human cells (Lee et al., 2017 PMID:28562169). __

We thank the reviewer for highlighting the papers that uncovered new mechanisms and players of centriole size and integrity control as well as for the detailed explanation of how different studies led to these discoveries in different organisms. We should have discussed these proteins, functional complexes and mechanisms in our manuscript and cited the relevant literature. We inadvertently focused on literature that uncovered centriole length regulation by MAPs and the inner scaffold. In the introduction section of the revised manuscript where we introduced centriole size regulation in pg. 5, we summarized the major findings on the role of different MAPs, cartwheel and PLK4 homeostatic clock in ensuring formation of centrioles at the correct size in different organisms.

__Minor points: __

__1) Introduction section: Literature reference missing for the sentence starting with "Importantly, the stable nature of centrioles enables them to withstand...". __

We cited research articles that show the importance of centriole motility during ciliary motility and cell division.

“Importantly, the stable nature of centrioles enables them to withstand mechanical forces during cell division and upon ciliary and flagellar motility (Abal et al., 2005; Bayless et al., 2012; Meehl et al., 2016; Pearson et al., 2009).

__2) Fig. S1 legend: A typo as follows: CRAPome banalysis should read CRAPome analysis. __

We corrected this typo.

__3) Fig. S2: Info on the scale bar in the legend is missing in Fig. S2A. Scale bars for different panels are missing in general in Fig. S2A. __

We added scale bar information for Fig. S2A and to all other supplementary figure legends that lack scale bar information.

__4) Fig. 3A and 3B: When displaying the data, coloured cartoon diagrams would be beneficial to guide the reader who are not fully familiar with the spatial orientation of these proteins. __

As suggested by the reviewer, we will remove the confocal imaging data for CCDC15 localization from Fig. 3A and 3B. For the revised version, we will include 3D-SIM imaging data along with a diagram that represents the spatial orientation of CCDC15 relative to the chosen centriole markers.

__5) Fig. 3H: No information about the sample number (number of cells or technical repeats examined) reported. __

We included information on the number of experimental replicates and cells analyzed.

__6) Fig. S3B legend: A typo as follows: CCD15-depelted RPE1 cells should read CCDC15-depleted RPE1 cells. __

We corrected this typo.

__7) Fig. S3B legend: A typo as follows: cellswere fixed with should read cells were fixed with. __

We corrected this typo.

__8) There are many spelling mistakes and typos throughout the paper. I have listed a few examples above, but please carefully read through the manuscript to correct all the errors. __

Thank you for indicating the spelling mistakes we missed to correct for initial submission. In the revised manuscript, we carefully read through the manuscript to correct the mistakes.

__9) Fig. S3E: The orange columns depicting % of cells with Sas-6 dots look awkward. Why the columns look larger than the mean line? Please correct as appropriate. __

The total percentage of cells in the two categories (orange and purple) we counted is 100%, which corresponds to the column value at the y-axis. Therefore, the value for each experimental replicate for the orange category is less than 100% and is marked below the 100% line.

__10) Although authors provide microscopy information for the U-ExM and FRAP experiments, there is no information about the microscopy on regular confocal imaging experiments which should be detailed in Materials and Methods. Also, there is no information about the lenses, laser lines and the filter sets that were used in the imaging experiments. These should be provided as well. __

In the methods section, we now included detailed information for the microscopes we used and imaging setup (lenses, laser lines, filter sets, detectors, z-stack size, resolution).

11)

• __ Fig. 2A: lacks a scale bar. __
• __ Fig. 2C legend: lacks info on the scale bar length. __
• __ Fig. 5A legend: lacks info on the scale bar length. __
• __ Fig. 7A: lacks a scale bar. __
• __ Fig. 7G legend: lacks info on the scale bar length. __
• __ Fig. S2C-E: lack scale bars. __
• __ Fig. S3D, F and H: lack scale bars. (Fig. S4 in the revised manuscript)__
• __ Fig. S3J legend: lacks info on the scale bar length. (Fig. S4 in the revised manuscript)__
• __ Fig. S4A, B, D and E: lack scale bars. (Fig. S5 in the revised manuscript)__
• __ Fig. S4C legend: lacks info on the scale bar length. (Fig. S5 in the revised manuscript)__
• __ Fig. S4G legend: lacks info on the scale bar length. (Fig. S5 in the revised manuscript)__ We added the scale bars and the size information to the figures and figure legends for the above figures.

Reviewer #2 (Significance (Required)): __The findings of this study join among the relatively new literature (e.g., Steib et al., 2020 and Le Guennec et al. 2020) on the nature of centriole inner scaffold and its potential roles in centriole formation, integrity and its propensity to form the primary cilium. Therefore, it will be of interest to a group of scientists studying these topics in the field of centrosomes/cilia.

My expertise is on the biochemistry and genetics of centriole formation in animals.__

We thank the reviewer for his/her comments and constructive feedback to improve our manuscript. We are encouraged to see that the reviewer acknowledges how the results from our manuscript advances our understanding of centriole length, integrity and function regulation.

References

Abal, M., G. Keryer, and M. Bornens. 2005. Centrioles resist forces applied on centrosomes during G2/M transition. Biol Cell. 97:425-434.

Atorino, E.S., S. Hata, C. Funaya, A. Neuner, and E. Schiebel. 2020. CEP44 ensures the formation of bona fide centriole wall, a requirement for the centriole-to-centrosome conversion. Nat Commun. 11:903.

Azimzadeh, J., P. Hergert, A. Delouvee, U. Euteneuer, E. Formstecher, A. Khodjakov, and M. Bornens. 2009. hPOC5 is a centrin-binding protein required for assembly of full-length centrioles. J Cell Biol. 185:101-114.

Bayless, B.A., T.H. Giddings, Jr., M. Winey, and C.G. Pearson. 2012. Bld10/Cep135 stabilizes basal bodies to resist cilia-generated forces. Mol Biol Cell. 23:4820-4832.

Chen, F., P.W. Tillberg, and E.S. Boyden. 2015. Optical imaging. Expansion microscopy. Science. 347:543-548.

Conkar, D., H. Bayraktar, and E.N. Firat-Karalar. 2019. Centrosomal and ciliary targeting of CCDC66 requires cooperative action of centriolar satellites, microtubules and molecular motors. Sci Rep. 9:14250.

Dantas, T.J., O.M. Daly, P.C. Conroy, M. Tomas, Y. Wang, P. Lalor, P. Dockery, E. Ferrando-May, and C.G. Morrison. 2013. Calcium-binding capacity of centrin2 is required for linear POC5 assembly but not for nucleotide excision repair. PLoS One. 8:e68487.

Firat-Karalar, E.N., N. Rauniyar, J.R. Yates, 3rd, and T. Stearns. 2014. Proximity interactions among centrosome components identify regulators of centriole duplication. Curr Biol. 24:664-670.

Gambarotto, D., V. Hamel, and P. Guichard. 2021. Ultrastructure expansion microscopy (U-ExM). Methods Cell Biol. 161:57-81.

Gambarotto, D., F.U. Zwettler, M. Le Guennec, M. Schmidt-Cernohorska, D. Fortun, S. Borgers, J. Heine, J.G. Schloetel, M. Reuss, M. Unser, E.S. Boyden, M. Sauer, V. Hamel, and P. Guichard. 2019. Imaging cellular ultrastructures using expansion microscopy (U-ExM). Nat Methods. 16:71-74.

Gheiratmand, L., E. Coyaud, G.D. Gupta, E.M. Laurent, M. Hasegan, S.L. Prosser, J. Goncalves, B. Raught, and L. Pelletier. 2019. Spatial and proteomic profiling reveals centrosome-independent features of centriolar satellites. EMBO J.

Gupta, G.D., E. Coyaud, J. Goncalves, B.A. Mojarad, Y. Liu, Q. Wu, L. Gheiratmand, D. Comartin, J.M. Tkach, S.W. Cheung, M. Bashkurov, M. Hasegan, J.D. Knight, Z.Y. Lin, M. Schueler, F. Hildebrandt, J. Moffat, A.C. Gingras, B. Raught, and L. Pelletier. 2015. A Dynamic Protein Interaction Landscape of the Human Centrosome-Cilium Interface. Cell. 163:1484-1499.

Hamel, V., E. Steib, R. Hamelin, F. Armand, S. Borgers, I. Fluckiger, C. Busso, N. Olieric, C.O.S. Sorzano, M.O. Steinmetz, P. Guichard, and P. Gonczy. 2017. Identification of Chlamydomonas Central Core Centriolar Proteins Reveals a Role for Human WDR90 in Ciliogenesis. Curr Biol. 27:2486-2498 e2486.

Heydeck, W., B.A. Bayless, A.J. Stemm-Wolf, E.T. O'Toole, A.S. Fabritius, C. Ozzello, M. Nguyen, and M. Winey. 2020. Tetrahymena Poc5 is a transient basal body component that is important for basal body maturation. J Cell Sci. 133.

Khouj, E.M., S.L. Prosser, H. Tada, W.M. Chong, J.C. Liao, K. Sugasawa, and C.G. Morrison. 2019. Differential requirements for the EF-hand domains of human centrin 2 in primary ciliogenesis and nucleotide excision repair. J Cell Sci. 132.

Kong, D., and J. Loncarek. 2021. Analyzing Centrioles and Cilia by Expansion Microscopy. Methods Mol Biol. 2329:249-263.

Laporte, M.H., I.B. Bouhlel, E. Bertiaux, C.G. Morrison, A. Giroud, S. Borgers, J. Azimzadeh, M. Bornens, P. Guichard, A. Paoletti, and V. Hamel. 2022. Human SFI1 and Centrin form a complex critical for centriole architecture and ciliogenesis. EMBO J. 41:e112107.

Le Guennec, M., N. Klena, D. Gambarotto, M.H. Laporte, A.M. Tassin, H. van den Hoek, P.S. Erdmann, M. Schaffer, L. Kovacik, S. Borgers, K.N. Goldie, H. Stahlberg, M. Bornens, J. Azimzadeh, B.D. Engel, V. Hamel, and P. Guichard. 2020. A helical inner scaffold provides a structural basis for centriole cohesion. Sci Adv. 6:eaaz4137.

Mahen, R. 2022. cNap1 bridges centriole contact sites to maintain centrosome cohesion. PLoS Biol. 20:e3001854.

Meehl, J.B., B.A. Bayless, T.H. Giddings, Jr., C.G. Pearson, and M. Winey. 2016. Tetrahymena Poc1 ensures proper intertriplet microtubule linkages to maintain basal body integrity. Mol Biol Cell. 27:2394-2403.

Mercey, O., C. Kostic, E. Bertiaux, A. Giroud, Y. Sadian, D.C.A. Gaboriau, C.G. Morrison, N. Chang, Y. Arsenijevic, P. Guichard, and V. Hamel. 2022. The connecting cilium inner scaffold provides a structural foundation that protects against retinal degeneration. PLoS Biol. 20:e3001649.

Moudjou, M., N. Bordes, M. Paintrand, and M. Bornens. 1996. gamma-Tubulin in mammalian cells: the centrosomal and the cytosolic forms. J Cell Sci. 109 ( Pt 4):875-887.

Odabasi, E., D. Conkar, J. Deretic, U. Batman, K.M. Frikstad, S. Patzke, and E.N. Firat-Karalar. 2023. CCDC66 regulates primary cilium length and signaling via interactions with transition zone and axonemal proteins. J Cell Sci. 136.

Paoletti, A., M. Moudjou, M. Paintrand, J.L. Salisbury, and M. Bornens. 1996. Most of centrin in animal cells is not centrosome-associated and centrosomal centrin is confined to the distal lumen of centrioles. J Cell Sci. 109 ( Pt 13):3089-3102.

Pearson, C.G., D.P. Osborn, T.H. Giddings, Jr., P.L. Beales, and M. Winey. 2009. Basal body stability and ciliogenesis requires the conserved component Poc1. J Cell Biol. 187:905-920.

Quarantotti, V., J.X. Chen, J. Tischer, C. Gonzalez Tejedo, E.K. Papachristou, C.S. D'Santos, J.V. Kilmartin, M.L. Miller, and F. Gergely. 2019. Centriolar satellites are acentriolar assemblies of centrosomal proteins. EMBO J.

Resendes, K.K., B.A. Rasala, and D.J. Forbes. 2008. Centrin 2 localizes to the vertebrate nuclear pore and plays a role in mRNA and protein export. Mol Cell Biol. 28:1755-1769.

Sahabandu, N., D. Kong, V. Magidson, R. Nanjundappa, C. Sullenberger, M.R. Mahjoub, and J. Loncarek. 2019. Expansion microscopy for the analysis of centrioles and cilia. J Microsc. 276:145-159.

Salisbury, J.L., K.M. Suino, R. Busby, and M. Springett. 2002. Centrin-2 is required for centriole duplication in mammalian cells. Curr Biol. 12:1287-1292.

Schweizer, N., L. Haren, I. Dutto, R. Viais, C. Lacasa, A. Merdes, and J. Luders. 2021. Sub-centrosomal mapping identifies augmin-gammaTuRC as part of a centriole-stabilizing scaffold. Nat Commun. 12:6042.

Steib, E., M.H. Laporte, D. Gambarotto, N. O’lieric, C. Zheng, S. Borgers, V. Olieric, M.L. Guennec, F. Koll, A.M. Tassin, M.O. Steinnmetz, P. Guichard, and V. Hamel. 2020. WDR90 is a centriolar microtubule wall protein important for centriole architecture integrity. eLife.

Steib, E., R. Tetley, R.F. Laine, D.P. Norris, Y. Mao, and J. Vermot. 2022. TissUExM enables quantitative ultrastructural analysis in whole vertebrate embryos by expansion microscopy. Cell Rep Methods. 2:100311.

Sydor, A.M., E. Coyaud, C. Rovelli, E. Laurent, H. Liu, B. Raught, and V. Mennella. 2018. PPP1R35 is a novel centrosomal protein that regulates centriole length in concert with the microcephaly protein RTTN. Elife. 7.

Tiryaki, F., J. Deretic, and E.N. Firat-Karalar. 2022. ENKD1 is a centrosomal and ciliary microtubule-associated protein important for primary cilium content regulation. FEBS J. 289:3789-3812.

Tsekitsidou, E., C.J. Wong, I. Ulengin-Talkish, A.I.M. Barth, T. Stearns, A.C. Gingras, J.T. Wang, and M.S. Cyert. 2023. Calcineurin associates with centrosomes and regulates cilia length maintenance. J Cell Sci. 136.

Van de Mark, D., D. Kong, J. Loncarek, and T. Stearns. 2015. MDM1 is a microtubule-binding protein that negatively regulates centriole duplication. Mol Biol Cell. 26:3788-3802.

Yang, C.H., C. Kasbek, S. Majumder, A.M. Yusof, and H.A. Fisk. 2010. Mps1 phosphorylation sites regulate the function of centrin 2 in centriole assembly. Mol Biol Cell. 21:4361-4372.

Ying, G., J.M. Frederick, and W. Baehr. 2019. Deletion of both centrin 2 (CETN2) and CETN3 destabilizes the distal connecting cilium of mouse photoreceptors. J Biol Chem. 294:3957-3973.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

In this study, Arslanhan et al. propose CCDC15 as a novel component of the centriole inner scaffold structure with potential roles in centriole length control, stability and the primary cilium formation in cultured epithelial cells. Using proximity labelling they explore the common interactors of Poc5 and Centrin-2, two resident molecules of the centriole inner scaffold, to hunt for novel regulators of this structure. The authors leverage expansion microscopy-based localization and siRNA-dependent loss-of-function experiments to follow up on one such protein they identify, CCDC15, with the aforementioned roles in centriole and cilia biology.

This study is designed and laid out nicely; however, to be able to support some of the important claims regarding their proximity labelling results and exploration on the roles of CCDC15, there are several major technical and reproducibility concerns that deem major revision. Similarly, the introduction (perhaps inadvertently) omits much of the recent studies on centriole size control that have highlighted the complexity of this biological problem. As such, addressing the following major points will be essential in further considering this work for publication.

Major comments:

1. The authors use Poc5 and Centrin-2 molecules as joint baits to reveal the interactome of the centriole inner scaffold, however the work lacks appropriate experimental and analytical controls to argue that this is a proximity mapping "at the centriole inner scaffold". In its current state, it is simply an interactome of total Poc5 and Centrin-2, and it might be misleading to call it an interactome at the centriole inner scaffold (the statistical identification of shared interactors cannot do full justice to their biology at the centrosome). Appropriate expression data needed to delineate how large the centrosomal vs. cytoplasmic (or nucleoplasmic) fraction is for either of these molecules, both without and upon the addition of biotin (to see whether the bulk of interaction data stem from the cytoplasm/nucleoplasm or the centrioles themselves). The authors can test this by selectively blotting a lysate fraction containing the centrosomes after centrifugation, and compare them with the simultaneous blot of the supernatant (which were readily used for the blots presented in Fig. 1B). This experiment also becomes very relevant for the case of Centrin-2, as it also heavily localizes to the nucleoplasm as the authors found out (see Fig. 1A and Fig. S1A).

Additionally, an orthogonal approach should be taken to perform bio-image analysis on their biotin/streptavidin imaging data to demonstrate the exact ratios between the centrosomal vs. cytoplasmic/nucleoplasmic biotin activation with appropriate signal normalization between the biotin/streptavidin images. This is particularly important, as although the authors claim that these cells stably express the V5BirA*, it seems that there is partial clonality to the expression. Some cells in both the Poc5 and Centrin-2 fusion constructs appear to lack the V5/Streptavidin signals upon Biotin addition (such as the two cells in the centre right in Poc5, and again a cell in the centre right for Centrin-2 images). In its current form, Fig. 1A lacks signal quantification and does not report any information about the replicates and distributions of the data. I worry that this may raise concerns on the reproducibility if published in its current form.

Similarly, it is not clear whether the expression of Poc5 and Centrin-2 fusion molecules somehow interfere with their endogenous interactions or function. At least some loss-of-function (e.g., RNAi) experiments should be performed where the depletion of endogenous proteins should be attempted to rescue by the fusion constructs. This will help evaluate whether the fusion proteins can rescue the depletion of their endogenous counterparts and behave as expected from a wild-type scenario.

Overall, as the entire claim around the proximity mapping revolve around its assumption about the centriole inner scaffold, these controls seem imperative to substantiate the ground truth of the biology presented in the manuscript. 2. I am curious about the choices of the cell lines in this work. The proximity mapping to reveal CCDC15 as a candidate protein for centriole inner scaffold was performed in HEK293T cells (human embryonic kidney), however its immunostaining was performed using RPE1 and U2OS cells (human retinal and osteosarcoma epithelial cells respectively). This raises questions regarding the generality of CCDC15 as a centriole inner scaffold protein. Could CCDC15 be simply unique to the centriole inner scaffold of epithelial cells such as RPE1 and U2OS cells? Or could the authors demonstrate any information/data on whether it's similarly localized to the inner scaffold in embryonic kidney cells or other cell types? If not, the claims should be moderated to reflect this fine detail. 3. Discussions and data on the localization of CCDC15 to centriolar satellites appear anecdotal and not fully convincing (Fig. S2D). Given that the authors test the relevance of PCM1 for CCDC15's centriolar localization, it is key to have quantitative data supporting their claim that centriolar satellites can help recruit CCDC15 to the centriole. Could the authors quantify what proportion of CCDC15 localize to the centriolar satellites? One way to do this could be to quantify the colocalization coefficience of CCDC15 and PCM1 signals. 4. Similar to above (#3), there is no quantitative information on the co-localization or partial co-localization of the signal foci in Fig. 3A and 3B. The authors readily study CCDC15's localization in wonderful detail in their expansion microscopy data, so they could actually consider taking out Fig. 3A and 3B, as the data seem redundant without any quantification. 5. Do the authors also feel that CCDC15 localize to the core lumen in a somehow helical manner (Fig. 1A, Fig. 1F top and bottom panels, Fig. 5A etc.)? Le Guennec et al. 2020's helical lattice proposal for the inner scaffold further reaffirms that CCDC15 is indeed a likely major component of the inner scaffold. In my view, authors should state this physical similarity explicitly to further support their findings on CCDC15. 6. The data on the link between the CCDC15 recruitment and the centriole growth (Fig. 4F) or the G2 phase of the cell cycle (Fig. 4H) are not fully convincing without quantitative data. For Fig. 4F, the authors should consider plotting the daughter centriole length vs the daughter CCDC15 intensities against each another, to see whether more elongated daughters truly tend to have more CCDC15. For Fig. 4H, the argument regarding the cell cycle regulation requires quantification of the bands from several WB repeats, normalized to the expression of GAPDH within each blot (this is particularly relevant, as the bands of CCDC15 do not look dramatically different enough to draw conclusions by eye). 7. The authors find herein that CCDC15 depletion lead to centrioles that are ~10% shorter than the controls. With the depletion of Poc5 and Wdr90 (other proposed components of the inner scaffold), the centrioles end up larger however (Steib et al., 2020). If the role of inner scaffold in promoting centriole elongation is structural, why are these two results the opposite of each other? I realize there is a brief discussion about this at the end of the paper, however, this requires a detailed discussion and speculation on the relevance of these findings. It would be key to clarify whether the inner scaffold as a structure inhibits or promotes centriole growth - or somehow both? If so, how?

There might be some intriguing opposing regulatory action of Poc5 and CCDC15 as demonstrated here, where CCDC15 depletion leads to slightly over-recruitment of Poc5, and vice versa. Does this suggest that a tug-of-war going on between different molecules that localize to the inner scaffold? Does this provide some dynamicity to this structure, which might in turn regulate centriole length both positively and negatively? This may be analogous to how opposing forces of dyneins and kinesins provide robust length control for mitotic spindles. I am speculating here, but hopefully these may provide some useful grounds for further discussion in the paper. If the authors deem it interesting experimentally, they can test whether the two molecules indeed regulate centriole length by opposing each other's action, by a double siRNA of CCDC15 and Poc5 to see if this retains the centriole length at its control siRNA size (like how they do a similar test for Poc1's potential co-operativity with CCDC15 in Fig. 6J). 8. In their introduction section, the authors discuss how relatively little is known about the size control of centrioles, however they fail to mention a series of recent primary literature that uncover striking, new mechanisms and novel molecular players that highlight the complexity of centriole size control. This complexity appears to arise from the existence of multitude of length control mechanisms that influence the cartwheel or the microtubule length individually, or simultaneously via yet-to-be further explored crosstalk mechanisms.

a. As such, when the authors talk about the procentriole size control in the introduction, they should discuss and refer to the following studies, in terms of: - How theoretical and experimental work demonstrate that procentriole length may vary dependent on the levels of its building block Sas-6 in animals (Dias Louro et al., 2021 PMID: 33970906; Grzonka and Bazzi, 2022 bioRxiv). - How a homeostatic Polo-like kinase 4 clock regulates centriole size during the cell cycle (Aydogan et al., 2018 JCB PMID: 29500190), and how biochemistry and genetics coupled with mathematical modelling unravel a conserved negative feedback loop between Cep152 and Plk4 that constitutes the oscillations of this clock in flies (Boese et al., 2018 PMID: 30256714; Aydogan et al., 2020 PMID: 32531200) and human cells (Takao et al., 2019 PMID: 31533936).

b. Similarly, when the authors refer to centriole size control induced by microtubule-related proteins, they should highlight the further complexity of this process by referring to: - How a molecule located at the microtubule wall, Cep295/Ana1, can regulate centriole length in flies (Saurya et al., 2016 PMID:27206860) and human cells (Chang et al., 2016 PMID:27185865) - like all the other centriolar MT molecules that the authors discuss in the manuscript. - How a crosstalk between Cep97 and Cep152 influences centriole growth in fly spermatids (Galletta et al., 2016 PMID:27185836). - How a crosstalk between CP110-Cep97 and Plk4 influences centriole growth in flies (Aydogan et al., 2022 PMID:35707992), and this molecular crosstalk is conserved, at least biochemically, in human cells (Lee et al., 2017 PMID:28562169).

Minor comments:

• Introduction section: Literature reference missing for the sentence starting with "Importantly, the stable nature of centrioles enables them to withstand...".
• Fig. S1 legend: A typo as follows: CRAPome banalysis should read CRAPome analysis.
• Fig. S2: Info on the scale bar in the legend is missing in Fig. S2A. Scale bars for different panels are missing in general in Fig. S2A.
• Fig. 3A and 3B: When displaying the data, coloured cartoon diagrams would be beneficial to guide the reader who are not fully familiar with the spatial orientation of these proteins.
• Fig. 3H: No information about the sample number (number of cells or technical repeats examined) reported.
• Fig. S3B legend: A typo as follows: CCD15-depelted RPE1 cells should read CCDC15-depleted RPE1 cells.
• Fig. S3B legend: A typo as follows: cellswere fixed with should read cells were fixed with.
• There are many spelling mistakes and typos throughout the paper. I have listed a few examples above, but please carefully read through the manuscript to correct all the errors.
• Fig. S3E: The orange columns depicting % of cells with Sas-6 dots look awkward. Why the columns look larger than the mean line? Please correct as appropriate.
• Although authors provide microscopy information for the U-ExM and FRAP experiments, there is no information about the microscopy on regular confocal imaging experiments which should be detailed in Materials and Methods. Also, there is no information about the lenses, laser lines and the filter sets that were used in the imaging experiments. These should be provided as well.
• Fig. 2A: lacks a scale bar.
• Fig. 2C legend: lacks info on the scale bar length.
• Fig. 5A legend: lacks info on the scale bar length.
• Fig. 7A: lacks a scale bar.
• Fig. 7G legend: lacks info on the scale bar length.
• Fig. S2C-E: lack scale bars.
• Fig. S3D, F and H: lack scale bars.
• Fig. S3J legend: lacks info on the scale bar length.
• Fig. S4A, B, D and E: lack scale bars.
• Fig. S4C legend: lacks info on the scale bar length.
• Fig. S4G legend: lacks info on the scale bar length.

Significance

The findings of this study join among the relatively new literature (e.g., Steib et al., 2020 and Le Guennec et al. 2020) on the nature of centriole inner scaffold and its potential roles in centriole formation, integrity and its propensity to form the primary cilium. Therefore, it will be of interest to a group of scientists studying these topics in the field of centrosomes/cilia.

My expertise is on the biochemistry and genetics of centriole formation in animals.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary:

In this manuscript, Arslanhan and colleagues use proximity proteomics to identify CCDC15 as a new centriolar protein that co-localizes and interacts with known inner scaffold proteins in cell culture-based systems. Functional characterization using state-of-the-art expansion microscopy techniques reveals defects in centriole length and integrity. The authors further reveal intriguing aberrations in the recruitment of other centriole inner scaffold proteins, such as POC1B and the SFI1/centrin complex, in CCDC15-deficient cells, and observe defects in primary cilia.

Major comments:

• The authors present a high-quality manuscript that identifies a novel centriolar protein by elegantly revealing and comparing the proximity proteomes of two known centriolar proteins, which represents an important component for the maintenance of centrioles.
• Data are often presented from two independent experiments (n = 2), which is nice, but also the minimum for experiments in biology. It is strongly recommended to perform at least three independent experiments.
• The protein interaction studies presented in Fig. 3 could be of higher quality. While it is great that the authors compared interactions to the centriolar protein SAS6, which is not expected to interact with CCDC15, the presented data raise many questions. 1) In most cases, co-expression of tagged CCDC15 stabilizes the tested interaction partners, such that the overall abundance seems to be higher. The increase in protein abundance is substantial for Flag-FAM161A (Fig. 3D) and GFP-Centrin-2 (Fig. 3E) and is even higher for the non-interactor SAS6 (Fig. 3G), while it cannot be assessed for GFP-POC1B (Fig. 3F). Hence, the higher expression levels under these conditions make it more likely that these proteins are "pulled down" and therefore do not represent appropriate controls. 2) All Co-IP experiments are lacking negative controls in the form of proteins that are not pulled down under the presented conditions. 3) The amounts of co-precipitation of the tested proteins appears very different. Could this reflect strong or weak interactors, or does it reflect the abundance of the respective proteins in centrioles?
• The observation that IFT88 is supposedly decreased at the base of cilia in CCDC15-depleted cells requires additional experiments/evidence. Fig. 7G shows the results of n = 2 and more importantly, a similar reduction of gamma-tubulin in siCCDC15. Could the observed reduction in IFT88 be explained by a decrease in accessibility to immunofluorescence microscopy? Would the reduction in IFT88 at the base also be apparent when the signals were normalized to gamma-tubulin signals?
• The observed Hedgehog signaling defects are described as follows: "CCDC15 depletion significantly decreased the percentage of SMO-positive cells". It is similarly described in the figure legend. If this was true, the simplest explanation would be that it reflects the reduction in ciliation rate (which is in a similar range). If SMO-positive cilia (instead of "cells") were determined, the text needs to be changed accordingly.
• OPTIONAL: While expansion microscopy is slowly becoming one of the standard super-resolution microscopy methods, which is particularly well validated for studying centrioles, the authors should consider confirming part of their findings (as a proof of principle, surely not in all instances) by more established techniques. This could serve to convince critical reviewers that may argue that the expansion process may induce architectural defects of destabilized centrioles, as observed after disruptions of components, such as in Fig. 6. Alternatively, the authors could cite additional work that make strong cases about the suitability of expansion microscopy for their studies, ideally with comparisons to other methods.

Minor points:

• Text, figures, and referencing are clear and accurate, apart from minor exceptions.
• The title suggests a regulator role for CCDC15 in centriole integrity and ciliogenesis, which has formally not been shown.
• As the authors observe changes in centriole lengths in the absence of CCDC15, it would be very insightful to compare these phenotypes to other components that affect centriolar length, such as C2CD3, human Augmin complex components (as HAUS6 is identified in Fig. 1) or others. These could be interesting aspects for discussion, additional experiments are OPTIONAL.
• While the reduced ciliation rate in the absence of CCDC15 is convincing, the authors did not investigate "ciliogenesis", i.e. the formation of cilia, and hence should re-phrase. The sentence in the discussion that "CCDC15 functions during assembly" should be removed.
• The existence of stably associated CCDC15 pools with centrosomes (Fig. 2) requires further evidence. The recovery of fluorescence after photobleaching in FRAP experiments is strongly dependent on experimental setups and is only semi-quantitative. A full recovery is unrealistic, hence, it is ideally compared to a known static or known mobile component. I personally think this experiment -as it is presented now- is of little value to the overall fantastic study. The authors may consider omitting this piece of data.
• The data that CCDC15 is a cell cycle-regulated protein is not very convincing (see Fig. 3H), as the signals area weak and the experiment has been performed only once (n= 1). This piece of data does not appear to be very critical for the main conclusions of the manuscript and may be omitted. Otherwise, this experiment should be repeated to allow for proper statistical analysis.
• Experimental details on how "defective centrioles" are determined are missing.
• For figures, in which the focus should be on growing centrioles (see Fig. 4), it could be helpful to guide the reader and indicate the respective areas of the micrographs by arrows.
• Page18: "centriole length shortening" could be changed to "centriole shortening".
• It is unclear how the authors determine distal from proximal ends of centrioles in presented micrographs (see Fig. 5D).
• Fig. 7A is missing scale bars and Fig.7 overall is lacking rectangle indicators of the areas that are shown at higher magnification in the insets.
• Fig. 7C displays cilia that appear very short, especially when comparing to the micrographs and bar graphs presented. The authors may want to explain this discrepancy.

Significance

• From a technical point of view the authors use two state-of-the-art technologies, namely proximity labeling combined with proteomics and ultrastructure expansion microscopy, that are both challenging and very well suited to address the main questions of this study.
• General assessment: The presented study is of highest experimental quality. Despite being very challenging, the expansion microscopy and proximity proteomics experiments have been designed and performed very well to allow solid interpretation. The results of the central data are consistent and allow strong first conclusions about the putative function of the newly identified centriolar protein CCDC15. The study presents a solid foundation for future hypothesis-driven, mechanistic analysis of CCDC15 and inner scaffold proteins in centriole length control and maintaining centriole integrity. The only limitation of the study is that the technically simpler experiments should be repeated to allow proper statistical assessment, which can be addressed easily.
• Advance: This is the first study that identifies CCDC15 as a centriolar protein and localizes it to the inner scaffold. It further describes a function for CCDC15 in centriole length control and shows its importance in maintaining centriole integrity with consequences for stable cilia formation in tissue culture. The study provides further functional insights into the interdependence of inner scaffold proteins and the role of CCDC15 in the recruitment of the SFI1/centrin distal complex.
• Audience: The manuscript will be of broad interest to the fields of centrosome and cell biology, both from a basic research and genetics/clinical point of view due to the association with human disorders. The state-of-the-art technologies applied will be of interest to a broader cell and molecular biology readership that studies subcellular compartments and microtubules.
• Reviewer's field of expertise: Genetics, imaging, and protein-protein interaction studies with a focus on centrosomes and cilia.

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

Learn more at Review Commons

---

Reply to the reviewers

Manuscript number: RC-2022-01716

Corresponding author(s): Thomas, Langer

1. General Statements

We have performed a comprehensive and time-resolved analysis of the mitochondrial proteome upon induction of cellular senescence and observed rapid metabolic rewiring of mitochondria. In particular, we identified two metabolic pathways, the 1C-folate metabolism and the branched-chain amino acid (BCAA) catabolism that were rapidly rewired upon senescence induction. Our analysis therefore provides unprecedented insight into the function and metabolic adaptations of mitochondria in senescent cells, which could also serve as a reference for future studies. Moreover, we demonstrate that quantification of mitochondrial abundance using the mitochondrial volume allows to reconcile seemingly contradicting conclusions on the function/fitness of mitochondria in senescent cells. In our opinion, our studies therefore represent a valuable and significant contribution to our understanding of the role of mitochondria in senescence.

2. Point-by-point description of the revisions

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

In the manuscript "Metabolic rewiring of mitochondria in senescence revealed by the time resolved analysis of the mitochondrial proteome", Kim et al performed the proteomic approach to investigate the mitochondrial programming during senescence induction of cultural fibroblast. The main findings mentioned by author are the enhanced catabolism of branched-chain amino acids and the reduction of 1C-folate metabolism.

1. * The author claimed their approaches are "time-resolved analysis" from day 1 to day 7 during the senescence induction. However, they did not adequately describe the transition of molecular signatures and cellular states from normal cells to senescent cells in each timepoint. The authors need to quantify the protein levels of p16, p21, and p53 and transcript levels of SASP factors in each timepoint to define the stages during induction and discuss how their finding related to this transition. * As per the reviewer's requests, we monitored the molecular markers during the transition to the senescent state by decitabine. The mRNA level of CDKN1A (encoding p21) was significantly increased on days 5 and 7, while that of CDKN2A (encoding p16) was moderately but significantly increased (~40 %) on day 7 (updated Suppl. Figure 1B). Furthermore, mRNA levels of two core SASP genes, IL1A and IL6 were significantly increased on day 7.

These results complement our analysis of protein levels of senescent markers including p16 and p21 and several SASP proteins (updated Suppl. Figure 2C). Of note, decreased intracellular protein levels of HMGB1 and HMGB2 indicate a senescent rather than a pre-senescent state (PMID: 27700366; 23649808) and demonstrate that cells display transcriptomic signatures of a senescent state after decitabine treatment for seven days. However, the mass spectrometric analysis of the cellular proteome revealed an accumulation of p21 on day 5 but not on day 7 (updated Suppl. Figure 2C). We therefore performed additional experiments to further corroborate the senescent state of the cells under these conditions and to collectively address concerns raised by reviewer 1 (points 1, 5) and reviewer 2 (points 2, 3). Senescence is defined by an irreversible cell-cycle arrest. Consistently, cells maintained the proliferation-deficient state and did not re-enter the cell cycle for 7 days after the removal of decitabine (updated Suppl. Figure 1D). Moreover, cells reached a plateau of SA-b-Gal positivity on day 7, which was maintained after 7 days of decitabine removal (updated Suppl. Figure 1E). We also examined the protein levels of several senescence markers (updated Suppl. Figure 1F). DNA damage was acutely induced upon decitabine treatment as evidenced by phosphorylated H2A.X at serine 139 (p-H2A.XS139). This was accompanied by the rapid reduction of lamin B1 (LMNB1) which is a hallmark of senescence (PMID: 22496421). In agreement with the EdU data (updated Suppl. Figure 1D), the loss of phosphorylated Rb at serine 807/811 (p-RbS807/811) and cyclin A2 (CCNA2), essential for G1-S transition, from day 3 indicated a cell cycle-arrested state.

Together, we conclude that cells start to become senescent on day 5 and reach the senescent state on day 7 after decitabine treatment. We now show these data in the updated Supplementary figure 1 and modified the text accordingly.

• The authors treated decitabine and doxorubicin in different periods. Sometimes they properly prepared two different control groups, and sometimes they use only one control group. This issue should be unified. *

Proteomic analysis revealed that DMSO treatment for 7 days did not affect the cellular proteome (Reviewer figure 1), indicating that DMSO is as ineffective as H2O in our experimental conditions. When cells treated with decitabine and doxorubicin were assayed in the same experiment, cells treated with DMSO were used as a common control. When decitabine and doxorubicin-treated cells were assayed independently, either DMSO or H2O was used as the control accordingly.

Reviewer figure 1. The pairwise comparison of proteome from cells treated with DMSO at different time points. The x-axis denotes log2FC and y-axis does –log(adj.P-value). The data are of the same origin from the updated Suppl. Figure 2.

• The mitochondria dysfunction and the increasing mitochondrial mass in cellular senescence have been reported in cell culture study previously. The authors described that the mitochondrial activity normalized to mitochondria volume decreased in senescent cells, which raises the importance of the exact value (mitochondria volume) they used for normalization. However, the quantification result in Figure 1 shows obvious batch effects and variation which make their estimation not reliable enough. The authors should increase replicate numbers in image analysis to provide robust quantification results. Besides, the orthogonal method for measuring mitochondrial mass should be also performed to confirm that the FC of volume is reliable. *

We have increased the number of experiments quantifying the mitochondrial volume in senescent cells (updated Figure 1B). Despite some variation between experiments, we observed a significant increase in mitochondrial volume, which could be explained by an increased length of mitochondrial tubules, while the mitochondrial width remained unaltered. New quantification revealed an about 8-fold increase of mitochondrial volume both upon decitabine and doxorubicin treatment (about 30% lower than our original calculation). These experiments are now shown as violin plots, showing the distribution of the experimental data along with the median and quartile values (updated Figure 1B). They further substantiate our conclusion that the bioenergetic activity of mitochondria is increased per senescent cell, but decreased when calculated relative to the mitochondrial volume.

• The normalization and scaling strategy of proteomic data were not described in method. Senescent cells show increasing cell size and protein abundance. The authors should describe how they processed the peptide counts in detail. It is confused whether the up regulation (FC>1) represents the increment per cell, per mitochondrion, or in the protein abundance. *

We apologize for not being clear at this point. For clarification, we have amended the text in the method part on p. 22:

´The proportion of proteins of a certain subcellular compartment with the total cellular proteome was calculated by dividing the sum of the TMT reporter intensities for all proteins of this compartment by the sum of TMT reporter intensities for all quantified proteins. Prior to differential expression analysis, TMT reporter intensities were normalized to within the TMT multiplex using VSN (PMID: 12169536). Intensity normalization and differential expression analysis was carried out using proteins quantified in all 32 samples (total peptide counts) or using the subset of mitochondrial proteins only (mitochondria-specific peptide counts). Thus, the fold-change denotes the change in protein abundance within a given (sub-)proteome.`

• The protein amounts of CDKN1A and CDKN2A were not increased on day 7 in Sup Fig 2C. The authors need to provide the explanation of them to prevent bias in their cellular senescence-dependent findings. In addition, ARF and INK4a should be separately quantified.

*

We have performed a series of additional experiments to corroborate that cells are senescent on day 7 (see point 1). We have not detected p14ARF, but p16INK4A in our proteomic analysis. CDKN2A thus refers to p16INK4A. We have updated Supplementary figure 2C for clarification.

• This comment is also related to comment #4. In line 223, the authors mentioned "This is accompanied by a decrease in mitochondrial translation, consistent with the observed decreased respiratory activity of mitochondria in senescent cells.". First, the linkage between mitochondrial translation and respiratory activity should be further illustrated. Second, the results in Fig. 2G and 2H clearly showed that the overall respiratory activity was enhanced in senescent cells (day 7), and only the activity normalized by mitochondrial volume showed decreased. How the authors normalize proteomic data from whole cell lysates with mitochondrial mass is missed in this manuscript.*

Mitochondrial DNA encodes essential subunits of OXPHOS complexes. Therefore, decreased mitochondrial translation is accompanied by reduced respiratory activity. We have clarified this point on p. 13 of the revised manuscript.

The proteomic data show the abundance of mitochondrial proteins within the cellular proteome. Our proteomic analysis established that the mitochondrial protein mass increases proportionally to that of the cell (updated Figure 3B), consistent with the increase in mitochondrial volume. For clarification, we have now improved the description of our quantification method in the manuscript on p. 22.

Reviewer #1 (Significance (Required)):

Overall, the central findings of the study provide informative evidence to support several concepts and biological observations in previous reports. However, because of the lack of sufficient description and experimental verification, part of the conclusion is not rigorous enough and needed to be further improved.

To address the concerns of the reviewer, we provide now more experimental evidence that the cells attain the senescent state when analyzed by time-resolved proteomics and improved the description of our experimental procedures.

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

*Kim et al. applied a new method to demonstrate that senescent cells accumulate with dysfunctional mitochondria. They isolated mitochondrial to profile mitochondrial proteome in a time-dependent manner and revealed a metabolic shift in mitochondria during the senescence process. However, many groups have widely studied mitochondrial biology in senescent cells (Joao F. Passos and Thomas Von Zglinicki et al.). Based on the previous finding, cellular senescence is also known for dramatic changes in mitochondrial mass, dynamics, structure, metabolism, and function. Thus, the current work is descriptive and incremental. *

We kindly disagree with this statement of the reviewer. We do appreciate and acknowledge previous studies on mitochondria in senescent cells, but are convinced that our work provides novel insight, as to how mitochondria are affected upon establishment of senescence. Rather than considering mitochondria as dysfunctional, our proteomic and metabolomic data indicate reprogramming of mitochondria: reduced respiratory activities are accompanied by enhanced BCAA catabolism, fatty acid metabolism, and Ca2+ transport (updated Figure 4A, updated Figure 5).

We would also like to point out that several reports on the role of mitochondria in senescent cells were seemingly contradicting and difficult to reconcile with each other. Some studies observed a decreased membrane potential and increased ROS production in senescent cells, while at the same time, mitochondrial respiration appeared to be increased (PMID: 15018610, 20160708). Other studies reported an increased OXPHOS activity due to enhanced mitochondrial metabolism of glucose/pyruvate and fatty acids (PMID: 23945590, 23685455, 30778219, 22421146). Our work provides a rationale for these apparently disparate findings, highlighting that the assessment of mitochondrial function/fitness in senescent cells requires the accurate determination of mitochondrial mass/volume. We show that senescent cells harbor more mitochondria with reduced respiratory activity.

As a side note, we did not isolate mitochondria but determined the cellular proteome and analyzed organellar proteomes (such as the mitochondrial proteome) based on reference databases (see also updated Suppl. Figure 2D).

I have specific comments listed below:

*Major: 1. The authors stated that mitochondrial DNA (mtDNA) was decreased per mitochondrion. However, in Fig 2A, there is no statistical significance. So, this statement is not valid. *

We have determined mtDNA levels in two additional, independent experiments but did not observe a statistically significant decrease in senescent cells due to a relatively high experimental variability (n=5). We have updated Figure 2A and corrected the text accordingly on p. 6.

• The authors investigated the mitochondrial proteome alterations during CS development (days 1 to 7 post-treatment of decitabine). However, in supplemental Fig. 2C, p16 and p21 did not increase on day 7. So, the mitochondria authors studied are the authors studied in a pre-senescent state. What is the rationale for this study, and why did the authors not examine mitochondria at the fully senescent stage?*

We kindly disagree with the statement of the reviewer that analyzed was a pre-senescent state. As outlined in detail in reply to reviewer 1 (point 1), the analysis of a series of established senescent markers on mRNA and protein levels demonstrates that cells have reached the senescent state after seven days of drug treatment. Most importantly, we show by EdU staining in Suppl. Figure 1D that cells are in cell cycle arrest on day 7 and now provide additional evidence (EdU- and SA-b-Gal staining, senescence and cell cycle markers) that this state is maintained for another seven days, excluding a quiescent state (updated Suppl. Figure 1D, 1E). This includes the loss of lamin B1, which is associated with senescence but not quiescence (PMID: 22496421).

• The authors claimed that the fraction of mitochondrial proteins did not significantly change (Shown in Fig. 3B). Again, this could be due to cells at the pre-senescent stage; I wonder if this change could be significant after cells are fully senescent.*

As discussed above, we provide now additional evidence demonstrating that cells are in a senescent state. Of note, our findings are in agreement with a recent report (PMID: 35987199), showing that mitochondrial and cellular proteomes are altered proportionally in senescent cells.

• The data in Fig. 4E is not statistically significant. Please increase n to confirm your conclusion.*

Considering the causal relationship between mitochondrial translation and OXPHOS deficiency, we feel that this point is rather confirmatory. We therefore present now instead two representative images (updated Figure 4D)

• Are these mitochondrial proteome alterations associated with the senescent cell's functional output or other features? Are these changes cell type-dependent.*

To address the functional impact of the mitochondrial reprogramming of metabolism in senescent cells, we blocked BCAA catabolism by knocking down the rate-limiting enzyme BCKDHA, before treating cells with decitabine or doxorubicin. However, the interpretation of these experiments was hampered by increased cell death upon BCKDHA knockdown independent of DNA damage. Thus, the BCKDHA appears to be essential for cell viability under our cell culture conditions, precluding an assessment of the role of the BCAA catabolism for the functional output of senescent cells. We included these data in the updated Suppl. Figure 5D and briefly mentioned them in the text on p. 10.

• I suggest the authors clarify explicitly the knowledge gap that the current study accomplished.*

We apologize for not being clear and have revised the manuscript to clarify this point. As pointed out in the General Statement, we have performed a comprehensive and time-resolved analysis of the mitochondrial proteome upon induction of cellular senescence and observed rapid metabolic rewiring of mitochondria. In particular, we identified two metabolic pathways, the 1C-folate metabolism and the BCAA catabolism that were rapidly rewired upon senescence induction. Our analysis therefore provides unprecedented insight into the function and metabolic adaptations of mitochondria in senescent cells, which could also serve as a reference for future studies. Moreover, we demonstrate that quantification of mitochondrial abundance using the mitochondrial volume allows to reconcile seemingly contradicting conclusions on the function/fitness of mitochondria in senescent cells. In our opinion, our studies therefore represent a valuable and significant contribution to our understanding of the role of mitochondria in senescence.

Minor: Please carefully check the statistical analysis. N of 2 is not sufficient for One-way ANOVA.

We have revisited the statistical analysis of our experiments and noted a mistake in the original Figure 4E (statistical analysis with n=2). We apologize for this mistake and have corrected it (see reply point 4) as updated Figure 4D.

Reviewer #2 (Significance (Required)):

I do not see the significance of the current work. The work is descriptive and incremental, which reduce the impact of this manuscript.

My expertise is in cellular senescence and aging.

As outlined in the General Statement and in reply to reviewer 2 (point 6), we provide insight into changes in the mitochondrial proteome in senescent fibroblasts in an unprecedented manner. Although metabolic reprogramming of mitochondria has been described, our results highlight the relevance of two metabolic pathways in senescent cells, the one-carbon and the BCAA metabolism, which to our knowledge has not been described before. Moreover, our work reveals the importance to adequately quantify mitochondrial abundance, when assessing the bioenergetic activity of mitochondria in (enlarged) senescent cells.

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

The manuscript by Kim and colleagues investigated the metabolic and proteomic changes in senescent fibroblasts. The authors performed a time-resolved analysis of the proteome and revealed the impacts on mitochondrial proteome. They also applied tracing approaches to further demonstrate the impact on mitochondrial metabolism and revealed a potential impact of branched-chain amino acid catabolism and carbon-folate metabolism in senescent fibroblasts. They concluded that the reprogramming of mitochondria influences the senescence-associated secretory phenotype (SASP) impacting diseases associated with senescent cells.

1. • The authors quantified the volume of mitochondria to examine mitochondrial functions. Some data were normalized "per cell" and "per mito". Should it read per mito volume? The comparison between the two normalization procedures in Fig 2 is important but also confusing. Can the authors speculate why non-mitochondrial respiration seems to be increased in decitabine/doxorubicin conditions (Fig. 2G/H)? Are the data significantly affected after normalization to non-mitochondrial respiration?* As anticipated by the reviewer, ´per mito` indeed refers to the mitochondrial volume. We have adjusted the text on p. 6 to clarify this point.

An increase in non-mitochondrial respiration in senescent cells was also reported by the von Zglinicki group (PMID: 28330601). A major source of non-mitochondrial OCR is thought to be cytosolic/peroxisomal H2O2 production by cyclooxygenase (COX), and NADPH oxidase, which consume oxygen to produce H2O2. Especially, the cytosolic H2O2 level is known to be higher in senescent cells (PMID: 10075689). Moreover, cyclooxygenase 1 (PTGS1) is significantly upregulated in our proteomics data (Reviewer figure 2), which could explain the reason for the increased non-mitochondrial OCR in senescent cells.

Reviewer figure 2. The pairwise comparison of PTGS1 level at each time point from the proteomics data in Figure 3. **: P Of note, normalization to non-mitochondrial respiration does not significantly alter our conclusions. We observed a significant decrease in basal respiration and spare respiratory capacity both upon decitabine and doxorubicin treatment, which recapitulates our original findings taking mitochondrial abundance into account (Reviewer figure 3).

Reviewer figure 3. Oxygen consumption rates of senescent fibroblasts after normalization to non-mitochondrial OCR. (A, B) Left: representative graphs of OCR data. Right: quantification of respiratory parameters based on the OCR graph. Welch t-test, Bonferroni-Dunn correction, n=5.

• The tracing data are impactful and critical to confirm metabolic changes. Can the authors explain why they added the tracer to the regular growth medium (instead of substituting the metabolite of interest)? The media composition of the proteomic and the metabolic experiments is not identical. For instance, additional 5.5mM 13C glucose has been added to the 13C glucose tracer experiment while all other experiments were performed with 5.5mM 12C glucose only. Changes in media composition certainly affect cell function and metabolism. The authors may want to repeat key tracing experiments to mimic experimental conditions used in proteomics analysis.*

The use of glucose-containing MEM had only practical reasons, because glucose-free MEM is not commercially available. However, this does not confound our proteomic analysis between day 1 and 7, since the tracing experiments in MEM containing glucose and 13C glucose were performed to the established senescent cells on day 7.

• The impact on branched-chain amino acids is interesting. Did the authors observe 13C incorporation into the TCA cycle from BCAA? Increased BCAA catabolism may increase mitochondrial respiration, but the authors observed decreased OCR in senescent cells. Further, does inhibition of BCAA catabolism rescue the phenotype observed in senescent cells?*

We did not observe any significant incorporation of BCAA carbons into malate but found an 3-5 fold increased flux into lipogenic TCA cycle intermediates, such as acetyl-CoA and citrate (updated Figure 5D). In agreement with glucose and glutamine being the major source of lipogenic acetyl-CoAs in cell culture (PMID: 31119666), the absolute changes are rather minor. We therefore carefully discuss the possibility that BCAA carbons are metabolized to supply the lipogenic carbon moieties in senescent cells rather than increasing mitochondrial respiration in the text on p. 12.

To address the functional impact of the mitochondrial reprogramming of metabolism in senescent cells, we blocked BCAA catabolism by knocking down the rate-limiting enzyme BCKDHA, before treating cells with decitabine or doxorubicin (see also Reviewer 2, point 5). However, the interpretation of these experiments was hampered by increased cell death upon BCKDHA knockdown independent of DNA damage. Thus, the BCKDHA appears to be essential for the cell viability under our cell culture conditions, precluding an assessment of the role of the BCAA catabolism for the functional output of senescent cells. We included these data in the updated Suppl. Figure 5D and briefly mentioned them in the text on p. 10.

Reviewer #3 (Significance (Required)):

Understanding the metabolic reprogramming of mitochondria in senescent cells is interesting and of high interest to the research community. However, some clarification is needed on experimental conditions, as media compositions in proteome and metabolome experiments were different which certainly affects cell metabolism and function.

We thank the reviewer for their positive evaluation of the significance of our work. In the revised manuscript, we have now clarified the technical issues raised showing that different culture conditions do not confound our proteomic and metabolomic analysis.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

The manuscript by Kim and colleagues investigated the metabolic and proteomic changes in senescent fibroblasts. The authors performed a time-resolved analysis of the proteome and revealed the impacts on mitochondrial proteome. They also applied tracing approaches to further demonstrate the impact on mitochondrial metabolism and revealed a potential impact of branched-chain amino acid catabolism and carbon-folate metabolism in senescent fibroblasts. They concluded that the reprogramming of mitochondria influences the senescence-associated secretory phenotype (SASP) impacting diseases associated with senescent cells.

1. The authors quantified the volume of mitochondria to examine mitochondrial functions. Some data were normalized "per cell" and "per mito". Should it read per mito volume? The comparison between the two normalization procedures in Fig 2 is important but also confusing. Can the authors speculate why non-mitochondrial respiration seems to be increased in decitabine/doxorubicin conditions (Fig. 2G/H)? Are the data significantly affected after normalization to non-mitochondrial respiration?
2. The tracing data are impactful and critical to confirm metabolic changes. Can the authors explain why they added the tracer to the regular growth medium (instead of substituting the metabolite of interest)? The media composition of the proteomic and the metabolic experiments is not identical. For instance, additional 5.5mM 13C glucose has been added to the 13C glucose tracer experiment while all other experiments were performed with 5.5mM 12C glucose only. Changes in media composition certainly affect cell function and metabolism. The authors may want to repeat key tracing experiments to mimic experimental conditions used in proteomics analysis.
3. The impact on branched-chain amino acids is interesting. Did the authors observe 13C incorporation into the TCA cycle from BCAA? Increased BCAA catabolism may increase mitochondrial respiration, but the authors observed decreased OCR in senescent cells. Further, does inhibition of BCAA catabolism rescue the phenotype observed in senescent cells?

Significance

Understanding the metabolic reprogramming of mitochondria in senescent cells is interesting and of high interest to the research community. However, some clarification is needed on experimental conditions, as media compositions in proteome and metabolome experiments were different which certainly affects cell metabolism and function

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Kim et al. applied a new method to demonstrate that senescent cells accumulate with dysfunctional mitochondria. They isolated mitochondrial to profile mitochondrial proteome in a time-dependent manner and revealed a metabolic shift in mitochondria during the senescence process. However, many groups have widely studied mitochondrial biology in senescent cells (Joao F. Passos and Thomas Von Zglinicki et al.). Based on the previous finding, cellular senescence is also known for dramatic changes in mitochondrial mass, dynamics, structure, metabolism, and function. Thus, the current work is descriptive and incremental. I have specific comments listed below:

Major:

1. The authors stated that mitochondrial DNA (mtDNA) was decreased per mitochondrion. However, in Fig 2A, there is no statistical significance. So, this statement is not valid.
2. The authors investigated the mitochondrial proteome alterations during CS development (days 1 to 7 post-treatment of decitabine). However, in supplemental Fig. 2C, p16 and p21 did not increase on day 7. So, the mitochondria authors studied are the authors studied in a pre-senescent state. What is the rationale for this study, and why did the authors not examine mitochondria at the fully senescent stage?
3. The authors claimed that the fraction of mitochondrial proteins did not significantly change (Shown in Fig. 3B). Again, this could be due to cells at the pre-senescent stage; I wonder if this change could be significant after cells are fully senescent.
4. The data in Fig. 4E is not statistically significant. Please increase n to confirm your conclusion.
5. Are these mitochondrial proteome alterations associated with the senescent cell's functional output or other features? Are these changes cell type-dependent?
6. I suggest the authors clarify explicitly the knowledge gap that the current study accomplished.

Minor: Please carefully check the statistical analysis. N of 2 is not sufficient for One-way ANOVA.

Significance

I do not see the significance of the current work. The work is descriptive and incremental, which reduce the impact of this manuscript. My expertise is in cellular senescence and aging.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In the manuscript "Metabolic rewiring of mitochondria in senescence revealed by the time resolved analysis of the mitochondrial proteome", Kim et al performed the proteomic approach to investigate the mitochondrial programming during senescence induction of cultural fibroblast. The main findings mentioned by author are the enhanced catabolism of branched-chain amino acids and the reduction of 1C-folate metabolism.

1. The author claimed their approaches are "time-resolved analysis" from day 1 to day 7 during the senescence induction. However, they did not adequately describe the transition of molecular signatures and cellular states from normal cells to senescent cells in each timepoint. The authors need to quantify the protein levels of p16, p21, and p53 and transcript levels of SASP factors in each timepoint to define the stages during induction and discuss how their finding related to this transition.
2. The authors treated decitabine and doxorubicin in different periods. Sometimes they properly prepared two different control groups, and sometimes they use only one control group. This issue should be unified.
3. The mitochondria dysfunction and the increasing mitochondrial mass in cellular senescence have been reported in cell culture study previously. The authors described that the mitochondrial activity normalized to mitochondria volume decreased in senescent cells, which raises the importance of the exact value (mitochondria volume) they used for normalization. However, the quantification result in Figure 1 shows obvious batch effects and variation which make their estimation not reliable enough. The authors should increase replicate numbers in image analysis to provide robust quantification results. Besides, the orthogonal method for measuring mitochondrial mass should be also performed to confirm that the FC of volume is reliable.
4. The normalization and scaling strategy of proteomic data were not described in method. Senescent cells show increasing cell size and protein abundance. The authors should describe how they processed the peptide counts in detail. It is confused whether the up regulation (FC>1) represents the increment per cell, per mitochondrion, or in the protein abundance.
5. The protein amounts of CDKN1A and CDKN2A were not increased on day 7 in Sup Fig 2C. The authors need to provide the explanation of them to prevent bias in their cellular senescence-dependent findings. In addition, ARF and INK4a should be separately quantified.
6. This comment is also related to comment #4. In line 223, the authors mentioned "This is accompanied by a decrease in mitochondrial translation, consistent with the observed decreased respiratory activity of mitochondria in senescent cells.". First, the linkage between mitochondrial translation and respiratory activity should be further illustrated. Second, the results in Fig. 2G and 2H clearly showed that the overall respiratory activity was enhanced in senescent cells (day 7), and only the activity normalized by mitochondrial volume showed decreased. How the authors normalize proteomic data from whole cell lysates with mitochondrial mass is missed in this manuscript.

Significance

Overall, the central findings of the study provide informative evidence to support several concepts and biological observations in previous reports. However, because of the lack of sufficient description and experimental verification, part of the conclusion is not rigorous enough and needed to be further improved.

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

Learn more at Review Commons

---

Reply to the reviewers

Detailed Answer to the Reviewers

Reviewer #1

__Summary __

The authors used a novel imaging technique to monitor glutamate release and correlated these measurements with gold standard electrophysiological measurements. The genetically encoded glutamate reporter, iGluSnFR, was expressed in mouse spiral ganglion neurons using the approach described in Ozcete and Moser (2021, EMBO J). The iGluSnFR signals and the postsynaptic currents were measured at the endbulb of Held synapse. A small effect of the expression of iGluSnFR on the mEPSC kinetics was found (but see comment 1). Furthermore, deconvolution of the iGluSnFR signals was performed enabling the comparison of some presynaptic properties assessed with either iGluSnFR or electrophysiology.

We thank the reviewer for her/his appreciation of our work and for the comments that have helped/will help us further improve our manuscript.

__Major comments __

1. The central finding of the study is the prolonged decay time constant of the mEPSC. The difference is small but astonishingly significant (0.172 {plus minus} 0.002 and 0.158 {plus minus} 0.001, P=0.003). The SEM is about 100 times smaller than the measured time constant. This is biologically not plausible. Therefore, I am skeptical about the statistical significance of the results.

We appreciate the feedback of the reviewer. We agree that our presentation of the data was easy to misunderstand and we changed it (see below). We modeled the statistical relationship of kinetic parameters with a mixed effects model (as described in methods). Since the presentation of regression parameters for this kind of data is not very usual in synaptic neuroscience (nor very informative in this study), we instead opted to report SEM and a p-value derived from the fit of the linear mixed effects model. For the SEM, there is no clear way to take into account the clustered nature of the data, so we calculated the SEM over all observations. Since the SEM is proportional to 1/sqrt(n) and the number of recorded mEPSCs is very large, this does indeed yield a very small SEM. We agree that reporting the SEM over all observations is unusual and leads to misunderstandings in this case. Now, we instead report the re-calculated the mean / SEM for all parameters over the median values per cell. We changed the presentation of the data also for the other values presented in the MS in all tables and the relevant parts of the main texts.. We note that the summary statistics do not directly influence the further statistical modeling.

1. Analysis of the size of RRP with electrophysiology and iGluSnFR is potentially interesting but iGluSnFR recordings could not resolve the spontaneous fusion of single vesicles. Therefore, it is not possible to estimate RRP with these iGluSnFR recordings. This limitation of the approach should be emphasised more clearly.

Yes, we think the inability to resolve single vesicles is one of the major limitations of the study and we note this in the introduction and in the relevant section of the discussion. We agree that it should be clear in the relevant section that we are not able to measure RRP size without resolving single vesicle release and modified the wording of the relevant results section to reflect this better (line 267, 497). We still believe that the cumulative release analysis is potentially interesting to researchers in the field, as RRP size is not the only parameter that can be estimated in this way. In particular, an estimation of the release probability in resting conditions is possible by dividing the amplitude of the first response (i.e. response to a single stimulus) by the RRP estimate even without knowing the exact number of vesicles that comprise either.

1. The control conditions (no surgery/no virus injection) are not the correct conditions for comparison with the experimental conditions (surgery/virus injection and sensor expression). The control group should be operated and injected with saline or ideally with a virus expressing GFP at the extracellular membrane. The authors addressed this issue by citing their previous work (Özcete and Moser, 2021). However, I am not convinced that surgery does not induce subtle changes that could explain the small differences in mEPSCs.

This is an excellent point that should be addressed in further research. A slowed decay would be consistent with the idea that iGluSnFR affects glutamatergic transmission by buffering glutamate, but we cannot rule out subtle changes due to the postnatal surgery or AAV-mediated transgene expression. In response to the reviewer’s comment, we modified the text to reflect the possibility of surgery and / or other parts of the expression system being responsible for the changes. We also discuss further control experiments (line 408). Finally, we believe that our comparison is still relevant for researchers using iGluSnFR in the system, as they will be asking if introducing a measurement system affects the underlying quantity.

__Minor comments __

The supplementary figures are not listed in the order in which they appear in the main text.

We now list the supplementary figures in the order in which they appear in the main text.

Figure 2B and 3 are not referenced in the main text.

We now reference the figures in the text.

The PPR in Figure 3 shows a PPR that cannot be evaluated because of the unusual plot with lines that are too thick.

We updated Fig.3 and chose a more straight forward way to display the PPRs.

Line 105: "...while simultaneously monitoring currents in postsynaptic cells". This sentence is not correct given that the EPSCs have not been shown yet at this point of the manuscript.

We removed this part of the sentence.

Line 110: "SV and are not cause are cause by spontaneous action potentials...". The sentence does not make sense.

We corrected the sentence.

Line 168-9: "...we did not find significant differences in amplitude and kinetics...". According to Table 2 (2mM Ca2+ condition), both Imax and Q appear to be almost twice as high in iGluSnFR as in control (2.05 {plus minus} 0.06 and 1.34 {plus minus} 0.03, respectively; P=0.241). Is this not a significant difference?

The difference was not significant. The misunderstanding likely stems from the same problem in the presentation of the values as for the mEPSCs. We replaced the SEMs with the SEMs of the cell median to avert this.

Table 4. 2mM Ca2+ condition. The Rrefill parameter is about an order of magnitude smaller in the iGluSnFR-expressing group. Is this correct or just a typo?

Thank you for spotting this: it was an error with regards to unit conversion. The value for the control group was off by a factor of 10. We corrected this mistake.

Referees cross-commenting

I also agree with the comments of the other reviewers.

Significance

General assessment

This topic is currently of interest because iGluSnFR techniques are widely used. However, the study is preliminary. The scientific progress in terms of quantity and quality is limited. For example, Figs. 1 and 5 show only images and traces with little scientific significance.

Advance

The main advance of the study is the implementation of the deconvolution of the iGluSnFR signal and the comparison of the back extrapolation with the first stimulus (Fig. 6). This comparison was similar between electrophysiology and iGluSnFR when deconvolution of the iGluSnFR data was performed. These data therefore argue against saturation of iGluSnFR, as expected from a large number of previous analyses of iGluSnFR.

There is little methodological improvements compared with the group's previous study (Ozcete and Moser, 2021 EMBO J). In this earlier study, a different synapse was analyzed but the same iGluSnFR was injected into the scala tympani of the right ear through the round window in the same way as in this study. Surprisingly, the authors do not refer to Ozcete and Moser (2021) in the relevant methods section.

Thank you for spotting this omission. We now cite Özçete and Moser (2021) in the appropriate place in the methods section as well.

Reviewer #2

Summary

In the manuscript 'Optical measurement of glutamate release robustly reports short-term plasticity at a fast central synapse' the authors present a careful analysis of whether direct measurements of transmitter release using the genetically-encoded indicator iGluSnFR, are suitable for assessing changes in transmitter release at the spiral ganglion neuron end bulbs of Held in the mouse cochlear nucleus. What sets this study apart from other studies, which have demonstrated the utility of iGluSnFR measurements, is the use of a camera-based fluorescence readout as opposed to confocal or 2P microscopy methods and that it is performed in the cochlear nucleus.

The primary methodology is the comparison of electrophysiological measurements of excitatory postsynaptic currents from bushy cells with fluorescence changes in the end bulbs of iGluSnFR expressing auditory nerve fibers with and without stimulation of the auditory nerve fibers. The experiments are technically demanding and introducing genetically encoded indicators in neurons of the cochlea is no small accomplishment. An important observation is that mEPSCs are slightly modified (prolonged) due to expression of iGluSnFR in the presynaptic end bulbs. This is perhaps not surprising as iGluSnFR binds glutamate and may act as a buffer to reduce the peak and slightly prolong the increase in cleft glutamate concentration after release from synaptic vesicles. To my knowledge, others have not reported iGluSnFR effects on mEPSCs. Perhaps earlier studies have not checked as carefully, alternatively previous studies had a too-low fraction of presynaptic terminals expressing iGluSnFR (or less expression of iGluSnFR) to detect a change in EPSC parameters, or this is a synapse-specific phenomenon. However, the authors demonstrate that EPSCs evoked by electrical stimulation of the auditory nerve fibers are unaffected by expression of iGluSnFR in the presynaptic neurons. Further findings are that the determined decay time constant is substantially longer than at other synapses (~16 ms at hippocampal synapses, Dürst et al., 2018). Synaptic depression was robustly reported by iGluSnFR at this synapse, but determination of single quantal events and thus quantal analysis was not really possible at this synapse using iGluSnFR in conjunction with the imaging and analysis techniques presented. The manuscript is carefully written and presented.

We thank the reviewer for her/his appreciation of our work and for the comments that have helped/will help us further improve our manuscript.

Major points

1) The ROIs are selected to be 'outer bounds' of the glutamate spread from the synapses being studied. My concern is that these generously-sized ROIs include signal from many iGluSnFR molecules which are distal to the release sites and thus will be reached only slowly by low concentrations of glutamate or be contributing only noise and no changes in fluorescence. I suggest the temporal resolution could be improved by restricting the analysis of fluorescence changes to fewer pixels within the ROIs with the fastest rising/highest amplitude responses.

Thanks for this helpful comment: The data in our data set should be well-suited to perform this analysis in addition to the presented analysis and so we added this new analysis to the Revision Plan.

2) The observation that despite a 2 fold increase in eEPSCs when changing from 2 mM to 4 mM extracellular calcium there is no change in iGluSnFR peak is curious as pointed out by the authors but not really discussed.

We don’t currently have an obvious explanation but consider saturation of the iGluSnFR peak response likely to contribute. In response to the comment of the reviewer, we have added the analysis of integrated iGluSnFR data, which we previously found to be more robust toward saturation than the peak, to the revision plan. We plan to add the relevant discussion along with the new analysis.

Are the traces presented in Figure 5 examples from the same recording?

Traces in fig. 5 are grand averages (wording modified for clarity). Unfortunately, it was not possible to routinely measure iGluSnFR responses from the same cell in 2mM Ca2+ and 4 mM Ca2+ as the time needed for the protocols was rather long which influenced cell stability and imaging conditions would deteriorate during the exchange of the bath solution. We think it is not quite possible to directly compare the absolute iGluSnFR responses at different extracellular Ca2+ levels.

Assuming the examples are from one cell I first assumed the lack of change of peak was saturation of iGluSnFR but the larger fluorescence change with 100 Hz stimulation suggests otherwise. How many endbulbs are contacting one BC? Do you capture iGluSnFR responses from only one or several? In the previous point I suggested that restricting analysis to the soonest reacting pixels might improve temporal resolution but in the case of detecting the peaks with higher and normal calcium, these fastest reacting signals are probably also more likely to be saturated with glutamate.

The eEPSCs elicited by this stimulation paradigm are monosynaptic (see methods / electrophysiology section), but there might be other iGluSnFR expressing endbulbs on the same bushy cell. Since we reduce the current just enough such that any further reduction leads to a complete failure to elicit an EPSC, we believe these additional endbulbs are not releasing glutamate. We cannot, however, exclude the possibility that iGluSnFR on neighboring structures captures any potential spillover glutamate.

Minor points

• mEPSCs are usually recorded in tetrodotoxin, I didn't find any mention in methods/results

In this system, sEPSCs are not affected by TTX (Oleskevich and Walmsley, 2002) and thus usually recorded without adding TTX. We discuss this more explicitly and added a clarification to reflect this assumption (on line 111).

• the large numbers of abbreviations make it difficult in places to follow the manuscript please at least define them again in the figure legends (e.g. BC, AVCN in figure 1, Q, FWHM in figure 2 etc.)

We went over the manuscript again and removed some abbreviations or redefined them in captions.

• it is a bit unusual to report results of a Wilcoxon test and at the same time mean and SEM instead of medians, if different tests were used then it is important to indicate this where the p values are given or make the sentence in the methods more definitive

We agree that the initial presentation of the data was ambiguous. We changed the presentation to reflect this (see also answer to reviewer 1).

• the liquid junction potential is reported as 12 mV, pretty sure it should be -12 mV (unless the QX-314 or some other of the more exotic ingredients in the extracellular solution is having a dramatic effect on the LJP).

We follow the usual conventions of P. H. Barry, Methods in Enzymology, Vol. 171, p. 678, as described in E. Neher, Methods in Enzymology, Vol. 207, p. 123, in which the LJP is defined as the potential of the bath solution with respect to the pipette solution. We subtracted this positive potential (+12mV) in the end to obtain the membrane potential which therefore was more hyperpolarized than the nominal potential.

I wonder if one of the faster/lower affinity iGluSnFR variants would be better suited for studying this synapse.

We agree with the reviewer that future studies should explore the potential of faster/lower affinity iGluSnFR variants for studying the endbulb synapse. The reasons why we employed the original version include: i) sharing the same mice for studies of cochlear ribbon synapses (Özcete & Moser, 2021) and cochlear nucleus synapses (this MS) for the sake of reducing animal experiments, ii) good signal to background facilitating our first study establishing the recording in brainstem slice, iii) less signal to background and shorter signals with the new variants (as found in preliminary recordings from cochlear ribbon synapses) that would make the endbulb recordings more challenging. We have added the following statement to discussion. “Future imaging studies of glutamate release at calyceal synapses should explore the potential of new iGluSnFR variants with lower affinity that provide more rapid signal decay. This will ideally go along with imaging at higher framerate and might require stronger intensities of the excitation light to boost the fluorescence signal.” on line 430.

The paper would benefit from a careful reading to shorten the text and to check for clarity. For instance page 15 line 436 I don't understand how 'the results can reduce the likelihood of biologically relevant changes'. I think the authors meant something different

Thank you for spotting this. We reworded the sentence (now on line 399): "The data on hand suggests that this is not the case. Firstly, even if a larger sample size may uncover more subtle effects neurotransmission of evoked events, our measurements suggest a small effect size. Secondly, even as we did find changes in mEPSC, it is probable that the biological significance is limited"

• page 5 'width' is misspelled

Fixed.

• page 18 'strychnine' is misspelled

Fixed.

• on many of the figures is text that it much too small

We went through the manuscript and increased the text size in the figures, where appropriate.

__Referees cross-commenting __

I agree with all the comments of the other reviewers - both raise the point that there should be a 'control' AAV injected for comparison of the mEPSCs which I missed but is of course quite important. See https://pubmed.ncbi.nlm.nih.gov/24872574/ for a study of AAV serotype-dependent effects on presynaptic release.

We now added a section on other possible factors influencing the results, citing the study above.

Significance

The main audience for this paper will be fairly specialized. Researchers interested in properties of presynaptic release and some specialists in synaptic transmission in the auditory system will be the main readers/citers of this work.

The work is an important technical/methodological report. It highlights an important effect of expressing iGluSnFR and also demonstrates that the effect is overall not very problematic. Additional problems using iGluSnFR are also indicated.

I am an electrophysiologist, studying synaptic transmission and plasticity with experience using a wide range of optogenetic tools

Reviewer #3

__Summary __

In the present manuscript, the authors explore the information that can be obtained using optical measurement of glutamate release with iGluSnFR on synaptic dynamics in the endbulb of Held.

They virally express iGluSnFR in presynaptic terminals, patch the postsynaptic cells and combine high-frame-rate optical recordings with electrophysiological measurements. Their first finding is that mEPCSs are prolonged when presynaptic cells express the glutamate indicator, which they interpret as buffering of extracellular glutamate by the indicator. Next, they repeated the experiment, this time with stimulating evoked EPSCs. In contrast to the previously observed effects, iGluSnFR did not affect the time course or the amplitude of the evoked EPSCs. The authors then asked whether iGluSnFR signals can be used to study synaptic dynamics, specifically, synaptic depression. In these experiments, the authors observed a change in the paired-pulse ratio with ISI of 10ms, but not longer intervals. They analyzed presynaptic release and did not find statistically significant differences.

Can iGluSnFR signals be used for the analysis of synaptic release? When stimulated at a low frequency of 10Hz (allowing the fluorescence to return close to baseline levels in between pulses), iGluSnFR dynamics were somewhat comparable to postsynaptic signals. At higher frequencies, the slow time course of the indicator prevented the identification of individual responses and the resulting fluorescence had a very different shape. To resolve this problem, the authors used deconvolution analysis (fig 6). This analysis revealed a linear relationship between the optical readout and the patch-clamp data.

I find the manuscript to be clearly written, the findings are well presented and discussed and are novel and of substantial interest to neuroscientists in the field. I do have a number of questions about experiments and analysis that may have an effect on the conclusions of this work.

We thank the reviewer for her/his appreciation of our work and for the comments that have helped/will help us further improve our manuscript.

1. In experiments comparing the effects of iGluSnFR expression on release dynamics (figure 1-4), the authors compare infected presynaptic cells to control (uninfected). The assumption is that synaptic buffering by iGluSnFR may affect glutamate diffusion in the synaptic cleft. However, it is possible that viral infection itself changes presynaptic properties. The authors should compare release from cells infected with GFP or a comparable indicator.

We agree that this is an important control experiment to be done in the future and that causal attribution is not in the scope of this study. A slowed decay would be consistent with the idea that iGluSnFR affects glutamatergic transmission by buffering glutamate, but we cannot rule out subtle changes due to the postnatal surgery or AAV-mediated transgene expression. In response to the reviewer’s comment, we modified the text to reflect the possibility of surgery and / or other parts of the expression system being responsible for the changes. We now also discuss further control experiments (line 408). Finally, we believe that our comparison is still relevant for researchers using iGluSnFR in the system, as they will be asking if introducing a measurement system affects the underlying quantity.

1. Analysis of pool parameters presented in table 4 indicates almost doubling of RRP size with iGLuSnFR with 2 mM Ca++. While not significant, this result may indicate a real effect that may have been missed due to low power (N=3 and 7 for these experiments). I do not believe the authors did a power analysis in this study. How was the number of experiments determined? I would suggest increasing the number of experiments to avoid type II errors.

We thank the reviewer for this critical comment. Indeed, we would also have liked to have a greater statistical power for these experiments, but had to face the situation that the establishing the method required more animals than expected and the animal license did not offer further animals for the analysis. Moreover, we note that the obtained RRP size estimates were generally lower compared to previous estimates of our lab for the endbulb synapse (e.g. Butola et al., 2021: ~20 nA for 2 mM Ca2+ in Fig. 5). This can partially be attributed to the use of cyclothiazide in previous studies, which we avoided given reports of presynaptic effects of cyclothiazide. As the series resistances of the included recordings were below 8 MOhm (mean series resistances: 2mM Ca, injected: 5.58 MOhm; 2mM Ca, control: 5.93 MOhm; 4mM Ca, injected: 5.75 MOhm; 4mM Ca, control: 6.0 MOhm) and series resistance compensation was set to 80% we do not expect clamp-quality to contribute to the smaller estimates in the present data set.

We have now added a statement noting the preliminary nature of these results and indicated that further experiments will be required to more certainly conclude on potential effects of iGluSnFR or the manipulation on endbulb transmission: “Our preliminary train stimulation analysis of vesicle pool dynamics in the presence and absence of AAV-mediated iGluSnFR expression in SGNs has not revealed significant differences between the two conditions. Further experiments, potentially involving faster versions of iGluSnFR and employing trains of different stimulation rates for model based analysis of vesicle pool dynamics (Neher and Taschenberger, 2021) will help to assess the value and impact of iGluSnFR in the analysis of transmission at calyceal synapses.” on line 381.

1. The deconvolution analysis assumes an instantaneous rise time. Yet previous work (Armbruster et al., 2020) that took into account diffusion, suggested potentially slower rise time dynamics. More importantly, the deconvolved waveforms do not match the shapes of the EPSCs (figures 5 and supp 6-2).What is the aim of the deconvolution? It was not clear from the text, but I assume it shows iGluSnFR binding to glutamate - in which case the slow waveforms are indicative of extrasynaptic iGluSnFR activation.

The deconvolution analysis was mainly used to recover the average responses to stimuli in the train without contamination by previous responses (see also Taschenberger et al. 2016, their figure 6).

We did also try to use the average singular response instead of the exponential fit as a kernel for the (Wiener) deconvolution analysis, which more closely resembled the observed (fast) rise. Unfortunately, this led to markedly worse results, likely because of the noise levels in the measurements. We believe that it would be beneficial to model the rise of the signal more precisely if glutamate imaging data is acquired at higher framerates.

The broad wave forms may be due to extrasynaptic binding of glutamate, but we also note that each frame corresponds to ~10ms and there is only ~10 data points between stimuli, so the responses are unlikely to be as sharp as eEPSCs.

However, I suppose that the more interesting question is whether iGluSnFR could be deconvolved to reveal the underlying release events, similar to how calcium signals can be used to inform about single action potentials.

We agree that it would be particularly interesting to use a "mini iGluSnFR" signal to deconvolve the resulting traces. Unfortunately, we failed to detect iGluSnFR signals reporting individual release events at this time, preventing this kind of analysis.

1. I suggest referencing and discussing (Aggarwal et al., 2022; Srivastava et al., 2022) . These highly relevant papers analyzed iGluSnFR to probe synaptic release.

References:

Aggarwal, A., Liu, R., Chen, Y., Ralowicz, A. J., Bergerson, S. J., Tomaska, F., Hanson, T. L., Hasseman, J. P., Reep, D., Tsegaye, G., Yao, P., Ji, X., Kloos, M., Walpita, D., Patel, R., Mohr, M. A., Tilberg, P. W., Mohar, B., Team, T. G. P., . . . Podgorski, K. (2022). Glutamate indicators with improved activation kinetics and localization for imaging synaptic transmission. bioRxiv, 2022.2002.2013.480251. https://doi.org/10.1101/2022.02.13.480251

Armbruster, M., Dulla, C. G., & Diamond, J. S. (2020). Effects of fluorescent glutamate indicators on neurotransmitter diffusion and uptake. Elife, 9. https://doi.org/10.7554/eLife.54441

Srivastava, P., de Rosenroll, G., Matsumoto, A., Michaels, T., Turple, Z., Jain, V., Sethuramanujam, S., Murphy-Baum, B. L., Yonehara, K., & Awatramani, G. B. (2022). Spatiotemporal properties of glutamate input support direction selectivity in the dendrites of retinal starburst amacrine cells. Elife, 11. https://doi.org/10.7554/eLife.81533

We thank the reviewer for the suggestions. Some of these studies were not available when we first drafted the manuscript. We now added a section discussing these studies starting on line 466:

Optimizing the imaging technique may reduce noise level, while the development of improved GEGIs could improve the signal to a level, at which spontaneous release events can be identified reliably in the cochlear nucleus. In retinal slices, where quantal events have been reliably observed with two-photon imaging, temporal deconvolution was successfully employed to estimate release rates from iGluSnFR signal (Srivastava et al., 2022; James et al., 2019). Subcellular targeting of iGluSnFR variants to the postsynaptic membrane may reduce measurement errors introduced by contributing extrasynaptic iGluSnFR signal and improve spatial resolution of glutamate imaging data(Hao et al., 2023; Aggarwal et al., 2022).

Referees cross-commenting

I also agree with the comments made by other reviewers!

Significance

Overall, this study addresses an important problem in basic neuroscience research. With the developing reliance on optical measurement of neuronal function, it is important to understand the impact of the indicators on physiological function and the limitations of the technique. The study is well-executed and will be informative to neuroscientists performing optical glutamate recording to study single-cell and circuit function in and beyond the auditory system.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

In the present manuscript, the authors explore the information that can be obtained using optical measurement of glutamate release with iGluSnFR on synaptic dynamics in the endbulb of Held.

They virally express iGluSnFR in presynaptic terminals, patch the postsynaptic cells and combine high-frame-rate optical recordings with electrophysiological measurements. Their first finding is that mEPCSs are prolonged when presynaptic cells express the glutamate indicator, which they interpret as buffering of extracellular glutamate by the indicator. Next, they repeated the experiment, this time with stimulating evoked EPSCs. In contrast to the previously observed effects, iGluSnFR did not affect the time course or the amplitude of the evoked EPSCs. The authors then asked whether iGluSnFR signals can be used to study synaptic dynamics, specifically, synaptic depression. In these experiments, the authors observed a change in the paired-pulse ratio with ISI of 10ms, but not longer intervals. They analyzed presynaptic release and did not find statistically significant differences. Can iGluSnFR signals be used for the analysis of synaptic release? When stimulated at a low frequency of 10Hz (allowing the fluorescence to return close to baseline levels in between pulses), iGluSnFR dynamics were somewhat comparable to postsynaptic signals. At higher frequencies, the slow time course of the indicator prevented the identification of individual responses and the resulting fluorescence had a very different shape. To resolve this problem, the authors used deconvolution analysis (fig 6). This analysis revealed a linear relationship between the optical readout and the patch-clamp data.

I find the manuscript to be clearly written, the findings are well presented and discussed and are novel and of substantial interest to neuroscientists in the field. I do have a number of questions about experiments and analysis that may have an effect on the conclusions of this work.

1. In experiments comparing the effects of iGluSnFR expression on release dynamics (figure 1-4), the authors compare infected presynaptic cells to control (uninfected). The assumption is that synaptic buffering by iGluSnFR may affect glutamate diffusion in the synaptic cleft. However, it is possible that viral infection itself changes presynaptic properties. The authors should compare release from cells infected with GFP or a comparable indicator.
2. Analysis of pool parameters presented in table 4 indicates almost doubling of RRP size with iGLuSnFR with 2 mM Ca++. While not significant, this result may indicate a real effect that may have been missed due to low power (N=3 and 7 for these experiments). I do not believe the authors did a power analysis in this study. How was the number of experiments determined? I would suggest increasing the number of experiments to avoid type II errors.
3. The deconvolution analysis assumes an instantaneous rise time. Yet previous work (Armbruster et al., 2020) that took into account diffusion, suggested potentially slower rise time dynamics. More importantly, the deconvolved waveforms do not match the shapes of the EPSCs (figures 5 and supp 6-2).What is the aim of the deconvolution? It was not clear from the text, but I assume it shows iGluSnFR binding to glutamate - in which case the slow waveforms are indicative of extrasynaptic iGluSnFR activation. However, I suppose that the more interesting question is whether iGluSnFR could be deconvolved to reveal the underlying release events, similar to how calcium signals can be used to inform about single action potentials.
4. I suggest referencing and discussing (Aggarwal et al., 2022; Srivastava et al., 2022) . These highly relevant papers analyzed iGluSnFR to probe synaptic release.

References:

Aggarwal, A., Liu, R., Chen, Y., Ralowicz, A. J., Bergerson, S. J., Tomaska, F., Hanson, T. L., Hasseman, J. P., Reep, D., Tsegaye, G., Yao, P., Ji, X., Kloos, M., Walpita, D., Patel, R., Mohr, M. A., Tilberg, P. W., Mohar, B., Team, T. G. P., . . . Podgorski, K. (2022). Glutamate indicators with improved activation kinetics and localization for imaging synaptic transmission. bioRxiv, 2022.2002.2013.480251. https://doi.org/10.1101/2022.02.13.480251

Armbruster, M., Dulla, C. G., & Diamond, J. S. (2020). Effects of fluorescent glutamate indicators on neurotransmitter diffusion and uptake. Elife, 9. https://doi.org/10.7554/eLife.54441

Srivastava, P., de Rosenroll, G., Matsumoto, A., Michaels, T., Turple, Z., Jain, V., Sethuramanujam, S., Murphy-Baum, B. L., Yonehara, K., & Awatramani, G. B. (2022). Spatiotemporal properties of glutamate input support direction selectivity in the dendrites of retinal starburst amacrine cells. Elife, 11. https://doi.org/10.7554/eLife.81533

Referees cross-commenting

I also agree with the comments made by other reviewers!

Significance

Overall, this study addresses an important problem in basic neuroscience research. With the developing reliance on optical measurement of neuronal function, it is important to understand the impact of the indicators on physiological function and the limitations of the technique. The study is well-executed and will be informative to neuroscientists performing optical glutamate recording to study single-cell and circuit function in and beyond the auditory system.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

In the manuscript 'Optical measurement of glutamate release robustly reports short-term plasticity at a fast central synapse' the authors present a careful analysis of whether direct measurements of transmitter release using the genetically-encoded indicator iGluSnFR, are suitable for assessing changes in transmitter release at the spiral ganglion neuron end bulbs of Held in the mouse cochlear nucleus. What sets this study apart from other studies, which have demonstrated the utility of iGluSnFR measurements, is the use of a camera-based fluorescence readout as opposed to confocal or 2P microscopy methods and that it is performed in the cochlear nucleus.

The primary methodology is the comparison of electrophysiological measurements of excitatory postsynaptic currents from bushy cells with fluorescence changes in the end bulbs of iGluSnFR expressing auditory nerve fibers with and without stimulation of the auditory nerve fibers. The experiments are technically demanding and introducing genetically encoded indicators in neurons of the cochlea is no small accomplishment. An important observation is that mEPSCs are slightly modified (prolonged) due to expression of iGluSnFR in the presynaptic end bulbs. This is perhaps not surprising as iGluSnFR binds glutamate and may act as a buffer to reduce the peak and slightly prolong the increase in cleft glutamate concentration after release from synaptic vesicles. To my knowledge, others have not reported iGluSnFR effects on mEPSCs. Perhaps earlier studies have not checked as carefully, alternatively previous studies had a too-low fraction of presynaptic terminals expressing iGluSnFR (or less expression of iGluSnFR) to detect a change in EPSC parameters, or this is a synapse-specific phenomenon. However, the authors demonstrate that EPSCs evoked by electrical stimulation of the auditory nerve fibers are unaffected by expression of iGluSnFR in the presynaptic neurons. Further findings are that the determined decay time constant is substantially longer than at other synapses (~16 ms at hippocampal synapses, Dürst et al., 2018). Synaptic depression was robustly reported by iGluSnFR at this synapse, but determination of single quantal events and thus quantal analysis was not really possible at this synapse using iGluSnFR in conjunction with the imaging and analysis techniques presented. The manuscript is carefully written and presented.

Major points:

1. The ROIs are selected to be 'outer bounds' of the glutamate spread from the synapses being studied. My concern is that these generously-sized ROIs include signal from many iGluSnFR molecules which are distal to the release sites and thus will be reached only slowly by low concentrations of glutamate or be contributing only noise and no changes in fluorescence. I suggest the temporal resolution could be improved by restricting the analysis of fluorescence changes to fewer pixels within the ROIs with the fastest rising/highest amplitude responses.
2. The observation that despite a 2 fold increase in eEPSCs when changing from 2 mM to 4 mM extracellular calcium there is no change in iGluSnFR peak is curious as pointed out by the authors but not really discussed. Are the traces presented in Figure 5 examples from the same recording? Assuming the examples are from one cell I first assumed the lack of change of peak was saturation of iGluSnFR but the larger fluorescence change with 100 Hz stimulation suggests otherwise. How many endbulbs are contacting one BC? Do you capture iGluSnFR responses from only one or several? In the previous point I suggested that restricting analysis to the soonest reacting pixels might improve temporal resolution but in the case of detecting the peaks with higher and normal calcium, these fastest reacting signals are probably also more likely to be saturated with glutamate.

Minor points:

• mEPSCs are usually recorded in tetrodotoxin, I didn't find any mention in methods/results
• the large numbers of abbreviations make it difficult in places to follow the manuscript please at least define them again in the figure legends (e.g. BC, AVCN in figure 1, Q, FWHM in figure 2 etc.)
• it is a bit unusual to report results of a Wilcoxon test and at the same time mean and SEM instead of medians, if different tests were used then it is important to indicate this where the p values are given or make the sentence in the methods more definitive
• the liquid junction potential is reported as 12 mV, pretty sure it should be -12 mV (unless the QX-314 or some other of the more exotic ingredients in the extracellular solution is having a dramatic effect on the LJP). I wonder if one of the faster/lower affinity iGluSnFR variants would be better suited for studying this synapse. The paper would benefit from a careful reading to shorten the text and to check for clarity. For instance page 15 line 436 I don't understand how 'the results can reduce the likelihood of biologically relevant changes'. I think the authors meant something different
• page 5 'width' is misspelled
• page 18 'strychnine' is misspelled
• on many of the figures is text that it much too small

Referees cross-commenting

I agree with all the comments of the other reviewers - both raise the point that there should be a 'control' AAV injected for comparison of the mEPSCs which I missed but is of course quite important. See https://pubmed.ncbi.nlm.nih.gov/24872574/ for a study of AAV serotype-dependent effects on presynaptic release.

Significance

The main audience for this paper will be fairly specialized. Researchers interested in properties of presynaptic release and some specialists in synaptic transmission in the auditory system will be the main readers/citers of this work.

The work is an important technical/methodological report. It highlights an important effect of expressing iGluSnFR and also demonstrates that the effect is overall not very problematic. Additional problems using iGluSnFR are also indicated.

I am an electrophysiologist, studying synaptic transmission and plasticity with experience using a wide range of optogenetic tools

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary

The authors used a novel imaging technique to monitor glutamate release and correlated these measurements with gold standard electrophysiological measurements. The genetically encoded glutamate reporter, iGluSnFR, was expressed in mouse spiral ganglion neurons using the approach described in Ozcete and Moser (2021, EMBO J). The iGluSnFR signals and the postsynaptic currents were measured at the endbulb of Held synapse. A small effect of the expression of iGluSnFR on the mEPSC kinetics was found (but see comment 1). Furthermore, deconvolution of the iGluSnFR signals was performed enabling the comparison of some presynaptic properties assessed with either iGluSnFR or electrophysiology.

Major comments

1. The central finding of the study is the prolonged decay time constant of the mEPSC. The difference is small but astonishingly significant (0.172 {plus minus} 0.002 and 0.158 {plus minus} 0.001, P=0.003). The SEM is about 100 times smaller than the measured time constant. This is biologically not plausible. Therefore, I am skeptical about the statistical significance of the results.
2. Analysis of the size of RRP with electrophysiology and iGluSnFR is potentially interesting but iGluSnFR recordings could not resolve the spontaneous fusion of single vesicles. Therefore, it is not possible to estimate RRP with these iGluSnFR recordings. This limitation of the approach should be emphasised more clearly.
3. The control conditions (no surgery/no virus injection) are not the correct conditions for comparison with the experimental conditions (surgery/virus injection and sensor expression). The control group should be operated and injected with saline or ideally with a virus expressing GFP at the extracellular membrane. The authors addressed this issue by citing their previous work (Özcete and Moser, 2021). However, I am not convinced that surgery does not induce subtle changes that could explain the small differences in mEPSCs.

Minor comments

The supplementary figures are not listed in the order in which they appear in the main text.

Figure 2B and 3 are not referenced in the main text.

The PPR in Figure 3 shows a PPR that cannot be evaluated because of the unusual plot with lines that are too thick.

Line 105: "...while simultaneously monitoring currents in postsynaptic cells". This sentence is not correct given that the EPSCs have not been shown yet at this point of the manuscript.

Line 110: "SV and are not cause are cause by spontaneous action potentials...". The sentence does not make sense.

Line 168-9: "...we did not find significant differences in amplitude and kinetics...". According to Table 2 (2mM Ca2+ condition), both Imax and Q appear to be almost twice as high in iGluSnFR as in control (2.05 {plus minus} 0.06 and 1.34 {plus minus} 0.03, respectively; P=0.241). Is this not a significant difference?

Table 4. 2mM Ca2+ condition. The Rrefill parameter is about an order of magnitude smaller in the iGluSnFR-expressing group. Is this correct or just a typo?

Referees cross-commenting

I also agree with the comments of the other reviewers.

Significance

General assessment This topic is currently of interest because iGluSnFR techniques are widely used. However, the study is preliminary. The scientific progress in terms of quantity and quality is limited. For example, Figs. 1 and 5 show only images and traces with little scientific significance.

Advance The main advance of the study is the implementation of the deconvolution of the iGluSnFR signal and the comparison of the back extrapolation with the first stimulus (Fig. 6). This comparison was similar between electrophysiology and iGluSnFR when deconvolution of the iGluSnFR data was performed. These data therefore argue against saturation of iGluSnFR, as expected from a large number of previous analyses of iGluSnFR.

There is little methodological improvements compared with the group's previous study (Ozcete and Moser, 2021 EMBO J). In this earlier study, a different synapse was analyzed but the same iGluSnFR was injected into the scala tympani of the right ear through the round window in the same way as in this study. Surprisingly, the authors do not refer to Ozcete and Moser (2021) in the relevant methods section.

Audience „basic research"

My field of expertise imaging and electrophysiology; basic research

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

Learn more at Review Commons

---

Reply to the reviewers

Manuscript number: RC-2022-01586

Corresponding author(s): Hammond, Gerald

1. General Statements

Our manuscript details a novel homeostatic feedback loop for the master plasma membrane regulatory molecule, PI(4,5)P2. In this loop, the PIP4K family of PI(4,5)P2-synthesizing enzymes act in a novel, non-enzymatic capacity: they sense PI(4,5)P2 levels and directly inhibit the lipid’s synthesis by inhibiting the major enzyme involved in the terminal step of synthesis, PIP5K. The three reviewers seem largely convinced of our data, and provided detailed, insightful and plausible suggestions for revision, which we have now comprehensively provided. This includes substantial new experimental work, including the generation of genomically tagged cell lines to localize all endogenous PIP4K isoforms.

However, all three reviewers questioned the paper’s novelty and significance based on recent studies in the literature demonstrating PIP5K inhibition by PIP4Ks [refs 25 & 53 in the manuscript]. We feel that this is an inaccurate and somewhat unfair assessment of our findings, since it does not consider our central (and completely unprecedented) finding that PIP4Ks directly sense PI(4,5)P2 levels through low-affinity binding. As well as being a novel finding, this places the previously observed inhibition of PIP5K by PIP4Ks into a completely new paradigm consisting of a complete, enclosed homeostatic feedback loop. This was not demonstrated previously in the literature.

Of course, the reviewers’ convergent opinions almost certainly reflect a deficit in our articulation of the novel findings in the original manuscript. We have therefore revised the current version to more clearly emphasize our novel findings.

2. Point-by-point description of the revisions

Reviewer #1

__Summary: __In this manuscript, authors address how PIP4K regulates tonic plasma membrane (PM) PI(4,5)P2 levels which are generated by major PI(4,5)P2 synthesis enzyme, PIP5K by using PIP4K and PIP5K overexpressing cells or acutely manipulating PM PI(4,5)P2 levels by the chemically induced dimerization (CID) system. Additionally, authors assessed effect of direct interaction between PIP4K and PIP5K by using supported lipid bilayers (SLBs) and purified PIP4K and 5K. Authors also were successful in monitoring dynamics of endogenous PIP4K by using a split fluorescent protein approach. Through this study, authors propose a model of PI(4,5)P2 homeostatic mechanism that PIP4Ks sense elevated PM PI(4,5)P2 by PIP5Ks, are recruited to the PM, and bind to PIP5Ks to inhibit PIP5Ks activity.

# 1.1: Although authors mention methods of statistical analysis in materials and methods, they did not present the results of statistical analysis in the figures. The quantitative data should be presented with statistical analysis data, which is important for showing where convincing differences between treatment groups are found.

We agree that statistics are important to fully interpret the data; we have now included the results of statistical tests (non-parametric statistics were used, as the data are not normally distributed) with correction for multiple comparisons. Significant changes are denoted using asterisk notation in figs. 1A-C, 2B, 5B & 7A. The full results are now reported as tables:

Fig 1A = table 1; Fig 1B = table 2; Fig. 1C = table 3; Fig. 2B = table 4; Fig. 5B = table 5; Fig 7A = tables 6 & 7.

__#1.2a: __Fig. 1D. Fig. 1D and Fig. 3A should be presented together because these are exactly same set of cells and information of each PIP4K and PIP5K membrane localization could be important for understanding mechanisms of inhibitory effect of PIP4Ks.

We struggled when writing the manuscript to reconcile these data into a single figure. The manuscript flows from showing inhibition of PIP5Ks by PIP4Ks in living cells (figs. 1 & 2), then showing low affinity PI(4,5)P2 binding by endogenous PIP4Ks (figs. 3-6) and finally to a direct interaction between PIP4K and PIP5K (fig. 7). We therefore felt that reconciling the data showing attenuated PI(4,5)P2 synthesis with the interaction between PIP4Ks and PIP5Ks, despite being demonstrated in the same experiments, would disrupt the flow of the paper. We therefore request to leave the data in Figs. 2B and 7A, whilst remaining explicit that the data derive from a single experiment.

#1.2b: Authors claimed that over-expression of all three PIP4K isoforms were able to attenuate the elevated PM PI(4,5)P2 levels caused by PIP5K over-expression. However, in Fig. 3A, PIP4K2A was recruited to PM by both PIP5K1A and PIP5K1C but looks only attenuated PIP5K1A, but not PIP5K1C, overexpression mediated PM PI(4,5)P2 elevation (Fig. 1D). PIP4K2C was less recruited to the PM than PIP4K2A and 2B in PIP5K1A overexpressing cell (Fig. 3A) but PIP4K2A, B and C isoforms equally attenuated increase of PM PI(4,5)P2 in PIP5K1A overexpressing cell (Fig. 1D). It is likely that efficiency of inhibitory effect of each PIP4K isoform is different by co-overexpressed PIP5K isoform. These images should be more carefully documented with Fig. 1D and Fig. 3A together.

As the reviewer suggests, we have now expanded our description of these data in both results and discussion; firstly, for the attenuating effects on PI(4,5)P2 synthesis, we write on the 3rd paragraph of p4: “We also reasoned that co-expression of PIP4K paralogs with PIP5K might attenuate the elevated PI(4,5)P2 levels induced by the latter. Broadly speaking, this was true, but with some curious paralog selectivity (fig. 2B, statistics reported in table 4): PIP4K2A and PIP4K2B both attenuated PI(4,5)P2 elevated by PIP5K1A and B, but not (or much less so) PIP5K1C; PIP4K2C, on the other hand, attenuated PIP5K1A and was the only paralog to significantly attenuate PIP5K1C’s effect, yet it did not attenuate PIP5K1B at all.”

On the relative ability of PIP5Ks to localize PIP4Ks we focus on the key result, writing on the 2nd paragraph of p7: “When co-expressing EGFP-tagged PIP5Ks and TagBFP2-tagged PIP4K2s, we found that PIP5K paralogs’ PM binding is largely unaffected by PIP4K over-expression (fig. 7A, upper panel and table 6), whereas all three paralogs of PIP4K are strongly recruited to the PM by co-expression of any PIP5K (fig. 7A, lower panel and table 7)…”

And finally, we describe a more nuanced discussion of the possible implications for differential inhibition of PIP5K isoforms by PIP4Ks in the discussion, starting in the first paragraph on p. 11: “Despite minor differences in the ability of over-expressed PIP5K paralogs to recruit over-expressed PIP4K enzymes (fig. 7A), we observed major differences in the ability of PIP4K paralogs to inhibit PI(4,5)P2 synthesis when over-expressed alone (fig. 1C) or in combination with PIP5K (fig. 2B). It is unclear what drives the partially overlapping inhibitory activity, where each PIP5K paralog can be attenuated by 2 or 3 PIP4Ks. This is however reminiscent of the biology of the PIPKs, where there is a high degree of redundancy among them, with few unique physiological functions assigned to specific paralogs [49]. There may be hints of paralog-specific functions in our data; for example, enhanced PI(4,5)P2 induced by over-expressed PIP5K1C is only really attenuated by PIP4K2C (fig. 2B). This could imply a requirement for PIP4K2C in regulating PI(4,5)P2 levels during PLC-mediated signaling, given the unique requirements for PIP5K1C in this process [50,51]. Regardless, a full understanding of paralog selectivity will need to be driven by a detailed structural analysis of the interaction between PIP4Ks and PIP5Ks - which is not immediately apparent from their known crystal structures, especially since PIP4Ks and PIP5Ks employ separate and distinct dimerization interfaces [49].”

#1.3: Fig. 1F. It seems that PIP4K2A accelerated PIP5K, but not Mss4, dependent PI(4,5)P2 generation before PI(4,5)P2 reaches 28,000 lipids/um2. Is this significant? If so, why did this happen?

We have answered this question with a sentence added to the 1st paragraph on p 8: *“The ability of PIP4K to bind to PIP5K on a PI(4,5)P2-containing bilayer also potentially explains the slightly accelerated initial rate of PI(4,5)P2synthesis exhibited by PIP5K1A that we reported in fig. 2C, since PIP4K may initially introduce some avidity to the membrane interaction by PIP5K, before PI(4,5)P2 reaches a sufficient concentration that PIP4K-mediated inhibition is effective.” *

#1.4: Fig. 3B. In this figure, authors only presented images after Rapa treatment. Therefore, it is not clear what these results mean. Before Rapa treatment, where did bait proteins and NG2-PIP4K2C localize? If ePIP4K2C delta PM intensity (ER:PM/PM) increase, does that mean increase in ER:PM intensity or decrease in PM intensity? According to Figure legend, PI(4,5)P2 indicator TubbycR332H was co-transfected, but those images are not shown in the figure. Images of PI(4,5)P2 indicator also should be presented to show whether after Rapa treatment PI(4,5)P2 increased at ER-PM contact sites, because that could be critical for the conclusion that "The use of Mss4 ruled out an effect of enhanced PI(4,5)P2 generation at contact sites, since this enzyme increases PI(4,5)P2 as potently as PIP5K1A (Fig. 1A), yet does not cause recruitment of PIP4K2C". Is this conclusion consistent with Fig. 2F and G?

These data now appear in Fig. 7B. We have added images showing the pre-rapamycin state to the revised figure. The reference to tubby­cR332H co-expression was an error. In fact, the cells expressed the ER:PM contact site marker MAPPER, which allowed us to quantify ER:PM contact site localization before and after rapamycin induced capture of the baits at these sites. The revised figure appears as follows:

The failure of Mss4 to recruit endogenous PIP4K2C is entirely consistent with the old Fig. 2F and G (now 5A and C), since these show PIP4K interaction with PI(4,5)P2 containing lipid bilayers (in Fig. 5C, the PI(4,5)P2 was synthesized by Mss4). We demonstrated that Mss4 is unable to interact with PIP4K2A in Fig. 7D.

#1.5: Fig. 3C and D. Based on results of Fig. 3C and D, authors concluded that "PIP4K2C binding to PI(4,5)P2-containing SLBs was greatly enhanced by addition of PIP5K to the membranes, but not Mss4". I don't think Fig. 3C and D are comparable because experimental conditions are different. While lipid composition of SLB used in Fig. 3C was 2% PI(4,5)P2, 98% DOPC, in Fig. 3D, it was 4% PI(4,5)P2, 96% DOPC. And also, in Fig. 3C, PIP5K1A was added to SLB at the time about 50 sec, whereas in Fig. 3D, Mss4 was added at 600 sec. It seems that in Fig. 3D, PIP4K2A was already saturated on SLB before adding Mss4. These two experiments must be performed under the same conditions.

We have repeated these experiments (which now appear in Fig. 7C & D) under identical conditions, with the same result.

---

#1.6: Overall results discussed in the text are very compressed referring readers to the 4 multi-panel complex figures with elaborate figure legends. While it is possible to figure out what the authors' studies and results are, it is quite a laborious process.

We have revised the manuscript to be less compressed and easier to read, with the data now organized as eight figures and the results section split into four sub-sections.

Minor comments:

#1.7: Fig. 2D. The purified 5-phosphatase used in Fig. 2D is INPP5E but described in figure legend and materials and methods ass OCRL. Which one is correct?

Purified OCRL was indeed used in the supported lipid bilayer experiments. The figure (now Fig. 4A) and legend have been corrected – thank you for spotting the error.

#1.8: Fig. 3B. Indicate which trace represents PIP5K1A, Lyn11 or Mss4.

The data now appears in Fig. 7B, with the traces separated into separate graphs for greater clarity (see response to #1.4).

#1.9: Fig. 4C. X-axis label. Is "Time (min)" correct? Or should it be "Time (sec)".

Thank you for spotting this typo. It should have indeed been seconds, and this is corrected in the new fig. 8C.

• *

Reviewer #1 (Significance (Required)): The finding that PIP4K itself is a low-affinity PI(4,5)P2 binding protein and sense increases of PM PI(4,5)P2 generated by PIP5K to control tonic PI(4,5)P2 levels by inhibiting PIP5K activity is a novel concept. However, inhibition of PIP5K by PIP4K and importance of the inhibitory effect of PIP4K in PI3K signaling pathway have previously been reported (ref 24). This reduces the novelty of the current work somewhat however, the authors do provide evidence for dual interactions of PIP4K (PIP2, PIP5K), which the previous report did not.

We appreciate the reviewer’s insightful comments and overall appreciation of our work. We agree that previous studies did not detect the dual interaction of PIP4Ks with PIP5Ks and PI(4,5)P2; as we argue strongly in the general comments, we think this actually fits as a complete, enclosed homeostatic feedback loop – which is a significant and novel finding.

• *

Reviewer #2

Summary: This paper proposes that the enzyme PIP4K2C is a negative regulator of the synthesis of PI(4,5)P2 and that it does so by dampening the activity of PIP5K which is the enzymatic activity responsible for producing the major pool of PI(4,5)P2 in cells.

• *

Reviewer #2 (Significance (Required)): Although the findings of the paper are presented as a major new advance, the observation that PIP4K might acts as a negative regulator of PIP2 synthesis has been previously presented in two previous publications. The significance of this paper is that it also shows the same point in another model system.

PIP4Ks Suppress Insulin Signaling through a Catalytic-Independent Mechanism

Diana G Wang 1, Marcia N Paddock 2, Mark R Lundquist 3, Janet Y Sun 3, Oksana Mashadova 3, Solomon Amadiume 3, Timothy W Bumpus 4, Cindy Hodakoski 3, Benjamin D Hopkins 3, Matthew Fine 3, Amanda Hill 3, T Jonathan Yang 5, Jeremy M Baskin 4, Lukas E Dow 6, Lewis C Cantley 7

PMID: 31091439; PMCID: PMC6619495;DOI: 10.1016/j.celrep.2019.04.070

and

Phosphatidylinositol 5 Phosphate 4-Kinase Regulates Plasma-Membrane PIP3 Turnover and Insulin Signaling.

Sharma S, Mathre S, Ramya V, Shinde D, Raghu P.Cell Rep. 2019 May 14;27(7):1979-1990.e7. doi: 10.1016/j.celrep.2019.04.084.PMID: 31091438

Both of these studies show that in cells lacking PIP4K, during signalling the levels of PIP2 rise much greater than in wild type cells. Indeed the Cantley lab paper (Wang et.al) have shown that this is likely due to an increase in PIP5K activity, using an in vitro assay. They have further disrupted the interaction between PIP4K and PIP5K and demonstrated the importance of this interaction in the enhanced levels of PIP2.

Respectfully, we disagree with this assessment, because we believe it doesn’t consider the novel, central findings we report: that PIP4Ks sense PI(4,5)P2 levels through direct interaction with the lipid, and that this is what facilitates PIP5K inhibition. These findings were not reported in the prior studies. Nonetheless, the studies are foundational for ours and were cited in our original manuscript (and are still, as refs 25 and 53).

• *

#2.1: Likewise although the authors have claimed that no mechanisms have claimed that there are no mechanisms reported to sense and downregulate PIP2 resynthesis. It is suggested that they read and consider the following recent paper which studies Pip2 resynthesis during GPCR triggered PLC signalling.

Kumari A, Ghosh A, Kolay S and Raghu P*. Septins tune lipid kinase activity and PI(4,5)P 2 turnover during G-protein-coupled PLC signalling in vivo. Life Sci Alliance. 2022 Mar 11;5(6):e202101293. doi: 10.26508/lsa.202101293. Print 2022 Jun.

We have now included a full discussion of this paper in the discussion starting on the last paragraph of p 9: “Since this paper was initially submitted for publication, another study has reported a similar homeostatic feedback loop in Drosophila photoreceptors, utilizing the fly homologue of septin 7 as the receptor and control center [38]. This conclusion is based on the observation that cells with reduced septin 7 levels have enhanced PIP5K activity in lysates, and exhibit more rapid PI(4,5)P2 resynthesis after PLC activation. However, changes in septin 7 membrane localization in response to acute alterations in PI(4,5)P2 levels, as well as direct interactions between PIP5K and septin 7, have yet to be demonstrated. Nevertheless, septin 7 has distinct properties as a potential homeostatic mediator; as a foundational member of the septin family, it is essential for generating all major types of septin filament [39]. Therefore, a null allele for this subunit is expected to reduce the prevalence of the septin cytoskeleton by half. Given that septin subunits are found in mammalian cells at high copy number, around ~106 each [29], and the fact that septins bind PI4P and PI(4,5)P2 [40,41], it is likely that septin filaments sequester a significant fraction of the PM PI4P and PI(4,5)P2 through high-avidity interactions. In addition, membrane-bound septins appear to be effective diffusion barriers to PI(4,5)P2 and other lipids [42]. We therefore speculate that septins may play a unique role in systems such as the fly photoreceptor with extremely high levels of PLC-mediated PI(4,5)P2 turnover: The septin cytoskeleton can act as a significant buffer for PI4P and PI(4,5)P2 in such systems, as well as corralling pools of the lipids for use at the rhabdomeres were the high rate of turnover occurs. This is in contrast to the role played by the PIP4Ks, where PI(4,5)P2 levels are held in a narrow range under conditions of more limited turnover, as found in most cells.”

__#2.2: __Likewise there are other earlier papers in the literature which have studied possible PIP2 binding proteins as sensors for this lipid.

We are only aware of a single, specific example of a similar negative feedback, which is discussed in the 3rdparagraph of p 10: “Curiously, although phosphatidylinositol phosphate kinases are found throughout eukarya, PIP4Ks are limited to holozoa (animals and closely related unicellular organisms) [47]. Indeed, we found the PIP5K from the fission yeast, Saccharomyces cerevisiae, does not interact with human PIP4Ks (fig. 7) and cannot modulate PI(4,5)P2 levels in human cells without its catalytic activity (fig. 1). This begs the question: how do S. cerevisiae regulate their own PI(4,5)P2 levels? Intriguingly, they seem to have a paralogous homeostatic mechanism: the dual PH domain containing protein Opy1 serves as receptor and control center [48], in an analogous role to PIP4K. Since there is no mammalian homolog of Opy1, this homeostatic mechanism appears to have appeared at least twice through convergent evolution. Combined with hints of a role for septins in maintaining PI(4,5)P2 levels [38], the possibility arises that there may yet be more feedback controls of PI(4,5)P2 levels to be discovered.”

• *

Technical standards: The work is done to a high technical standard.

#2.3: Does catalytically dead isoform of PIP4K2B and 2C also yield the same result as a catalytically dead version of PIP4K2A in Fig 1B?

In a word: yes. We have added these experiments, which are now presented in Fig. 2A:

The results are described in the results in the 2nd paragraph of p. 4: “To directly test for negative regulation of PIP5K activity by PIP4K in cells, we wanted to assay PI(4,5)P2 levels after acute membrane recruitment of normally cytosolic PIP4K paralogs. To this end, we triggered rapid PM recruitment of cytosolic, FKBP-tagged PIP4K by chemically induced dimerization (CID) with a membrane targeted FRB domain, using rapamycin [27]. As shown in fig. 2A, all three paralogs of PIP4K induce a steady decline in PM PI(4,5)P2 levels within minutes of PM recruitment. Catalytically inactive mutants of all three paralogs produce identical responses (fig. 2A).”

#2.4: The labelling on y-axis for PI(4,5)P2 biosensor intensity ratio is PM/cell at some places, PM/Cyt or PM/Cyto in some places. It is recommended to make it uniform across all the panels.

PM/Cyto was a typo, now corrected to PM/Cyt. PM/Cell and PM/Cyt are two subtly different metrics used to normalize PM fluorescence intensity across varying transient expression levels. This is clarified in the methods in the 3rdparagraph on p.22: “For confocal images, the ratio of fluorescence intensity between specific compartments was analyzed as described previously [59]. In brief, a custom macro was used to generate a compartment of interest specific binary mask through à trous wavelet decomposition[68]. This mask was applied to measure the fluorescence intensity within the given compartment while normalizing to the mean pixel intensity in the ROI. ROI corresponded to the whole cell (denoted PM/Cell ratio) or a region of cytosol (PM/Cyt), as indicated on the y axis of individual figures.”

#2.5: The claim that PI(4,5)P2 production is sufficient to recruit PIP4K2C to the PM can be ascertained further if one is able to do an experiment where PI(4,5)P2 is ectopically expressed in some compartment of the cell which is non-native to PI(4,5)P2 and as a consequence of this PIP4K2C is recruited to this non-native compartment.

We have now removed the assertion that PI(4,5)P2 is sufficient to localize PIP4Ks to the membrane, since our conclusion is that the coincident presence of PI(4,5)P2 and PIP5Ks in the PM is what ultimately localizes the PIP4Ks. We did not detect recruitment of endogenous PIP4Ks to lysosomes when ectopic PI(4,5)P2 synthesis was induced, although fluorescence levels are so low as to be inconclusive, and therefore not appropriate for inclusion in the manuscript.

#2.6: In the entire figure 2, to establish that PI(4,5)P2 is necessary and sufficient for PM localisation of PIP4K, PIP4K2C is used as the PIP4K isoform on the basis that it is highly abundant in HEK293 cells. But PIP4K2A is localised mainly at the plasma membrane and here we are discussing about PI(4,5)P2 regulation at the PM . Can experiments be done with isoforms 2A and 2B as well? Can acute depletion of PI(4,5)P2 lead to the membrane dissociation of the isoform 2A as well? This will help us in understanding if there is an isoform specific difference in sensing PI(4,5)P2 levels which will help us in targeting specific isoform as therapeutic targets.

We have now generated endogenously tagged PIP4K2A and PIP4K2B; these cell lines are characterized in the revised fig. 3:

With the dependence on PI(4,5)P2 for PM binding for all isoforms shown in fig. 4:

32 cells that were imaged across three independent experiments. (E) Depletion of PI(4,5)P2 causes NG2-PIP4K2C to dissociate from the membrane. As in C, NG2-PIP4K2C (blue) cells were transfected with FKBP-tagged proteins, TubbyC (orange) and Lyn11-FRB, scale bar is 2.5 µm; cells were stimulated with 1µM rapa, as indicated. TubbyC traces represent mean change in fluorescence intensity (Ft/Fpre) ± s.e. The NG2-PIP4K2C traces represent the mean change in puncta per µm2 ± s.e. of > 38 cells that were imaged across three independent experiments. " v:shapes="Text_x0020_Box_x0020_5">

And increased binding by elevated PI(4,5)P2 levels shown in fig. 5B:

The results are described in the accompanying results text “PIP4K are low affinity sensors of PM PI(4,5)P2”, pp.4-7. In short, endogenous PIP4K isoforms behave similarly with respect to PI(4,5)P2-dependent PM recruitment.

• *

#2.7: In Figure 1A, it is shown that overexpression of a catalytically dead PIP5K 1A/1B/1C is still able to increase PI(4,5)P2 levels. In the figure 2E, expression of homodimeric mutant of PIP5K domain which is a way to increase catalytic activity of PIP5K, increases PI(4,5)P2 levels which is consistent with the inferences from Fig, 1 , but what is surprising is a catalytically dead variant not being able to do so. Why is there a discrepancy between Fig. 1A and Fig. 2E? If the homodimeric mutant is the reason, then it is not clear in the explanation.

We have added the following clarification to the results on the second paragraph of p.6: “We next tested for rapid binding to acutely increasing PI(4,5)P2 levels in living cells, using CID of a homodimeric mutant PIP5K domain (PIP5K-HD), which can only dimerize with itself and not endogenous PIP5K paralogs [34]. This domain also lacks two basic residues that are crucial for membrane binding [35], and only elevates PM PI(4,5)P2 when it retains catalytic activity (fig. 5D), unlike the full-length protein (fig. 1A).” We currently do not fully understand why these well characterized residues of PIP5Ks are necessary for PM binding and inhibition by PIP4K. This is a focus of ongoing studies in the lab for the structural basis of PIP5K inhibition by PIP4K.

• *

#2.8: Show the loading control in Fig 2A western.

We have added the loading control using alpha tubulin in the revised fig. 3B.

---

#2.9: In the figure 2D, in the legend OCRL is written. So, the labelling in the panel should also be changed to OCRL from INPP5E. It is intermixed.

Reviewer 1 also spotted this inconsistency (#1.7): Purified OCRL was indeed in the supported lipid bilayer experiments. The figure (now Fig. 4A) and legend have been corrected – thank you for spotting the error.

• *

#2.10: In the figure 2E, can the labelling be changed from HD to something more self-explanatory for homodimeric mutant of PIP5K domain?

We prefer to keep the “HD” notation in the revised figure 5D for brevity’s sake, but now define the abbreviation in the text in the second paragraph of p.6: “…a homodimeric mutant PIP5K domain (PIP5K-HD)…”.

#2.11: In Fig. 2E, PIP5K expression is acute and in Fig. 2F Mss4 expression is chronic, both of which is able to recruit PIP4K2C to the plasma membrane. How can a likewise argument be drawn out of these two experiments when one is acute and the other one is a chronic expression? It is suggested to do an FRB-FKBP experiment for Mss4 as well.

We agree with the reviewer that an FKBP-Mss4 would have been an excellent experiment. As can be seen from Fig. __1A, Mss4 is constitutively PM localized in mammalian cells. However, we were unable to identify a truncation of Mss4 that lost constitutive membrane binding whilst retaining catalytic activity. Therefore, we could only perform chronic overexpression as shown in __fig. 5B. The lack of an acute demonstration is why we went on to develop the PIP5K-HD constructs, results of which are reported in __fig. 5D. __

#2.12: In the text, Fig. 2G and 2H is written for PIP4K2C, but in the corresponding panels and legends, it is an assay for purified PIP4K2A on SLBs. Kindly resolve the discrepancy.

We thank the reviewer for spotting this discrepancy. PIP4K2A is the protein that was used in the SLB experiments now reported in fig. 5A & C and the accompanying results on pp.5-6. This is now corrected in the manuscript.

#2.13: Kindly explain a bit in detail why the baits were now targeted to ER-PM contact sites. It is not self-explanatory.

We have now added a more detailed description to the third paragraph of p. 7: “We therefore sought to distinguish between a direct PIP5K-PIP4K binding interaction versus PI(4,5)P2-induced co-enrichment on the PM. To this end, we devised an experiment whereby a bait protein (either PIP5K or control proteins) could be acutely localized to subdomains of the PM, with the same PI(4,5)P2 concentration. This was achieved using CID of baits with an endoplasmic reticulum (ER) tethered protein, causing restricted localization of the bait protein to ER-PM contact sites – a subdomain of the PM (fig. 7B).”

• *

#2.14: The conclusions for Fig. 3 most likely hints towards the possibility of PIP4K and PIP5K interaction being independent of PI(4,5)P2 levels. Well, Fig. 3C and 3D does suggest a direct interaction, but can other protein-protein interaction assays be used to establish the direct interaction of PIP4K with PIP5K such as FRET or Yeast two hybrid as assays scoring for interaction?

We respectfully diverge from the reviewer’s assessment of the data, presented in the revised fig. 7. Figs. 7A & B__show PIP4K and PIP5K interacting in the context of a PI(4,5)P2 replete PM; __fig. 7C shows this in the context of a PI(4,5)P2 replete SLB. Therefore, we make no assertion that the PIP4K/PIP5K is independent of PI(4,5)P2 levels. We also contend that the latter experiment is a more direct demonstration than a Y2H assay, or even FRET (which can occur among non-interacting proteins localized to a membrane surface, see e.g. 10.1074/jbc.m007194200).

#2.15: Conceptually a direct interaction can be explained to some extent from Fig. 3 but extending it to be an inhibitory interaction is not right without a direct experiment. Can an experiment be done with PI4P enriched SLB, wherein you put just PIP5K purified protein vs PIP5K+PIP4K combination and measure the % mol of PI(4,5)P2 produced using a probe. That will be suggestive of a negative interaction.

This is a great experiment, the results of which are reported in fig. 2C, described in the third full paragraph of p. 4: “To more directly examine inhibition of PIP5K by PIP4K, we tested activity of purified PIP5K1A on PI4P-containing supported lipid bilayers (SLBs). Addition of PIP4K2A exhibited delayed inhibition of PIP5K1A activity (fig. 2C): Once PI(4,5)P2 reached approximately 28,000 lipids/µm2 (~2 mol %), PIP5K dependent lipid phosphorylation slowed down, which doubled the reaction completion time (fig. 2C, right). In contrast, we observed no PIP4K dependent inhibition of Mss4 (fig. 2C, inset). These data recapitulate the prior finding that PIP4K only inhibited purified PIP5K in the presence of bilayer-presented substrate [25]. We therefore hypothesized that inhibition of PIP5K by PIP4K requires recruitment of the latter enzyme to the PM by PI(4,5)P2 itself.”

• *

__#2.15: __ In Figure 3B, the FRB tagged constructs are magenta coded and PIP4K2C is cyan. Kindly change the labelling of the FRB constructs on the y axis to magenta so that it goes with what is written in the legend. It will also be appreciated to show a colocalization quantification between the magenta (FRB constructs) and cyan (PIP4K2C) post rapamycin addition and not just the intensity for ER-PM recruited PIP4K2C.

These modifications and some additional points have been added in response to reviewer 1’s #1.4 to the revised fig. 7B. Note, we quantified the co-localization with an ER-PM contact site marker, MAPPER. Co-localization with the FRB-tagged construct would be misleading, because this construct is localized across the membrane at the start of the experiment and would thus have a high degree of co-localization. As can be seen from the inset graphs in the new analysis, however, all FRB-tagged constructs co-localize with MAPPER after rapamycin addition, but only FRB-PIP5K1A causes endogenous PIP4K2C to increase co-localization with this compartment.

# 2.16: Again, in the text , the description is written for PIP4K2C but in the result panel and legend (Fig. 3C and Fig. 3D), PIP4K2A is mentioned. Kindly resolve the discrepancy

We have corrected the results text on the last paragraph of p. 7: “Finally, we also demonstrate that PIP4K2A binding to PI(4,5)P2-containing supported lipid bilayers was greatly enhanced by addition of PIP5K to the membranes (fig. 7C), but not by Mss4 (fig. 7D).”

• *

# 2.17: In the Fig. 4B, it will be appreciated to show statistical significance in terms of R2 value for commenting on the linear response.

“Linear response” was not the best description of what we were trying to articulate in the revised fig. 8B; we have now amended the results in the 2nd paragraph of p.8 to read: “Of these, Tubbyc showed the largest degree of change in PM localization across all changes in PI(4,5)P2 levels (fig. 8B).”

• *

#2.18: Discussion can be in general a bit more detailed which is suggestive of future experiments to do that can shed more light on the interaction such as which residues in PIP4K interacts with PIP5K to negatively regulate it.

The revised manuscript contains a greatly expanded discussion, as described in detail in our responses to comments #1.2b, #2.1 and __#2.2. __

#2.19: In the discussion, more light can be shed on the fact that Mss4 in spite of being a 5- kinase is not negatively regulated by PIP4K and the fact that PIP4K is present only in metazoans suggests that this fine tuning of PI(4,5)P2 levels is specific to metazoans. Another insight could be in the direction, that Fig 4. tells PI3K, but not calcium signaling is modulated by this fine tuning and interestingly class I PI3K is also an enzyme specific to metazoans. Hence, unlike yeast, metazoans rely on growth factor signalling processes, hence regulation of PI(4,5)P2 by PIP4K and hence Class I PI3K and PI(3,4,5)P3 could be a process relevant to metazoans.

We have addressed the restriction of PIP4K to holozoa as described in our response to #2.2, wherein we describe a previously proposed paralogous mechanism in fungi. The reviewer’s point about the homeostatic process being related to class I PI3K signaling in growth control of multicellular organisms is interesting, but the presence of the PIP4Ks in some unicellular organisms complicates this view. We are of the view that a discussion of this important topic is a little nuanced for inclusion in the current manuscript.

• *

Reviewer #3

__Summary: __Using state of the art imaging techniques the authors try to address how cells sense PI(4,5)P2 levels and regulate PIP5Ks to maintain an optimal level since any dysregulation of PI(4,5)P2 levels can have significant effects on the functioning of the cell and led to numerous disease states, such as cancers.

The key conclusions are convincing and importantly validate previous disputed findings made by Wang et al. (Cell Reports 2019) using different and more rigorous methods, however unfortunately due to the Wang et al publication the overall novelty of this study is lacking. A suggestion to the authors is to state/explain with text more clearly how their findings are more precise and higher quality than the previous report and why their findings are necessary and significant to drive the field forward.

We have revised the manuscript to more clearly state our novel finding that PIP4Ks are PI(4,5)P2 sensing proteins that inhibit PIP5Ks on the membrane in a PI(4,5)P2-dependent manner, which was not previously described in the literature.

Further, experiments in the study were performed in vitro in cultured cells using overexpression methods making the physiological significance a bit unclear and the enthusiasm of the main discovery dampened. With that being said these findings are worthy of publication in order to advance the field and understanding of how the PIP kinase families are regulated and maintain PIP2 homeostasis which is important for life.

We feel that this assessment is slightly unfair, since most of the key experiments have been validated using purified proteins in supported lipid bilayers, and endogenous proteins were studied using genomic tagging approaches, rather than over-expression.

Minor and easily addressable experiments should be performed by the authors the following. Further, many of these experimental issues can easily go in supplemental materials

#3.1: Include western blots for the constructs to compare expression levels.

We agree that it is important to take into account differences in expression levels for the experiments presented in fig. 1. However, since these are single cell assays, Western blotting of whole populations of transiently transfected cells is not the best control. Instead, having acquired the images under consistent excitation and detection parameters, we compared the fluorescence intensity, expressed as relative expression in Fig. 1A and C, which is discussed in the results text in the first two paragraphs of the results on p. 3: “Notably, expression of the catalytically inactive mutants was usually somewhat less strong compared to the wild-type enzymes, yet effects on PI(4,5)P2 levels were similar (fig. 1A).” and “Again, differences in expression level between isoforms do not explain differences in activity, since all achieved comparable expression levels as assessed by fluorescence intensity (fig. 1C).”

#3.2: For Figure 1A, what is the source of the observed increase in PI(4,5)P2, how do the authors take into account the role of endogenous PIP5Ks?

We added a new experiment in the revised Fig. 1B showing that the increased PI(4,5)P2 occurs at the expense of PM PI4P:

This is described in the first paragraph of the results on p.3: “PI(4,5)P2 levels are expected to increase at the expense of PM PI4P levels when over-expressing any of the three isoforms of human PIP5K (A-C) or the single paralog from the budding yeast, Saccharomyces cerevisiae (Mss4). Indeed, this was precisely what we observed (fig. 1A and B, statistics reported in tables 1 and 2).”

The role for endogenous PIP5Ks is clarified on the sentence that spans pp. 3-4: “We therefore reasoned that saturation of endogenous, inhibitory PIP4K molecules by PIP5K over-expression, regardless of catalytic activity of the PIP5K, would free endogenous, active PIP5K enzyme from negative regulation (fig. 1D).”

• *

#3.3: For Figure 1B, could the authors comment on the intracellular distribution of PI(4,5)P2. How are they able to reliably distinguish their signal between plasma membrane and intracellular localizations and conclude that PIP2 on the plasma membrane is decreased?

As detailed in the now expanded methods section covering image analysis on p. 22, our analysis specifically quantifies fluorescence in the plasma membrane.

#3.4: Please include statistics for all image- based quantitation analysis.

We have added details of statistical analysis and tabulated the results, as detailed in our response to __#1.1. __

__#3.5: __ Could the authors comment on the ability of PIP4K to have affinity for its own product? How does PIP4K sense membrane PI(4,5)P2 since these kinases are mostly cytoplasmic?

We have added a comment to the 1st paragraph of the Discussion on p.9: “PIP4K’s low affinity and highly co-operative binding to PI(4,5)P2 makes it an excellent sensor for tonic PI(4,5)P2 levels. It is poised to sense PI(4,5)P2generated in excess of the needs of the lipids’ legion effector proteins, ensuring these needs are met but not exceeded. Nevertheless, the relatively low PIP4K copy number of around 2.5 x 105 per cell [29] is a small fraction of the total PI(4,5)P2 pool, estimated to be ~107 [33], ensuring little impact on the capacity of the lipid to interact with its effectors.”

__#3.6: __Do the authors have any other experiments to substantiate the binding of the two PIP kinases, similar to the Wang et al findings? Is the N-term motif required? Is it possible to disrupt that interaction and show the phenotype?

We do not have additional, conclusive experiments to share at this time, and believe that characterization of the inhibitory interaction is beyond the scope of the current manuscript. We do however add a comment on this topic to the 1st paragraph of p. 11: “Regardless, a full understanding of paralog selectivity will need to be driven by a detailed structural analysis of the interaction between PIP4Ks and PIP5Ks - which is not immediately apparent from their known crystal structures, especially since PIP4Ks and PIP5Ks employ separate and distinct dimerization interfaces [50].”

#3.7: With the overexpression studies in Figure 1, do the authors see any changes in signaling when they just overexpress PIP5Ks versus in combination with PIP4Ks to show that the changes in plasma membrane PI(4,5)P2 can affect downstream signaling?

We agree with the reviewer that attenuating PIP5K-mediated PI(4,5)P2 increases with PIP4K should affect downstream signaling. However, we believe that these will not add additional insight compared to the already included experiments (fig. 8), whereby signaling output in response to graded changes in PI(4,5)P2 levels was investigated.

• *

Reviewer #3 (Significance (Required)): Overall, as mentioned above because of the 2019 Wang et al report the novelty is diminished, however using completely alternate methods and sophisticated microscopy this body of work indeed advances the field and provides further believable evidence of the PIP kinase families communicating in higher organisms which is required to maintain PIP2 levels shedding light on many of the findings that were previously unexplained surrounding the PIP4K studies. Further, the use of biosensors to describe these findings are new and will enable others in the field to begin to use such tools to investigate potential crosstalk between other lipid kinases.

As we argued in the general comments, we do feel that this evaluation misses the key finding that PIP4Ks are PI(4,5)P2 sensors, and that this regulates PIP5K regulation as part of a feedback loop.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Using state of the art imaging techniques the authors try to address how cells sense PI(4,5)P2 levels and regulate PIP5Ks to maintain an optimal level since any dysregulation of PI(4,5)P2 levels can have significant effects on the functioning of the cell and led to numerous disease states, such as cancers.

The key conclusions are convincing and importantly validate previous disputed findings made by Wang et al. (Cell Reports 2019) using different and more rigorous methods, however unfortunately due to the Wang et al publication the overall novelty of this study is lacking. A suggestion to the authors is to state/explain with text more clearly how their findings are more precise and higher quality than the previous report and why their findings are necessary and significant to drive the field forward. Further, experiments in the study were performed in vitro in cultured cells using overexpression methods making the physiological significance a bit unclear and the enthusiasm of the main discovery dampened. With that being said these findings are worthy of publication in order to advance the field and understanding of how the PIP kinase families are regulated and maintain PIP2 homeostasis which is important for life.

Minor and easily addressable experiments should be performed by the authors the following. Further, many of these experimental issues can easily go in supplemental materials

1. Include western blots for the constructs to compare expression levels.
2. For Figure 1A, what is the source of the observed increase in PI(4,5)P2, how do the authors take into account the role of endogenous PIP5Ks?
3. For Figure 1B, could the authors comment on the intracellular distribution of PI(4,5)P2. How are they able to reliably distinguish their signal between plasma membrane and intracellular localizations and conclude that PIP2 on the plasma membrane is decreased?
4. Please include statistics for all image- based quantitation analysis.
5. Could the authors comment on the ability of PIP4K to have affinity for its own product? How does PIP4K sense membrane PI(4,5)P2 since these kinases are mostly cytoplasmic?
6. Do the authors have any other experiments to substantiate the binding of the two PIP kinases, similar to the Wang et al findings? Is the N-term motif required? Is it possible to disrupt that interaction and show the phenotype?
7. With the overexpression studies in Figure 1, do the authors see any changes in signaling when they just overexpress PIP5Ks versus in combination with PIP4Ks to show that the changes in plasma membrane PI(4,5)P2 can affect downstream signaling?

Significance

Overall, as mentioned above because of the 2019 Wang et al report the novelty is diminished, however using completely alternate methods and sophisticated microscopy this body of work indeed advances the field and provides further believable evidence of the PIP kinase families communicating in higher organisms which is required to maintain PIP2 levels shedding light on many of the findings that were previously unexplained surrounding the PIP4K studies. Further, the use of biosensors to describe these findings are new and will enable others in the field to begin to use such tools to investigate potential crosstalk between other lipid kinases.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

This paper proposes that the enzyme PIP4K2C is a negative regulator of the synthesis of PI(4,5)P2 and that it does so by dampening the activity of PIP5K which is the enzymatic activity responsible for producing the major pool of PI(4,5)P2 in cells.

Significance

Nature and Significance of the advance:

Although the findings of the paper are presented as a major new advance, the observation that PIP4K might acts as a negative regulator of PIP2 synthesis has been previously presented in two previous publications. The significance of this paper is that it also shows the same point in another model system.

PIP4Ks Suppress Insulin Signaling through a Catalytic-Independent Mechanism Diana G Wang 1, Marcia N Paddock 2, Mark R Lundquist 3, Janet Y Sun 3, Oksana Mashadova 3, Solomon Amadiume 3, Timothy W Bumpus 4, Cindy Hodakoski 3, Benjamin D Hopkins 3, Matthew Fine 3, Amanda Hill 3, T Jonathan Yang 5, Jeremy M Baskin 4, Lukas E Dow 6, Lewis C Cantley 7 PMID: 31091439; PMCID: PMC6619495;DOI: 10.1016/j.celrep.2019.04.070

and

Phosphatidylinositol 5 Phosphate 4-Kinase Regulates Plasma-Membrane PIP3 Turnover and Insulin Signaling. Sharma S, Mathre S, Ramya V, Shinde D, Raghu P.Cell Rep. 2019 May 14;27(7):1979-1990.e7. doi: 10.1016/j.celrep.2019.04.084.PMID: 31091438

Both of these studies show that in cells lacking PIP4K, during signalling the levels of PIP2 rise much greater than in wild type cells. Indeed the Cantley lab paper (Wang et.al) have shown that this is likely due to an increase in PIP5K activity, using an in vitro assay. They have further disrupted the interaction between PIP4K and PIP5K and demonstrated the importance of this interaction in the enhanced levels of PIP2.

Likewise although the authors have claimed that no mechanisms have claimed that there are no mechanisms reported to sense and downregulate PIP2 resynthesis. It is suggested that they read and consider the following recent paper which studies Pip2 resynthesis during GPCR triggered PLC signalling.

Kumari A, Ghosh A, Kolay S and Raghu P*. Septins tune lipid kinase activity and PI(4,5)P 2 turnover during G-protein-coupled PLC signalling in vivo. Life Sci Alliance. 2022 Mar 11;5(6):e202101293. doi: 10.26508/lsa.202101293. Print 2022 Jun.

Likewise there are other earlier papers in the literature which have studied possible PIP2 binding proteins as sensors for this lipid.

Technical standards: The work is done to a high technical standard.

Major and Minor Comments

Comments for Figure 1

1. Does catalytically dead isoform of PIP4K2B and 2C also yield the same result as a catalytically dead version of PIP4K2A in Fig 1B?
2. The labelling on y-axis for PI(4,5)P2 biosensor intensity ratio is PM/cell at some places, PM/Cyt or PM/Cyto in some places. It is recommended to make it uniform across all the panels.

Comments for Figure 2

1. The claim that PI(4,5)P2 production is sufficient to recruit PIP4K2C to the PM can be ascertained further if one is able to do an experiment where PI(4,5)P2 is ectopically expressed in some compartment of the cell which is non-native to PI(4,5)P2 and as a consequence of this PIP4K2C is recruited to this non-native compartment.
2. In the entire figure 2, to establish that PI(4,5)P2 is necessary and sufficient for PM localisation of PIP4K, PIP4K2C is used as the PIP4K isoform on the basis that it is highly abundant in HEK293 cells. But PIP4K2A is localised mainly at the plasma membrane and here we are discussing about PI(4,5)P2 regulation at the PM . Can experiments be done with isoforms 2A and 2B as well? Can acute depletion of PI(4,5)P2 lead to the membrane dissociation of the isoform 2A as well? This will help us in understanding if there is an isoform specific difference in sensing PI(4,5)P2 levels which will help us in targeting specific isoform as therapeutic targets.
3. In Figure 1A, it is shown that overexpression of a catalytically dead PIP5K 1A/1B/1C is still able to increase PI(4,5)P2 levels. In the figure 2E, expression of homodimeric mutant of PIP5K domain which is a way to increase catalytic activity of PIP5K, increases PI(4,5)P2 levels which is consistent with the inferences from Fig, 1 , but what is surprising is a catalytically dead variant not being able to do so. Why is there a discrepancy between Fig. 1A and Fig. 2E? If the homodimeric mutant is the reason, then it is not clear in the explanation.
4. Show the loading control in Fig 2A western.
5. In the figure 2D, in the legend OCRL is written. So, the labelling in the panel should also be changed to OCRL from INPP5E. It is intermixed.
6. In the figure 2E, can the labelling be changed from HD to something more self-explanatory for homodimeric mutant of PIP5K domain?
7. In Fig. 2E, PIP5K expression is acute and in Fig. 2F Mss4 expression is chronic, both of which is able to recruit PIP4K2C to the plasma membrane. How can a likewise argument be drawn out of these two experiments when one is acute and the other one is a chronic expression? It is suggested to do an FRB-FKBP experiment for Mss4 as well.
8. In the text, Fig. 2G and 2H is written for PIP4K2C, but in the corresponding panels and legends, it is an assay for purified PIP4K2A on SLBs. Kindly resolve the discrepancy.

Comments for Figure 3

1. Kindly explain a bit in detail why the baits were now targeted to ER-PM contact sites. It is not self-explanatory.
2. The conclusions for Fig. 3 most likely hints towards the possibility of PIP4K and PIP5K interaction being independent of PI(4,5)P2 levels. Well, Fig. 3C and 3D does suggest a direct interaction, but can other protein-protein interaction assays be used to establish the direct interaction of PIP4K with PIP5K such as FRET or Yeast two hybrid as assays scoring for interaction?
3. Conceptually a direct interaction can be explained to some extent from Fig. 3 but extending it to be an inhibitory interaction is not right without a direct experiment. Can an experiment be done with PI4P enriched SLB, wherein you put just PIP5K purified protein vs PIP5K+PIP4K combination and measure the % mol of PI(4,5)P2 produced using a probe. That will be suggestive of a negative interaction.
4. In Figure 3B, the FRB tagged constructs are magenta coded and PIP4K2C is cyan. Kindly change the labelling of the FRB constructs on the y axis to magenta so that it goes with what is written in the legend. It will also be appreciated to show a colocalization quantification between the magenta (FRB constructs) and cyan (PIP4K2C) post rapamycin addition and not just the intensity for ER-PM recruited PIP4K2C.
5. Again, in the text , the description is written for PIP4K2C but in the result panel and legend (Fig. 3C and Fig. 3D), PIP4K2A is mentioned. Kindly resolve the discrepancy

Comments for Figure 4

1. In the Fig. 4B, it will be appreciated to show statistical significance in terms of R2 value for commenting on the linear response.

Comments for Discussion

1. Discussion can be in general a bit more detailed which is suggestive of future experiments to do that can shed more light on the interaction such as which residues in PIP4K interacts with PIP5K to negatively regulate it.
2. In the discussion, more light can be shed on the fact that Mss4 in spite of being a 5- kinase is not negatively regulated by PIP4K and the fact that PIP4K is present only in metazoans suggests that this fine tuning of PI(4,5)P2 levels is specific to metazoans. Another insight could be in the direction, that Fig 4. tells PI3K, but not calcium signaling is modulated by this fine tuning and interestingly class I PI3K is also an enzyme specific to metazoans. Hence, unlike yeast, metazoans rely on growth factor signalling processes, hence regulation of PI(4,5)P2 by PIP4K and hence Class I PI3K and PI(3,4,5)P3 could be a process relevant to metazoans.

Audience: cell biologists and biochemists interested in PIp2 signalling and PIP kinases

My expertise: PIP2 and PIP kinases

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In this manuscript, authors address how PIP4K regulates tonic plasma membrane (PM) PI(4,5)P2 levels which are generated by major PI(4,5)P2 synthesis enzyme, PIP5K by using PIP4K and PIP5K overexpressing cells or acutely manipulating PM PI(4,5)P2 levels by the chemically induced dimerization (CID) system. Additionally, authors assessed effect of direct interaction between PIP4K and PIP5K by using supported lipid bilayers (SLBs) and purified PIP4K and 5K. Authors also were successful in monitoring dynamics of endogenous PIP4K by using a split fluorescent protein approach. Through this study, authors propose a model of PI(4,5)P2 homeostatic mechanism that PIP4Ks sense elevated PM PI(4,5)P2 by PIP5Ks, are recruited to the PM, and bind to PIP5Ks to inhibit PIP5Ks activity.

1. Although authors mention methods of statistical analysis in materials and methods, they did not present the results of statistical analysis in the figures. The quantitative data should be presented with statistical analysis data, which is important for showing where convincing differences between treatment groups are found.
2. Fig. 1D. Fig. 1D and Fig. 3A should be presented together because these are exactly same set of cells and information of each PIP4K and PIP5K membrane localization could be important for understanding mechanisms of inhibitory effect of PIP4Ks. Authors claimed that over-expression of all three PIP4K isoforms were able to attenuate the elevated PM PI(4,5)P2 levels caused by PIP5K over-expression. However, in Fig. 3A, PIP4K2A was recruited to PM by both PIP5K1A and PIP5K1C but looks only attenuated PIP5K1A, but not PIP5K1C, overexpression mediated PM PI(4,5)P2 elevation (Fig. 1D). PIP4K2C was less recruited to the PM than PIP4K2A and 2B in PIP5K1A overexpressing cell (Fig. 3A) but PIP4K2A, B and C isoforms equally attenuated increase of PM PI(4,5)P2 in PIP5K1A overexpressing cell (Fig. 1D). It is likely that efficiency of inhibitory effect of each PIP4K isoform is different by co-overexpressed PIP5K isoform. These images should be more carefully documented with Fig. 1D and Fig. 3A together.
3. Fig. 1F. It seems that PIP4K2A accelerated PIP5K, but not Mss4, dependent PI(4,5)P2 generation before PI(4,5)P2 reaches 28,000 lipids/um2. Is this significant? If so, why did this happen?
4. Fig. 3B. In this figure, authors only presented images after Rapa treatment. Therefore, it is not clear what these results mean. Before Rapa treatment, where did bait proteins and NG2-PIP4K2C localize? If ePIP4K2C delta PM intensity (ER:PM/PM) increase, does that mean increase in ER:PM intensity or decrease in PM intensity? According to Figure legend, PI(4,5)P2 indicator TubbycR332H was co-transfected, but those images are not shown in the figure. Images of PI(4,5)P2 indicator also should be presented to show whether after Rapa treatment PI(4,5)P2 increased at ER-PM contact sites, because that could be critical for the conclusion that "The use of Mss4 ruled out an effect of enhanced PI(4,5)P2 generation at contact sites, since this enzyme increases PI(4,5)P2 as potently as PIP5K1A (Fig. 1A), yet does not cause recruitment of PIP4K2C". Is this conclusion consistent with Fig. 2F and G?
5. Fig. 3C and D. Based on results of Fig. 3C and D, authors concluded that "PIP4K2C binding to PI(4,5)P2-containing SLBs was greatly enhanced by addition of PIP5K to the membranes, but not Mss4". I don't think Fig. 3C and D are comparable because experimental conditions are different. While lipid composition of SLB used in Fig. 3C was 2% PI(4,5)P2, 98% DOPC, in Fig. 3D, it was 4% PI(4,5)P2, 96% DOPC. And also, in Fig. 3C, PIP5K1A was added to SLB at the time about 50 sec, whereas in Fig. 3D, Mss4 was added at 600 sec. It seems that in Fig. 3D, PIP4K2A was already saturated on SLB before adding Mss4. These two experiments must be performed under the same conditions.
6. Overall results discussed in the text are very compressed referring readers to the 4 multi-panel complex figures with elaborate figure legends. While it is possible to figure out what the authors' studies and results are, it is quite a laborious process.

Minor comments:

1. Fig. 2D. The purified 5-phosphatase used in Fig. 2D is INPP5E but described in figure legend and materials and methods ass OCRL. Which one is correct?
2. Fig. 3B. Indicate which trace represents PIP5K1A, Lyn11 or Mss4.
3. Fig. 4C. X-axis label. Is "Time (min)" correct? Or should it be "Time (sec)".

Significance

The finding that PIP4K itself is a low-affinity PI(4,5)P2 binding protein and sense increases of PM PI(4,5)P2 generated by PIP5K to control tonic PI(4,5)P2 levels by inhibiting PIP5K activity is a novel concept. However, inhibition of PIP5K by PIP4K and importance of the inhibitory effect of PIP4K in PI3K signaling pathway have previously been reported (ref 24). This reduces the novelty of the current work somewhat however, the authors do provide evidence for dual interactions of PIP4K (PIP2, PIP5K), which the previous report did not.

The reviewers have expertise in PLC- and PIP5K-related signaling pathways.

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

Learn more at Review Commons

---

Reply to the reviewers

Manuscript number: RC-2022-01723

Corresponding author(s): Daphne Avgousti, Srinivas Ramachandran

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

Summary This study by Lewis et al. examines the role of heterochromatin in the nuclear egress of herpesvirus capsids. They show that heterochromatin markers macroH2A1 and H3K27me3 are enriched at specific genome regions during the infection. They also show that when macroH2A1 is removed or H3K27me3 is depleted (both of which reduce the amount of heterochromatin at the nuclear periphery), the capsids are not able to egress as effectively. This is interesting since it could be argued that heterochromatin acts as a hindrance to the transport of viral capsids to the nuclear envelope and that the loss of it would allow capsids to reach the nuclear envelope more easily. However, this paper seems to show that heterochromatin formation, on the contrary, is necessary for efficient egress. Overall, the study seems comprehensive. The methodology is solid, and the experiments are very well controlled. However, some issues need to be addressed before publication.

Major comments

1) In line 49, the authors state, "Like most DNA viruses, herpes simplex virus (HSV-1) takes advantage of host chromatin factors both by incorporating histones onto its genome to promote gene expression and by reorganizing host chromatin during infection". In addition, HSV1 expression can be hindered by the host's interferon response via histone modifications. Ref. Johnson KE, Bottero V, Flaherty S, Dutta S, Singh VV, Chandran B. IFI16 restricts HSV-1 replication by accumulating on the HSV-1 genome, repressing HSV-1 gene expression, and directly or indirectly modulating histone modifications. PLoS Pathog. 2014 Nov 6;10(11):e1004503. doi: 10.1371/journal.ppat.1004503. Erratum in: PLoS Pathog. 2018 Jun 6;14(6):e1007113. PMID: 25375629; PMCID: PMC4223080.

We agree with the reviewer and have amended our text and added the reference. See line 57.

2) Reference 5 is misquoted in the sentence, "This redistribution of host chromatin results in a global increase in heterochromatin". In that reference, the amount of heterochromatin is not analyzed in any way. However, that particular paper shows that the transport of capsid through chromatin is the rate-limiting step in nuclear egress, which is important considering this study. Further, the article by Aho et al. shows that when the infection proceeds capsids can more easily traverse from the replication compartment into the chromatin, which means that infection can modify chromatin for easier capsid transport. For that reason, the article is an important reference, but it needs to be cited correctly.

We agree with the reviewer that this citation was misquoted and have corrected the citation. See lines 55 and 62-64.

3) The term heterochromatin channel at lines 54, 102, and 303 is misleading since the channels seen in the original referred paper are less dense chromatin areas. Also, this term is not used in the original paper where the phenomenon was first described. These less dense interchromatin channels were found by soft-X-ray tomography imaging and analyses, not by staining.

We thank the reviewer for pointing out this discrepancy and have amended the text to accurately describe the methods used in the appropriate citations. See lines 65, 115, and 383.

4) It is difficult to visualize chromatin using TEM microscopy. The values of peripheral chromatin thickness given in Figure 1e (5-15 nm) do not seem realistic given that the thickness of just one strand of histone-wrapped DNA is 11 nm. Why are the two values for WT different? If you can get so different values for WT, it is a bit worrisome (switching the WT results between the top and bottom parts of Fig. 1e would for example result in very different conclusions on the effect of macroH2A1 KO for the thickness of the chromatin layer).

*We agree with the reviewer that it is difficult to visualize chromatin by TEM. It is also important to note that comparisons can only be made between samples treated on the same day in the same way. Taking this into account, we chose to compare macroH2A1 KO cell stains to controls done at the same time, and the same for H3K27me3 depleted conditions compared to DMSO treated and prepare for EM at the same time. Visually, it is apparent that the staining in the macroH2A1 KO control cells is somewhat different than those of the H3K27me3 depleted control cells, which represents the inherent variability of this method. It is also true that one nucleosome is around 11nm, however, since the cells contain highly compacted chromatin with many other proteins present, this measurement is not appropriate to apply. Adding up the millions of nucleosomes that make up the chromosomes at 11nm each would result in a space much larger than the nucleus, therefore we focus on comparing between control and experimental conditions restricted to this assay as a relative qualitative comparison. Nevertheless, we agree with the reviewer that the notion of changing chromatin is difficult to quantify by EM and so we have taken an additional approach to test our hypothesis and confirm EM interpretations (discussed lines 391-393). We have utilized live capsid trafficking to visualize capsid movement in nuclei in the presence or absence of macroH2A1. The results from these new experiments are presented in new Figure 5 and EV5 and support our model. *

5) In lines 134-137 it says that "The enrichment of macroH2A1 and H3K27me3 was observed as large domains that were gained upon viral infection (Fig 2a), suggesting that the host landscape is altered upon infection. These gains were reflected in an increase in total protein levels measured by western blot (Fig 2b)." However, the protein levels of H3K27me3 do not seem to increase during infection. In other presented data as well (Figs. 2a, 2b, 2c, S2a) it is difficult to justify the statement that H3K27me3 is enriched in infection. When this is the case, the conclusion that the amount of heterochromatin increases in the infection (the quotation above and the one in line 315) is not supported. The statement in line 315 is also not specific since it is unclear what "newly formed heterochromatin increases" means.

We agree with the reviewer that our original description was misleading. We now have edited the text to clarify that there is redistribution of macroH2A1 and H3K27me3. In the revised manuscript, we have also included mass spectrometry data mined from Kulej et al. that show peptide counts that reflect increases in the heterochromatin markers described (see new Figure EV1a). Despite this quantitative measure, upon rigorous replicates of our western blots as requested by Reviewer 2, we concluded that the increases originally described are somewhat inconsistent by western blot. This discrepancy between mass spectrometry data and western blot is likely due to the non-linear nature of antibody binding and developing of western blots by the ECL enzymatic reaction. Therefore, our revised manuscript focuses on this redistribution as a reaction to infection and stress responses instead of a global increase as the original manuscript stated. See lines 174, 182, 196, 397 and Fig EV4d in main text and discussion sections.

• *

6) Quantitation of viral capsid location in H3K27me3-depleted cells seems somewhat arbitrary. It would have been more robust to calculate the number of capsids per unit length of the nuclear envelope with and without depletion.

We agree with the reviewer that the quantification of capsids in the H3K27me3-depleted conditions was arbitrary. In our revised manuscript, we have now repeated this quantification to accurately measure the phenotype observed, that is the chains of capsids lined up at the inner nuclear membrane. To do this, we used two measures: 1) the distance from the INM as less than 200nm and 2) the distance from other capsids as less than 300nm. Taking into account these two measures, we quantified the frequency with which multiple capsids lined up at the INM in WT and H3K27me3-depleted conditions. This is represented in the new Figure 5d. In the WT setting, we observe most often 1 single capsid at the INM, with a small fraction of 2 capsids. However, in the H3K27me3-depleted condition, we observe much greater numbers of capsids at the INM more frequently, as many as 16 at a time, leading to an average of 2-3 capsids at any single location. The source data for this figure are also provided. See lines 589 and Fig5d.

7) In lines 300-302 it says "Elegant electron microscopy work showed that HSV-1 infection induces host chromatin redistribution to the nuclear periphery2,8." However, the redistribution data in reference 8 is based on soft x-ray tomography and not on electron microscopy."

We have amended the text to accurately describe the methods used in the citations. See line 384.

8) The authors bundle together the effects of macroH2A1 removal and H3K27me3 depletion by saying that they both decrease the amount of heterochromatin at the nuclear periphery and therefore hinder capsid egress. This seems overly simplistic and macroH2A1 and H3K27me3 seem to act very differently, which is manifested in the drastic difference in nuclear capsid localization between the two cases. This difference needs to be discussed more.

We agree with the reviewer that there is a nuanced difference in the effect on nuclear egress in the absence of the two heterochromatin marks. Specifically, that macroH2A1 loss results in greater numbers of capsids dispersed throughout the nucleus, whereas depletion of H3K27me3 results in capsids reaching the INM and not escaping. To examine these differences further, we have carried out live imaging of capsid trafficking in macroH2A1 KO cells compared to control and found that capsids move much more slowly, consistent with our model, see new Figure 5h-I and EV5h-i. Conversely, H3K27me3 depletion does not prevent the capsids from reaching the INM, raising the question of whether they are successfully able to dock at the nuclear egress complex (NEC). To investigate this further, we obtained an antibody against the NEC component UL34 and probed during infection in our heterochromatin disrupted conditions. We found that UL34 levels are unchanged upon loss of macroH2A1 or depletion of H3K27me3, suggesting the levels of UL34 do not account for the decrease in titers. These data are now presented in new Figure EV3g-h. Furthermore, we have amended our model to include the two different scenarios upon loss of different types of heterochromatin (see new Figure 6) and discussion of these differences. See line 428.

Minor comments Line 45: Nuclear replicating viruses -> Nuclear-replicating viruses Line 56: is -> are Line 64: 25kDa -> 25 kDa Line 159: macroH2A1 cells -> macroH2A1 KO cells Line 289: The term gDNA is rarely used for viral DNA. Replace gDNA with viral DNA. Line 405: 8hpi -> 8 hpi Line 449: mm2 -> μm2 "Scale bar as indicated" words can be removed in the figure legends or at least should not be repeated many times within one figure legend.

We have amended the text to address these comments. See lines 52, 68, 76, 179, 334, 513, and 585.

Reviewer #1 (Significance (Required)):

These findings would appeal to a broad audience in the field of virology. Specifically, the researcher in the fields of virus-cell and virus-nucleus interactions. This manuscript analyses herpesvirus-induced structural changes in the chromatin structure and organization in the nucleus that are also likely to affect the intranuclear transport of viral capsids.

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

The manuscript "HSV-1 exploits heterochromatin for egress" describes the effects of heterochromatin at the nuclear periphery, macroH2A1 or H3K27me3 on HSV-1 replication and egress. Knocking out macroH2A1 or depleting H3K27me3 with high concentrations of tazemetostat depleted heterochromatin at the nuclear periphery, may not have affected HSV-1 protein expression and modestly inhibited the production of cell-free infectivity and HSV-1 genomes. macroH2A1 deposition was affected by infection, creating new heterochromatin domains which did not correlate directly with the levels of expression of the genes in them. The authors conclude that heterochromatin at the nuclear periphery dependent on macroH2A1 and H3K27me3 are critical for nuclear egress of HSV-1 capsids.

The experiments leading to the conclusion that HSV-1 capsids egress the nucleus through channels in the peripheral chromatin confirm previously published results (https://doi.org/10.1038/srep28844). The previously published EM micrographs show a much larger number of nuclear capsids, more consistent with the images in the classical literature, even in conditions when nuclear egress was not inhibited. Figures 1 and 4 show scarce nuclear capsids, even under the conditions when nuclear egress should be inhibited according to the model and analyses. The large enrichment in nuclear capsids in KO cells predicted by the model is not reflected in figure 4a, which shows only a modest increase in nuclear capsid density (the total number of nuclear capsids would be more informative). The number or density of nuclear capsids is not shown in H3K27 "depleted" cells. The robustness of the analyses of the number of capsids at the membrane in H3K27 "depleted" cells is unclear. For example, the analyses could be repeated with different cut offs, such as 2 or 4. If they are robust, then the conclusions will not change when the cutoff value is changed.

We appreciate the reviewer’s observation that to number of capsids we show differs from those published in the publication by Myllys et al. (Sci Rep 2016 PMID 27349677). It is important to note there are several differences between our study and that of Myllys et al. that explain the difference. First, as reviewer 1 pointed out, the Myllys et al. study used three-dimensional soft X-ray tomography combined with cryogenic fluorescence and electron microscopy to observe capsids in 3D rendered nuclei. Since our method uses only single ultrathin 50nm slices of cells, we cannot visualize the total number of capsids per nucleus, rather only per slice, which is why we have averaged slices of many nuclei to generate a statistical comparison between macroH2A1 KO or H3K27me3-depleted and control cells treated at the same time (see response to reviewer 1). Furthermore, these other methods are specialized techniques for 3D imaging that are beyond the scope of our study. Second, the Myllys et al. paper used B cells which are much smaller than HFFs, lending themselves to better tomography studies but not commonly used to study HSV-1 biology. Third, the Myllys et al. paper also used a different MOI and time point than we have. Taken together, these differences account for the disparity in visualizing capsids which is why we quantified capsid number across many images.

We agree with the reviewer that our quantification in the H3K27me3-depleted cells compared to control was somewhat arbitrary. As stated in the response to Reviewer 1 above, in our revised manuscript we have now repeated this quantification to accurately reflect the phenotype observed, that is the chains of capsids lined up at the inner nuclear membrane. To do this, we used two measures: 1) the distance from the INM as less than 200nm and 2) the distance from other capsids as less than 300nm. Taking into account these two measures, we quantified the frequency with which multiple capsids lined up at the INM in WT and H3K27me3-depleted conditions. This is represented in the new Figure 5d. In the WT setting, we observe most often 1 single capsid at the INM, with a small fraction of 2 capsids. However, in the H3K27me3-depleted condition, we observe much greater numbers of capsids at the INM more frequently, as many as 16 at a time, leading to an average of 2-3 capsids at any single location. The source data for this figure are also provided. See lines 589 and Fig 5d.

Furthermore, we have now also carried out live-imaging analysis of single capsids during infection which show the appropriate number of capsids expected when the full nucleus is visible. These results are presented in the new Figure 5 and EV5.

The quantitation of the western blots present no evidence of reproducibility and/or variability. The number of biologically independent experiments analyzed must be stated in each figure and the standard deviation must be presented. As presented, the results do not support the conclusions reached. The quality of western blots should also be improved. it is unclear why figure 2b shows viral gene expression in wild-type cells only, and not in KO or H3K27me3 depleted cells, which are only shown in the supplementary information. These blots presented in Figure S5a and S5b are difficult to evaluate as the signal is rather weak and the controls appear to indicate different loading levels. These blots do not appear to be consistent with the conclusions reached. Some blots (VP16, ICP0 in HFF) appear to indicate a delay in protein expression whereas others (VP16, ICP0 in RPE) appear to indicate earlier expression of higher levels. The claimed "depletion of H3K27me3 is not clear in in figure S5d, in which the levels appear to be highly variable in all cases, without a consistent pattern, with no evidence of reproducibility and/or variability, and using a mostly cytoplasmic protein as loading control. All western blots should be repeated to a publication level quality, the number of independent experiments must be clearly stated in each figure, and the reproducibility and/or variability must be indicated by the standard deviation.

*As reviewer 1 also pointed out, we appreciate that there is some variability with respect to the stated ‘increase’ in these heterochromatin marks during infection. As stated in response to reviewer 1, in our revised manuscript we have included a deeper analysis of these marks from global mass spectrometry that indicates an increase in total levels. Please see response to reviewer 1. *

• *

In the revised manuscript, we have now included mass spectrometry data mined from Kulej et al. that show peptide counts that reflect increases in the heterochromatin markers described (see new Figure EV1a). Despite this quantitative measure, upon rigorous replicates of our western blots as requested by Reviewer 2, we concluded that the increases originally described are somewhat inconsistent by western blot. This discrepancy between mass spectrometry data and western blot is likely due to the non-linear nature of antibody binding and developing of western blots by the ECL enzymatic reaction. Nevertheless, our genome-wide chromatin profiling showed consistent, reproducible, and statistically significant redistribution of macroH2A1 and H3K27me3 upon HSV-1 infection. Therefore, our revised manuscript now focuses on this redistribution as a reaction to infection and stress responses instead of a global increase as the original manuscript stated. See lines 174, 182, 196, 397 and Fig EV4b-c.

• *

With respect to viral protein levels, although there is slight variation in the levels of VP16 or ICP0 in RPEs compared to HFFs, we do not feel that this difference is biologically significant as several other measures of viral infection progression are unchanged (viral RNA, viral genome accumulation within infected cells). Furthermore, the significant difference in titers we observe is not explained by slight differences in ICP0 or VP16. Nevertheless, to document this variability in western blot and assuage any concern of impact infection progression, we have repeated each western blot presented in the paper three separate times and used these blots to quantify each relevant protein. Graphs of western blot quantitation can be found in each figure accompanying a western blot as follows:

Western blots:

Figures 3b-c, 4ab, EV1b, EV5a

Quantitation of western blots:

Figures 3d, 4c, EV1c, EV5b-f

• *

An enhanced analyses of the RNA-seq data, analyzing all individual genes rather than pooling them together, would provide better support to these conclusions. Then, the western blots are useful to show that the changes in mRNA result in changes in the levels of selected proteins.

• *

*We appreciate the reviewer’s interest in the RNA-seq data, however, we feel that reviewer has not understood the analysis we presented in the initial submission. To clarify, we calculated fold changes for individual genes and did not pool RNA-seq data anywhere in the manuscript. We show boxplots of log2 fold changes of individual genes. Boxplots enable summarization of the salient features of a distribution while still representing individual gene analysis. Here, the distribution being plotted is the log2 fold change of individual genes that intersect with macroH2A1 domains that change due to infection. As such, clusters 1-3 of macroH2A1 domains feature a loss in macroH2A1 due to infection and the boxplots show that the majority of genes are upregulated. To highlight this point further, in our revised manuscript we have included volcano plots of genes intersecting with each cluster also showing the split between the number of genes significantly upregulated and downregulated in each cluster at each time point (see new Figure EV3c). As expected from the boxplots, clusters 1-3 feature a much higher fraction of genes are significantly upregulated, whereas cluster 5 features a higher fraction of genes downregulated with concomitant increase in macroH2A1 due to infection. Taken together with the gene ontology analysis (new Figure Sd), these results support our model in which macroH2A1 is deposited in active regions to block transcription and promote heterochromatin formation. To further support these conclusions, we have also carried out analysis of 4sU-RNA data generated upon salt stress or heat shock and found that the regions defined by gain of macroH2A1 (i.e. clusters 5 and 6) also exhibit significant decreases in new transcription at just 1-2 hours after treatment. These data, which are presented in new Figure EV3b-c, strongly support our model in which macroH2A1 is deposited in genes downregulated upon stress response to generate new heterochromatin. *

Figure S1 raises some questions about the specificity of the macroH2A1 antibody used for CUT&Tag. As expected CUT&Tagging the cellular genome in the KO cells with the specific antibody results in lower signal than with the IgG control antibody. In contrast, viral DNA is CUT&Tagged as efficiently in the KO as in the WT cells, and in both cases significantly above the IgG controls. The simplest interpretation of these results is that the antibody cross-reacts with a protein that binds to HSV-1 genomes. The manuscript must experimentally address this possibility.

We agree with the reviewer that there is a possibility that antibodies cross react. However, we are confident that this is not the case in this scenario for the following reasons:

• *

*1 – We have carried out immunofluorescence analysis of macroH2A1 or H3K27me3 during HSV-1 infection and observe no overlap with ICP8 staining. We have included these images together with a histogram documenting the lack of overlap in the new Figure EV2f-g. *

• *

2 – CUT&Tag relies on the Tn5 transposase to insert barcodes into accessible regions of the genome. An inherent limitation of this method during viral infection is that the replicating viral genome is very dynamic and accessible, leading to easier and less specific insertion by the transposase. This is evidenced by the pattern of signal across the viral genome that is completely overlapping in the macroH2A1, H3K27me3 and IgG conditions. Snapshots of the full viral genome are now included in the new Figure EV2c-d.

• *

*Furthermore, using CUT&Tag with macroH2A1 antibody, we expect the transposition rate to be identical between WT and macroH2A1 KO conditions for the Ecoli and viral genomes. This is because we assume that the transposition in these two genomes is non-specific since there is no macroH2A1 present. Then, we expect the spike-in normalized CUT&Tag enrichment on the viral genome to be the same between WT and macroH2A1 KO conditions. Since IgG should not be affected by macroH2A1 KO, we expect the IgG enrichment to be same between WT and macroH2A1 KO conditions. Thus, non-specific background would result in higher enrichment in an apparent signal on viral genome in the macroH2A1 KO condition. *

• *

Combined with this expectation for background transposition and the following: 1) the distribution of the CUT&Tag signal across the viral genome is virtually identical between IgG, macroH2A1, and H3K27me3 CUT&Tag signal in WT and macroH2A1 KO cells (see new Figure EV2c-d), 2) that there is no colocalization between macroH2A1 or H3K27me3 with viral genomes by immunofluorescence (see new Figure EV2f-g), and 3) the whole genome correlation of the signals across CUT&Tag samples on the viral genome, but not the host, are virtually identical as presented in a heat map (see new Figure EV1g vs EV2e), we conclude that the viral CUT&Tag signal is noise. Therefore, any analysis of the signal on the viral genomes would not be biologically meaningful.

• *

Also, Figure S1 shows that the viral genome is CUT&Tag'ed with H3K27me3 antibody as efficiently in macro H2A1 WT and KO cells, and in both cases above the background signal from IgG control antibody. The authors conclude that the signal with the specific antibody "mirrors" that of the control antibody, but "mirroring" is not defined and the actual data show that there is a large increase in signal with the specific antibody. Not surprisingly, the background signal also increases, as the number of genomes increase while infection progresses. The authors conclude that "these results indicated that there was a significant background signal from the viral genome that could not be accounted for", but no evidence supporting this conclusion is presented. The data show clear signal above the background from the viral genome and that this signal is not affected by the presence or absence of macroH2A1. This section of the manuscript has to be thoroughly re-analyzed as there is clear H3K27 signal.

*We agree with the reviewer that as presented in the current manuscript it seems as though there is a real H3K27me3 signal. However, as stated in the above comment, the pattern of this signal matches that of all other conditions, including IgG, suggesting it is not a real signal, cross-reacted or otherwise, but rather an artifact of the methodology. See new Figure EV2. *

The concentration of tazemetostat used is high. Normally, concentrations of around 1µM are used in cells, and 10µM is often cytotoxic (for examplehttps://doi.org/10.1038/s41419-020-03266-3; https://doi.org/10.1158/1535-7163.MCT-16-0840). The effects on H3K27me3 presented in figure S1b appear to be normalized to mock infected treated cells. If so, they do not allow to evaluate the effectivity of the treatment. Cell viability after the four days treatment must be evaluated, the claimed "depletion" of H3K27me3 must be clearly demonstrated (the blots in figure S5 are not sufficient as presented), and levels of different histone methylations must be tested to support the claimed specificity of tazemetostat for H3K27me3 at the high concentrations used.

*While we agree with the reviewer that the cytotoxicity of any inhibitor is an important aspect to take into account, in this instance the reviewer is incorrect. The reviewer has cited papers that highlight the potential use of tazemetostat as a cancer-cell specific treatment for colorectal and B-cell cancers. In both of these cases, the primary conclusion is that tazemetostat’s cytotoxic property is largely corelated to mutation in EZH2. In fact, WT EZH2 treated cells had a more “cytostatic” response, which shows that tazemetostat is not toxic with WT EZH2 (Brach et al. Mol Cancer Ther. 2017, PMID 28835384) as is the case in our system. Furthermore, the Tan et al. study shows a non-transformed human fibroblast (CCD-18co) and embryonic colon epithelial (FHC) as “healthy controls” for their work in colorectal cancer cell lines in Figure 1D. These 2 cell lines, which are comparable to the WT HFF cells we used, show no reduction in viability at a log fold greater concentration than the 10 µM used in our paper. *

• *

*Nevertheless, we agree with the reviewer that cytotoxicity should be formally ruled out. In our original experiment, we recorded cell counts at the harvested mock, 4-, 8-, and 12 hpi and found no difference in the number of cells over the course of infection (see new Figure EV3e). We also used trypan blue staining as a measure of cell viability upon tazemetostat treatment and found no toxicity. These results are presented in new Figure EV3f. *

Furthermore, we agree with the reviewer that total H3 levels by western blot should be included in any comparison of H3 modification. While these were included in some figures, they were unintentionally omitted in others. In our revised manuscript we have now included these blots together with quantification of triplicate biological samples of H3K27me3 levels normalized to total H3. See new Figures 3, 4, EV1, and EV5.

• *

Minor comments. Reference No.27 is misquoted in lines 250-251, which state that it shows that "HSV-1 titers, but not viral replication, where reduced upon EZH2 inhibition." The reference actually shows inhibition of HSV-1 infectivity, DNA levels and mRNA for ICP4, ICP22 and ICP27. This reference uses much shorter treatments (12 h and only after infection). It also shows that inhibition of EZH2/1 up regulates expression of antiviral genes.

*We appreciate that the reviewer has pointed out a discrepancy between our results using an EZH2 inhibitor (tazemetostat) and those from reference 27 (Arbuckle et al., mBio, 2017 PMID 28811345) that requires clarification. The reviewer states that the treatments were 12 hours after infection, however, this is incorrect. In the Arbuckle et al. study, the authors used multiple different inhibitors at high doses for short treatments before infection and noted that this caused an upregulation in antiviral genes that blocked infection progression of multiple viruses including HCMV, Ad5 and ZIKA. Importantly, these genes include multiple immune signaling and interferon stimulated genes. In our study, we specifically use a much lower dose of EZH2 inhibitor, with respect to the IC50 value, and waited 3 days to ensure a steady state. In our system, any initial burst of immune response from the inhibitor would likely have subsided by the time we do our infection. Furthermore, supplemental figure EV1 from the Arbuckle et al. study states that EZH1/2 inhibitors do not affect nuclear accumulation of viral genomes and suppress HSV-1 IE expression in an MOI-independent manner (Arbuckle et al. Supplemental Figure 1). These results in fact support our conclusions that it is not any antiviral effect of inhibition of EZH2 that causes the decrease in titers that we observe. *

• *

To clarify, the IC50 value of the inhibitors used in the Arbuckle et al. study are 10 nmol/L (GSK126) and 4 nmol/L (GSK343). The IC50 is a measurement used to denote the amount of drug needed to inhibit a biological process by 50% and is commonly used in pharmacology to compare drug potency. In the Arbuckle et al. study, GSK126 was used at a concentration range of 15-30 µM, that is 1500-3000x more than the IC50 level as converted from nmol/L to µM, and GSK343 was used at a concentration range of 20-35 µM, that is 5000-8750x more than the IC50 level, to see changes in viral mRNA levels. The IC50 value for tazemetostat is 11 nmol/L which means that one would need to use a much higher molarity of tazemetostat, at least 28 µM which would be 2500x the IC50 value, to achieve the comparable biological changes as the inhibitors used in the Arbuckle et al. study. Thus, we are confident that the 10 µM concentration used in our study is an appropriate and non-toxic amount that would not impact antiviral responses at the dose and times that we used. As shown above and reported in multiple studies (for example: Knutson et al. Molecular Cancer Therapy 2014 PMID 24563539, Tan et al. Cell Death and Disease 2020 PMID 33311453 cited above, and Zhang et al. Neoplasia 2021 PMID 34246076, among others) the concentration of tazemetostat that we used is not toxic to the cells. Importantly, it was also reported that a global decrease in H3K27me3 by EZH2 inhibition using a 10 µM concentration of tazemetostat (here referred to by the identifier EPZ6438) did not impact HSV-1 RNA transcript accumulation measured by bulk sequencing (Gao et al. Antiviral Res 2020 PMID 32014498), consistent with our findings.

• *

In our revised manuscript, we have now included a discussion of these important points. See lines 409-428.

HFF are primary human cells but they are fibroblasts whereas the primary target of HSV-1 replication is epithelial cells. The wording used "they represent a common site of infection in humans" must be edited

We agree with the reviewer and have updated the text. See lines 109.

Disruption of macroH2A (1 and 2) results in general defects in nuclear architecture, not just peripheral chromatin (https://doi.org/10.1242/jcs.199216;, see also figure 1c and 5a, presenting invaginated and lobulated nuclei). The manuscript would benefit from including a broader discussion of the effects of macroH2A defects on the general nuclear architecture.

• *

We agree with the reviewer and our revised manuscript now includes a more in-depth discussion of the impact of macroH2A and other heterochromatin marks on nuclear structure. See lines 373-374 and 394.

The title should be edited, as "egress" in virology is commonly used to refer to the egress of virions from the cell, not to the nuclear egress of capsids. Adding the words nuclear and capsid should be sufficient to address this issue.

*We agree with the reviewer and will update the title to read “HSV-1 exploits host heterochromatin for nuclear egress”. Given that we are measuring multiple aspects of infection, we feel that adding the word ‘capsid’ is not necessary. *

It is unclear why preferential changes in expression of housekeeping genes would indicate "stress responses to infection". The rationale for this conclusion must be fully articulated and supported.

We agree with the reviewer that it may not be immediately clear as to why changes in house-keeping gene expression represent a stress response. In a recent study that we cite in our manuscript, Hennig et al. (PLOS Path 2018 PMID 29579120) demonstrate that changes in chromatin accessibility and gene transcription during HSV-1 infection resemble those that occur upon heat shock or salt stress. These results strongly support the model that global transcription changes caused upon stress (heat, salt, infection etc.) result in dramatic alterations to chromatin structure. In support of this notion, in our revised manuscript we now include analysis of these datasets based on our macroH2A1-defined clusters. Importantly, we found that the regions defined by gain of macroH2A1 (i.e. clusters 5 and 6) also exhibit significant decreases in new transcription at just 1-2 hours of exposure to salt and heat stress. These data, which are presented in new Figure EV3b-c, strongly support our model in which macroH2A1 is deposited on active genes to generate heterochromatin as a response to the stress of infection. We also discuss these results further in the revised manuscript, see lines 210-220, 233-236, and 424-426.

Statistical methods must be fully described in materials and methods and the number of biologically independent experiments must be stated in each figure.

*We agree with the reviewer and have included these details in each figure legend. *

Reviewer #2 (Significance (Required)):

The major strengths of the manuscript lie on the comprehensive analyses of the effects of knocking histone macroH2A in the nuclear architecture and chromatin organization. These analyses indicate that peripheral heterochromatin is defective in the KO. Another strength lies on the analyses of the news heterochromatin domains in HSV-1 infected cells. The relationship between the lack of correlation between the changes in gene expression and global heterochromatin domains defined by macroH2A1 with the main conclusion is less clear.

The major weakness is that the data presented do not strongly support the conclusions. Additional experiments are required to support the main conclusion that the effects in peripheral heterochromatin result in a biologically significant effect on capsid egress. The authors should also consider that the additional experimentation may not support the conclusion that macroH2A or H3K27me3 play critical roles in the nuclear egress of capsids.

• *

*To support our conclusions, we have carried out an entirely different set of experiments to track capsid movement. Bosse et al. PNAS 2015 PMID 26438852 and Aho et al. PLOS Path 2021 PMID 34910768 use live-imaging and single-particle tracking to characterize capsid motion relative to host chromatin. These approaches allowed the authors to discover that infection-induced chromatin modifications promote capsid translocation to the INM. They showed that 1) HSV-1 infection alters host heterochromatin such that open space is induced at heterochromatin boundaries, termed "corrals", in which viral capsids diffuse and 2) the movement of viral capsids through the host heterochromatin is the rate limiting step in HSV-1 nuclear egress. *

• *

To test our hypothesis that macroH2A1-dependent heterochromatin specifically is required, we collaborated with Dr. Jens Bosse to carry out these same experiments in our macroH2A1 KO and paired control cells. We tracked RFP-VP26 using spinning-disk confocal live imaging to track individual capsid movement within the nucleus. We found that capsids in cells lacking macroH2A1 traveled much shorter distances on average. This is represented graphically by the mean-square displacement (MSD) of capsid movement in macroH2A1 KO cells plateauing at ~0.4 µm2 vs 0.6 µm2 in WT cells, which represents the size of the “corral”, or space through which capsids diffuse. The average corral size in macroH2A1 KO cells is ~300 nm less than the average corral size in WT cells (two-thirds the size). These results are consistent with the finding that macroH2A1 limits chromatin plasticity both in vitro (Muthurajan et al. J Biol Chem 2011 PMID 21532035) and in cells (Kozlowski et al. EMBO Rep 2018 PMID 30177554). These data strongly support our hypothesis that macroH2A1-dependent heterochromatin is critical for the translocation of HSV-1 capsids through the host chromatin to reach the INM. Furthermore, these data support the model in which macroH2A1 allows for the increase of open space induced during infection. Loss of this open space restricts the movement of capsids in the nucleus, as quantified by our live-imaging experiments. These data are now included in the new Figure 5 and EV5 and described in lines 348-372 and 1011-1037.

• *

NOTE: These experiments were done in a separate lab using the same cells and MOI we used for our TEM studies. It is important to note that because this was done by live imaging where the full nucleus and cell are visible, the appropriate number of capsids is apparent.

Another major weakness is that the results of CUT&Tag of the viral genome are dismissed without proper justification. The authors conclude that the results invalidate the assays, but the results are consistent with cross-reactivity of the macroH2A1 antibody with another protein that interacts with the viral genomes and with H3K27me3 being associated with the viral genomes irrespectively of macroH2A1.

*We agree with the reviewer that as presented the viral genome reads were dismissed without thorough justification. As stated above, we are confident that the patterns we detected do not represent a biologically relevant signal but rather an artifact of the experimental set up. Furthermore, it is well known in the field that normalizing replicating viral genomes during lytic infection in any kind of chromatin profiling technique is fraught with inconsistencies as each cell may have a different copy number of viral genomes at any given time point. Therefore, we feel strongly that any analysis of the viral genome chromatin profile during a lytic replication at this point in time would require single cell sequencing which is beyond the scope of this study. We appreciate that this was not clearly presented in the original manuscript and in our revised submission we have included a full supplemental figure documenting the negative data that support our conclusions (see new Figure EV2). *

If the authors had additional data supporting the claim that these results do not reflect cross-reactivity or association with the viral genomes, these data must be presented. Without that additional data, the conclusions are not supported and these discussions must be removed from the manuscript. The authors may still opt to not analyze any association with the viral genomes, but they should not dismiss them as artifactual without actual evidence to support this claim. Previously published literature is also misquoted.

This study makes an incremental contribution to the previously published evidence showing that HSV-1 capsids egress the nucleus through channels in between the peripheral chromatin. It shows that disruption of the heterochromatin at the nuclear periphery, and the nuclear architecture in general, may have a modest effect on capsid egress. This information may be of interest mostly to a specialized audience focused on the egress of nuclear capsids.

While we agree with the reviewer on many points as stated above, we respectfully disagree that our study is merely an incremental contribution of interest only to a specialized audience focused on nuclear egress. As reviewer 2 states earlier, the strength of our study lies in the “comprehensive analyses of the effects of knocking histone macroH2A in the nuclear architecture and chromatin organization”, which would be of interest to a broad chromatin audience as well as virologists. Together with the new data presented here and a revised manuscript, we feel that our study would be of interest to a broad audience in the chromatin and virology fields as reviewers 1 and 3 also pointed out. Chromatin is generally analyzed in the context of how it might affect gene expression and the impact of chromatin on biological processes such as viral infections, and its structural role in the nucleus is not commonly considered. Here, we demonstrate an important example of the glaring effects of chromatin structure on the biological nuclear process of infection.

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

Lewis et al. reveal an unexpected role for heterochromatin formation in remodeling the nucleus to facilitate egress of the nuclear-replicating virus HSV1. By performing TEM in HSV1-infected primary human fibroblasts, the authors show that capsids accumulate at the inner nuclear membrane in regions of less densely stained heterochromatin, in agreement with studies in established cell lines. The authors go on to reveal that heterochromatin in the nuclear periphery of HSV1-infected primary fibroblasts was dependent on the histone variant macroH2A1 and is enriched with H3K27me3.CUT & Tag was used to profile macroH2A1 over time during lytic HSV1 infection and showed that both macroH2A1 and H3K27me3 were enriched over newly formed heterochromatic regions 10s-100s of Kb in length in active compartments. Remarkably, loss of macroH2A1 or H3K27me3 reduced released, cell free infection virus progeny and increased intranuclear capsid accumulation without detectably impacting the proportion of mature genome containing capsids, virus genome or protein accumulation. Their finding that newly remodeled heterochromatin forms in HSV infected cells and is a critical determinant for the association of capsids with the inner nuclear membrane is consistent with a critical role in egress.

I have only relatively minor editorial suggestions listed below to improve the manuscript:

Line 92: This subtitle should be revised to more precisely state the findings shown in the Fig 1 data. While the first part of the statement "HSV1 capsids associate with regions of less dense chromatin" is consistent with what is shown, the final phrase "...to escape the nucleus" is an interpretation of the data inferred from the static image.

We agree with the reviewer and have amended our text to more accurately describe the figure. See lines 138-139.

Line 96: I am not sure the statement that fibroblasts represent a "common" site of infection is supported by ref 15. FIbroblasts do, as indicated in ref 15, express the appropriate receptor(s) for virus entry and in culture support robust virus productive growth. However, in human tissue, infection of dermal fibroblasts appears rare, suggesting it may not be a "common" site of infection (PMCID: PMC8865408). Maybe simply revise wording to indicate fibroblasts represent "a site of infection or can be infected in tissue?".

We agree with the reviewer, as was also pointed out by reviewer 2, and have amended the text. See lines 109.

Line 126-127: As written it states that "....regions of the host genome that increase during infection", implying these genome regions are amplified (increase). I think the authors mean that infection increases binding of mH2A1 and H3K27me3 to broad regions of the host genome. Please clarify.

We agree with the reviewer that this was written ambiguously. As was pointed out by reviewers 1 and 2, the increase in these marks depends on the type of measurement. Therefore, we have modified the text in a revised manuscript to focus instead on the redistribution of these marks during infection. See line 138-139.

FIgS1, a,b,c,d: please indicate that 4,8,12 indicate hpi, correct? And indicate that in the legend M indicates Mock.

This is correct and we have updated this in the figure legend. See lines 625-627.

Line 197: "active compartments". Do the authors mean transcriptionally active compartments? Please clarify

This is correct and have clarified this in the text. See line 248.

Line 232: please replace "productive" with "infectious"

We agree with the reviewer and have amended our text. See line 295.

Line 233 - The authors conclude mH2A1 is important for egress, ruling out assembly before even bringing it up. As I read on, it is clear the authors addressed this important issue later on in the manuscript. That said, it was a bit jarring to conclude egress is important without addressing the assembly possibility at this juncture in the manuscript. One way to remedy this would be to move the Fig S6 assembly/capsid type data (lines 286-297, Fig S6) and surrounding text earlier to support the conclusion that mH2A1 did not detectably influence assembly, but is important for egress.

*We agree with the reviewer that the order of presentation makes it difficult to follow. Our revised manuscript now includes these important data within the same figure. See new Figure 5. *

Line 244: "progeny production" - it would be helpful to specify "cell free or released infectious virus progeny"

Line 248: change "produced" to released"

Line 273 replace "productive" with "infectious virus progeny released from infected cells"

Fig S5c: Was the plaque assay performed on cell free supernatants? This should be indicated.

We agree with the reviewer and have made all these changes in the text. See lines 285-287.

Reviewer #3 (Significance (Required)):

The experiments are well executed, the data are solid with appropriate statistical analysis and their analysis sufficiently rigorous, and the manuscript is clearly written. Moreover, the finding that HSV manipulates host heterochromatin marks to facilitate nuclear egress is significant and exciting. The work reveals an unexpected role for newly assembled heterochromatin in egress of nuclear replicating viruses like HSV1.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Lewis et al. reveal an unexpected role for heterochromatin formation in remodeling the nucleus to facilitate egress of the nuclear-replicating virus HSV1. By performing TEM in HSV1-infected primary human fibroblasts, the authors show that capsids accumulate at the inner nuclear membrane in regions of less densely stained heterochromatin, in agreement with studies in established cell lines. The authors go on to reveal that heterochromatin in the nuclear periphery of HSV1-infected primary fibroblasts was dependent on the histone variant macroH2A1 and is enriched with H3K27me3.CUT & Tag was used to profile macroH2A1 over time during lytic HSV1 infection and showed that both macroH2A1 and H3K27me3 were enriched over newly formed heterochromatic regions 10s-100s of Kb in length in active compartments. Remarkably, loss of macroH2A1 or H3K27me3 reduced released, cell free infection virus progeny and increased intranuclear capsid accumulation without detectably impacting the proportion of mature genome containing capsids, virus genome or protein accumulation. Their finding that newly remodeled heterochromatin forms in HSV infected cells and is a critical determinant for the association of capsids with the inner nuclear membrane is consistent with a critical role in egress.

I have only relatively minor editorial suggestions listed below to improve the manuscript:

Line 92: This subtitle should be revised to more precisely state the findings shown in the Fig 1 data. While the first part of the statement "HSV1 capsids associate with regions of less dense chromatin" is consistent with what is shown, the final phrase "...to escape the nucleus" is an interpretation of the data inferred from the static image.

Line 96: I am not sure the statement that fibroblasts represent a "common" site of infection is supported by ref 15. FIbroblasts do, as indicated in ref 15, express the appropriate receptor(s) for virus entry and in culture support robust virus productive growth. However, in human tissue, infection of dermal fibroblasts appears rare, suggesting it may not be a "common" site of infection (PMCID: PMC8865408). Maybe simply revise wording to indicate fibroblasts represent "a site of infection or can be infected in tissue?".

Line 126-127: As written it states that "....regions of the host genome that increase during infection", implying these genome regions are amplified (increase). I think the authors mean that infection increases binding of mH2A1 and H3K27me3 to broad regions of the host genome. Please clarify.

FIgS1, a,b,c,d: please indicate that 4,8,12 indicate hpi, correct? And indicate that in the legend M indicates Mock.

Line 197: "active compartments". Do the authors mean transcriptionally active compartments? Please clarify

Line 232: please replace "productive" with "infectious"

Line 233 - The authors conclude mH2A1 is important for egress, ruling out assembly before even bringing it up. As I read on, it is clear the authors addressed this important issue later on in the manuscript. That said, it was a bit jarring to conclude egress is important without addressing the assembly possibility at this juncture in the manuscript. One way to remedy this would be to move the Fig S6 assembly/capsid type data (lines 286-297, Fig S6) and surrounding text earlier to support the conclusion that mH2A1 did not detectably influence assembly, but is important for egress.

Line 244: "progeny production" - it would be helpful to specify "cell free or released infectious virus progeny"

Line 248: change "produced" to released"

Line 273 replace "productive" with "infectious virus progeny released from infected cells"

Fig S5c: Was the plaque assay performed on cell free supernatants? This should be indicated.

Significance

The experiments are well executed, the data are solid with appropriate statistical analysis and their analysis sufficiently rigorous, and the manuscript is clearly written. Moreover, the finding that HSV manipulates host heterochromatin marks to facilitate nuclear egress is significant and exciting. The work reveals an unexpected role for newly assembled heterochromatin in egress of nuclear replicating viruses like HSV1.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

The manuscript "HSV-1 exploits heterochromatin for egress" describes the effects of heterochromatin at the nuclear periphery, macroH2A1 or H3K27me3 on HSV-1 replication and egress. Knocking out macroH2A1 or depleting H3K27me3 with high concentrations of tazemetostat depleted heterochromatin at the nuclear periphery, may not have affected HSV-1 protein expression and modestly inhibited the production of cell-free infectivity and HSV-1 genomes. macroH2A1 deposition was affected by infection, creating new heterochromatin domains which did not correlate directly with the levels of expression of the genes in them. The authors conclude that heterochromatin at the nuclear periphery dependent on macroH2A1 and H3K27me3 are critical for nuclear egress of HSV-1 capsids.

The experiments leading to the conclusion that HSV-1 capsids egress the nucleus through channels in the peripheral chromatin confirm previously published results (https://doi.org/10.1038/srep28844). The previously published EM micrographs show a much larger number of nuclear capsids, more consistent with the images in the classical literature, even in conditions when nuclear egress was not inhibited. Figures 1 and 4 show scarce nuclear capsids, even under the conditions when nuclear egress should be inhibited according to the model and analyses. The large enrichment in nuclear capsids in KO cells predicted by the model is not reflected in figure 4a, which shows only a modest increase in nuclear capsid density (the total number of nuclear capsids would be more informative). The number or density of nuclear capsids is not shown in H3K27 "depleted" cells. The robustness of the analyses of the number of capsids at the membrane in H3K27 "depleted" cells is unclear. For example, the analyses could be repeated with different cut offs, such as 2 or 4. If they are robust, then the conclusions will not change when the cutoff value is changed.

The quantitation of the western blots present no evidence of reproducibility and/or variability. The number of biologically independent experiments analyzed must be stated in each figure and the standard deviation must be presented. As presented, the results do not support the conclusions reached. The quality of western blots should also be improved. it is unclear why figure 2b shows viral gene expression in wild-type cells only, and not in KO or H3K27me3 depleted cells, which are only shown in the supplementary information. These blots presented in Figure S5a and S5b are difficult to evaluate as the signal is rather weak and the controls appear to indicate different loading levels. These blots do not appear to be consistent with the conclusions reached. Some blots (VP16, ICP0 in HFF) appear to indicate a delay in protein expression whereas others (VP16, ICP0 in RPE) appear to indicate earlier expression of higher levels. The claimed "depletion of H3K27me3 is not clear in in figure S5d, in which the levels appear to be highly variable in all cases, without a consistent pattern, with no evidence of reproducibility and/or variability, and using a mostly cytoplasmic protein as loading control. All western blots should be repeated to a publication level quality, the number of independent experiments must be clearly stated in each figure, and the reproducibility and/or variability must be indicated by the standard deviation. An enhanced analyses of the RNA-seq data, analyzing all individual genes rather than pooling them together, would provide better support to these conclusions. Then, the western blots are useful to show that the changes in mRNA result in changes in the levels of selected proteins.

Figure S1 raises some questions about the specificity of the macroH2A1 antibody used for CUT&Tag. As expected CUT&Tagging the cellular genome in the KO cells with the specific antibody results in lower signal than with the IgG control antibody. In contrast, viral DNA is CUT&Tagged as efficiently in the KO as in the WT cells, and in both cases significantly above the IgG controls. The simplest interpretation of these results is that the antibody cross-reacts with a protein that binds to HSV-1 genomes. The manuscript must experimentally address this possibility.

Also, Figure S1 shows that the viral genome is CUT&Tag'ed with H3K27me3 antibody as efficiently in macro H2A1 WT and KO cells, and in both cases above the background signal from IgG control antibody. The authors conclude that the signal with the specific antibody "mirrors" that of the control antibody, but "mirroring" is not defined and the actual data show that there is a large increase in signal with the specific antibody. Not surprisingly, the background signal also increases, as the number of genomes increase while infection progresses. The authors conclude that "these results indicated that there was a significant background signal from the viral genome that could not be accounted for", but no evidence supporting this conclusion is presented. The data show clear signal above the background from the viral genome and that this signal is not affected by the presence or absence of macroH2A1. This section of the manuscript has to be thoroughly re-analyzed as there is clear H3K27 signal.

The concentration of tazemetostat used is high. Normally, concentrations of around 1µM are used in cells, and 10µM is often cytotoxic (for example https://doi.org/10.1038/s41419-020-03266-3; https://doi.org/10.1158/1535-7163.MCT-16-0840). The effects on H3K27me3 presented in figure S1b appear to be normalized to mock infected treated cells. If so, they do not allow to evaluate the effectivity of the treatment. Cell viability after the four days treatment must be evaluated, the claimed "depletion" of H3K27me3 must be clearly demonstrated (the blots in figure S5 are not sufficient as presented), and levels of different histone methylations must be tested to support the claimed specificity of tazemetostat for H3K27me3 at the high concentrations used.

Minor comments.

Reference No.27 is misquoted in lines 250-251, which state that it shows that "HSV-1 titers, but not viral replication, where reduced upon EZH2 inhibition." The reference actually shows inhibition of HSV-1 infectivity, DNA levels and mRNA for ICP4, ICP22 and ICP27. This reference uses much shorter treatments (12 h and only after infection). It also shows that inhibition of EZH2/1 up regulates expression of antiviral genes.

HFF are primary human cells but they are fibroblasts whereas the primary target of HSV-1 replication is epithelial cells. The wording used "they represent a common site of infection in humans" must be edited

Disruption of macroH2A (1 and 2) results in general defects in nuclear architecture, not just peripheral chromatin (https://doi.org/10.1242/jcs.199216;, see also figure 1c and 5a, presenting invaginated and lobulated nuclei). The manuscript would benefit from including a broader discussion of the effects of macroH2A defects on the general nuclear architecture.

The title should be edited, as "egress" in virology is commonly used to refer to the egress of virions from the cell, not to the nuclear egress of capsids. Adding the words nuclear and capsid should be sufficient to address this issue.

It is unclear why preferential changes in expression of housekeeping genes would indicate "stress responses to infection". The rationale for this conclusion must be fully articulated and supported.

Statistical methods must be fully described in materials and methods and the number of biologically independent experiments must be stated in each figure.

Significance

The major strengths of the manuscript lie on the comprehensive analyses of the effects of knocking histone macroH2A in the nuclear architecture and chromatin organization. These analyses indicate that peripheral heterochromatin is defective in the KO. Another strength lies on the analyses of the news heterochromatin domains in HSV-1 infected cells. The relationship between the lack of correlation between the changes in gene expression and global heterochromatin domains defined by macroH2A1 with the main conclusion is less clear.

The major weakness is that the data presented do not strongly support the conclusions. Additional experiments are required to support the main conclusion that the effects in peripheral heterochromatin result in a biologically significant effect on capsid egress. The authors should also consider that the additional experimentation may not support the conclusion that macroH2A or H3K27me3 play critical roles in the nuclear egress of capsids. Another major weakness is that the results of CUT&Tag of the viral genome are dismissed without proper justification. The authors conclude that the results invalidate the assays, but the results are consistent with cross-reactivity of the macroH2A1 antibody with another protein that interacts with the viral genomes and with H3K27me3 being associated with the viral genomes irrespectively of macroH2A1. If the authors had additional data supporting the claim that these results do not reflect cross-reactivity or association with the viral genomes, these data must be presented. Without that additional data, the conclusions are not supported and these discussions must be removed from the manuscript. The authors may still opt to not analyze any association with the viral genomes, but they should not dismiss them as artifactual without actual evidence to support this claim. Previously published literature is also misquoted.

This study makes an incremental contribution to the previously published evidence showing that HSV-1 capsids egress the nucleus through channels in between the peripheral chromatin. It shows that disruption of the heterochromatin at the nuclear periphery, and the nuclear architecture in general, may have a modest effect on capsid egress. This information may be of interest mostly to a specialized audience focused on the egress of nuclear capsids.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary

This study by Lewis et al. examines the role of heterochromatin in the nuclear egress of herpesvirus capsids. They show that heterochromatin markers macroH2A1 and H3K27me3 are enriched at specific genome regions during the infection. They also show that when macroH2A1 is removed or H3K27me3 is depleted (both of which reduce the amount of heterochromatin at the nuclear periphery), the capsids are not able to egress as effectively. This is interesting since it could be argued that heterochromatin acts as a hindrance to the transport of viral capsids to the nuclear envelope and that the loss of it would allow capsids to reach the nuclear envelope more easily. However, this paper seems to show that heterochromatin formation, on the contrary, is necessary for efficient egress. Overall, the study seems comprehensive. The methodology is solid, and the experiments are very well controlled. However, some issues need to be addressed before publication.

Major comments

1. In line 49, the authors state, "Like most DNA viruses, herpes simplex virus (HSV-1) takes advantage of host chromatin factors both by incorporating histones onto its genome to promote gene expression and by reorganizing host chromatin during infection". In addition, HSV1 expression can be hindered by the host's interferon response via histone modifications. Ref. Johnson KE, Bottero V, Flaherty S, Dutta S, Singh VV, Chandran B. IFI16 restricts HSV-1 replication by accumulating on the HSV-1 genome, repressing HSV-1 gene expression, and directly or indirectly modulating histone modifications. PLoS Pathog. 2014 Nov 6;10(11):e1004503. doi: 10.1371/journal.ppat.1004503. Erratum in: PLoS Pathog. 2018 Jun 6;14(6):e1007113. PMID: 25375629; PMCID: PMC4223080.
2. Reference 5 is misquoted in the sentence, "This redistribution of host chromatin results in a global increase in heterochromatin5". In that reference, the amount of heterochromatin is not analyzed in any way. However, that particular paper shows that the transport of capsid through chromatin is the rate-limiting step in nuclear egress, which is important considering this study. Further, the article by Aho et al. shows that when the infection proceeds capsids can more easily traverse from the replication compartment into the chromatin, which means that infection can modify chromatin for easier capsid transport. For that reason, the article is an important reference, but it needs to be cited correctly.
3. The term heterochromatin channel at lines 54, 102, and 303 is misleading since the channels seen in the original referred paper are less dense chromatin areas. Also, this term is not used in the original paper where the phenomenon was first described. These less dense interchromatin channels were found by soft-X-ray tomography imaging and analyses, not by staining.
4. It is difficult to visualize chromatin using TEM microscopy. The values of peripheral chromatin thickness given in Figure 1e (5-15 nm) do not seem realistic given that the thickness of just one strand of histone-wrapped DNA is 11 nm. Why are the two values for WT (in the top and bottom parts) different? If you can get so different values for WT, it is a bit worrisome (switching the WT results between the top and bottom parts of Fig. 1e would for example result in very different conclusions on the effect of macroH2A1 KO for the thickness of the chromatin layer).
5. In lines 134-137 it says that "The enrichment of macroH2A1 and H3K27me3 was observed as large domains that were gained upon viral infection (Fig 2a), suggesting that the host landscape is altered upon infection. These gains were reflected in an increase in total protein levels measured by western blot (Fig 2b)." However, the protein levels of H3K27me3 do not seem to increase during infection. In other presented data as well (Figs. 2a, 2b, 2c, S2a) it is difficult to justify the statement that H3K27me3 is enriched in infection. When this is the case, the conclusion that the amount of heterochromatin increases in the infection (the quotation above and the one in line 315) is not supported. The statement in line 315 is also not specific since it is unclear what "newly formed heterochromatin increases" means.
6. Quantitation of viral capsid location in H3K27me3-depleted cells seems somewhat arbitrary. It would have been more robust to calculate the number of capsids per unit length of the nuclear envelope with and without depletion.
7. In lines 300-302 it says "Elegant electron microscopy work showed that HSV-1 infection induces host chromatin redistribution to the nuclear periphery2,8." However, the redistribution data in reference 8 is based on soft x-ray tomography and not on electron microscopy."
8. The authors bundle together the effects of macroH2A1 removal and H3K27me3 depletion by saying that they both decrease the amount of heterochromatin at the nuclear periphery and therefore hinder capsid egress. This seems overly simplistic and macroH2A1 and H3K27me3 seem to act very differently, which is manifested in the drastic difference in nuclear capsid localization between the two cases. This difference needs to be discussed more.

Major comments

Line 45: Nuclear replicating viruses -> Nuclear-replicating viruses

Line 56: is -> are

Line 64: 25kDa -> 25 kDa

Line 159: macroH2A1 cells -> macroH2A1 KO cells

Line 289: The term gDNA is rarely used for viral DNA. Replace gDNA with viral DNA.

Line 405: 8hpi -> 8 hpi

Line 449: mm2 -> μm2 "Scale bar as indicated" words can be removed in the figure legends or at least should not be repeated many times within one figure legend.

Significance

These findings would appeal to a broad audience in the field of virology. Specifically, the researcher in the fields of virus-cell and virus-nucleus interactions. This manuscript analyses herpesvirus-induced structural changes in the chromatin structure and organization in the nucleus that are also likely to affect the intranuclear transport of viral capsids.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

__We thank the reviewers for their detailed comments, constructive feedback and insightful suggestions to improve our research and manuscript. Our detailed responses are in bold. __

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

Summary This paper by Styles et al describes the effect of propylene glycol (PG) on a range of enveloped viruses, focusing mainly on IAV and SARS-Cov2. The direct virucidal effect is shown both in vitro and in vivo, as well as the effect of vaporized PG on airborne IAV and SARS-Cov-2. This is a well-performed and very clear study.

Major comments

The claims and conclusions are in general well supported by the data, yet I have some comments.

• Interestingly, there were variations in the effect of PG on the different pseudoviruses. Surprising as I assumed that the effect of PG would be mainly on the lipid envelope, as you also mention in the discussion. It is suggested that PG therefore also has an effect on the conformation of the surface proteins. This can be easily tested, does the addition of PG result in a changed recognition by antibodies targeting these surface proteins? This information would be very valuable for your manuscript.

__Many thanks for this excellent suggestion. To address this we are working in collaboration with Dr Joe Grove (Centre for Virus Research, University of Glasgow) using immobilised purified pseudovirus particles +/- PG treatment. Immunostaining against pseudovirus glycoproteins and their HIV gagpol capsid core enables comprehensive analysis of particle composition and integrity. Our preliminary experiments since submission have shown treatment with >60% PG increased gagpol signal, indicating compromised envelope integrity of equivalent magnitude to direct permeabilisation (positive control condition). At PG concentrations between 25 – 60% we’ve observed a significant reduction in glycoprotein signal preceding substantial envelope disruption. We are repeating these experiments in parallel with infectivity assays to assess how these changes in glycoprotein antibody recognition and envelope permeabilization relate to pseudovirus entry. We anticipate these studies will be completed within 8 weeks. __

__In addition, we are optimising cryo-EM protocols for analysis of SARS-CoV-2 virus particles +/- PG treatment to directly visualise the impact of PG on virus structure, in collaboration with Dr Paul Simpson (EM Centre, Imperial College). __

• Figure S1 shows that although the clinical scores are better, the viral load is not reduced by the addition of PG to the inoculum. This is not mentioned in the main text. This seems very relevant information and should be presented in the main text. Can you please elaborate on this.

We agree that the virological and immunological findings are of interest, and have added the following to the main text “____Analysis of the remaining mice showed PG inhalation reduced clinical score and bronchioalveolar lavage cell counts on day 5, despite equivalent nasal and airway viral loads at this late time post-infection.” Examining longitudinal weight loss and survival was the primary objective of this experimental design, which only allowed analysis of BAL cell counts and viral loads in the remaining mice at the experimental end point (day 5 post-infection). As 3 mice in the IAV only group reached the severity limit on day 3, we did not have the statistical power to detect significant differences in clinical scores, lung infiltrates or viral loads. IAV burden in the nose and lung usually reach peak levels at day 2/3 post-infection, and further experiments to look at the magnitude of viral replication and immunopathology over the course of infection in PG-treated mice will be important for future translational studies. As such, we have added the following to the Fig. S1 legend “____Further investigations are required to determine whether PG treatment reduces inflammatory cell infiltrates or peak IAV loads on days 2/3 post-infection.”

• Optional: It would be very interesting to see the effect of vaporized PG on viral transmission between mice/ferrets of a highly infectious virus (eg Sendai virus). This would give an idea if the PG also works on more biological respiratory droplets.

We wholeheartedly agree. To perform these additional studies requires additional funding and project licence amendments so unfortunately must lie beyond the scope of this study.

Minor comments

• The caption of Fig1D mentions that the mice are inoculated with 50ml total volume. I presume this should be µl.

__Absolutely, thank you for spotting this error and it is now corrected. __

• Please put the axis of the viral titer graph now in Fig s1 in PFU/lung or nasal cavity? This would be more informative.

__Many thanks for this suggestion, now changed to PFU/lung along with BAL cells/lung. __

• Typo in 'PG has broad-spectrum virucidal activity' paragraph: titre should be titer.

__Titre is the British English spelling whereas titer is the American English spelling. We will comply with the journal specification on American versus British English in the final text. __

• Is allergy to PG an issue that could have an effect on further development of this antiviral strategy?

A recent review (Pemberton & Kimber ____https://doi.org/10.1016/j.yrtph.2023.105341____) of the evidence on this topic concludes “..the weight of evidence points to PG possessing no, or only extremely modest, skin sensitising properties.” There is anecdotal evidence of PG potentially acting as a very weak allergen in humans with underlying/pre-disposing skin conditions. Further clinical development of PG as a virucide would need to balance its significant benefits against this when “..the risk for sensitisation to PG on uncompromised skin seems to be extremely low.”

Reviewer #1 (Significance (Required)):

This study is highly relevant. It proposes an easy and possibly efficacious method to prevent airborne transmission of a broad range of (enveloped) viruses.

A strength of this study is the set-up with the virus transmission tunnel. This nicely shows the effect of the vaporized PG on airborne IAV and SARS-Cov-2. A limitation is that the researchers did not analyze non-enveloped viruses, which should be done in a follow-up study.

__Many thanks for these comments. Since submission we have extended our investigations to non-enveloped rotaviruses in collaboration with Dr Alex Borodavka (Department of Biochemistry, University of Cambridge). Preliminary experiments have shown that rotavirus infectivity decreased from ~108 PFU/mL to undetectable levels (2 PFU/mL) following incubation with 50% PG at 37oC for 1h compared to 0% PG control. We are now testing different temperatures, PG concentrations and incubation durations and will include this new data in our revised submission. __

There does not seem to be any recent research investigating this compound as an antiviral tool. Although the compound was tested for its antimicrobial properties in the 1940s, this has not been continued for unclear reasons (?).

__We speculate the rapid development and clinical testing of conventional antibiotics during this period (culminating the Nobel prize for Medicine for Flemming, Florey and Chain in 1945) is responsible for the abrupt cessation of research into PG as an antimicrobial despite its obvious efficacy. We were unaware of these studies when we commenced our research into PG and this literature seems to have flown completely under the radar. __

This is very valuable research in this world recovering from the recent SARS-Cov-2 pandemic. More mechanistic studies need to be performed to understand how this compound exerts its antiviral and antimicrobial effect.

A more specialized audience will be interested in this study.

My field of expertise: RSV, PIV, antivirals

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

SUMMARY: This study investigates the inactivation potential of propylene glycol against a number of important human respiratory viruses, including influenza A virus and SARS-CoV-2. Potent antiviral activity of propylene glycol was demonstrated for viruses in bulk solutions, in 2uL droplets on a variety of surfaces, and within aerosols. Additionally, propylene glycol was added to viruses prior to infection of live animals, and this treatment significantly improved animal survival outcomes and reduced their weight loss and clinical scores compared to animals infected with non-treated virus. Levels of propylene glycol utilised in the study were substantially less than those classified as well-tolerated by mammals. Overall, authors demonstrate that propylene glycol is a strong yet safe virucide that could be used to control the spread of viral infections.

MAJOR COMMENTS:

Claims and conclusions are very well-supported by the data presented. Both the data presentation and method details are sufficient to reproduce the experiments. Authors might like to add precise gram concentrations of PG in addition to the percentages listed in the manuscript, either g/L or g/kg as appropriate. This would also allow easier comparison of the tested PG does to those published as safe doses for mammals, which are often given as g/kg.

Many thanks for this suggestion. These calculations and the concentration are now in a new “supplementary calculations” section in the supplementary material and the Figure 1 legend directs readers to them: “____See supplementary materials for % PG solution (v/v) to g/L conversion and mouse PG dose in g/kg.”

Replicates are more than adequate, and statistical analysis is mostly robust. I have one concern of statistical analyses performed in Figure 2D. In the SARS1 coronavirus panel of this figure, statistical analysis indicates that 10% PG causes significantly more infection (in RLU) than 0% PG. The same is true for the SARS-CoV-2 Alpha panel. The differences in infection level between 0% and 10% PG by eye are minimal, and authors should consider whether this statistical test is the most robust/appropriate, and whether these particular statistical differences are biologically meaningful to report. If analyses remain as is, authors should postulate why this small increase in infection level was observed for some lentivirus pseudotypes at 10% PG compared to controls.

__Many thanks to the reviewer for highlighting this. Whilst Dunnet's post-test we employed does indeed suggest a significant difference between 10% and 0% PG for some of the pseudoviruses, another post-test (Bonferroni's) does not. The reviewer correctly questions the biological significance of these very small differences in luciferase activity. In our experience, effect sizes of less than half a log unit are not biologically meaningful and certainly we make no claims that 10% PG is affecting pseudovirus entry in these experiments. To avoid any ambiguity therefore, we have removed the "*" from the relevant graphs and have restricted reporting of statistical significance to those conditions with p

OPTIONAL - The study could be further enhanced by preliminary mechanistic investigation into the action of propylene glycol (PG) - for example, authors postulate that PG may act via disruption of the lipid membrane, and this could be supported by EM imaging of a few PG-treated viruses. Alternatively, non-enveloped viruses could be tested for sensitivity to determine the importance of the envelope. Authors mention this in the discussion as being outside the scope, though testing one or two non-enveloped viruses would elevate the work substantially, as it opens the possibility for PG to also protect against non-enveloped respiratory viruses such as rhinovirus, as well as enteric viruses.

__As explained in detail in our response to reviewer 1 and our revision plan, we are addressing these excellent suggestions in three ways for the revised manuscript: __

• __Collaboration with Dr Joe Grove’s group to address PG's action on viral surface glycoproteins, virus composition and envelope integrity. __
• EM imaging of PG-treated SARS-CoV-2 with Dr Paul Simpson.
• Collaboration with Dr Alex Borodavka’s group to investigate PG-mediated virucidal activity against non-enveloped rotaviruses.

  OPTIONAL - All influenza work in this study used the lab-adapted isolate A/PR8. This is a very useful and representative virus, though the study may benefit from comparison of the PG-mediated inactivation kinetic of A/PR8 with that of a more recent clinical isolate of influenza A. This could be a useful supplementary figure.

__Many thanks for this suggestion. We agree that examining the inactivation kinetics of different flu strains in vitro would be informative and will endeavour to include this in our revised manuscript if these experiments can be performed within an appropriate time frame. We know PG is effective against ____the 2009 pandemic strain H1N1 A/California/7/2009 as this was used for our in vivo experiments as listed in the methods section. For clarity we have added this information into the main text. __

MINOR COMMENTS:

Could authors generate (or reference) a robust correlation curve between IAV plaque area (px^2) that is calculated by the ColonyArea plugin, in relation to actual PFU/mL as counted by eye? This would help the reader to understand the decrease in viral titre that PG mediates from experiments conducted in the aerosol tunnel, e.g. does a decrease from 10^5 px^2 down to 10^3 px^2 equate to a 2-log10 reduction in PFU, or it is less than this?

This correlation could also be used to generate a detection limit for Plaque Area (px^2) that is equivalent to 1 or 10 PFU/well, which could then be added to graphs in Figure 3, S4, S5. This will help the reader understand if detection of 10^3 px^2 (for example) is quite high, or if it is already close to detection limits.

__The reviewer makes an interesting suggestion. As can be seen from the requested representation of those data (reviewer figure 1, replotted from Fig 3D), for lower nebulised input virus amounts where infection did not transmit beyond the first plate there is a very respectable linear correlation between plaque number and plaque area (R2=0.72, Spearman’s r=0.98, p2 – 103 px2 reflects the 1 – 10 plaque detection limit. However, when performing initial controls we found this relationship soon departs from linearity as input virus amount increases; essentially plaques begin to overlap with each other on the plates closest to the nebuliser. We are concerned it might be inappropriate to present PFU/well extrapolated from plaque areas collected from lower input virus experiments, since this could potentially mislead or confuse the reader about the fidelity and nature of the transmission tunnel experimental system. Moreover, for this particular type of assay plaque area affords a much greater dynamic range than PFU/well, which is essential for demonstrating the range of PG’s virucidal activity with higher input virus, range of vapor concentrations and over different transmission distances. Taking these considerations into account, and unavoidable idiosyncrasies of testing virus transmission under more “natural” conditions (e.g. direct virus deposition on cell monolayers prohibiting serial dilution for exact PFU determination), we have decided to continue reporting plaque area only for quantification of transmission tunnel data - the parameter that was actually measured. This accurately reflects the extensive cell clearance under control conditions, where the monolayer is completely disrupted, compared to the protective effect of PG vapor, where the monolayer is largely intact with limited plaques. We hope the reviewer understands our reasoning. __

To communicate this, we have added the following sentence to the methods section:

__“Plaque area shows a strong linear correlation with plaque number at lower viral inputs (R2=0.72, Spearman’s r=0.98, p

Can authors specify if the PEG was mixed with virus immediately prior to inoculation of animals, or was it mixed ahead of time (e.g. 30 mins or 1 hour ahead of administration)? In this case, low-level inactivation of viruses in bulk solution may have occurred. Figure 1C shows minimal inactivation should have occurred with 20% PG over this time-course at room-temperature, but this detail would be helpful to specify.

__Many thanks for bringing this oversight to our attention. IAV was mixed with PG immediately prior to inoculation of animals (

Figure legend of Fig 1. D) currently reads "... inoculated with 50mL..." , should this be 50uL?

__It should, thank you for spotting this error and it is now corrected. __

Reviewer #2 (Significance (Required)):

This study offers a robust proof of concept that PG can be applied to viral disinfection of a large range of enveloped viruses in a variety of settings (aerosol, fomites, in vivo, etc). This work now requires follow-on clinical studies to establish optimal and safe PG levels for viral inactivation in real-world settings, and comprehensive testing of safety risks from long term PG-exposure before this virucide can become clinically applied. This impact of this primary research could be considered broad and translational.

The audiences anticipated to be interested in these results are varied, and include those in virology fields, medical personnel, aerosol scientists, disinfection experts, and public health/ policy makers.

Field of expertise of reviewer: virology, animal infections, immunology, aerosol-borne influenza virus.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary

This study investigates the inactivation potential of propylene glycol against a number of important human respiratory viruses, including influenza A virus and SARS-CoV-2. Potent antiviral activity of propylene glycol was demonstrated for viruses in bulk solutions, in 2uL droplets on a variety of surfaces, and within aerosols. Additionally, propylene glycol was added to viruses prior to infection of live animals, and this treatment significantly improved animal survival outcomes and reduced their weight loss and clinical scores compared to animals infected with non-treated virus. Levels of propylene glycol utilised in the study were substantially less than those classified as well-tolerated by mammals. Overall, authors demonstrate that propylene glycol is a strong yet safe virucide that could be used to control the spread of viral infections.

Major comments

Claims and conclusions are very well-supported by the data presented. Both the data presentation and method details are sufficient to reproduce the experiments. Authors might like to add precise gram concentrations of PG in addition to the percentages listed in the manuscript, either g/L or g/kg as appropriate. This would also allow easier comparison of the tested PG does to those published as safe doses for mammals, which are often given as g/kg.

Replicates are more than adequate, and statistical analysis is mostly robust. I have one concern of statistical analyses performed in Figure 2D. In the SARS1 coronavirus panel of this figure, statistical analysis indicates that 10% PG causes significantly more infection (in RLU) than 0% PG. The same is true for the SARS-CoV-2 Alpha panel. The differences in infection level between 0% and 10% PG by eye are minimal, and authors should consider whether this statistical test is the most robust/appropriate, and whether these particular statistical differences are biologically meaningful to report. If analyses remain as is, authors should postulate why this small increase in infection level was observed for some lentivirus pseudotypes at 10% PG compared to controls.

OPTIONAL - The study could be further enhanced by preliminary mechanistic investigation into the action of propylene glycol (PG) - for example, authors postulate that PG may act via disruption of the lipid membrane, and this could be supported by EM imaging of a few PG-treated viruses. Alternatively, non-enveloped viruses could be tested for sensitivity to determine the importance of the envelope. Authors mention this in the discussion as being outside the scope, though testing one or two non-enveloped viruses would elevate the work substantially, as it opens the possibility for PG to also protect against non-enveloped respiratory viruses such as rhinovirus, as well as enteric viruses.

OPTIONAL - All influenza work in this study used the lab-adapted isolate A/PR8. This is a very useful and representative virus, though the study may benefit from comparison of the PG-mediated inactivation kinetic of A/PR8 with that of a more recent clinical isolate of influenza A. This could be a useful supplementary figure.

Minor comments

Could authors generate (or reference) a robust correlation curve between IAV plaque area (px^2) that is calculated by the ColonyArea plugin, in relation to actual PFU/mL as counted by eye? This would help the reader to understand the decrease in viral titre that PG mediates from experiments conducted in the aerosol tunnel, e.g. does a decrease from 10^5 px^2 down to 10^3 px^2 equate to a 2-log10 reduction in PFU, or it is less than this?

This correlation could also be used to generate a detection limit for Plaque Area (px^2) that is equivalent to 1 or 10 PFU/well, which could then be added to graphs in Figure 3, S4, S5. This will help the reader understand if detection of 10^3 px^2 (for example) is quite high, or if it is already close to detection limits.

Can authors specify if the PEG was mixed with virus immediately prior to inoculation of animals, or was it mixed ahead of time (e.g. 30 mins or 1 hour ahead of administration)? In this case, low-level inactivation of viruses in bulk solution may have occurred. Figure 1C shows minimal inactivation should have occurred with 20% PG over this time-course at room-temperature, but this detail would be helpful to specify.

Figure legend of Fig 1. D) currently reads "... inoculated with 50mL..." , should this be 50uL?

Significance

This study offers a robust proof of concept that PG can be applied to viral disinfection of a large range of enveloped viruses in a variety of settings (aerosol, fomites, in vivo, etc). This work now requires follow-on clinical studies to establish optimal and safe PG levels for viral inactivation in real-world settings, and comprehensive testing of safety risks from long term PG-exposure before this virucide can become clinically applied. This impact of this primary research could be considered broad and translational.

The audiences anticipated to be interested in these results are varied, and include those in virology fields, medical personnel, aerosol scientists, disinfection experts, and public health/ policy makers.

Field of expertise of reviewer: virology, animal infections, immunology, aerosol-borne influenza virus.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary

This paper by Styles et al describes the effect of propylene glycol (PG) on a range of enveloped viruses, focusing mainly on IAV and SARS-Cov2. The direct virucidal effect is shown both in vitro and in vivo, as well as the effect of vaporized PG on airborne IAV and SARS-Cov-2. This is a well-performed and very clear study.

Major comments

The claims and conclusions are in general well supported by the data, yet I have some comments. - Interestingly, there were variations in the effect of PG on the different pseudoviruses. Surprising as I assumed that the effect of PG would be mainly on the lipid envelope, as you also mention in the discussion. It is suggested that PG therefore also has an effect on the conformation of the surface proteins. This can be easily tested, does the addition of PG result in a changed recognition by antibodies targeting these surface proteins? This information would be very valuable for your manuscript. - Figure S1 shows that although the clinical scores are better, the viral load is not reduced by the addition of PG to the inoculum. This is not mentioned in the main text. This seems very relevant information and should be presented in the main text. Can you please elaborate on this.<br /> - Optional: It would be very interesting to see the effect of vaporized PG on viral transmission between mice/ferrets of a highly infectious virus (eg Sendai virus). This would give an idea if the PG also works on more biological respiratory droplets.

Minor comments

• The caption of Fig1D mentions that the mice are inoculated with 50ml total volume. I presume this should be µl.
• Please put the axis of the viral titer graph now in Fig s1 in PFU/lung or nasal cavity? This would be more informative.
• Typo in 'PG has broad-spectrum virucidal activity' paragraph: titre should be titer.
• Is allergy to PG an issue that could have an effect on further development of this antiviral strategy?

Significance

This study is highly relevant. It proposes an easy and possibly efficacious method to prevent airborne transmission of a broad range of (enveloped) viruses.

A strength of this study is the set-up with the virus transmission tunnel. This nicely shows the effect of the vaporized PG on airborne IAV and SARS-Cov-2. A limitation is that the researchers did not analyze non-enveloped viruses, which should be done in a follow-up study.

There does not seem to be any recent research investigating this compound as an antiviral tool. Although the compound was tested for its antimicrobial properties in the 1940s, this has not been continued for unclear reasons (?). This is very valuable research in this world recovering from the recent SARS-Cov-2 pandemic. More mechanistic studies need to be performed to understand how this compound exerts its antiviral and antimicrobial effect.

A more specialized audience will be interested in this study.

My field of expertise: RSV, PIV, antivirals

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer #1

1-1) Authors conclude that "MFN2 and MTCH2 compensate for each other's absence", but the compensation works only when they are overexpressed. On the other hand, upon HBSS or CHX treatment, MFN2 KO or MTCH2 KO cells have elongated/hyperfused mitochondria, but this is not observed in double deficient cells. In this case, MFN2 and MTCH2 show compensatory effects on mitochondria elongation. The authors believe the two conditions, unstressed & OE and stressed conditions, activating same molecular machineries, but it is not fully supported by the data. For example, in unstressed condition, MTCH2 OE can recover MFN2 KO but not of MFN1 KO, suggesting MTCH2-dependent mitochondria fusion requires MFN1, but it is not tested for stressed condition.

Furthermore, even if the machineries are common among stressed and unstressed conditions, authors should discuss why the endogenous MFN2/MTCH2 expression is not enough to activate mitochondria fusion in MTCH2/MFN2 KO cells, respectively, and how the compensatory effects are activated upon HBSS or CHX treatment.

__R: We thank the reviewer for this important comment. __

1) The reviewer’s comment is correct; in resting conditions the endogenous expression of MFN2 in MTCH2 KO and vice-versa are not sufficient to compensate for each other’s absence, since in steady state they both remain largely fragmented. Thus, we corrected the text accordingly by removing this concept. Interestingly, in the revised MS we show that MFN2 KO results in an increase in MTCH2 expression levels in the mitochondria and to a strong decrease in the GPAT3 and 4 expression levels in the ER (Rev Fig. 3B, C). These results suggest that: 1) Expression of MFN2 is important for the stabilization of GPAT3 and 4 (perhaps they form a functional complex together); 2) the expression levels of endogenous MTCH2 are possibly elevated to compensate for the decreased biosynthesis of LPA and increased demand of LPA funneling for mitochondrial fusion. This hypothesis was added to the discussion.

---

2) In the revised MS, we tested the possibility that MTCH2 overexpression together with mitochondrial fusion stress could enforce mitochondrial fusion in MFN1 KO cells, and found that it could not (Rev Supp Fig. 4 J-K).

---

3) We have also discussed what might be the compensatory effects of MTCH2 and MFN2 when activated upon enforced mitochondrial fusion induced by protein overexpression and stress (page 20 last paragraph).

1-2) Authors found out that ER-targeted MFN2 can rescue the mitochondria fragmentation in MTCH2 KO MEF cells, but mitochondria-targeted MFN2 has a lower effect than Wt MFN2. (Fig 2A&B). This finding suggests that MTCH2 loss might impair MFN2 localization at the ER. The authors should investigate endogenous MFN2 localization in MTCH2 KO MEFs.

R: We thank the reviewer for this insightful comment. We addressed this point and show that MTCH2 deletion does not change the expression levels or the subcellular localization of endogenous MFN2 (Rev Fig. 3B).

1-3) The analysis to test whether recovery of mitochondrial morphology by ER-targeted MFN2 in MTCH2 KO depends on LPA synthesis or not is missing (Fig 3G). Authors should examine whether mitochondrial elongation induced by ER-localized MFN2 in MTCH2 KO cells is impaired by the GPAT inhibitor.

R: This analysis was performed (we also included the other mutants), and presented the new data in the revised MS (Rev Supp Fig. 3N, O).

1-4) In the discussion section, authors suggest that mitochondrial LPA would be a crucial factor for MFN2 dependent mitochondrial fusion. To test this hypothesis, authors should overexpress mitochondrial GPAT and evaluate its effect on mitochondrial morphology.

R: We thank the reviewer for this important suggestion. We ordered two commercially available plasmids encoding GPAT1 to address this point (DNASU ____HsCD00082324 and HsCD00082324)____ but unfortunately the proteins were not expressed in our cells. In addition, we revised the text and changed the angle of the interpretation of these results, and clarified that inhibiting LPA impairs MTCH2-independent mitochondrial plasticity in response to MFN2 overexpression, rather than MFN2-dependent mitochondrial fusion.

1-5) In the discussion section, authors indicated that ER-targeted MFN2 could recover mito-ER contacts leading to LPA flux from ER to mitochondria and mitochondria elongation in MTCH2 KO. However, MTCH2 KO itself already have more mito-ER contacts (Fig 2D-H), and an artificial linker fails to recover mitochondria fragmentation in MTCH2 KO cells (Fig S2C, D). Thus, increased number of contacts appears not sufficient to recover the phenotype. The authors should consider this point in the discussion.

R: We added a comment on this important point in the results section (page 10, line 9)

2) Certain methods are not appropriate to support the stated conclusions.

2-1) Authors assess "mitochondria fusion" by evaluating mitochondrial morphology. The authors also describe mitochondrial clumping as a fusion-impaired phonotype (Fig 4A&B). Mitochondrial fusion should be evaluated using a PEG assay or a mtPA-GFP analysis.

R: We now provide in the revised MS results of a mtPA-GFP analysis done for MTCH2 KO MEFs exposed to GPATs inhibitor and treated with CHX (Rev Fig. 4G, H). This experiment supports the notion that loss of MTCH2 along with LPA synthesis inhibition largely impairs mitochondrial fusion in response to CHX.

2-2) In figure 2D-G, authors show that MTCH2 KO cells have more and longer mitochondria-ER contacts. The correct experiment is not to compare these cells to WT, but to KO reconstituted with MTCH2.

R: The reviewer is correct however it will take us many more months to generate a stable MTCH2 rescue cell line and to perform EM analysis of these cells, which would significantly slow down the revision of our MS.

__ __2-3) Since staining of MitoTracker depends on mitochondrial membrane potential, mitochondria with low potential would be invisible and excluded from the analysis. Authors should investigate mitochondrial morphology by immunostaining also in Fig 4D, S4A, D, and K.

__R: We agree with the reviewer’s comment, ____but re-doing all these experiments will be too labor-intensive and time consuming. We therefore focused on Supp Fig. 4A, in which the combination of FSG67 (GPATi in the revised MS) and HBSS treatment impaired mitochondrial membrane potential and mitochondria did not uptake the MitoTracker dye. We repeated this experiment using immunofluorescence, performed new quantifications, and incorporated the new data into the MS (Rev Fig. 4A). We did not remove Supp Fig. 4A since we wanted to emphasize the point that the combination of LPA synthesis inhibition and amino-acid deprivation results in loss of mitochondrial membrane potential. __

Minor comments

1) Since authors use FSG67, an inhibitor against GPAT1, 2 and 3, knocking down of each of GPATs will improve the significance of this work.

__ R: We thank the reviewer for suggesting these experiments. Since the contribution of mitochondrial GPATs to mitochondrial fusion was already established, we complemented our studies by silencing ER GPATs 3 and 4. We tested the contribution of ER-GPATs to MTCH2-independent mitochondrial fusion elicited by MFN2 overexpression (Rev Fig. 3L-N) and induced by either HBSS or CHX (Rev Fig. 4I-K).__

2) Recently, a paper about MTCH2 is published (Guna et al., Science), which shows its insertase activity on tail anchored proteins. Authors should include this point of view in the discussion.

R: We performed new experiments to address this important point. We evaluated the effect of MTCH2 deletion on the expression and localization of the fusion and fission proteins and on the LPA synthesis proteins, and found minor changes (Rev Supp Fig. 1D and Rev Fig. 3B, C). Thus, the effects we are seeing in the MTCH2 KO cells do not seem to be directly related to its insertase activity.

3-1) There are some discrepancies with the previous Labbé et al article.: Labbé et al. suggest that MTCH2 activity on mitochondria (from HCT116 cells) fusion is dependent on LPA, based on in vitro fusion assay. On the other hand, this manuscript shows that inhibition of LPA synthesis could not block MTCH2-induced mitochondria elongation in WT MEF and HEK293T cells. MTCH2 KO HCT116 cells are resistant to HBSS- but sensitive to CHX-induced mitochondria elongation. In this manuscript, MTCH2 KO MEF cells are sensitive to both stimuli, and only when MTCH2- and MFN2-deficient MEF cells are resistant to both stimuli. These discrepancies would be caused by difference of assay system or cell lines, but it is not clearly addressed.

R: We thank the reviewer for raising these discrepancies. ____Despite the discrepancies in the response of MTCH2 KO MEFs to HBSS, both manuscripts largely support the model that MTCH2 funnels LPA towards mitochondrial fusion sites. Our data suggests that loss of MTCH2 unmasks the requirement of sufficient LPA synthesis to sustain mitochondrial fusion, and we also show that MTCH2 overexpression can enforce mitochondrial fusion in the presence of GPATs inhibitor. These two results can be conciliated by the model that MTCH2 is able to optimize or funnel the levels of LPA towards the mitochondrial fusion sites, and when MTCH2 is absent and not able to catalyze this process, mitochondria rely on LPA levels to stay elevated to enable mitochondrial fusion but in a less efficient way. Interestingly, since we show that mitochondrial fusion enforced by HBSS shows full dependency on LPA synthesis, it is expected that loss of MTCH2 will have a stronger impact on HBSS-mediated mitochondrial fusion than on CHX. Nevertheless, MEFs clearly and consistently are sensitive to HBSS, yet this does not exclude that MTCH2 deletion in combination with HBSS treatment may be synergistically detrimental for the cell in other aspects of its well-functioning.

---

Reviewer #2

Specific points

1. The authors should comment on recently published work in Science that MTCH1 and likely MTCH2 are outer membrane insertases. The data in the manuscript seem consistent with the model that an undetermined protein may be poorly inserted in a MTCH2 knockout leading to reduced fusion activity mediated by MFN1. Could this explain how MTCH2 overexpression selectively restores fusion to MFN2 KO cells?

R: We thank the reviewer for raising this important point, and we have now performed new experiments to address it. We evaluated the effect of MTCH2 deletion on the expression and localization of the fusion and fission proteins and on the LPA synthesis proteins, and found minor changes (Rev Supp Fig. 1D and Rev Fig. 3B, C). Thus, the effects we are seeing in the MTCH2 KO cells do not seem to be directly related to its insertase activity.

The authors make the claim that "MFN2 and MTCH2 compensate for each other's absence" though this is not supported by their data. For example, MFN2 expression is not affected in an MTCH2 KO but cannot compensate to promote fusion. Rather, the authors find that MTCH2 and MFN2 are capable of promoting fusion when overexpressed in the absence of the other.

__R: We thank the reviewer for this important comment. __

The reviewer’s comment is correct, and in resting conditions the endogenous expression of MFN2 in MTCH2 KO cells and vice-versa are not sufficient to compensate for each other’s absence, since in steady state they both remain largely fragmented. Thus, we corrected the text accordingly by removing this concept. Interestingly, in the revised MS we show that MFN2 KO results in an increase in MTCH2 expression levels in the mitochondria and to a strong decrease in the GPAT3 and 4 expression levels in the ER (Rev Fig. 3B, C). These results suggest that: 1) Expression of MFN2 is important for the stabilization of GPAT3 and 4 (perhaps they form a functional complex together); 2) the expression levels of endogenous MTCH2 are possibly elevated to compensate for the decreased biosynthesis of LPA and increased demand of LPA funneling for mitochondrial fusion. This hypothesis was added to the discussion.

---

The authors rely heavily on claims that tagged MFN1 and MFN2 are fully functional, but is it possible that MFN1-GFP is only partially functional? Does untagged MFN1 overexpression cause mitochondrial fusion in a MTCH2 KO? The quantification of this is done only with aspect ratio, and not by categorization of mitochondrial morphology (Fig. S1). The authors should present both analyses in this and all other experiments in the manuscript, particularly since aspect ratio is only performed on 15 cells per condition and not on experimental replicates. The authors should clarify if sample identity was blinded prior to analysis.

__R: As suggested by the reviewer, we repeated all the MFN1 overexpression experiments using an untagged version of the protein, which was overexpressed in MFN1 KO, MTCH2 KO and MFN2 KO MEFs. The new data was included in Rev Fig. 1C-E. The data is consistent with our previous observations using MFN1-GFP. Also, mitochondrial morphology classification was added to the majority of the experiments presented in our MS, which represents quantification of three separate technical repetitions of the experiments (this analysis was not performed blinded). __

The constructs that form the basis of conclusions of ER versus mitochondrial-targeted MFN2 require additional controls to support robust conclusions. The immunofluorescence data suggests that MFN2-ACTA to some degree targets outside of mitochondria (Fig. 2A), and also that MFN2-YIFFT seems to localize to some degree to mitochondria (Fig. 2A). The YIFFT construct may also localize more prevalently to mitochondria in the presence of ACTA (Fig. S2A). Mistargeting would make interpretation of these experiments not possible, as these constructs must exclusively localize to their intended organelle to make strong conclusions. Triple labeling with ER and mitochondrial markers would be helpful, as well as western blots to confirm consistent expression levels and protein stability of each construct. The ACTA MFN2 also appears to promote fusion in a MTCH2 KO. How is this reconciled with the conclusion that ER-targeting of MFN2 is required?

R: We thank the reviewer for these important concerns. We repeated all the MFN2 mutant experiments, including the GTPase mutant MFN2 K109A, and included ER labelling to the MFN2-IYFFT and MFN2-ACTA transfections (Rev Fig. 2B; Supp Fig. 2A, C, D). We also included line fluorescence plot analysis to show localization of MFN2 mutants along with ER or mitochondrial markers (Rev Fig. 2C). Because these studies are cell-based analysis, which were performed using transient transfection, and the efficiency of expression of some of these constructs was very low, it would be technically very difficult to detect their expression using subcellular fractionation. Nevertheless, the constructs MFN2-IYFFT-FLAG and MFN2-ACTA used in our MS were previously reported and their subcellular localization was confirmed (____Sugiura, A. et al. (2013) ‘MITOL Regulates Endoplasmic Reticulum-Mitochondria Contacts via Mitofusin2’, Molecular Cell, 51(1), pp. 20–34. doi:https://doi.org/10.1016/j.molcel.2013.04.023)____.

The authors conclude that "MFN2-mediated fusion requires LPA synthesis", but what is shown instead is that GPATi is epistatic to fusion caused by overexpression of MFN2. The authors should be careful about drawing strong conclusions from their overexpression studies. While MTCH2 overexpression causes hyperfusion in the presence of GPATi, this does not mean that LPA doesn't promote MTCH2-dependent fusion, merely that GPATi does not block hyperfusion caused by MTCH2 overexpression.

R: We thank the reviewer for pointing out these issues. We have revised the text accordingly, and instead of "MFN2-mediated fusion requires LPA synthesis" we wrote “GPATi is epistatic to fusion caused by overexpression of MFN2” (page 13, line 14). ____We also added “____reducing LPA levels by inhibiting GPAT activity is insufficient to impair mitochondrial fusion enforced by MTCH2-overxpression____” (page 13, line 9).

The collapse of the mitochondrial network in MTCH2 KO cells treated with cycloheximide + GPATi does not indicate the cells "require newly-synthesized LPA to respond to SIMH". Instead, it suggests that mitochondria in MTCH2 KO cells are sensitized to combined GPATi/cycloheximide treatment. Could this collapse be fusion-independent? If mitochondria are treated with nocodazole to relax the mitochondrial network (as in Smirnova et al, MBoC, 2001 or Yang et al, Nat Comm, 2022), does the mitochondrial network appear hyperfused?

R: Thank you and as the reviewer suggested we rephrased to “mitochondria in MTCH2 KO cells are sensitized to combined ____GPATi/CHX treatment____.” (page 16, line 4).____ In addition, we performed the nocodazol experiment suggested by the reviewer and included the results in the MS (Rev Supp Fig. 4G-I). The results suggest that MTCH2 KO/GPATi generate clumping and collapse of the mitochondrial network that can be relaxed by blocking microtubule polymerization. Importantly even though the network is still highly fragmented, mitochondrial morphology has changed from large rounded and fragmented to short and tubulated mitochondria, suggesting that MTCH2 KO/GPATi does not impair other mitochondrial architectural changes induced by CHX.

The fact that FSG67 kills starved cells in the absence of MTCH2 does not mean LPA is not required for starvation induced fusion as concluded by the authors (p.12, first paragraph).

__R: Thank you and we corrected this part of the MS _(page 14, line 19)_. __

__ __ Reviewer #3

Major comments:

1. As FSG67, an inhibitor of glycerol-phosphate acyl transferase (GPAT) for LPA synthesis, blocks GPAT1/2 in the OMM and GPAT3/4 in the ER, it remains possible that OMM-anchored MFN2 cooperates with GPAT1/2 for concentrating LPA at the mitochondrial fusion site. Thus, the authors should test if loss of GPAT3/4, but not GPAT1/2, suppresses mitochondrial elongation in MTCH2 KO cells overexpressing MFN2.

R: We appreciate the reviewer’s comment and value this observation. We now included in the revised MS two different sets of experiments: silencing ER resident GPATs in MTCH2 KO MEFs, and enforcing mitochondrial fusion either by MFN2 overexpression (Rev Fig. 3L-N) or by HBSS/CHX stress (Rev Fig. 4I-K). We found that ER GPATs do have a contribution to MTCH2-independent mitochondrial fusion enforced by MFN2 overexpression, and to HBSS-induced mitochondrial fusion. These new results suggest that in the absence of MTCH2, LPA synthetized at the ER, is utilized for mitochondrial fusion. Unfortunately, we were unsuccessful in silencing GPAT1 (we tried two different sets of siRNAs, each composed of four different oligos).

It seems conceivable that FSG67 treatment causes a decrease in the protein levels of MFN2 and/or MTCH2, thereby leading to mitochondrial fragmentation. The authors should clarify this point by western blotting.

R: We thank the reviewer for this insightful comment. We performed these experiments and the results appear in the MS (Rev Supp Fig. 3A). We analyzed by Western blot the effect of FSG67 (named GPATi in the revised MS) on the expression levels of MTCH2, MFN2, fusion/fission proteins, GPATs, and a few other proteins. Interestingly, GPATi actually resulted in a small increase in MFN2 expression levels but had no effect on MTCH2 expression levels. Moreover, thanks to the reviewer’s suggestion, we also revealed that in non-treated and GPATi-treated MFN2 KO cells, MTCH2 expression levels were increased and GPATs3/4 expression levels were largely decreased (Rev Fig. 3B, C). These results suggest that: 1) Expression of MFN2 is important for the stabilization of GPAT3 and 4 (perhaps they form a functional complex together); 2) the expression levels of endogenous MTCH2 are possibly elevated to compensate for the decreased biosynthesis of LPA and increased demand of LPA funneling for mitochondrial fusion. This hypothesis was added to the discussion.

---

Minor comments:

1. It would be interesting to investigate whether an ER-anchored MFN2 variant defective in GTP hydrolysis can restore mitochondrial elongation in MTCH2 KO cells.

R: We thank the reviewer for this suggestion. We generated a MFN2-IYYFT K109A mutant, but unfortunately it was expressed at very low levels, and in the expressed cells it elicited mitochondrial aggregation.

The authors should add single-color fluorescent images into Figs. 1A, 1C, 1F, S1A, 2H, S2A, S2C, 3C, 3F, S3F, and S3H.

__R: We thank the reviewer and added single-color fluorescent images into all the Figures requested.

__

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary:

In this manuscript, Goldman et al. report the role of mitochondrial carrier homolog 2 (MTCH2), a mammalian atypical transporter, in regulating mitochondrial shape. Although numerous studies have previously suggested that MTCH2 localizes to the outer membrane of mitochondria (OMM) and acts in a myriad of processes including apoptosis, energy production, mitochondrial dynamics, lipid metabolism, and calcium signaling, its primary function remains uncertain. Using extensive fluorescence imaging, the authors reveal a functional relationship between MTCH2 and MFN1/2, large GTPases regulating mitochondrial fusion. Overexpression of MTCH2 restored mitochondrial elongation in MFN2 KO cells, but not MFN1 KO or MFN1/2 DKO cells. Overexpression of MFN2, but not MFN1, recovered mitochondrial elongation in MTCH2 KO cells. These results suggest that MTCH2 and MFN1 cooperatively act in mitochondrial elongation, and that MFN2 promotes mitochondrial fusion independently of MTCH2 and MFN1. Strikingly, ER-anchored MFN2 can rescue mitochondrial shaping defects in MTCH2 KO cells. The authors also investigated the role of lysophosphatidic acid (LPA), a mitochondrial fusion-promoting lipid, in MTCH2- and MFN2-mediated processes, and found that inhibition of LPA synthesis led to suppression of mitochondrial elongation in MTCH2 KO cells overexpressing MFN2, but not MFN2 KO cells overexpressing MTCH2. Collectively, they propose that MFN1 and MFN2 mediates mitochondrial fusion via two distinct mechanisms: one in the OMM depending on MTCH2 and MFN1, and the other in the ER depending on MFN2 and LPA synthesis.

Major comments:

1. As FSG67, an inhibitor of glycerol-phosphate acyl transferase (GPAT) for LPA synthesis, blocks GPAT1/2 in the OMM and GPAT3/4 in the ER, it remains possible that OMM-anchored MFN2 cooperates with GPAT1/2 for concentrating LPA at the mitochondrial fusion site. Thus, the authors should test if loss of GPAT3/4, but not GPAT1/2, suppresses mitochondrial elongation in MTCH2 KO cells overexpressing MFN2.
2. It seems conceivable that FSG67 treatment causes a decrease in the protein levels of MFN2 and/or MTCH2, thereby leading to mitochondrial fragmentation. The authors should clarify this point by western blotting.

Minor comments:

1. It would be interesting to investigate whether an ER-anchored MFN2 variant defective in GTP hydrolysis can restore mitochondrial elongation in MTCH2 KO cells.
2. The authors should add single-color fluorescent images into Figs. 1A, 1C, 1F, S1A, 2H, S2A, S2C, 3C, 3F, S3F, and S3H.

Significance

General assessment:

The findings in this study are potentially interesting and could provide new insights into the molecular mechanisms of mitochondrial fusion in mammals. The fluorescence imaging data are of high quality with quantification and statistical evaluation, mostly supporting the conclusion. There are, however, some missing points regarding the relationships among MTCH2, MFN1, MFN2, and GPAT1/2/3/4 in more detail. For example, does ER-anchored MFN2 interact with GPAT3/4? Does MTCH2 interact with MFN1 to promote mitochondrial elongation? Is LPA required for MFN1-mediated mitochondrial fusion? Nevertheless, this study would significantly be strengthened if the authors clarify the major and minor comments.

Advance:

This study raises the possibilities that ER-anchored MFN2 may act in transport of LPA from the ER to mitochondria in cooperation with GPAT3/4, and that MTCH2 may promote MFN1-mediated mitochondrial fusion independently of LPA.

Audience:

Given the findings that ER-anchored MFN2 and LPS synthesis cooperatively acts in promoting mitochondrial fusion, it will attract a broad range of researchers who study mitochondria, ER, membrane fusion, lipids, and interorganellar communication.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

In this manuscript, Goldman et al examine the role of the mitochondrial protein MTCH2 in mitochondrial fusion. Using a combination of overexpression of MTCH2, MFN1, and MFN2 in respective knockout cells, pharmacological inhibition of GPATs, and stress-induced mitochondrial hyperfusion, the authors explore the relationship between MTCH2, the mitofusins, and LPA levels. These experiments are particularly interesting in light of recent work by Labbe et al which propose that MTCH2 coordinates with LPA to promote mitochondrial fusion. The authors find a key difference from the work of Labbe et al, presenting data suggesting that the role of MTCH2 in starvation-induced mitochondrial hyperfusion varies depending on cell line. They also make the interesting observation that overexpression of MTCH2 is able to bypass GPATi treatment and promote mitochondrial fusion. The authors instead utilize differentially targeted MFN2 to argue that ER-localized MFN2 promotes mitochondrial fusion cooperatively with LPA synthesis. Based on these and other results, they conclude that MTCH2 works with MFN1 to stimulate fusion in a pathway parallel to MFN2 and LPA.

While the assays are mostly well performed and the findings will be of interest to those in the mitochondrial dynamics field, many of the conclusions drawn by the authors are not supported by the experiments shown. This in turn causes the the text to suffer from a lack of clarity and made the logic of the manuscript hard to follow. Additional controls are also needed to draw robust conclusions from MFN2 targeting and MFN1 overexpression experiments. Finally, the rigor of quantification should be clarified and expanded to all assays in the manuscript.

Specific points

1. The authors should comment on recently published work in Science that MTCH1 and likely MTCH2 are outer membrane insertases. The data in the manuscript seem consistent with the model that an undetermined protein may be poorly inserted in a MTCH2 knockout leading to reduced fusion activity mediated by MFN1. Could this explain how MTCH2 overexpression selectively restores fusion to MFN2 KO cells?
2. The authors make the claim that "MFN2 and MTCH2 compensate for each other's absence" though this is not supported by their data. For example, MFN2 expression is not affected in an MTCH2 KO but cannot compensate to promote fusion. Rather, the authors find that MTCH2 and MFN2 are capable of promoting fusion when overexpressed in the absence of the other.
3. The authors rely heavily on claims that tagged MFN1 and MFN2 are fully functional, but is it possible that MFN1-GFP is only partially functional? Does untagged MFN1 overexpression cause mitochondrial fusion in a MTCH2 KO? The quantification of this is done only with aspect ratio, and not by categorization of mitochondrial morphology (Fig. S1). The authors should present both analyses in this and all other experiments in the manuscript, particularly since aspect ratio is only performed on 15 cells per condition and not on experimental replicates. The authors should clarify if sample identity was blinded prior to analysis.
4. The constructs that form the basis of conclusions of ER versus mitochondrial-targeted MFN2 require additional controls to support robust conclusions. The immunofluorescence data suggests that MFN2-ACTA to some degree targets outside of mitochondria (Fig. 2A), and also that MFN2-YIFFT seems to localize to some degree to mitochondria (Fig. 2A). The YIFFT construct may also localize more prevalently to mitochondria in the presence of ACTA (Fig. S2A). Mistargeting would make interpretation of these experiments not possible, as these constructs must exclusively localize to their intended organelle to make strong conclusions. Triple labeling with ER and mitochondrial markers would be helpful, as well as western blots to confirm consistent expression levels and protein stability of each construct. The ACTA MFN2 also appears to promote fusion in a MTCH2 KO. How is this reconciled with the conclusion that ER-targeting of MFN2 is required?
5. The authors conclude that "MFN2-mediated fusion requires LPA synthesis", but what is shown instead is that GPATi is epistatic to fusion caused by overexpression of MFN2. The authors should be careful about drawing strong conclusions from their overexpression studies. While MTCH2 overexpression causes hyperfusion in the presence of GPATi, this does not mean that LPA doesn't promote MTCH2-dependent fusion, merely that GPATi does not block hyperfusion caused by MTCH2 overexpression.
6. The collapse of the mitochondrial network in MTCH2 KO cells treated with cycloheximide + GPATi does not indicate the cells "require newly-synthesized LPA to respond to SIMH". Instead, it suggests that mitochondria in MTCH2 KO cells are sensitized to combined GPATi/cycloheximide treatment. Could this collapse be fusion-independent? If mitochondria are treated with nocodazole to relax the mitochondrial network (as in Smirnova et al, MBoC, 2001 or Yang et al, Nat Comm, 2022), does the mitochondrial network appear hyperfused?
7. The fact that FSG67 kills starved cells in the absence of MTCH2 does not mean LPA is not required for starvation induced fusion as concluded by the authors (p.12, first paragraph).

Significance

see above

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In this manuscript, authors investigate the role of MTCH2 in mitochondrial morphology, in several conditions. The authors showed compensatory effect of MFN2 and MTCH2 on stress induced mitochondria hyperfusion (SIMH) in HBSS or CHX treated condition. Since mitochondria hyperfusion upon CHX treatment is impaired in MTCH2 KO cells treated with a GPAT inhibitor, but not in MFN2 KO cells, authors suggest two modes of SIMH, one MTCH2 dependent, the second MFN2/LPA dependent.

This effect seems to be phenocopied in unstressed condition using overexpression system. Mitochondrial fragmentation in MFN2 KO cells can be recovered by MTCH2 overexpression, and vice versa. The fragmentation of mitochondria in MTCH2 KO MEF is reversed also by an ER-targeted MFN2, suggesting the importance of MFN2 ER localization. The authors also point out that MTCH2 KO have increased mitochondria-ER contacts.

Major comments

1. Each of the finding is interesting, but the results are not well discussed and logical links leading to the key conclusions are sometimes missing.

1-1) A previous report (Bahat et al., 2018) and this manuscript show that MFN2 OE can restore mitochondrial elongation in MTCH2 KO cells and MTCH2 OE can do the same in MFN2 KO cells. Based on these data, authors conclude that "MFN2 and MTCH2 compensate for each other's absence", but the compensation works only when they are overexpressed. On the other hand, upon HBSS or CHX treatment, MFN2 KO or MTCH2 KO cells have elongated/hyperfused mitochondria, but this is not observed in double deficient cells. In this case, MFN2 and MTCH2 show compensatory effects on mitochondria elongation. The authors believe the two conditions, unstressed&OE and stressed conditions, activating same molecular machineries, but it is not fully supported by the data. For example, in unstressed condition, MTCH2 OE can recover MFN2 KO but not of MFN1 KO, suggesting MTCH2-dependent mitochondria fusion requires MFN1, but it is not tested for stressed condition. Furthermore, even if the machineries are common among stressed and unstressed conditions, authors should discuss why the endogenous MFN2/MTCH2 expression is not enough to activate mitochondria fusion in MTCH2/MFN2 KO cells, respectively, and how the compensatory effects are activated upon HBSS or CHX treatment.

1-2) Authors found out that ER-targeted MFN2 can rescue the mitochondria fragmentation in MTCH2 KO MEF cells, but mitochondria-targeted MFN2 has a lower effect than Wt MFN2. (Fig 2A&B). This finding suggests that MTCH2 loss might impair MFN2 localization at the ER. The authors should investigate endogenous MFN2 localization in MTCH2 KO MEFs.

1-3) The analysis to test whether recovery of mitochondrial morphology by ER-targeted MFN2 in MTCH2 KO depends on LPA synthesis or not is missing (Fig 3G). Authors should examine whether mitochondrial elongation induced by ER-localized MFN2 in MTCH2 KO cells is impaired by the GPAT inhibitor.

1-4) In the discussion section, authors suggest that mitochondrial LPA would be a crucial factor for MFN2 dependent mitochondrial fusion. To test this hypothesis, authors should overexpress mitochondrial GPAT and evaluate its effect on mitochondrial morphology.

1-5) In the discussion section, authors indicated that ER-targeted MFN2 could recover mito-ER contacts leading to LPA flux from ER to mitochondria and mitochondria elongation in MTCH2 KO. However, MTCH2 KO itself already have more mito-ER contacts (Fig 2D-H), and an artificial linker fails to recover mitochondria fragmentation in MTCH2 KO cells (Fig S2C, D). Thus, increased number of contacts appears not sufficient to recover the phenotype. The authors should consider this point in the discussion. 2. Certain methods are not appropriate to support the stated conclusions.

2-1) Authors assess "mitochondria fusion" by evaluating mitochondrial morphology. The authors also describe mitochondrial clumping as a fusion-impaired phonotype (Fig 4A&B). Mitochondrial fusion should be evaluated using a PEG assay or a mtPA-GFP analysis.

2-2) In figure 2D-G, authors show that MTCH2 KO cells have more and longer mitochondria-ER contacts. The correct experiment is not to compare these cells to WT, but to KO reconstituted with MTCH2.

2-3) Since staining of MitoTracker depends on mitochondrial membrane potential, mitochondria with low potential would be invisible and excluded from the analysis. Authors should investigate mitochondrial morphology by immunostaining also in Fig 4D, S4A, D, and K. 3. Other points

3-1) There are some discrepancies with the previous Labbé et al article.: Labbé et al. suggest that MTCH2 activity on mitochondria (from HCT116 cells) fusion is dependent on LPA, based on in vitro fusion assay. On the other hand, this manuscript shows that inhibition of LPA synthesis could not block MTCH2-induced mitochondria elongation in WT MEF and HEK293T cells. MTCH2 KO HCT116 cells are resistant to HBSS- but sensitive to CHX-induced mitochondria elongation. In this manuscript, MTCH2 KO MEF cells are sensitive to both stimuli, and only when MTCH2- and MFN2-deficient MEF cells are resistant to both stimuli. These discrepancies would be caused by difference of assay system or cell lines, but it is not clearly addressed.

Minor comments

1. Since authors use FSG67, an inhibitor against GPAT1, 2 and 3, knocking down of each of GPATs will improve the significance of this work.
2. Recently, a paper about MTCH2 is published (Guna et al., Science), which shows its insertase activity on tail anchored proteins. Authors should include this point of view in the discussion.

Significance

Mitochondrial carrier homologue 2 (MTCH2/MIMP/SLC25A50) was found as a mitochondrial solute carrier family member but the substrates are unknown. MTCH2 has roles on apoptosis with Bid (Zaltsman et al., 2010, the authors' group), lipid homeostasis (Rottiers et al., 2017), and mitochondrial morphology (Bahat et al., 2018, the authors' group, and Labbé et al., 2021).

Stress induced mitochondria hyperfusion (SIMH) was reported in 2009 by Tondera et al.. Under stress condition, such as UV-C, cycloheximide (CHX), or actinomycin D treatment, hyperfused mitochondria were observed and the event was named as SIMH. They also showed that SIMH is dependent on L-Opa1, MFN1 and SLP-2, which is later found as Opa1 regulator, but not on MFN2, BAX/BAK.

In this manuscript, authors show compensatory effect of MFN2 and MTCH2 on SIMH in HBSS or CHX treated condition. This compensatory effect seems to be reproduced in unstressed condition: mitochondrial fragmentation in MFN2 KO cells can be recovered with MTCH2 overexpression, and vice versa. Authors indicate that LPA synthesis and its mitochondrial localization would be crucial for MFN2 dependent fusion. The compensatory effect of MFN2 and MTCH2 is potentially interesting for a large audience in multiple cell biology fields (mitochondrial biology, ER-mitochondria contact sites, lipid biology).

Expertise: Mitochondria; fusion/fission; ER-mitochondria contact sites

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

Learn more at Review Commons

---

Reply to the reviewers

Manuscript number: RC-2022-01784

Corresponding author(s): Felipe, Court

1. General Statements [optional]

We submit a revision plan for our manuscript “Senescent Schwann cells induced by aging and chronic denervation impair axonal regeneration after peripheral nerve injury” by Fuentes-Flores et al. from the groups of Felipe Court, Judith Campisi, Jose Gomez, and Ahmet Hoke.

One of the greatest challenges in the field of peripheral nerve regeneration is the decrease in the nerve regenerative capacity in aged patients or after delayed repair, a condition also known as chronic denervation. For the last two decades, several research groups have focused on understanding this phenomenon, but the main drivers of unsuccessful regeneration and poor functional recovery have been elusive, remaining an important clinical problem.

In the work described in this manuscript we found an unexpected property of Schwann cells in the denervated nerves. Aged and chronically denervated Schwann cells are not just passive participants in the impaired regeneration process, but they actively inhibit the regeneration of peripheral axons. Using a combination of morphological, behavioral and molecular techniques in a collaborative multi-lab approach we demonstrate for the first time that senescent Schwann cells accumulate in aged or chronically denervated peripheral nerves modifying the nerve environment, increasing proinflammatory and regeneration-inhibitory factors. Elimination of senescent Schwann cells using a systemic intervention with senolytic or genetically targeting p16-positive senescent cells, greatly improve axonal regeneration in both chronic denervation and aging conditions. Importantly, the enhanced axonal regeneration observed after senescent cell elimination is accompanied by improved functional recovery after chronic denervation. Chronic denervation and aging are the main clinical problems associated to peripheral nerve injuries. Our approach, using FDA approved drugs currently in clinicals trials for its application as senotherapeutics, effectively broadens the spectrum of its clinical use and effectiveness.

We foresee this work will be of interest to a wide audience, including experts in nerve regeneration, senescent cells, aging and those studying the effect of chronic insults in regenerative medicine.

We have now received the comments from two reviewers and we are prepared to experimentally approach the issues raised. We thank their criticism and suggestions, as well as their very enthusiastic comments. We are extremely pleased as both reviewers recognized the important implication of this work, from reviewer 1:

“The findings reported in this manuscript are very interesting and will move the field of nerve repair forward. This paper will be of interest for basic science audience in the fields of aging and neurobiology and has also potential interest to the broader clinical and translational fields. Indeed, this paper provides data that Schwann cells entering a senescent stage not only fail to support axon regeneration in aged animals, but actively inhibit axon regeneration…. Furthermore, the use of an FDA approved drug, currently in clinicals trials for its application as senotherapeutics, to increase axon regeneration in aged and chronic denervation conditions will provide new avenues for clinical applications”.

Which is backed up by reviewer 2:

“Overall, this is an interesting study that undertake fundamental question in the field of nerve physiopathology and also could open a good opportunity in developing therapeutic strategies for translational research”.

We understand the reviewers have raised issues associated with the manuscript format and we are prepared to profoundly edit the manuscript as suggested. In addition, after discussion the experimental issues raised by the reviewers, we are prepared to perform all the experiments and controls suggested (some of them are currently underway), including new animal experiments and in vitro work. This information is detailed in the point-by point revision plan below.

Thank you in advance for the consideration and we look forward to hearing from you in due course. Please do not hesitate to contact me if you want to discuss anything associated to the manuscript and the revision plan.

2. Description of the planned revisions

Point-by point Reponses and revision plan in blue

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

Summary This manuscript by Fuentes-Flores et al reports that elimination of senescent Schwann cells by systemic senolytic drug treatment or genetic targeting improves nerve regeneration and functional recovery in aging and chronic denervation. This improved regeneration is associated with an upregulation of c-Jun expression. Mechanistically the authors provide data to show that senescent Schwann cells secrete factors that are inhibitory to axon regeneration. These findings are very interesting and move the field forward beyond the notion that Schwann cells fail to support axon regeneration in aged animals and identify potential targets to enhance nerve repair. The use of a senolytic drug to increase regeneration in aged and chronic denervation conditions provides new avenues for clinical interventions. However, some of the claims are overstated since this is not the first characterization of senescent Schwann cells and the manipulations used in the study are not entirely specific for Schwann cells. The manuscript is also poorly written and difficult to follow, given the complex set of surgeries and terminology, and lack of explanation of the rationale for the surgery model used. Figures are poorly labeled and difficult to follow without figure legends, and figure legends do not match the figures.

We thank the reviewer for the positive comments, we also acknowledge the problems detected in the manuscript format, including the lack of a clear explanation of the complex procedures used. We are prepared to work carefully on the format, including clear explanations of the procedures and new schemes to complement the text. As detailed below we are also prepared to perform all the experimental work proposed by the reviewer, which will strengthen the conclusion of this manuscript.

Below are suggestions for improvements:

Major comments:

• In the axon regeneration assays, how is the reconnection site defined in longitudinal images stained for SCG10? Of particular concern is that Figure 1B "adult chronic dmg" nerve section image appears to be identical to the image in Figure 5A "Vehicle adult (47dpi)". However, the reconnection site is located at different sites along the nerve. Also, the scale bar appears identical but the legend states different sizes. In Figure 1 chronic damage is 42dpi, and figure 5 is 47dpi, yet with what appears as the same image.

Revisions incorporated in the transferred manuscript, see section 3, below.

• The authors need to provide a rationale for the choice of this complex injury model and what are the advantages over other models. Please clearly describe the timepoints for each experiment and why the time points were chosen for analysis. Provide the scheme of injury in the main Figures to ease comprehension. A scheme is provided in what appears to be Figure S2, but the legend of Figure S2 does not match. Please compare same time points between aged and adults. Days post injury is sometimes referred to 7 or 42, and it is difficult to follow if it is days post initial transection of the tibial nerve or days post reconnection of transected tibial to peroneal. Revise all Figure legends and supplementary Figure legends to match figures.

We thank the reviewer for this comment. In a revised manuscript we will provide a clear explanation for the injury model used, as well as references, including one from the group of Tessa Gordon that describe for the first time this model in rats (PMID: 30215557), and the one applying this model to the mouse from the groups of Rhona Mirsky and Kristjan Jessen (PMID: 33475496). Briefly, this model has two advantages: first, it allows to generate chronic denervation for months and then be able to connect the distal denervated stump with a proximal one without the need of a nerve bridge; And second, neurons in the different groups that have different denervation times (1 week versus 6 weeks) are all damaged at the same time, eliminating variability associated to chronic axonal damage. We will include this information in the results section of a revised manuscript along with the above references.

We will include schemes of the injuries performed in each experiment in each figure, also adding a timeline. This is an excellent suggestion to clearly understand the different procedures performed. We will also check all figures and legends, including correcting the problem detected by the reviewer (legends of Figures 2 and 5 were swapped). We understand the problem of referring to days post-injury, then we will introduce a new form of referring to the initial transection and the experiment, which include reconnection. Adding schemes per figure will also help to understand the different timelines used in different experiments, including the ones using the senolytics.

As detailed above, we will perform a very careful revision of the text and legends for consistency.

• The authors' main conclusion is that Senescent Schwann cells inhibit axon regeneration. The authors need to tune down this statement and acknowledge that their manipulations are not entirely Schwann cells specific. While the data nicely shows a contribution of senescent Schwann cells, it does not sufficiently acknowledge the possibility that other senescent cells in the nerve contribute to this effect. First, the authors refer in the discussion that 60% of the senescent cells are SOX10 negative, and thus represent other cells beyond Schwann cells. This quantification needs to be shown in Figure 1. Second, the genetic and pharmacology manipulation eliminate all senescent cells, including Schwann cells. Third, while the culture experiment may be Schwann cells specific, the authors need to provide detailed information on how they purify these cells, how they induce repair Schwann cells (rSC) as claimed in Figure 3, and demonstrate whether these are pure Schwann cells. This is important because other cells in the nerve contribute to nerve repair, including mesenchymal cells. Finally, the claim that using Mpz-cre will lead to c-jun overexpression only in SC also needs to be demonstrated, since Mpz is also expressed in satellite glial cells in the DRG.

We thank the reviewer for these comments and suggestions. We will tone down the statement that senescence Schwann cells are the only cell candidates for modulating regeneration. We discussed this in the original manuscript, but we agree we need to review this statement, including new data detailed below.

We will include the data requested by the reviewer (60% SOX-10 negative senescent cells) in a new graph in Figure 1. Also, we are currently performing new experiments and quantifications using specific markers for macrophages, epithelial cells, and fibroblasts, to identify the cell identity of the 40% SOX-10 negative senescent cells in aging and chronic denervation.

Regarding in vitro experiments, we will provide detailed methods for Schwann cell purification, and induction of rSC phenotype. Related to the purity of these cultures, in past experiments we have obtained numbers ranging from 95-98% of Schwann cell purity; we will repeat these experiments and quantification for this manuscript and include this data in the method section of a revised text.

Regarding c-jun overexpression in the Mpz-cre, we agree with the reviewer that there is probably overexpression in satellite glial cells in the DRG. Satellite glial cells (SGCs) are a subset of cells in the Schwann cell (SC) lineage that express several early myelination markers, such as Mbp, Mag, and Plp, and the transcription factor Sox10. SGCs express early SC markers, such as CDH19, and are transcriptionally and morphologically similar to SCs, even in the absence of axonal contact. Regarding the possibility that SGCs are contributing to the enhanced regeneration presented by mice with c-jun overexpression, this issue was somehow approached previously by Wagstaff et al. (PMID: 33475496), as they showed that in this mouse strain, increased axonal regeneration was equally observed in sensory neurons, in contact with SGCs, and motor neurons, which are not associated to SGCs. This observation suggests that the effect is associated with c-jun overexpressing Schwann cells in the distal stump. In addition, in our work, the changes in senescent cells observed in the c-jun overexpressed mouse, were associated to the distal nerve stump, which was mechanically separated from the proximal region. We agree it is important to include this discussion, and we will do so in the revised manuscript.

• In Figure 5, c-jun is shown after denervation (42 dpi). The results describe 28 days of denervation, 5 days of GCV and 7 days post reconnection, which makes 40 days. If that is not the case, results need to better explain timeline of this procedure. Also, what is the basal c-jun expression in p16-3MR mice? In addition to the number of c-jun positive cells shown in Figure 5G, the authors need to quantify the percent of c-jun puncta that co-localize with Sox10. The size of the c-jun puncta appears different in size in vehicle and GCV, is that an expected phenotype?

As expressed above, we will include schemes and timelines for all the surgical experiments in a revised manuscript, including detailed information for the experiments using the p16-3MR mice.

Regarding c-jun expression in the p16-3MR mice, we are currently performing the suggested control experiment which is important to draw conclusions of this research. We will use immunofluorescence, but also include western blots as an extra analytical method in this and other experiments. All this information will be included in a revised manuscript.

The observation of the apparent difference c-jun puncta is intriguing. Is not an expected phenotype, but it will important to check if there is a quantifiable change in the pattern of expression. We will quantify this in the different groups and include the results in a revised Figure 5.

Minor comments

• Improve labeling of Figures or at the very least describe in the Figure legend. For example: Figure 5B-C, which of the graphs is from adult mice and which is from aged mice?

We are sorry about the lack of clear labeling in the figures; we will carefully review all figures in the manuscript and their corresponding legends, adding better labeling. Labelling of Figure 5B-C has been corrected.

• The authors need to carefully describe where the high magnification images were taken in the injured nerve and keep the comparison at same site between groups. Please check the scale bar for each image. For example, the images in Figure 1D/F/I/K used same scale, but the cell size and cell morphology are different. The images for split individual channels need to match the merge channel images. For example, the individual channel and merge images are not properly aligned in Figure 4C, ABY-263 group.

As suggested by the reviewer, we will show the regions in which high magnification were taken. All quantifications were performed in comparable sites, but we will include information in a revised manuscript to clearly describe the methodology used. We will check all scale bars in a revised manuscript.

For the comment on alignment problems we have incorporated this in the transferred manuscript, see section 3 below.

• The analysis method used to quantify axon regeneration should be consistent throughout. For example, in Figure 1C, number of axons/nerve width(um) was used for regeneration assay, but axon density (width corrected) was used in Figure 5B-C in regeneration assay.

Revisions incorporated in the transferred manuscript, see section 3, below.

**Referees cross-commenting**

I agree with all Referee #2's comments. Both sets of comments are important, complementary and point to the same major concerns that need to be addressed. Agree as well that both reviewer think this is an interesting and relevant study for the field of nerve repair, if revised appropriately.

Reviewer #1 (Significance (Required)):

__Significance____ __The findings reported in this manuscript are very interesting and will move the field of nerve repair forward. This paper will be of interest for basic science audience in the fields of aging and neurobiology and has also potential interest to the broader clinical and translational fields. Indeed, this paper provides data that Schwann cells entering a senescent stage not only fail to support axon regeneration in aged animals, but actively inhibit axon regeneration. While this reviewer raises questions on whether only senescent Schwann cells or other senescent cells in the nerve contribute to this effect, the identification of potential targets to enhance nerve repair is highly significant. Furthermore, the use of an FDA approved drug, currently in clinicals trials for its application as senotherapeutics, to increase axon regeneration in aged and chronic dennervation conditions will provide new avenues for clinical applications.

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

Summary The reported study by Fuentes-Flores et al. shows that Schwann cells (SCs) in peripheral nerves undergo senescence with aging or in chronic denervation. This senescent SC phenotype correlates with downregulation of c-Jun expression and axon regeneration capacity and consequently, affecting functional recovery. The study has been undertaken by using in vivo mice model of chronically denervated sciatic nerve and in vitro of rat-primary cell cultures (Schwann cell, DRG-explants, and their coculture). Schwann cell exosome manipulation was also included to exploit their released factor as media for cell cultures.

Major comments:

1. __ __ Chronic denervated nerve model and Schwann cell phenotype

*Tibial nerve transection and chronic denervated nerve: As the referee have no information about this model (no reference is cited), a detailed description should be provided highlighting the interest of such model compared to standard sciatic nerve lesion model.

We thank the reviewer for this comment. We will provide a clear explanation for the injury model used, as well as references, including one from the group of Tessa Gordon that describe for the first time this model in rats (PMID: 30215557), and the one applying this model to the mouse from the groups of Rhona Mirsky and Kristjan Jessen (PMID: 33475496). Briefly, this model has two advantages: first, it allows to generate a chronic denervation for months and then be able to connect the distal denervated stump with a proximal one without the need of a nerve bridge; second, neurons in the different groups that have different denervation times (1 week versus 6 weeks) are all damaged at the same time, eliminating variability associated to chronic axonal damage. We will include this information in the result section of a revised manuscript along with the above references.

*Should be shown, histological analysis of denervated tibial branch prior reconnection with freshly cut proximal peroneal branch with specific immunostainings of rSCs v.s. sSC associated with dapi nuclear staining (as used along this study). Specific staining for other cells should be also provided (i.e., macrophage and endothelial cells). In simple words, "how chronically denervated nerve looks like and what is his cellular content? This is necessary to responds to the following main referee question: does the increase/decrease of rSCs or sSC under specific condition all through the study concerns the SCs that have migrated from freshly cut peroneal branch into denervated tibial distal branch, or resident SCs that have survived in chronically denervated tibial distal branch. In other words, whether rSCs that migrate (and accompanying regenerating axons) into chronically denervated branch nerve undergo phenotype change into sSC because of the environment of chronically denervated nerve. This is not clearly described or discussed, and remain confusing for reader.

We are sorry about the lack of clarity in the text and figures. The immunostaining analysis of denervated tibial branch prior reconnection is included in the original manuscript, specifically in Figure 1D-K, and Figure 4. In a revised manuscript we will include schemes in the Figures to shown the regions analyzed in each case.

We thank for the suggestion of including staining for other cell types. As suggested by the reviewer we are currently performing these experiments and analysis for macrophages, endothelial cells and fibroblasts, together with staining for cell senescence (p16), in both aged and chronically denervated conditions. We will include this data in a revised manuscript

About the specific question of the reviewer: “does the increase/decrease of rSCs or sSC under specific condition all through the study concerns the SCs that have migrated from freshly cut peroneal branch into denervated tibial distal branch, or resident SCs that have survived in chronically denervated tibial distal branch”. Our data demonstrate that senescence Schwann cells appear in the distal nerve stump in aged mice and after chronic denervation. The distal stump is physically disconnected from the proximal part of the nerve. Therefore, after reconnection the regenerating axons encounters a tissue which is already populated with senescent cells. To clearly explain this, we will add extra text in the results and discussion section to clarify these findings.

Support of up- and down-regulation of gene expression illustrated in fig.4

The conclusions and statements on up- or down-regulation of c-jun, yH2AX, beta-gal, P16, arise from quantitative and qualitative analysis from immunostaining of these specific markers by determining the number of positive cells rSCs vs. sSC. For these strong statements appropriate methods for quantification of protein levels such as western blots are required. For example, the statement of down regulation of c-jun expression, the quantitative graph shows strong increase in c-jun cell number under ABT263 treatment but the histological photo does not illustrate such decrease in number of the cell. It shows rather an increase in brightness of c-jun staining. Thus, only appropriate method for protein level quantification could be conclusive. This would also remove the doubt that some photos are under- or over- exposed as it appears in the figure. For example, in aged animal under vehicle condition, there is no variability in staining intensity. Accordingly, one question to the authors: for quantification of cell number, are weakly stained cells considered as positive cell?

We agree with the reviewer that a more quantitative method is required to complement the immunofluorescence data. As the reviewer correctly states, quantification of the immunofluorescence data corresponds to cell positive for the specific marker, expressed as % of cell positive for that maker. Therefore, we will perform western blot for c-jun, yH2AX, and p16 for the different models, including treatments with senolytics.

Regarding the method for quantification, we performed all these quantifications using Imaris software, in which we set up the same threshold for all conditions for a specific antibody marker. Then, in addition to the quantitative western blot analysis, we will include a graph representing the distribution of the labelling (intensity histogram) for all cell number quantification from immunofluorescence data, comparing control with the experimental condition. Finally, the methods used for quantification will be expanded in a revised manuscript.

The method section should be revised in general

Methods could be described in brief only when are supported by provided refs in which the reader could find details. Several refs are missing, i.e., 4.4 for thermal allodynia, 4.5 for ABT263 gavage administration; senescence induction, ...

Quantification methods should be more detailed, several information are missing and not found in result section or legends (i.e., number of nerve section per animal, neurite length, ...).

We completely agree with the reviewer that the method section was not developed adequately in the original manuscript. As described in previous responses, we will detail the methods used, especially those associated with quantifications performed. We will also include references for the different methods used, including those detailed by the reviewer.

The use of rat for in vitro DRG and SC culture while in vivo study is undertaken on mice The switch of species from in vivo to in vitro (mice vs. rat), is not justified as mice DRG and SC culture are also commonly used. In addition, the use of transgenic mice (used here only for in vivo) could also be exploited to address specific and reinforce the data.

We agree with the reviewer that using mouse SC in in vitro experiments will be a better approximation to support our findings. We have been using rat SC in this and other publications as they were the standard model used in in vitro experiments. Nevertheless, as the author states, now there are suitable methods for culturing mouse SC, that we have incorporated in our lab. Therefore, we will perform key in vitro experiments using mouse SC together with mouse DRGs.

Regarding the use of the transgenic mice (3MR), we thank the reviewer for this suggestion; we will perform new experiments using SC derived from 3MR mice in order to demonstrate induction of senescence (by expression of the red fluorescent protein in this transgenic line) and senolysis in vitro.

Use of conditioned exosome/media

Should be explained why the use of exosomes directly in cell culture was not tested. This would be close to physiological condition, regarding the concentration of released factors.

This is an important point that was not explored. As we have plenty of experience using SC-derived exosomes, we will perform the suggested experiments comparing the effect of exosomes from conditioned media from senescent-induced SC and include the results from these experiments in a revised manuscript.

The statement on the effect of rSC vs. sSC cell on growth cone dynamic

The provided data illustrated in fig. 3 are not in support that sSC affect growth cone dynamics. Only what would be "suggested" is that the decrease in neurite length could be associated to changes of growth cone morphology, on fixed tissue, that appeared to be affected. If such statement has to be maintained, time-laps is required. The image does not reflect a retracting neurite nor collapsed growth cone. In addition, other mechanisms could be at the basis of observed decrease in neurite length, which are not evaluated here. This is an important point to address as the authors state that sSC release inhibitory factors.

We completely agree with the reviewer: we are not exploring growth cone dynamics. We will change the manner these results are presented as we are not demonstrating a dynamic process in our results. We prefer to modify the text associated with these experiments rather than perform a time-lapse analysis at this moment. This is part of a future exploration we want to achieve, that will take some time to develop, and we consider at this moment lies outside the scope of the present work. Included in section 4, below.

other comments

• The surgical description is complicated, also annotation to be added in supp fig #2A; provide

We will work in the description of the model, including references to other papers using this nerve anastomosis model for assessing regenerative potential. As stated above we will also include schemes in all figures to help the reader with the surgical procedure and different timelines used.

• M&M 4.2: lign #8, error referring to Fig 2A, correct by. Supp fig 2A

Revisions incorporated in the transferred manuscript, see section 3, below.

• Review refs list, ref#17, full info needed

Revisions incorporated in the transferred manuscript, see section 3, below.

• Fig 3H would be interesting only if contain a column of the 21 proteins exclusively expressed in senescent-induced SCs

Revisions incorporated in the transferred manuscript, see section 3, below.

• The immunostaining of Lamin B1 positive nuclear invaginations in fig S4 should better described in results for non-familiar reader

We will describe better this staining pattern in the result section of the revised manuscript.

• Title of Fig 3, the expression "neuronal growth" is not appropriate here (neurite outgrowth)

Revisions incorporated in the transferred manuscript, see section 3 below.

**Referees cross-commenting**

I agree with referee#1's comments. He/she has raised complementary and important points that should be taken into account by the authors as well as we share the same major concerns. Furthermore, we both expressed the interest of such study if revised appropriately.

Reviewer #2 (Significance (Required)):

Significance

Overall, this is an interesting study that undertake fundamental question in the field of nerve physiopathology and also could open a good opportunity in developing therapeutic strategies for translational research. However, additional investigations are needed to support the main conclusions.

3. Description of the revisions that have already been incorporated in the transferred manuscript

Reviewer 1

Major comments

• In the axon regeneration assays, how is the reconnection site defined in longitudinal images stained for SCG10? Of particular concern is that Figure 1B "adult chronic dmg" nerve section image appears to be identical to the image in Figure 5A "Vehicle adult (47dpi)". However, the reconnection site is located at different sites along the nerve. Also, the scale bar appears identical but the legend states different sizes. In Figure 1 chronic damage is 42dpi, and figure 5 is 47dpi, yet with what appears as the same image.

We thank the reviewer for detecting this issue. The problem arises as the data shown in Figure 1a corresponds to the controls (vehicle) of Figure 5a. The data in Figure 1 is a known phenomenon in the field of peripheral regeneration (i.e., decreased regeneration in aged animals as well as in chronically denervated nerves); nevertheless, we decided to add this data at the start of the manuscript to clearly shown the reader (thinking in a broader scientific audience) the baseline of the evident decrease in axonal regeneration in these two conditions. To make this very clear, we have included the same image Figure 1 and 5 for the controls and detailed this in the legends of both Figure 1 and 5 in the uploaded manuscript.

We have checked the scales and modify the scale in Figure 5 as it was not correct. We have also corrected the nomenclature for days post injury in this image as well as in the corresponding legend.

In addition, for transparency, we have uploaded in a public repository (EBI BioStudies database, https://www.ebi.ac.uk/biostudies/) all the microscopy images used in this work, which is detailed in the uploaded manuscript in a new section named data availability.

Regarding the localization of the reconnection site, this is identified using the whole z-stack of the nerve and not a single section, the region in the z-stack can be recognized using two parameters: the difference in diameter between the proximal and distal stump and by identifying the filament used to suture both stumps. We have included a description of this procedure in the method section of the revised manuscript. As this is not always a perpendicular line, in the revised Figures we have now used an arrowhead to denote the reconnection site.

We are sorry for the confusion generated by the labeling of some images. We will review the text and figures and fix errors. In a revised manuscript we will also add schemes for several figures in order to explain better the experimental procedure and timelines.

Minor comments

• Improve labeling of Figures or at the very least describe in the Figure legend. For example: Figure 5B-C, which of the graphs is from adult mice and which is from aged mice?

We have included the suggested labelling for Figure 5B-C.

• The images for split individual channels need to match the merge channel images. For example, the individual channel and merge images are not properly aligned in Figure 4C, ABY-263 group.

We thank the reviewer for spotting the error in the split channels, we have now fixed this in Fig 4C, but also corrected other alignment problems detected in Fig 4A and 5F.

• The analysis method used to quantify axon regeneration should be consistent throughout. For example, in Figure 1C, number of axons/nerve width(um) was used for regeneration assay, but axon density (width corrected) was used in Figure 5B-C in regeneration assay.

We have included the procedure used to quantify axonal regeneration in the method section of the uploaded manuscript, which is the same throughout the manuscript. We are sorry for the different texts in the axes of graphs included in Figure 1 and Figure 5. In the first version of the figures, we were using the term “axon density (width corrected)”, but then we decided to change it to “number of axons/nerve width (mm)”, which was more precise. Unfortunately, the text of the graph axis in Figure 5 was not changed by mistake. We have now fix this in the revised version.

Reviewer 2

• M&M 4.2: lign #8, error referring to Fig 2A, correct by. Supp fig 2A

We have corrected this error in the uploaded manuscript.

• Review refs list, ref#17, full info needed

We have fixed this reference in the uploaded manuscript.

• Fig 3H would be interesting only if contain a column of the 21 proteins exclusively expressed in senescent-induced SCs

We are sorry about this omission in Figure 3H. We included a list of all the identified proteins of repair and senescent-induced SCs in Supplementary Table 4 (Table S4) of the original manuscript, including the identity of the 21 proteins exclusively expressed in senescent-induced SCs. In the revised version, we have incorporated the information of the 21 proteins in Figure 3H as suggested by the reviewer.

• Title of Fig 3, the expression "neuronal growth" is not appropriate here (neurite outgrowth)

We thank the reviewer for detecting this error, we have changed the expression as suggested in the uploaded manuscript.

Other changes included in the revised manuscript and Figures

1. Numeric data for graphs We have included a new supplementary excel file: Supplementary Table 8, including the data for each replicate associated to the graphs of text and supplementary figures.

Revision Figure 5G.

During our review on all the individual replicates of the manuscript data to upload into BioStudies, the first author noticed that the data in graph of Figure 5G corresponded to a pilot experiment performed to set up the protocol. The final experiment was not included, then we uploaded the correct images and included the final quantification. Data is comparable, and statistical differences remains.

Revision Figure 3A.

We have modified the pseudocolors of the S100 antibody channel from magenta to green for ease of visualization. The image and quantification remain exactly the same.

4. Description of analyses that authors prefer not to carry out

Reviewer 2

The statement on the effect of rSC vs. sSC cell on growth cone dynamic

The provided data illustrated in fig. 3 are not in support that sSC affect growth cone dynamics. Only what would be "suggested" is that the decrease in neurite length could be associated to changes of growth cone morphology, on fixed tissue, that appeared to be affected. If such statement has to be maintained, time-laps is required. The image does not reflect a retracting neurite nor collapsed growth cone. In addition, other mechanisms could be at the basis of observed decrease in neurite length, which are not evaluated here. This is an important point to address as the authors state that sSC release inhibitory factors.

We completely agree with the reviewer: we are not exploring growth cone dynamics. We will change the manner these results are presented as we are not demonstrating a dynamic process in our results. We prefer to modify the text associated with these experiments rather than perform a time-lapse analysis at this moment. This is part of a future exploration we want to achieve, that will take some time to develop, and we consider at this moment lies outside the scope of the present work.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

The reported study by Fuentes-Flores et al. shows that Schwann cells (SCs) in peripheral nerves undergo senescence with aging or in chronic denervation. This senescent SC phenotype correlates with downregulation of c-Jun expression and axon regeneration capacity and consequently, affecting functional recovery. The study has been undertaken by using in vivo mice model of chronically denervated sciatic nerve and in vitro of rat-primary cell cultures (Schwann cell, DRG-explants, and their coculture). Schwann cell exosome manipulation was also included to exploit their released factor as media for cell cultures.

Major comments:

1. Chronic denervated nerve model and Schwann cell phenotype

2. Tibial nerve transection and chronic denervated nerve: As the referee have no information about this model (no reference is cited), a detailed description should be provided highlighting the interest of such model compared to standard sciatic nerve lesion model.

3. Should be shown, histological analysis of denervated tibial branch prior reconnection with freshly cut proximal peroneal branch with specific immunostainings of rSCs v.s. sSC associated with dapi nuclear staining (as used along this study). Specific staining for other cells should be also provided (i.e., macrophage and endothelial cells). In simple words, "how chronically denervated nerve looks like and what is his cellular content? This is necessary to responds to the following main referee question: does the increase/decrease of rSCs or sSC under specific condition all through the study concerns the SCs that have migrated from freshly cut peroneal branch into denervated tibial distal branch, or resident SCs that have survived in chronically denervated tibial distal branch. In other words, whether rSCs that migrate (and accompanying regenerating axons) into chronically denervated branch nerve undergo phenotype change into sSC because of the environment of chronically denervated nerve. This is not clearly described or discussed, and remain confusing for reader.
4. Support of up- and down-regulation of gene expression illustrated in fig.4 The conclusions and statements on up- or down-regulation of c-jun, yH2AX, beta-gal, P16, arise from quantitative and qualitative analysis from immunostaining of these specific markers by determining the number of positive cells rSCs vs. sSC. For these strong statements appropriate methods for quantification of protein levels such as western blots are requiered. For example, the statement of down regulation of c-jun expression, the quantitative graph shows strong increase in c-jun cell number under ABT263 treatment bu the histological photo does not illustrate such decrease in number of the cell. It shows rather an increase in brightness of c-jun staining. Thus, only appropriate method for protein level quantification could be conclusive. This would also remove the doubt that some photos are under- or over- exposed as it appears in the figure. For example, in aged animal under vehicle condition, there is no variability in staining intensity. Accordingly, one question to the authors: for quantification of cell number, are weakly stained cells considered as positive cell?
5. The method section should be revised in general Methods could be described in brief only when are supported by provided refs in which the reader could find details. Several refs are missing, i.e., 4.4 for thermal allodynia, 4.5 for ABT263 gavage administration; senescence induction, ... Quantification methods should be more detailed, several information are missing and not found in result section or legends (i.e., number of nerve section per animal, neurite length, ...).
6. The use of rat for in vitro DRG and SC culture while in vivo study is undertaken on mice The switche of species from in vivo to in vitro (mice vs. rat), is not justified as mice DRG and SC culture are also commonly used. In addition, the use of transgenic mice (used here only for in vivo) could also be exploited to address specific and reinforce the data.
7. Use of conditioned exosome/media Should be explained why the use of exosomes directly in cell culture was not tested. This would be close to physiological condition, regarding the concentration of released factors.
8. The statement on the effect of rSC vs. sSC cell on growth cone dynamic The provided data illustrated in fig. 3 are not in support that sSC affect growth cone dynamics. Only what would be "suggested" is that the decrease in neurite length could be associated to changes of growth cone morphology, on fixed tissue, that appeared to be affected. If such statement has to be maintained, time-laps is required. The image does not reflect a retracting neurite nor collapsed growth cone. In addition, other mechanisms could be at the basis of observed decrease in neurite length, which are not evaluated here. This is an important point to address as the authors state that sSC relase inhibitory factors.

Other comments

• The surgical description is complicated, also annotation to be added in supp fig #2A; provide
• M&M 4.2: lign #8, error referring to Fig 2A, correct by. Supp fig 2A
• Review refs list, ref#17, full info needed
• Fig 3H would be interesting only if contain a column of the 21 proteins exclusively expressed in senescent-induced SCs
• The immunostaining of Lamin B1 positive nuclear invaginations in fig S4 should better described in results for non-familiar reader
• Title of Fig 3, the expression "neuronal growth" is not appropriate here (neurite outgrowth)

Referees cross-commenting

I agree with referee#1's comments. He/she has raised complementary and important points that should be taken into account by the authors as well as we share the same major concerns. Furthermore, we both expressed the interest of such study if revised appropriately.

Significance

Overall, this is an interesting study that undertake fundamental question in the field of nerve physiopathology and also could open a good opportunity in developing therapeutic strategies for translational research. However, additional investigations are needed to support the main conclusions.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary

This manuscript by Fuentes-Flores et al reports that elimination of senescent Schwann cells by systemic senolytic drug treatment or genetic targeting improves nerve regeneration and functional recovery in aging and chronic denervation. This improved regeneration is associated with an upregulation of c-Jun expression. Mechanistically the authors provide data to show that senescent Schwann cells secrete factors that are inhibitory to axon regeneration. These findings are very interesting and move the field forward beyond the notion that Schwann cells fail to support axon regeneration in aged animals and identify potential targets to enhance nerve repair. The use of a senolytic drug to increase regeneration in aged and chronic denervation conditions provides new avenues for clinical interventions. However, some of the claims are overstated since this is not the first characterization of senescent Schwann cells and the manipulations used in the study are not entirely specific for Schwann cells. The manuscript is also poorly written and difficult to follow, given the complex set of surgeries and terminology, and lack of explanation of the rationale for the surgery model used. Figures are poorly labeled and difficult to follow without figure legends, and figure legends do not match the figures. Below are suggestions for improvements:

Major comments:

• In the axon regeneration assays, how is the reconnection site defined in longitudinal images stained for SCG10? Of particular concern is that Figure 1B "adult chronic dmg" nerve section image appears to be identical to the image in Figure 5A "Vehicle adult (47dpi)". However, the reconnection site is located at different sites along the nerve. Also, the scale bar appears identical but the legend states different sizes. In Figure 1 chronic damage is 42dpi, and figure 5 is 47dpi, yet with what appears as the same image.
• The authors need to provide a rationale for the choice of this complex injury model and what are the advantages over other models. Please clearly describe the timepoints for each experiment and why the time points were chosen for analysis. Provide the scheme of injury in the main figures to ease comprehension. A scheme is provided in what appears to be Figure S2, but the legend of Figure S2 does not match. Please compare same time points between aged and adults. Days post injury is sometimes referred to 7 or 42, and it is difficult to follow if it is days post initial transection of the tibial nerve or days post reconnection of transected tibial to peroneal. Revise all figure legends and supplementary figure legends to match figures.
• The authors' main conclusion is that Senescent Schwann cells inhibit axon regeneration. The authors need to tune down this statement and acknowledge that their manipulations are not entirely Schwann cells specific. While the data nicely shows a contribution of senescent Schwann cells, it does not sufficiently acknowledge the possibility that other senescent cells in the nerve contribute to this effect. First, the authors refer in the discussion that 60% of the senescent cells are SOX10 negative, and thus represent other cells beyond Schwann cells. This quantification needs to be shown in Figure 1. Second, the genetic and pharmacology manipulation eliminate all senescent cells, including Schwann cells. Third, while the culture experiment may be Schwann cells specific, the authors need to provide detailed information on how they purify these cells, how they induce repair Schwann cells (rSC) as claimed in Figure 3, and demonstrate whether these are pure Schwann cells. This is important because other cells in the nerve contribute to nerve repair, including mesenchymal cells. Finally, the claim that using Mpz-cre will lead to c-jun overexpression only in SC also needs to be demonstrated, since Mpz is also expressed in satellite glial cells in the DRG.
• In Figure 5, c-jun is shown after denervation (42 dpi). The results describe 28 days of denervation, 5 days of GCV and 7 days post reconnection, which makes 40 days. If that is not the case, results need to better explain timeline of this procedure. Also, what is the basal c-jun expression in p16-3MR mice? In addition to the number of c-jun positive cells shown in Figure 5G, the authors need to quantify the percent of c-jun puncta that co-localize with Sox10. The size of the c-jun puncta appears different in size in vehicle and GCV, is that an expected phenotype?

Minor comments

• Improve labeling of Figures or at the very least describe in the Figure legend. For example: Figure 5B-C, which of the graphs is from adult mice and which is from aged mice?
• The authors need to carefully describe where the high magnification images were taken in the injured nerve and keep the comparison at same site between groups. Please check the scale bar for each image. For example, the images in Figure 1D/F/I/K used same scale, but the cell size and cell morphology are different. The images for split individual channels need to match the merge channel images. For example, the individual channel and merge images are not properly aligned in Figure 4C, ABY-263 group.
• The analysis method used to quantify axon regeneration should be consistent throughout. For example, in Figure 1C, number of axons/nerve width(um) was used for regeneration assay, but axon density (width corrected) was used in Figure 5B-C in regeneration assay.

Referees cross-commenting

I agree with all Referee #2's comments. Both sets of comments are important, complementary and point to the same major concerns that need to be addressed. Agree as well that both reviewer think this is an interesting and relevant study for the field of nerve repair, if revised appropriately.

Significance

The findings reported in this manuscript are very interesting and will move the field of nerve repair forward. This paper will be of interest for basic science audience in the fields of aging and neurobiology and has also potential interest to the broader clinical and translational fields. Indeed, this paper provides data that Schwann cells entering a senescent stage not only fail to support axon regeneration in aged animals, but actively inhibit axon regeneration. While this reviewer raises questions on whether only senescent Schwann cells or other senescent cells in the nerve contribute to this effect, the identification of potential targets to enhance nerve repair is highly significant. Furthermore, the use of an FDA approved drug, currently in clinicals trials for its application as senotherapeutics, to increase axon regeneration in aged and chronic dennervation conditions will provide new avenues for clinical applications.

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

Learn more at Review Commons

---

Reply to the reviewers

General Statements

Thank you for providing an initial assessment of our manuscript. We went through all the raised comments and suggestions aiming to improve our manuscript. Our manuscript will benefit from addressing them.

Our main impression is that the concerns regarding the novelty of our work by Reviewers #1 and #3 come from the fact that we apply a known flexible statistical framework (group factor analysis) to novel applications in single-cell data analysis, namely the estimation of multicellular programs and sample-level unsupervised analysis. The core methodology of our work is indeed based on the popular tool Multi-omics factor analysis (MOFA). We see the novelty of our study in the formulation of these relatively new applications within this framework, and the demonstration of the added value that this formulation provides building on MOFA’s strengths, in particular by expanding the possibilities of downstream analysis of single-cell data including the meta-analysis of distinct single-cell patient cohorts and its integration to complementary bulk and spatial data modalities.

The simultaneous estimation of multicellular programs together with sample-level unsupervised analysis is only possible with a single available tool, scITD, which is limited by its modeling strategy, based on tensor decomposition: with tensor decomposition, multicellular programs can not be estimated from distinct feature sets across cell-types, making this method less flexible and sensitive to technical effects, such as background expression. We compared our proposed methodology with scITD and showed the benefits of using group factor analysis as implemented in MOFA for this task. Moreover, as of now, no other methodology is able to estimate multicellular programs and perform sample-level unsupervised analysis, simultaneously in multiple independent single-cell atlases. We also showed how multicellular programs are traceable in bulk transcriptomics data and show that they are better fit to classify heart failure patients compared to classic cell-type deconvolution approaches.

Altogether, we believe that our current manuscript complements existing literature and puts forward an approach with distinct features to analyze single-cell atlases. We will edit the text to make more explicit the novelty and advantages of our proposed methodology, and we will emphasize that our work does not mean to propose a new method, but rather demonstrate how group factor analysis can be used for novel sample-level analysis of single-cell data. We plan to incorporate the suggestions by Reviewer #1 regarding the inclusion of additional datasets, model validations, and novel applications involving a direct modeling of cell-compositions and spatial organization of cells. Moreover, we plan to discuss perspectives on how cell communications can be incorporated in the analysis of multicellular programs as suggested by Reviewer #2. Additionally, we will correct all the figure and text typos identified by the reviewers. Finally, we will provide an R package (https://github.com/saezlab/MOFAcellulaR) and python implementations (https://liana-py.readthedocs.io/en/latest/notebooks/mofacellular.html) that facilitate the use of our approach.

Please find below the point-by-point response to the reviewers in blue, numbered for convenience.

__Reviewer #1 (Evidence, reproducibility and clarity (Required)): __

Remark to authors

Flores et al. present a pipeline in which they leverage MOFA framework, a matrix factorization algorithm to infer multi-cellular programs (MCPs). Learning and using MCP has already been proposed by others. Yet, authors pursue a similar goals by using MOFA, providing a cell*sample matrix for different cell types as different views (instead of multiple modalities/views) as the input. They later apply MOFA using this data format on a series of applications to analyze acute and chronic human heart failure single-cell datasets using MCPs. Authors further try to expand their analysis by incorporating other modalities.

Major points:

1.1 As briefly outlined in the remarks, the current manuscript needs novel findings and methodology to grant a research article which I can' see here. The underlying matrix factorization is the original MOFA (literally imported in the code) with no modification to further optimize the method toward the task. While I appreciate and acknowledge the author's efforts resulting in a detailed analysis of heart samples, I think all of these could have been part of MOFA's existing tutorials.

Response 1.1 As the reviewer correctly states, we used the framework and code of MOFA. The novelty lies in its application for the unsupervised analysis of samples from cross-condition single-cell data and the inference of MCPs. MOFA is a statistical framework implementing a generalization of group factor analysis with fast inference and its current version fits the task of MCP inference and unsupervised analysis of samples across cell-types that provides a more flexible modeling alternative than current available methods (as presented in Table 1 of the manuscript). Current work on MCP inference is based on the premise of multi-view factorization with distinct statistical modeling alternatives. As mentioned in the discussion of our manuscript, three main points distinguish our discussed methodology from present alternatives and provide evidence about its relevance and uniqueness over available tools:

Simultaneous unsupervised analysis of samples across cell-types and inference of MCPs, together with comprehensive interpretable descriptions of the reconstruction of the original multi-view dataset. This is only currently possible with scITD (Mitchel et al, 2022) and is compared in the manuscript. DIALOGUE (Jerby-Arnon & Regev, 2022) is limited to the generation of MCPs and Tensor-cell2cell (Armingol et al, 2021) is only focused in cell-communications with limited interpretability.

Flexible non-overlapping feature set that handles better technical effects such as background expression, as discussed in section “__2.2 Multicellular factor analysis for an unsupervised analysis of samples in single-cell cohorts”. __Moreover, as mentioned by the reviewer in a later point (Reviewer comment 1.2), this enables joint modeling of distinct aspects of the tissue, such as cell compositions, cell communications (preliminary work: https://liana-py.readthedocs.io/en/latest/notebooks/mofatalk.htm) and spatial organization.

Joint-modeling of independent atlases that enables meta-analysis at the sample level of cross-condition single-cell data. No currently available methodology is capable of performing similar modeling. For these reasons, we believe that our work is worth being discussed and presented to the community as a research article. We will modify the discussion to put more emphasis on the added value of group factor analysis as implemented in MOFA.

Moreover, we now provide an R package (https://github.com/saezlab/MOFAcellulaR) and python implementations within our analysis framework LIANA (https://liana-py.readthedocs.io/en/latest/notebooks/mofacellular.html) that facilitates the usage of our proposed methodology. The R and python implementations are compatible with current Bioconductor and scverse pipelines, respectively.

Application of our methodology to heart failure datasets also revealed novel knowledge about heart disease processes:

In myocardial infarction, we found that our MCPs associated with cardiac remodeling capture cell-state-independent gene expression changes. This provides a novel understanding on the effect of disease contexts in the expression profiles of specialized cells. This finding was not reported in the original atlas publication.

In chronic heart failure, we identified a conserved MCP of cardiac remodeling across patient cohorts and etiologies, suggesting a common chronic phase between distinct initial causes of heart failure.

Moreover, we showed that deconvoluted chronic heart failure MCPs from bulk transcriptomics better classify patients in comparison to classic cell-type composition deconvolution of bulk data. To our knowledge, this finding was not presented in any of the manuscripts of other methodologies focused on MCPs.

Altogether, our current work shows a novel application of group factor analysis for the simultaneous estimation of MCPs and the sample-level unsupervised analysis of cross-condition single cell data. We showed the unique features compared to current available tools. Distinct post-hoc analysis in combination with other data modalities shows the biological relevance of our proposed methodology to complement the tissue-centric knowledge of disease.

1.2 How can you explain that the results in donor-level analyses are not due to technical artifacts (batch variation)? Can this be used to infer a new patient similarity map? For example, I would test this by leaving out a few patients from training, projecting them, and seeing where they would end up in the manifold or classifying disease conditions for new patients and explaining the classification by MCPs responsible for that condition.

Response 1.2 When knowledge of the technical batches is available it is possible to test for association between these labels and the factors encoding MCPs as shown in Figure 2.

In our current applications, we additionally showed the biological relevance of our estimated MCPs by mapping them to spatial and bulk data sets, which is a direct way of testing how generalizable were our findings:

In the application of MOFA to human myocardial infarction data, we mapped the gene loadings conforming the MCP associated with cardiac remodeling to paired spatial transcriptomics datasets. We showed that in general, the cell-type specific expression of the MCP of cardiac remodeling encompassed larger areas in ischemic and fibrotic samples compared to myogenic (control) samples.

In the application of MOFA to chronic human end-stage heart failure data, we mapped the gene loadings conforming the MCP associated with cardiac remodeling to 16 independent bulk transcriptomics datasets of heart failure. There we showed that the cell-type specific expression of the MCP of cardiac remodeling separates heart failure patients from control individuals. Regarding the generation of new patient similarity maps, it is possible to estimate the positions of new samples in the manifold formed by the factors representing the MCPs. As suggested by the reviewer we will show this by classifying heart failure single-cell samples using MCPs of two independent patient cohorts (presented in section 2.7).

1.3 The bulk and spatial analysis are used posthoc after running MOFA, I think since MOFA can use non-overlapping features set, it would be interesting to see if deconvoluted bulk or ST data can be encoded as another view (one view from scRNAseq data for each cell-type and another view from bulk RNA-seq or ST, you can get normalized expression per spot (for ST) or per sample (for bulk) and use them as input.

Response 1.3 Thanks for the suggestion. We agree that the possibility of using non-overlapping features opens options of complex models that include the cell-type compositional and organizational aspects of tissues. However these features must be quantified in the same sample, thus it is limited to samples profiled simultaneously at different scales.

We will present the results of a sample-level joint model of multicellular programs together with cell-proportions and spatial dependencies using the myocardial infarction dataset presented in section 2.2. For this dataset based on our previous work we have the compositions of major cell-types and their spatial relationships based on spatially contextualized models (Kuppe et al, 2022). We will run a MOFA model and show how it can be used to find factors associated with structural and molecular features of tissues.

__Minor: __

1.4 Some figure references are not correct (e.g., "the single-cell data into a multi-view data representation by estimating pseudo bulk gene expression profiles for each cell-type across samples (Figure 1b)." should be figure 2b)

Response 1.4 Thanks for pointing this out. We apologize for these mistakes and we will adjust all labels correctly.

1.5 The paper is well written, but there could be some more clarifications about what authors consider as cell-type and cell-state, condition, MCPs which I think is critical to current analysis (see here https://linkinghub.elsevier.com/retrieve/pii/S0092867423001599) for the reader not familiar with those concepts.

Response 1.5 We agree with the reviewer that it is important to introduce these concepts in more detail to avoid confusion. We will adapt the current manuscript to incorporate these definitions in the introduction.

__Reviewer #1 (Significance (Required)): __

1.6 While I find the concept of MCPs interesting, the current work seems like a series of vignettes and tutorials by simply applying MOFA on different datasets (The authors rightfully state this). However, It needs to be clarified what the novelty is since there is no algorithmic improvement to current MCP methods (because there is no new method) nor novel biological findings. Additionally, even in the current form, the applications are limited to the heart, and the generalization of this proposed analysis pipeline to other tissues and datasets is not explored. Overall, the paper lacks focus and novelty, which is required to grant a publication at this level.

Response 1.6 As mentioned in response 1.1, we show that group factor analysis as implemented in MOFA has advantages given its flexibility of the feature space, the joint-modeling of independent datasets, and the interpretability of the model. We will make these advantages clearer in the discussion, and we will explicitly mention the disadvantages and lack of functionalities of available methods.

The applications were mainly done in heart data for consistency although they represent four distinct single-cell datasets, one spatial transcriptomics dataset, and 16 independent bulk transcriptomics datasets. For completeness, as suggested by the reviewer, we will show the application of our methodology to peripheral blood mononuclear cell data of lupus samples (preliminary results: https://liana-py.readthedocs.io/en/latest/notebooks/mofacellular.html)

__expertise: Computational biology, single-cell genomics, machine learning __

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

Summary:

The authors use MOFA, an unsupervised method to analyze multi-omics data, to create multicellular programs of cross-condition multi-sample studies. First, for each cell-type, a pseudobulk expression matrix per sample is created. The cell-type now functions as the separate view, typically reserved for the different omics layers in MOFA. This then results in a latent space with a certain number of factors across samples. The factors, representing coordinated gene expression changes across cell-types, can then be checked for associations with covariates of interest across the samples.

MOFA is well-suited for this task, as it can handle missing data and it is a linear model facilitating the interpretation of the factors. Users should be aware that MOFA can estimate the number of factors, but the pseudobulk profiles require a rigorous selection of cell-type specific marker genes. The result will be most suited for downstream analysis if there is a clear association with one factor and a clinical covariate of interest. In a final step, a positive or negative gene signature can be created by setting a cut-off on the gene weights for that specific factor.

The method is applied on 3 separate data sets of heart disease, each time demonstrating that at least one of the factors is associated with a disease covariate of interest. The authors also compare the method to a competitor tool, scITD, and explore to what extent a factor mainly captures variance associated with (i) a general condition covariate or rather (ii) specific cell states.

The multicellular programs are also mapped to spatial data with spot resolution. Though this analysis does not bring any novel biological insight in the use case, it does support the claim that the programs are associated with the covariate of interest.

The most interesting applications of MOFA are in my opinion the potential for meta-analysis of single-cell studies and validation of cell-type specific gene signatures with publicly available bulkRNAseq data sets.

The authors provide various data sets and data types to support their claims and the paper is well written. The relevant code and data has been made available.

We thank the reviewer for the positive comments to our work.

__Major comments __

2.1 What is the added value of the gene signatures obtained from MOFA compared to e.g. a naive univariate approach? In theory, a similar collection of genes or gene signature could be obtained by running a differential gene expression analysis across the samples for each cell-type (e.g. myogenic vs ischemic ) and applying a set of relevant cut-offs or filters on the results. In other words, does MOFA detect genes that would otherwise be missed?

Response 2.1 Thank you for the relevant comment. The original motivation of our work is the unsupervised analysis of samples based on a manifold formed by a collection of multicellular molecular programs. We envisioned that this unsupervised analysis would be relevant in situations where a clear histological or clinical classification of samples is not possible with reliability. As mentioned by Reviewer #1 in comment 1.2, one advantage of these approaches is that they create patient similarity maps, which have been shown useful to stratify patients in a recent analogous work in multiple sclerosis (Macnair et al, 2022). The cell-type signatures obtained from relevant factors explaining the patient stratification avoid the likelihood of performing “double dipping” by avoiding the need of a direct differential expression analysis between newly formed groups.

In our applications, the generation of cell-type signatures (here called multicellular programs) associated to a specific clinical covariate (eg. control vs perturbation) are post-hoc analyses of the generated manifold. And as the reviewer correctly points out, these signatures should be similar to performing direct differential expression analysis between those patient conditions. In the related work of scITD (Mitchel et al, 2022) the authors showed high concordance between the cell-type signatures and the results of differential expression analysis. For completion, we will similarly quantify the degree of overlap between genes of our generated signatures with the ones coming from differential expression analysis.

It is relevant to mention that in complex experimental designs with multiple conditions, our approach facilitates patient ordering, which allows the understanding of one condition in the context of all the others, avoiding the need of multiple testing and the definition of multiple contrasts, as mentioned in the text.

We will incorporate these points in the discussion section of the manuscript.

2.2 Could scITD also be used for meta-analysis or could the obtained gene signatures of that method also be mapped to bulkRNAseq data? If so, it would be interesting to show the relative performance with MOFA. If not, this specific advantage should be highlighted.

Response 2.2 Thank you for pointing this out. scITD does not provide a group-based model to perform meta-analysis, and this feature is one of the main advantages of group factor analysis as currently implemented in MOFA. We will highlight this feature in Table 1 and in the discussion.

Although scITD signatures of a single study could be mapped to bulk transcriptomics data, the stringent tensor representation leads to the generation of signatures that may be influenced by technical effects as shown in the manuscript section 2.2. Thus we believe that the flexibility of the feature space in MOFA is an advantage for this task. We will add this observation to the discussion.

2.3 Users need to specify gene set signatures based on the weights for a factor of interest. This might suggest a limitation to categorical covariates of interest. If the authors see potential for a continuous covariate of interest, this should at least be highlighted in the text and if possible demonstrated on a use case.

Response 2.3 In our applications we limited ourselves to categorical variables, however, it is possible to associate factors to continuous variables. An implementation of the association with continuous variables is already available in our newly created R package “MOFAcellulaR”: https://github.com/saezlab/MOFAcellulaR/blob/main/R/get_associations.R.

The datasets we analyzed have no continuous clinical covariates to showcase this functionality, but as suggested by the reviewer we will highlight this feature in the text.

__Minor comments __

2.4 In Figure 2c the association between factor 2 and the technical factor shows a very strong outlier. Please verify that the association is still significant after applying a more robust statistical test (e.g. non-parametric test as Wilcoxon).

Response 2.4 Thanks for the observation, we will test these differences with a non-parametric test.

2.5 For mapping the cell-type specific factor signatures to bulk transcriptomics, the exact performed comparison or model is unclear. There are seven cell-type signatures for each sample in every study. Was there a t-test run for each cell-type or was a summary measure taken across the cell-types? he thresholding is also rather lenient (adj. p-val 0.1).

Response 2.5 We are sorry for not being clear about our procedure. After identifying the multicellular program associated with heart failure estimated from the two single cell studies meta-analyzed, we calculated the weighted mean expression of the seven cell-type signatures independently to every sample of the 16 bulk studies. In other words each sample within each bulk study will be represented by a vector of 7 values representing the relative expression of a cell-type specific signature (Figure 6D-left). For each bulk transcriptomics study, first, we centered the gene expression data before calculating the weighted mean.

In supplementary figure 4-e we show the results of performing a t-test of the cell-type scores between heart failure and control samples within each study. Given the relative low sample size of most of the studies (affecting the power of the test), we chose a not so stringent adjusted p-value. For completion, we will show the results of a more classical threshold (adj. p-value

2.6 typo in abstract: In sum, our framework serves as an exploratory tool for unsupervised analysis of cross-condition single-cell ***atlas*** and allows for the integration of the measurements of patient cohorts across distinct data modalities

Response 2.6 Thanks for pointing out this typo. We will modify the text.

2.7 In Figure 4a it is not clear to me why on the one hand we see marker enrichment vs loading enrichment with healthy and disease.

Response 2.7 We apologize, this is a typo after editing the labels. Both should contain the marker enrichment label. We will fix this.

2.8 IN Figure 4b it would help if the same color scheme would be maintained throughout the paper (here now black and white) and if for the cell states the boxplots would be connected per condition, emphasizing the (absence) of change across cell states within a condition.

Response 2.8 We thank the reviewer for the suggestion. We will reorganize the panels showing the gene expression per condition and fix the color scheme.

__Reviewer #2 (Significance (Required)): __

__General assessment: __

2.9 MOFA is well-suited for detecting multicellular programs because it can handle missing data and allows for easy interpretation of the factors as a linear method. It might have particular potential for meta-analysis across multiple studies and reevaluating bulkRNAseq data sets, but in the current manuscript it is unclear to what extent this is a specific advantage of MOFA or could also be done with competitors. The authors show how the obtained results and associations with clinical covariates can be validated across multiple data types. How the resulting multicellular programs can provide additional biological insight or form the starting point for additional downstream analysis (e.g. cell communication) is not covered in the paper.

Response 2.9 We thank the reviewer for highlighting the methodological advantages of group factor analysis for the estimation of multicellular programs and the unsupervised analysis of samples from cross-condition single-cell atlas. As mentioned in response 1.1 and 2.2, the added value of our methodology is the flexibility of feature views (that goes beyond gene expression) and simultaneous modeling of independent single-cell datasets, a feature not present in any of the currently available methods that facilitates the meta-analysis of datasets across modalities.

While we interpret the presented multicellular programs in the context of cellular functions and the division of labor of cell states, it is true that we did not attempt to provide mechanistic hypotheses, for example, via cell-cell communication, on how this coordination across cell-types emerges.

Previous work of the related tool Tensor-cell2cell (Armingol et al, 2021) has presented the idea of the estimation of multicellular programs from cell-cell communications and group factor analysis can also be used for this task (preliminary work: https://liana-py.readthedocs.io/en/latest/notebooks/mofatalk.html). We will discuss in the text perspectives on how the estimation of multicellular programs can be linked to the inference of cell communications from single-cell data together with analysis alternatives previously proposed by scITD and Tensor-cell2cell. However, we believe that this question requires further work and it is out of scope of our current manuscript.

__Audience: This paper will be mainly of interest to a specialized public interested in unsupervised methods for large scale multi-sample and multi-condition studies. __

__Reviewer: main background in the analysis of scRNAseq data. __

__Reviewer #3 (Evidence, reproducibility and clarity (Required)): __

This manuscript by Saez-Rodriguez and colleagues proposes to repurpose Multi-Omics Factor Analysis for the use of single cell data. The initial open problem stated by the paper is the need for a framework to map multicellular programs (such as derived from factor analysis) to other modalities such as spatial or bulk data. The authors propose to repurpose MOFA for use in single cell data. Case studies involve human heart failure datasets (and focuses on spatial and bulk comparisons).

There are particular issues with clarity regarding the key methodological contribution (and assessment of it), discussed under significance.

__Reviewer #3 (Significance (Required)): __

3.1 I am very puzzled by the repeated claims the manuscript makes that their central methodological contribution and innovation is to use MOFA for single cell data. One of their citations for MOFA is to MOFA+, which is precisely that (in a relatively popular manuscript published by the original authors of MOFA and not overlapping with the present authors). I am left to wonder what I missed.

Response 3.1 We apologize for the misunderstanding, as mentioned in the response to review 1.1 and explained by reviewer 2’s summary, the main objective of our work is to use the statistical framework of group factor analysis for the inference of multicellular programs and the sample-level unsupervised analysis of cross-condition single-cell data, which is a distinct task to multimodal integration (Argelaguet et al, 2021).

While it is true that MOFA+ introduced expansions to the model for the modeling of single-cell data, namely fast inference and group-based modeling, the main focus in their applications is the multimodal integration of data, where each cell is represented by a collection of distinct collection of features (e.g. chromatin accessibility and gene expression). Unlike multimodal integration, here we propose a different approach to analyze single-cell data at the sample level instead of the cell level, without modifying the underlying statistical model (see section 2.1 of the manuscript).

In detail, what we assume is that samples of single-cell transcriptomics data (e.g. tissue from a patient) can be represented by a collection of independent vectors collecting the gene expression information of cell types composing the tissue analyzed. Decomposition of these multiple views with group factor analysis produces a manifold that captures multicellular programs (coordinated expression processes across cell-types), or shared variability across cell-types simultaneously. Altogether, this represents a novel usage of group factor analysis in an application for the inference of multicellular programs, where the main focus is not at the cell-level but at the patient level.

As a side note, Britta Velten, one of main developers of MOFA and coauthor of both the MOFA and MOFA+ papers, is a contributor and coauthor of this manuscript, and Ricard Argelaguet, who also led both versions of MOFA, gave us helpful feedback and is acknowledged as such on this work.

3.2 Multimodal integration methods are fairly numerous and even if they're not all exactly factor analyses, it's strange to argue that MOFA fills some unique conceptual gap. I agree it fills something of an interesting gap (except for MOFA+ already filling it), but it's not like the quite popular spatial to single-cell integration approaches aren't doing similar things. If this is a methods paper (as it is presented) then there would have to be very substantially more comparative evaluation to these other approaches.

Response 3.2 As presented in the previous response (3.1) our current work is not focused on multimodal integration, but rather the inference of multicellular programs and the sample-level unsupervised analysis of single-cell data. Given this, in the current manuscript we compared our proposed methodology with the only three other available methods that address at least partially the inference of multicellular programs (see Table 1 in our manuscript). In response 1.1 and 3.2 we discussed the advantages of our proposed methodology compared to available methods. In the manuscript section 2.2 we compared group factor analysis with tensor decomposition and showed that the former better deals with technical artifacts and better identifies known patient groups.

We will distinguish our work from multimodal integration explicitly in the introduction and the manuscript section 2.1 to avoid confusions.

3.3 The biological use cases are comparatively interesting and dominate the manuscript (but are still presented principally as use cases rather than a compelling biological narrative of their own).

Response 3.3 The focus of our manuscript was the reintroduction of group factor analysis for the novel applications of the inference of multicellular programs and the sample-level unsupervised analysis from single-cell data. Given the distinct possibilities of post-hoc analyses, we mainly used acute and chronic heart failure data to showcase the utility of MOFA to connect spatial and bulk modalities with single-cell data.

That said, as discussed in response 1.1, our analyses allowed to generate novel hypotheses of these datasets:

In myocardial infarction, we found that our estimated multicellular programs associated with cardiac remodeling capture cell-state-independent gene expression changes. This provides a novel understanding of the effect of disease contexts in the expression profiles of specialized cells. In other words, we found that cell-states, regardless of their specialized function, share a common response in the tissue context.

In chronic heart failure, we identified a conserved multicellular program of cardiac remodeling across patient cohorts and etiologies, suggesting a common chronic phase between distinct initial causes of heart failure, which again may be linked to the dominating response to the tissue context that is shared across etiologies.

These two results support the observation that deconvoluted chronic heart failure multicellular programs from bulk transcriptomics better classify patients in comparison to classic cell-type composition deconvolution of bulk data. To our knowledge, this finding was not presented in any of the manuscripts of other methodologies focused on MCPs. We summarize these results in the third paragraph of the discussion in the manuscript:

“In an application to a collection of public single-cell atlases of acute and chronic heart failure, we found evidence of dominant cell-state independent transcriptional deregulation of cell-types upon myocardial infarction. This may suggest that while certain functional states within a cell-type are more favored in a disease context, most of the cells of a specific type have a shared transcriptional profile in disease tissues. If part of this shared transcriptional profile is interpreted as a signature of the tissue microenvironment that drives cells in tissues towards specific functions, this result may also indicate that a major source of variability across tissues, besides cellular composition, is the degree in which the homeostatic transcriptional balance of the tissue is disturbed. By combining the results of multicellular factor analysis with spatial transcriptomics datasets, we explored this hypothesis and identified larger areas of cell-type-specific transcriptional alterations in diseased tissues. Given these observations on global alterations upon myocardial infarction, we meta-analyzed single-cell samples from two additional studies of healthy and heart failure patients with multiple cardiomyopathies. Here, we found a conserved transcriptional response across cell-types in failing hearts, despite technical and clinical variability between patients. Further, we could find traces of these cell-type alterations in independent bulk data sets. These observations suggest that our approach can estimate cell-type-specific transcriptional changes from bulk data that, together with changes in cell-type compositions, describe tissue pathophysiology. Altogether, these results highlight how MOFA can be used to integrate the measurements of independent single-cell, spatial, and bulk datasets to measure cell-type alterations in disease.”

To fully assess the relevance of these observations, they should be investigated in more datasets and analyses, where shared functional cell-states across distinct heart failure etiologies are identified and then compared at their compositional and molecular level. This, in our opinion, represents an independent study on its own.

3.4 Altogether, I found the framing of this manuscript very puzzling. It is possible the result would be more clearly presented if the use case was the major focus rather than the more conceptual point about factor analysis.

Response 3.4 Thanks for the suggestion. The major aim of this manuscript is to highlight the versatility of the generalization of group factor analysis as implemented in MOFA for novel applications in single-cell data analysis, beyond multimodal integration of single cells. The definition of multicellular programs from single-cell data and its sample-level unsupervised analysis are relatively new analyses in the field, and thus we believe that it is timely to show how a known statistical framework can be used for these applications.

We believe that a detailed analysis of single-cell datasets of heart failure deserves its own focus and it is out of scope of our current objective with this manuscript. We apologize for the apparent misunderstanding of the objective of our methodology. We will add these distinctions in the introduction of the manuscript.

References

Argelaguet R, Cuomo ASE, Stegle O & Marioni JC (2021) Computational principles and challenges in single-cell data integration. Nat Biotechnol 39: 1202–1215

Armingol E, Baghdassarian H, Martino C, Perez-Lopez A, Knight R & Lewis NE (2021) Context-aware deconvolution of cell-cell communication with Tensor-cell2cell. BioRxiv

Jerby-Arnon L & Regev A (2022) DIALOGUE maps multicellular programs in tissue from single-cell or spatial transcriptomics data. Nat Biotechnol 40: 1467–1477

Kuppe C, Ramirez Flores RO, Li Z, Hayat S, Levinson RT, Liao X, Hannani MT, Tanevski J, Wünnemann F, Nagai JS, et al (2022) Spatial multi-omic map of human myocardial infarction. Nature 608: 766–777

Macnair W, Calini D, Agirre E, Bryois J, Jaekel S, Kukanja P, Stokar-Regenscheit N, Ott V, Foo LC, Collin L, et al (2022) Single nuclei RNAseq stratifies multiple sclerosis patients into three distinct white matter glia responses. BioRxiv

Mitchel J, Gordon MG, Perez RK, Biederstedt E, Bueno R, Ye CJ & Kharchenko P (2022) Tensor decomposition reveals coordinated multicellular patterns of transcriptional variation that distinguish and stratify disease individuals. BioRxiv

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

This manuscript by Saez-Rodriguez and colleagues proposes to repurpose Multi-Omics Factor Analysis for the use of single cell data. The initial open problem stated by the paper is the need for a framework to map multicellular programs (such as derived from factor analysis) to other modalities such as spatial or bulk data. The authors propose to repurpose MOFA for use in single cell data. Case studies involve human heart failure datasets (and focuses on spatial and bulk comparisons).

There are particular issues with clarity regarding the key methodological contribution (and assessment of it), discussed under significance.

Significance

1. I am very puzzled by the repeated claims the manuscript makes that their central methodological contribution and innovation is to use MOFA for single cell data. One of their citations for MOFA is to MOFA+, which is precisely that (in a relatively popular manuscript published by the original authors of MOFA and not overlapping with the present authors). I am left to wonder what I missed.
2. Multimodal integration methods are fairly numerous and even if they're not all exactly factor analyses, it's strange to argue that MOFA fills some unique conceptual gap. I agree it fills something of an interesting gap (except for MOFA+ already filling it), but it's not like the quite popular spatial to single-cell integration approaches aren't doing similar things. If this is a methods paper (as it is presented) then there would have to be very substantially more comparative evaluation to these other approaches.
3. The biological use cases are comparatively interesting and dominate the manuscript (but are still presented principally as use cases rather than a compelling biological narrative of their own).

Altogether, I found the framing of this manuscript very puzzling. It is possible the result would be more clearly presented if the use case was the major focus rather than the more conceptual point about factor analysis.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

The authors use MOFA, an unsupervised method to analyze multi-omics data, to create multicellular programs of cross-condition multi-sample studies. First, for each cell-type, a pseudobulk expression matrix per sample is created. The cell-type now functions as the separate view, typically reserved for the different omics layers in MOFA. This then results in a latent space with a certain number of factors across samples. The factors, representing coordinated gene expression changes across cell-types, can then be checked for associations with covariates of interest across the samples. MOFA is well-suited for this task, as it can handle missing data and it is a linear model facilitating the interpretation of the factors. Users should be aware that MOFA can estimate the number of factors, but the pseudobulk profiles require a rigorous selection of cell-type specific marker genes. The result will be most suited for downstream analysis if there is a clear association with one factor and a clinical covariate of interest. In a final step, a positive or negative gene signature can be created by setting a cut-off on the gene weights for that specific factor. The method is applied on 3 separate data sets of heart disease, each time demonstrating that at least one of the factors is associated with a disease covariate of interest. The authors also compare the method to a competitor tool, scITD, and explore to what extent a factor mainly captures variance associated with (i) a general condition covariate or rather (ii) specific cell states. The multicellular programs are also mapped to spatial data with spot resolution. Though this analysis does not bring any novel biological insight in the use case, it does support the claim that the programs are associated with the covariate of interest. The most interesting applications of MOFA are in my opinion the potential for meta-analysis of single-cell studies and validation of cell-type specific gene signatures with publicly available bulkRNAseq data sets. The authors provide various data sets and data types to support their claims and the paper is well written. The relevant code and data has been made available.

Major comments

• What is the added value of the gene signatures obtained from MOFA compared to e.g. a naive univariate approach? In theory, a similar collection of genes or gene signature could be obtained by running a differential gene expression analysis across the samples for each cell-type (e.g. myogenic vs ischemic ) and applying a set of relevant cut-offs or filters on the results. In other words, does MOFA detect genes that would otherwise be missed?
• Could scITD also be used for meta-analysis or could the obtained gene signatures of that method also be mapped to bulkRNAseq data? If so, it would be interesting to show the relative performance with MOFA. If not, this specific advantage should be highlighted.
• Users need to specify gene set signatures based on the weights for a factor of interest. This might suggest a limitation to categorical covariates of interest. If the authors see potential for a continuous covariate of interest, this should at least be highlighted in the text and if possible demonstrated on a use case.

Minor comments

• In Figure 2c the association between factor 2 and the technical factor shows a very strong outlier. Please verify that the association is still significant after applying a more robust statistical test (e.g. non-parametric test as Wilcoxon).
• For mapping the cell-type specific factor signatures to bulk transcriptomics, the exact performed comparison or model is unclear. There are seven cell-type signatures for each sample in every study. Was there a t-test run for each cell-type or was a summary measure taken across the cell-types? he thresholding is also rather lenient (adj. p-val 0.1).
• typo in abstract: In sum, our framework serves as an exploratory tool for unsupervised analysis of cross-condition single-cell atlas and allows for the integration of the measurements of patient cohorts across distinct data modalities
• In Figure 4a it is not clear to me why on the one hand we see marker enrichment vs loading enrichment with healthy and disease.
• IN Figure 4b it would help if the same color scheme would be maintained throughout the paper (here now black and white) and if for the cell states the boxplots would be connected per condition, emphasizing the (absence) of change across cell states within a condition.

Significance

General assessment:

MOFA is well-suited for detecting multicellular programs because it can handle missing data and allows for easy interpretation of the factors as a linear method. It might have particular potential for meta-analysis across multiple studies and reevaluating bulkRNAseq data sets, but in the current manuscript it is unclear to what extent this is a specific advantage of MOFA or could also be done with competitors. The authors show how the obtained results and associations with clinical covariates can be validated across multiple data types. How the resulting multicellular programs can provide additional biological insight or form the starting point for additional downstream analysis (e.g. cell communication) is not covered in the paper.

Audience: This paper will be mainly of interest to a specialized public interested in unsupervised methods for large scale multi-sample and multi-condition studies.

Reviewer: main background in the analysis of scRNAseq data.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Remark to authors

Flores et al. present a pipeline in which they leverage MOFA framework, a matrix factorization algorithm to infer multi-cellular programs (MCPs). Learning and using MCP has already been proposed by others. Yet, authors pursue a similar goals by using MOFA, providing a cell*sample matrix for different cell types as different views (instead of multiple modalities/views) as the input. They later apply MOFA using this data format on a series of applications to analyze acute and chronic human heart failure single-cell datasets using MCPs. Authors further try to expand their analysis by incorporating other modalities.

Major points:

As briefly outlined in the remarks, the current manuscript needs novel findings and methodology to grant a research article which I can' see here. The underlying matrix factorization is the original MOFA (literally imported in the code) with no modification to further optimize the method toward the task. While I appreciate and acknowledge the author's efforts resulting in a detailed analysis of heart samples, I think all of these could have been part of MOFA's existing tutorials.

How can you explain that the results in donor-level analyses are not due to technical artifacts (batch variation)? Can this be used to infer a new patient similarity map? For example, I would test this by leaving out a few patients from training, projecting them, and seeing where they would end up in the manifold or classifying disease conditions for new patients and explaining the classification by MCPs responsible for that condition.

The bulk and spatial analysis are used posthoc after running MOFA, I think since MOFA can use non-overlapping features set, it would be interesting to see if deconvoluted bulk or ST data can be encoded as another view (one view from scRNAseq data for each cell-type and another view from bulk RNA-seq or ST, you can get normalized expression per spot (for ST) or per sample (for bulk) and use them as input.

Minor:

Some figure references are not correct (e.g., "the single-cell data into a multi-view data representation by estimating pseudo bulk gene expression profiles for each cell-type across samples (Figure 1b)." should be figure 2b)

The paper is well written, but there could be some more clarifications about what authors consider as cell-type and cell-state, condition, MCPs which I think is critical to current analysis (see here https://linkinghub.elsevier.com/retrieve/pii/S0092867423001599) for the reader not familiar with those concepts.

Significance

While I find the concept of MCPs interesting, the current work seems like a series of vignettes and tutorials by simply applying MOFA on different datasets (The authors rightfully state this). However, It needs to be clarified what the novelty is since there is no algorithmic improvement to current MCP methods (because there is no new method) nor novel biological findings. Additionally, even in the current form, the applications are limited to the heart, and the generalization of this proposed analysis pipeline to other tissues and datasets is not explored. Overall, the paper lacks focus and novelty, which is required to grant a publication at this level.

expertise: Computational biology, single-cell genomics, machine learning

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

Learn more at Review Commons

---

Reply to the reviewers

The work presented here examined the combined contribution of intermediate gray matter spinal interneurons of the spinal lumbar enlargement (L2-L4) to locomotion in rats. By targeting this region with kainic acid, we were able to produce a specific locomotor signature that was not compensated for over time, indicating the need for cellular replacement therapies in the treatment of such spinal cord injuries leading to the loss of spinal enlargement intermediate gray matter. Further, the newly developed techniques of a combinatorial behavioral assessment using Random Forest classification and a machine learning intermediate gray matter neuronal loss assessment established in this work add an unbiased, in-depth approach that we are making available to others.

The reviewers have critically evaluated our work and highlighted points of weakness either in the research itself or in connecting with our audience. Below is our detailed response to all the comments as well as our revision plan for submission. We believe we have been able to sufficiently address the concerns that were voiced to strengthen our manuscript and express our gratitude for the feedback.

Reviewer #1 (Evidence, reproducibility and clarity (Required)): __ In this paper, Kuehn and colleagues report on the analysis of functional impairments following intermediate gray matter lesion with kainic acid. The image convincingly show that mostly purely grey matter lesion can be achieved throughout the paper. The authors took care to do a battery of well-designed behavioral tests and sophisticated analysis in order to access functional impairment. They then correlate their behavioral assessment to lesion size, the number of NeuN positive cells in layers V-VII epicenters as well motoneuron numbers and the percentage of white matter. Overall, the manuscript is well written, nicely framed in the existing literature, very clear and the experiments are simple but well designed. The behavioral testing and evaluations including random forest ranking are well performed. The methodology is complete and would allow reproducing the experiments. Statistics are used appropriately. We have however some reserves and comments on some of the results and interpretations. Addressing these comments would not involve new experiments but new re-analysis of the existing datasets.

Major comments:__

__ While the claims that grey matter lesions trigger major behavioral impairments is convincing in particular with the refine behavioral experiments performed, the key claim that only interneuron loss in layer V-VII mediates those deficits is currently not supported by the presented data. In particular, we would suggest that the lesions performed, in contrast to the claims, are not purely and selectively impacting layer V-VII but might also impact layers VIII-IX. We think that presenting neuronal counts based on NeuN staining separately for layer I-IV, V-VII, VIII-IX and comparing control vs KA is necessary. Only with these data can conclusions be supported either in the direction suggested by the authors or otherwise.__

• Although primarily targeting laminae V-VII, we realize this is not exclusively doing so with our lesion model. We understand the value of what you request and are retraining our computer models to be able to do the additional neuronal quantification in laminae I-IV, VIII, IX. We will then combine lamina VIII with laminae V-VII to make up the intermediate gray matter NeuN counts. Completion of all manually validated new analysis is ongoing and will be finished shortly. We plan on adding this additional analysis to the paper, which means much of Figure 6 and Supplementary Figure 3 will be altered and partially for Figure 7, but we won’t know exactly how until we finish the analysis. Tracked changes are shown in the updated manuscript PDF and highlighted text may change depending on results of this analysis.

Another claim relative to the lack of involvement of motoneurons in the related behavioral deficits is also difficult to resolve with the current data. Motoneurons have been identified based on NeuN staining and size. While this is not the state of the art (ChAT staining would have been preferable), it remains acceptable. However, the data presented figures 7 and 8 show a very wide range in the motoneuron count (15 to 50) indicating either motoneuron loss or a count performed at different lumbar levels in the animals. This raises questions on the model (is it really involving only layers V-VII?) or on the interpretation of the data. Therefore we believe that motoneurons counts need to be presented separately (see above) in control vs KA groups and data need to be discussed in this perspective. Authors should also tone down the specificity of the model and involvement of motoneurons accordingly (page 20 for example).

• Although we agree with the reviewers that ChAT staining would have been preferable, we had a limited amount of tissue available. Our unbiased, machine-learning-based analysis of neuronal loss by NeuN required much of the existing tissue. However, neuronal staining has been previously established to identify motoneurons based on size inclusion (Hadi et al., 2000; Wen et al., 2015), as we have used here. Additionally, we will be including total neuronal analysis from lamina IX as requested (please see answer to previous comment).

• By including the Controls along with the KA rats, we postulate that the wide range of motoneuron numbers is due to natural individual variation as well as due to variation at each spinal level, and not due to the KA lesion, as the KA animals have a range of motoneuron counts, sometimes even greater than the controls (Figure 7 and 8). However, as requested, we have split L2, L3 and L4 (graphs below) and still do not see a correlation with behavioral performance (BBB and inclined beam). The variation due to spinal level may partially be explained by the fact that there are different numbers of motoneurons at each spinal level, dependent upon the number of muscles each spinal level is responsible for and the number of motor columns at a given level (Mohan et al., 2015; Nicolopoulos-Stournaras & Iles, 1983). These counts are taken from a given section and not the entirety of the spinal level, adding further possible variation. Moreover, we have removed the controls as suggested (graphed below) for motoneuron analysis and still do not see a correlation between the number of motoneurons and behavioral performance (BBB and inclined beam). We do not find this the correct way to graphically represent the data as it does not allow the reader to see the natural number of motoneurons that exist at each spinal level and variation within as well as knowing that this is not due to injury correlating with behavioral differences, and therefore we would like to keep these graphs with controls in the manuscript.

We have toned down the specificity of the model and involvement of motoneurons as requested on pages 20-21.

Most of the conclusions rely on correlations that include control animals (injected with saline hence with no lesions and no behavioral deficits; Fig 6 and 7). This artificially skews the correlations as those animals show no lesions and good performance in the behavioral tests. These correlations need to be performed only with KA injected animals to determine the respective involvements of interneurons and motoneurons.

• To address your concern, we first did as you asked and removed the controls and performed the correlation analysis for Figure 6, shown below. There are no significant correlations between neurons at each spinal level and behavior. We would further argue that unlike a contusion injury where control animals only receive a laminectomy, our control animals have very minor neuronal loss due to the saline injection itself and therefore do have a minor lesion. An example of this is seen in Figure 6 for the control animal at spinal level L2 where the pipette track is visible. Therefore, to show that the observed behavioral deficits are from the kainic acid and not the injection itself, we would argue that it is important that the control animals remain in the correlation analysis.

The long-term study (Fig 8) is performed with very few animals and hence, drawing conclusions from these animal numbers is difficult. All correlations are performed including control animals which is even more of a problem here as in Figure 6 and 7 due to the low number of animals. The authors should either add animals or remove the figure. When control animals (injected with saline) are removed (as they do not show any lesion and perform accurately in the behavior), one would actually see a correlation between the number of motoneurons and the behavioral performance (Fig. 8E,F) but not with the lesion size (Fig.8C,D).

• The long-term study was planned with more animals, but due to exclusion criteria by lesion length, the numbers remain low. We had discussed extensively whether to include this data in the manuscript or not. We decided for several reasons to include it in the manuscript within the main figures. First, it demonstrates that once these interneurons are lost, there are no compensatory mechanisms that restore function, which is quite striking given that the ones that lose weight support by 2 weeks do not regain it over a 3-month observational period. Further indicating that loss of lumbar gray matter interneurons is essential to locomotor function of hindlimbs and should be targeted in SCI replacement therapeutics. However, we do not agree with removing controls to examine the motoneuron number as there is motoneuron number variation within the lesion area and the motoneuron number from the KA animals is within the Control motoneuron range, which can be seen with the graph including the Controls. We can provide the individual spinal lesion level correlations, but this does not provide the entire picture as one level alone has not been found to be essential to the behavioral deficits. We are currently processing these animals to also provide NeuN numbers from laminae I-IV, V-VIII and IX.

Minor comments:

__ Figure 1A: if lesions are bilateral, it would be nice to illustrate this on the schematic.__

• This has been fixed. Figure 1B-D: scale bars are missing

• This has been fixed. Figure 3H: What represents the y-axis? % of completion or number of completion?

• This has been fixed. Figure 4 Table: Please specific what the acronym stands for: pLDA.

• This has been clarified in the figure legend. Figure 6 A: scale bars are missing

• This will be fixed when the data for the analysis is finished and the figure is redone. Figure 6B/C/D: please add the spinal level analyzed directly on the graphs. This will ease the comprehension.

• This has been adjusted. Figure 7 and Figure 8: While it is quite convincing that the model is purely a grey matter injury (panel C and D), the data are very much spread out for the number of motoneurons per mice (see major comments above). We would suggest to plot those data to present the number of neurons (interneurons in layer I-IV, V-VII and motoneurons) control vs KA.

• Thank you for the suggestion. We will plan on presenting the additional neuronal quantification data mentioned above by comparing Controls and KA animals.

Dots are missing on those figures (probably superimposed on top of each other). This should be changed to see all data points

• Thank you for the observation. They were superimposed but we have fixed this. Figure 8E,F: the number of motoneurons is very low also in controls. How is this explained?

• Depending on where the section was taken at each spinal level, there is variation in the number of motoneuron columns innervating targeted muscles (Mohan et al., 2015), Figure 6). Therefore, it is not surprising to see a range of motoneurons. In addition, we would like to clarify that these motoneuron counts are taken from only three sections across the lesion (from the three lesion injection epicenters), not the whole lumbar section. Often the motoneuron number in the KA group was equal to or greater than the Control group, indicating more often variation than motoneuron loss. Regardless motoneuron numbers do not correlate with the observed behavioral deficits.

__Reviewer #1 (Significance (Required)):

This paper by Kuehn and colleagues reports on the functional impairments that follow intermediate gray matter lesions using kainic acid. This work is largely confirmatory of previous studies (Magnuson et al., 1999; Hadi et al., 2000) with modern behavioral evaluation. After revision, it would provide a description of the functional impairments following those specific lesions. The paper would be informative for a specific audience in particular scientists in the field of spinal cord injury and spinal interneuron. Our field of expertise is spinal cord injury, inflammation, behavior and axon outgrowth.__

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

This manuscript reports on a pair of well-designed and well-carried out studies investigating a Kainic Acid (KA)-mediated gray matter lesion in the lumbar enlargement of adult female SD rats. The investigators demonstrate, using NeuN immunohistochemistry, that the KA lesion reduces NeuN positive cells along the length of the lumbar spinal cord from rostral to L2 to slightly caudal to L4 following 6 separate injections made on the right and left sides of the spinal cord at L2, L3 and L4. The investigators made significant efforts to avoid depleting neurons in the dorsal and ventral horns, and the evidence provided suggests they were successful. The methodology described is sound and sufficient details are provided to allow the reader to fully understand the studies. It is outstanding that the study was done while following all of the PREPARE and ARRIVE guidelines. A second major component of the work is the use of multiple outcome measures and efforts (using a Forest analysis) to develop a relatively quick, accurate and efficient system to screen or classify the injuries in individual animals within 2 weeks of the injury so that subsequent treatments could be done on animals which received injuries of sufficient severity (within a relatively narrow range) and with balanced experimental and groups. Again, with this effort the investigators were largely successful. The KA lesion results in persistent locomotor and sensorimotor deficits, that plateau early without substantial sensory dysfunction.__

Major Comments:

__ Introduction: Overall, the rationale presented and the review of the pertinent literature is solid, with the following exception: The authors state that their model should allow them to thoroughly investigate the behavioral readout of premotor IN loss. It is generally accepted that the designation of premotor interneurons refer to those directly connected to motor neurons, and while the chosen KA lesion certainly targets some premotor neurons, it also targets many other interneurons that do not directly contact motoneurons. Please revise how the lesions are referred to. In the very next paragraph the targets are defined somewhat differently as "INs and propriospinal INs in laminae V-VII in spinal levels L2-L4".__

• We agree that our wording does cause confusion to the reader and to avoid this we have now made the change from premotor INs to SpINs (pages 3-5).

• On a side note, we would like to state these are adult female Fischer rats and not adult female SD rats, also described in the methods.

Spared white matter. In many (but not all) labs, spared white matter at the epicenter is an important measurement because it presumably represents all the spared axons, such that any/all rostrocaudal communication is represented. Thus, it is the single point (or section, in this case) that has the smallest number of axons represented as stained white matter. So, to indicate that you assessed "three epicenters per spinal cord" doesn't make sense in this context, Even if you are referring to three separate KA injection sites (L2, L3 and L4). Thus, averaging three sections also doesn't really make sense because the actual epicenter should be represented by the single cross section that has the smallest area of stained white matter. Also related to spared white matter, in many labs they calculate %SWM based on a section from a control animal, and this should reduce variability because some cords shrink (injured gray matter) more than others after the injury, whether it be a contusion or mild excitotoxic injury. Please either re-calculate your SWM or provide additional justification for your current method.

• We agree with the reviewer that normally only the epicenter of the lesion needs to be examined for white matter damage as once the connection is severed it does not matter what is rostral or caudal to this site. However, in our case we do not find any significant differences in white matter between the Controls and the KA groups. To be certain we looked at all three lesion epicenters where the damage occurred. If you examine the graphs below, you will notice that in fact the KA animals have a higher % white matter of the CSA than the Controls. Given how this analysis is done we are looking at % white matter of the cross sectional area (CSA). In the KA animals the loss of gray matter causes a collapse that makes it appear as though the white matter covers more of the CSA area than it normally does. Even if we were to normalize to the Controls you would see the same as what you already observe in these graphs.

For this reason, we have compared the average area of white matter at the three lesion epicenters between the Control and KA groups and did not find significant differences (new Figure 7C). We also evaluated average area of white matter at the individual spinal levels (L2-L4) and did not find significant differences between the two groups and therefore averaged them. This indicates that we are not seeing any white matter alterations with our lesion model.

Results: Within the results (and elsewhere) there are a number of un-supported statements that should be removed, softened or supported. For example, on page 18 the authors talk about how the CatWalk "further investigates the role of propriospinal INs connecting the cervical and lumbar enlargements" and no reference is provided.

• The requested references have now been added.

It is important to note that two animals were not included overall because they were unable to perform the CatWalk assessment. Additional information about these animals might be helpful to further characterize the KA lesions, for example, when they are too large.

• Yes, we have looked into this. Lesion size appears to play a role (Figure 2C) but does not appear to be the only determining factor as two animals (KA#6, KA#7) with and without weight support had the same lesion length (10,325um). We predict this is due to the amount of neuronal loss; KA#7 had greater neuronal loss in all three levels compared to KA#6.

Figure 6 brings up a number of questions including how the three "epicenters" were determined and how some KA lesioned spinal cords appear to have more than 100% the number of neurons in the control spinal cords. Yes, there is variability in normal animals, but still this seems unlikely. Is it possible that the KA injection sites were not accurate in these animals? I know it is unlikely, however, the large number of neurons in some animals at L2 is bothersome. Did the investigators always inject L2, L3 and L4 in that order? Pipettes tend to wick up liquid thus diluting the drug/cells/whatever at the tip.

• We understand your concern of being greater than 100%, therefore we have changed the normalization to the greatest control value vs the average of controls (except for lesion size which is done to largest lesion size overall, new Supplemental Figure 3 and Figures 6 and 7 will be altered once our new analysis is finished).

• Animal KA #1 that you are referring to could have been a technical error due to injections but it is hard to say at this point as we have found nothing from our surgery records that indicate why this animal would be different from others. Yes, bilateral injections were always performed in the same order (L2, L3, L4). However, we think it is unlikely that this created a significant drug dilution problem as we see animals with more damage in L4 than L3 or L2 (KA #3, #6 and #7 in new Supplemental Figure 3). But clearly L2 in animal KA 1 is not significantly damaged.

Also for Figure 6, I am not convinced that the color coding is really very useful here. I think what might be more useful would be some higher magnification images of the intermediate gray matter. This figure also appears to show pipette tracks in some sections suggesting that the KA was leaking up the track either during injection or when the pipette was withdrawn. This is not a serious issue, but might be worth mentioning as a confound.

• First, we would like to clarify that Figure 6 is already a higher magnification image of only laminae V-VII not the entire gray matter (please see figure legend). Figure 1 is a lower magnification but here in Figure 6 we wanted to highlight the region of interest that was analyzed for neuronal loss. Pipette tracks were also observed in the Controls and not thought to be due to KA leakage, as we don’t see neuronal damage beyond the injection tracts in the dorsal horns. With the new figure we will see if the color coding will be added or not dependent on the space available.

Finally, for Figure 6, the correlations shown are quite poor, and would be even worse of control animals were not included. Too much strength is given to these findings.

These issues with Figure 6 become even more serious as we move to Figure 7. Here, looking at the correlation to loss of MNs is weak because this reviewer is not convinced that looking at the "three epicenters" is a valid approach. Were the epicenters identified by particular criteria? Also, I think images showing how MNs were identified and counted would be important, in particular since you did not use ChAT staining but relied on NeuN and size.

• These epicenters were chosen after reviewing all coronal sections in a 1:7 series of the lumbar cord (T12-L5). The three epicenters were the three coronal slices with the greatest neuronal loss (methods, page 12). This is supported by the inflammatory response in these sections (not shown).

• Please see the schematic below that explains the motoneuron analysis that is also performed in our work which is detailed in the methods. Briefly, the cell soma area of NeuN+ cells in lamina IX were measured in Image J. NeuN+ cells with an area greater than 916µm2 were used for the motoneuron analysis (Wen et al, 2015).

• We agree that in Figure 6 the correlations for each spinal level although significant are moderate but this is due to the fact that one given spinal level was not found to be responsible for the behavioral deficits. This is supported by our work on correlation with lesion length, the lesion must span multiple levels to produce the behavioral deficit. Finally, the correlations may change when we add in lamina VIII, but we won’t know until the analysis is finished.

• As for Figure 7, we agree that we do not see correlations and our argument is that motoneuron and white matter area are not responsible for the behavioral deficits we observe (new Figure 7). Therefore, you are reading those correctly, these are not significant correlations.

Discussion Yes, interneurons in the intermediate gray matter throughout the lumbar enlargement "regulate lower motoneurons" but they also do other things, most notably communicating both intra and intersegmentally (short and long propriospinals). Please adjust this statement.

• We appreciate this detailed feedback, we have adjusted this statement to the following:

“Damage to this area, which includes regulation of lower motoneurons leads not only to gross motor deficits (BBB score), but rhythmic and skilled walking (even and uneven horizontal ladders), coordination (BBB subscore), balance (inclined beam) and gait deficits (CatWalk), as well.” (page 25)

On page 25, you talk again about premotor SpINs. I understand that you are using this term/nomenclature to distinguish these INs from motoneurons, but this is problematic because many if not most of your readers will assume the premotor SpINs synapse directly onto MNs, which of course many of the INs that are eliminated by KA do not. Calling them simply SpINs would be sufficient and still distinguish them from MNs.

• We have adjusted this to the term “SpIN and premotor circuitry” on pages 26 and 27.

On page 27 you talk about the RI, and while there is a statistically significant drop in RI, it must be admitted that the RI remains above 90% (0.9) which means that 9 out of 10 steps use a normal sequence. Thus, I think it is misleading to indicate that this indicates a difference for the KA animals. In fact, I think it is more important to consider how these animals were able to maintain an RI in excess of 90% despite the loss of substantial numbers of INs.

• Thank you for the comment, we have adjusted this in the discussion:

“In addition to gait rhythm changes, we also saw significant differences in pattern generation. The regularity index (RI) measures correctly sequenced footsteps and is used to analyze recovery in mild to moderate injuries and coordination (Koopmans et al., 2005; Kuerzi et al., 2010; Shepard et al., 2021). While KA-animals have a significantly lower RI in comparison to the controls, the RI remains above 90% which is still relatively high given the amount of neuronal loss. However, we would argue that a single parameter is not the defining factor of gait/coordination, but a combination of parameters and tests provides a more comprehensive picture, as we have seen with our pLDA analysis and Random Forest classification approaches.” (Pages 28-29)

The rationale for determining classification prior to histological analysis is somewhat weak, and I think it would be worthwhile strengthening this rationale at the beginning of this paragraph...it becomes more obvious later why this classification is important. Is the variability of the KA model greater than an NYU or IH contusion model? If so, why? The early functional plateau is key to this argument.

• We postulate that less severe SCIs and our milder KA lesion tend to have more variability than more severe SCI models. In the contusion models this is due to the delayed natural compensatory functional recovery plateau that can last up to 5-6 weeks. However with the KA model, variability arises from titrating down KA and adding multiple injection sites increasing variable success rate per injection. In the KA model, the early functional plateau at two weeks allows for correctly excluding or classifying animals into equally lesioned groups prior to treatment with our Random Forest Eco model. We agree that we need to clarify this reasoning in the results and have now done so on page 22. “To test the efficacy of experimental SCI therapies, it is important to effectively evaluate recovery performance through the combination of behavioral tests. In addition to carefully classifying groups at the end of the study, there is a need to provide exclusion criteria and equal sorting of variability between groups prior to treatment (after deficits have stabilized at two weeks).” (page 22)

Minor Comments:

__ Heatmap Analysis: The term "lesion size" is insufficiently accurate to be used in this context. Do you mean lesion length?__

• This term has now been adjusted to lesion length throughout the manuscript and figures.

Kainic Acid injuries are known to be accompanied by cell division and neurogenesis in the brain, and if that kind of thing is happening in the presented model, it could be an interesting confound/addition to the alluded to cellular replacement __therapies.____

__

• KA has been shown to be accompanied by cell division and neurogenesis in the brain, however from our own work and previous work with KA in the spinal cord if this occurs it is not at a level that is relevant to functional recovery as evidenced in our long-term study. A previous study by Magnuson et al compared E14 cerebral rat precursor cell transplantation 40 minutes and 4 weeks post-KA injury and did not find significant differences in cell survival/division (Magnuson et al., 2001). Therefore, we do not believe this would hamper or confound our future work with cellular replacement therapies. In addition, cell transplantation would take place 2 weeks post-KA injury when KA would no longer be able to hamper the transplanted cells.

__Reviewer #2 (Significance (Required)):

__

__ Overall, this is a well-designed and performed set of studies that takes the KA lesion model into new territory, well set-up to perform delayed (sub-acute or early chronic) neuron replacement studies. The work characterizes a multi-segment but mild KA injury model that demonstrates persistent dysfunction that plateaus early, and a rapid and efficient system to classify the injury with a high predictability of long-term dysfunction by 2 weeks post-injury.

This model should be of interest because it focuses on gray-matter specific tissue loss and functional deficits that should be amenable to neuron replacement strategies without the complications of white-matter dependent functional losses.

My expertise: I have been using a variety of spinal cord injury models, in rats, for many years including contusions, lacerations and excitotoxic (KA) lesions. I have a lot of experience with locomotor, motor and sensory outcome measures. However, I have very limited experience with the Random Forest analysis employed and am not an expert in statistics.__

__References: __

Hadi, B., Zhang, Y. P., Burke, D. A., Shields, C. B., & Magnuson, D. S. (2000). Lasting paraplegia caused by loss of lumbar spinal cord interneurons in rats: no direct correlation with motor neuron loss. J Neurosurg, 93(2 Suppl), 266-275. https://doi.org/10.3171/spi.2000.93.2.0266

Koopmans, G. C., Deumens, R., Honig, W. M., Hamers, F. P., Steinbusch, H. W., & Joosten, E. A. (2005). The assessment of locomotor function in spinal cord injured rats: the importance of objective analysis of coordination. J Neurotrauma, 22(2), 214-225. https://doi.org/10.1089/neu.2005.22.214

Kuerzi, J., Brown, E. H., Shum-Siu, A., Siu, A., Burke, D., Morehouse, J., Smith, R. R., & Magnuson, D. S. (2010). Task-specificity vs. ceiling effect: step-training in shallow water after spinal cord injury. Exp Neurol, 224(1), 178-187. https://doi.org/10.1016/j.expneurol.2010.03.008

Mohan, R., Tosolini, A. P., & Morris, R. (2015). Segmental Distribution of the Motor Neuron Columns That Supply the Rat Hindlimb: A Muscle/Motor Neuron Tract-Tracing Analysis Targeting the Motor End Plates. Neuroscience, 307, 98-108. https://doi.org/10.1016/j.neuroscience.2015.08.030

Nicolopoulos-Stournaras, S., & Iles, J. F. (1983). Motor neuron columns in the lumbar spinal cord of the rat. J Comp Neurol, 217(1), 75-85. https://doi.org/10.1002/cne.902170107

Pitzer, C., Kurpiers, B., & Eltokhi, A. (2021). Gait performance of adolescent mice assessed by the CatWalk XT depends on age, strain and sex and correlates with speed and body weight. Sci Rep, 11(1), 21372. https://doi.org/10.1038/s41598-021-00625-8

Shepard, C. T., Pocratsky, A. M., Brown, B. L., Van Rijswijck, M. A., Zalla, R. M., Burke, D. A., Morehouse, J. R., Riegler, A. S., Whittemore, S. R., & Magnuson, D. S. (2021). Silencing long ascending propriospinal neurons after spinal cord injury improves hindlimb stepping in the adult rat. Elife, 10. https://doi.org/10.7554/eLife.70058

Wen, J., Sun, D., Tan, J., & Young, W. (2015). A consistent, quantifiable, and graded rat lumbosacral spinal cord injury model. J Neurotrauma, 32(12), 875-892. https://doi.org/10.1089/neu.2013.3321

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

This manuscript reports on a pair of well-designed and well-carried out studies investigating a Kainic Acid (KA)-mediated gray matter lesion in the lumbar enlargement of adult female SD rats. The investigators demonstrate, using NeuN immunohistochemistry, that the KA lesion reduces NeuN positive cells along the length of the lumbar spinal cord from rostral to L2 to slightly caudal to L4 following 6 separate injections made on the right and left sides of the spinal cord at L2, L3 and L4. The investigators made significant efforts to avoid depleting neurons in the dorsal and ventral horns, and the evidence provided suggests they were successful. The methodology described is sound and sufficient details are provided to allow the reader to fully understand the studies. It is outstanding that the study was done while following all of the PREPARE and ARRIVE guidelines. A second major component of the work is the use of multiple outcome measures and efforts (using a Forest analysis) to develop a relatively quick, accurate and efficient system to screen or classify the injuries in individual animals within 2 weeks of the injury so that subsequent treatments could be done on animals which received injuries of sufficient severity (within a relatively narrow range) and with balanced experimental and groups. Again, with this effort the investigators were largely successful. The KA lesion results in persistent locomotor and sensorimotor deficits, that plateau early without substantial sensory dysfunction.

Major Comments:

Introduction: Overall, the rationale presented and the review of the pertinent literature is solid, with the following exception: The authors state that their model should allow them to thoroughly investigate the behavioral readout of premotor IN loss. It is generally accepted that the designation of premotor interneurons refer to those directly connected to motor neurons, and while the chosen KA lesion certainly targets some premotor neurons, it also targets many other interneurons that do not directly contact motoneurons. Please revise how the lesions are referred to. In the very next paragraph the targets are defined somewhat differently as "INs and propriospinal INs in laminae V-VII in spinal levels L2-L4".

Spared white matter. In many (but not all) labs, spared white matter at the epicenter is an important measurement because it presumably represents all the spared axons, such that any/all rostrocaudal communication is represented. Thus, it is the single point (or section, in this case) that has the smallest number of axons represented as stained white matter. So, to indicate that you assessed "three epicenters per spinal cord" doesn't make sense in this context, Even if you are referring to three separate KA injection sites (L2, L3 and L4). Thus, averaging three sections also doesn't really make sense because the actual epicenter should be represented by the single cross section that has the smallest area of stained white matter. Also related to spared white matter, in many labs they calculate %SWM based on a section from a control animal, and this should reduce variability because some cords shrink (injured gray matter) more than others after the injury, whether it be a contusion or mild excitotoxic injury. Please either re-calculate your SWM or provide additional justification for your current method.

Results: Within the results (and elsewhere) there are a number of un-supported statements that should be removed, softened or supported. For example, on page 18 the authors talk about how the CatWalk "further investigates the role of propriospinal INs connecting the cervical and lumbar enlargements" and no reference is provided.

It is important to note that two animals were not included overall because they were unable to perform the CatWalk assessment. Additional information about these animals might be helpful to further characterize the KA lesions, for example, when they are too large.

Figure 6 brings up a number of questions including how the three "epicenters" were determined and how some KA lesioned spinal cords appear to have more than 100% the number of neurons in the control spinal cords. Yes, there is variability in normal animals, but still this seems unlikely. Is it possible that the KA injection sites were not accurate in these animals? I know it is unlikely, however, the large number of neurons in some animals at L2 is bothersome. Did the investigators always inject L2, L3 and L4 in that order? Pipettes tend to wick up liquid thus diluting the drug/cells/whatever at the tip.

Also for Figure 6, I am not convinced that the color coding is really very useful here. I think what might be more useful would be some higher magnification images of the intermediate gray matter. This figure also appears to show pipette tracks in some sections suggesting that the KA was leaking up the track either during injection or when the pipette was withdrawn. This is not a serious issue, but might be worth mentioning as a confound.

Finally, for Figure 6, the correlations shown are quite poor, and would be even worse of control animals were not included. Too much strength is given to these findings.

These issues with Figure 6 become even more serious as we move to Figure 7. Here, looking at the correlation to loss of MNs is weak because this reviewer is not convinced that looking at the "three epicenters" is a valid approach. Were the epicenters identified by particular criteria? Also, I think images showing how MNs were identified and counted would be important, in particular since you did not use ChAT staining but relied on NeuN and size.

Discussion Yes, interneurons in the intermediate gray matter throughout the lumbar enlargement "regulate lower motoneurons" but they also do other things, most notably communicating both intra and intersegmentally (short and long propriospinals). Please adjust this statement.

On page 25, you talk again about premotor SpINs. I understand that you are using this term/nomenclature to distinguish these INs from motoneurons, but this is problematic because many if not most of your readers will assume the premotor SpINs synapse directly onto MNs, which of course many of the INs that are eliminated by KA do not. Calling them simply SpINs would be sufficient and still distinguish them from MNs.

On page 27 you talk about the RI, and while there is a statistically significant drop in RI, it must be admitted that the RI remains above 90% (0.9) which means that 9 out of 10 steps use a normal sequence. Thus, I think it is misleading to indicate that this indicates a difference for the KA animals. In fact, I think it is more important to consider how these animals were able to maintain an RI in excess of 90% despite the loss of substantial numbers of INs.

The rationale for determining classification prior to histological analysis is somewhat weak, and I think it would be worthwhile strengthening this rationale at the beginning of this paragraph...it becomes more obvious later why this classification is important. Is the variability of the KA model greater than an NYU or IH contusion model? If so, why? The early functional plateau is key to this argument.

Minor Comments:

Heatmap Analysis: The term "lesion size" is insufficiently accurate to be used in this context. Do you mean lesion length?

Kainic Acid injuries are known to be accompanied by cell division and neurogenesis in the brain, and if that kind of thing is happening in the presented model, it could be an interesting confound/addition to the alluded to cellular replacement therapies.

Significance

Overall, this is a well-designed and performed set of studies that takes the KA lesion model into new territory, well set-up to perform delayed (sub-acute or early chronic) neuron replacement studies. The work characterizes a multi-segment but mild KA injury model that demonstrates persistent dysfunction that plateaus early, and a rapid and efficient system to classify the injury with a high predictability of long-term dysfunction by 2 weeks post-injury.

This model should be of interest because it focuses on gray-matter specific tissue loss and functional deficits that should be amenable to neuron replacement strategies without the complications of white-matter dependent functional losses.

My expertise: I have been using a variety of spinal cord injury models, in rats, for many years including contusions, lacerations and excitotoxic (KA) lesions. I have a lot of experience with locomotor, motor and sensory outcome measures. However, I have very limited experience with the Random Forest analysis employed and am not an expert in statistics.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In this paper, Kuehn and colleagues report on the analysis of functional impairments following intermediate gray matter lesion with kainic acid. The image convincingly show that mostly purely grey matter lesion can be achieved throughout the paper. The authors took care to do a battery of well-designed behavioral tests and sophisticated analysis in order to access functional impairment. They then correlate their behavioral assessment to lesion size, the number of NeuN positive cells in layers V-VII epicenters as well motoneuron numbers and the percentage of white matter. Overall, the manuscript is well written, nicely framed in the existing literature, very clear and the experiments are simple but well designed. The behavioral testing and evaluations including random forest ranking are well performed. The methodology is complete and would allow reproducing the experiments. Statistics are used appropriately. We have however some reserves and comments on some of the results and interpretations. Addressing these comments would not involve new experiments but new re-analysis of the existing datasets.

Major comments:

While the claims that grey matter lesions trigger major behavioral impairments is convincing in particular with the refine behavioral experiments performed, the key claim that only interneuron loss in layer V-VII mediates those deficits is currently not supported by the presented data. In particular, we would suggest that the lesions performed, in contrast to the claims, are not purely and selectively impacting layer V-VII but might also impact layers VIII-IX. We think that presenting neuronal counts based on NeuN staining separately for layer I-IV, V-VII, VIII-IX and comparing control vs KA is necessary. Only with these data can conclusions be supported either in the direction suggested by the authors or otherwise.. Another claim relative to the lack of involvement of motoneurons in the related behavioral deficits is also difficult to resolve with the current data. Motoneurons have been identified based on NeuN staining and size. While this is not the state of the art (ChAT staining would have been preferable), it remains acceptable. However, the data presented figures 7 and 8 show a very wide range in the motoneuron count (15 to 50) indicating either motoneuron loss or a count performed at different lumbar levels in the animals. This raises questions on the model (is it really involving only layers V-VII?) or on the interpretation of the data. Therefore we believe that motoneurons counts need to be presented separately (see above) in control vs KA groups and data need to be discussed in this perspective. Authors should also tone down the specificity of the model and involvement of motoneurons accordingly (page 20 for example). Most of the conclusions rely on correlations that include control animals (injected with saline hence with no lesions and no behavioral deficits; Fig 6 and 7). This artificially skews the correlations as those animals show no lesions and good performance in the behavioral tests. These correlations need to be performed only with KA injected animals to determine the respective involvements of interneurons and motoneurons.<br /> The long-term study (Fig 8) is performed with very few animals and hence, drawing conclusions from these animal numbers is difficult. All correlations are performed including control animals which is even more of a problem here as in Figure 6 and 7 due to the low number of animals. The authors should either add animals or remove the figure. When control animals (injected with saline) are removed (as they do not show any lesion and perform accurately in the behavior), one would actually see a correlation between the number of motoneurons and the behavioral performance (Fig. 8E,F) but not with the lesion size (Fig.8C,D).

Minor comments:

Figure 1A: if lesions are bilateral, it would be nice to illustrate this on the schematic.

Figure 1B-D: scale bars are missing

Figure 3H: What represents the y-axis? % of completion or number of completion?

Figure 4 Table: Please specific what the acronym stands for: pLDA.

Figure 6 A: scale bars are missing

Figure 6B/C/D: please add the spinal level analyzed directly on the graphs. This will ease the comprehension.

Figure 7 and Figure 8: While it is quite convincing that the model is purely a grey matter injury (panel C and D), the data are very much spread out for the number of motoneurons per mice (see major comments above). We would suggest to plot those data to present the number of neurons (interneurons in layer I-IV, V-VII and motoneurons) control vs KA.

Dots are missing on those figures (probably superimposed on top of each other). This should be changed to see all data points

Figure 8E,F: the number of motoneurons is very low also in controls. How is this explained?

Significance

This paper by Kuehn and colleagues reports on the functional impairments that follow intermediate gray matter lesions using kainic acid. This work is largely confirmatory of previous studies (Magnuson et al., 1999; Hadi et al., 2000) with modern behavioral evaluation. After revision, it would provide a description of the functional impairments following those specific lesions. The paper would be informative for a specific audience in particular scientists in the field of spinal cord injury and spinal interneuron.

Our field of expertise is spinal cord injury, inflammation, behavior and axon outgrowth.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

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

*Randhawa and co-authors have studied various aspects of the regulation of lignocellulose degradation by the filamentous ascomycete fungus Penicillium funiculosum. Over-expression of the well-known transcription factor clr2 (which regulates cellulase gene expression in Neurospora and other ascomycetes) in a delta-mig1 strain did not result in an increase in cellulase activity. However, when combined with an increased Ca2+ concentration the cellulase activity in the medium did increase. Using RNA-Seq, the authors have identified a candidate regulator: Snf1. Indeed, a knockout confirms that this gene is involved in the posttranscriptional regulation of cellulase production, specifically by regulating the secretion of the cellulases. *

Major comments:

In general, the topic and results are interesting. There are a few issues that need to be addressed, however. The manuscript would benefit from some careful proofreading. For example, articles ('the', 'a') are frequently missing. Very informal language is sometimes used ('zilch effect'). Put a space between '1000bp', etc. It is 'kDa', not 'kD', etc.

Response – Thank you very much for the encouraging remarks. We have thoroughly checked the manuscript and have added the articles at appropriate places. We have also improved the manuscript’s language and removed any informal language used.

I am a bit puzzled by the choice of calcium source: CaCO3, up to 10 g/L. Calcium carbonate does not efficiently dissolve in water unless the pH is low. Fungi generally acidify their culture medium during growth. As such, calcium carbonate likely has a pH buffering effect. Therefore, the described effects may also be attributed to a more neutral pH of the medium, and not necessarily to an increase in calcium ions.

Response – We completely agree with the reviewer and had the same thought that the pH buffering effect of CaCO3 could be the reason for increased cellulase production. We ruled out this by using 50 mg/l CaCl2 solely in rest of the experiments performed in Fig. 3 and afterwards. We have also mentioned the same in the manuscript (lines 175-178).

The authors have performed RNA-Seq, but as far as I can tell the data has not been made publicly available. At least, the raw reads should be deposited in the Short Read Archive of NCBI (or a similar repository), and preferably also the expression values in GEO of NCBI (or a similar repository).

Response – We will comply and deposit the raw reads in the short read archive of NCBI. We will also be providing the differential analysis of transcription factors expressed under glucose and Avicel in NCIM1228 and ∆Mig1 in the supplementary information.

P21. Very little information is provided in the M&M regarding the gene expression analysis. Provide references to all the tools, as well as the version numbers. Were any non-default parameters used?

Response – We have added the complete information on tools and procedures used for RNA-seq data analysis. For differential expression profiling, all FPKM values were normalized to the library size using the R package, Edge R. The expression value for the transcript was calculated using the reads aligned & normalised it on library size (Total sequencing reads generated) & transcript length giving us FPKM value (Fragments Per Kilobase of transcript per Million mapped reads), and TPM value (Transcript per million reads), which is regarded as normalized expression value for a particular transcript. We have taken the number of reads which got aligned to the conserved transcripts (Present in both the comparison group i.e Wild Type Glu & Cellulose samples (S1, S2, S7 & S8) Vs MIG1 glu & Cellulose sample (S3, S4, S5 & S6) and performed the differential gene expression between the two groups. The excel sheet having differential expression profiling of transcription factors is available as supplementary data.

The authors claim that SSP1 CaMKK phosphorylates SNF1 AMPK (last title of the Results section). I don't see any evidence for a direct interaction between these two proteins. I will believe that they are in the same pathway, but if the authors want to claim a direct interaction then additional experiments will be required. E.g. Y2H.

Response – Ssp1 is known to phosphorylate SNF1 during nutritional stress in S. pombe and they were found to interact directly by Co-IP studies. Based on the literature, we planned to over-express Ssp1 in P. funiculosum.

Minor comments:

• Please add line numbers to the manuscript, this facilitates the review process.*

Response - Line numbers have been added.

*P14 "in all yeasts and filamentous fungi". I doubt that all fungi have been tested. *

Response - The phrase has been modified.

P18. "in diverse yeasts and fungi". Yeasts are also fungi.

Response - The phrase has been modified.

P16. "solves dual purpose". I think this is meant: "serves a dual purpose"?

Response - The phrase has been modified.

*P17, first paragraph: this seems very speculative to me, so it should probably be labeled as such. *

Response - The phrase has been modified.

P21. What reference genome is used? Please cite the paper.

Response - We have our own reference genome in lab which is yet to be published.

Fig 1B. These are reported as volcano plots, but to me it looks like an empty graph (no data points), only a number of genes.

Response - The pictures have been changed.

Fig 1D. What do the colors on the right represent? The colors on the right represents k-means clustering of the genes of transcription factors.

Response - The same has been added to the figure legend also.

On various places in the manuscript the term "three times in triplicate" is used. What is meant here, three technical replicates of each of the three biological replicates?

Response - Yes we mean the same and the phrase has been modified.

P46. "We aimed to sought"

Response - The phrase has been modified.

Abstract: The sentence "Further, Ca2+-signaling" should be rewritten, because currently is seems to suggest that SSP1 downregulates the phospho-HOG1 levels.

Response – As suggested by the Western blot in the Fig.4b, Snf1 gets phosphorylated only when dual signal of calcium and cellulose are present. Since we observed upregulated Ssp1 expression in Avicel (Fig. 4a), and increased Ssp1 expression could increase the phosphorylated Snf1 in the cell (Fig. 7i), our data suggests that Ssp1 phosphorylates Snf1 in a Ca2+-dependent manner. Further Hog1 was found in hyperphosphorylated state in ∆Snf1 (Fig 6e), thus we believe Snf1 AMPK downregulates phospho-Hog1 levels.

Reviewer #1 (Significance (Required)):

*In general, the topic and results are interesting. There are a few issues that need to be addressed, however. The manuscript would benefit from some careful proofreading. *

Response – We highly appreciate the encouraging words of the reviewer. We have addressed all the issues raised by the reviewer. The major ones included the language and readability of the text, which has been improvised. We have replaced the volcano plot figures, and will be uploading the RNA-seq data to the SRA database of NCBI and excel sheet of differential expression analysis of transcription factor will be added as a supplementary file to the manuscript.

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

• Randhawa et al. study the effect of loss of function of Snf1 kinase and calcium on the production of enzymes related to cellulose degradation in the fungus Penicillium funiculosum. *
• The manuscript is well structured and the researchers have done an enormous amount of work in constructing a number of mutant strains in this fungus. *
• Transcriptomics and proteomics support the conclusions reached with the strains generated.*

Response - Thank you very much for showing confidence in our research work, we are highly obliged by positive remarks on the manuscript.

The manuscript is long and suffers from an excess of results presented in figures. My main criticism focuses on the presentation of data on the cellular distribution of the ER and Golgi apparatus. The micrographs are inconclusive and it is not really clear what the authors are trying to show in these experiments. These results are not really necessary for the article and I suggest that they be removed from the article.

Response – We agree with the reviewers comments on data on the cellular distribution of the ER and Golgi apparatus. We have removed the micrograph data on the cellular distribution of the ER and Golgi apparatus (earlier Figure 3j and Figure 4r).

Reviewer #2 (Significance (Required)):

The authors have done an excellent job in producing a large number of strains carrying null alleles. In addition, they have used two broad analysis techniques that allow them to establish coherent hypotheses and corroborate them with the results.

Response – Thank you very much for the positive comments

The manuscript is difficult to understand in some sections because of the excessive amount of data and panels in the figures. The names designating each strain and given in full length in the graphs do not help either.

Response – Thank you very much for the valuable suggestion. We have reduced the number of graphs by including all enzymes assays in one concise graph in Figure 4. We have also shortened the names of strains and enzymes, in all the figures.

This work is of interest to all researchers interested in the integrity of signaling and regulatory pathways on extracellular enzymes of biotechnological interest.

*My interests focus on the cell biology of filamentous fungi, in particular on the molecular mechanisms and subcellular localization of elements involved in intracellular transport, signaling against environmental stresses and changes in transcriptional regulatory patterns. *

Response – Thank you once again for the encouraging remarks.

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

The manuscript submitted by Randhawa et al focus on the mechanism of cellulases secretion, the very important and basal question in the filamentous fungi, particularly for cellulases biotechnology. As the author said the molecular basis of cellulases production previously study mainly focuses on regulation mechanism at transcription level, the study of molecular mechanism of cellulases translation and secretion are much rare. Therefore the submitted work is very impressive me on the progress of this area. What they presented shown the Ca2+ is critical for the regulation of cellulases secretion by SNF-1, SSP1 and HOG1. The regulation might be caused by affecting the protein trafficking in ER and Golgi, the manuscript found the development of ER and Golgi shown changes by staining by ER-tracker and Bodipy under different conditions and mutants. The manuscript constructed a model about regulatory mechanism of Ca2+ on cellulases translation and secretion level. The present study is close to make significant progress in the cellulases regulation area.

Response – We appreciate the positive comments of the reviewer.

Major comments: I am really impressive for the great work in the manuscript, however, I think the more work do need for give the conclusion of paper.

1.In terms of dynamics development of ER and Golgi of strains, the very critical data for the conclusion of the paper, the current data is only by chemical staining. It is not robust, it will be needed by other methods, for example, GFP-labeling the marker of ER or Golgi.

Response – The manuscript focuses on the signaling events governing cellulase production, and secretion. Since ER and Golgi are the sites of protein production and secretion, we hypothesized, if the Ca2+ signaling affects post-transcriptional events, it must have had some impact on the dynamics of these organelles; and microscopy experiments suggested us the same. In the next set of experiments, we proved our hypothesis with the proteomics and functional analysis of Snf1, Ssp1, and Hog1 MAPK. Hog1 MAPK pathway is known to regulate protein trafficking and secretion in yeast. We here showed that Ca2+- dependent regulation of Hog1 MAPK and its downregulation by Snf1 AMPK is crucial to cellulase secretion.

2.Also the author try to suggest the cellulases were detained in the ER, not went into Golgi, therefore the secretome protein decreased. It is very much possibly but the evidence is not robust either, to trafficking the GFP-labelled CBH1 might be a good experiment to make it clear.

Response – Thank you very much for raising the query. The manuscript majorly focuses on the role of calcium signaling on cellulase translation and secretion. Further, we have studied two signaling proteins, Snf1 AMPK and Hog1 MAPK which are downstream to calcium signaling, and we found their crosstalk vital to cellulase secretion. We have not talked about cellulases being detained in the ER or Golgi, rather we focused on the signaling events regulating cellulase production and transport.

Since we had already ruled out the role of calcium in cellulase transcriptional activation, and ER and Golgi being major site of protein production in the cell; we performed microscopy experiments to see if the calcium signaling modifies ER and Golgi morphology during carbon stress. We found under-developed Golgi in the absence of calcium in wild type. This experiment helped us to build a hypothesis that calcium signaling might have role in downstream events like protein translation, and secretion. The hypothesis was proved by functional analysis of signaling proteins, Western blot and proteomics experiments. Further, microscopy experiments further strengthened our observation that Snf1 AMPK is downstream target of calcium signaling and has no role in the cellulase translation, but cellulase secretion.

Considering that we are not focusing on the protein trafficking of cellulase, the confocal microscopy experiments are not decisive, rather build supporting evidence for our hypothesis, as suggested by the second reviewer. We have proved our hypothesis of Ca2+-dependent post-transcriptional regulation of cellulase by proteomics, and other biochemical experiments. Nevertheless, we plan to perform the confocal experiments again to achieve pictures with higher resolution.

1.On page 9, please indicate the fold changes of the kinases genes talked about, snf1 and so on.

Response – We have added the Fold change in the expression of Snf1 and Ssp1 (line number 221).

2.The quality of microscopic figure is not good, should have one with higher resolution, even consider to present the electron microscope picture to give the er and Golgi dynamics changes the manuscript talked about(optional).

Response: We agree with the reviewer’s suggestion to add high resolution confocal images of mycelia in Fig 3j and Fig. 4o. We are in the process of repeating the confocal microscopy experiment. We will update the manuscript with improved microscopic pictures.

*3. The quality of Western plot need to be improved, particularly figure 4f，figure 7i， it is hard to give the conclusion based on the picture presented *

Response – We have replaced the pictures of western blots (Fig 4f, and Fig 7i) with high resolution images.

Reviewer #3 (Significance (Required)):

The manuscript submitted by Randhawa et al focus on the mechanism of cellulases secretion, the very important and basal question in the filamentous fungi, particularly for cellulases biotechnology. As the author said the molecular basis of cellulases production previously study mainly focuses on regulation mechanism at transcription level, the study of molecular mechanism of cellulases translation and secretion are much rare. Therefore the submitted work is very impressive me on the progress of this area. What they presented shown the Ca2+ is critical for the regulation of cellulases secretion by SNF-1, SSP1 and HOG1. The regulation might caused by affecting the protein trafficking in ER and Golgi, the manuscript found the development of ER and Golgi shown changes by staining by ER-tracker and Bodipy under different conditions and mutants. The manuscript constructed a model about regulatory mechanism of Ca2+ on cellulases translation and secretion level. The present study is close to make significant progress in the cellulases regulation area.

Response - Thank you for the positive comments on the manuscript.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

The manuscript submitted by Randhawa et al focus on the mechanism of cellulases secretion, the very important and basal question in the filamentous fungi, particularly for cellulases biotechnology. As the author said the molecular basis of cellulases production previously study mainly focuses on regulation mechanism at transcription level, the study of molecular mechanism of cellulases translation and secretion are much rare. Therefore the submitted work is very impressive me on the progress of this area. What they presented shown the Ca2+ is critical for the regulation of cellulases secretion by SNF-1, SSP1 and HOG1. The regulation might caused by affecting the protein trafficking in ER and Golgi, the manuscript found the development of ER and Golgi shown changes by staining by ER-tracker and Bodipy under different conditions and mutants. The manuscript constructed a model about regulatory mechanism of Ca2+ on cellulases translation and secretion level. The present study is close to make significant progress in the cellulases regulation area.

Major comments

I am really impressive for the great work in the manuscript, however, I think the more work do need for give the conclusion of paper.

1.In terms of dynamics development of ER and Golgi of strains, the very critical data for the conclusion of the paper, the current data is only by chemical staining. It is not robust, it will be needed by other methods, for example, GFP-labeling the marker of ER or Golgi 2.Also the author try to suggest the cellulases were detained in the ER, not went into Golgi，therefore the secretome protein decreased. It is very much possibly but the evidence is not robust either, to trafficking the GFP-labelled CBH1 might be a good experiment to make it clear

Minors

1.On page 9, please indicate the fold changes of the kinases genes talked about, snf1 and so on. 2.The quality of microscopic figure is not good, should have one with higher resolution, even consider to present the electronmicroscope picture to give the er and Golgi dynamics changes the manuscript talked about(optional) . 3.The quality of western plot need to be improved, particularly figure 4f，figure 7i， it is hard to give the conclusion based on the picture presented

Significance

The manuscript submitted by Randhawa et al focus on the mechanism of cellulases secretion, the very important and basal question in the filamentous fungi, particularly for cellulases biotechnology. As the author said the molecular basis of cellulases production previously study mainly focuses on regulation mechanism at transcription level, the study of molecular mechanism of cellulases translation and secretion are much rare. Therefore the submitted work is very impressive me on the progress of this area. What they presented shown the Ca2+ is critical for the regulation of cellulases secretion by SNF-1, SSP1 and HOG1. The regulation might caused by affecting the protein trafficking in ER and Golgi, the manuscript found the development of ER and Golgi shown changes by staining by ER-tracker and Bodipy under different conditions and mutants. The manuscript constructed a model about regulatory mechanism of Ca2+ on cellulases translation and secretion level. The present study is close to make significant progress in the cellulases regulation area.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Randhawa et al. study the effect of loss of function of snf1 kinase and calcium on the production of enzymes related to cellulose degradation in the fungus Penicillium funiculosum. The manuscript is well structured and the researchers have done an enormous amount of work in constructing a number of mutant strains in this fungus. Transcriptomics and proteomics support the conclusions reached with the strains generated. The manuscript is long and suffers from an excess of results presented in figures. My main criticism focuses on the presentation of data on the cellular distribution of the ER and Golgi apparatus. The micrographs are inconclusive and it is not really clear what the authors are trying to show in these experiments. These results are not really necessary for the article and I suggest that they be removed from the article.

Significance

The authors have done an excellent job in producing a large number of strains carrying null alleles. In addition, they have used two broad analysis techniques that allow them to establish coherent hypotheses and corroborate them with the results. The manuscript is difficult to understand in some sections because of the excessive amount of data and panels in the figures. The names designating each strain and given in full length in the graphs do not help either. This work is of interest to all researchers interested in the integrity of signalling and regulatory pathways on extracellular enzymes of biotechnological interest.

My interests focus on the cell biology of filamentous fungi, in particular on the molecular mechanisms and subcellular localisation of elements involved in intracellular transport, signalling against environmental stresses and changes in transcriptional regulatory patterns.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Randhawa and co-authors have studied various aspects of the regulation of lignocellulose degradation by the filamentous ascomycete fungus Penicillium funiculosum. Over-expression of the well-known transcription factor clr2 (which regulates cellulase gene expression in Neurospora and other ascomycetes) in a delta-mig1 strain did not result in an increase in cellulase activity. However, when combined with an increased Ca2+ concentration the cellulase activity in the medium did increase. Using RNA-Seq, the authors have identified a candidate regulator: Snf1. Indeed, a knockout confirms that this gene is involved in the posttranscriptional regulation of cellulase production, specifically by regulating the secretion of the cellulases.

Major comments:

In general, the topic and results are interesting. There are a few issues that need to be addressed, however. The manuscript would benefit from some careful proofreading. For example, articles ('the', 'a') are frequently missing. Very informal language is sometimes used ('zilch effect'). Put a space between '1000bp', etc. It is 'kDa', not 'kD', etc.

I am a bit puzzled by the choice of calcium source: CaCO3, up to 10 g/L. Calcium carbonate does not efficiently dissolve in water unless the pH is low. Fungi generally acidify their culture medium during growth. As such, calcium carbonate likely has a pH buffering effect. Therefore, the described effects may also be attributed to a more neutral pH of the medium, and not necessarily to an increase in calcium ions. The authors have performed RNA-Seq, but as far as I can tell the data has not been made publicly available. At least, the raw reads should be deposited in the Short Read Archive of NCBI (or a similar repository), and preferably also the expression values in GEO of NCBI (or a similar repository). P21. Very little information is provided in the M&M regarding the gene expression analysis. Provide references to all the tools, as well as the version numbers. Were any non-default parameters used? The authors claim that SSP1 CaMKK phosphorylates SNF1 AMPK (last title of the Results section). I don't see any evidence for a direct interaction between these two proteins. I will believe that they are in the same pathway, but if the authors want to claim a direct interaction then additional experiments will be required. Eg Y2H.

Minor comments:

Please add line numbers to the manuscript, this facilitates the review process.

P14 "in all yeasts and filamentous fungi". I doubt that all fungi have been tested.

P18. "in diverse yeasts and fungi". Yeasts are also fungi.

P16. "solves dual purpose". I think this is meant: "serves a dual purpose"?

P17, first paragraph: this seems very speculative to me, so it should probably be labeled as such.

P21. What reference genome is used? Please cite the paper.

Fig 1B. These are reported as volcano plots, but to me it looks like an empty graph (no data points), only a number of genes.

Fig 1D. What do the colors on the right represent?

On various places in the manuscript the term "three times in triplicate" is used. What is meant here, three technical replicates of each of the three biological replicates?

P46. "We aimed to sought"

Abstract: The sentence "Further, Ca2+-signaling" should be rewritten, because currently is seems to suggest that SSP1 downregulates the phosphor-HOG1 levels.

Significance

In general, the topic and results are interesting. There are a few issues that need to be addressed, however. The manuscript would benefit from some careful proofreading.

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

Learn more at Review Commons

---

Reply to the reviewers

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

* Negreira et al. have studied aneuploidy in Leishmania selected using a "flash selection" with SbIII or miltefosine (MF). They provided evidence for the SbIII arm that a few parasites in the population with a specific genotype were enriched during drug selection, and these selected parasites with continuous drug pressure further present modifications in their ploidy. For MF selection they show a different scenario where first a minor population with a mutation in the MT gene is selected and with further passages with drugs, parasites with changes in ploidy are further enriched.*

* Here are some comments that hopefully will be helpful for the authors.*

* The plasticity of the Leishmania genome is fascinating. It is remarkable that these parasites can tolerate so many and frequent changes in ploidy. Either these changes are stochastic and serendipitous or as convey by the authors are part of the parasite arsenal to respond to a changing environment. They cleverly used single cell sequencing and bar-coded parasites in this well designed and well conducted study to assess the role of ploidy in parasite biology.*

1. Drugs are not inducing any of the changes observed, instead the drugs are selecting for parasites with different genotypes (e.g. polyploidy of chromosome 23 for SbIII or parasites with mutations in MT). This is an important conceptual difference and the authors need to change their text throughout starting at line 28.

R: We agree with the reviewer and adapted the text. These changes were introduced as follow:

Line 27 (line 19 in the new version):

“____we revealed that ____antimony-induced aneuploidy changes ____under antimony pressure____ result from the polyclonal selection of pre-existing karyotypes”

Line 201 (line 187 in the new version):

“____This approach revealed that____ the flash selection with SbIII ____induced led to____ a fourfold reduction in lineage diversity that stabilized between passages 3 to 4, leaving between 101 to 131 of detectable lineages”

Line 354 (line 381 in the new version)

“____The flash selection performed with miltefosine revealed a contrasting scenario where aneuploidy remained unchanged ____even after a stronger bottleneck induced by associated with the drug at passage 1, 25 µM and illustrated by the strong decrease in barcode diversity (from 453 to 7 lineages).”

• Line 170. Its is probably expected that no cells have increased copy of chromosome 23, 27 and 31 after single cell genomics. None of the first passages of the four SePOP are polyploid for chromosome 27. One possibility is that a subpopulation of cells with increased copy of chr. 23 (because of MRPA?) and 31 (because of ?) are first selected and in subsequent passages cells triploid for 27 are selected. Of note the ploidy of chr. 27 appears to decrease from passage 4 to 5 in SePOP1 which is unusual if the drug pressure is maintained.*

R: We agree with the reviewer that the aneuploidy changes seen in the SePOP1-4 can be explained by the initial selection of subpopulations of cells with a beneficial pre-existing dosage increase in one or two chromosomes (e.g., chromosome 23 and 31) followed by the selection of additional cumulative modifications emerging in subsequent time points. This conclusion was previously stated throughout the text and is also depicted by the minimum spanning tree in figure 1C, but we made some alterations in the text in order to better state this conclusion:

Line 100 (line 87 in new version):

“____Using single-cell genome sequencing, we could uncover the evolutionary paths that might have led to the emergence of such aneuploidy changes,____ which involved ____indicating a process of ____selection of pre-existing karyotypes complemented by further ____de novo cumulative ____alterations in chromosome copy number along evolution”

Line 168 (line 156 in new version):

“However, none of the sequenced promastigotes showed amplification of chromosomes 23, 27 and 31 concomitantly, and no pre-existing karyotype was identified with a pentasomy in chromosome 23 as observed in the SePOP3, suggesting that some of the aneuploidy ____modifications were generated along adaptation to SbIII changes seen in SePOP1-4 happened after initial exposure to SbIII.____”

Line 191 (line 177 in new version):

“Altogether, our single-cell data suggest that (i) aneuploidy changes observed in the SbIII-exposed populations are explained by the selection of pre-existing aneuploid cells, complemented by additional somy changes generated de novo during the experiment and initial selection of subpopulations bearing ____pre-existing chromosomal amplifications followed by the further selection of cumulative karyotypic modifications emerging in subsequent time points____ and (ii) that the aneuploidy changes seen in SePOP1-4 would have a polyclonal origin.”

Regarding the decrease of chr.27 in SePOP1 from passage 4 to 5, we believe this decrease is not very significant as its somy value (2.71) indicates that the majority of cells still display a trisomy for this chromosome. Moreover, this decrease coincides with the moment where a dosage increase (from ~3 to ~10 copies per haploid genome) in the MRPA locus happens exclusively in that population and in that passage (see supplementary figure S2B), which likely has a stronger impact in SbIII tolerance compared to the trisomy of chr27.

• Lane 194. I agree with the concept of the selection of pre-existing aneuploid cells but the additional somy changes observed are, in my opinion, just selected because these changes occur continuously.** *

R: The changes mentioned above starting at line 191 were also done in response to this comment.

Their barcoded strategy was interesting but it would appear that different lineages are enriched in the 4 SePOP. It would be of interest to test whether those lineages have similar ploidy at the onset. I am unclear of why they have to amplify the barcode prior sequencing. Could they just not get this info from the SePOP data; it is my understanding that the drug selection was done with the barcoded population. This would have facilitated the correlation barcode-specific ploidy.

R: We agree that it would have been interesting to integrate the single-cell genomics and the barcode data in order to determine if the selected lineages had similar karyotypes at the onset of the experiment. However, although the genome coverage of individual cells in the single-cell genomics method used in our study is enough to determine differences in chromosome copy number, it is not enough to evaluate, at sequence level, individual genomic loci such as the lineage barcodes. This is because the genome coverage per cell is too low (in our case 0,8x) meaning that most genomic loci are mapped by just a single sequence read or not mapped at all (10X Genomics, 2020). Thus, it was not possible to determine the lineage barcode of individual cells from the single-cell data.

Regarding the need for amplifying the barcodes: in contrast to WGS, a targeted amplification of the barcodes enabled us to obtain millions of reads covering the barcodes. This, in turn allowed quantifying accurately the frequency of each barcoded lineage.

This is now mentioned in the text starting in line 514 (548 in the new version):

“____Barcode amplification was done using the same DNA samples used for bulk whole genome sequencing. Targeted amplification of the barcodes is needed as the number of reads containing a lineage barcode (~50 pair end reads per sample on average in our case) in the whole genome sequencing data is insufficient for the determination of the frequency of each barcoded lineage in the parasite pool.____”

• The MF screen was harsh and the parasites selected (derived from few clones within the population when considering the time needed to expand) contained SNPs in MT. Difficult to compare the two screens. Passages with higher MF concentration led to major changes in ploidy but with few common features between the MePOP lines.*

R: The screen of the BPK282 strain under SbIII or miltefosine pressure provides two contrasting models and this is one of the interests of the present study. The BPK282 strain belongs to a population of L. donovani parasites from the lowlands of the Indian subcontinent, where parasites were exposed to strong SbIII pressure for decades, even more since these parasites are transmitted from human to human. This population is characterized by strong genomic variations affecting SbIII susceptibility, of which the intra-chromosomal amplification of MRPA is a well-known driver of SbIII pre-adaptation. BPK282 has this intrachromosomal amplification of MRPA and thus it is strongly pre-adapted to SbIII. In contrast, at the time of isolation of BPK282, miltefosine was not yet implemented in clinical practice in the Indian sub-continent (ISC). BPK282 is considered highly susceptible to miltefosine and pre-adaptation to this drug was not, until the present study, identified in this strain and in the ISC population it was isolated from. We performed the flash selection with both drugs to investigate if aneuploidy modulations would follow similar patterns in these two contrasting environments, one where the strain is pre-adapted, and another where it is highly susceptible.

We state this starting in line 242 (line 230 in the new version):

“The results described above demonstrated the importance of aneuploidy for parasite adaptation to high SbIII pressure together with the polyclonality of corresponding molecular adaptations. We aimed here to verify if the same features would be observed with another anti-leishmania drug, miltefosine. In contrast to SbIII, there was – at least before present study – no pre-adaptation known to miltefosine in the BPK282 strain, which is considered very susceptible to the drug (23).”

We also added two sentences in the discussion reiterating this contrast between SbIII and MF in BPK282:

Starting in line 353 (377 in the new version):

__“Finally, we assessed the role and dynamics of aneuploidy under strong pressure of another drug, miltefosine. _Noteworthy, BPK282 was isolated from the population endemic in the Gangetic plain, before miltefosine was implemented in the region (in sharp contrast to SbIII). Hence different results were expected for the scenario of genomic adaptation and clonal dynamics._” __

In addition, we believe that the results of the miltefosine flash selection further corroborate the notion that aneuploidy modulations seen in these drug selection experiments can happen de novo along adaptation to the drug. This was not well stressed in the manuscript and thus we included the following statement during in the discussion:

Starting at line 360 (line 387 in the new version):

“__This demonstrated that the strong bottleneck associated with initial exposure to miltefosine in the first passage did not impair the potential for aneuploidy modulations in later passages, and that these modifications depend on the strength of the stress caused by the drug. These observations are also in agreement with the notion of aneuploidy modulations happening de novo during adaptation to the drug as the aneuploidy profiles seen at passage 9 in the MePOPs exposed to 100 µM are also very different from the pre-existing karyotypes identified in the single-cell data of BPK282____.” __

• I am not asking for extra work but as a suggestion to help in linking ploidy with phenotype it would have been very interesting to look at 5 passages without drug (SbIII or miltefosine) to see whether a decrease in ploidy is correlated to a decrease in resistance.*

R: Unfortunately, we do not have access to the selected populations anymore, but we agree that characterizing these selected populations after keeping them for a few passages without drug would further strengthen the understanding of the relationship between aneuploidy modulations and SbIII tolerance.

Minor points

1. The environment studied (high drug pressure) is unlikely to occur in nature. The authors may wish to comment on how this may translate in the sand fly or in animals.

First of all, the population of L. donovani from which strain BPK282 originated has been naturally under high drug pressure since decades, given the anthroponotic nature of transmission in the Indian sub-continent and the absence of reported animal reservoir. An additional pressure came from the strong pollution with Arsenic, that is present in the lowlands where BPK282 was isolated (Perry et al., 2011). The same authors showed that chronic exposure to arsenic in drinking water can lead to resistance to antimonial drugs (cross resistance) in a mouse model of visceral leishmaniasis and concluded that arsenic contamination in the Gangetic plain may have played a significant role in the development of Leishmania antimonial resistance (Perry et al., 2013). This might explain why antimony resistance drivers like amplification of MRPA were already present in the populations even before antimony was implemented in the region (Imamura et al., 2016).

This is now mentioned in the text starting at line 311 (320 in the new version)

__ “_This pre-adaptation likely comes from the combination of high antimony pressure for decades, highly endemic pollution with arsenic – which can cause cross-resistance to antimonials (33, 34) – and anthroponotic transmission without animal reservoir_.”__

Secondly, in current study, we pushed further the parasite and experimentally exposed it to even higher drug pressure. Our flash selection approach was done as a general model to investigate the mechanisms that Leishmania exploits in order to adapt to sudden and strong environmental stresses, with a focus on aneuploidy changes. This is stated in the manuscript.

Starting at line 93 (line 79 in the new version):

“____In the present study we aimed to address these questions using a reproducible in vitro evolutionary model to study aneuploidy modulations and karyotype evolution in the context of adaptation to sudden environmental stresses, invoked here by the direct exposure to high concentrations of 2 drugs, trivalent antimonial (SbIII) or miltefosine (further called ‘flash selection’).”

In addition, for miltefosine, the concentrations used in our flash selection are lower than the concentrations found in the blood of treated patients or inside macrophages. Thus, for miltefosine, parasites are likely to be exposed to similar or even higher concentrations than those used in our study. We now highlight this in the discussion of the manuscript:

Starting at line 291 (line 290 in new version):

“This abrupt change in environments is also a characteristic of drug treatment. In the case of antimonials, measures made in patients treated for visceral leishmaniasis estimate a peak of 10 mg/L or ~82 µM of Sb in the blood after only 2 hours post drug administration (26). For miltefosine, blood concentrations can be as high as 70 µg/ml, or 172 µM after 72h (27). Moreover, bone marrow-derived macrophages exposed to 10 µM of miltefosine in vitro display intracellular concentrations of the drug as high as 323 µM after 72h (28). This illustrates that Leishmania parasites are directly exposed to sharp increases in drug concentrations – in the case of miltefosine, even higher than the concentrations used in this study – in patients upon drug administration.”

With respect to the importance of sand flies or animals in the environmental pressure, (i) animals play a negligible role given that transmission of L. donovani in the ISC is anthroponotic, without animal reservoir and (ii) the sand fly hosts the parasite for a short period of time (max 10 days), during which the parasite is not exposed to drugs.

• In Fig. S2 MRPA in SePOP1 is a signature of extrachromosomal amplification. *Was that studied?

R: We previously showed that amplification of MRPA in L. donovani encountered in the Indian sub-continent was intrachromosomal (Imamura et al., 2016); further amplification of that specific gene could occur by intrachromosomal expansion/contraction or indeed by episomal amplification. However, one of the core messages of present paper is that increased somy of chr23 automatically leads to increased dosage of the intra-chromosomal MRPA amplicon. We adapted the text in order to acknowledge the possibility of episomal amplification:

Starting at line 140 (line 127 in the new version):

“The BPK282 strain already contains a natural intra-chromosomal amplification of the MRPA gene that may bring a pre-adaptation to SbIII (____14____), and the locus might be subject to further ____intrachromosomal ____expansion ____or____,____ contraction____, or episomal amplification____.”

• For Chromosome 31 in the Sb screen, it would appear that the proximal (left) part is of lower copy number than the distal (right) portion of the chromosome. How could this have happened? Deletion of a portion of chromosome 31 for one allele? This has been described before (Mukherjee et al., 2013) in SbIII resistant lines as one telomeric end of Chr. 31 encodes AQP1, the route of entry of SbIII.*

R: The figures 1A and 2F and 3A do not indicate the copy number of intra-chromosomal segments as they reflect a single numeric value representing the somy of each chromosome at different time points (the x axis of the graphs). Thus, there is no information on differences between distal or proximal copy numbers inside a chromosome in those figures. The only figure showing read depth along the chromosome is fig S2B and corresponds to chr23 and not 31. It is indeed possible that there are telomeric deletions affecting AQP1 but this was not the scope of our study, since we were interested in understanding the reasons and possible drivers of increased gene dosage of chr31.

Reviewer #1 (Significance (Required)):

* The plasticity of the Leishmania genome is fascinating. It is remarkable that these parasites can tolerate so many and frequent changes in ploidy. Either these changes are stochastic and serendipitous or as convey by the authors are part of the parasite arsenal to respond to a changing environment. They cleverly used single cell sequencing and bar-coded parasites in this well designed and well conducted study to assess the role of ploidy in parasite biology.*

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

* Negreira et al. present a study that aims to understand the early evolution of aneuploidy. They use Leishmania, a protozoon parasite known for its genome plasticity, as model, and two drugs as stress inducers. In this work, they use single-cell genomics and lineage tracing to detect changes in chromosome copy numbers. They conclude that, although parasites tend to have genomes with unusual plasticity, aneuploidy dynamics depend on the stressor more than the organism.

* Further experiments:

1. Lines 121-124: I believe the authors should corroborate the statement that expansion of lineages that were fitter prior to drug exposure is stochastically by doing a statistical test comparing their obtained data and randomly generated simulated values. Given that there is still a considerable proportion of lineages with higher fitness and found in more than one passage, I believe this experiment/test would add strength to the conclusion.

R: We believe that the stochasticity per se is not the relevant aspect of our results, but the fact that the expansion of different lineages in different populations is followed by the emergence of the same somy changes in a set of chromosomes (23, 27, and 31), thus showing a process of convergent evolution. Therefore, we decided to reduce the emphasis on the stochasticity itself and adapted the text to highlight this process of convergence. This was done in the following parts of the manuscript:

Starting at line 98 (line 84 in new version):

“we revealed that changes in aneuploidy under ____SbIII____ pressure have a polyclonal origin, arising from the reproducible survival of a specific subset of lineages, which further expand stochastically differentially between independent replicates but converge to similar aneuploidy modifications”.

Starting at line 220 (line 205 in new version):

“____Moreover____, ____most of the positively affected lineages were enriched in only one of the SePOPs ____(Fig. 2C and fig. S3B)____ (figure 2C and supplementary figure S3B).____, suggesting that____. Altogether, these data indicate that____ (i) a subset of lineages was fitter to SbIII prior the drug exposure and (ii)____ their ____the further____ expansion ____of these surviving lineages was ____stochastically driven. divergent between independent replicates.____”

Starting at line 344 (line 366 in the new version):

“From 453 different traceable lineages, 303 consistently disappeared during SbIII exposure and 60 showed an increased frequency in at least one replicate. Most of these positively affected lineages were enriched in only one of the SePOP replicates, suggesting (i) higher tolerance to SbIII in a subset of lineages that reproducibly survived the flash selection and (ii) further expansion of these surviving lineages being stochastically driven. , including lineages which were dominating the population at the onset of the experiment (figure 2F), thus indicating that these lineages had a fitness disadvantage to SbIII compared to the other lineages. Among the surviving lineages, 60 could further expand in at least one of the SePOPs, leading to different clonal compositions in each population. Interestingly, changes in clonal composition in each SePOP coincide with the moments where changes in aneuploidy are observed in these populations, suggesting that these aneuploidy changes are due to the emergence of subsets of fitter lineages. Moreover, the observation that the same set of 3 chromosomes displayed dosage increases in all SePOP despite the fact that different lineages dominated each SePOP points to a process of convergent evolution, which further supports the notion of these chromosomes being under positive selection.”

Minor issues:

Fig. 1B: Add label to top horizontal axis, showing frequency of each karyotype.

R: A label stating ‘Number of Cells’ was added at figure 1B.

Lines 92-96: Could the authors postulate how and why pre-existing aneuploid cells seem to be selected upon SbIII exposure?

R: We believe that some aneuploidy changes, like the dosage increase of chromosome 23 (from 3 to 4 copies) offer an adaptive advantage to the cells bearing it by over-expressing genes related to SbIII tolerance. This was discussed in the manuscript.

starting at line 304 (314 in the new text):

“____Chromosome 23 bears the MRPA genes which encode an ABC-thiol transporter involved in the sequestration of Sb-thiol conjugates into intracellular vesicles (28). Amplification of MRPA genes through extra-or intra-chromosomal amplification is a well-known driver of experimental SbIII resistance. The line here used (BPK282) is remarkably pre-adapted to SbIII (18) – like other strains of the Gangetic plain – thanks to a pre-existing intra-chromosomal amplification of MRPA genes encountered in 200 sequenced L. donovani isolates of that region (13). The recurrent dosage increase of chromosome 23 observed here under SbIII pressure is a rapid way to further amplify the MRPA gene and this mechanism was likely selected instead of further amplifying MRPA genes intra-chromosomally.”

Fig. 3: Are panels B and C swapped in the figure or the reference swapped in the text? Fig. 3C seems to refer to the mutation (lines 173-179), whereas Fig. 3B seems to relate to the surviving lineages (lines 183-186).

R: Indeed, figures 3B and 3C were erroneously positioned in the panel. This is now fixed in the new version.

Lines 94-97: Could the authors comment on the advantages and disadvantages of such an aggressive selection method? I am not surprised with such a drastic decrease in lineage diversity in this context.

R: We now added a section at the beginning of the discussion commenting this:

starting at line 291 (line 281 in the new version):

“____Historically, adaptation in Leishmania was mainly addressed using a ‘gentle’ stepwise approach where parasite populations are exposed to progressively increasing drug concentrations in vitro over the course of months, allowing these populations to adapt to each concentration before proceeding to the next increment (19, 23-25). This approach is useful to reveal mechanisms promoting full resistance against that drug which emerge at the later time points where drug concentration is high, but it precludes the evaluation of mechanisms allowing parasites to cope with sudden and strong environmental changes as initial concentrations are often too permissive. Importantly, in nature, changes in environmental pressures are often abrupt rather than gradual, and therefore, demand for mechanisms which allow parasite populations to quickly adapt to the new environment.____”

And then on line 300 in the new version:

“____In____ the present study, we investigated the mechanisms governing the early adaptation of Leishmania promastigote populations to a direct exposure to high concentrations of two drugs – SbIII and miltefosine – as models of sudden environmental stresses.____”

Could the authors elaborate on what is different in chromosome 31 that makes it so prone to change?

R: We improved our discussion about the potential drivers of dosage increases for the other 2 chromosomes (chr 27 and chr 31) which, apart from chr23, are also consistently amplified under SbIII exposure.

Starting at line 320 (line 331 in the new version):

“_Regarding the other 2 chromosomes, chromosome 31 also bears a gene involved in antimony resistance, the sodium stibogluconate resistance protein gene (LdBPK310951.1). Interestingly, the ortholog of this gene displayed an increased copy number in L. braziliensis promastigotes experimentally selected for antimony resistance in vitro compared to non-selected lines (31). Moreover, this same study found a 50 kb intrachromosomal amplification affecting 23 genes (out of a total of 31 amplified genes) in chromosome 27 in the SbIII resistant line, with many of these genes displaying a copy number more than 10 times higher compared to the SbIII sensitive line (31). Among these genes, a WW domain/Zinc finger C-x8-C-x5-C-x3-H type - protein gene (LdBPK_270130.1 ortholog in L. donovani) was also the gene with the most upregulated expression compared to the SbIII-sensitive line. Importantly, CCCH type zinc finger proteins are known targets of antimony (32), and therefore, a higher expression of this gene might mitigate its inactivation by the drug.____” __

And for chromosome 31, we also discussed further its potential role in general response against drug-induced stresses.

Starting at line 360 (line 394 in new version):

“____At 100 µM, aneuploidy changes were specific to each of the 4 MePOP replicates, with the exception of chromosome 31 that consistently showed a higher somy than the control. The fact that an increase in copy number of chromosome 31 was observed under strong SbIII and miltefosine pressure, as well as under pressure of other drugs (24) might indicate that the dosage increase in this chromosome has also a general role against multiple types of stresses. ____Noteworthy, there are several ABC transporters in that chromosome (ABCC4-7 and ABCD3) which could play a role in drug efflux (36). Moreover, ontology analysis of chromosome 31 in L. braziliensis have demonstrated an enrichment of genes involved in iron metabolism which could play a role in general adaptation to oxidative stresses (37), but empirical evidence is still lacking.____”

Reviewer #2 (Significance (Required)):

* Aneuploidy can be well-tolerated, beneficial, or deleterious. Particularly, they can confer resistance against environment stresses, including drug pressures. This study aims to understand how aneuploidy arises. The authors approach this question using a model organism, Leishmania donovani, and two distinct drugs as environmental stressors. Using single-cell DNA sequencing and lineage tracing, the authors find that the appearance of aneuploidy is dependent on the drug used, which makes it dependent on the environmental stressor, rather than pre-determined. Importantly, they present a new barcoding method that may be useful to the field of experimental genome evolution.*

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

* This interesting, well written paper uses cutting edge technologies to address the evolutionary dynamics of changes in Leishmania donovani genomes in response to high drug pressure. Using single-cell genome sequencing and lineage tracing with a newly adapted cell barcoding system, the authors were able to follow aneuploid changes and lineage selection following exposures to high concentrations of either antimony or miltefosine. The main conclusions drawn from the careful bioinformatic analyses and methodic representation of 864 single cell genomes and 453 different traceable lineages were that for each drug exposure there was polyclonal selection of pre-adapted parasites complemented by de novo adaptions. Consistent changes in aneuploidy were associated with the populations selected by antimony, while miltefosine selected for populations that had a point mutation in a miltefosine transporter gene. These conclusions are well supported by the data.*

* Reviewer #3 (Significance (Required)):*

*3 comments, 3 responses *

Comment 1

*One general comment is that the contribution of pre-adapted lineages to the emergence of drug resistant populations under conditions of natural exposure is apt to be overstated from the current analysis. As the authors discuss, the L. donovani line used is already pre-adapted to resist antimony due, at least in part, to the amplification of the MRPA gene on chromosome 23. So it is expected that lineages adapted to strong antimony pressure will pre-exist in this line. It seems possible that the de novo adaptions that were observed, involving further copy number amplification of chromosome 23 and other chromosomes (e.g., chr 31), might be facilitated by their pre-existing aneuploides. Thus, the evolutionary dynamics observed might be very particular to these sorts of pre-conditioned cells. *

R: Although BPK282 is indeed pre-adapted to antimony due to an amplification of the MRPA locus, this strain is a clone, so this intra-chromosomal amplification is shared among all cells in the population. Thus, it is probable that this intra-chromosomal amplification alone is not the only reason why some lineages are better adapted to antimony than others, but its combination with variations in aneuploidy affecting chromosome 23. We agree that de novo adaptations were likely facilitated by the presence of pre-existing aneuploidies. This was already commented in answers to comments 2 and 3 of reviewer 1.

Comment 2

It should also be discussed that the culture condtions themselves may pre-condition the parasites for antimony resistance (and possibly other drugs). Continuous passage of L. donovani in axenic culture produced consistent patterns of aneuploid changes, including amplification of Chr 23 (Barja et al., Nat Ecol evol, 2017). Thus a potential caveat of the use of cultured promastigotes is that their culture adaptions might involve genes on the same chromosomes that confer drug resistance.

R: Indeed, and we are aware of the work of Barja et al 2017. However, the flash selection models characterize a competition assay between (sub)clonal lineages which are exactly in the same environment (lineages within each SePOP were in the same culture flasks). Thus, although culture adaptation might indeed lead to pre-conditioning against SbIII due to amplification of chr23, this pre-conditioning should affect the entire population and does not explain the differences in susceptibility to SbIII between the lineages within each SePOP. Moreover, the controls (maintenance in the same culture medium but without drug pressure) did not show any change in their aneuploidy, while SePOP showed an increase in somy of several chromosomes, including chromosome 23 (see fig.1A).

Comment 3

For the miltefosine selection, of the 7 lineages surviving in at least one of the MePOP replicates, only lineage 302 is represented more than once. What is the evidence that the adaptive mutations in the other 6 lineages were pre-existing and did not arise de novo?

R: We agree that evidence for pre-existing mutations is only present for lineage 302 and changed that in the text.

At line 29 (line 22 in the new version):

“In the case of miltefosine, early parasite adaptation was associated with independent pre-existing point mutations in a miltefosine transporter gene.”

Figs 3b and 3c are incorrectly referenced in the text.

R: Fixed in new version.

Discussion p. 8 - "Interestingly, the Gly160Asp mutation also correlated with the frequency of a specific lineage (lineage 27) and appeared in 3 of the 4 MePOPs, indicating that this was a pre-existing mutation found in that lineage." Lineage 302 would appear to be the correct lineage, not 27. Please clarify.

R: Indeed, the correct is lineage 302. This has now been fixed in the new version.

Additional modifications in the manuscript:

1) The mutation in the LdMT gene affecting the codon of amino acid 1016 was described as a Glutamate to stop codon mutation (Glu1016Stop), while in fact the original amino acid is a Serine (Ser1016Stop). This was corrected in the new version.

References

10X Genomics, 2020. How much of a single cell’s genome is amplified? [WWW Document]. 10X Genomics. URL https://kb.10xgenomics.com/hc/en-us/articles/360005108931-How-much-of-a-single-cell-s-genome-is-amplified- (accessed 4.3.23).

Imamura, H., Downing, T., Van den Broeck, F., Sanders, M.J., Rijal, S., Sundar, S., Mannaert, A., Vanaerschot, M., Berg, M., De Muylder, G., Dumetz, F., Cuypers, B., Maes, I., Domagalska, M., Decuypere, S., Rai, K., Uranw, S., Bhattarai, N.R., Khanal, B., Prajapati, V.K., Sharma, S., Stark, O., Schönian, G., De Koning, H.P., Settimo, L., Vanhollebeke, B., Roy, S., Ostyn, B., Boelaert, M., Maes, L., Berriman, M., Dujardin, J.-C., Cotton, J.A., 2016. Evolutionary genomics of epidemic visceral leishmaniasis in the Indian subcontinent. eLife 5, e12613. https://doi.org/10.7554/eLife.12613

Perry, M.R., Wyllie, S., Prajapati, V.K., Feldmann, J., Sundar, S., Boelaert, M., Fairlamb, A.H., 2011. Visceral Leishmaniasis and Arsenic: An Ancient Poison Contributing to Antimonial Treatment Failure in the Indian Subcontinent? PLoS Negl. Trop. Dis. 5, e1227. https://doi.org/10.1371/journal.pntd.0001227

Perry, M.R., Wyllie, S., Raab, A., Feldmann, J., Fairlamb, A.H., 2013. Chronic exposure to arsenic in drinking water can lead to resistance to antimonial drugs in a mouse model of visceral leishmaniasis. Proc. Natl. Acad. Sci. 110, 19932–19937. https://doi.org/10.1073/pnas.1311535110

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

This interesting, well written paper uses cutting edge technologies to address the evolutionary dynamics of changes in Leishmania donovani genomes in response to high drug pressure. Using single-cell genome sequencing and lineage tracing with a newly adapted cell barcoding system, the authors were able to follow aneuploid changes and lineage selection following exposures to high concentrations of either antimony or miltefosine. The main conclusions drawn from the careful bioinformatic analyses and methodic representation of 864 single cell genomes and 453 different traceable lineages were that for each drug exposure there was polyclonal selection of pre-adapted parasites complemented by de novo adaptions. Consistent changes in aneuploidy were associated with the populations selected by antimony, while miltefosine selected for populations that had a point mutation in a miltefosine transporter gene. These conclusions are well supported by the data.

Significance

One general comment is that the contribution of pre-adapted lineages to the emergence of drug resistant populations under conditions of natural exposure is apt to be overstated from the current analysis. As the authors discuss, the L. donovani line used is already pre-adapted to resist antimony due, at least in part, to the amplification of the MRPA gene on chromosome 23. So it is expected that lineages adapted to strong antimony pressure will pre-exist in this line. It seems possible that the de novo adaptions that were observed, involving further copy number amplification of chromosome 23 and other chromosomes (eg chr 31), might be facilitated by their pre-existing aneuploides. Thus the evolutionary dynamics observed might be very particular to these sorts of pre-conditioned cells. It should also be discussed that the culture condtions themselves may pre-condition the parasites for antimony resistance (and possibly other drugs). Continuous passage of L. donovani in axenic culture produced consistent patterns of aneuploid changes, including amplification of Chr 23 (Barja et al., Nat Ecol evol, 2017). Thus a potential caveat of the use of cultured promastigotes is that their culture adaptions might involve genes on the same chromosomes that confer drug resistance.<br /> For the miltefosine selection, of the 7 lineages surviving in at least one of the MePOP replicates, only lineage 302 is represented more than once. What is the evidence that the adaptive mutations in the other 6 lineages were pre-existing and did not arise de novo?

Figs 3b and 3c are incorrectly referenced in the text.

Discussion p. 8 - "Interestingly, the Gly160Asp mutation also correlated with the frequency of a specific lineage (lineage 27) and appeared in 3 of the 4 MePOPs, indicating that this was a pre-existing mutation found in that lineage." Lineage 302 would appear to be the correct lineage, not 27. Please clarify.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Negreira et al. present a study that aims to understand the early evolution of aneuploidy. They use Leishmania, a protozoon parasite known for its genome plasticity, as model, and two drugs as stress inducers. In this work, they use single-cell genomics and lineage tracing to detect changes in chromosome copy numbers. They conclude that, although parasites tend to have genomes with unusual plasticity, aneuploidy dynamics depend on the stressor more than the organism.

Further experiments:

Lines 121-124: I believe the authors should corroborate the statement that expansion of lineages that were fitter prior to drug exposure is stochastically by doing a statistical test comparing their obtained data and randomly generated simulated values. Given that there is still a considerable proportion of lineages with higher fitness and found in more than one passage, I believe this experiment/test would add strength to the conclusion.

Minor issues:

Fig. 1B: Add label to top horizontal axis, showing frequency of each karyotype. Lines 92-96: Could the authors postulate how and why pre-existing aneuploid cells seem to be selected upon SbIII exposure? Fig. 3: Are panels B and C swapped in the figure or the reference swapped in the text? Fig. 3C seems to refer to the mutation (lines 173-179), whereas Fig. 3B seems to relate to the surviving lineages (lines 183-186). Lines 94-97: Could the authors comment on the advantages and disadvantages of such an aggressive selection method? I am not surprised with such a drastic decrease in lineage diversity in this context.

Could the authors elaborate on what is different in chromosome 31 that makes it so prone to change?

Significance

Aneuploidy can be well-tolerated, beneficial, or deleterious. Particularly, they can confer resistance against environment stresses, including drug pressures. This study aims to understand how aneuploidy arises. The authors approach this question using a model organism, Leishmania donovani, and two distinct drugs as environmental stressors. Using single-cell DNA sequencing and lineage tracing, the authors find that the appearance of aneuploidy is dependent on the drug used, which makes it dependent on the environmental stressor, rather than pre-determined. Importantly, they present a new barcoding method that may be useful to the field of experimental genome evolution.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Negreira et al. have studied aneuploidy in Leishmania selected using a "flash selection" with SbIII or miltefosine (MF). They provided evidence for the SbIII arm that a few parasites in the population with a specific genotype were enriched during drug selection, and these selected parasites with continuous drug pressure further present modifications in their ploidy. For MF selection they show a different scenario where first a minor population with a mutation in the MT gene is selected and with further passages with drugs, parasites with changes in ploidy are further enriched.

Here are some comments that hopefully will be helpful for the authors.

The plasticity of the Leishmania genome is fascinating. It is remarkable that these parasites can tolerate so many and frequent changes in ploidy. Either these changes are stochastic and serendipitous or as convey by the authors are part of the parasite arsenal to respond to a changing environment. They cleverly used single cell sequencing and bar-coded parasites in this well designed and well conducted study to assess the role of ploidy in parasite biology.

1. Drugs are not inducing any of the changes observed, instead the drugs are selecting for parasites with different genotypes (e.g. polyploidy of chromosome 23 for SbIII or parasites with mutations in MT). This is an important conceptual difference and the authors need to change their text throughout starting at line 28.
2. Line 170. Its is probably expected that no cells have increased copy of chromosome 23, 27 and 31 after single cell genomics. None of the first passages of the four SePOP are polyploid for chromosome 27. One possibility is that a subpopulation of cells with increased copy of chr. 23 (because of MRPA?) and 31 (because of ?) are first selected and in subsequent passages cells triploid for 27 are selected. Of note the ploidy of chr. 27 appears to decrease from passage 4 to 5 in SePOP1 which is unusual if the drug pressure is maintained.
3. Lane 194. I agree with the concept of the selection of pre-existing aneuploid cells but the additional somy changes observed are, in my opinion, just selected because these changes occur continuously.
4. Their barcoded strategy was interesting but it would appear that different lineages are enriched in the 4 SePOP. It would be of interest to test whether those lineages have similar ploidy at the onset. I am unclear of why they have to amplify the barcode prior sequencing. Could they just not get this info from the SePOP data; it is my understanding that the drug selection was done with the barcoded population. This would have facilitated the correlation barcode-specific ploidy.
5. The MF screen was harsh and the parasites selected (derived from few clones within the population when considering the time needed to expand) contained SNPs in MT. Difficult to compare the two screens. Passages with higher MF concentration led to major changes in ploidy but with few common features between the MePOP lines.
6. I am not asking for extra work but as a suggestion to help in linking ploidy with phenotype it would have been very interesting to look at 5 passages without drug (SbIII or miltefosine) to see whether a decrease in ploidy is correlated to a decrease in resistance.

Minor points

1. The environment studied (high drug pressure) is unlikely to occur in nature. The authors may wish to comment on how this may translate in the sand fly or in animals.
2. In Fig. S2 MRPA in SePOP1 is a signature of extrachromosomal amplification. Was that studied?
3. For Chromosome 31 in the Sb screen, it would appear that the proximal (left) part is of lower copy number than the distal (right) portion of the chromosome. How could this have happened? Deletion of a portion of chromosome 31 for one allele? This has been described before (Mukherjee et al., 2013) in SbIII resistant lines as one telomeric end of Chr. 31 encodes AQP1, the route of entry of SbIII.

Significance

The plasticity of the Leishmania genome is fascinating. It is remarkable that these parasites can tolerate so many and frequent changes in ploidy. Either these changes are stochastic and serendipitous or as convey by the authors are part of the parasite arsenal to respond to a changing environment. They cleverly used single cell sequencing and bar-coded parasites in this well designed and well conducted study to assess the role of ploidy in parasite biology.

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

Learn more at Review Commons

---

Reply to the reviewers

We thank all reviewers for their comments and suggestions. The revised manuscript included new experiments they suggested and extensive text edits. Our point-by-point response is shown in bold.

Point-by-point description of the revisions

—----------------------------------------------------------------------------------------------------------------

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

Summary

In this manuscript, Blank et al. propose a link between cell-cycle dependent changes in metabolic flux and corresponding changes in TORC1 activity in yeast cells. Based on their findings, the authors propose that Bat1-dependent leucine synthesis from glucose increases as cells progress through G1 and that this activates TORC1 to drive cell cycle progression. Although the existence of cell-cycle dependent synthesis of leucine is a novel and exciting finding, several aspects of the proposed model are not sufficiently supported by experimental evidence, in particular the fact that the increase in Leu synthesis is causing the increase in TORC1 activity in late G1.

Major comments:

1. To show that the increase in Leu biosynthesis in S-phase is activating TOR, one would ideally want to blunt this increase in biosynthesis and assay TORC1 activity. Admittedly, this is difficult. So, instead, the authors study bat1- cells which have strongly impaired synthesis of BCAA including Leucine. The relevance of these bat1- cells to the proposed cell-cycle dependent model, however, is questionable for two reasons: 1) Although the authors state that "exogenous supplementation of BCAAs in all combinations suppressed the growth defect of bat1- cells, especially when valine was present", the spot assays in Figure 3 show visible rescues only when valine is present either alone or in combination, while supplementation of leucine or isoleucine does not seem to have any effect. Hence it appears that the bat1- phenotype is mainly due to limiting valine levels, not leucine levels. 2) The relevance of these results for understanding TORC1 regulation are questionable, since valine does not typically activate TORC1. Does addition of Leu to bat1- cells increase TORC1 activity ? RESPONSE: The reviewer’s comments were very valuable. We performed the suggested experiments (adding not only Leu but also Ile and Val) to bat1 cells and measuring phosphorylation of Rps6 (see new Figure 4D) and the DNA content of those cells (see new Figure 3C). We found that Leu weakly promotes cell cycle progression, compared to the addition of Val, which also leads to pronounced activation of TORC1 (>10-fold activation; see Figure 4D). We discuss these findings in the revised text.

We also note, as published by others and now discussed in the text, that in WT cells, exogenous addition of Leu (or any other BCAA) does not lead to sustained activation of TORC1 (see new Figure 4D). This is not surprising. As reported by the Hall lab (see PMID: 25063813, which we now cite), the Gtr-dependent activation of TORC1 by BCAAs mentioned by the reviewer is very transient. Hence, our new data, showing sustained TORC1 activation and cell cycle effects upon Val addition in bat1 cells, is exciting. They argue that bat1 cells serve as a highly sensitized background of low TORC1 activity, enabling the display of effects that are difficult to measure in WT cells.

TORC1 activity is known to depend on steady-state leucine concentrations in the cell rather than on leucine flux. Although the authors observe that the synthesis rate of leucine increases during G1 progression, this does not necessarily translate into increased leucine concentrations in the cell. To support the claim that the increase in TORC1 activity during G1 progression depends on leucine, the authors would need to show that, not only leucine synthesis, but also overall leucine levels in the cell increase during G1 progression.

RESPONSE: We did this experiment and now report the data (see new Figure EV2), using the Edman degradation-based assay. We found that changes in the steady-state levels of BCAAs had a similar pattern, and those changes were most significant for valine (rising 30-40% from late G1 to G2/M). Nonetheless, we note also that the kinetics of amino acid synthesis measured by our isotope tracing experiment need not match the steady-state levels of amino acids. Steady-state levels are affected by a multitude of parameters, only one of which is the rate of synthesis, as we now discuss in detail in the manuscript.

To test whether the increase in Leu biosynthesis in S-phase activates TORC1, a few different approaches could be tested: 1) Since leucine activates TORC1 through the Gtr proteins, the authors could test whether rendering TORC1 resistant to low leucine through expression of constitutively active Gtrs abolishes the cell-cycle dependence in TORC1 activity. 2) Leu could be added to the medium of wildtype cells in G1 to the amount necessary to cause an increase in intracellular Leu levels similar to those seen in S-phase to test whether this increases TORC1 activity.

RESPONSE: We did the suggested experiments, which are now shown in the new Figure 5. Leucine and valine accelerated the rise in TORC1 activity in G1. However, there were no noticeable downstream consequences in the kinetics of cell cycle progression. As we discuss in the text:

“A small acceleration of the rise in the levels of phosphorylated Rps6 was evident in both the leucine- and valine supplemented cells (Figure 5A,B). Nonetheless, there were no noticeable downstream consequences in the kinetics of cell cycle progression, in either the rate the cells increased in size or their critical size (Figure 5A; see values above the corresponding blots), consistent with the notion that TORC1 activity already is at a maximal level in these conditions…”

In Fig 2B one sees that Leu biosynthesis peaks at 150min and then drops again. The p-RpS6 blot in Fig. 5D, however, only goes up to 140 min and shows that TORC1 activity increases up to 140 min, but it doesn't show timepoints beyond 150 min when Leu biosynthesis drops again, and hence one would expect TORC1 activity to drop. If TORC1 activity were to drop from 150min onwards, this would strengthen the correlation between Leu biosynthesis and TORC1 activity.

RESPONSE: The reason for the drop in Figure 2 is trivial and does not affect the interpretation. As seen in Figure 1 (the experiment from which the data in Figure 2 are shown), by 180 min, the cells were entering a new cell cycle, evidenced by a reduction in cell size (Figure 1B) and in the fraction of budded cells (Figure 2B). At that point, there is a mix of mothers and daughters with very poor synchrony, making it impossible to conclude much about the drop in Leu synthesis (i.e., does it arise from the lack of new synthesis in mothers, daughters, or both?). In the experiment in Figure 5, the reviewer mentions (now those figures have moved to File S8 because we added more experiments in the figure) the experiment terminated when peak budding was reached, which was 140 min, within one cell cycle. Lastly, it is important to stress that every elutriation experiment is different. While the times are close, comparing various experiments on a time basis alone is inaccurate. Instead, the metric used in the field to compare different experiments is usually cell size, which we use in all other Figures except Figure 1 because, in that case, the experiment was a time-based, pulse-chase one.

Minor concerns:

1. In Figure EV4, the authors should highlight some of the metabolites that are significantly changed, in particular the BCAA. The figure is not very informative as currently presented. __RESPONSE: We have now labeled the BCAAs, and a few more metabolites as suggested (note the Figure is now EV5). __

Fig 2 - are "expressed ratios" the best term for metabolite levels? Unlike genes, where such heat maps are often used, the metabolites are not 'expressed'. How about 'relative metabolite level' instead?

RESPONSE: Good point. The axis now reads “relative abundance”.

Page 8: "We also measured the MID values from the media of the same cultures used to prepare the cell extracts." Where are these data? We don't see them in File S2?

RESPONSE: The data are in File S2 (there are many ‘sheets’ in the file). In sheets 3,4 are the MID values and the analysis from metabolites in the media.

Fig 4B - the x-axis labeling is missing for the bat1- cells

RESPONSE: Corrected. Note that new DNA content measurements are now shown in Figure 3C.

Although the authors state repeatedly that they show "for the first time in any system" that TORC1 activity is dynamic in the cell cycle, similar observations have already been made before, for instance showing high mTORC1 activity in the G1/S transition in the Drosophila wing disc or low mTORC1 activity during mitosis in mammalian cells (see PMIDs 28829944, 28829945, and 31733992). The text should be amended accordingly.

RESPONSE: Thank you. Corrected.

There are two entries for valine in File S1/Sheet8. Why?

RESPONSE: The reason is that they were detected in both analytical pipelines (primary metabolites and biogenic amines; primary metabolites were measured with GC-TOF MS, while biogenic amines with HILIC-QTOF MS/MS), which were combined in the Table. We did not describe it adequately in the previous version. We do now, in the Methods. We also note that the raw data from each method are shown in the corresponding supplemental files. We combined them in the Table used in the Figure for display purposes. We also note that the amino acids were also measured by another method (PTH-based HPLC). Hopefully, the new edits in the Methods clarify these points.

Reviewer #1 (Significance (Required)):

Significance

Despite the well-known effects of pharmacological or genetic manipulations of TORC1/mTORC1 on cell cycle progression, whether and how mTORC1 activity itself is physiologically coupled to cell cycle progression is still an insufficiently studied aspect. Hence this study provides an interesting link between cell-cycle dependent regulation of amino acid biosynthesis and TORC1 regulation. Importantly, the results of this study rely on centrifugal elutriation to obtain cell cycle synchronization, thus ruling out potential metabolic artifacts due to pharmacological methods. The observed changes in metabolic flux are therefore likely genuine and represent the major strength of the study. The major limitation is the lack of strong evidence supporting the notion that the increase in Leu biosynthesis at late G1 or S-phase is causing the increase in TORC1 activity.

The major advance is conceptual - that amino acid biosynthesis rates are cell-cycle dependent.

These results will be of interest to a broad audience of people studying the cell cycle, cell growth, TORC1 activity, cell metabolism and cancer.

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

This paper provides evidence that branched chain amino acid (BCAA) in the G1 phase of the cell cycle, fueled by pyruvate generated by glucose catabolism activates cell growth and allows cells to reach the critical size required for entry into S phase by activation of TORC1 signaling. Previous work had indicated that Leucine supplementation of a bat1 bat2 mutant, lacking both enzymes that catalyze BCAA from the alpha-keto acid precursors and starved on minimal medium, led to TORC1 activation. This work is significant in suggesting that BCAA synthesis from glucose is responsible for a cyclic activation of TORC1 necessary for a normal rate of cell growth in the G1 phase of the cell cycle.

The study employs metabolic flux analysis of metabolites derived from glucose following a pulse-chase with different isotopes of glucose in synchronized early G1 cells (obtained by elutriation) throughout one cell cycle. They claim that the only compelling changes in metabolites observed as the cell cycle proceeds was a decline in pyruvate containing only one heavy 13C carbon atom and a corresponding increase in Leu (M6) with 6 heavy carbon atoms, which is interpreted to indicate Leu synthesis from pyruvate that begins in early G1 and peaks at mitosis. They show that a bat1 mutant exhibits a slow-growth phenotype that can be mitigated only by valine (although they infer similar effects for Leu and Ile that I find unconvincing) and they observed reductions in all three BCAAs in different experiments that measure steady amino acid levels in different ways (although the results are compelling only for Val). They go on to show evidence that the bat1 mutation reduces birth and mean cell size and leads to an increased proportion of G1 cells in asynchronous cultures, and they claim that bat1 cells take much longer than WT to achieve the same size found when a synchronized WT culture reaches 50% budding (although they don't show the data for this last point.) Interestingly, they find that deleting BAT1 suppresses sensitivity to the TORC1 inhibitor rapamycin (Rap), consistent with the idea that the bat1 mutation impairs TORC1 activity in the same manner as Rap and that BCAA are required to activate TORC1 in WT cells to the level that can be impaired by Rap, as summarized in the model in Fig. 5F. Consistent with this, they present evidence that the bat1 mutation reduces TORC1 signaling as judged by diminished Rps6 phosphorylation (although it was not shown that this effect could be reversed by Val addition). They also show that TORC1 signaling/Rps6-P increases as the cell cycle progresses using elutriated early G1 cells, suggesting that TORC1 activity is periodic in the cell cycle (although they don't establish this periodicity through a second cell cycle).

General critique:

The conclusion that BCAA synthesis from glucose is responsible for a cyclic activation of TORC1 necessary for a normal rate of cell growth in the G1 phase of the cell cycle is potentially of considerable significance. There are however a number of puzzling aspects of the data that seem to weaken this conclusion. As described in greater detail below, it is difficult to explain why only Leu is synthesized from glucose during the cell cycle, and why only Val shows a marked reduction in the bat1 mutant that appears to be responsible for the slow-growth phenotype. In addition, there are important controls lacking of showing that a Val supplement can suppress the G1 delay and reduction in TORC1 signaling in the bat1 mutant. In addition, the evidence that TORC1 activity is periodic in the cell cycle is lacking and it needs to be shown that Rps6-P levels are periodic through at least a second cell cycle.

Major comments:

-Why don't they observe synthesis of Ile and particularly Val in the metabolic flux experiment of Fig. 1, especially considering that only Val appears to be critically required for normal cell growth in the bat1 mutant based on the results in Fig. 3B?

RESPONSE: We now show the actual plots and the errors of all the measurements in Figure 2 (instead of a heatmap we had shown before). Valine (M5) levels show a very similar trend to leucine (M6). The variance in the measurements was higher, though, and statistically, the valine changes were less significant. Hence, it was more appropriate to highlight the leucine changes. Lastly, the new DNA content data (Figure 3C) show an effect upon the addition of leucine, albeit less significant than that of valine addition.

-The data in Fig. 3B do not show a convincing increase in growth of the bat1 mutant with addition of Leu and Ile; and the stimulation by Val alone seems identical to that seen with Val in combination with Leu and Ile. Thus, it appears that the slow-growth of the bat1 mutant results only from reduced Val levels, not all 3 BCAAs, which is at odds with their interpretation of the data.

__RESPONSE. As mentioned above, the effect of valine is more pronounced than leucine's, but leucine does have consequences, best shown in the DNA content analysis (new Figure 3C). We also note that valine alone is insufficient to suppress the growth and cell cycle defects of bat1 cells. The latest data we have added (see Figures 3 and 5) are consistent with the interpretation that at least some de novo synthesis of BCAAs in the cell may be needed, explaining why exogenous BCAAs, including valine, are unable to correct the defects of bat1 cells fully. __

-they claim to see reductions in all three BCAAs in the bat1 mutant; however, no significant reduction was found for Leu in Fig. EV3, and only Val was altered by the 1.5-fold cut-off imposed on the MS metabolomics data in Fig. EV4 (which could be appreciated only by an in-depth examination of the supplementary data in File S1-the Val, Leu, and Ile dots should be labeled in Fig. EV4). In addition, the reductions in Ala and Gly showin in Fig. EV3 were not found in the MS analysis of Fig. EV4. It needs to be acknowledged that the metabolomics data show a marked reduction in the bat1 mutant only for Valine with little or no change in Leucine levels. This result is difficult to explain with the simple models shown in Fig. 3A and 5F, which requires additional comment. The authors should acknowledge the much greater effect of the bat1 mutation on Val levels versus Leu and Ile, revealed both by measuring the levels of BCAAs in the mutant and comparing the BCAAs for rescuing the slow-growth of the mutant, and explain how this can be reconciled with the results in Fig. 2 where only Leu and not Val or Ile synthesis was detected.

__RESPONSE. The perceived discrepancy in the steady-state measurements could easily arise from the different analytical methods used in each case. The differences are less substantial than the reviewer implies. For steady-state measurements in BAT1 vs. bat1 cells, we used the PTH-based method (which only detects amino acids) and two different MS-based pipelines (which detect various metabolites). From the MS-based analyses, the drop for all BCAAs was statistically significant. Although the magnitude of the drop was greater for valine (about 60% for valine vs. ~30% for isoleucine and leucine). Why is this a problem? __

As for the valine changes in the isotope tracing experiments, as we mentioned above, the trend for valine (M5) was similar to that of leucine (M6) (now, hopefully, that data is shown better in Figure 2). Furthermore, as we commented above (see response to Reviewer 1) and now stated in the text, our isotope tracing experiments measure only the rate of synthesis, which need not match the steady-state abundances. The latter are affected by a multitude of variables, including the turnover of proteins and amino acids, not to mention their partition into distinct intracellular pools.

__Lastly, please note that we have now added PTH-based measurements of amino acid levels in the cell cycle of wild type cells (new Figure EV2). As mentioned in our response to Reviewer 1, we found that changes in the steady-state levels of BCAAs had a similar pattern, and those changes were most significant for valine (rising 30-40% from late G1 to G2/M). __

-They need to add the data indicating that the bat1 mutant requires longer than WT cells to reach the ~35 fL volume at which 50% of WT cells are budded.

__RESPONSE: We added all that data (new Figure EV6) and discussed it better in the text. Note that our elutriation analyses allow accurate estimates of the G1 duration, which is at least 2x longer in bat1 vs. BAT1 cells. __

-It seems important to show that Val supplementation can suppress the overabundance of G1 cells in bat1 mutant cells shown in Fig. 4C; and can restore sensitivity to Rap and Rps6-P accumulation in bat1 mutant cells (in Fig.s 5A & B).

__RESPONSE: Excellent suggestions. We now present the requested experiments. The DNA content data are in Figure 3C, and the phospho-Rps6 data in the new Figure 4D are discussed in the text. Briefly, exogenous valine, and to a lesser extent leucine, suppressed the G1 accumulation, but not to wild type levels. Exogenous valine also substantially increased TORC1 activity (>10-fold). __

-It seems important to show that Rps6-P will decline in M phase and increase during a second cell cycle to establish that TORC1 activity actually fluctuates in the cell cycle instead of just by reduced by the manipulations involved in collecting young G1 cells by elutriation.

RESPONSE: The second cycle comment is not pertinent to our elutriation setup. The two-cycle approach should be used in arrest-and-release synchronizations to minimize arrest-related artifacts when cells continue to grow in size. This is why we used elutriation in the first place, as described in the text, to avoid such artifacts. In elutriations it is the first cycle, exclusively of daughter cells, that can be meaningfully scored. After that, the cells lose synchrony very fast because you have mothers (which grow in size very little) and daughters (which need to double in size until mitosis). Hence, the second cycle will be meaningless and impossible to interpret.

Reviewer #2 (Significance (Required)):

General Assessment:

Strengths: Evidence for BCAA biosynthesis from glucose in the G1 phase of the cell cycle, and evidence obtained from analyzing the bat1 mutant that BCAA synthesis underlies activation of TORC1 early in the cell cycle in a manner required to achieve the critical cell size necessary for G1 to S transition.

Weaknesses: Lack of evidence for Val biosynthesis in G1 despite evidence that Val limitation is more crucial than Leu limitation in the bat1 mutant; lack of confirmation that Val limitation underlies the delayed G1-S transition and reduced TORC1 signaling in the bat1 mutant; and lack of compelling evidence that TORC1 activity is periodic in WT cells.

Advance: This would be the first evidence that TORC1 activity varies through the cell cycle in a manner controlled by synthesis of BCAAs

Audience: This advance would be of great interest to a wide range of workers studying how the cell cycle is regulated and the role of TORC1 in controlling cell growth and division in normal cells and in human disease.

My expertise: Mechanisms of metabolic regulation of gene expression at the transcriptional and translational levels in budding yeast

**Referees cross-commenting**

Ref. #1's major comment 1 echoes my request for clarification about whether Leu, and not just Val, is limiting growth in the bat1 mutant, and also the need to determine which BCAA supplement to bat1 cells will restore TORC1 activity (which was also requested by Ref. #3).

I agree with this reviewer's request to provide evidence that Leu levels actually increase during G1 progression (comment #2). I also think the suggested experiments in Comment #3 are reasonable for their potential to provide stronger evidence that Leu production in the G1 phase of wild-type cells activates TORC1, as currently the argument is based on the finding of low TORC1 activation in bat1 cells (that seem to be limiting for Val vs. Leu). Comment #4 echoes similar requests made by both me and Ref. #3. Ref. #3's major comments 1 and 3 mirror two of my major comments. I wasn't convinced of the need to monitor Sch9 versus Rps6 phosphorylation as a read-out of TORC1 activity-does being a direct substrate truly matter? Regarding comment 5, I wasn't convinced of the need to include Rap-sensitive or -resistant control strains for the analysis in Fig. 5A. And regarding comment 4, while it would be interesting to examine if TORC1 regulates BCAA synthesis during cell cycle progression, this seems to be outside the scope of a demonstration that BCAA synthesis stimulates TORC1.

Thus, it seems we all agree on certain experiments that need to be carried out, and Ref. #1 has rightly proposed a few others with the potential to strengthen the evidence that Leu production during G1 phase mediates cyclic activation of TORC1

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

In this manuscript, Blank and colleagues measure the synthesis of various metabolites from glucose during cell cycle progression and observe an increased synthesis of branched-chain amino acids (BCAA) from the early G1 to late G1 phase. Interestingly, they also found a gradual increase in TORC1 activity from the early G1 to the S phase which is proposed to be dependent on BCAA synthesis.

Major comments:

1. The authors show that TORC1 activity increases from the early G1 to the S phase. TORC1 activity is sensitive to short-term starvations caused during changing media or centrifugations. Hence, the concern arises regarding the increased pattern of TORC1 activity during the cell cycle. Is it really a biological phenomenon or a cellular adaptation to experimental conditions? Can authors provide more support for this observation? Can authors monitor the cell cycle for the two cell cycles to confirm that TORC1 activity shows a wavy pattern? RESPONSE: The same point was also made by Reviewer #2. As we noted in our response above, “____The second cycle comment is not pertinent to our elutriation setup. The two-cycle approach should be used in arrest-and-release synchronizations to minimize arrest-related artifacts when cells continue to grow in size. This is why we used elutriation in the first place, as described in the text, to avoid such artifacts. In elutriations it is the first cycle, exclusively of daughter cells, that can be meaningfully scored. After that, the cells lose synchrony very fast because you have mothers (which grow in size very little) and daughters (which need to double in size until mitosis). Hence, the second cycle will be meaningless and impossible to interpret.____”

The authors use Rps6 phosphorylation as a read-out of TORC1 activity, which is not a direct substrate of TORC1. Analysis of the direct substrates of TORC1, such as phosphorylation of Sch9 will solidify the author's claim.

RESPONSE: The reviewers discussed this point (see their comments above). We agree with the opinion that Rps6 phosphorylation accurately reports on TORC1 activity (also used in the fly experiments we now cite, as requested by Reviewer 1). For all our experiments' objectives and conclusions, it doesn't matter if the phosphorylation of Rps6 lies more downstream than Sch9 phosphorylation.

Authors show that Bat1 lacking strain have reduced TORC1 activity. Can authors restimulate these cells with Leucin, Valine, and Isoleucine individually or in combination to identify the critical amino acid for the TORC1 activity?

RESPONSE: Yes, that is an excellent suggestion. We show the experiment in Figure 4D (see previous response). Valine showed pronounced activation (>10-fold).

The authors claim that increased BCAA synthesis is necessary for TORC1 activation. Since TORC1 is shown to be upstream of amino acid biosynthesis pathways, it will be interesting to check if TORC1 per se regulates BCAA synthesis during cell cycle progression. The authors could inhibit TORC1 by rapamycin treatment and monitor if the BCAA synthesis still shows cell cycle-dependent modulation.

RESPONSE: The reviewers also discussed this point (see their comments above). We agree with the view that it is a very substantial undertaking, well beyond the scope of this work.

In Figure 5A, the use of any rapamycin-sensitive and rapamycin-resistant strains as controls will strengthen their claim of TORC1 inhibition being epistatic to Bat1 deletion, since the rapamycin in minimal media might be less effective.

RESPONSE: Again, the reviewers also discussed this point (see their comments above). We agree that it will not add much to the conclusions in the context of all the data we show and the existing literature.

Minor comments:

1. The data of metabolic labeling, especially various species M1, M2, M3, etc., of an individual metabolite is difficult to understand for the general readers. Hence, a schematic explaining various species might be helpful. RESPONSE: We added a new Figure (EV1) delineating the carbons from glucose to valine and leucine.

Please describe the elutriation approach in more detail with media conditions and buffer conditions to understand the overall experimental setup.

RESPONSE: We now added this information (see the second section of the Materials and Methods).

Reviewer #3 (Significance (Required)):

Significance:

Overall, this study presents an interesting observation to the researchers working in TORC1 and cell cycle regulation.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

In this manuscript, Blank and colleagues measure the synthesis of various metabolites from glucose during cell cycle progression and observe an increased synthesis of branched-chain amino acids (BCAA) from the early G1 to late G1 phase. Interestingly, they also found a gradual increase in TORC1 activity from the early G1 to the S phase which is proposed to be dependent on BCAA synthesis.

Major comments:

1. The authors show that TORC1 activity increases from the early G1 to the S phase. TORC1 activity is sensitive to short-term starvations caused during changing media or centrifugations. Hence, the concern arises regarding the increased pattern of TORC1 activity during the cell cycle. Is it really a biological phenomenon or a cellular adaptation to experimental conditions? Can authors provide more support for this observation? Can authors monitor the cell cycle for the two cell cycles to confirm that TORC1 activity shows a wavy pattern?
2. The authors use Rps6 phosphorylation as a read-out of TORC1 activity, which is not a direct substrate of TORC1. Analysis of the direct substrates of TORC1, such as phosphorylation of Sch9 will solidify the author's claim.
3. Authors show that Bat1 lacking strain have reduced TORC1 activity. Can authors restimulate these cells with Leucin, Valine, and Isoleucine individually or in combination to identify the critical amino acid for the TORC1 activity?
4. The authors claim that increased BCAA synthesis is necessary for TORC1 activation. Since TORC1 is shown to be upstream of amino acid biosynthesis pathways, it will be interesting to check if TORC1 per se regulates BCAA synthesis during cell cycle progression. The authors could inhibit TORC1 by rapamycin treatment and monitor if the BCAA synthesis still shows cell cycle-dependent modulation.
5. In Figure 5A, the use of any rapamycin-sensitive and rapamycin-resistant strains as controls will strengthen their claim of TORC1 inhibition being epistatic to Bat1 deletion, since the rapamycin in minimal media might be less effective.

Minor comments:

1. The data of metabolic labeling, especially various species M1, M2, M3, etc., of an individual metabolite is difficult to understand for the general readers. Hence, a schematic explaining various species might be helpful.
2. Please describe the elutriation approach in more detail with media conditions and buffer conditions to understand the overall experimental setup.

Significance

Overall, this study presents an interesting observation to the researchers working in TORC1 and cell cycle regulation.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

This paper provides evidence that branched chain amino acid (BCAA) in the G1 phase of the cell cycle, fueled by pyruvate generated by glucose catabolism activates cell growth and allows cells to reach the critical size required for entry into S phase by activation of TORC1 signaling. Previous work had indicated that Leucine supplementation of a bat1 bat2 mutant, lacking both enzymes that catalyze BCAA from the alpha-keto acid precursors and starved on minimal medium, led to TORC1 activation. This work is significant in suggesting that BCAA synthesis from glucose is responsible for a cyclic activation of TORC1 necessary for a normal rate of cell growth in the G1 phase of the cell cycle.

The study employs metabolic flux analysis of metabolites derived from glucose following a pulse-chase with different isotopes of glucose in synchronized early G1 cells (obtained by elutriation) throughout one cell cycle. They claim that the only compelling changes in metabolites observed as the cell cycle proceeds was a decline in pyruvate containing only one heavy 13C carbon atom and a corresponding increase in Leu (M6) with 6 heavy carbon atoms, which is interpreted to indicate Leu synthesis from pyruvate that begins in early G1 and peaks at mitosis. They show that a bat1 mutant exhibits a slow-growth phenotype that can be mitigated only by valine (although they infer similar effects for Leu and Ile that I find unconvincing) and they observed reductions in all three BCAAs in different experiments that measure steady amino acid levels in different ways (although the results are compelling only for Val). They go on to show evidence that the bat1 mutation reduces birth and mean cell size and leads to an increased proportion of G1 cells in asynchronous cultures, and they claim that bat1 cells take much longer than WT to achieve the same size found when a synchronized WT culture reaches 50% budding (although they don't show the data for this last point.) Interestingly, they find that deleting BAT1 suppresses sensitivity to the TORC1 inhibitor rapamycin (Rap), consistent with the idea that the bat1 mutation impairs TORC1 activity in the same manner as Rap and that BCAA are required to activate TORC1 in WT cells to the level that can be impaired by Rap, as summarized in the model in Fig. 5F. Consistent with this, they present evidence that the bat1 mutation reduces TORC1 signaling as judged by diminished Rps6 phosphorylation (although it was not shown that this effect could be reversed by Val addition). They also show that TORC1 signaling/Rps6-P increases as the cell cycle progresses using elutriated early G1 cells, suggesting that TORC1 activity is periodic in the cell cycle (although they don't establish this periodicity through a second cell cycle).

General critique:

The conclusion that BCAA synthesis from glucose is responsible for a cyclic activation of TORC1 necessary for a normal rate of cell growth in the G1 phase of the cell cycle is potentially of considerable significance. There are however a number of puzzling aspects of the data that seem to weaken this conclusion. As described in greater detail below, it is difficult to explain why only Leu is synthesized from glucose during the cell cycle, and why only Val shows a marked reduction in the bat1 mutant that appears to be responsible for the slow-growth phenotype. In addition, there are important controls lacking of showing that a Val supplement can suppress the G1 delay and reduction in TORC1 signaling in the bat1 mutant. In addition, the evidence that TORC1 activity is periodic in the cell cycle is lacking and it needs to be shown that Rps6-P levels are periodic through at least a second cell cycle.

Major comments:

• Why don't they observe synthesis of Ile and particularly Val in the metabolic flux experiment of Fig. 1, especially considering that only Val appears to be critically required for normal cell growth in the bat1 mutant based on the results in Fig. 3B?
• The data in Fig. 3B do not show a convincing increase in growth of the bat1 mutant with addition of Leu and Ile; and the stimulation by Val alone seems identical to that seen with Val in combination with Leu and Ile. Thus, it appears that the slow-growth of the bat1 mutant results only from reduced Val levels, not all 3 BCAAs, which is at odds with their interpretation of the data.
• they claim to see reductions in all three BCAAs in the bat1 mutant; however, no significant reduction was found for Leu in Fig. EV2, and only Val was altered by the 1.5-fold cut-off imposed on the MS metabolomics data in Fig. EV3 (which could be appreciated only by an in-depth examination of the supplementary data in File S1-the Val, Leu, and Ile dots should be labeled in Fig. EV3). In addition, the reductions in Ala and Gly showin in Fig. EV2 were not found in the MS analysis of Fig. EV3. It needs to be acknowledged that the metabolomics data show a marked reduction in the bat1 mutant only for Valine with little or no change in Leucine levels. This result is difficult to explain with the simple models shown in Fig. 3A and 5F, which requires additional comment. The authors should acknowledge the much greater effect of the bat1 mutation on Val levels versus Leu and Ile, revealed both by measuring the levels of BCAAs in the mutant and comparing the BCAAs for rescuing the slow-growth of the mutant, and explain how this can be reconciled with the results in Fig. 2 where only Leu and not Val or Ile synthesis was detected.
• They need to add the data indicating that the bat1 mutant requires longer than WT cells to reach the ~35 fL volume at which 50% of WT cells are budded.
• It seems important to show that Val supplementation can suppress the overabundance of G1 cells in bat1 mutant cells shown in Fig. 4C; and can restore sensitivity to Rap and Rps6-P accumulation in bat1 mutant cells (in Fig.s 5A & B).
• It seems important to show that Rps6-P will decline in M phase and increase during a second cell cycle to establish that TORC1 activity actually fluctuates in the cell cycle instead of just by reduced by the manipulations involved in collecting young G1 cells by elutriation.

Referees cross-commenting

Ref. #1's major comment 1 echoes my request for clarification about whether Leu, and not just Val, is limiting growth in the bat1 mutant, and also the need to determine which BCAA supplement to bat1 cells will restore TORC1 activity (which was also requested by Ref. #3). I agree with this reviewer's request to provide evidence that Leu levels actually increase during G1 progression (comment #2). I also think the suggested experiments in Comment #3 are reasonable for their potential to provide stronger evidence that Leu production in the G1 phase of wild-type cells activates TORC1, as currently the argument is based on the finding of low TORC1 activation in bat1 cells (that seem to be limiting for Val vs. Leu). Comment #4 echoes similar requests made by both me and Ref. #3. Ref. #3's major comments 1 and 3 mirror two of my major comments. I wasn't convinced of the need to monitor Sch9 versus Rps6 phosphorylation as a read-out of TORC1 activity-does being a direct substrate truly matter? Regarding comment 5, I wasn't convinced of the need to include Rap-sensitive or -resistant control strains for the analysis in Fig. 5A. And regarding comment 4, while it would be interesting to examine if TORC1 regulates BCAA synthesis during cell cycle progression, this seems to be outside the scope of a demonstration that BCAA synthesis stimulates TORC1.

Thus, it seems we all agree on certain experiments that need to be carried out, and Ref. #1 has rightly proposed a few others with the potential to strengthen the evidence that Leu production during G1 phase mediates cyclic activation of TORC1

Significance

General Assessment:

Strengths: Evidence for BCAA biosynthesis from glucose in the G1 phase of the cell cycle, and evidence obtained from analyzing the bat1 mutant that BCAA synthesis underlies activation of TORC1 early in the cell cycle in a manner required to achieve the critical cell size necessary for G1 to S transition.

Weaknesses: Lack of evidence for Val biosynthesis in G1 despite evidence that Val limitation is more crucial than Leu limitation in the bat1 mutant; lack of confirmation that Val limitation underlies the delayed G1-S transition and reduced TORC1 signaling in the bat1 mutant; and lack of compelling evidence that TORC1 activity is periodic in WT cells.

Advance: This would be the first evidence that TORC1 activity varies through the cell cycle in a manner controlled by synthesis of BCAAs

Audience: This advance would be of great interest to a wide range of workers studying how the cell cycle is regulated and the role of TORC1 in controlling cell growth and division in normal cells and in human disease.

My expertise: Mechanisms of metabolic regulation of gene expression at the transcriptional and translational levels in budding yeast

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary

In this manuscript, Blank et al. propose a link between cell-cycle dependent changes in metabolic flux and corresponding changes in TORC1 activity in yeast cells. Based on their findings, the authors propose that Bat1-dependent leucine synthesis from glucose increases as cells progress through G1 and that this activates TORC1 to drive cell cycle progression. Although the existence of cell-cycle dependent synthesis of leucine is a novel and exciting finding, several aspects of the proposed model are not sufficiently supported by experimental evidence, in particular the fact that the increase in Leu synthesis is causing the increase in TORC1 activity in late G1.

Major comments:

1. To show that the increase in Leu biosynthesis in S-phase is activating TOR, one would ideally want to blunt this increase in biosynthesis and assay TORC1 activity. Admittedly, this is difficult. So, instead, the authors study bat1- cells which have strongly impaired synthesis of BCAA including Leucine. The relevance of these bat1- cells to the proposed cell-cycle dependent model, however, is questionable for two reasons: 1) Although the authors state that "exogenous supplementation of BCAAs in all combinations suppressed the growth defect of bat1- cells, especially when valine was present", the spot assays in Figure 3 show visible rescues only when valine is present either alone or in combination, while supplementation of leucine or isoleucine does not seem to have any effect. Hence it appears that the bat1- phenotype is mainly due to limiting valine levels, not leucine levels. 2) The relevance of these results for understanding TORC1 regulation are questionable, since valine does not typically activate TORC1. Does additionn of Leu to bat1- cells increase TORC1 activity ?
2. TORC1 activity is known to depend on steady-state leucine concentrations in the cell rather than on leucine flux. Although the authors observe that the synthesis rate of leucine increases during G1 progression, this does not necessarily translate innto increased leucine concentrations in the cell. To support the claim that the increase in TORC1 activity during G1 progression depends on leucine, the authors would need to show that, not only leucine synthesis, but also overall leucine levels in the cell increase during G1 progression.
3. To test whether the increase in Leu biosynthesis in S-phase activates TORC1, a few different approaches could be tested: 1) Since leucine activates TORC1 through the Gtr proteins, the authors could test whether rendering TORC1 resistant to low leucine through expression of constitutively active Gtrs abolishes the cell-cycle dependence in TORC1 activity. 2) Leu could be added to the medium of wildtype cells in G1 to the amount necessary to cause an increase in intracellular Leu levels similar to those seen in S-phase to test whether this increases TORC1 activity.
4. In Fig 2B one sees that Leu biosynthesis peaks at 150min and then drops again. The p-RpS6 blot in Fig. 5D, however, only goes up to 140 min and shows that TORC1 activity increases up to 140 min, but it doesn't show timepoints beyond 150 min when Leu biosynthesis drops again, and hence one would expect TORC1 activity to drop. If TORC1 activity were to drop from 150min onwards, this would strengthen the correlation between Leu biosynthesis and TORC1 activity.

Minor concerns:

1. In Figure EV3, the authors should highlight some of the metabolites that are significantly changed, in particular the BCAA. The figure is not very informative as currently presented.
2. Fig 2 - are "expressed ratios" the best term for metabolite levels? Unlike genes, where such heat maps are often used, the metabolites are not 'expressed'. How about 'relative metabolite level' instead?
3. Page 8: "We also measured the MID values from the media of the same cultures used to prepare the cell extracts." Where are these data? We don't see them in File S2?
4. Fig 4B - the x-axis labeling is missing for the bat1- cells
5. Although the authors state repeatedly that they show "for the first time in any system" that TORC1 activity is dynamic in the cell cycle, similar observations have already been made before, for instance showing high mTORC1 activity in the G1/S transition in the Drosophila wing disc or low mTORC1 activity during mitosis in mammalian cells (see PMIDs 28829944, 28829945, and 31733992). The text should be amended accordingly.
6. There are two entries for valine in File S1/Sheet8. Why?

Significance

Despite the well-known effects of pharmacological or genetic manipulations of TORC1/mTORC1 on cell cycle progression, whether and how mTORC1 activity itself is physiologically coupled to cell cycle progression is still an insufficiently studied aspect. Hence this study provides an interesting link between cell-cycle dependent regulation of amino acid biosynthesis and TORC1 regulation. Importantly, the results of this study rely on centrifugal elutriation to obtain cell cycle synchronization, thus ruling out potential metabolic artifacts due to pharmacological methods. The observed changes in metabolic flux are therefore likely genuine and represent the major strength of the study. The major limitation is the lack of strong evidence supporting the notion that the increase in Leu biosynthesis at late G1 or S-phase is causing the increase in TORC1 activity.

The major advance is conceptual - that amino acid biosynthesis rates are cell-cycle dependent.

These results will be of interest to a broad audience of people studying the cell cycle, cell growth, TORC1 activity, cell metabolism and cancer.