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

Learn more at Review Commons

---

Reply to the reviewers

1. Point-by-point description of the revisions

Reviewer #1:

Evidence, reproducibility and clarity (Required):

In this manuscript Czajkowski et al explore the role of the doublecortin-family kinase ZYG-8 during meiosis in C. elegans Oocytes. First by studying available temperature-sensitive mutants and then by generating their own strain expressing ZYG-8 amenable to auxin-inducible degradation, they establish that defects in ZYG-8 lead to defects in spindle assembly, such as the formation of multipolar spindles, and spindle maintenance, in which spindles elongate, fall apart, and deform in meiosis. Based on these observations the authors conclude that ZYG-8 depletion leads to excessive outward force. As the lab had previously found that the motor protein KLP-18 generates outside directed forces in meiosis, Czajkowski et al initially speculate that ZYG-8 might regulate KLP-18. KLP-18 depletion generally leads to the formation of monopolar spindles in meiosis. Intriguingly, when the authors co-deplete ZYG-8 they find that in some cases bipolarity was reestablished. This led to the hypothesis that yet another kinesin, BMK-1, the homolog of the mammalian EG-5, could provide redundant outward directed forces to KLP-18. The authors then study the effect of ZYG-8 and KLP-18 co-depletion in a BMK-1 mutant background strain and observe that bipolarity is no longer reestablished under these conditions, suggesting that BMK-1 generates additional outward directed forces. The authors also conclude that ZYG-8 inhibits BMK-1. To follow up on this Czajkowski et al generate a ZYG-8 line that carries a mutation in the kinase domain, which should inhibit its kinase activity. This line shows similar effects in terms of spindle elongation but reduced impact on spindle integrity, reflected in minor effects on the number of spindle poles and spindle angle. The authors conclude that ZYG-8's kinase activity is required for the function of ZYG-8 in meiosis and mitosis. Overall, the paper is well written, and the data is presented very clearly and reproducible. The experiments are adequately replicated, and statistical analysis are adequate. *The observations are very interesting. However, the authors could provide some additional insight into the function of ZYG-8. This paper is strongly focused on motor generated forces within the spindle and tries to place ZYG-8 within this context, but there is compelling evidence from other studies that ZYG-8 also affects microtubule dynamics, which would have implications for spindle assembly and structure. The paper would strongly benefit from the authors exploring this role of ZYG-8 in the context of meiosis further. If the authors feel that this would extend beyond the scope of this paper, I would suggest that the authors rephrase some of their introduction and discussion to reflect the possibility that changes in microtubule growth and nucleation rates could explain some of the phenotypes (think of katanin) and effects and that therefore it can not necessarily be concluded that BMK-1 is inhibited by ZYG-8. *

We thank the reviewer for these positive comments on our manuscript and on the rigor of our data. We also thank them for the excellent suggestion to explore a potential role for microtubule dynamics. As detailed below in response to the specific points, we performed new experiments to explore this possibility, and found via FRAP analysis that there were substantial changes in microtubule dynamics upon ZYG-8 depletion. We have therefore added these new data and have re-written major parts of the manuscript to incorporate a discussion of microtubule dynamics throughout the paper (introduction, results, model, discussion). Our data now support two roles for ZYG-8 in regulating acentrosomal spindle assembly and stability - one in modulating microtubule dynamics and the other in tuning forces (either directly or indirectly). We are grateful to the reviewer for motivating us to do these experiments, as they have added a whole new angle to the manuscript and have greatly increased its impact, as we now have a fuller understanding of how ZYG-8 contributes to oocyte meiosis.

Major points:

*1.) Zyg-8, as well as the mammalian homolog DCLK-1, has been reported to play an important role for microtubule dynamics. While the introduction mentions its previously shown role in meiosis and mitosis, it is totally lacking any background on the effect on microtubule dynamics. The authors mention these findings in the discussion, but it would be helpful to incorporate this in the introduction as well. As an example, Goenczy et al 2001 demonstrated that ZYG-8 is involved in spindle positioning but also showed its ability to bind microtubules and promote microtubule assembly. Interestingly, like the authors here, Goenczy et al concluded that while the kinase domain contributes to, it is not essential ZYG-8's function. Also, Srayko et al 2005 (PMID 16054029) demonstrated that ZYG-8 depletion led to reduced microtubule growth rates and increased nucleation rates in C. elegans mitotic embryos. And in mammalian cells DCLK-1 was shown to increase microtubule nucleation rate and decrease catastrophe rate, leading to a net stabilization of microtubules (Moores et al 2006, PMID: 16957770). It would be great if the authors could add to the introduction that ZYG-8 has been suggested to affect microtubule dynamics. *

We agree that this is a great idea. As the reviewer suggested, we decided to explore the possibility that ZYG-8 impacts microtubule dynamics within the oocyte spindle. We depleted ZYG-8 and performed FRAP experiments to determine if there were effects on microtubule turnover. We found that loss of ZYG-8 caused a dramatic decrease in the spindle's ability to recover tubulin, both at the spindle center and at the spindle poles (shown in a new Figure 7). We made substantial changes to the manuscript when adding these new data - the manuscript now discusses ZYG-8's role in modulating microtubule dynamics in the introduction, results, discussion, and model (Figure 9), and we added all of the references suggested by the reviewer. We think that the manuscript is greatly improved due to these additions and changes.

*2a.) The authors initially study two different ts alleles, or484ts and b235ts. The experiments clearly show a significant increase in spindle length in both strains. However, the or484 strain had been previously studied (McNally et al 2016, PMID: 27335123), and only minor effects on spindle length were reported (8.5µm in wt metaphase and 10µm in zyg-8 (or484)). How do the authors explain these differences in ZYG-8 phenotype. Even though the ZYG-8 phenotype is consistent throughout this paper it would be good to explain why the authors observe spindle elongation, fragmentation and spindle bending in contrast to previous observations. *

The reviewer is correct that McNally et.al. (2016) noted only minor effects on spindle length and did not report observing spindle bending or pole defects. However, the images presented in their paper of spindles in the zyg-8(or484) mutant (in Figure 8B) only showed spindles after they had already shrunk in preparation for anaphase; it is possible that these spindles had pole or midspindle defects prior to this shrinking, and that the authors did not note those phenotypes because their analysis focused on anaphase. In contrast, since the goal of our study focused on how ZYG-8 impacts spindle assembly and maintenance, we looked carefully at spindle morphology and quantified a larger number of metaphase spindles (in their study, only 12 metaphase spindles were measured, since metaphase was not the focus of their manuscript). Recently (after we submitted our manuscript), another study from the McNally lab was published, where they did note metaphase defects following ZYG-8 inhibition (though they did not describe the defects in detail or explore why they happened). We now mention and cite this new paper (Li et.al., 2023) in our manuscript, to show that our findings are consistent with the work of others in the field.

*2b.) As a general note, it would be helpful if the authors could indicate if the spindles are in meiosis I or II. The only time where this is specifically mentioned is in Video 7, showing a Meiosis II spindle, which makes me assume all other data is in Meiosis I. Adding this to the figures would also help to distinguish if some of the images, i.e. Figure 1B, show multipolar spindles due to failed polar body extrusion. If this is the case then the quantification of number of poles should maybe reflect different possibilities, such as fragmented poles vs. multiple poles because two spindles form around dispersed chromatin masses. *

We agree that it is a good idea to clarify this issue. For all of our experiments, we analyzed both MI and MII spindles. However, there were no noticeable differences in phenotype between MI and MII spindles for any of our mutant/depletion conditions - we observed bent spindles, elongated spindles, and extra poles in both MI and MII following ZYG-8 inhibition. Therefore, for the quantifications presented in the manuscript (spindle length, spindle angle, number of poles), we pooled our MI and MII data. We have now added this information to the manuscript for clarity (lines 97-99 and 139-141). In addition, we have added new images to Figure 1B that show examples of MII spindles (both at the permissive and restrictive temperature), to show that the phenotypes are indistinguishable between MI and MII.

We agree with the reviewer that one of the spindles in the original Figure 1B looked like it could have resulted from failed polar body extrusion (the chromosomes appeared to be in two masses, something we did not originally notice, so theoretically each mass could have organized its own spindle). To determine if this was the case, we looked closely at the chromosomes in this image; we confirmed that there were only 6 chromosomes, and that all were bivalents (these can be distinguished from MII chromosomes based on size). Therefore, this spindle was not multipolar due to an issue with polar body extrusion. However, to prevent future confusion, we picked a different representative spindle (where the bivalents we not grouped into two masses), and we added a new column to the figure that shows the DNA channel in grayscale (so it is easier to see and count the chromosomes). We also now note in the materials and methods how we were able to distinguish between MI and MII in our experiments (chromosome count, size, presence of polar bodies), so that it is clear that none of our phenotypes result from failed polar body extrusion (lines 600-603).

*3) The authors generate a line that carries a mutation leading to a kinase dead version of ZYG-8. It would be great if the authors could further test if this version is truly kinase dead. What is interesting is that the kinase dead version the authors create has less effect on the numbers of pole than the zyg-8 (b235)ts strain, which carries a mutation in a less conserved kinase region. Overall, it seems that the phenotypes are very similar, independent on mutations in the microtubule binding area, kinase area or after AID. This could of course be due to all regions being important, i.e. microtubule binding is required for localizing kinase-activity. Generating mutant versions of the target proteins, for example here BMK-1, that can not be phosphorylated or are constitutively active as well as assessment of changes in protein phosphorylation levels in the kinase dead strain would be helpful to provide deeper insight into potential regulation of proteins by ZYG-8. *

We agree that it would be ideal to test whether the D604N mutant is truly kinase dead. However, in the interest of time, we ask to be allowed to skip that experiment. The analogous residue has been mutated in mammalian ZYG-8 (DCLK1), and has been shown to cause DCLK1 to be kinase dead in vitro; this is a highly conserved aspartic acid in the central part of the catalytic domain, so we infer that the mutation we made in ZYG-8 should be kinase dead as well. However, since we did not test this directly, we softened our language in the manuscript, explaining that we "infer" that it is kinase dead rather than stating definitively that it is. With regard to the zyg-8(b235)ts mutant having a stronger phenotype, we think that it is possible that this mutation destabilizes a larger portion of the protein (rather than just affecting the catalytic activity), since the phenotypes in this mutant are similar to depletion of the protein in the ZYG-8 AID strain. Therefore, we think that our D604N mutant reveals new information about the role of kinase activity, since it is a more specific mutation that should likely only affect catalytic activity and not the rest of the protein (based on the previous work on DCLK1).

While we appreciate the suggestion from the reviewer to generate mutant versions of potential target proteins, we ask that this be considered beyond the scope of the study. Now that we know that ZYG-8 not only affects forces within the spindle (maybe BMK-1) but also microtubule dynamics, there are many potential targets - it would require a lot of work to figure out what the relevant targets are. Instead of exploring this experimentally in this manuscript, we added a new section to the discussion where we speculate on what some of these targets could be, to motivate future studies.

*4a) The authors state that "BMK-1 provides redundant outward force to KLP18". Redundancy usually suggests that one protein can take over the function of another one when the other is not there. In these scenarios a phenotype is often only visible when both proteins are depleted as each can take over the function of the other one. Here however the situation seems slightly different, as depletion of BMK-1 has no phenotype while depletion of KLP-18 leads to monopolar spindles. If BMK-1 would normally provide outward directed forces, would this not be visible in KLP-18 depleted oocytes if they were truly redundant? I assume the authors hypothesize that ZYG-8 inhibits BMK-1 and thus it can not generate outward directed forces. In this case, do the authors envision that ZYG-8 inhibits BMK-1 prior to or in metaphase or only in anaphase or throughout meiosis? Do they speculate, that BMK-1 is inhibited in anaphase and only active in metaphase? *

The reviewer makes an excellent point - we agree that we should not use the word "redundant" in this context, so we have removed this phrasing from the manuscript. We hypothesize that BMK-1 can provide outward forces during spindle assembly but is not capable of providing as much force as KLP-18 (the primary force-generating motor). We infer this based on our experiments where we co-deplete KLP-18 and ZYG-8 (using long-term depletion). Although BMK-1 is presumably activated under these conditions, it is not able to restore spindle bipolarity (there are outward forces generated, which results in minus ends being found at the periphery of the monopolar spindle, but spindles are not bipolar).

Therefore, BMK-1 is not able to fully replace the function of KLP-18 during spindle assembly. Interestingly, our experiments imply that BMK-1 can better substitute for KLP-18 later on (when ZYG-8 is inhibited); when we remove ZYG-8 from formed monopolar spindles, bipolarity can be restored (an activity dependent on BMK-1). These findings suggest that ZYG-8 plays a more important role in suppressing BMK-1 activity after the spindle forms, to prevent spindle overelongation in metaphase. We have edited the manuscript to better explain these points.

*4b) In addition, Figure S4 somewhat argues against a role for ZYG-8 in regulating BMK-1. ZYG-8 depletion supposedly leads to increased outward forces due to loss of BMK-1 inhibition, thus co-depletion of ZYG-8 with BMK-1 should rescue the increased spindle size at least to some extent, however neither increase in spindle length nor increase in additional spindle pole formation are prevented by co-depletion of BMK-1 suggesting that BMK-1 is not generating the forces leading to spindle length increase. Thus, arguing that after all ZYG-8 does not regulate BMK-1. This should be discussed further in the paper and the authors should consider changing the title. At this point the provided evidence that ZYG-8 is regulating motor activity is not strong enough to make this claim. *

The reviewer is correct that Figure S4 shows that the effects of depleting ZYG-8 on bipolar spindles (spindle elongation and pole/midspindle defects) cannot solely be explained by a role for ZYG-8 in regulating BMK-1 - this was the point that we were trying to make when we included this data in the original manuscript. However, we previously did not know what this other role could be, and therefore we only speculated on other potential roles in the discussion. Now that we have done FRAP experiments and have found that ZYG-8 also affects microtubule dynamics in the oocyte spindle, we now have a better explanation for the data presented in Figure S4 - it makes sense that deleting BMK-1 would not rescue the effects of ZYG-8 depletion, since we have evidence that ZYG-8 also regulates microtubule dynamics. We now clearly explain this in the revised manuscript and we have changed the title to make it clear that ZYG-8 plays multiple roles in oocytes.

*5) The authors are proposing that ZYG-8 regulates/ inhibits BMK-1, however convincing evidence for an inhibition is not provided in my opinion and the effect of ZYG-8 on BMK-1 could be indirect. To make a compelling argument for a regulation of BMK-1 the authors would have to investigate if ZYG-8 interacts and/ or phosphorylates BMK-1 (see 7) and if this affects its dynamics. In addition, given the reported role of ZYG-8 on microtubule dynamics it would be very important that the authors consider studying the effect of ZYG-8 degradation on microtubule dynamics. Tracking of EBP-2 would be good, however this is very difficult to do inside meiotic spindles due to their small size. In addition, the authors could maybe consider some FRAP experiments, which could provide insights into microtubule dynamics and motions, which could be indicative of outward directed forces/ sliding. *

We thank the reviewer for these comments as they motivated us to explore a role for ZYG-8 in modulating microtubule dynamics. The reviewer is correct that tracking EBP-2 in the very small meiotic spindle is not possible due to technical limitations, so we took the suggestion to perform FRAP. These experiments revealed that microtubule turnover in the spindle is greatly slowed following ZYG-8 depletion, suggesting a global stabilization of microtubules (data presented in a new Figure 7). This change in dynamics could contribute to the observed spindle phenotypes, which we now explain in detail in the manuscript. Given these new findings, we also now note that the effects we see on BMK-1 activity could be indirect (i.e. maybe increasing the stability of microtubules allows motors to exert excess forces). We now clearly discuss these various possibilities in the discussion.

Summary: Additional requested experiments:

• Interaction/ phosphorylation of BMK-1 by ZYG-8, i.e. changes of BMK-1 phosphorylation in absence of ZYG-8, BMK-1 mutations that may prevent phosphorylation by ZYG-8.
• Assessment of microtubule dynamics (EBP-2, FRAP, length in monopolar spindles...)
• Kinase activity of the kinase dead ZYG-8 strain (OPTIONAL) We assessed the role of ZYG-8 in microtubule dynamics (bullet point #2). Because this new analysis revealed that ZYG-8 plays multiple roles in the spindle, we decided not to further investigate whether ZYG-8 phosphorylates BMK-1, since the manuscript now no longer argues that this is ZYG-8's major function. We also did not assess the kinase activity of the D604N mutant since this has been done previously for DCLK1, and instead we softened the language in manuscript when describing this mutant.

Minor points:

*1) In Figure 4C it seems that the ZYG-8 AID line as well as the zyg-8 (or848)ts already have a phenotype (increased ASPM-1 foci) in absence of auxin/ at the permissive temperature. Does this suggest that the ZYG-8 AID as well as the zyg-8 (or848) strains are after all slightly defective (even if Figure 1, S1 and S2 argue otherwise) and thus more responsive to the loss of KLP-18? *

The reviewer is correct that the ZYG-8 AID strain (without auxin) and zyg-8(or848)ts strain (at the permissive temperature) are slightly defective in the klp-18(RNAi) monopolar spindle assay. To more rigorously determine whether these strains were also defective in other assays, we generated new graphs comparing the spindle lengths and angles of the two temperature sensitive strains at the permissive temperature to wild-type (N2) worms. These data are now shown in Figure S1 (new panels F and G). A comparison of our ZYG-8 AID strain to a control strain (both in the absence of auxin) are shown in Figure S2 (panels C and D). In this analysis, there wasn't a significant difference for either of these comparisons (i.e. the spindle lengths and angles were all equivalent). We do not know why these strains appear to be slightly defective in the monopolar spindle assay, though perhaps this assay is more sensitive and can detect very mild defects in protein function.

*2) The authors observe that in preformed monopolar spindles degradation of ZYG-8 can sometimes restore bipolarity. This observation is very interesting but why do the authors not observe a similar phenotype in long-term ZYG-8 AID; klp-18 (RNAi) or zyg-8(or484)ts; klp-18(RNAi). In the latter conditions bipolarity does not seem to occur at all. Do the authors think this is due to differences in timing of events? *

We thank the reviewer for highlighting this point. We do think that our data suggest that ZYG-8 plays a more important role maintaining the spindle that it does in spindle formation; we have now more clearly explained this in the manuscript (detailing the differences in phenotypes we observe when we deplete ZYG-8 prior to spindle assembly or after the spindle has already formed, lines 180-189 and 227-231). To emphasize this point further, we have also included a graph in Figure S3G that directly compares the number of poles per spindle in long-term auxin treated spindles to short-term auxin treated spindles (with and without metaphase arrest).

*3) Based on the Cavin-Meza 2022 paper it looks like depletion of KLP-18 in a BMK-1 mutant background does not look different from klp-18 (RNAi) alone. However, looking at Video 8, it looks like spindles "shrink" in absence of KLP-18 and BMK-1. Or is this due to any effects from the ZYG-8 AID strain? This can also be seen in Video 9. *

The reviewer highlights a fair point that was not clearly explained in our manuscript. In normal monopolar anaphase, chromosomes move in towards the center pole as the spindle gets smaller (C. elegans oocyte spindles shrink in both bipolar and monopolar anaphase); this was previously described in Muscat et.al. 2015, and, as the reviewer noted, in Cavin-Meza et.al. 2022 (in a strain with the bmk-1 mutation). We see this same monopolar anaphase behavior in the ZYG-8 AID strain (Figure 6). We have now better explained normal monopolar anaphase progression and we have cited the Muscat et.al. paper in the relevant sections of the manuscript (lines 221-223 and 714-717).

*4) Line 311: " ZYG-8 loads onto the spindle along with BMK-1, and functions to inhibit BMK-1 from over elongating microtubules during metaphase." Maybe this sentence could be re-phrased as it currently sounds like BMK-1 elongates (polymerizes) microtubules. *

In re-writing the manuscript and emphasizing that there are multiple for ZYG-8 (in addition to regulating forces within the spindle), we removed this sentence.

*5) Line 313: "Intriguingly, in C. elegans oocytes and mitotically-dividing embryos, BMK-1 inhibition causes faster spindle elongation during anaphase, suggesting that BMK-1 normally functions as a brake to slow spindle elongation (Saunders et al., 2007; Laband et al., 2017). Further, ZYG-8 has been shown to be required for spindle elongation during anaphase B (McNally et al., 2016). Our findings may provide an explanation for this phenotype, since if ZYG-8 inhibits BMK-1 as we propose, then following ZYG-8 depletion, BMK-1 could be hyperactive, slowing anaphase B spindle elongation." This paragraph could be modified for better clarity. It is not clear how the findings of the authors, BMK-1 provides outward force but is normally inhibited by ZYG-8, align with the last sentence saying "following zyg-8 depletion, BMK-1 could be hyperactive slowing anaphase B spindle elongation", should it not increase elongation according to the authors observations? *

In re-writing the manuscript to incorporate our new data showing that ZYG-8 plays a role in modulating microtubule dynamics, we also re-wrote this discussion so that there would be less emphasis on the potential connection between ZYG-8 and BMK-1. In making these edits to expand the focus of the manuscript, we removed this section of the discussion.

Reviewer #1 (Significance (Required)): *In this manuscript Czajkowski et al explore the role of the doublecortin-family kinase ZYG-8 during meiosis in C. elegans Oocytes. The authors conclude that BMK-1 generates outward directed force, redundant to forces generated by KLP-18, and that ZYG-8 inhibits BMK-1. The authors conclude that ZYG-8's kinase activity is required for the function of ZYG-8 in meiosis and mitosis. This research is interesting and provides some novel insight into the role of ZYG-8. In particular the observed spindle elongation and subsequent spindle fragmentation are novel and had not yet been reported. Also, the observation that degradation of ZYG-8 in monopolar klp-18(RNAi) spindles can restore bipolarity is novel and interesting, as well as the observation that this is somewhat dependent on the presence of BMK-1. This will be of interest to a broad audience and provides some new insight into the role of importance of ZYG-8 and BMK-1. The limitation of the study is the interpretation of the results and the lack of solid evidence that the observed phenotypes are due to ZYG-8 regulation motor activity, as the title claims. To support this some more experiments would be required. In addition, ZYG-8 has been reported to affect microtubule dynamics, which can certainly affect the action of motors on microtubules. This line of research is not explored in the paper but would certainly add to its value.

Field of expertise: Research in cell division *

We thank the reviewer for their positive comments on the impact and novelty of our findings. We hope that the additional experiments we performed and the revisions we made to the text thoroughly address the reviewer's concerns and that they deem the revised manuscript ready for publication.

Reviewer #2 (Evidence, reproducibility and clarity (Required)): *In this manuscript, the authors explore the requirement for doublecortin kinase Zyg8 in C elegans oocytes. Oocytes build meiotic spindles in the absence of centrosomes, and therefore unique regulation occurs during this process. Therefore, how spindles are built and its later stability are an area of active investigation in the field. Using mutant alleles of Zyg8 and auxin-induced degron alleles, the authors demonstrate that this kinase is required to negatively regulate outward pole forces through BMK1 kinesin and that it has other functions to still explore. Overall, I find that this study takes an elegant genetic approach to tackling this important question in oocyte biology. I have some comments to consider for making the MS clear to a reader. *

We thank the reviewer for these positive comments on our approach and the importance of our research question. We have attempted to address all of the reviewer's suggestions and we think that they increase the clarity of the manuscript.

Major Comments:

*1.) Although I like the graphs describing the altered angles of the spindles, it falls short in fully assessing the phenotype in a meaningful statistical way. Could the authors also graph the data to show statistical significance in the angles between conditions? Perhaps by grouping them into angle ranges and performing an Anova test? This is important in Figure 2E where it is not obvious that there is a difference. *

The reviewer makes a good point - we have now addressed this concern by performing ANOVA tests to compare conditions on each of the angle graphs. Results of these tests have been reported in the corresponding figure legends. This analysis has confirmed all of the statements we made in the original manuscript. In Figure 1D and S1D, spindle angles were significantly different in the zyg-8 temperature sensitive mutants at the restrictive temperature, and in Figure 2, the angles were significantly different between the "minus auxin" and "plus auxin" conditions. This differs from Figure 7, where there was no significant difference in spindle angle between control spindles and kinase dead mutant spindles (p-value >0.1).

*2.) The authors do not discuss the significance of the altered spindle angles which I think is an interesting phenotype. Would this be a problem upon Anaphase onset? What is known about spindle angle and aneuploidy or cell viability? Has this phenotype been described before in oocytes or somatic cells? Does depletion of other kinesin motors cause this? *

The reviewer brings up a good point that warrants more discussion in the manuscript. We agree that the angled spindles are an interesting phenotype; we believe that they could be a result of the spindle elongating to a point where the spindle center becomes weakened, suggesting that the severity of the angle is representative of the severity of spindle elongation. Alternatively, the angled spindles could be a result of the loss of spindle stability factors, such as the doublecortin domain of ZYG-8. This domain is known to have microtubule binding activity; this could be required to maintain stable crosslinked microtubules in the spindle center, such that when ZYG-8 is depleted, the spindle more easily comes apart as the spindle elongates. We now discuss these possibilities in the revised manuscript.

To the reviewer's second point, we did not examine anaphase outcomes in our manuscript. However, this was recently explored by another lab (in a study that was published after we submitted our manuscript). This study showed that spindles lacking ZYG-8 were able to initiate anaphase and segregate chromosomes (McNally et.al., 2023, https://doi.org/10.1371/journal.pgen.1011090). Perhaps when the spindle shrinks at anaphase onset, the spindle is able to reorganize and largely correct the angle defect, enabling bi-directional chromosome segregation. Interestingly, however, McNally et.al. did report conditions under which spindle bending in anaphase resulted in polar body extrusion errors. The authors reported that BMK-1, which is known to act as a brake to prevent spindle oveelongation in anaphase, is required to prevent bent spindles during anaphase by resisting the forces of cortical myosin on the spindle. Thus, there is precedence for the idea that spindle needs to remain straight throughout anaphase, to ensure proper chromosome segregation.

*3.) How is embryo spindle positioning determined? It is not clear from the images that there is a defect so I'm not sure what to look for. Is there a way to quantify this? *

In the original manuscript, spindle positioning within the embryo was determined qualitatively by eye, which we agree was not a precise measure. To address the reviewer's comment, we re-analyzed our images and assessed the position of the spindle within each embryo quantitatively - these data are now shown in Figure 8H and Figure S2B. Spindle position was quantified by analyzing images using Imaris software. The center of the spindle was set by creating a Surface of the DNA signal, and finding the center of that signal. The cell center was determined by measuring the length of the embryo along the long axis and the width of the embryo along the short axis, and setting the center as the halfway point of the total length and width of the embryo. Distance from spindle center to cell center was then measured and graphed. This quantification confirmed the claims we made in our original manuscript - both auxin-treated ZYG-8 AID spindles and ZYG-8 kinase-dead mitotic spindles were significantly mispositioned. The details of how we performed this quantification have been added to the materials and methods.

*4.) In Figure 1, it appears that there are 2 spindles. Are these MI and MII spindles or ectopic spindles? How do the authors know which one to measure? *

We thank the reviewer for pointing this out. Reviewer 1 had a similar comment, and we now understand that using that image was misleading, as it looked like as if were two separate MII spindles formed following a failed polar body extrusion event. We have gone through all of our images to stage the oocytes by looking at their chromosome morphology (i.e., to distinguish MI and MII) - the image in question had 6 bivalents and was therefore in Meiosis I; we think that this was a single spindle where the chromosomes happened to cluster into two masses. However, to prevent further confusion, we have replaced this image with a different representative image. In spindles like this with multiple poles, we measure the dominant axis of the spindle (if there are multiple poles, we pick the most prominent ones for the angle measurement). For additional details please see our response to Reviewer 1 major point #2b.

*5.) The authors show depletion of Zyg8 by western (long) and loss of Gfp (long and short), but don't do so for the acute treatment. I'm guessing this is because the Gfp tag is taken by the spindle marker. The authors should either demonstrate or explain how they know that the acute depletion is effective in removing Zyg8 protein. *

The reviewer makes a valid point. However, we are unable to see ZYG-8 depletion via acute auxin treatment using live imaging, as ZYG-8 localization is too dim and diffuse to see on the spindle using our typical live imaging parameters (we attempted to do this in a version of the ZYG-8 AID strain that has mCherry::tubulin and GFP::ZYG-8, so that there was no other spindle protein tagged with GFP). To see any GFP::ZYG-8 signal, we had to increase the laser power and exposure time well above what we typically use for live imaging - in doing this, we noticed that there was a limit to how high we could go before the cell began dying during the imaging time course, evident by a lack of chromosome movement, lack of tubulin turnover, and a general increase in tubulin signal throughout the cytoplasm. We do believe that ZYG-8 is being depleted using acute auxin treatment, however, as we see spindle defects very quickly upon dissection of the oocytes into auxin - we just unfortunately don't have a good way of quantifying this given these technical limitations. We have now added information to the materials and methods noting that we cannot see GFP::ZYG-8 under our live imaging conditions (lines 552-561), so that the reader better understands this caveat.

*6.) In video 2, the chromosome signal is dimmer in the auxin treatment compared to video 1. Why is this? Is it just an experimental artifact or is there something significant about this? If it is because of video choice, consider replacing this one. *

We thank the reviewer for their keen observations. The chromosome signal being dimmer in the auxin treatment is an experimental artifact - the brightness of the signal can vary depending on how far the spindle is from the slide (this can vary from video to video, and can also change over the time course of one video if the spindle moves during filming). Because of this, movies taken at the same intensity and exposure conditions may appear to have varying levels of brightness. So that readers of the manuscript can better see the chromosomes in this video, we have brightened the chromosome channel in this movie and noted this in the materials and methods (lines 549-551).

7.) Please consider color palette changes for color-blind readership.

We agree that it is important to present data in a way that can be appreciated by color-blind readers. Although we would prefer not to have to alter every image in our paper at this point, we have provided all important individual channels in grey scale. We are also planning to adopt a change in color palette for future papers.

Reviewer #2 (Significance (Required)):

*The strengths of this manuscript include use of multiple genetic approaches to establish temporal requirements of ZYG8 and which pathway it is acting through. Additionally, the videos and images make the phenotypes clear to evaluate. A minor limitation is that we don't know if the ZYG8 and BMK1 genetic interaction is a direct phosphorylation or not. This MS is an advancement to the field of spindle building and stability, and is particularly relevant to human oocyte quality and fertility. Previous work has shown that human oocyte spindles are highly unstable, but it is challenging to dissect genetic interactions and to conduct mechanistic studies in human oocytes. Therefore, the work here, although conducted in a nematode, can shed light on mechanism as to why human oocyte spindles are unstable and associated with high aneuploidy rates. Based on my expertise in mammalian oocyte biology, I am confident that work presented here will be of high interest to people in the field of meiotic spindle building, aneuploidy and fertility. It also will have broader interest to folks in the areas of kinesin biology, general microtubule and spindle biology. *

We thank the reviewer for these positive comments on the strength of our data and the significance of our findings reported in our original manuscript. We think that the improvements that we have made in response to suggestions from all three reviewers has further increased this impact.

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

*Summary: The focus of this paper is the function of a relatively understudied (at least in meiosis) kinase in acentrosomal spindle assembly (Zyg-8, or DCLK1 in mammals) in C. elegans oocytes. The authors use existing ts alleles and a newly generated GFP-Auxin fusion protein, and find that the ts alleles and auxin degron have similar phenotypes. They also examine the interaction with two related kinesins, KLP-18 and BMK-1 in order to investigate the mechanism behind the zyg-8 mutant phenotype. One can probably debate the significance and focus of their conclusions (force balance on the spindle). However, this is an important study because its the first on the meiotic function of a ZYG-8 kinase, and it may open the way to further studies of this kinase and how it regulates multiple kinesins and meiotic spindle assembly. *

We thank the reviewer for these positive comments and for pointing out the potential future impacts of our work. In revising the manuscript, we have broadened the focus of the manuscript - we no longer solely focus on force balance within the spindle. Thus, our revisions have substantially increased the significance and impact of our work, since the manuscript is no longer narrowly focused.

Major points:

*1.) The main concern is the focus that the main defect in zyg-8 depleted oocytes is on outward forces (eg line 134, 277, but many other places in Results and Discussion). The arguments in favor (eg line 269-271) are reasonable. However, these data are not conclusive, and do not rule out regulation of other motor activities, such as bundling, depolymerization or chromosome movement. These are complex phenotypes, and a kinase could have multiple targets and there are often multiple interpretations. This is briefly alluded to in line 372-373 but the authors could do more. Spindle length changes could be caused by different rates of depolymerization or polymerization at the poles or chromosomes. Its not clear how poleward force regulation explains the multiple pole phenotype, although a lack of central spindle integrity could do that. In most of the Results and Discussion, it is not clearly stated on what structures these outward forces are acting. Are these forces effecting kinetochore associated microtubules, or antiparallel overlap microtubules? What do the authors mean by proper force balance? Figure 8 suggests the defect is associated with the amount of overlap and force among antiparallel microtubules - that the forces effected are from the sliding of these microtubules. *

We agree that our original manuscript was too narrowly focused on the idea of force balance and that we did not discuss other potential roles for ZYG-8 in enough detail (except for briefly in our discussion). In response to both this comment and to a suggestion by Reviewer #1, we decided to investigate a potential role for ZYG-8 in modulating microtubule dynamics (which could be another explanation for some of the phenotypes we observed). We performed FRAP to measure the rate of tubulin turnover within the spindle near the center and at the poles. Interestingly, these experiments revealed that loss of ZYG-8 slows the rate of tubulin turnover, suggesting a general stabilization of microtubules. Thus, we have re-written our manuscript to clearly explain that ZYG-8 plays multiple roles in oocyte spindles - with these changes throughout the manuscript (in the introduction, results, and discussion), the paper is now no longer focused primarily on forces. We hypothesize in the discussion that the phenotypes we observe could be a combination of the effects on microtubule dynamics and spindle forces; if microtubules become more stable and motors produce excess outward forces, this may cause stress on the spindle structure that could cause the midspindle to bend and the poles to split (lines 379-382). We also now more clearly explain that the effect ZYG-8 has on spindle forces could be either direct or indirect (e.g., ZYG-8 could directly regulate motors or, by affecting the microtubule tracks themselves, it could affect their ability to exert forces). As for which population of microtubules are affected, we hypothesize that the excess forces act primarily on overlapping antiparallel microtubules (these microtubules run laterally alongside chromosomes in this system), as is represented in the model figure (Figure 9); we attempted to more clearly explain this in the re-written manuscript.

*2.) Based on differences between the long term and short term knockdown phenotypes, the authors suggest ZYG-8 is more important for spindle maintenance. For example, in line 299 the authors note that there is a more severe phenotypes with zyg-8 removed from pre-formed spindles. The authors could improve the presentation of this to allow the reader to appreciate this observation. The data is spread between Figures 2 and 3 without a direct comparison of the data. One solution would be to graph the data (eg # of poles) together in one graph and indicate if there is statistical significance. In the Discussion, the authors could refer to specific figure panels. *

The reviewer is correct that our data suggests that ZYG-8 is more important for spindle maintenance than it is for assembly. As suggested, we made a graph that includes all the pole data from Figures 2 and 3 (long-term auxin, short-term auxin, and metaphase-arrested short-term auxin) - this is now shown in Figure S3D. This makes it easier for the reader to compare these data and appreciate this point. In addition, we added text to the results section, to more clearly explain our rationale for thinking that ZYG-8 plays a more prominent role in spindle maintenance than in assembly (lines 180-189 and 227-231).

*3.) What is the practical difference between acute and short term depletion. Does acute show weaker phenotypes because there is more residual protein? Unfortunately, the effectiveness of Auxin treatment does not appear to be measured for acute or short term. If the acute depletion adds little to Figure 3, or is not much different than long term, then its not clear what it adds to the paper. Later, in Figure 6, why is only short term and acute analyzed. In general, the authors need to provide better rationale for the different auxin conditions, particularly acute and short term (eg. line 135). If they don't add anything, they should consider not presenting them because readers may get confused by the different conditions, why they were done, and what is learned from each one. *

The reviewer brings up a fair point that we agree requires clarification. Descriptions of the different types of auxin experiments is provided in Figure 1A. Long-term AID depletes proteins overnight, so the protein of interest is already missing from the oocyte when the spindle begins to form - this allows us to assess whether the protein is required for spindle assembly. However, to determine if a protein is required to stabilize pre-formed spindles, we need to remove the protein quickly after the spindle forms (using either acute or short-term AID). Acute AID is performed by dissecting oocytes directly into auxin-containing media; this allows us to watch what happens to the spindle live, as the protein is being depleted. However, one limitation is that we can only film for a short time before the oocytes begin to die (oocytes become unhappy with extended light exposure, so we cut off the videos after 15 minutes or so, to ensure that we are not filming past the point where they begin to arrest or die). Therefore, to assess what happens to spindles beyond this point, we perform short-term auxin treatment, where whole worms are soaked in auxin containing solution for 30-45 minutes and then the oocytes are dissected for immunofluorescence; this technique allows us to look at what happens to the spindle after more extended protein depletion (since we are not limited to the 10-15 minute window of filming). We have now clarified this in the manuscript by adding these details to the materials and methods. Unfortunately, it is not technically possible to quantify the extent of protein depletion in acute AID via western blotting since we would not be able to easily collect enough dissected oocytes to make a protein sample. (It is also technically challenging to quantify this via imaging; see our response to Reviewer #2, point #5). However, we assume that we are depleting ZYG-8 since we see dramatic spindle defects immediately upon dissection into auxin.

Minor points:

*3.) I am a little confused about imaging for GFP::tubulin in auxin experiments. Doesn't the ZYG-8 protein also have GFP? Should this be visible in controls? Is it measurable in the experiments? *

The reviewer is correct that the ZYG-8 protein is also tagged with GFP in the GFP::tubulin; mCherry::histone live imaging experiments. However, we found that the GFP::ZYG-8 signal is undetectable using the live imaging conditions we are using. We determined this by analyzing a version of the ZYG-8 AID strain in which tubulin was tagged with mCherry (and thus the only GFP-tagged protein was ZYG-8). Using the same live imaging parameters we use for our movies of GFP::tubulin (same exposure time, laser power, etc), we did not detect any GFP::ZYG-8. We have now added this information to the materials and methods (lines 552-561) to clarify these points for the reader, to prevent further confusion.

*4.) It is nice that the authors validated the results in an emb-30 background with unarrested oocytes. The authors note that the wild-type oocytes undergo anaphase (line 150). The images seem to suggest the auxin treated oocytes do not. Can the authors comment on anaphase in the depletion experiments. Even better, would be to comment on the accuracy of chromosome alignment and segregation. If zyg-8 mutant oocytes complete meiosis, is there any aneuploidy? These are important questions because otherwise the defects in zyg-8 mutants have less significance. *

We thank the reviewer for their comment. Previous work on ZYG-8 in C. elegans examined a role for ZYG-8 in anaphase and showed that this protein is required for anaphase B spindle elongation (McNally et.al. 2016); because this was known when we launched our study, we purposely did not extensively study ZYG-8 in anaphase and instead focused on understanding how ZYG-8 contributes to spindle formation and stability. Our fixed imaging long-term AID experiments revealed that spindles were able to go through anaphase and segregate chromosomes bidirectionally despite the metaphase spindle phenotypes, consistent with this previous work (McNally et.al. 2016) and with another recent paper from the same lab (McNally et.al. 2023). However, we did not examine whether there were chromosome segregation errors. Given that anaphase is not the focus of our paper, we ask that this be deemed beyond the scope of our study.

5.) Later, in line 184, the authors indicate that zyg-8 bipolar spindles "segregate chromosomes". Which images show anaphase I? As noted above, a limitation of these studies is not knowing the outcome of meiosis in these Zyg-8 depletions.

We agree that in the original manuscript it was difficult to see that chromosomes were segregating bidirectionally in our movies and in the still timepoint images presented in Figure 5. Therefore, we brightened the chromosome channel in the relevant videos to make it easier to see the segregating chromosomes. Video 6 shows an oocyte in Meiosis II, as the first polar body can be seen near the spindle in this movie. At 2 minutes, the monopolar spindle becomes bipolar and begins to shrink as it goes into anaphase. Chromosomes begin to move apart and then the spindle elongates. At 11 minutes, you can see that the chromosomes have segregated bidirectionally. Thus, when monopolar spindles reorganize into bipolar spindles under these conditions, they can drive bidirectional chromosome segregation. We did not assess the fidelity of chromosome segregation under these conditions (i.e., whether chromosomes segregated accurately), as the question we were trying to answer in this experiment was whether outward forces sufficient to re-establish bipolarity could be activated upon ZYG-8 depletion (as explained above in response to point #4, we focused our study on trying to understand the effects of ZYG-8 depletion on the spindle, rather than on anaphase). We agree that analyzing anaphase outcomes would be interesting, but we ask that it be considered beyond the scope of this study.

*6.) Line 206 suggests that ZYG-8 inhibits BMK-1. Is a simple explanation that BMK-1 is required for the bipolar spindles observed in the klp-18 zyg-8 AID oocytes? *

Yes, the reviewer is correct that BMK-1 is required for the generation of bipolar spindles in the klp-18(RNAi) ZYG-8 AID conditions. In the original manuscript we extrapolated this result to propose that ZYG-8 regulates BMK-1. However, this comment, as well as feedback from the other reviewers and our new experiments (showing that ZYG-8 also modulates microtubule dynamics) has made us re-think the way we discuss this result, as we now agree that it does not prove this regulation (it is only suggestive). Therefore, in the revised manuscript, we no longer definitely claim that ZYG-8 regulates BMK-1 - we have switched to softer language (stating that ZYG-8 "may regulate" BMK-1, etc.). In the results section we now describe our conclusions as follows: "These data demonstrate that BMK-1 produces the outward forces that are activated upon ZYG-8 and KLP-18 co-depletion and raise the possibility that ZYG-8 regulates BMK-1 either directly or indirectly" (lines 250-252).

*7.) Given that many mitotic and meiotic kinases are localized to specific regions or domains of the spindle, there is only limited discussion of the ZYG-8 localization pattern. Does the ZYG-8 localization pattern provide any insights into its mechanism of promoting spindle assembly? *

The reviewer makes a good point - while we did report ZYG-8 localization, the discussion on the importance of its localization pattern was limited. To address this, we now remind readers in the discussion that ZYG-8 and BMK-1 co-localize throughout meiosis, consistent with the possibility that ZYG-8 could regulate BMK-1. Notably, this localization pattern is also consistent with the observation that ZYG-8 modulates microtubule dynamics across the spindle; this is now also noted in the discussion (lines 358-361).

*8.) Line 96-97 - how much is the ZYG-8 depletion? *

To address this question, we have quantified the amount of ZYG-8 protein in our ZYG-8 AID strain in control, long-term, and short-term auxin treated conditions. The western blot was quantified by comparing the raw intensity of the bands and subtracting the background signal. Short-term auxin depletion resulted in an ~63% reduction in ZYG-8 GFP signal, and long-term depletion resulted in an ~93% reduction in ZYG-8 GFP signal. This has now been reported in the manuscript on lines 785-786.

*9.) Line 140: the authors say spindle length could not be measured, but perhaps it makes more sense to measure half spindle (chromosome to spindle pole). The images do give the impression that the chromosome to pole distance is shorter. *

While we liked this idea and tried to perform these measurements, it turned out to be difficult in practice, since the spindle length measurements are obtained by finding the distance from pole (center of the ASPM-1 staining) to center of the chromosome signal. If you look carefully at our images you will notice that the chromosomes lose alignment following short-term AID; therefore, the chromosomes do not form one mass, which made it very difficult to determine an accurate "center" of the DNA signal. Additionally, in most cases the poles are disrupted such that ASPM-1 is found in many separate masses and/or is diffusely localized around the periphery of the spindle. Because of this, we unfortunately felt that these measurements would not be very accurate and would be hard to interpret.

*10.) Don't see the point of lines 323-330. Could be deleted? *

In revising our manuscript, we have rephrased these lines in an attempt to provide more context. Because DCLK1 has been shown to be upregulated in a wide variety of cancers, there are ongoing efforts to find chemical inhibitors that specifically block the kinase activity of this protein to be used as cancer therapeutics. However, no one has previously shown that the kinase activity of DCLK1 is important for its in vivo function (in any organism). Therefore, we were trying to make the point that, since we demonstrated that kinase activity is important for the functions of a DCLK1 family member in vivo, this suggests that these kinase inhibitors may in fact be beneficial in knocking down DCLK1 activity.

11.) Figure 1: Because ts alleles could have a defective phenotype at "permissive" temperature, a wild-type control should be included. This data does appear in a later figure.

The reviewer is correct that this data does appear in a later figure, but we agree this direct comparison would provide clarity to the reader. To address this comment, we compared the spindle lengths and angles of the two temperature-sensitive (TS) strains (at both the permissive and restrictive temperatures) to wild-type (N2) worms - these data have been added to Figure S1 (new panels F and G). The spindle lengths of both TS strains at the permissive temperature did not significantly differ from wild-type spindle lengths (p>0.1), while both TS strains at the restrictive temperature were significantly different than wild-type (p0.1), but there was a significant difference between wild-type spindles and the TS mutants at the restrictive temperature (doublecortin domain mutant (p Reviewer #3 (Significance (Required)): The strengths of this paper are the novelty of studying Zyg-8. It also addresses important questions regarding acentrosmal spindle assembly in oocytes. The weakness is mostly in the limited interpretation of results and not enough consideration of alternative interpretations. Related to this, the authors only test the force balance hypothesis with the knockout of two related kinesins. They don't experimentally investigate other mechanisms for the zyg-8 phenotype. This research should be of broad interest to anyone interested in oocyte spindle assembly, and also in a more specialized way to those who study kinases or Zyg-8 homologs in other cell types or organisms.

We thank the reviewer for these positive comments on the strengths and novelty of our manuscript. We also appreciate the constructive suggestions of all three reviewers, which motivated us to perform new experiments that revealed additional functions for ZYG-8 - these revisions have greatly improved the manuscript and have broadened its impact.

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 focus of this paper is the function of a relatively understudied (at least in meiosis) kinase in acentrosomal spindle assembly (Zyg-8, or DCLK1 in mammals) in C. elegans oocytes. The authors use existing ts alleles and a newly generated GFP-Auxin fusion protein, and find that the ts alleles and auxin degron have similar phenotypes. They also examine the interaction with two related kinesins, KLP-18 and BMK-1 in order to investigate the mechanism behind the zyg-8 mutant phenotype. One can probably debate the significance and focus of their conclusions (force balance on the spindle). However, this is an important study because its the first on the meiotic function of a ZYG-8 kinase, and it may open the way to further studies of this kinase and how it regulates multiple kinesins and meiotic spindle assembly.

Major:

1) The main concern is the focus that the main defect in zyg-8 depleted oocytes is on outward forces (eg line 134, 277, but many other places in Results and Discussion). The arguments in favor (eg line 269-271) are reasonable. However, these data are not conclusive, and do not rule out regulation of other motor activities, such as bundling, depolymerization or chromosome movement. These are complex phenotypes, and a kinase could have multiple targets and there are often multiple interpretations. This is briefly alluded to in line 372-373 but the authors could do more. Spindle length changes could be caused by different rates of depolymerization or polymerization at the poles or chromosomes. Its not clear how poleward force regulation explains the multiple pole phenotype, although a lack of central spindle integrity could do that. In most of the Results and Discussion, it is not clearly stated on what structures these outward forces are acting. Are these forces effecting kinetochore associated microtubules, or antiparallel overlap microtubules? What do the authors mean by proper force balance? Figure 8 suggests the defect is associated with the amount of overlap and force among antiparallel microtubules - that the forces effected are from the sliding of these microtubules.

2) Based on differences between the long term and short term knockdown phenotypes, the authors suggest ZYG-8 is more important for spindle maintenance. For example, in line 299 the authors note that there is a more severe phenotypes with zyg-8 removed from pre-formed spindles. The authors could improve the presentation of this to allow the reader to appreciate this observation. The data is spread between Figures 2 and 3 without a direct comparison of the data. One solution would be to graph the data (eg # of poles) together in one graph and indicate if there is statistical significance. In the Discussion, the authors could refer to specific figure panels.

3) What is the practical difference between acute and short term depletion. Does acute show weaker phenotypes because there is more residual protein? Unfortunately, the effectiveness of Auxin treatment does not appear to be measured for acute or short term. If the acute depletion adds little to Figure 3, or is not much different than long term, then its not clear what it adds to the paper. Later, in Figure 6, why is only short term and acute analyzed. In general, the authors need to provide better rationale for the different auxin conditions, particularly acute and short term (eg. line 135). If they don't add anything, they should consider not presenting them because readers may get confused by the different conditions, why they were done, and what is learned from each one.

Minor:

1) I am a little confused about imaging for GFP::tubulin in auxin experiments. Doesn't the ZYG-8 protein also have GFP? Should this be visible in controls? Is it measurable in the experiments?

2) It is nice that the authors validated the results in an emb-30 background with unarrested oocytes. The authors note that the wild-type oocytes undergo anaphase (line 150). The images seem to suggest the auxin treated oocytes do not. Can the authors comment on anaphase in the depletion experiments. Even better, would be to comment on the accuracy of chromosome alignment and segregation. If zyg-8 mutant oocytes complete meiosis, is there any aneuploidy? These are important questions because otherwise the defects in zyg-8 mutants have less significance.

3) Later, in line 184, the authors indicate that zyg-8 bipolar spindles "segregate chromosomes". Which images show anaphase I? As noted above, a limitation of these studies is not knowing the outcome of meiosis in these Zyg-8 depletions.

4) Line 206 suggests that ZYG-8 inhibits BMK-1. Is a simple explanation that BMK-1 is required for the bipolar spindles observed in the klp-18 zyg-8 AID oocytes?

5) Given that many mitotic and meiotic kinases are localized to specific regions or domains of the spindle, there is only limited discussion of the ZYG-8 localization pattern. Does the ZYG-8 localization pattern provide any insights into its mechanism of promoting spindle assembly?

6) Line 96-97 - how much is the ZYG-8 depletion?

7) Line 140: the authors say spindle length could not be measured, but perhaps it makes more sense to measure half spindle (chromosome to spindle pole). The images do give the impression that the chromosome to pole distance is shorter.

8) Don't see the point of lines 323-330. Could be deleted?

9) Figure 1: Because ts alleles could have a defective phenotype at "permissive" temperature, a wild-type control should be included. This data does appear in a later figure.

Significance

The strengths of this paper are the novelty of studying Zyg-8. It also addresses important questions regarding acentrosmal spindle assembly in oocytes. The weakness is mostly in the limited interpretation of results and not enough consideration of alternative interpretations. Related to this, the authors only test the force balance hypothesis with the knockout of two related kinesins. They don't experimentally investigate other mechanisms for the zyg-8 phenotype. This research should be of broad interest to anyone interested in oocyte spindle assembly, and also in a more specialized way to those who study kinases or Zyg-8 homologs in other cell types or organisms.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

In this manuscript, the authors explore the requirement for doublecortin kinase Zyg8 in C elegans oocytes. Oocytes build meiotic spindles in the absence of centrosomes, and therefore unique regulation occurs during this process. Therefore, how spindles are built and its later stability are an area of active investigation in the field. Using mutant alleles of Zyg8 and auxin-induced degron alleles, the authors demonstrate that this kinase is required to negatively regulate outward pole forces through BMK1 kinesin and that it has other functions to still explore. Overall, I find that this study takes an elegant genetic approach to tackling this important question in oocyte biology. I have some comments to consider for making the MS clear to a reader.

Major Comments:

1. Although I like the graphs describing the altered angles of the spindles, it falls short in fully assessing the phenotype in a meaningful statistical way. Could the authors also graph the data to show statistical significance in the angles between conditions? Perhaps by grouping them into angle ranges and performing an Anova test? This is important in Figure 2E where it is not obvious that there is a difference.

2. The authors do not discuss the significance of the altered spindle angles which I think is an interesting phenotype. Would this be a problem upon Anaphase onset? What is known about spindle angle and aneuploidy or cell viability? Has this phenotype been described before in oocytes or somatic cells? Does depletion of other kinesin motors cause this?

3. How is embryo spindle positioning determined? It is not clear from the images that there is a defect so I'm not sure what to look for. Is there a way to quantify this?

4. In Figure 1, it appears that there are 2 spindles. Are these MI and MII spindles or ectopic spindles? How do the authors know which one to measure?

5. The authors show depletion of Zyg8 by western (long) and loss of Gfp (long and short), but don't do so for the acute treatment. I'm guessing this is because the Gfp tag is taken by the spindle marker. The authors should either demonstrate or explain how they know that the acute depletion is effective in removing Zyg8 protein.

6. In video 2, the chromosome signal is dimmer in the auxin treatment compared to video 1. Why is this? Is it just an experimental artifact or is there something significant about this? If it is because of video choice, consider replacing this one.

7. Please consider color palette changes for color-blind readership.

Significance

The strengths of this manuscript include use of multiple genetic approaches to establish temporal requirements of ZYG8 and which pathway it is acting through. Additionally, the videos and images make the phenotypes clear to evaluate. A minor limitation is that we don't know if the ZYG8 and BMK1 genetic interaction is a direct phosphorylation or not.

This MS is an advancement to the field of spindle building and stability, and is particularly relevant to human oocyte quality and fertility. Previous work has shown that human oocyte spindles are highly unstable, but it is challenging to dissect genetic interactions and to conduct mechanistic studies in human oocytes. Therefore, the work here, although conducted in a nematode, can shed light on mechanism as to why human oocyte spindles are unstable and associated with high aneuploidy rates.

Based on my expertise in mammalian oocyte biology, I am confident that work presented here will be of high interest to people in the field of meiotic spindle building, aneuploidy and fertility. It also will have broader interest to folks in the areas of kinesin biology, general microtubule and spindle biology.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In this manuscript Czajkowski et al explore the role of the doublecortin-family kinase ZYG-8 during meiosis in C. elegans Oocytes. First by studying available temperature-sensitive mutants and then by generating their own strain expressing ZYG-8 amenable to auxin-inducible degradation, they establish that defects in ZYG-8 lead to defects in spindle assembly, such as the formation of multipolar spindles, and spindle maintenance, in which spindles elongate, fall apart, and deform in meiosis. Based on these observations the authors conclude that ZYG-8 depletion leads to excessive outward force. As the lab had previously found that the motor protein KLP-18 generates outside directed forces in meiosis, Czajkowski et al initially speculate that ZYG-8 might regulate KLP-18. KLP-18 depletion generally leads to the formation of monopolar spindles in meiosis. Intriguingly, when the authors co-deplete ZYG-8 they find that in some cases bipolarity was reestablished. This led to the hypothesis that yet another kinesin, BMK-1, the homolog of the mammalian EG-5, could provide redundant outward directed forces to KLP-18. The authors then study the effect of ZYG-8 and KLP-18 co-depletion in a BMK-1 mutant background strain and observe that bipolarity is no longer reestablished under these conditions, suggesting that BMK-1 generates additional outward directed forces. The authors also conclude that ZYG-8 inhibits BMK-1. To follow up on this Czajkowski et al generate a ZYG-8 line that carries a mutation in the kinase domain, which should inhibit its kinase activity. This line shows similar effects in terms of spindle elongation but reduced impact on spindle integrity, reflected in minor effects on the number of spindle poles and spindle angle. The authors conclude that ZYG-8's kinase activity is required for the function of ZYG-8 in meiosis and mitosis.

• Overall, the paper is well written, and the data is presented very clearly and reproducible. The experiments are adequately replicated, and statistical analysis are adequate. The observations are very interesting. However, the authors could provide some additional insight into the function of ZYG-8. This paper is strongly focused on motor generated forces within the spindle and tries to place ZYG-8 within this context, but there is compelling evidence from other studies that ZYG-8 also affects microtubule dynamics, which would have implications for spindle assembly and structure. The paper would strongly benefit from the authors exploring this role of ZYG-8 in the context of meiosis further. If the authors feel that this would extend beyond the scope of this paper, I would suggest that the authors rephrase some of their introduction and discussion to reflect the possibility that changes in microtubule growth and nucleation rates could explain some of the phenotypes (think of katanin) and effects and that therefore it can not necessarily be concluded that BMK-1 is inhibited by ZYG-8.

Major comments:

1) Zyg-8, as well as the mammalian homolog DCLK-1, has been reported to play an important role for microtubule dynamics. While the introduction mentions its previously shown role in meiosis and mitosis, it is totally lacking any background on the effect on microtubule dynamics. The authors mention these findings in the discussion, but it would be helpful to incorporate this in the introduction as well. As an example, Goenczy et al 2001 demonstrated that ZYG-8 is involved in spindle positioning but also showed its ability to bind microtubules and promote microtubule assembly. Interestingly, like the authors here, Goenczy et al concluded that while the kinase domain contributes to, it is not essential ZYG-8's function. Also, Srayko et al 2005 (PMID 16054029) demonstrated that ZYG-8 depletion led to reduced microtubule growth rates and increased nucleation rates in C. elegans mitotic embryos. And in mammalian cells DCLK-1 was shown to increase microtubule nucleation rate and decrease catastrophe rate, leading to a net stabilization of microtubules (Moores et al 2006, PMID: 16957770).

It would be great if the authors could add to the introduction that ZYG-8 has been suggested to affect microtubule dynamics.

2) The authors initially study two different ts alleles, or484ts and b235ts. The experiments clearly show a significant increase in spindle length in both strains. However, the or484 strain had been previously studied (McNally et al 2016, PMID: 27335123), and only minor effects on spindle length were reported (8.5µm in wt metaphase and 10µm in zyg-8 (or484)). How do the authors explain these differences in ZYG-8 phenotype. Even though the ZYG-8 phenotype is consistent throughout this paper it would be good to explain why the authors observe spindle elongation, fragmentation and spindle bending in contrast to previous observations.

As a general note, it would be helpful if the authors could indicate if the spindles are in meiosis I or II. The only time where this is specifically mentioned is in Video 7, showing a Meiosis II spindle, which makes me assume all other data is in Meiosis I. Adding this to the figures would also help to distinguish if some of the images, i.e. Figure 1B, show multipolar spindles due to failed polar body extrusion. If this is the case then the quantification of number of poles should maybe reflect different possibilities, such as fragmented poles vs. multiple poles because two spindles form around dispersed chromatin masses.

3) The authors generate a line that carries a mutation leading to a kinase dead version of ZYG-8. It would be great if the authors could further test if this version is truly kinase dead. What is interesting is that the kinase dead version the authors create has less effect on the numbers of pole than the zyg-8 (b235)ts strain, which carries a mutation in a less conserved kinase region. Overall, it seems that the phenotypes are very similar, independent on mutations in the microtubule binding area, kinase area or after AID. This could of course be due to all regions being important, i.e. microtubule binding is required for localizing kinase-activity. Generating mutant versions of the target proteins, for example here BMK-1, that can not be phosphorylated or are constitutively active as well as assessment of changes in protein phosphorylation levels in the kinase dead strain would be helpful to provide deeper insight into potential regulation of proteins by ZYG-8.

4) The authors state that "BMK-1 provides redundant outward force to KLP18". Redundancy usually suggests that one protein can take over the function of another one when the other is not there. In these scenarios a phenotype is often only visible when both proteins are depleted as each can take over the function of the other one. Here however the situation seems slightly different, as depletion of BMK-1 has no phenotype while depletion of KLP-18 leads to monopolar spindles. If BMK-1 would normally provide outward directed forces, would this not be visible in KLP-18 depleted oocytes if they were truly redundant? I assume the authors hypothesize that ZYG-8 inhibits BMK-1 and thus it can not generate outward directed forces. In this case, do the authors envision that ZYG-8 inhibits BMK-1 prior to or in metaphase or only in anaphase or throughout meiosis? Do they speculate, that BMK-1 is inhibited in anaphase and only active in metaphase? In addition, Figure S4 somewhat argues against a role for ZYG-8 in regulating BMK-1. ZYG-8 depletion supposedly leads to increased outward forces due to loss of BMK-1 inhibition, thus co-depletion of ZYG-8 with BMK-1 should rescue the increased spindle size at least to some extent, however neither increase in spindle length nor increase in additional spindle pole formation are prevented by co-depletion of BMK-1 suggesting that BMK-1 is not generating the forces leading to spindle length increase. Thus, arguing that after all ZYG-8 does not regulate BMK-1. This should be discussed further in the paper and the authors should consider changing the title. At this point the provided evidence that ZYG-8 is regulating motor activity is not strong enough to make this claim.

5) The authors are proposing that ZYG-8 regulates/ inhibits BMK-1, however convincing evidence for an inhibition is not provided in my opinion and the effect of ZYG-8 on BMK-1 could be indirect. To make a compelling argument for a regulation of BMK-1 the authors would have to investigate if ZYG-8 interacts and/ or phosphorylates BMK-1 (see 7) and if this affects its dynamics. In addition, given the reported role of ZYG-8 on microtubule dynamics it would be very important that the authors consider studying the effect of ZYG-8 degradation on microtubule dynamics. Tracking of EBP-2 would be good, however this is very difficult to do inside meiotic spindles due to their small size. In addition, the authors could maybe consider some FRAP experiments, which could provide insights into microtubule dynamics and motions, which could be indicative of outward directed forces/ sliding.

Summary:

Additional requested experiments:

• Interaction/ phosphorylation of BMK-1 by ZYG-8, i.e. changes of BMK-1 phosphorylation in absence of ZYG-8, BMK-1 mutations that may prevent phosphorylation by ZYG-8. -Assessment of microtubule dynamics (EBP-2, FRAP, length in monopolar spindles...) -Kinase activity of the kinase dead ZYG-8 strain (OPTIONAL)

Minor comments:

1) In Figure 4C it seems that the ZYG-8 AID line as well as the zyg-8 (or848)ts already have a phenotype (increased ASPM-1 foci) in absence of auxin/ at the permissive temperature. Does this suggest that the ZYG-8 AID as well as the zyg-8 (or848) strains are after all slightly defective (even if Figure 1, S1 and S2 argue otherwise) and thus more responsive to the loss of KLP-18?

2) The authors observe that in preformed monopolar spindles degradation of ZYG-8 can sometimes restore bipolarity. This observation is very interesting but why do the authors not observe a similar phenotype in long-term ZYG-8 AID; klp-18 (RNAi) or zyg-8(or484)ts; klp-18(RNAi). In the latter conditions bipolarity does not seem to occur at all. Do the authors think this is due to differences in timing of events?

3) Based on the Cavin-Meza 2022 paper it looks like depletion of KLP-18 in a BMK-1 mutant background does not look different from klp-18 (RNAi) alone. However, looking at Video 8, it looks like spindles "shrink" in absence of KLP-18 and BMK-1. Or is this due to any effects from the ZYG-8 AID strain? This can also be seen in Video 9.

4) Line 311: " ZYG-8 loads onto the spindle along with BMK-1, and functions to inhibit BMK-1 from over elongating microtubules during metaphase." Maybe this sentence could be re-phrased as it currently sounds like BMK-1 elongates (polymerizes) microtubules.

5) Line 313: "Intriguingly, in C. elegans oocytes and mitotically-dividing embryos, BMK-1 inhibition causes faster spindle elongation during anaphase, suggesting that BMK-1 normally functions as a brake to slow spindle elongation (Saunders et al., 2007; Laband et al., 2017). Further, ZYG-8 has been shown to be required for spindle elongation during anaphase B (McNally et al., 2016). Our findings may provide an explanation for this phenotype, since if ZYG-8 inhibits BMK-1 as we propose, then following ZYG-8 depletion, BMK-1 could be hyperactive, slowing anaphase B spindle elongation." This paragraph could be modified for better clarity. It is not clear how the findings of the authors, BMK-1 provides outward force but is normally inhibited by ZYG-8, align with the last sentence saying "following zyg-8 depletion, BMK-1 could be hyperactive slowing anaphase B spindle elongation", should it not increase elongation according to the authors observations?

Significance

In this manuscript Czajkowski et al explore the role of the doublecortin-family kinase ZYG-8 during meiosis in C. elegans Oocytes. The authors conclude that BMK-1 generates outward directed force, redundant to forces generated by KLP-18, and that ZYG-8 inhibits BMK-1. The authors conclude that ZYG-8's kinase activity is required for the function of ZYG-8 in meiosis and mitosis. This research is interesting and provides some novel insight into the role of ZYG-8. Inm particular the observed spindle elongation and subsequent spindle fragmentation are novel and had not yet been reported. Also, the observation that degradation of ZYG-8 in monopolar klp-18(RNAi) spindles can restore bipolarity is novel and interesting, as well as the observation that this is somewhat dependent on the presence of BMK-1. This will be of interest to a broad audience and provides some new insight into the role of importance of ZYG-8 and BMK-1. The limitation of the study is the interpretation of the results and the lack of solid evidence that the observed phenotypes are due to ZYG-8 regulation motor activity, as the title claims. To support this some more experiments would be required. In addition, ZYG-8 has been reported to affect microtubule dynamics, which can certainly affect the action of motors on microtubules. This line of research is not explored in the paper but would certainly add to its value.

Field of expertise: Research in cell division

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

Learn more at Review Commons

---

Reply to the reviewers

The authors sincerely appreciate the editors’ and the reviewers’ dedication in providing constructive and insightful comments aimed at enhancing the quality of the manuscript. In response to the valuable feedback received, we have implemented significant revisions to the manuscript, including the addition of key experiments, reorganization of the figures as well as providing detailed point-to-point responses to address the reviewers’ concerns. With these changes, we are confident that we have effectively addressed the comments raised by all three reviewers and have strengthened the overall quality of the manuscript.

Below are the major improvements we have made in the revised manuscript:

1. Figure 4  new figure with polysome profiling assay to strengthen the link between translational regulation and mitochondrial defects.
2. Figure 7  added confocal images showing the transfer of mitochondria into recipient cells.
3. Figure S2  added RER data further supporting a shift of metabolism to favor fatty acid oxidation as shown by proteomics data.
4. Figure S4  added WB data showing that protein degradation was not affected, strengthening a protein synthesis defect due to Fam210a KO.
5. Figure S5B, S6C  added quantification to the staining and blots.

1. Point-by-point description of the revisions

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

In the manuscript entitled "FAM210A mediates an inter-organelle crosstalk essential for protein synthesis and muscle growth in mouse", Chen et al, found that knocking out of FAM210A specifically in muscle using Myl Cre resulted in abnormal mitochondria, hyperacetylation of cytosolic proteins, and translation defects. The manuscript uncovered the new functions of FAM210A in regulating metabolism and translation. I have the following the concerns about the manuscript.

Comments

One of the major phenotypes of FAM210A is the decrease of muscle mass after 6 weeks after birth. Is this phenotype caused by the accumulation of progressive loss of muscle mass from birth? Are the body weight and muscle mass reduced in FAM210A knocking out new-born mice? Is the muscle mass growth curve the same in FAM210A and WT mice from birth to 6 weeks after birth? These results will reveal more mechanism of FAM210A mediated muscle mass control. Answer: Indeed, the phenotype of the Fam210aMKO was caused by the progressive loss of muscle mass. The body weight of the mice was not different before 3-weeks of age (Figure 2B). We reasoned that myonuclei accretion occurred before Myl1Cre induced knockout of Fam210a, accounting for the relative normal muscle development and nuclei accretion prior to 21 days after birth (refer to Response Figure 2). However, due to the small muscle mass, it is hard to accurately evaluate whether the muscle mass in very young mice. Regardless, we believe that body weight and muscle weight closely mimic each other and exhibit similar slopes in WT and KO mice (Response Figure 1).

Beyond 21 days, muscle growth is mainly attributed to hypertrophy of myofibers, a process that relies on protein synthesis. Yet the Fam210aMKO myofibers has defects in protein synthesis, explaining why the muscles cannot gain weight after 3 weeks and started to lose weight. We have shown that at 4 weeks the TA muscle weight was 13 mg in Fam210aMKO compared to 25 mg in WT control. At 6-weeks, the TA weight in the Fam210aMKO mouse was 10 mg compared to 28 mg in the WT control. Furthermore, the TA weight of the Fam210aMKO mouse was 8.7 mg compared to 36mg in the WT control. These results provide compelling evidence that the Fam210aMKO muscles are progressively wasted.

Response Figure 1. Changes of body weights and TA muscle weights during postnatal growth. The muscle weights increased (in wildtype mice) or decreased (in KO mice) with body weights at similar trends.

Does the muscle mass continue to decrease after 8 weeks?

Answer: Based on the trend (see Response Figure 1), we believe the answer is “yes”. However, we were not allowed to monitor the Fam210aMKO mice after 8 weeks of age, as they were severely lethargic and can barely move, reaching the humane endpoint determined by the IACUC guidelines.

FAM210A knockout mice displayed high lethal rate. Is there any potential mechanism for the high lethality?

Answer: We performed extensive necropsy and could not identify a direct cause. The potential cause for the lethality could be the difficulty of breathing as the diaphragm muscle was very thin in the Fam210aMKO mouse compared to the WT control. Besides, the diminished muscle contraction force (Figure 3) might have prohibited normal activities (including eating), leading to exhaustive death.

In Figure 2, the muscle mass decreased significantly, while the fat mass only decreased slightly in FAM210A knockout mice. However, the ratio of the lean mass and fat mass to body mass did not change in FAM210A knockout mice compared to WT mice. How do the authors reconcile this?

Answer: Just to clarify, Figure 2D-E shows that fat mass was significantly reduced at 4-week old but not reduced at 6-week old. We interpret the significant reduction of the mass but not the ratio (to body weight) as the result of the concomitant reduction of the body weight in the Fam210aMKO mice.

Are there changes of the number of nuclei per myotube? Is the muscle atrophy in FAM210A knockout mice caused by the defects of fusion, or the degradation of protein, or both?

Answer: We thank the reviewer for this question. To answer this question, we isolated myofibers from WT and Fam210aMKO mouse at 4-week-old and quantify the myonuclei number. We did not observe a significant reduction of myonuclei number per myofiber in the Fam210aMKO mouse, suggesting that the myoblast fusion into myofibers was not affected in the Fam210aMKO model. (Response Figure 2)

Response Figure 2. DAPI staining and quantification in the single myofiber isolated from WT and Fam210aMKO mice.

The number of myonuclei in the WT and Fam210aMKO was not different, suggesting normal fusion of satellite cells in Fam210aMKO mice.

We also did western blot to check the atrophy related protein expression in WT and Fam210aMKO mouse at different ages. Interestingly, we did not observe a significant induction of these proteins (Atrogin-1, MuRF1) in the Fam210aMKOmuscle. Therefore, we conclude that the muscle atrophy was due to protein translation defects in the Fam210aMKO, independent of myoblast fusion and protein degradation (Figure S4C).

Are the growth curves of muscle mass growth in EDL and SOL the same in FAM210A knockout mice?

Answer: We thank the reviewer for the question. In the Myl1Cre mediated Fam210a KO model, Fam210a was deleted in both fast (EDL) and slow (SOL) muscles (see response to Reviewer 3, second point). We think that the “growth curve” of the EDL and SOL muscle should be same (stagnant and even reduced) upon Fam210a KO as the mouse grows from 4-week to 8-week.

The oxygen consumption and carbon dioxide production are higher in FAM210A knockout mice, suggesting a high metabolism rate. In contrast, the heat production of FAM210A knockout mice is lower, suggesting a low metabolism rate. Any explanation?

Answer: The VCO2 and VO2 values were normalized to the body weight, and the KO value appeared high because their body weights were much lower at the time of test. While for heat production (unit: Kcal/hr), body weight was not a factor in the calculation. The seemingly contradicting/surprising result that a weak KO mouse could have higher VCO2 and VO2could be recapitulated in other mouse models (for example PMID: 22307625).

Given the high glucose consumption in FAM210A, why is the clearance rate of blood glucose low?

Answer: We believe there is a misunderstanding here. A smaller AUC (as seen in the KO) suggest faster blood glucose clearance. The circulating glucose level after fasting is lower in the KO mice, which suggests that the Fam210aMKO mice were consuming more glucose compared to the WT mice. In the GTT test, the Fam210aMKO mice showed a lower AUC after the injection of glucose, implying that the Fam210aMKO mice cleared the injected glucose at a faster rate, probably due to a pseudo-fasting state which would promote the uptake of circulating glucose when available.

Are there any changes of the abilities for the FAM210A knockout mice in running endurance?

Answer: Indeed, the Fam210aMKO mice ran less distance, shorter time, and at a lower speed when tested on a treadmill endurance running program (Figure 3)

In page 5, the last sentence of the 2nd paragraph, the authors concluded "There results suggest that Fam210aMKO induces a metabolic switch to a more oxidative state." It is better to describe it as muscle metabolic since the whole-body metabolism has not been carefully examined.

Answer: We thank the reviewer for pointing this out, we will change the wording to better reflect the changes observed in the Fam210aMKO mouse regarding the metabolism.

In Fig. 6, what is the link between increased transcription level of Fgf21 and the elevated level of aberrant acetylation of proteins?

Answer: We thank the reviewer for this interesting question! However, we did not pursue a direct causal relationship between Fgf21 level and aberrant protein acetylation. In our model, we are proposing that mitochondrial defects in the Fam210aMKO model can trigger the integrated stress response which leads to a higher Fgf21 transcript level in the muscle. This is coinciding with the acetylation increase in the muscle due to the excessive production of acetyl-CoA. A potential relationship between Fgf21 and protein acetylation warrant examination in a future study.

After careful considerations on the mechanism proposed in the study, we decided to remove qPCR data showing the modest increase of Fgf21 mRNA level. The removal of this data will not change the conclusions we draw nor lessen the significance of the mitochondria transfer experiment.

Is there any link between the increased acetylation level of rebolsome proteins and the translation defects?

Answer: Indeed, there are ample studies showing that ribosomal proteins can be acetylated, and that the acetylation of ribosomal proteins can affect the protein synthesis process, for example in PMID: 35604121 and PMID: 37742082. Here in this paper, we showed by ribosome profiling assay that the muscle has defects in the polysome formation (at 4-week and 6-week), when the protein acetylation was significantly increased in the Fam210aMKO mice (Figure 4D-4G).

How do the abnormal mitochondria lead to increased protein acetylation? And how do these defects further cause translation problem?

Answer: As elaborated in the discussion, we propose that upon Fam210a KO in mature myofiber, the TCA cycle in the mitochondria was disrupted, blocking utilization of acetyl-CoA and resulting in the accumulation of acetyl-CoA in the muscle. The excess acetyl-CoA lead to increased protein acetylation in the cytosol. We identified that ribosomal proteins are hyperacetylated in the muscle. We also observed that the polysome formation in the muscle was impaired, which exacerbates the translation efficiency.

Consistently, when we treated C2C12 during in vitro culture with sodium acetate to mimic the increase of acetylation of proteins, we showed that excessive levels of acetyl-CoA can block the differentiation of C2C12 cells (Response Figure 3).

Response Figure 3. The effect of sodium acetate on the differentiation of C2C12 myoblasts.

The differentiation of C2C12 myoblasts into myotubes were probed by the protein abundance of Myog and MF20, which showed a decrease in the expression level when sodium acetate was added in increasing amounts.

The defects in translation will cause general problems besides mitochondria defects. Are there any phenotypes related to the overall translation inhibition observed? If not, why?

Answer: Just to clarify, our model suggests that mitochondrial defects in the Fam210a KO causes cytosolic translation defects, not the other way around. We showed by SUnSET experiment that the global translation was indeed reduced in the Fam210aMKO muscle at 4-week. We also observed that the p-S6 level which indicates the global protein translation was decreased. It is also true that the global translational arrest can exacerbate the mitochondrial defects and fewer mitochondrial proteins can be synthesized. This feed forward loop can explain the aggravating phenotype in the Fam210aMKO mouse as the mouse gets older.

Are the abnormal mitochondria, increased protein acetylation, and translation inhibition observed in 2-6 weeks old mice? When were these defects first found? Are they correlated with muscle atrophy?

Answer: At 2-week-old, the protein synthesis or degradation was not changed between WT and Fam210aMKO mice (Figure S4C). The mitochondria abnormality was first observed at 4 weeks of age, concomitant with the decrease of protein translation (decreased p-S6), polysome formation, and protein hyperacetylation. The acetylation increase was apparent at 6-week together with decreased p-S6 level, polysome assembly and mitochondrial defects. Decreased protein translation has been shown to cause muscle atrophy (PMID: 19046572).

Reviewer #1 (Significance (Required)):

This manuscript described many interesting phenotypes of Fam210a knockout mice. However, the links between these phenotypes are obscure. The logic of the manuscript will be greatly improved if the authors could provide explanations to logically link the phenotypes.

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

Summary: In this manuscript, Chen et al., investigate the functions of FAM210A in skeletal muscle physiology and metabolism. FAM210A is a mitochondria-localized protein in which mutations have been associated with sarcopenia and osteoporosis. Using publicly available gene expression datasets from human skeletal muscle biopsies the authors first demonstrate that the expression of FAM210 is reduced in muscle atrophy-associated diseases and increased in muscle hypertrophy conditions. Based on this, they show that a muscle specific Fam210a deletion leads to muscle atrophy/weakness, systemic metabolic defects, and premature lethality in mouse. Further examination of the knockout myofibers reveals impaired mitochondrial respiration and translation program. Additionally, the authors demonstrate that the flow of TCA cycle is disrupted in the FAM210A-deleted myofibers, which causes abnormal accumulation of acetyl-coA and hyperacetylation of a subset of proteins. The authors claim that Fam210a deletion in skeletal muscle induces the hyper-acetylation of several small ribosomal proteins that leads to ribosomal disassembly and translational deficiency. However, this conclusion is not supported by adequate experimentation and rigorous analysis of ribosomal proteins acetylation and ribosome assembly.

Major comments:

-In general, figure legends are lacking information regarding number of biological replicates used and details about statistical analysis. What does three * vs. one * mean in terms of p-value? Exact p-values should be indicated.

Answer: We thank the reviewer for pointing this out, we have added the information to the revised figure legends.

-The mechanistic studies linking muscle phenotypes with ribosomal protein hyperacetylation and mRNA translation defects are underdeveloped and not rigorously carried.

Answer: We agree with the reviewer and have added new data in the revised manuscript to strengthen this link. For example, we have now provided direct evidence on the defective polysome assembly in the Fam210a KO muscles (Figure 4D-4G), which should profoundly impact mRNA translation. In addition, other groups have also shown that ribosomal protein acetylation can impact mRNA translation and polysome formation (PMID: 35604121).

We also explored the effect of acetylation on differentiation (a process accompanied by extensive protein synthesis) related to our mouse model. We used sodium acetate to elevate acetylation during C2C12 differentiation. We found that increased acetylation indeed impaired the differentiation as can be seen by the reduced expression of MF20 (myosin protein) by WB and IF. The differentiation marker Myogenin was also reduced (Response Figure 3, 4).

Response Figure 4. Immunofluorescence staining of Myog and MF20 in the differentiated C2C12 myotubes treated with different amounts of sodium acetate.

The number of MF20 (green) positive myotubes and Myog (red) positive nuclei was significantly reduced in the cells treated with 15mM and 30mM sodium acetate.

-Fig S1: The validation WB of FAM210A KO is not the most convincing. Why are the FAM210A levels so low in TA compared to other tissues?

Answer: This is due to the insufficient proteins loaded as it was obvious from the Tubulin marker. We have replaced the WB blot with more convincing blots as requested (Figure S1C).

-Fig 2G: The authors state "Hematoxylin and eosin (H&E) staining did not reveal any obvious myofiber pathology in the Fam210a KO mice up to 8 weeks". However there seems to be a progressive increase in nuclei up to 8-weeks in the KO. What is the significance of this?

Answer: Thank you for pointing this out. We have now changed the wording and quantified the myonuclei number per myofiber. The increase of myonuclei in the H&E images is likely due to the smaller myofiber size in the Fam210aMKOmouse compared to the WT (Response Figure 5).

Response Figure 5. Quantification of the myonuclei number in the H&E images.

-IP-MS analysis for FAM210A interacting proteins requires validation with IP and reverse IP + WB experiment.

Answer: We did perform the co-IP with SUCLG2 and FAM210A antibodies to try to confirm the interaction. To be more specific, we transduced C2C12 myoblasts cells with an Fam210a overexpression virus and differentiated the cells for 3 days. The myotubes were used to test the interaction by pulling down Fam210a with a myc antibody (FAM210A has a myc tag) and blot with SUCLG2 antibody. Unfortunately, the results were not promising (Response Figure 6). We reasoned that the interaction might be indirect or too transient to be reliably detected.

Response Figure 6. co-IP of SUCLG2 and FAM210A.

• Figure 4A requires quantification of the SDH signals from multiple samples.

Answer: We thank the reviewer for this suggestion. We have added the quantification of the staining (Figure S5B).

• Figure 6F: To clearly demonstrate an increase in protein acetylation in the FAM210 MKO, the authors must provide quantification data generated with more then N=1. Please add the molecular weights markings on the side of the blots.

Answer: We thank the reviewer for this suggestion, we have provided the quantification of the Acetylated-lysine blots, and added the molecular weight markers (Figure 6F, Figure S6C).

• Figure 6H and S5: The mitochondria transfer experiment appears to be quite efficient compared to previously published studies. It would be important to control that the signal observed in the recipient cells is not due to the leakage of the MitoTracker dye from the donor mitochondria.

Answer: This is an interesting point though MitoTracker dye is not supposed to leak as it covalently binds to mitochondrial proteins. Even though the dye may leak to mark the endogenous mitochondrial, it does not affect our goal to demonstrate that transfer of Fam210aMKO mitochondria into healthy cells can induce protein hyperacetylation. Additional evidence argues against the leakiness of Mitotracker dye to subsequently mark other mitochondria in the recipient cells: 1) mtDNA and MitoTracker signal both increase linearly with the increasing amounts of mitochondria transferred (Figure S7A); 2) We have now also included confocal images to show the presence of both MitoTracker labeled and non-labeled mitochondria in the recipient cells. We reason that if MitoTracker leaks within a cell then it would have labeled all mitochondrial in that cell (Figure 7C).

• Figure 6J: The increase in Fgf21 is modest. Although the difference is statistically significant, is it biologically important?

Answer: We thank the reviewer for this question; indeed, the increase is modest. We think the reason of the modest increase compared to the drastic increase seen in vivo was because when we transplanted the WT and Fam210aMKOmitochondria to the recipient cell, the original mitochondria in the recipient were not depleted, which could explain the milder effect. However, we were able to show that the recipient cells readily increase the acetylation of proteins after receiving the Fam210aMKO mitochondria, recapitulating the phenotype we saw in the Fam210aMKO muscle.

After careful considerations on the mechanism proposed in the study, we decided to remove qPCR data showing the modest increase of Fgf21 mRNA level. The removal of this data will not change the conclusions we draw nor lessen the significance of the mitochondria transfer experiment.

• Figure 6C: How significant is the difference in acetylation of RPL30 in WT vs. KO. RPS13 was not found in the WT MS? Was this normalized to Input?

Answer: the MS was performed with same loading. The mass spectrometry results for protein identification after AcK-IP were from pooled samples from 3 independent replicates (as the KO muscles are very scarce). Therefore, there was not a significance test.

• Figure 7D: What are the MW of the bands shown on this blot? This experiment is by no means sufficient to demonstrate and confirm that ribosomal proteins are acetylated. An increase in RPL30 and RPS13 acetylation must be directly assessed.

Answer: We thank the reviewer for suggesting the more direct assays to look at RPL30 and RPS13 acetylation. We have shown that the ribosome fractions were indeed hyperacetylated in the Fam210aMKO mouse compared to the WT control (Figure S6D). We agree that this result cannot lead to the conclusion that the RPL30 and RPS13 are specifically hyperacetylated. Indeed, we have tried to use Acetylated lysine antibody pull down and RPS13/RPL30 blot to show the increase in the acetylated RPS13/RPL30 protein. However, we cannot show a robust increase in the acetylation, potentially due to the low number of acetylation sites on RPS13 and RPL30 protein. We therefore have reworded the conclusion in the revised manuscript to better reflect the results.

• Fig7E: This experiment is not properly executed and in its current state does not rigorously support that "hyper-acetylation of several small ribosomal proteins leads to ribosomal disassembly". A) UV profiles of the fractionation must be provided to assess the quality of the profile. B) Provide MW markers. Which band is RPL30? The Input and free fraction bands are not at the same size. RPL30 should at least be visible on the 60S and polysomes from the WT. C) These results do not match the acetylation MS data, which seem to show that the increase in acetylation is much greater for RPS13. However, RPS13 presence on polysomes (assuming they are polysomes) is not affected in the KO. D) This type of experiment must be done for three independent biological replicates, blots from single lanes must be quantified and normalized to total signal (from all the lanes) for the same antibody.

Answer: we appreciate the great advice on improving the experiment. As suggested, we have now added proper experimentation (UV profile, and better WB), with the help of Dr. Kotaro Fujii (included as co-author in the revised manuscript). The following results showed that in the 4-week sample, there was a decrease in the 80S monosome and polysome in the Fam210aMKO mice compared to the WT. The change was more drastic at 6-week (Figure 4D-4G). Similarly, due to the scarce amount of muscle in the KO mice, we need to pool samples from the 6-week-old mice for the experiment, and hope the reviewer can understand the situation. With the clear peaks shown in the UV profile as well as the WB results, we provide more convincing evidence that the polysome assembly was indeed impaired in the Fam210aMKO (Figure 4D-4G).

• Fig 7F: Global translation rates are assessed by puro incorporation at week 4, a time point when differences in protein acetylation were not observed. This does not support the hypothesis that increased acetylation of ribosomal proteins causes defect in protein translation. (Referencing the authors statement p.7 lines 321-24.).

Answer: We thank the reviewer for this question. When we quantified the protein acetylation increase in the muscle at 4-weeks, we showed that there was a significant increase. But like the reviewer said, the ribosomal fractions were not significantly acetylated by WB at 4-week. We reasoned that, at early stages (4-weeks), the ISR signaling can lead to the translational arrest, along with the polysome formation defects, leading to the decreased protein translation. These are included in the discussion.

• Other studies have implicated Fam210A in the regulation of mitochondrial protein synthesis through an interaction with EF-Tu. The authors also identified EF-Tu as an interactor in their LC-MS analysis (FigS4). A role for this interaction accounting for mitochondrial and translation defects seems to be underestimated and unexplored here.

Answer: We agree with this point and believe the cytoplasmic translation defects are in addition to the mitochondrial translational defects. We have shown that FAM210A KO leads to the decrease of the MTCO1 which is encoded by the mitochondrial genome. Besides, we also showed by mitochondrial proteomics that TUFM was reduced in the KO, which also contributed to translational arrest in the mitochondria (Figure 5J). To answer whether mitochondrial encoded proteins are decreased in upon Fam210a KO, we blotted the protein lysates at different stages with antibodies for a few mitochondrial encoded proteins and showed that they decreased with ages (Response Figure 7).

Response Figure 7. WB analysis and quantification of mitochondrial encoded proteins in WT and Fam210aMKO muscle at different ages.

The mitochondrial proteins were indeed decreased in Fam210aMKO starting from 6-weeks of age compared to the WT.

Minor comments:

-What is known about FAM210A, other studies assessing its role, and the rational for studying its function should be better introduced.

Answer: We thank the reviewer for the suggestion to have more information of FAM210A functions/mechanisms in the introduction. We have added more background to the introduction.

-In the discussion the authors states: "Moreover, when the proportion of ribosomal protein phosphorylation buildup in the Fam210aMKO, the assembly of the translational machinery is impaired therefore further dampen the cellular translation". Do they mean acetylation and not phosphorylation?

Answer: We are sorry about the typo and have changed it. We thank the reviewer for catching this.

• Please use the term "mRNA translation" or "protein synthesis" instead of "protein translation" in the text.

Answer: We thank the reviewer for the suggestion to properly refer to these processes. We have changed the terms in the manuscript.

-The methods section for RT-qPCR: It should ne M-MLV RT and not M-MLC. If the qPCR data was normalized with 18S, please provide the sequence of the primers in the table. Information on how primer efficiency was tested must be included in the method section.

Answer: We thank the reviewer for pointing this out. We have changed the texts. We also have provided 18S sequence and provide texts about how primer efficiency was tested.

Reviewer #2 (Significance (Required)):

General assessment: Previous genome-wide association studies have found that mutations in FAM210A were associated with sarcopenia and osteoporosis. Because FAM210A is not expressed in the bone and highly expressed in skeletal muscle, it suggests that FAM210A likely plays an important role in muscle, which could also affect bone regulation. The authors here provide further evidence of an important role for FAM210A in diseases affecting muscle function by demonstrating that the expression of FAM210A decreases with age and in patients affected by Pompe disease, Duchenne muscular dystrophy and hereditary recessive myopathy. FAM210A is a mitochondria-localized protein and given the crucial role of mitochondria in supporting muscle metabolism, elucidating the molecular function of FAM210A may provide important insights into diseases biology that could lead to the development of therapeutic approaches. Thus, a significant protein and regulatory pathway are explored in this study that can potentially impact human health. In this manuscript, the authors provide compelling evidence of the importance of Fam210a in muscle homeostasis with their newly generate mouse model. The experiments looking at muscle physiology, function and metabolism are well-executed and for the most part rigorous, which are the strengths of this manuscript. However, the conclusion that Fam210a deletion in skeletal muscle induces the hyper-acetylation of several small ribosomal proteins, which leads to ribosomal disassembly and translational deficiency is not supported by the data presented here. As noted in the comments above, these experiments need major improvement. Additionally, there are other concerns about general scientific rigor and conclusions inconsistent with the data presented as also noted in the comments section.

Advance: Although a previous study explored the role of FAM210A using a skeletal muscle-specific KO induced at postnatal 28 days under a HSA promoter, the model used by the authors here provide a cleaner approach and more insights into the molecular functions of FAM210A in muscle physiology. The findings that Fam210a MKO disrupts the flow of TCA cycle, which leads to an abnormal accumulation of acetyl-CoA is interesting and provide new conceptual advance on the roles of FAM210A in mitochondria function in muscle. Acetyl-CoA production is an important source of acetyl-group that can be transferred to proteins and regulate gene expression programs. Thus, this is an important finding. However, molecular mechanism by which FAM210A regulates this process through an interaction with SUCLG2 is not provided and the nature this interaction is superficially explored.

Audience: Findings from this manuscript are likely to interest both basic research and translational/clinical audiences as it explores the physiological and molecular function of a disease-linked protein. The findings are also likely to impact the fields of metabolism, mitochondria function and regulation of gene expression by protein acetylation (if concerns raised regarding these experiments are addressed).

The fields of expertise of this reviewer are protein and RNA modifications, ribosome biogenesis and mRNA translation.

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

The authors state that in their manuscript "the role of mitochondria in regulating cytosolic protein translation in skeletal muscle cells (myofibers)" has been explored (Line 19-20). As experimental model, they used mice expressing Cre recombinase under the control of the myosin light chain 1 promoter. The first conclusion was that "FAM210A is positively associated with muscle mass in mice and humans". The authors say that the presented data "reveal a novel crosstalk between the mitochondrion and ribosome mediated by FAM210A".

I recognize the potential of this work since the role of FAM210a has been more deeply investigated in skeletal muscle. In fact, the study by Tanaka et al, 2018 presented only a preliminary characterization of the role of FAM210a in muscle. However, I think that this work is not complete and each aspect that has been investigated is not well connected with each other. In particular, it is not clear whether the disrupted ribosomal assembly by hyperacetylation causes muscle atrophy or it is altered under catabolic states during atrophy (primary cause or consequence of?).

Answer: We thank the reviewer for recognizing the importance of the study that characterizes the effect of FAM210A in muscle mass maintenance. In this study, we have shown that polysome formation was impaired at 4-week and therefore the translational efficiency was reduced in the muscle. This translational decrease coincides with the acetylation increase. Moreover, we showed by mitochondrial transfer experiment that the mitochondria from the Fam210aMKO mice can carry the phenotype and lead to acetylation increase in the recipient cells. Since muscle protein synthesis defects have been known to lead to muscle dystrophy, and we have shown that in the Fam210aMKO model, protein synthesis was indeed defective while there was not an induction of atrophy. Therefore, we conclude that the in the KO model, the protein synthesis defects lead to muscle atrophy.

The other major point is represented by the fact that the Myl1-CRE expressing model provides selectivity in fast muscle fibers (see for example Barton PJR, Harris AJ, Buckingham M. Myosin light chain gene expression in developing and denervated fetal muscle in the mouse. Development. 1989;107: 819-824). Then the authors knocked out FAM210a only in fast fibers and they never take in consideration this key point! This is crucial since fast and slow muscles have different content of mitochondria with different size, shape, and metabolism! The muscle fibers can be classified based on the mitochondrial metabolism (see for example Chemello et al., 2019; PMID: 30917329).

Regarding this point, they simply wrote at Line 75-76 "using a skeletal muscle specific Myl1 (myosin, light polypeptide 1) driven Cre recombinase specifically expressed in post-differentiation myocytes and multinucleated myofibers,...". It would be more correct to write multinucleated type 2 myofibers showing the reduction of FAM210a in different fiber types.

I think that the authors must solve these aspect and then organize the findings accordingly. The data are in general interesting for broad type of audience.

Answer:

We appreciate the reviewer’s comment on the Myl1 knock-in Cre (Myl1Cre) model, which prompted us to more explicitly clarify some of the confusions around this model. We fully respect the validity of the 1989 study by Dr. Buckingham and other studies showing fast muscle specific expression of Myl1. However, we and others have shown that Myl1 not only mark the fast but also the slow myofibers (elaborated below). The discrepancy can be explained by the fact that using the Myl1Cre as a lineage marker is different from directly examining Myl1 expression at static timepoints by in situ hybridization (ISH). This is because Cre recombinase can accumulate and diffuse to all the myonuclei in a multinucleated myofiber, subsequently leading to deletion of LoxP-flanked DNA in all nuclei. Also, in the Cre/LoxP system, only a small amount of Cre recombinase is needed to induce the recombination of the target loxP sites and lead to gene KO. Another example of the discrepancy between the static mRNA pattern and the dynamic gene expression during development is the Hox gene expression. When the corresponding author (SK) of this manuscript was trained with Dr. Joshua R Sanes, he developed 3 Cre lines driven by three different Hox genes– that have been shown by ISH to be expressed in a specific rostral to caudal domain in the spinal cord during development. However, each of these Cre model ended up marking all the spinal cord without any domain specificity. In the case of Myl1Cre mouse model, we have previously published a paper on the lineage-tracing results using the Myl1Cre and showed that Myl1Cre marked all fast AND slow myofibers in mice (Wang et al, 2015, PMID: 25794679). In another lineage tracing study using nuclear GFP reporter, we report that Myl1Cre marks 96% nuclei in myofibers regardless of fiber types (Bi et al., 2016, PMID: 27644105), the remainder 4% non-marked nuclei potentially represent satellite cells. Other groups have also used the Myl1Cre model to induce KO in both fast and slow muscles (Pereira et al, 2020, PMID: 31916679). Therefore, we believe that the Myl1Cre mouse model allows us to efficiently knockout the Fam210a gene in both slow and fast muscle.

To directly confirm that Fam210a was efficiently knocked out in both slow and fast muscles using the Myl1Cre mouse model, we isolated different muscle groups (Soleus and diaphragm that contains a large fraction of slow myofibers, TA and EDL that contain predominantly fast myofibers) and checked the expression level and the KO efficiency of Fam210a by WB. We have shown that even in slow muscles like diaphragm and SOL, the KO was very efficient, as there were no visible FAM210A bands in the WB (Figure S1C).

In more detail:

The data must be analyzed and discussed based on the fact that FAM210a has been deleted specifically in fast fibers. First the authors must show the protein levels of FAM210a in both fast, slow and mixed fast-slow muscles. Then for example in Figure S1C EDL, GAS and SOL muscles must be included.

Answer: This is related to the misunderstanding of the Myl1Cre model. We understand the reviewer’s concern and therefore isolated proteins from different muscles in WT and Fam210aMKO mice at 4-weeks and checked the expression level of FAM210A. We have shown that regardless of fast or slow muscles, FAM210A was deleted.

The blot in general must be repeated since it has poor quality (continuum of FAM210a band in the samples).

Answer: We thank the reviewer for this suggestion and increase the data quality. We have changed the original blot with the following blots showing that FAM210A was not deleted in other non-muscle tissues (Figure S1C).

Please provide staining of TA, GAS and SOL muscles to show how Myl1CRE-directed deletion of FAM210a affect the different myofibers.

Answer: This point is also related to assumption that Myl1Cre only induce deletion in fast myofibers. We have done staining in both EDL and SOL muscle to show the relative changes in myofiber compositions. We found that the myofibers in EDL and SOL muscle have shifted to a more oxidative type upon Fam210a KO (Figure S3).

In Figure 2F where decreased TA muscle weight was showed in the Fam210aMKO mice, the authors must include also the other muscles (EDL, GAS and SOL).

Answer: We thank the reviewer for helping us be more rigorous on the phenotype examination. We understand that the reviewer initially raised this question because of the concern on Myl1Cre model. Now that we have shown the MylCremarks both the fast and slow muscles, we believe this question is no longer a concern. Besides, to indirectly answer the question, we would like for the reviewer to appreciate the size difference of the EDL as well as the SOL muscle in Figure S3 in the manuscript. As can be seen from the images, the size of the SOL muscles in the KO was significantly reduced compared to the WT, speaking in favor of the KO effect on slow muscles.

In general, since the HSA-CRE model is generally used for gene manipulation in skeletal muscles the authors must characterize their model considering that the myosin light chain 1 promoter Myl1-Cre is mainly active in postmitotic type II myofibers. The last model can also give advantage for mosaic gene manipulation in muscles with mixed fiber types.

Answer: We thank the reviewer for bringing this point up. We hope by the multiple lines of evidence that we provided in the previous questions, we can convince the reviewer that the KO model using the Myl1Cre does not lead to a mosaic gene manipulation in the muscle. On the contrary, the KO model is a homogeneous KO in both fast and slow muscles.

Line 118-119 Fam210a level is positively corelated with muscle mass, as it is reduced in muscle atrophy conditions and increased in muscle hypertrophy conditions. Fig 1: I don't like since there are many different models in which the muscle mass reduction is associated with different mechanisms. Then independently of mechanisms associated with changes in muscle mass Fam210a is always linked to? Which common mechanism can explain this?

Answer: We understand that the reviewer would like to pursue a conserved mechanism governing muscle mass maintenance, however, we by no means wanted to make a direct causal relationship between FAM210A level and different muscle disease/atrophy conditions. Indeed, the atrophic conditions presented have different mechanisms leading to muscle mass reduction, yet we wanted to present the possible connection that Fam210a level and muscle mass are co-regulated, and we later confirmed by KO mouse model that FAM210A KO indeed reduces muscle mass.

Line 144-146 Hematoxylin and eosin (H&E) staining did not reveal any obvious myofiber pathology in the Fam210aMKO mice up to 8 weeks (Figure 2G). I totally disagree! It seems that there is more inflammation upon deletion of Fam210aMKO. Please check it.

Answer: We thank the reviewer for pointing this out to help us more rigorously describe our results. We have changed the wording to better reflect the changes observed with H&E images.

Fig3E-L there is a huge difference between EDL and SOL. The authors can't avoid to discuss their data considering the real expression of CRE upon Myl promoter: specific deletion in fast fibers. I think that the data in FIGS3 are very important and must be linked to data in Fig3. Organize in a different way all the presented data to really describe what is happening upon deletion of Fam210a.

Again, the authors MUST organize better their data in the manuscript: to each paragraph must correspond data in the main figures. For example: at Line 189 Fam210aMKO mice exhibit systemic metabolic defects and at Line 208 Fam210aMKO increases oxidative myofibers and decreases glycolytic myofibers. These two paragraphs discuss data showed only in supplementary figures.

Answer: We thank the reviewer for this suggestion. As shown in the previous responses, the Myl1Cre indeed induce efficient deletion of Fam210a in slow muscles. Therefore, we did not consider this to be a myofiber-specific deletion model. We consider these two results as the effect of a mitochondrial protein (FAM210A) on the myofiber metabolism (independent of myofiber type specific deletion), and that the deletion of Fam210a results in mitochondrial stress, which can lead to myofiber switch (Figure S3).

Physical activity mast be monitored. Show respiratory exchange ratio (RER = VCO2/VO2) and discuss the results.

Answer: We thank the reviewer for this suggestion. By these results, we would like to demonstrate that muscle homeostasis is important for the systemic metabolism, disruption of muscle mass maintenance in the Fam210aMKO mice leads to defects in the whole-body metabolism. We have now included the RER results (Figure S2F, S2G). The results show that the Fam210aMKO mice had significantly lower RER (VCO2/VO2) value at daytime, indicating that the mice rely more on utilizing fat as the fuel source. This is consistent with the proteomics results (Figure 5K) that the Fam210aMKOmice have increased FAO pathway. Unfortunately, our metabolic chamber does not have the capacity to monitor activity. We instead include data on heat production (Figure S2E).

"Fam210aMKO increases oxidative myofibers and decreases glycolytic myofibers". The data mast be associated with the evaluation of the expression levels of FAM210 in different fiber type to really understand what is happening upon FAM210a loss.

Answer: We understand the reviewer’s concern on the different expression level of Fam210a as well as the KO efficiency using the Myl1Cre model. We have shown that Fam210a is knocked out in fast and slow muscles, therefore, we did not consider the effects on fast and slow myofibers separately.

As SDH activity in type 1 fibers is higher than type 2 the and since the authors are using a model in which Fam210a is deleted only in type 2 fiber they should understand what is happening: fiber 1

Answer: We agree with the reviewer that the SDH activity is different in different myofibers. We have shown by western blot that FAM210A was similarly KO in both fast and slow muscles. When we performed fiber type staining in EDL and SOL muscle, we saw that there was a shift towards the slower myofiber types both in the EDL and SOL muscle, due to mitochondrial defects.

Associate a cox assay with the sdh assay

Answer: We thank the reviewer for this suggestion. We have shown by SDH staining as well as seahorse experiments using isolated mitochondria that the complex II activity was impaired in the muscle. We understand the reviewer would like to see a COX assay to show the defects of the mitochondrial function. Though we were not able to perform the COX assay, we showed from other aspects that the mitochondrial function was impaired by running WB of the mitochondrial encoded proteins (ATP6, MTCO2, mtCYB) and showed these proteins were decreased with ages. Along with the morphological changes of the mitochondria shown by electron microscope (Figure 5 and Figure S5), we conclude that these changes must have impacted mitochondrial function.

Figure 4b blot tubulin and FAM210a look strange. Look especially at first and second and fourth form the left side.

Answer: We are sorry about the mistake in the images, we have changed the Tubulin blot in the Oxphos blots.

Figure 4B OXPHOS protein levels look similar between wt and KO. Include the quantification with the significance (min 3-5 mice per genotype).

Answer: we have quantified the change between WT and KO on different proteins (Reponse Figure 8).

Response Figure 8. Quantification of the OxPHOS proteins in WT and Fam210aMKO muscle at different ages.

Quantification of the blots showed that indeed the mitochondrial proteins were decreased in the Fam210aMKO. The change of mitochondrial encoded protein MTCO1 was earlier detected in the Fam210aMKO.

Provide TEM analysis for SOL muscle. I would understand whether mitochondria are differently affected in fast and slow muscles.

Answer: We understand the reviewer was originally concerned about the KO efficiency of Fam210a in fast and slow muscles, based on the assumption about the MylCre model. We have shown that the FAM210A protein was similarly depleted in both fast and slow muscles by western blot. In this case, we would speculate that the mitochondrial change in fast and slow muscles would be similar because the mitochondrial changes were due to the inherent defects in the mitochondria.

In all experiments must be clear which muscle type or types was/were used:

Line 268: "isolated from WT and Fam210aMKO muscles at 6 weeks of age".

Line 587 "Muscle lysate acetyl-CoA contents"

For Seahorse Mitochondrial Respiration Analysis at Line 599 "isolated mitochondria from muscle"

For TCA cycle metabolomics at Line 615 "muscle tissue was weighed and homogenized"

For SCS activity assay at Line 632 "mitochondria from muscles were isolated"

For LC-MS/MS at Line 647 "Mitochondria were purified from skeletal muscles and subjected to proteomics analysis".

For Ribosome isolation at Line 676 "Skeletal muscle from mice"

For Polysome profiling experiment at Line 696 "muscle tissues from mice were dissected"

It is important to know which muscles were used since confounding effects of the specific deletion of FAM210a in type 2 fibers must be identified and discussed.

Answer: We thank the reviewer for considering the different muscle groups in our mouse model. For experiments requiring a large amount of muscle tissue, such as ribosome isolation, mitochondrial isolation and polysome profiling, we used all the muscles from the mouse. For WB experiments, we used the TA muscle. We have included this information in the method section in the manuscript. Since we have shown that FAM210A was similarly depleted in different muscles (see previous responses), we think it is justified to pool muscles from the same mouse.

Line 296-297 The authors wrote "Consistently, the mRNA levels of Atf4, Fgf21 and the associated transcripts were highly induced in the Fam210aMKO 296 both in the 4-week and 6-week-old muscle samples". Is Fgf21 responsible for the reduction of body weight? (see for example PMID: 28552492, PMID: 28607005 and PMID: 33944779). Measure the circulating Fgf21 protein in Ko and wt mice.

Answer: We thank the reviewer for this great suggestion. Indeed, Fgf21 can potentially lead to body weight reduction, and this can explain the smaller body weight in our mouse model as well. However, we are more concerned about the muscle changes in our mouse model, therefore we did not further validate the changes of Fgf21 in the circulation.

After careful considerations on the mechanism proposed in the study, we decided to remove qPCR data showing the modest increase of Fgf21 mRNA level. The removal of this data will not change the conclusions we draw nor lessen the significance of the mitochondria transfer experiment.

Moreover the authors must check Opa1 total protein level and also the ratio between long and short isoforms. Is Fam210a interacting with Opa1?

Answer: We thank the reviewer for this interesting question. Another publication from our lab has shown that Fam210a can modulate the cleavage of OPA1 in brown adipose tissue and influence the cold-induced thermogenesis (PMID: 37816711). Indeed, OPA1 deletion in muscle can lead to muscle atrophy and postnatal death at about day 10 (PMID: 28552492) through the induction of UPR (ISR) and the induction of Fgf21. We did not check the interaction between FAM210A and OPA1 in the muscle context, and FAM210A was not found to be interacting with OPA1 in brown adipose tissue (PMID: 37816711). However, the focus of this study was the acetylation change and the FAM210A effect on muscle mass maintenance. Therefore, we did not pursue the OPA1 related mechanism in skeletal muscle.

The final part of the paper is really interesting but need to be discussed knowing exactly the used experimental model. Then check in which fiber types FAM210a is loss.

Answer: We thank the reviewer for the stringency on the model used. Indeed, the mitochondria can be different from different muscle groups. However, since the muscle isolated from WT and KO mice was properly controlled and therefore can balance the effects of different mitochondria. We have consistently observed the increased acetylation when mutant mitochondria were transferred.

Regarding the mitochondrial transplantation I'm surprise to see that it happens in the direction of unhealthy mitochondria to healthy cells. Are you able to rescue the phenotype of Fam210a KO cells with healthy mitochondria?

Answer: We thank the reviewer for bringing this interesting yet important question up! Our mitochondrial transfer results support a “gain-of-function” model where excessive Acetyl CoA produced by the Fam210a-KO mitochondrial induces hyperacetylation. Regarding the question to transfer healthy mitochondria to rescue the KO cells, we reason that even when we transfer the healthy mitochondria to the KO cells, the healthy mitochondria will not stop the mutant mitochondria from making excessive amounts of acetyl-CoA and thus protein acetylation. A clean transfer would require depletion of the mitochondria in the KO cells and concomitant restoring FAM210A level in the KO cells (as the lack of Fam210a gene in the KO cells will eventually convert the transferred mitochondrial into mutants with the normal turnover of FAM210A). This is technically highly challenging and nearly impossible to do. We hope that the reviewer can understand the difficulties.

Reviewer #3 (Significance (Required)):

In conclusion, the strength of the presented paper is the novelty: the authors explored the role of FAM210a in skeletal muscle. However, the major limitation is represented by the fact that they did not show in which fiber types Fam210a is knocked out. In fact, the used CRE recombinase expressing model is well-known to be specific for type 2 fibers. Then since mitochondria and metabolism are central in this manuscript and they are different in the fast and slow fiber types, the authors must dissect in details this point.

Moreover, there are many data but they are not linked each other and discussed properly. The paper must be completely re-organized.

This manuscript can be interesting for a broad type of audience.

I'm an expert on mitochondria, metabolism and skeletal muscle.

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 authors state that in their manuscript "the role of mitochondria in regulating cytosolic protein translation in skeletal muscle cells (myofibers)" has been explored (Line 19-20). As experimental model, they used mice expressing Cre recombinase under the control of the myosin light chain 1 promoter. The first conclusion was that "FAM210A is positively associated with muscle mass in mice and humans". The authors say that the presented data "reveal a novel crosstalk between the mitochondrion and ribosome mediated by FAM210A".

I recognize the potential of this work since the role of FAM210a has been more deeply investigated in skeletal muscle. In fact, the study by Tanaka et al, 2018 presented only a preliminary characterization of the role of FAM210a in muscle. However, I think that this work is not complete and each aspect that has been investigated is not well connected with each other. In particular, it is not clear whether the disrupted ribosomal assembly by hyperacetylation causes muscle atrophy or it is altered under catabolic states during atrophy (primary cause or consequence of?).

The other major point is represented by the fact that the Myl1-CRE expressing model provides selectivity in fast muscle fibers (see for example Barton PJR, Harris AJ, Buckingham M. Myosin light chain gene expression in developing and denervated fetal muscle in the mouse. Development. 1989;107: 819-824). Then the authors knocked out FAM210a only in fast fibers and they never take in consideration this key point! This is crucial since fast and slow muscles have different content of mitochondria with different size, shape, and metabolism! The muscle fibers can be classified based on the mitochondrial metabolism (see for example Chemello et al., 2019; PMID: 30917329).

Regarding this point, they simply wrote at Line 75-76 "using a skeletal muscle specific Myl1 (myosin, light polypeptide 1) driven Cre recombinase specifically expressed in post-differentiation myocytes and multinucleated myofibers,...". It would be more correct to write multinucleated type 2 myofibers showing the reduction of FAM210a in different fiber types.

I think that the authors must solve these aspect and then organize the findings accordingly. The data are in general interesting for broad type of audience.

In more detail:

The data must be analyzed and discussed based on the fact that FAM210a has been deleted specifically in fast fibers. First the authors must show the protein levels of FAM210a in both fast, slow and mixed fast-slow muscles. Then for example in Figure S1C EDL, GAS and SOL muscles must be included. The blot in general must be repeated since it has poor quality (continuum of FAM210a band in the samples). Please provide staining of TA, GAS and SOL muscles to show how Myl1CRE-directed deletion of FAM210a affect the different myofibers. In Figure 2F where decreased TA muscle weight was showed in the Fam210aMKO mice, the authors must include also the other muscles (EDL, GAS and SOL). In general, since the HSA-CRE model is generally used for gene manipulation in skeletal muscles the authors must characterize their model considering that the myosin light chain 1 promoter Myl1-Cre is mainly active in postmitotic type II myofibers. The last model can also give advantage for mosaic gene manipulation in muscles with mixed fiber types. Line 118-119 Fam210a level is positively corelated with muscle mass, as it is reduced in muscle atrophy conditions and increased in muscle hypertrophy conditions. Fig 1: I don't like since there are many different models in which the muscle mass reduction is associated with different mechanisms. Then independently of mechanisms associated with changes in muscle mass Fam210a is always linked to? Which common mechanism can explain this? Line 144-146 Hematoxylin and eosin (H&E) staining did not reveal any obvious myofiber pathology in the Fam210aMKO mice up to 8 weeks (Figure 2G). I totally disagree! It seems that there is more inflammation upon deletion of Fam210aMKO. Please check it. Fig3E-L there is a huge difference between EDL and SOL. The authors can't avoid to discuss their data considering the real expression of CRE upon Myl promoter: specific deletion in fast fibers. I think that the data in FIGS3 are very important and must be linked to data in Fig3. Organize in a different way all the presented data to really describe what is happening upon deletion of Fam210a. Again, the authors MUST organize better their data in the manuscript: to each paragraph must correspond data in the main figures. For example: at Line 189 Fam210aMKO mice exhibit systemic metabolic defects and at Line 208 Fam210aMKO increases oxidative myofibers and decreases glycolytic myofibers. These two paragraphs discuss data showed only in supplementary figures. Physical activity mast be monitored. Show respiratory exchange ratio (RER = VCO2/VO2) and discuss the results. "Fam210aMKO increases oxidative myofibers and decreases glycolytic myofibers". The data mast be associated with the evaluation of the expression levels of FAM210 in different fiber type to really understand what is happening upon FAM210a loss. As SDH activity in type 1 fibers is higher than type 2 the and since the authors are using a model in which Fam210a is deleted only in type 2 fiber they should understand what is happening: fiber 1 Associate a cox assay with the sdh assay Figure 4b blot tubulin and FAM210a look strange. Look especially at first and second and fourth form the left side. Figure 4B OXPHOS protein levels look similar between wt and KO. Include the quantification with the significance (min 3-5 mice per genotype). Provide TEM analysis for SOL muscle. I would understand whether mitochondria are differently affected in fast and slow muscles. In all experiments must be clear which muscle type or types was/were used: Line 268: "isolated from WT and Fam210aMKO muscles at 6 weeks of age". Line 587 "Muscle lysate acetyl-CoA contents" For Seahorse Mitochondrial Respiration Analysis at Line 599 "isolated mitochondria from muscle" For TCA cycle metabolomics at Line 615 "muscle tissue was weighed and homogenized" For SCS activity assay at Line 632 "mitochondria from muscles were isolated" For LC-MS/MS at Line 647 "Mitochondria were purified from skeletal muscles and subjected to proteomics analysis". For Ribosome isolation at Line 676 "Skeletal muscle from mice" For Polysome profiling experiment at Line 696 "muscle tissues from mice were dissected" It is important to know which muscles were used since confounding effects of the specific deletion of FAM210a in type 2 fibers must be identified and discussed. Line 296-297 The authors wrote "Consistently, the mRNA levels of Atf4, Fgf21 and the associated transcripts were highly induced in the Fam210aMKO 296 both in the 4-week and 6-week-old muscle samples". Is Fgf21 responsible for the reduction of body weight? (see for example PMID: 28552492, PMID: 28607005 and PMID: 33944779). Measure the circulating Fgf21 protein in Ko and wt mice. Moreover the authors must check Opa1 total protein level and also the ratio between long and short isoforms. Is Fam210a interacting with Opa1? The final part of the paper is really interesting but need to be discussed knowing exactly the used experimental model. Then check in which fiber types FAM210a is loss. Regarding the mitochondrial transplantation I'm surprise to see that it happens in the direction of unhealthy mitochondria to healthy cells. Are you able to rescue the phenotype of Fam210a KO cells with healthy mitochondria?

Significance

In conclusion, the strength of the presented paper is the novelty: the authors explored the role of FAM210a in skeletal muscle. However, the major limitation is represented by the fact that they did not show in which fiber types Fam210a is knocked out. In fact, the used CRE recombinase expressing model is well-known to be specific for type 2 fibers. Then since mitochondria and metabolism are central in this manuscript and they are different in the fast and slow fiber types, the authors must dissect in details this point. Moreover, there are many data but they are not linked each other and discussed properly.The paper must be completely re-organized.

This manuscript can be interesting for a broad type of audience.

I'm an expert on mitochondria, metabolism and skeletal muscle.

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 manuscript, Chen et al., investigate the functions of FAM210A in skeletal muscle physiology and metabolism. FAM210A is a mitochondria-localized protein in which mutations have been associated with sarcopenia and osteoporosis. Using publicly available gene expression datasets from human skeletal muscle biopsies the authors first demonstrate that the expression of FAM210 is reduced in muscle atrophy-associated diseases and increased in muscle hypertrophy conditions. Based on this, they show that a muscle specific Fam210a deletion leads to muscle atrophy/weakness, systemic metabolic defects, and premature lethality in mouse. Further examination of the knockout myofibers reveals impaired mitochondrial respiration and translation program. Additionally, the authors demonstrate that the flow of TCA cycle is disrupted in the FAM210A-deleted myofibers, which causes abnormal accumulation of acetyl-coA and hyperacetylation of a subset of proteins. The authors claim that Fam210a deletion in skeletal muscle induces the hyper-acetylation of several small ribosomal proteins that leads to ribosomal disassembly and translational deficiency. However, this conclusion is not supported by adequate experimentation and rigorous analysis of ribosomal proteins acetylation and ribosome assembly.

Major comments:

• In general, figure legends are lacking information regarding number of biological replicates used and details about statistical analysis. What does three * vs. one * mean in terms of p-value? Exact p-values should be indicated.
• The mechanistic studies linking muscle phenotypes with ribosomal protein hyperacetylation and mRNA translation defects are underdeveloped and not rigorously carried.
• Fig S1: The validation WB of FAM210A KO is not the most convincing. Why are the FAM210A levels so low in TA compared to other tissues?
• Fig 2G: The authors state "Hematoxylin and eosin (H&E) staining did not reveal any obvious myofiber pathology in the Fam210a KO mice up to 8 weeks". However there seems to be a progressive increase in nuclei up to 8-weeks in the KO. What is the significance of this?
• IP-MS analysis for FAM210A interacting proteins requires validation with IP and reverse IP + WB experiment.
• Figure 4A requires quantification of the SDH signals from multiple samples.
• Figure 6F: To clearly demonstrate an increase in protein acetylation in the FAM210 MKO, the authors must provide quantification data generated with more then N=1. Please add the molecular weights markings on the side of the blots.
• Figure 6H and S5: The mitochondria transfer experiment appears to be quite efficient compared to previously published studies. It would be important to control that the signal observed in the recipient cells is not due to the leakage of the MitoTracker dye from the donor mitochondria.
• Figure 6J: The increase in Fgf21 is modest. Although the difference is statistically significant, is it biologically important?
• Figure 6C: How significant is the difference in acetylation of RPL30 in WT vs. KO. RPS13 was not found in the WT MS? Was this normalized to Input?
• Figure 7D: What are the MW of the bands shown on this blot? This experiment is by no means sufficient to demonstrate and confirm that ribosomal proteins are acetylated. An increase in RPL30 and RPS13 acetylation must be directly assessed.
• Fig7E: This experiment is not properly executed and in its current state does not rigorously support that "hyper-acetylation of several small ribosomal proteins leads to ribosomal disassembly". A) UV profiles of the fractionation must be provided to assess the quality of the profile. B) Provide MW markers. Which band is RPL30? The Input and free fraction bands are not at the same size. RPL30 should at least be visible on the 60S and polysomes from the WT. C) These results do not match the acetylation MS data, which seem to show that the increase in acetylation is much greater for RPS13. However, RPS13 presence on polysomes (assuming they are polysomes) is not affected in the KO. D) This type of experiment must be done for three independent biological replicates, blots from single lanes must be quantified and normalized to total signal (from all the lanes) for the same antibody.
• Fig 7F: Global translation rates are assessed by puro incorporation at week 4, a time point when differences in protein acetylation were not observed. This does not support the hypothesis that increased acetylation of ribosomal proteins causes defect in protein translation. (Referencing the authors statement p.7 lines 321-24.).
• Other studies have implicated Fam210A in the regulation of mitochondrial protein synthesis through an interaction with EF-Tu. The authors also identified EF-Tu as an interactor in their LC-MS analysis (FigS4). A role for this interaction accounting for mitochondrial and translation defects seems to be underestimated and unexplored here.

Minor comments:

• What is known about FAM210A, other studies assessing its role, and the rational for studying its function should be better introduced.
• In the discussion the authors states: "Moreover, when the proportion of ribosomal protein phosphorylation buildup in the Fam210aMKO, the assembly of the translational machinery is impaired therefore further dampen the cellular translation". Do they mean acetylation and not phosphorylation?
• Please use the term "mRNA translation" or "protein synthesis" instead of "protein translation" in the text.
• The methods section for RT-qPCR: It should ne M-MLV RT and not M-MLC. If the qPCR data was normalized with 18S, please provide the sequence of the primers in the table. Information on how primer efficiency was tested must be included in the method section.

Significance

General assessment: Previous genome-wide association studies have found that mutations in FAM210A were associated with sarcopenia and osteoporosis. Because FAM210A is not expressed in the bone and highly expressed in skeletal muscle, it suggests that FAM210A likely plays an important role in muscle, which could also affect bone regulation. The authors here provide further evidence of an important role for FAM210A in diseases affecting muscle function by demonstrating that the expression of FAM210A decreases with age and in patients affected by Pompe disease, Duchenne muscular dystrophy and hereditary recessive myopathy. FAM210A is a mitochondria-localized protein and given the crucial role of mitochondria in supporting muscle metabolism, elucidating the molecular function of FAM210A may provide important insights into diseases biology that could lead to the development of therapeutic approaches. Thus, a significant protein and regulatory pathway are explored in this study that can potentially impact human health. In this manuscript, the authors provide compelling evidence of the importance of Fam210a in muscle homeostasis with their newly generate mouse model. The experiments looking at muscle physiology, function and metabolism are well-executed and for the most part rigorous, which are the strengths of this manuscript. However, the conclusion that Fam210a deletion in skeletal muscle induces the hyper-acetylation of several small ribosomal proteins, which leads to ribosomal disassembly and translational deficiency is not supported by the data presented here. As noted in the comments above, these experiments need major improvement. Additionally, there are other concerns about general scientific rigor and conclusions inconsistent with the data presented as also noted in the comments section.

Advance: Although a previous study explored the role of FAM210A using a skeletal muscle-specific KO induced at postnatal 28 days under a HSA promoter, the model used by the authors here provide a cleaner approach and more insights into the molecular functions of FAM210A in muscle physiology. The findings that Fam210a MKO disrupts the flow of TCA cycle, which leads to an abnormal accumulation of acetyl-CoA is interesting and provide new conceptual advance on the roles of FAM210A in mitochondria function in muscle. Acetyl-CoA production is an important source of acetyl-group that can be transferred to proteins and regulate gene expression programs. Thus, this is an important finding. However, molecular mechanism by which FAM210A regulates this process through an interaction with SUCLG2 is not provided and the nature this interaction is superficially explored.

Audience: Findings from this manuscript are likely to interest both basic research and translational/clinical audiences as it explores the physiological and molecular function of a disease-linked protein. The findings are also likely to impact the fields of metabolism, mitochondria function and regulation of gene expression by protein acetylation (if concerns raised regarding these experiments are addressed). The fields of expertise of this reviewer are protein and RNA modifications, ribosome biogenesis and mRNA translation.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In the manuscript entitled "FAM210A mediates an inter-organelle crosstalk essential for protein synthesis and muscle growth in mouse", Chen et al, found that knocking out of FAM210A specifically in muscle using Myl Cre resulted in abnormal mitochondria, hyperacetylation of cytosolic proteins, and translation defects. The manuscript uncovered the new functions of FAM210A in regulating metabolism and translation. I have the following the concerns about the manuscript.

Comments

1. One of the major phenotypes of FAM210A is the decrease of muscle mass after 6 weeks after birth. Is this phenotype caused by the accumulation of progressive loss of muscle mass from birth? Are the body weight and muscle mass reduced in FAM210A knocking out new born mice? Is the muscle mass growth curve the same in FAM210A and WT mice from birth to 6 weeks after birth? These results will reveal more mechanism of FAM210A mediated muscle mass control.
2. Does the muscle mass continue to decrease after 8 weeks?
3. FAM210A knockout mice displayed high lethal rate. Is there any potential mechanism for the high lethality?
4. In Figure 2, the muscle mass decreased significantly, while the fat mass only decreased slightly. In FAM210A knockout mice. However, the ratio of the lean mass and fat mass to body mass did not change in FAM210A knockout mice compared to WT mice. How do the authors reconcile this?
5. Are there changes of the number of nuclei per myotube? Is the muscle atrophy in FAM210A knockout mice caused by the defects of fusion, or the degradation of protein, or both?
6. Are the growth curves of muscle mass growth in EDL and SOL the same n FAM210A knockout mice?
7. The oxygen consumption and carbon dioxide production are higher in FAM210A knockout mice, suggesting a high metabolism rate. In contrast, the heat production of FAM210A knockout mice is lower, suggesting a low metabolism rate. Any explanation?
8. Given the high glucose consumption in FAM210A, why is the clearance rate of blood glucose low?
9. Are there any changes of the abilities for the FAM210A knockout mice in running endurance?
10. In page 5, the last sentence of the 2nd paragraph, the authors concluded "There results suggest that Fam210aMKO induces a metabolic switch to a more oxidative state." It is better to describe it as muscle metabolic since the whole body metabolism has not been carefully examined.
11. In Fig. 6, what is the link between increased transcription level of Fgf21 and the elevated level of aberrant acetylation of proteins?
12. Is there any link between the increased acetylation level of rebolsome proteins and the translation defects?
13. How do the abnormal mitochondria lead to increased protein acetylation? And how do these defects further cause translation problem?
14. The defects in translation will cause general problems besides mitochondria defects. Are there any phenotypes related to the overall translation inhibition observed? If not, why?
15. Are the abnormal mitochondria, increased protein acetylation, and translation inhibition observed in 2-6 weeks old mice? When were these defects first found? Are they correlated with muscle atrophy?

Significance

This manuscript described many interesting phenotypes of Fam210a knockout mice. However, the links between these phenotypes are obscure. The logic of the manuscript will be greatly improved if the authors could provide explanations to logically link the phenotypes.

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

Learn more at Review Commons

---

Reply to the reviewers

__Reviewer #1 (Evidence, reproducibility and clarity (Required)):____ __ Deka and colleagues report that a non-canonical NFKb signaling operates in DCs in the context of inflammation and inhibits a tolerogenic mechanism driven by b-catenin-Raldh2. The following comments are made to clarify the findings presented.

1. The authors re-analyzed published scRNAseq data from DSS colitis to identify the expression of Relb and NFKB2 in myeloid cells. 1a- The authors are encouraged to expand this analysis to other published datasets.

We sincerely appreciate the comment from the knowledgeable reviewer. Unfortunately, we did not find any other publicly available scRNAseq dataset from DSS-treated mice. To circumvent this problem, we instead examined a previously published microarray-based bulk transcriptomic dataset obtained using FACS-sorted DCs isolated from the mouse colon (GSE58446, Muzaki et al. 2016; doi:10.1038/mi.2015.64). We consistently found an increased expression of Relb, and to some extent Nfkb2, mRNAs in intestinal DCs upon DSS treatment (Fig R1). Because microarray analysis lacks the quantitative attributes that scRNAseq offers, we provided this newly analysed dataset for reviewer's eyes and refrained from including this data in the manuscript per se. Of note, we have also provided our own experimental data directly demonstrating p100 processing in DC isolated from colitogenic gut. (Fig 1F)

Importantly, we could identify an additional scRNAseq dataset derived colitogenic human ulcerative colitis patients (GSE162335, Devlin et al. 2021, doi.org/10.1053/j.gastro. 2020.12.030). Interrogation of this dataset indeed confirmed that RELB and NFKB2 mRNAs were majorly expressed in intestinal DCs and not in intestinal macrophages and that IBD was associated with increased expression of multiple RelB-important genes in intestinal DCs. These analyses further supported the notion that heightened non-canonical NF-kB signalling in DCs could be fuelling aberrant gut inflammation. We have now incorporated this newly acquired data in the supplementary Figure S5A-S5C. (line# 456-460)

1b. Additionally, the expression of Relb and NFKB2 in other cells - especially other myeloid cells -should be explored and included, even if then the authors later choose to test their function in DCs.

Adhering to this brilliant suggestion, we have now further interrogated the mouse scRNAseq dataset (GSE148794 ; Ho et al., 2021) to compare macrophages and DCs for the expression of the non-canonical signal transducers. Indeed, we found a relatively insignificant level of Relb and Nfkb2 mRNAs in intestinal macrophages in comparison to intestinal DCs. Our data suggested that the non-canonical NF-kB pathway is likely to play a more prominent role in DCs than in macrophages in the gut. This new analysis has now been presented in the revised draft in Figure 1B. (revised text line#136-140) This comparison indeed proved useful in motivating subsequent in-depth analyses of the non-canonical NF-kB pathway in DCs in the context of experimental colitis.

1c. Please note that there is a transition from Fig 1A to Fig 1B to focus on DCs, which is not apparent from the figure.

Please find our response to #1b.

Please include scale bars for all histological analyses.

We thank the reviewer for alerting us. The scale bars were already included in the histological analyses; we have now appropriately highlighted them in this revised version for better visual clarity.

In Fig S1I, the authors show that loss of body weight upon DSS treatment in Nfkb2DCD11c is indistinguishable from control. Why is the starting weight at 110%? Please clarify.

We sincerely apologize for this inadvertent error. We have now rectified the axis label, representing the starting weight at day 0 as 100% (currently Figure S1J).

In Figure 2, please indicate the database/s used for the identification of top biological pathways.

We used "WikiPathways subset of cellular processes" available at www.gsea-msigdb.org/gsea/msigdb/mouse/collections.jsp?targetSpeciesDB=Mouse#M8 for the pathway enrichment analysis presented in Figure 2B. We also utilized a previously published RA-target gene set for the gene set enrichment analysis presented in Figure 2C (Balmer and Blomhoff, 2002; 10.1194/jlr.r100015-jlr200). While this information was included in the materials and methods section in the original draft, we have now included these descriptions in the figure legend for further clarity. (please see revised figure legends 2B and 2C)

The authors show a more significant expansion in Tregs upon DSS treatment when non-canonical NFKb is ablated in DCs. Is this at the expense of a reduction of specific Th cells? Can the authors also report the number of cells beyond the % of cells?

In response to the reviewer's comment, we have examined the abundance of Th17 cells in the colon of our knockout mice. As also observed earlier upon DC-specific ablation of NIK function (Jie et al., 2018), disruption of non-canonical NF-kB signaling in DCs in RelbDDC or Nfkb2DDC mice led to a reduced frequency of RoRgt+ Th17 cells in the LP (Figure S3F, new data). Our in vitro (Figure 2I) and in vivo (Figure 3F-G, 6B-6C) studies conclusively linked DC-intrinsic non-canonical NF-kB signaling to intestinal Treg via the RA pathway. Therefore, we conjectured that the observed decline in the Th17 compartment in our knockouts could be secondary to Treg expansion. We have now further discussed this point in the revised manuscript. (line#296-300)

As the reviewer suggested, in addition to the Treg frequency, we have also presented the number of intestinal FoxP3+ CD4 T cells in the supplement (Figure S3E). Our data revealed a similar increase in the total Treg numbers in the mouse colon upon ablation of the non-canonical NF-kB pathway in DCs. (line#296-300)

In figure 6A, it appears that not only the amount of beta-catenin expressed but also the percentage of beta-catenin positive MNL DCs is significantly expanded upon ablation of non-canonical NFkb. Please verify and if so, include.

We thank the reviewer for this very insightful comment. We have now catalogued MLN DCs into b-cateninlow and b-cateninhigh compartments. Indeed, we found a substantial more than two-fold increase in the frequency of b-cateninhigh DCs in RelbDDC mice. Accordingly, we have revised Figure 6A and emphasised this point in the text.

(line#428-430)

In analogy to comment #1 above, please expand the analyses in human samples to include the expression of Relb and Nfkb2 to other myeloid cells.

Adhering to the valuable suggestion by the reviewer, we have now analysed the scRNAseq dataset (SCP 259) comparing DCs, macrophages, and inflammatory and cycling monocytes present in the human gut for the expression of RELB and NFKB2 mRNA (Figure 7B). Consistent with our observation involving the mouse colon, we found that mRNAs encoding these non-canonical signal transducers were mostly expressed in DCs among various MNPs. This point has also been emphasized in the revised draft. (line#446-448)

Reviewer #1 (Significance (Required)):

Strengths of the manuscript include the conceptual novelty of the intersection between non-canonical NFkb and the tolerogenic b-catenin-Raldh2 axis. And additional strength is the methodical approach, which includes various immunological and biochemical assessments as well as genetic perturbations to dissect such relationships. While it remains unknown the relevant triggers for the non-canonical axis described, this study advances our mechanistic understanding on how activation of this axis overrides regulatory mechanisms in DCs. As such, this manuscript should be of broad interest to immunologists and in particular mucosal immunologists. We sincerely thank the reviewer for lauding our work as conceptually novel and methodical. The encouragement from the knowledgeable reviewer would certainly motivate us further to identify the relevant trigger of this pathway in the gut.

---

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

The following issues are noted.

- all animal strains and their provenance should be described and properly referenced (for example, there are at least two CD11c-Cre strains with different specificity). Along the same lines, the specificity of Cre recombination should be confirmed, at least in major cell types (DCs vs T effector or regulatory cells).

We sincerely appreciate the reviewer's attention to these important details. We would like to point out that in our original draft, Table 1 in the Materials and Methods section provides information on the source and the identifier of all mouse strains used. In particular, we utilized CD11c-Cre mice with the identifier 008068 from the Jackson Laboratories. Alternately known as B6.Cg-Tg (Itgax-cre)1-1Reiz/J, this strain displays Cre-mediated recombination in more than 95% of conventional DCs while exhibiting only minor recombination in lymphocytes (low-activated T cells (www.jax.org/strain/008068). Importantly, our immunoblot analyses revealed efficient depletion of RelB in specifically splenic CD11c+ cells of RelbDDC mice with only a negligible reduction in CD11c- cells (Figure S1E). Our analyses involving Nfkb2DDC mice also assured of similar gene disruption specificity (Figure S1H). Notably, our results were consistent with those documented on RelbDDC and Nfkb2DDC strains earlier (Andreas et al., 2019). To further address the reviewer's concern pertaining to the T-cell compartment, we have now compared splenic CD4+ cells from Nfkb2fl/fl and Nfkb2DDC mice for the expression of p100 (Figure S1I, newly added in the revised draft). Our results confirmed that CD11c-Cre-driven ablation of the non-canonical NF-kB pathway did not perturb p100 expressions in T cells. Taken together, these allow us to emphasize that knockout phenotypes observed in our study were attributed to non-canonical NF-kB deficiency in DCs. We have accordingly modified the text to highlight gene deletion specificities in our knockouts. (line#166-167, 180-182)

- the DSS model is prone to "batch effects" of individual cages, and proper comparison between genotypes is possible only if mice of different genotypes (eg littermates) are housed together in the same cages. The authors should clearly confirm whether this was the case, and if not, key experiments should be repeated in this setting.

As mentioned in the materials and methods section of the original draft, littermate male mice of indicated genotypes were indeed cohoused for at least one week prior to experiments. We have now further emphasised this point in the legend of Figure 1.

- BMDCs represent a heterogeneous mixture of DCs and macrophages (Helft et al., Immunity 2015). These populations should be clearly defined and compared between genotypes, to make sure that they do not underlie the observed gene expression differences.

The knowledgeable reviewer has raised a very pertinent issue. We would like to emphasize that instead of generating BMDCs using GM-CSF alone following the protocol prescribed by Helft et al. (2015), we differentiated bone marrow cells to BMDCs using a cocktail of GM-CSF+IL4 adhering to the protocol published by Jin and Sprent (2018). Following the reviewer's suggestion, we have now compared BMDCs generated in these two protocols in our laboratory. As reported earlier (Jin and Sprent, 2018), unlike BMDCs generated using GM-CSF alone, BMDCs generated using the GM-CSF+IL4 cocktail did not contain CD115high macrophage-like cells (Figure S2C). (line#230-232) However, they displayed equivalent expressions of the DC marker CD135 on their surface. Moreover, when we compared BMDCs derived from Relbfl/fl and RelbDCD11c mice in flow cytometry analyses, we found comparable surface expression of CD135, assuring intact BMDC generation from bone marrow cells ex vivo in spite of the absence of RelB (Figure S2I). (line#266-268) These studies argue that macrophage-like cells did not contribute to the observed gene expression differences between WT and RelB-deficient BMDCs.

• the analysis of DCs in mutant strains (e.g. in Fig. 3) would benefit from a better definition of populations, e.g. resident vs migratory DCs in the MLN, the Notch2-dependent CD103+ CD11b+ DCs in the LP and MLN, etc. Again, this would be important to justify differences in gene expression (e.g. Fig. 3D).

We sincerely appreciate the comment from the knowledgeable reviewer. In a landmark paper from Prof. Fiona Powrie's group (Coombes et al., 2007), it was earlier demonstrated that CD103+ DCs present in the intestine migrate to local MLNs and play a key role in producing RA and supporting Tregs. While our BMDC data strongly supported a cell-intrinsic mechanism underlying Raldh2 upregulation upon non-canonical NF-kB deficiency (Figure 2F), our in vivo studies (Figure 3D-3E) did not entirely rule out also a possible expansion of RA-producing CD103+ DC compartment in our knockout mice. Although the proposition that non-canonical NF-kB signaling regulates the generation of specific intestinal DC subsets seems attractive, we must point out that previous studies showed a relatively unaltered frequency of CD103+ cells among steady-state migratory DCs in skin-draining lymph nodes (Döhler et al., 2017). Nevertheless, following the reviewer's suggestion, we now plan to perform advanced flow cytometry analyses to compare Relbfl/fl and RelbDCD11c mice for the frequency of CD103+CD11b-, CD103+CD11b+ and CD103-CD11b+ DCs in the intestine. To this end, we have already optimized our experimental protocol for staining intestinal DCs with anti-CD103 antibody (BD Bioscience). In the coming weeks, we are expecting to gather adequate numbers of littermate knockout mice to perform a side-by-side comparison. [NOTE: also the section - "Description of the planned revisions"]

• the analysis of b-catenin protein expression and cellular localization at the single-cell level (e.g. by IF) would greatly strengthen the mechanistic connection between NF-kB and Wnt/b-catenin pathways.

Adhering to the reviewer's suggestion, we have now performed immunofluorescence assay (IFA) to capture the impact of RelB deficiency on b-catenin expression and cellular localization. Because BMDCs pose challenges for IFA owing to their non-adherent nature, we instead examined mouse embryonic fibroblasts (MEFs), which provide for a genetically amenable model cell system. As presented below (Figure R2), our IFA data conclusively demonstrated an increased cellular abundance and nuclear localization of b-catenin in Relb-/- MEFs. While we are truly excited to find that our proposed mechanism is functional in another cell type, we feel that the inclusion of MEF data in the main manuscript, which describes DC-mediated immune controls, may cause significant distractions for the general audience. Accordingly, we have provided this data for the reviewer's eyes only.

Minor: - The reanalysis of previous single-cell data is in Figs. 1 and 7 are much less convincing or exciting than the new experimental data relegated to the supplements. The distribution of the results between main and experimental figures may be reconsidered in this light.

We concur with the knowledgeable reviewer that our scRNAseq analyses may have appeared less convincing in the original draft. In response to comments by reviewer-1 and reviewer-3, we have now added additional data panels (Figure 1B, Figure 7B and Figure 7G) and examined additional publicly available datasets (Figure S5). In the revised draft, these analyses helped us to more firmly establish a link between non-canonical NF-kB signaling in DCs to aberrant intestinal inflammation in mice and humans.

However, we slightly diverge that many key experimental datasets were relegated to the supplement. Except for the FITC-dextran experiment, data from all other experimental analyses were presented in the main text (Figure 1). To suitably manage space in our figure panels, we opted to present quantified data averaged from experimental replicates in the main text while providing representative raw data in the supplement. Besides, immunoblot analyses confirming DC-specific ablation of target genes in our knockouts were placed in the supplement. Notably, these knockout strains were also examined earlier (Andreas et al., 2019). Those studies, along with our own analyses (Figure S1E, S1H and S1I - additional data), confirmed the most efficient gene deletion in CD11c+ cells. While maintaining these data in the supplement for want of space, we have now cited this reference in the main text to emphasize that knockout phenotypes observed in our study were attributed to non-canonical NF-kB dysfunctions in DCs.

Reviewer #2 (Significance (Required)):

The manuscript by Deka et al. explores the role of the non-canonical NF-kB pathway, specifically of its key mediators RelB and NF-kB2, in dendritic cells (DCs) during intestinal inflammation. The key strength of the paper is the demonstration that DC-specific deletion of RelB or NF-kB2 leads to improved acute or chronic DSS colitis. It is also shown that reducing the dose of b-catenin rescues the phenotype of RelB deletion, providing an important genetic connection between NF-kB and Wnt/b-catenin pathways. As such, the work is novel, important and of potential significance to the field.

We express our deepest gratitude to the reviewer for his/her valuable time and insightful comments. We are indeed extremely excited that the knowledgeable reviewer finds our work novel, important and of potential significance to the field. These positive comments would inspire us to look further into potential interventions targeting the non-canonical NF-kB pathway in human ailments.

---

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

Summary:

This manuscript from Deka et al. investigates the role of dendritic cell noncanonical NFκB signaling on intestinal inflammation. Based on prior data showing altered DC function in intestinal inflammation, they interrogated existing scRNAseq data and found that DSS treatment (which yields chemical colitis) increased the expression of non-canonical NFkB family members in dendritic cells. This led to the generation of a DC-specific RelB deficient mouse and use of a DC specific NFkB2 deficient mouse, each of which showed varying degrees of protection from chemical colitis.

Overall, they do a very nice job identifying a mechanism by which noncanonical NFκB signaling in dendritic cells contributes to intestinal inflammation via transcriptional regulation of Axin1, downregulation of β-catenin, restraint of Raldh2 synthesis, impaired retinoic acid synthesis and subsequent decrease in protective Tregs, IgA+ B cells, and microbial dysbiosis. The importance of this pathway is well supported by their focused targeting of β-catenin. After pharmacologic inhibition of β-catenin showed restoration of Raldh2 abundance, they made a DC-specific β-catenin haploinsufficiency RelBDCD11c mouse which showed impaired Raldh2 activity with restoration of colonic Tregs and fecal sIgA. When challenged with DSS, the protective phenotype seen with the RelBDCD11c was lost and the colitis phenotype returned to that of the Relbfl/fl control, further solidifying the role of β-catenin, Raldh2 and RA on intestinal inflammation. Additionally, the discussion provides a robust mechanistic explanation for the phenotypic differences between the RelB and Nfkb2 genotypes, drawing on the authors' deep knowledge of the non-canonical NFκB pathway.

Major comments:

1. Although they propose a novel mechanism by which dendritic cells can contribute to intestinal inflammation, it is in a model of acute epithelial injury that accentuates the contribution of the innate immune system. Would recommend including a discussion of the limitations of this model.

We most sincerely thank the knowledgeable reviewer for raising this important issue. We argue that while erosive epithelial injury initiates colitis in the DSS model, T cells were shown to aggravate intestinal pathologies, particularly at DSS doses used in our study (Kim et al., 2006; doi: 10.3748/wjg.v12.i2.302). Furthermore, our chemically-induced colitis model offered a convenient tool for genetically dissecting the DC-intrinsic role of the non-canonical NF-kB pathway in the intestine. However, we agree entirely that no single animal model fully captures the clinical complexities of human IBD and that other models of experimental colitis should also be employed in the future to assess the generalisability of the proposed DC mechanism in regulating intestinal inflammation. In particular, future studies ought to examine composite knockout strains in the T-cell transfer model of experimental colitis to establish further the role of non-canonical NF-kB signaling in DCs in alleviating intestinal inflammation. As suggested by the reviewer, we have now articulated this point in the discussion section. (line# 553-557)

The human work (Figure 7) shows solid evidence of heightened non-canonical NFκB signaling in DCs via abundance of RELB and NFKB2 along with a few RelB important genes, however, the RA-specific pathway identified in the mouse work is not strongly corroborated by the human data. There is demonstration of one β-activated gene (CCND1) showing decreased expression in IBD patients, however no other gene along with RA pathway was clearly identified to be differentially expressed as one would predict from the mouse work.

We sincerely thank the knowledgeable reviewer for articulating this deficiency in our analyses of single-cell RNA-seq data derived from IBD patients (SCP259, Smillie et al., 2019). We would like to clarify that many well-known b-catenin target genes, including MYC, were not detectable in this dataset. Nevertheless, to address the reviewer's concern, we subjected this dataset to GSEA using a previously published list of RA target genes (Balmer et al., 2002). Our analyses revealed a significant enrichment of RA targets among genes that were downmodulated in DCs derived from inflamed colonic tissues of IBD patients as compared to those from non-inflamed tissues (Figure 7G, newly added in the revised version). We have now discussed this data in the result section. (line#470-475) These studies further substantiated the inverse correlation between noncanonical NF-kB signalling and the RA pathway in DCs in the inflamed human gut.

Minor comments: 1. Their NFκB2DCD11c mouse underwent a regimen of chronic DSS treatment after acute DSS treatment only displayed subtle phenotypic changes. Was the same chronic colitis regimen also tested in the RelBDCD11c ?

Indeed, we also examined RelbDCD11c mice in the chronic DSS treatment regime. As compared to Relbfl/fl mice, these knockout mice displayed significantly less bodyweight changes upon chronic DSS challenge. Because RelbDCD11c mice readily showed acute DSS phenotype, we did not further pursue investigations involving this strain in the chronic DSS settings and rather focused on Nfkb2DCD11c mice to illustrate chronic DSS phenotypes.

In the introduction, it was stated that patients with UC have a marked reduction in intestinal DCs. If DCs (particularly non-canonical NFκB signaling) promote inflammation, how do you explain a decrease in this cell type in patients with active disease?

Depending on the expression of immunogenic or tolerogenic factors, DCs may both promote or subdue inflammation in the colon. We have now revisited the relevant reference published by Magnusson et al., (2016). Indeed, the authors noted a marked reduction in the intestine of the CD103+ DC subset, which has been majorly linked to tolerogenic RA synthesis. While it is generally thought that aberrant inflammation promotes the death of mononuclear phagocytes in the intestine, it seems that either a contraction of the tolerogenic DC compartment or downmodulation of tolerogenic pathways in DCs incites gut inflammation in IBD patients. We have now revised the text in the introduction section to clarify this point. (line#81-83)

The focus on retinoic acid is interesting, however may be oversimplifying the role of non-canonical NFκB in DCs on the mucosal immune system. It must also be mentioned that there is crosstalk between the non-canonical and canonical NFκB signaling systems, for example Nfkb2 is capable of functioning as a IkB protein and inhibiting RelA-p50 (from the last author's prior work - Basak et al, Cell, 2007). Thus would include some mention of possible effects on the canonical system that contribute to intestinal inflammation.

We thank the reviewer for raising this important point. As mentioned in the introduction section of our original draft, the canonical NF-kB pathway in DCs aggravates experimental colitis mice (Visekruna, A. et al. 2015). Indeed, Nfkb2-dependent crosstalk was shown to modulate inflammatory RelA activity in a variety of cell types (Basak et al., 2007; Shih et al., 2009; Chawla et al., 2021). Although such cross-regulatory RelA controls by non-canonical NF-kB signaling are yet to be established in DCs, our studies involving RelB-deficient cells confirmed an essential role of p100-mediated RelB regulations in DC functions. We admit that further studies are required to determine if, independent of RelB, p100 directs immunogenic DC attributes via also RelA or another factor. We have now elaborated on p100-mediated crosstalks in the discussion section. (line#561-562)

In the single-cell DSS data they analyzed, there was a distinct DC population seen with DSS colitis treatment. Although they are categorized as cDC2s, what genes separate them from the other DC populations?

We curated a list of genes from Brown et al., (2019) to categorize cDC1 and cDC2 subsets in our study. We would like to clarify that the list was provided in Supplementary Table 1 in our original draft. In view of the reviewer's comment, we have now referred to this Table in the legend of Supplementary Figure 1 and also in the main text. (line#153)

Why was the RNAseq work on BMDCs (that identified RA metabolism as a top-ranking differentially expressed pathway) done only on Nfkb2-/- BMDCs and not RelB-/-? RelB-/- had a more pronounced protected phenotype in the cell type-specific knockout and is a cleaner target (does not have the IkB capability of Nfkb2).

We broadly agree with the knowledgeable reviewer that comparing WT and Relb-/- BMDCs for global gene expressions could have been worthwhile. We would like to clarify that we initially utilized a dataset derived using Nfkb2-/- BMDCs already available in the laboratory. These analyses were instrumental in developing a notion that non-canonical NF-kB signalling could be modulating Radh2 expression in DCs. Because previous studies involving germline Relb-/- mice suggested a role of RelB in the nonhematopoietic niche in instructing myeloid development (Briseño et al., 2017), we focused our subsequent analyses on BMDCs generated using bone marrow cells from cell type-specific knockouts. Indeed, we could confirm elevated Raldh2 expressions in BMDCs generated from both RelbDDC and Nfkb2DDC mice. Taken together, our studies suggested that Nfkb2-encoded p100 controlled Raldh2 expressions in DCs by providing RelB:p52 and less so as a regulator of the RelA activity. Although we admit that further studies are required to determine if, independent of RelB, p100 directs immunogenic DC attributes via also RelA or another factor. We have now deliberated this point in the discussion section. (line#566-567)

Reviewer #3 (Significance (Required): As a physician-scientist who clinically cares for patients with inflammatory bowel disease, and scientifically studies signaling within innate immune cells, this manuscript does a rigorous job of identifying a mechanism by which canonical NFκB signaling in dendritic cells contributes to intestinal inflammation. This study would be very informative for both basic and translational researchers as it identifies a clear pathway by which the innate immune system contributes to intestinal inflammation, and opens up room for inquiry into triggers of non-canonical NFκB in IBD and modulation of the RA pathway as a potential novel therapeutic target.

We are humbled that the knowledgeable reviewer finds our work to be informative for basic and translational research. These encouragements would undoubtedly motivate us further to identify the relevant trigger of this pathway in the gut and explore potential interventions.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary:

This manuscript from Deka et al. investigates the role of dendritic cell noncanonical NFκB signaling on intestinal inflammation. Based on prior data showing altered DC function in intestinal inflammation, they interrogated existing scRNAseq data and found that DSS treatment (which yields a chemical colitis) increased expression of non-canonical NFkB family members in dendritic cells. This led to the generation of a DC specific RelB deficient mouse and use of a DC specific NFkB2 deficient mouse, each of which showed varying degrees of protection from chemical colitis. Overall, they do a very nice job identifying a mechanism by which noncanonical NFκB signaling in dendritic cells contributes to intestinal inflammation via transcriptional regulation of Axin1, downregulation of β-catenin, restraint of Raldh2 synthesis, impaired retinoic acid synthesis and subsequent decrease in protective Tregs, IgA+ B cells, and microbial dysbiosis. The importance of this pathway is well supported by their focused targeting of β-catenin. After pharmacologic inhibition of β-catenin showed restoration of Raldh2 abundance, they made a DC specific β-catenin haploinsufficiency RelBCD11c mouse which showed impaired Raldh2 activity with restoration of colonic Tregs and fecal sIgA. When challenged with DSS, the protective phenotype seen with the RelBCD11c was lost and the colitis phenotype returned to that of the Relbfl/fl control, further solidifying the role of β-catenin, Raldh2 and RA on intestinal inflammation. Additionally, the discussion provides a robust mechanistic explanation for the phenotypic differences between the RelB and Nfkb2 genotypes, drawing on the authors' deep knowledge of the non-canonical NFκB pathway.

Major comments:

1. Although they propose a novel mechanism by which dendritic cells can contribute to intestinal inflammation, it is in a model of acute epithelial injury that accentuates the contribution of the innate immune system. Would recommend including a discussion of the limitations of this model.
2. The human work (Figure 7) shows solid evidence of heightened non-canonical NFκB signaling in DCs via abundance of RELB and NFKB2 along with a few RelB important genes, however the RA specific pathway identified in the mouse work is not strongly corroborated by the human data. There is demonstration of one β-activated gene (CCDN1) showing decreased expression in IBD patients, however no other gene along with RA pathway was clearly identified to be differentially expressed as one would predict from the mouse work.

Minor comments:

1. Their NFκB2CD11c mouse underwent a regimen of chronic DSS treatment after acute DSS treatment only displayed subtle phenotypic changes. Was the same chronic colitis regimen also tested in the RelBCD11c ?
2. In the introduction, it was stated that patients with UC have a marked reduction in intestinal DCs. If DCs (particularly non-canonical NFκB signaling) promote inflammation, how do you explain a decrease in this cell type in patients with active disease?
3. The focus on retinoic acid is interesting, however may be oversimplifying the role of non-canonical NFκB in DCs on the mucosal immune system. It must also be mentioned that there is crosstalk between the non-canonical and canonical NFκB signaling systems, for example Nfkb2 is capable of functioning as a IkB protein and inhibiting RelA-p50 (from the last author's prior work - Basak et al, Cell, 2007). Thus would include some mention of possible effects on the canonical system that contribute to intestinal inflammation.
4. In the single cell DSS data they analyzed, there was a distinct DC population was seen with DSS colitis treatment. Although they are categorized as cDC2s, what genes separate them from the other DC populations?
5. Why was the RNAseq work on BMDCs (that identified RA metabolism as a top ranking differentially expressed pathway) done only on Nfkb-/- BMDCs and not RelB-/-? The RelB-/- had a more pronounced protected phenotype in the cell type specific knockout, and is a cleaner target (does not have the IkB capability of Nfkb2).

Significance

As a physician scientist who clinically cares for patients with inflammatory bowel disease, and scientifically studies signaling within innate immune cells, this manuscript does a rigorous job of identifying a mechanism by which canonical NFκB signaling in dendritic cells contributes to intestinal inflammation. This study would be very informative for both basic and translational researchers as it identifies a clear pathway by which the innate immune system contributes to intestinal inflammation, and opens up room for inquiry into triggers of non-canonical NFκB in IBD and modulation of the RA pathway as a potential novel therapeutic target.

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 following issues are noted.

• all animal strains and their provenance should be described and properly referenced (for example, there are at least two CD11c-Cre strains with different specificity). Along the same lines, the specificity of Cre recombination should be confirmed, at least in major cell types (DCs vs T effector or regulatory cells).
• the DSS model is prone to "batch effects" of individual cages, and proper comparison between genotypes is possible only if mice of different genotypes (eg littermates) are housed together in the same cages. The authors should clearly confirm whether this was the case, and if not, key experiments should be repeated in this setting.
• BMDCs represent a heterogeneous mixture of DCs and macrophages (Helft et al., Immunity 2015). These populations should be clearly defined and compared between genotypes, to make sure that they do not underlie the observed gene expression differences.
• the analysis of DCs in mutant strains (e.g. in Fig. 3) would benefit from better definition of populations, e.g. resident vs migratory DCs in the MLN, the Notch2-dependent CD103+ CD11b+ DCs in the LP and MLN, etc. Again, this would be important to justify differences in gene expression (e.g. Fig. 3D).
• the analysis of b-catenin protein expression and cellular localization at single-cell level (e.g. by IF) would greatly strengthen the mechanistic connection between NF-kB and Wnt/b-catening pathways.

Minor:

• the reanalyses of previous single-cell data in Figs. 1 and 7 are much less convincing or exciting than the new experimental data relegated to the supplements. The distribution of the results between main and experimental figures may be reconsidered in this light.

Significance

The manuscript by Deka et al. explores the role of the non-canonical NF-kB pathway, specifically of its key mediators RelB and NF-kB2, in dendritic cells (DCs) during intestinal inflammation. The key strength of the paper is the demonstration that DC-specific deletion of RelB or NF-kB2 lead to improved acute or chronic DSS colitis. It is also shown that reducing the dose of b-catenin rescues the phenotype of RelB deletion, providing an important genetic connection between NF-kB and Wnt/b-catenin pathways. As such, the work is novel, important and of potential significance to the field.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Deka and colleagues report that a non-canonical NFKb signaling operates in DCs in the context of inflammation and inhibits a tolerogenic mechanism driven by b-catenin-Raldh2. The following comments are made to clarify the findings presented.

1. The authors re-analyzed published scRNAseq data from DSS colitis to identify the expression of Relb and NFKB2 in myeloid cells.
   • a. The authors are encouraged to expand this analysis to other published datasets.
   • b. Additionally, the expression of Relb and NFKB2 in other cells - especially other myeloid cells -should be explored and included, even if then the authors later choose to test their function in DCs.
   • c. Please note the there is a transition from Fig 1A to Fig1B to focus on DCs, which is not apparent from the figure.
2. Please include scale bars for all histological analyses.
3. In Fig S1I, the authors show that loss of body weight upon DSS treatment in Nfkb2deltaCD11c is indistinguishable from control. Why is the starting weight at 110%? Please clarify.
4. In Figure 2, please indicate the database/s used for identification of top biological pathways.
5. The authors show a more significant expansion in Tregs upon DSS treatment when non-canonical NFKb is ablated in DCs. Is this at the expense of a reduction of specific Th cells? Can the authors also report the number of cells, beyond the % of cells?
6. In figure 6A, it appears that not only the amount of beta-catenin expressed, but also the percentage of beta-catenin positive MNL DCs is significantly expanded upon ablation of non-canonical NFkb. Please verify and if so, include.
7. In analogy to the comment #1 above, please expand the analyses in human samples to include the expression of Relb and Nfkb2 to other myeloid cells.

Significance

Strengths of the manuscript include the conceptual novelty of the intersection between non-canonical NFkb and the tolerogenic b-catenin-Raldh2 axis. And additional strength is the methodic approach, which includes various immunological and biochemical assessments as well as genetic perturbations to dissect such relationships. While it remains unknow the relevant triggers for the non-canonical axis described, this study advances our mechanistic understanding on how activation of this axis overrides regulatory mechanisms in DCs. As such, this manuscript should be of broad interest to immunologists and in particular mucosal immunologists.

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

Learn more at Review Commons

---

Reply to the reviewers

We thank all the reviewers for their comments on our manuscript. We have attempted to address all the points raised by the reviewers and are happy to note that the manuscript is significantly strengthened with the additional experiments that we have performed and from significant restructuring of the manuscript.

Reviewer #1

Major Comments

1. The choice of cells looks confusing. Drosophila are indeed widely used in research of neurodegeneration mechanisms, since they well reflect the behavioral characteristics of a wide range of brain diseases, but why authors used insect immune cells to study the effect of mHTT on cellular processes? Huntington's disease has a well-established site of origin, in the spiny neurons of the striatum, and they certainly have a different protein context than in insect cells. __Author's response: __We thank the reviewer for this comment. Patients with Huntington's disorder display a variety of symptoms affecting peripheral, non-neuronal cells, including alterations in the function of immune cells. Hemocytes isolated from Drosophila expressing pathogenic forms of Huntingtin also display altered immune responses. Through our manuscript we explore the effect of Huntingtin aggregates on cellular functions of hemocytes. Additionally, we have now included data showing that we are able to observe similar phenotypes in mammalian cells such as neuronal SHSY5Y and HEK293T (Supp. Fig. 3). This is indicative of similar effects being exerted by Huntingtin aggregates across cell types and organisms. Finally, we demonstrate that we are able to rescue neurodegeneration in the fly eye upon overexpression of either Hip1 or components of the Arp2/3 complex (Fig. 4F), further solidifying our results that Huntingtin aggregates alter CME in an actin-dependent manner and that this largely is responsible for the toxicity. This validates our observations that effects on CME appear to be independent of cell type and that non-neuronal cells such as hemocytes can also be used to study the effects of pathogenic aggregates.

The interrelationship between mutant huntingtin and actin cytoskeleton and clathrin-mediated endocytosis that are convincingly demonstrated in other earlier studies in the m/s are described in rather morphological level and there is no description of molecular interactions of proteins belonging to three systems considered, Htt (control vs mutant), actin cytoskeleton and CME. Lack of these data renders the morphological observations unsupported

__Author's response: __Previous data shown in Hosp et al, 2017 indicates that a large number of proteins involved in both actin remodelling and clathrin mediated endocytosis are sequestered within Huntingtin aggregates. While the mechanism of sequestration remains unknown, is has also been observed that loss of Huntingtin results in altered organization of the actin cytoskeleton. We have now added points discussing this in the results section.

Three last figures of total eight demonstrate the effect of proteins, responsible for the initiation of certain neurodegenerative pathologies, on the activity of clathrin-mediated endocytosis, and on the properties of actin cytoskeletal system, however neither in the abstract nor in the introduction there is no any word about these proteins; in the discussion only a few words are devoted to one of these proteins TDP-43. When starting the article, did the authors plan to enter this data into the manuscript?

__Author's response: __We have now amended this by revising the abstract and the text.

It is important to work on the style of the manuscript, the article is difficult to read, it is a collection of data that does not seem related to each other.

__Author's response: __We have reorganized the manuscript and have improved on the flow to make it easier for the reader. We apologize for the rather tedious and confusing flow in the previous draft.

Reviewer #2

This manuscript endeavours to explore the link between mutant Huntingtin, clathrin-mediated membrane transport and the actin cytoskeleton: both its dynamics and overall mechanics. As I read it, it carries the interesting idea that pathogenic protein aggregates alter actin cytoskeletal dynamics by sequestering Arp2/3 nucleator. This has two consequences in the authors' experiments: disruption of clathrin-coated vesicle movement and an increase in cellular stiffness. An interesting question is whether these two effects are related: Is the disruption of vesicular movement due to the change in cytoplasmic stiffness? Or could they be features that both reflect the underlying change in actin dynamics. This may be hard to tease apart and beyond the purview of this manuscript.

I have some suggestions that could strengthen the MS.

Major Comments

1. Further characterizing Arp2/3 sequestration. The notion seems to be that actin nucleators would be sequestered (and inactivated) by mutant protein aggregates, as supported by co-localization studies. In addition, could the authors:

2. a) Test if the dynamics of Arp2/3 are altered, comparing e.g. Arp3-GFP FRAP in the aggregates vs that elsewhere. Author's response: We indeed attempted the FRAP experiment. However due to some technical difficulties we were not convinced by the extent of FRAP in the transgenic fly line. It appeared as an artifact and we were not comfortable including the data in the manuscript. We have instead provided example files for the reviewer to examine.

3. b) Test more directly if actin nucleation is altered in cells that have pathogenic mutant aggregates. This could be done by barbed-end labelling (e.g. measuring incorporation of labelled actin in live cells that are lightly permeabilized with saponin). __Author's response: __We have performed barbed-end labeling for HTT Q15 and HTT Q138 expressing cells. Images and quantification have now been added to the revised manuscript as Figures 2H and 2I. While this was a challenging experiment, it was deeply satisfying to observe such dramatic changes indicating a change in the state of the actin cytoskeleton.

Does manipulating actin nucleation alter cellular mechanics as it does for clathrin-coated vesicle transport? For example, does inhibition of Arp2/3 (e.g. with CK666) increase cellular stiffness and would stiffness be amelioriated in mutant cells if Arp3 is overexpressed?

__Author's response: __We have used LatA to look at whether alteration in the actin cytoskeleton affects cellular stiffness. We found that disruption of the actin cytoskeleton leads to a decrease in cellular stiffness in WT as well as in HTT Q138 expressing cells (shown in Figure 5 and discussed in the results section). We have also now performed AFM on CK666 treated cells and showed that treatment of CK666 leads to a decrease in cellular stiffness similar to LatA treated cells. This further strengthens our hypothesis that a 'Goldilocks' state of actin remodeling and consequently cellular stiffness is required for CME to proceed. We have not performed AFM on cells overexpressing Arp2/3 in HTT Q138 background. However, we believe that it will rescue cellular stiffness as overexpression of Arp2/3 rescues filopodia formation in HTT Q138 expressing cells (Figure 4E) as well as neurodegeneration. AFM data obtained from CK666 treated cells is now added in Supplementary figure 8.

Although it may be difficult to determine if the defect in vesicle transport is due to the change in rheology, I wonder if the authors could reinforce their analysis by showing the overall relationship between the two features. It would be interesting if they could plot CCV velocity against elasticity for all the various conditions that they have tested. Would this cumulative analysis be informative?

__Author's response: __This data is already present across the manuscript as part of different figures. We are not sure whether we can reuse the same data to put it as part of a different figure which plots the relationship between elasticity and CCS velocity. We would be grateful for advice on whether this is allowed and how to mention that the data is also part of different figures.

Focus of the MS. I think that the MS is a little longer and more discursive than it needs to be. I rather struggled to find the focus of the story (which could well be me). There is a deal of repetition that could be profitably cut (the reader may actually find it easier to follow). As well, some anticipation and summaries could be shortened. The final paragraph of the introduction largely summarizes the paper; it could be shortened quite considerably, so that the reader can get directly into the Results themselves. Similarly, the final paragraph of the results is a summary which could work better elsewhere - perhaps, e.g. at the beginning of the discussion.

__Author's response: __We have now trimmed and rearranged the text in the manuscript. We have reorganized the manuscript and have improved on the flow to make it easier for the reader. We apologize for the rather tedious and confusing flow in the previous draft. We are open to further suggestions to improve the writing style.

Specific points

i) Fig 3E. The changes in F-actin flow revealed by PIV are quite dramatic. How reproducible are these changes. (The data presented were from single cells?) __Author's response: __ Changes in F- actin flow obtained from PIV analysis (now figure 2J, 7E in MS) were performed on atleast 5 cells of each type, and the results were observed to be consistent across all. The representative figure is a true representative of the data observed.

1. ii) If TPD43, does it also affect Arp2/3? Author's response: __We thank the reviewer for this comment. Unfortunately, __we could not perform this experiment due to the unavailability of a fluorescently tagged TDP43 fly line which which would enable us to visualize whether Arp3 was sequestered within the aggregates.

Minor points:

1. a) Fig 5a, b - why change the order on the x-axis? __Author's response: __We have fixed this now. We have removed figure 5b, since, in the revised MS we are only talking about stiffness instead of viscoelastic properties of the cells.

Reviewer #2

Overall, I think that the significance of the MS lies in its evidence that sequestration of actin nucleators may be a key effect of mutant protein aggregation, with implications for cellular function. This would provide a useful conceptual framework to understand the cell biological consquences of creating pathogenic protein aggregates.

__Reviewer #3 __

Summary

In the paper, the authors showed that huntingtin aggregates, which play a critical role in initiating neurodegenerative diseases, impair clathrin-mediated endocytosis (CME). Using live cell imaging and AFM, the authors demonstrated that CME is affected by the alteration in actin cytoskeletal organization and cellular viscosity. Further, the authors concluded that there was a strong link between dynamic actin organization and functional CME in the context of neurodegeneration. While the data is interesting and novel, the study in its current form needs major revision before it is accepted.

Major comments:

1) Figure 2: The authors should show the compromised actin cytoskeleton structure after Lat A and cytoD treatment to back up the findings.

__Author's response: __We have included the representative micrographs of compromised cytoskeleton in terms of filopodia formation upon treatment of LatA and CytoD in Supplementary figure 3E.

2) Figure 2g and 2h: Quantification data of filopodia must be supported with representative images.

__Author's response: __Figure number has been changed to 2D and 2E. Representative image for the quantification of filopodia has been now included in supplementary figure 3D.

3) RNAi studies must be performed using control siRNA to check off-target effects.

__Author's response: __Luc VAL10 was used as a control for all the RNAi experiments. However, data for RNAi is not shown as the phenotype for Luc VAL10 was comparable to WT. We have included Luc VAL10 as a control for Profilin RNAi in the FRAP experiment (Supplementary figure 4C).

4) The result section needs to be reorganized to maintain flow. In the current format, the results of a similar set of experiments are spread across different figures, making it a bit difficult to understand.

__Author's response: __We apologize for the inconvenience. This issue has been addressed now.

5) Figure 3d: The expression level and spatial distribution of HTTQ138 transfection were not convincing compared to the httQ15 expression level and the distribution.

Author's response Figure 3D (Figure 3A in this MS) shows the data obtained from hemocytes isolated from third instar larva of the same age. These are not transfected cells and are obtained from Drosophila larvae using the same Gal4 driver, Cg-Gal4. Thus, the level of expression will be same. However, the distribution may show a change due to the aggregating nature of HTT Q138, while HTT Q15 is non-aggregating and therefore remains diffused.

6) Suppl Fig 2a data must be supported using images showing myosin VI distribution in wild-type vs. HTTQ138 transfected cells.

__Author's response: __This data (Supplementary figure 4D in present MS) has been obtained from genetic knockdown of myosin VI. The aim of the experiment was to show that we see similar effects on CCSs movement as we see upon disruption of the actin cytoskeleton.

7) Suppl movie videos are not labeled correctly in the source. It is not possible to locate them and know which videos are referred to in the manuscript.

__Author's response: __We apologize for the inconvenience. This issue has been fixed now.

8) Page 8: How do HTT aggregates sequester the actin-binding proteins? An explanation should be provided in the result section.

__Author's response: __Previous data shown in Hosp et al, 2017 indicates that a large number of proteins involved in both actin remodelling and clathrin mediated endocytosis are sequestered within Huntingtin aggregates. While the mechanism of sequestration remains unknown, the types of proteins involved in actin remodelling are diverse and do not represent specific types or classes. We have now added points discussing this in the results section.

9) Page 10: The authors concluded that "increasing the availability of proteins involved in actin reorganization is capable of restoring CME even in the presence of pathogenic aggregates." Since several actin-associated proteins are involved in actin reorganization, which types/classes of proteins are involved in CME restoration? The authors should expand it in the discussion.

__Author's response: __As we have only investigated the roles of Hip1 and the Arp2/3 complex, we are confident of only reporting their roles in the context of this manuscript. However, previous data shown in Hosp et al, 2017 indicates that a large number of proteins involved in both actin remodelling and clathrin mediated endocytosis are sequestered within Huntingtin aggregates. While the mechanism of sequestration remains unknown, the types of proteins involved in actin remodelling are diverse and do not represent specific types or classes. Therefore this indicates that modulation of actin, through the sequestration of proteins involved in this process is affected in the presence of Huntingtin aggregates. We have added points detailing this in the results and discussion sections.

10) The schematic of the proposed model depicting critical steps by which pathogenic proteins inhibit CME is required. It will help readers to understand the molecular mechanism easily.

Author's response: We have now included a model in the manuscript (Fig. 8B).

Minor:

1) Figure panel referencing in the text needs to be more consistent, for example, fig 3e is referred to before fig 3d., and fig 2 panels are referred to before fig. 3 panels.

__Author's response: __We have reordered the figures and maintained a consistent order throughout.

2) The authors should use similar phrasing throughout the manuscript to avoid confusion. For instance, either use 'HTTQ138' or 'htt Q138'.

__Author's response: __We apologize for this. We have now maintained uniform nomenclature through the text.

3) Page 10: AFM indentation experimental part and its discussion in the result section is unnecessary. Shift it to the 'Materials and Method' section.

__Author's response: __We have now trimmed this portion and we are now only showing elasticity data and not viscoelasticity.

4) This statement looks a bit exaggerated. There is not sufficient evidence to support the statement- "It can be said that the cells in general behave like a soft glass. The presence of aggregates lowers the effective temperature pushing it nearer to the glass transition, affecting transport."

__Author's response: __We have now removed all figures resulting from an analysis that assumes glassy behaviour. Instead, we have now provided a more conventional and well-established analysis to obtain Young's modulus of cells exhibiting different transport properties.

5) Page 12: What is the basis for selecting proteins Aβ-42, FUSR521C, αSynA30P, αSynA53T, and TDP-43 over other proteins? An explanatory sentence must be added to support the selection.

__Author's response: __We have modified the text to clarify this point.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary

In the paper, the authors showed that huntingtin aggregates, which play a critical role in initiating neurodegenerative diseases, impair clathrin-mediated endocytosis (CME). Using live cell imaging and AFM, the authors demonstrated that CME is affected by the alteration in actin cytoskeletal organization and cellular viscosity. Further, the authors concluded that there was a strong link between dynamic actin organization and functional CME in the context of neurodegeneration. While the data is interesting and novel, the study in its current form needs major revision before it is accepted.

Major comments:

1. Figure 2: The authors should show the compromised actin cytoskeleton structure after Lat A and cytoD treatment to back up the findings.
2. Figure 2g and 2h: Quantification data of filopodia must be supported with representative images.
3. RNAi studies must be performed using control siRNA to check off-target effects.
4. The result section needs to be reorganized to maintain flow. In the current format, the results of a similar set of experiments are spread across different figures, making it a bit difficult to understand.
5. Figure 3d: The expression level and spatial distribution of HTTQ138 transfection were not convincing compared to the httQ15 expression level and the distribution.
6. Suppl Fig 2a data must be supported using images showing myosin VI distribution in wild-type vs. HTTQ138 transfected cells.
7. Suppl movie videos are not labeled correctly in the source. It is not possible to locate them and know which videos are referred to in the manuscript.
8. Page 8: How do HTT aggregates sequester the actin-binding proteins? An explanation should be provided in the result section.
9. Page 10: The authors concluded that "increasing the availability of proteins involved in actin reorganization is capable of restoring CME even in the presence of pathogenic aggregates." Since several actin-associated proteins are involved in actin reorganization, which types/classes of proteins are involved in CME restoration? The authors should expand it in the discussion.
10. The schematic of the proposed model depicting critical steps by which pathogenic proteins inhibit CME is required. It will help readers to understand the molecular mechanism easily.

Minor:

1. Figure panel referencing in the text needs to be more consistent, for example, fig 3e is referred to before fig 3d., and fig 2 panels are referred to before fig. 3 panels.
2. The authors should use similar phrasing throughout the manuscript to avoid confusion. For instance, either use 'HTTQ138' or 'htt Q138'.
3. Page 10: AFM indentation experimental part and its discussion in the result section is unnecessary. Shift it to the 'Materials and Method' section.
4. This statement looks a bit exaggerated. There is not sufficient evidence to support the statement- "It can be said that the cells in general behave like a soft glass. The presence of aggregates lowers the effective temperature pushing it nearer to the glass transition, affecting transport."
5. Page 12: What is the basis for selecting proteins Aβ-42, FUSR521C, αSynA30P, αSynA53T, and TDP-43 over other proteins? An explanatory sentence must be added to support the selection.

Significance

• Fundamental study in neurogenerative diseases explaining the mechanobiological aspects of the diseases
• The audience will be interested in knowing the mechanobiological aspects (altered actin cytoskeleton) of the neurogenerative disease.
• Field of reviewer expertise: mechanobiology; mechanotransduction; cell-ECM interactions

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

This manuscript endeavours to explore the link between mutant Huntingtin, clathrin-mediated membrane transport and the actin cytoskeleton: both its dynamics and overall mechanics. As I read it, it carries the interesting idea that pathogenic protein aggregates alter actin cytoskeletal dynamics by sequestering Arp2/3 nucleator. This has two consequences in the authors' experiments: disruption of clathrin-coated vesicle movement and an increase in cellular stiffness. An interesting question is whether these two effects are related: Is the disruption of vesicular movement due to the change in cytoplasmic stiffness? Or could they be features that both reflect the underlying change in actin dynamics. This may be hard to tease apart and beyond the purview of this manuscript.

I have some suggestions that could strengthen the MS.

1. Further characterizing Arp2/3 sequestration. The notion seems to be that actin nucleators would be sequestered (and inactivated) by mutant protein aggregates, as supported by co-localization studies. In addition, could the authors:
   • a) Test if the dynamics of Arp2/3 are altered, comparing e.g. Arp3-GFP FRAP in the aggregates vs that elsewhere.
   • b) Test more directly if actin nucleation is altered in cells that have pathogenic mutant aggregates. This could be done by barbed-end labelling (e.g. measuring incorporation of labelled actin in live cells that are lightly permeabilized with saponin).
2. Does manipulating actin nucleation alter cellular mechanics as it does for clathrin-coated vesicle transport? For example, does inhibition of Arp2/3 (e.g. with CK666) increase cellular stiffness and would stiffness be amelioriated in mutant cells if Arp3 is overexpressed?
3. Although it may be difficult to determine if the defect in vesicle transport is due to the change in rheology, I wonder if the authors could reinforce their analysis by showing the overall relationship between the two features. It would be interesting if they could plot CCV velocity against elasticity for all the various conditions that they have tested. Would this cumulative analysis be informative?
4. Focus of the MS. I think that the MS is a little longer and more discursive than it needs to be. I rather struggled to find the focus of the story (which could well be me). There is a deal of repetition that could be profitably cut (the reader may actually find it easier to follow). As well, some anticipation and summaries could be shortened. The final paragraph of the introduction largely summarizes the paper; it could be shortened quite considerably, so that the reader can get directly into the Results themselves. Similarly, the final paragraph of the results is a summary which could work better elsewhere - perhaps, e.g. at the beginning of the discussion.

Specific points

• i) Fig 3E. The changes in F-actin flow revealed by PIV are quite dramatic. How reproducible are these changes. (The data presented were from single cells?)
• ii) If TPD43, does it also affect Arp2/3?

Minor points

a) Fig 5a, b - why change the order on the x-axis?

Significance

Overall, I think that the significance of the MS lies in its evidence that sequestration of actin nucleators may be a key effect of mutant protein aggregation, with implications for cellular function. This would provide a useful conceptual framework to understand the cell biological consquences of creating pathogenic protein aggregates.

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 "Pathogenic aggregates alter actin organization and cellular viscosity resulting in stalled clathrin mediated endocytosis" by Singh at al attempts to examine the relationship between the mutant huntington aggregates formation, vesicular transport and cytoskeletal rearrangements in cells bearing mHTT aggregates. The authors did a lot of work, used modern methods and obtained a large amount of data, but the purpose of this work remained unclear.

The information that during the aggregates' formation a huge number of cellular proteins playing an important role in cell physiology are involved into aggregates, and that many processes, including vesicle transport, are disrupted in cells carrying mHTT, is not new. Many details of the study, such as demonstration that HIP1 colocalizes with markers of clathrin-mediated endocytosis in neuronal cells and is highly enriched on clathrin-coated vesicles (CCVs) also was already published (e.g. doi: 10.1074/jbc.C100401200). The m/s itself is not well written, and I was not able to understand what specific neurodegenerative process the authors are studying. I believe that the current version of the requires major revisions in its scientific content as well as its writing.

Major:

1. The choice of cells looks confusing. Drosophila are indeed widely used in research of neurodegeneration mechanisms, since they well reflect the behavioral characteristics of a wide range of brain diseases, but why authors used insect immune cells to study the effect of mHTT on cellular processes? Huntington's disease has a well-established site of origin, in the spiny neurons of the striatum, and they certainly have a different protein context than in insect cells.
2. The interrelationship between mutant huntingtin and actin cytoskeleton and clathrin-mediated endocytosis that are convincingly demonstrated in other earlier studies in the m/s are described in rather morphological level and there is no description of molecular interactions of proteins belonging to three systems considered, Htt (control vs mutant), actin cytoskeleton and CME. Lack of these data renders the morphological observations unsupported
3. Three last figures of total eight demonstrate the effect of proteins, responsible for the initiation of certain neurodegenerative pathologies, on the activity of clathrin-mediated endocytosis, and on the properties of actin cytoskeletal system, however neither in the abstract nor in the introduction there is no any word about these proteins; in the discussion only a few words are devoted to one of these proteins TDP-43. When starting the article, did the authors plan to enter this data into the manuscript? In addition, the authors did not show at what level these proteins are expressed in transgenic flies or in cells derived from flies.
4. It is important to work on the style of the manuscript, the article is difficult to read, it is a collection of data that does not seem related to each other. Since the manuscript needs a major overhaul, I consider discussing minor comments unnecessary.

Significance

The manuscript "Pathogenic aggregates alter actin organization and cellular viscosity resulting in stalled clathrin mediated endocytosis" by Singh at al attempts to examine the relationship between the mutant huntington aggregates formation, vesicular transport and cytoskeletal rearrangements in cells bearing mHTT aggregates. The authors did a lot of work, used modern methods and obtained a large amount of data, but the purpose of this work remained unclear.

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

Learn more at Review Commons

---

Reply to the reviewers

Reply to Reviewers

We are grateful to the three reviewers for their careful and constructive critiques of our preprint. We will address all of their comments and suggestions, which help to make our paper more precise and understandable. In our replies, we use 'Patterson, eLife (2021)' as shorthand for Patterson, Basu, Rees & Nurse, eLife 2021:10.

Reviewer #1 (Evidence, reproducibility and clarity (Required)): Novák and Tyson present a model-based analysis of published data that had claimed to demonstrate bistable activation of CDK at the G2/M transition in fission yeast. They point out that the published data does not distinguish between ultra-sensitive (switch-like, but reversible) and bistable (switch-like, but irreversible) activation. They back up their intuition with robust quantitative modeling. They then point out that, with a simple experimental modification, the published experiments could be repeated in a way that would test between the ultra-sensitive and bistable possibilities.

This is an accurate and concise summary of our paper.

Therefore, this is a rare paper that makes a specific modeling-based prediction and proposes a straightforward way to test it. As such, it will be of interest to a broad range of workers involved in the fields cell cycle and regulatory modeling.

We agree that our work will be of interest to a broad range of scientists studying cell cycle regulation and mathematical modeling of bistable control systems.

Nonetheless, attention to the following points would improve the manuscript. The authors should be more careful about how they describe protein abundance. They often refer to protein level. I believe in every case they mean protein concentration, but this is not explicitly stated; it could be interpreted as number of protein molecules per cell. The authors should either explicitly state that level means concentration or, more simply, use concentration instead of level.

A valid criticism that has been addressed in the revised version.

The authors should explain why they include stoichiometric inhibition of CDK by Wee1 in their model. Is it required to make the model work in the wild-type case, or only in the CDK-AF case? My intuition is it should only be required in the AF case, but I would like to know for sure. Also, they should state if there is any experimental data for such regulation.

Bistability of the Tyr-phosphorylation switch requires 'sufficient' nonlinearity, which may come from the phosphorylation and dephosphorylation reactions that interconvert Cdk1, Wee1 and Cdc25. The easiest way to model these interconversion reactions is to use Hill- or Goldbeter-Koshland functions for the phosphorylation and dephosphorylation of Wee1 and Cdc25, but this approach is not appropriate for Gillespie SSA, which assumes elementary reactions. Both Wee1 and Cdc25 are phosphorylated on multiple sites, which we approximate by double phosphorylation; but this level of nonlinearity is not sufficient to make the switch bistable. In addition, stochiometric inhibition is a well-known source of nonlinearity, and in the Wee1:Cdk1 enzyme:substrate complex, Cdk1 is inhibited because Wee1 binds to Cdk1 near its catalytic site. In our model, stoichiometric inhibition of Cdk1 by Wee1 is required for bistability even in the wild-type case because the regulations of Wee1 and Cdc25 by phosphorylation are not nonlinear enough. There is experimental evidence that stoichiometric inhibition of Cdk1 by Wee1 is significant: mik1D wee1ts double mutant cells at the restrictive temperature (Lundgren, Walworth et al. 1991) are less viable than AF-Cdk1 (Gould and Nurse 1989). Furthermore, Patterson (eLife, 2021) found weak 'bistability' when they used AF-Cdk1 to induce mitosis. This puzzling observation suggests a residual feedback mechanism in the absence of Tyr-phosphorylation. Our model accounts for this weak bistability by assuming that free CDK1 can phosphorylate and inactivate the Wee1 'enzyme' in the Wee1:Cdk1 complex, which makes CDK1 and Wee1 mutual antagonists. This reaction is based on formation of a trimer, Cdk1:Wee1:Cdk1, which is possible since CDK1 phosphorylation of Wee1 occurs in its N-terminal region, which lies outside the C-terminal catalytic domain of Wee1 (Tang, Coleman et al. 1993). These ideas have been incorporated into the text in the subsection describing the model (see lines120-125).

The authors should explicitly state, on line 131, that the fact that "the rate of synthesis of C-CDK molecules is directly proportional to cell volume" results in a size-dependent increase in the concentration of C-CDK.

The accumulation of C-CDK molecules in fission yeast cells is complicated. In general, we may assume that larger cells have more ribosomes and make all proteins faster than do smaller cells. Absent other regulatory effects, the number of protein molecules is proportional to cell volume, and the concentration is constant. But, in Patterson's experiments, the number of C-CDK molecules is zero at the start of induction and rises steeply thereafter (see lines 147-148), and the rate of increase (#molec/time) is proportional to the size of the growing cell.

The authors should explain, on line 100, why they are "quite sure the bistable switch is the correct interpretation".

Line 105-106: "Although we suspect that the mitotic switch is bistable,.."

On line 166, include the units of volume.

Done

On lines 152 and 237, "smaller protein-fusion levels "should be replaced with "lower protein-fusion concentrations".

Done

**Referee cross-commenting** *I concur with the other two reviews. *

Reviewer #1 (Significance (Required)): *The paper is significant in that it points out an alternative interpretation for an important result in an important paper. Specifically, it points out that the published data is consistent with activation of CDK at the G2/M transition in fission yeast could be ultra-sensitive (switch-like, but reversible) instead of bistable (switch-like, but irreversible). The distinction is important because it has been claimed, by the authors of the submitted manuscript among others, that bistability is required for robust cell-cycle directionality. *

We agree with this assessment.

However, activation of CDK at the G2/M transition in other species has been shown to be bistable and the authors state that they are "quite sure the bistable switch is the correct interpretation". So, the paper is more likely an exercise in rigor than an opportunity to overturn a paradigm.

We were the first authors to predict that the G2/M switch is bistable (J. Cell Sci., 1993) and among the first to prove it experimentally in frog egg extracts (PNAS, 2004). Our models (Novak and Tyson 1995, Novak, Pataki et al. 2001, Tyson, Csikasz-Nagy et al. 2002, Gerard, Tyson et al. 2015) of fission yeast cell-cycle control rely on bistability of the G2/M transition; so, understandably, we believe that the transition in fission yeast is a bistable switch. But the 'bistable paradigm' has never been directly demonstrated by experimental observations in fission yeast cells. The Patterson paper (eLife, 2021) claims to provide experimental proof, but we demonstrate in our paper that Patterson's experiments are not conclusive evidence of bistability. Furthermore, we suggest that a simple change to Patterson's protocol could provide convincing evidence that the G2/M switch is either monostable or bistable. We are not proposing that the switch is monostable; we would be quite surprised if the experiment, correctly done, were to indicate a reversible switch. Our point is simply that the published experiments are inconclusive. The point we are making is neither a mere 'exercise in rigor' nor a suggestion to 'overturn a paradigm.' Rather it is a precise theoretical analysis of a central question of cell cycle regulation that should be of interest to both experimentalists and mathematical modelers.

Reviewer #2 (Evidence, reproducibility and clarity (Required)): Summary: The manuscript asks whether the data reported in Patterson et al. (2021) is consistent with a bistable switch controlling the G2/M transition in fission yeast. Patterson et al. (2021) use an engineered system to decouple a non-degradable version of Cyclin-dependent kinase (CDK) from cell growth and concomitantly measure CDK activity (by the nuclear localization of a downstream target, Cut3p). They observe cells with indistinguishable CDK levels but two distinct CDK activities, which they posit shows bistable behavior. In this study, the authors ask if other models can also explain this data. The authors use both deterministic and Gillespie based stochastic simulations to generate relationships between CDK activities and protein levels for various cell sizes. They conclude that the experiments performed in Patterson et al. are insufficient to distinguish between a bistable switch and a reversible ultrasensitive switch. They propose additional experiments involving the use a degradable CDK construct to also measure the inactivation kinetics.

This is an accurate summary of our paper.

They propose that a bistable switch will have different forward (OFF->ON) and backward (ON->OFF) switching rates. A reversible ultrasensitive switch will have indistinguishable switching rates.

Our analysis of Patterson's (2021) experiments is based on the well-known fact that the threshold for turning a bistable switch on is significantly different from the threshold for turning it off (in Patterson's case, the 'threshold' is the level of fusion protein in the cell when CDK is activated), whereas for a reversible, ultrasensitive switch, the two thresholds are nearly indistinguishable. The 'rate' at which the switch is made is a different issue, which we do not address explicitly. In the experiments and in our model, the switching rates are fast, whether the switch is bistable or monostable. The results are interesting and worth publication in a computational biology specific journal, as they might only appeal to a limited audience.

We think our results should also be brought to the attention of experimentalists studying cell cycle regulation, because Patterson's paper (eLife, 2021) presents a serious misunderstanding of the existence and implications of 'bistability' of the G2/M transition in fission yeast. Whereas Patterson's work is an elegant and creative application of genetics and molecular biology to an important problem, it is not backed up by quantitative mathematical modeling of the experimental results. In that sense, Patterson's work is incomplete, and its shortcomings need to be addressed in a highly respected journal, so that future cell-cycle experimentalists will not make the same-or similar-mistakes.

Several ideas need to be clarified and additional information needs to be provided about the specific parameters used for the simulations: Major comments: #1 The parameters need to be made more accessible by means of a supplementary table and appropriate references need to be cited.

Two new supplementary tables (S1 and S2) summarize the dynamic variables and parameter values.

It is not clear why Michaelis Menten kinetics will not be applicable to this system. Has it been demonstrated that the Km s of the enzymes are much greater than the substrate concentrations for all the reactions? If yes, please cite.

MM kinetics are not appropriate for such protein interaction networks because one protein may be both an enzyme and a substrate for a second protein (e.g., Wee1 and CDK, or Cdc25 and CDK). So, the condition for validity of MM kinetics (enzyme concen ≪ substrate concen) cannot be satisfied for both reactions. Indeed, enzyme concen ≈ substrate concen is probably true for most reactions in our network. Hence, it is advisable to stick with mass-action rate laws. Furthermore, MM kinetics are a poor choice for 'propensities' in Gillespie SSA calculations, as has been shown by many authors (Agarwal, Adams et al. 2012, Kim, Josic et al. 2014, Kim and Tyson 2020).

It will not be surprising if the simulation with Michaelis Menten would alter the dynamics shown in this study. A reversible switch with two different enzymes (catalyzing the ON->OFF and OFF->ON transitions) having different kinetics can give asymmetric switching rates. This would directly contradict what has been shown in Figure 7A-D.

We don't follow the reviewer's logic here. The two transitions, off → on and on → off, are already driven by different molecular processes (dephosphorylation of inactive CDK-P by Cdc25 and phosphorylation of active CDK by Wee1, respectively). Positive feedback of CDK activity on Cdc25 and Wee1 (++ and −−, respectively) causes bistability and asymmetric switching thresholds. Switching rates, which are determined by the kinetic rate constants of the up and down processes, are of secondary importance to the primary question of whether the switch is monostable or bistable.

#2 Line 427: The authors use a half-time of 6 hours in their model as Patterson et al. used a non-degradable construct. It is not clear why dilution due to cell growth has not been considered. The net degradation rate of a protein is the sum of biochemical degradation rate and growth dilution rate. The growth dilution rate seems significant (140 mins doubling time or 0.3 h-1 dilution rate) relative to assumed degradation rate (0.12 h-1). Please clarify why was the effect of dilution neglected in the model or show by sensitivity analysis this does not change the predicted CDK activation thresholds.

The reviewer highlights an important effect, but it is not relevant to our calculations. In the deterministic model used to calculate the bifurcation diagrams, both cell volume and the concentration of the non-degradable Cdc13:Cdk1 dimer are kept constant; therefore, there is no dilution effect. The stochastic model deals with changing numbers of molecules per cell; the dilution effect is taken into account by the appearance of cell volume, V(t), at appropriate places in the propensity functions. In other words: in the deterministic model, which is written for concentration changes, the dilution term, −(x/V)(dV/dt), is zero because V=constant; in the stochastic model, written in terms of numbers of molecules, dilution effects are implicit in the propensity functions.

*#3 Line 402 The authors state that the production rate of the Cdk protein is 'assumed' proportional to the cell volume. The word 'assumed' is incorrect here as a simple conversion of concentration-based differential equation (with constant production rate) to molecular numbers would show that production rate is proportional to the volume. This is not an assumption. *

Correct; we modified the text (see line 450-462). The role of cell volume in production rate is more relevant to the case of Cdc25, where we assume that its production rate, Δconcentration/Δt, is proportional to V, because the concentration of Cdc25 in the cell increases as the cell grows. We added two references (Keifenheim, Sun et al. 2017, Curran, Dey et al. 2022) to justify this assumption. In the stochastic code, the propensity for synthesis of Cdc25 molecules is proportional to V2.

#4 Line 423 Please cite the appropriate literature that shows that fission yeast growth during cell division is exponential. If the dynamics are more complicated, involving multiple phases of growth during cell division, please state so.

We now acknowledge that volume growth in fission yeast, rather than exponential, is bilinear with a brief non-growing phase at mitosis (Mitchison 2003). However, we suggest that our simplifying assumption of exponential growth is appropriate for the purposes of these calculations. See line 473-476: "In our stochastic simulations, we assume that cell volume is increasing exponentially, V(t) = V0eμt. Although fission yeast cells actually grow in a piecewise linear fashion (Mitchison 2003), the simpler exponential growth law (with doubling time @ 140 min) is perfectly adequate for our purposes in this paper.."

*#5 Line 250 The authors convert the bistable version of the CDK switch to reversible sigmoidal by assuming that Wee1 and Cdc25 phosphorylation is proportional to the CDK level rather than activity, which seems biochemically unrealistic. This invokes an altered circuit architecture where inactive CDK has enough catalytic activity to phosphorylate the two modifying enzymes (Wee1/Cdc25) but not enough to drive mitosis. This might be possible if the Km of CDK for Wee1/Cdc25 is lower relative to other downstream substrates that drive mitosis. The authors can reframe this section of the paper to state this possibility, which might be interesting to experimentalists. *

The reviewer is correct that the molecular biology underlying our 'reversible sigmoidal' model is biochemically unrealistic. But, in our opinion, this is the simplest way to convert our bistable model into a monostable, ultrasensitive switch while maintaining the basic network structure in Fig. 1. Our purpose is to show that a monostable model-only slightly changed from the bistable model-can account for Patterson's experimental data equally well. If Nurse's group modifies the experimental protocol as we suggest and their new results indicate that the G2/M transition in fission yeast is bistable, then our reversible sigmoidal model, having served its purpose, can be forgotten. If they show that the transition is not bistable, then both experimentalists and theoreticians will have to think about biochemically realistic mechanisms that can account for the new data...and everything else we already know about the G2/M transition in fission yeast.

#6 It is difficult to phenomenologically understand a bistable switch just based on differences in activation and inactivation thresholds. For example, a reversible ultrasensitive switch also shows a difference in activation and inactivation thresholds (Figure 7D). How much of a difference should be expected of a bistable switch versus reversible switch?

We show how much of a difference can be expected by contrasting Fig. 7 to Fig. 8. For the largest cells (panel D of both figures), the difference is small and probably undetectable experimentally. For medium-sized cells (panel C), the difference is larger but probably difficult to distinguish experimentally. Only the smallest cells (panel B) provide an opportunity for clearly distinguishing experimentally between monostable and bistable switching.

*Moreover, as the authors clearly understand (line 275), time-delays in activation and inactivation reactions can inflate these differences. In the future, if the authors can convert the equations to potential energy space as done in Acar et al. 2005 (Nature 435:228) in Figure 3c-d, it will be useful. Also, predicting the distribution of switching rates from the Gillespie simulation might be informative and can be directly compared to experimental measurements in the future (if the Cut3p levels in nucleus and cytosol equilibrates fast enough or other CDK biosensors are developed). *

The famous paper by Acar et al. (2005) is indeed an elegant experimental and theoretical study of bistability ('cellular memory') in the galactose-signalling network of budding yeast. We have included a comparison of Patterson et al. with Acar et al. in our Conclusions section (lines 353-368):

"It is instructive, at this point, to compare the work of Patterson et al. (2021) to a study by Acar et al. (Acar, Becskei et al. 2005) of the galactose-signaling network of budding yeast. Combining elegant experiments with sophisticated modeling, Acar et al. provided convincing proof of bistability ('cellular memory') in this nutritional control system. They measured PGAL1-YFP expression (the response) as a function of galactose concentration in the growth medium (the signal), analogous to Patterson's measurements of CDK activity as a function of C-CDK concentration in fission yeast cells. In Acar's experiments, the endogenous GAL80 gene was replaced by PTET-GAL80 in order to maintain Gal80 protein concentration at a constant value determined by doxycycline concentration in the growth medium. The fixed Gal80p concentration in Acar's cells is analogous to cell volume in Patterson's experiments. In Fig.3b of Acar's paper, the team plotted the regions of monostable-off, monostable-on and bistable signaling in dependence on their two control parameters, external galactose concentration and intracellular Gal80p concentration, analogous to our Fig.4. Because Acar's experiments explored both the off → on and on → off transitions, they could show that their observed thresholds (the red circles) correspond closely to both saddle-node bifurcation curves predicted by their model. On the other hand, Patterson's experiments (as analyzed in our Fig.4) probe only the off → on transition."

The purpose of our paper is to show that Patterson-type experiments can and should be done so as to probe both thresholds, as was done by van Oudenaarden's team. They went further to characterize their bistable switch in terms of 'the concept of energy landscapes'. We think it is premature to pursue this idea in the context of the G2/M transition in fission yeast until there is firm, quantitative data characterizing the nature of the 'presumptive' bistable switch in fission yeast.

Minor comments: #1 Line 2: Please replace "In most situations" to "In favorable conditions"

Done.

**Referee cross-commenting** I agree with Reviewer 1 that this falls more under pointing out an alternative interpretation of a single experiment than challenging widely supported orthodoxy about how the eukaryotic cell cycle leaves mitosis.

As we said earlier, our 1993 paper in J Cell Sci is the source of this orthodox view, and it is widely supported at present because there is convincing experimental evidence for bistability in frog egg extracts, budding yeast cells and mammalian cells. Patterson's paper is not sound evidence for bistability of the G2/M transition in fission yeast cells. It is important for experimentalists to know why the experiments fail to confirm bistability, and important for someone to do the experiment correctly in order to confirm (or, what would be really interesting, to refute) the expectation of bistability at the G2/M transition in fission yeast cells.

Reviewer #2 (Significance (Required)): Suitable for specialist comp bio journal eg PLoS Comp Bio

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

The paper by Novak and Tyson revisits a recent paper from Nurse group on the bistability of mitotic switch in fission yeast using mathematical modelling. The authors extend their older models of mitotic entry check point and implement both deterministic and stochastic version of new model. They show this model does indeed possess bistability and show that combined with stochastic fluctuations the model can show bimodality for the cyclin-CDK activity at a particular cell size consistent with the recent experimental data. However, the authors also show alternative model that has mono-stable ultrasensitivity can also explain the data and suggest experiments that can prove the existence of hysteresis and therefore bistability.

Right on.

While the biological implication of the study is well explained, the authors can improve the presentation of their model and the underlying assumptions. I have the following comments and suggestions for improvement of the paper.

1. • The cartoon of the mathematical model is confusing at places, for example the wee1-CDK complex according to the equations either dissociates back to wee1 and CDK or gives rise to pCDK and wee1, the arrow below is confusing as it implies it can also give rise to wee1p, the CDK phosphorylation of wee1 is already included in the diagram. Also, the PP2A is put on the arrow for all reactions but for wee1p2 to wee1p its action shown with a dashed line. Also, I wondered if wee1p and wee1p2 can also bind CDK and sequester or phosphorylate CDK?* We are sorry for the confusion and have improved Fig. 1.

2. The rates and variables in the ODEs are not fully described. Also sometimes unclear what is parameter and what is a variable, I had to look at the code.*

We now include tables of variables and parameter values, with explanatory notes.

• The model has quite a few parameters, but these are not at all discussed in the paper. How did the authors come up with these particular set of parameters, has there been some systematic fitting, or tuning by hand to produce a good fit to the data? I could only see the value of the parameters in the code, but perhaps a table with the parameters of the model, what they mean and their value (and perhaps how the values is obtained) is missing.*

The parameters were tuned by hand to fit Patterson's data, based, of course, on our extensive experience fitting mathematical models to myriad data sets on the cell division cycles of fission yeast, budding yeast, and frog egg extracts. We now provide a table of parameter values.

• The authors are using the Gillespie algorithm with time varying parameters (as some rates depend on volume and volume is not constant). Algorithm needs to be modified slightly to handle this (see for example Shahrezaei et al Molecular Systems Biology 2008). *

A valid criticism, but the rate of cell volume increase is very slow compared to the propensities of the biochemical reactions. We write (lines 492-498):

"In each step of the SSA, the volume of the cell is increasing according to an exponential function, and, consequently, the propensities of the volume-dependent steps are, in principle, changing with time; and this time-dependence could be taken into account explicitly in implementing Gillespie's SSA (Shahrezaei, Ollivier et al. 2008). However, the step-size between SSA updates is less than 1 s compared to the mass-doubling time (140 min) of cell growth. So, it is warranted to neglect the change in V(t) between steps of the SSA, as in our code."

• The authors correctly point out, ignoring mRNA has resulted in underestimation of noise, however another point is that mRNA life times are short and that also affects the timescale of fluctuations and this may be relevant to the switching rates between the bistable states. *

A valid point, but to include mRNA's would double the size of the model. Furthermore, we have little or no data about mRNA fluctuations in fission yeast cells, so it would be impossible to estimate the values of all the new parameters introduced into the model. Finally, the switching rates between bistable states (or across the ultrasensitive boundary) are not the primary focus of Patterson's experiments or our theoretical investigations. So, we propose to delay this improvement to the model until the relevant experimental data is available.

• In the introduction add, "In this study" to "Intrigued by these results, we investigated their experimental observations with a model of bistability in the activation of cyclin-CDK in fission yeast." *

Done

Reviewer #3 (Significance (Required)): Overall, this is an interesting study that revisits an old question and some recent experimental data. The use of stochastic modelling in explaining variability and co-existence of cell populations in the context of cell cycle and comparison to experimental data is novel and of interest to the communities of cell cycle researchers, systems biologists and mathematical biologists.

We agree. Thanks for the endorsement

References

Acar, M., A. Becskei and A. van Oudenaarden (2005). "Enhancement of cellular memory by reducing stochastic transitions." Nature 435(7039): 228-232.

Agarwal, A., R. Adams, G. C. Castellani and H. Z. Shouval (2012). "On the precision of quasi steady state assumptions in stochastic dynamics." J Chem Phys 137(4): 044105.

Curran, S., G. Dey, P. Rees and P. Nurse (2022). "A quantitative and spatial analysis of cell cycle regulators during the fission yeast cycle." Proc Natl Acad Sci U S A 119(36): e2206172119.

Gerard, C., J. J. Tyson, D. Coudreuse and B. Novak (2015). "Cell cycle control by a minimal Cdk network." PLoS Comput Biol 11(2): e1004056.

Gould, K. L. and P. Nurse (1989). "Tyrosine phosphorylation of the fission yeast cdc2+ protein kinase regulates entry into mitosis." Nature 342(6245): 39-45.

Keifenheim, D., X. M. Sun, E. D'Souza, M. J. Ohira, M. Magner, M. B. Mayhew, S. Marguerat and N. Rhind (2017). "Size-Dependent Expression of the Mitotic Activator Cdc25 Suggests a Mechanism of Size Control in Fission Yeast." Curr Biol 27(10): 1491-1497 e1494.

Kim, J. K., K. Josic and M. R. Bennett (2014). "The validity of quasi-steady-state approximations in discrete stochastic simulations." Biophys J 107(3): 783-793.

Kim, J. K. and J. J. Tyson (2020). "Misuse of the Michaelis-Menten rate law for protein interaction networks and its remedy." PLoS Comput Biol 16(10): e1008258.

Lundgren, K., N. Walworth, R. Booher, M. Dembski, M. Kirschner and D. Beach (1991). "mik1 and wee1 cooperate in the inhibitory tyrosine phosphorylation of cdc2." Cell 64(6): 1111-1122.

Mitchison, J. M. (2003). "Growth during the cell cycle." Int Rev Cytol 226: 165-258.

Novak, B., Z. Pataki, A. Ciliberto and J. J. Tyson (2001). "Mathematical model of the cell division cycle of fission yeast." Chaos 11(1): 277-286.

Novak, B. and J. J. Tyson (1995). "Quantitative Analysis of a Molecular Model of Mitotic Control in Fission Yeast." J Theor Biol 173: 283-305.

Patterson, J. O., S. Basu, P. Rees and P. Nurse (2021). "CDK control pathways integrate cell size and ploidy information to control cell division." Elife 10.

Shahrezaei, V., J. F. Ollivier and P. S. Swain (2008). "Colored extrinsic fluctuations and stochastic gene expression." Mol Syst Biol 4: 196.

Tang, Z., T. R. Coleman and W. G. Dunphy (1993). "Two distinct mechanisms for negative regulation of the Wee1 protein kinase." EMBO J 12(9): 3427-3436.

Tyson, J. J., A. Csikasz-Nagy and B. Novak (2002). "The dynamics of cell cycle regulation." Bioessays 24(12): 1095-1109.

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 paper by Novak and Tyson revisits a recent paper from Nurse group on the bistability of mitotic switch in fission yeast using mathematical modelling. The authors extend their older models of mitotic entry check point and implement both deterministic and stochastic version of new model. They show this model does indeed possess bistability and show that combined with stochastic fluctuations the model can show bimodality for the cyclin-CDK activity at a particular cell size consistenent with the recent experimental data. However, the authors also show alternative model that has mono-stable ultrasensitivity can also explain the data and suggest experiments that can prove the existence of hysteresis and therefore bistability.

While the biological implication of the study is well explained, the authors can improve the presentation of their model and the underlying assumptions. I have the following comments and suggestions for improvement of the paper. <br /> 1. The cartoon of the mathematical model is confusing at places, for example the wee1-CDK complex according to the equations either dissociates back to wee1 and CDK or gives rise to pCDK and wee1, the arrow below is confusing as it implies it can also give rise to wee1p, the CDK phosphorylation of wee1 is already included in the diagram. Also, the PP2A is put on the arrow for all reactions but for wee1p2 to wee1p its action shown with a dashed line. Also, I wondered if wee1p and wee1p2 can also bind CDK and sequester or phosphorylate CDK? 2. The rates and variables in the ODEs are not fully described. Also sometimes unlcear what is parameter and what is a variable, I had to look a the code. 3. The model has quite a few parameters, but these are not at all discussed in the paper. How did the authors come up with these particular set of parameters, has there been some systematic fitting, or tuning by hand to produce a good fit to the data? I could only see the value of the parameters in the code, but perhaps a table with the parameters of the model, what they mean and their value (and perhaps how the values is obtained) is missing. 4. The authors are using the Gillespie algorithm with time varying parameters (as some rates depend on volume and volume is not constant). Algorithm needs to be modified slightly to handle this (see for example Shahrezaei et al Molecular Systems Biology 2008). 5. The authors correctly point out, ignoring mRNA has resulted in underestimation of noise, however another point is that mRNA life times are short and that also affects the timescale of fluctuations and this may be relevant to the switching rates between the bistable states. 6. In the introduction add, "In this study" to "Intrigued by these results, we investigated their experimental observations with a model of bistability in the activation of cyclin-CDK in fission yeast.

Significance

Overall, this is an interesting study that revisits an old question and some recent experimental data. The use of stochastic modelling in explaining variability and co-existence of cell populations in the context of cell cycle and comparison to experimental data is novel and of interest to the communities of cell cycle researchers, systems biologists and mathematical biologists.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary: The manuscript asks whether the data reported in Patterson et al. (2021) is consistent with a bistable switch controlling the G2/M transition in fission yeast. Patterson et al. (2021) use an engineered system to decouple a non-degradable version of Cyclin-dependent kinase (CDK) from cell growth and concomitantly measure CDK activity (by the nuclear localization of a downstream target, Cut3p). They observe cells with indistinguishable CDK levels but two distinct CDK activities, which they posit shows bistable behavior. In this study, the authors ask if other models can also explain this data. The authors use both deterministic and Gillespie based stochastic simulations to generate relationships between CDK activities and protein levels for various cell sizes. They conclude that the experiments performed in Patterson et al. are insufficient to distinguish between a bistable switch and a reversible ultrasensitive switch. They propose additional experiments involving the use a degradable CDK construct to also measure the inactivation kinetics. They propose that a bistable switch will have different forward (OFF->ON) and backward (ON->OFF) switching rates. A reversible ultrasensitive switch will have indistinguishable switching rates.

The results are interesting and worth publication in a computational biology specific journal, as they might only appeal to a limited audience. Several ideas need to be clarified and additional information needs to be provided about the specific parameters used for the simulations:

Major comments:

1. The parameters need to be made more accessible by means of a supplementary table and appropriate references need to be cited. It is not clear why Michaelis Menten kinetics will not be applicable to this system. Has it been demonstrated that the Km s of the enzymes are much greater than the substrate concentrations for all the reactions? If yes, please cite. It will not be surprising if the simulation with Michaelis Menten would alter the dynamics shown in this study. A reversible switch with two different enzymes (catalyzing the ON->OFF and OFF->ON transitions) having different kinetics can give asymmetric switching rates. This would directly contradict what has been shown in Figure 7A-D.
2. Line 427: The authors use a half-time of 6 hours in their model as Patterson et al. used a non-degradable construct. It is not clear why dilution due to cell growth has not been considered. The net degradation rate of a protein is the sum of biochemical degradation rate and growth dilution rate. The growth dilution rate seems significant (140 mins doubling time or 0.3 h-1 dilution rate) relative to assumed degradation rate (0.12 h-1). Please clarify why was the effect of dilution neglected in the model or show by sensitivity analysis this does not change the predicted CDK activation thresholds.
3. Line 402 The authors state that the production rate of the Cdk protein is 'assumed' proportional to the cell volume. The word 'assumed' is incorrect here as a simple conversion of concentration-based differential equation (with constant production rate) to molecular numbers would show that production rate is proportional to the volume. This is not an assumption.
4. Line 423 Please cite the appropriate literature that shows that fission yeast growth during cell division is exponential. If the dynamics are more complicated, involving multiple phases of growth during cell division, please state so.
5. Line 250 The authors convert the bistable version of the CDK switch to reversible sigmoidal by assuming that Wee1 and Cdc25 phosphorylation is proportional to the CDK level rather than activity, which seems biochemically unrealistic. This invokes an altered circuit architecture where inactive CDK has enough catalytic activity to phosphorylate the two modifying enzymes (Wee1/Cdc25) but not enough to drive mitosis. This might be possible if the Km of CDK for Wee1/Cdc25 is lower relative to other downstream substrates that drive mitosis. The authors can reframe this section of the paper to state this possibility, which might be interesting to experimentalists.
6. It is difficult to phenomenologically understand a bistable switch just based on differences in activation and inactivation thresholds. For example, a reversible ultrasensitive switch also shows a difference in activation and inactivation thresholds (Figure 7D). How much of a difference should be expected of a bistable switch versus reversible switch? Moreover, as the authors clearly understand (line 275), time-delays in activation and inactivation reactions can inflate these differences. In the future, if the authors can convert the equations to potential energy space as done in Acar et al. 2005 (Nature) in Figure 3c-d, it will be useful. Also, predicting the distribution of switching rates from the Gillespie simulation might be informative and can be directly compared to experimental measurements in the future (if the Cut3p levels in nucleus and cytosol equilibrates fast enough or other CDK biosensors are developed).

Minor comments:

1. Line 2: Please replace "In most situations" to "In favorable conditions"

Referee cross-commenting

I agree with Reviewer 1 that this falls more under pointing out an alternative interpretation of a single experiment than challenging widely supported orthodoxy about how the eukaryotic cell cycle leaves mitosis.

Significance

Suitable for specialist comp bio journal eg PLoS Comp Bio

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

Novák and Tyson present a model-based analysis of published data that had claimed to demonstrate bistable activation of CDK at the G2/M transition in fission yeast. They point out that the published data does not distinguish between ultra-sensitive (switch-like, but reversible) and bistable (switch-like, but irreversible) activation. They back up their intuition with robust quantitative modeling. They then point out that, with a simple experimental modification, the published experiments could be repeated in a way that would test between the ultra-sensitive and bistable possibilities. Therefore, this is a rare paper that makes a specific modeling-based prediction and proposes a straightforward way to test it. As such, it will be of interest to a broad range of workers involved in the fields cell cycle and regulatory modeling. Nonetheless, attention to the following points would improve the manuscript.

The authors should be more careful about how they describe protein abundance. They often refer to protein level. I believe in every case they mean protein concentration, but this is not explicitly stated; it could be interpreted as number of protein molecules per cell. The authors should either explicitly state that level means concentration or, more simply, use concentration instead of level.

The authors should explain why they include stoichiometric inhibition of CDK byWee1 in their model. Is it required to make the model work in the wild-type case, or only in the CDK-AF case. My intuition is it should only be required in the AF case, but I would like to know for sure. Also, they should state if there is any experimental data for such regulation.

The authors should explicitly state, on line 131, that the fact that "the rate of synthesis of C-CDK molecules is directly proportional to cell volume" results in a size-dependent increase in the concentration of C-CDK.

The authors should explain, on line 100, why they are "quite sure the bistable switch is the correct interpretation".

On line 166, include the units of volume.

On lines 152 and 237, "smaller protein-fusion levels "should be replaced with "lower protein-fusion concentrations".

Referee cross-commenting

I concur with the other two reviews.

Significance

The paper is significant in that it points out a alternative interpretation for an important result in an important paper. Specifically, it points out that the published data is consistent with activation of CDK at the G2/M transition in fission yeast could be ultra-sensitive (switch-like, but reversible) instead of bistable (switch-like, but irreversible). The distinction is important because it has been claimed, by the authors of the submitted manuscript among others, that bistability is required for robust cell-cycle directionality. However, activation of CDK at the G2/M transition in other species has been shown to be bistable and the authors state that they are "quite sure the bistable switch is the correct interpretation". So, the paper is more likely an exercise in rigor than an opportunity to overturn a paradigm.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

Please see attached point-by-point response file, it contains essential figures and formatting that cannot be pasted here.

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:

Bataclan et al. provide extensive in vitro work on Regnase1/3 function in mast cells. They make use of in vitro differentiated mast cells from bone marrow (BMMC) and in vitro cultures from primary peritoneal mast cells from mice. They show robustly Reg1/3 upregulation by transcript and protein upon activation stimulation in these cultured mast cells. Knockdown and CRISPR editing of Reg1/3 show enhanced inflammatory signaling. TNF is identified as a direct target of Reg1/3 nuclease activity. Reg3 is implicated as a negative regulator of Reg1. Overall, this work highlights Reg1/3 as novel players in mast cell biology that control inflammatory signaling including TNF.

Major comments:

• The result section requires more consistent & sufficient information on the assays (in vitro, ex vivo, species, etc.) particularly for the meta-analysis in the beginning. Relatedly, can the authors assume cross-species conservation of Reg1/3 in mast cells across mammals to make general conclusions on the functions or should this be limited mostly to mouse observations? It would be helpful to mention the species studied in the abstract.
• Reg1 protein levels appear to be more stable than Reg3 protein levels (mirroring the transcript) Fig1a+c. Is the half-life of Reg3 shorter than Reg1? What is functional impact of Reg1 regulation at the transcriptional level considering the protein half-lives?
• MFI comparisons can only be used on unimodal populations (not bimodal i.e. Fig3b). Also, contour flow plots should show outliers.
• Is TNF functionally secreted from WT vs KD/KO mast cells under these in vitro conditions? This is not shown and seems quite an important aspect given the focus on TNF regulation (other secreted factors could also be considered). Degranulation appears to be lower in Fig5g. Does Reg1 overexpression reduces TNF expression and secretion?
• Catalytic variants of Reg1/3 appear to be only tested in WT cells. Wouldn't it be worthwhile to test these in KO cells? The absolute protein levels of ectopic and endogenous are not entirely clear and only shown for WT Reg3 in Fig4b.
• Fig2b shows upregulation of Reg1 transcript is not enhanced upon Reg3 knockdown, but it is enhanced in Fig4a? How are these assays different?
• The authors implicate Reg3 to regulate Reg1 by transcript. What is the role of the protein interaction between Reg1 and 3? Why is the protein interaction not seen in the IF experiments? Would other cell systems (maybe mouse) be more suitable.
• Cell death increases after Reg1/3 knockdowns; can this be rescued with Reg1/3 reinstation? Is the protein level of Reg1 important for this; can this be dosed? Are the assays for cell death/proliferation done in comparison to unstimulated resting cells, cultured conditions and IgE stimulated mast cells? Minor point: A single assay of cell death is sufficient in the main figures; same for proliferation assays.
• It is not entirely clear how the RNA-seq data with CRISPR KO in Fig6 is different than the nanostring data with Knockdowns shown earlier in Fig2? Were unstimulated cells not used as a control before? A floxed mouse model for Reg1 is introduced at the end, but would have been useful for many assays. OPTIONAL: are in vivo experiments of interest to confirm the findings? The conclusions that Reg1/3 regulates physiological mast cell responses might be a reach otherwise.

Minor comments:

• Some of the Knockdown and CRISPR validation main figures are redundant and might be better suited for the supplement.
• The MW in the western blot requires lines to indicate the exact location of the ladder.
• Fig2 title refers as loss of Reg1/3 using siRNA. The term is typically used in the context of genetic ablation and not for knockdowns.
• Gene locus figures of Reg1/3 are confusing why are only E3 and E6 shown and not all Exons (Fig2a et al)?
• Are biological replicates or technical replicates used in Fig1?
• Is a student's t test the appropriate statistical analysis for most of the analyses?
• An explanation why nanostring and RNA-seq are both used would be helpful.

Significance

Bataclan et al investigate Reg1/3 function in mast cells, that are of interest to elucidate the regulation of mast cell effector functions. Reg1/3 were identified as novel regulators of murine mast cells. Independent and complementary use of Knockdowns and CRISPR deletions provide robust data. Also, the use of two independent mast cell populations and different stimuli is of interest, albeit they are not used in every assay. Altogether, the in vitro data are robust and rigorous. On the flipside, in vivo data is not provided. The translational impact would be higher if findings are tested in in vivo models including preclinical applications (as discussed in text). Mouse studies are often indicative of mechanisms in higher mammals, but it would help to lay this out more clearly. Therefore, the connection to humans is less clear. The type of work presented here would be best described as basic research. This review has been assessed with the following expertise: mouse/human immunology, mouse models, immune cell signaling, immune effector functions, flow cytometry.

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:

Bataclan et al. studied the role of Regnase-1 and -3 in the mast cell (MC) function in vitro. Using siRNA, CRISPR-mediated depletion, and overexpression models in, mainly, IL-3-cultured bone-marrow-derived mast cells, they beautifully demonstrated that Regnase-1 and, to a lesser extent, Regnase-3 suppressed the expression of TNF in MCs in response to IgE-stimulation. Regnase-1 also had roles in MC survival and IgE-stimulation-induced degranulation. Regnase-3 seems to have mixed effects on MC functions as the protein's primary targets include both TNF and Regnase-1 mRNAs.

Major comments:

1. The authors mainly used BMMCs generated by four weeks of cultivation of bone marrow cells with IL-3; at this point, MCs are considered "less mature". Could the authors reproduce the primary data (the role of Regnase-1 on TNF, cell survival, and degranulation) even if they used more mature MCs (e.g., extended cultivation with IL-3 (6-7 weeks))?
2. TNF mRNA is not considered a primary target for Regnase-1 in other cell types, such as macrophages or fibroblasts. Although this reviewer does not doubt that the TNF expression level is controlled by Regnase-1 in MCs, more evidence is needed to conclude that TNF is a "primary target" for Regnases. They can determine the stability of "endogenous" TNF mRNA in Regnase-depleted cells, as in Figure 4d.
3. The role of Regnase-1 on the MC degranulation is less pronounced. Could the authors show the same result using another assay, such as a FACS-based assay excluding dead or dying cells?
4. Have the authors had a chance to look into which protease(s) cleaved Regnase-1 in IgE-stimulated MCs?

Significance

General assessment:

This study investigated the role of Regnase-1 and -3 in the mast cell (MC) function in vitro. The strength of the study is that they used several methods to manipulate the expression levels of Regnases in the cells (siRNA, CRISPR, and Regnase-1 overexpression) and obtained consistent results. Although the study is well designed and the results are beautiful, the BMMCs the authors used are less mature MCs and do not represent the cells in the mammalian body well. Also, this study only focused on in vitro mouse-derived MCs, and human MCs or in vivo roles of MC Regnases were not studied.

Advance:

The role of Regnase-1 in several cell types has already been shown, including the populations involved in type-2 immunity (Th2 and ILC2). However, this study might be the first to investigate the role of Regnases in MCs. The roles of Regnases in MCs (control of mRNA stability through the RNase activity) presented are in line with the previous studies.

Audience:

The primary audience of this study may be basic researchers studying type-2 immunity or MC biology. Because MCs are an essential cell population in several allergic disorders, some clinicians who care for allergic patients might also be interested in this study. However, main audiences may be relatively limited to specific fields like immunology and allergology.

This reviewer's main field of expertise is basic research in immunology and allergology. More specific keywords are MCs, ILC2, IgE, type-2 cytokines, and mouse models.

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:

Mast cells play a role in exerting effector functions in the immune response. However, the functioning of RNA-binding proteins (RBPs) in the mast cells is not well understood. The authors focused on the Regnase family of RNA-degrading enzymes and worked towards unraveling their functions. Upon activating mast cells, the authors observed a significant induction of Regnase-3 among RBPs. Furthermore, they found that Regnase-3 directly controls the well-known Regnase-1, revealing its role in regulating homeostasis and inflammatory responses in mast cells.

Major comments:

This experiment has been meticulously designed, and the reliability of the presented data is quite high. While the observations are specific to mast cells, the insights gained could provide valuable information about the interrelation within the Regnase family across other immune cell types. However, there are three main concerns raised by the reviewer:

The first concern revolves around the use of the IVT system for the expression of Regnase-1 and Regnase-3. Is there evidence confirming the expression of Regnase-3 as a full-length protein, as shown in Figures 3g and 3h? The APC-HA signal by FACS could be positive even for degradation products alone. Assuming Regnase-3 is not expressed in its full-length, it might lead to results similar to the control. Given the larger molecular weight of Regnase-3 compared to Regnase-1, it is crucial to demonstrate sufficient expression through IVT. In particular, the Western blot data for Regnase-3 in some cases confirm the full length of the product, while in other cases more degradation products appear.

The second concern arises from the co-immunoprecipitation experiment in Figure 4i. While the experiment detects Regnase-1 co-precipitating with Regnase-3, the reverse-precipitating Regnase-1 with Regnase-3 shows a signal comparable to IgG, indicating background noise. Further improvement is needed in this aspect.Is it possible that your antibody against Regnase-1 also binds to Regnase-3?

The third concern is related to the evaluation of Regnase-1 degranulation in Figure 5g, where there appears to be some variability. Including Regnase-3 and conducting mutual evaluations could enhance the reliability of the results, possibly addressing this variability.

Minor comments:

Figure 6e-g also seems to vary, and the effects on cell death and cell proliferation tend to be somewhat milder.

Certainly, their IF images (Figure S2 and Figure S5) suggests that Regnase-1 and Regnase-3 have no or only a weak interaction.

Is the input to the immunoprecipitation from whole cell lysate or is it cleared with a low-speed (or high-speed) centrifugation?

Significance

The following aspects are important: Understanding the regulatory mechanisms mediated by RNA-binding proteins (RBPs) has gained significant attention in recent years. While it is inferred that RBPs control specific RNAs, many uncertainties remain about how they function in various RNA metabolism processes. The Regnase family is known to operate within the mechanism of RNA stability. While lower organisms have only one type, mammals have evolved to have four types. Understanding the implications of this family expansion is highly valuable, shedding light on how the RNA within our body's cells is regulated.

• Advance: In this study, a strength lies in the meticulous examination of the relationship between Regnase-1 and Regnase-3 by handling them similarly in the context of mast cells. However, because of the multifaceted effects of Regnase-1 on cell proliferation, cell death, and the cell cycle, significant progress in understanding it has yet to be made. Going forward, the replication of immune responses in mast cells at the animal level holds the potential to further deepen our comprehension of RBPs through the Regnase family. This study complements two previous investigations on Regnase-3 (PMID: 34215755, 31126966) while specifically focusing on mast cells. The findings align with the prevailing perspective that emphasizes the significance of Regnase-1 among the Regnase family.
• Audience: Basic research
• Immunology, Molecular Biology, Genetic Engineering

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

Learn more at Review Commons

---

Reply to the reviewers

Dear Editor,

We have addressed the points and concerns raised by the reviewers and wish to thank them for their effort and time. We agree with all the comments and suggestions, which resulted in a significant improvement of the manuscript. Below, we provide a point-by-point response to all comments.

Sincerely,

Anders Hofer, corresponding author

---

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

In their paper Ranjbarian and colleagues provide a tour de force at characterizing dAK from G. intestinalis using both enzymology and structural biology. G. intestinalis does not have RNR and therefore this organism relies on dAK which catalyzes formation of dAMP and ADP from deoxyadenosine and ATP (among other substrate pairs). The authors performed a terrific job at testing this reaction in depth using a recombinant dAK and a battery of various co-substrates (both natural as well as synthetic ones, Table 1). Extensive structural information on dAK was obtained using a combination of X-ray crystallography and cryo-EM. Overall, this work will be paramount aid in better understanding of the reaction mechanism, especially in the context of molecules which can be used as inhibitors of such a crucial enzyme (metabolic vulnerability for this parasite).

This manuscript, in its current form does not require additional experiments but I would like to have a few aspects corrected/clarified, before it can be accepted for publication:

Line 30: "whereas the affinities for deoxyguanosine, deoxyinosine and deoxycytidine were 400-2000 times lower." Better not to use term "affinity" when KM or kcat/KM are implied (unless ITC was used to measure true Kds).

-This is a good point, and we are now using KM values in all instances were actual numbers are implied and only kept the word affinity in cases where it is discussed in more general terms.

Line 31: "Deoxyadenosine analogues halogenated at the 2- and/or 2´-positions were also potent substrates, with comparable EC50 values as the main drug used today, metronidazole, but with the advantage of being usable on metronidazole-resistant parasites." Not sure this sentence is clear as written.

-We have now rewritten the sentence as follows: "Deoxyadenosine analogues halogenated at the 2- and/or 2´-positions were also potent substrates with comparable EC50 values on cultured G. intestinalis cells as metronidazole, the first line treatment today, with the additional advantage of being effective against metronidazole-resistant parasites."

Line 55: "..G. intestinalis (synonymous to G. lamblia and G. duodenalis)..". Very nice that authors provide this information as it is usually a point of confusion i.e. multiple names for the same organism.

-Thanks a lot, we are happy that you liked it.

Line 61 and above as well: "Treatment regimes are mainly based on metronidazole and to a lesser extent other 5-nitroimidazoles...". MT is introduced a bit sporadically, and not completely clear which enzyme it inhibits and its mode of action." Common knowledge is that MT is known for its action in aerobic parasites/bacteria and known as Flagyl, where it is mode of action was linked to "activation" due to microaerophilic conditions. Maybe MT can be introduced after text starting from Line 71?

-The description of metronidazole is adjusted as following: "Metronidazole (Flagyl) is the most commonly used drug to treat giardiasis and selectively kills the parasite and other anaerobic organisms by forming free radicals under oxygen-limited conditions, but it has side effects such as nausea, abdominal pain, diarrhea, and in some cases neurotoxicity reactions."

Line 70: what is "cyst-wall"?

-It is a cell wall consisting of three major cyst wall proteins and N-acetylgalactosamine. We have adjusted the sentence to the following to make the term clearer: The trophozoites can also secrete material to form a cyst wall and go through two rounds of DNA replication to form cysts, which contain four nuclei and a 16N genome per cell (4N in each nucleus).

Line 90: "The reaction is catalyzed by deoxyribonucleoside kinases (dNKs), which are.." I really do not like when in order to find a reaction which is catalyzed by an enzyme in a particular study one needs to dive into the literature, sometimes it requires a lot of time as in most of recent papers on the subject reactions catalyzed are not listed. Please add a Figure or a panel with reactions catalyzed by both dNKs families.

-It is a good idea and we have now added a figure (Fig. 1), which compares the deoxyribonucleotide metabolism of G. intestinalis with mammalian cells. The different deoxyribonucleoside kinases in the parasite and mammalian cells are included in the figure.

Line 96: "..was found to have a ~10-fold higher affinity to thymidine.." as I mentioned above I really do not like the usage of "affinity", when actually low KM is implied.

-It is corrected now (see above).

Line 113: "This does not match the current knowledge that there are three dNKs in total whereof one completely specific for thymidine. The lack of knowledge about these essential enzymes in the parasite has hampered the understanding of Giardia deoxyribonucleoside metabolism and hence its exploitation as a target for antiparasitic drugs." Very good rationale, as I mentioned above, I think a Figure needs to be introduced that depicts different enzymes involved in deoxyribonucleoside metabolism (both TK1 and non- TK1 members) in Giardia with clearly labeled all known paralogs and corresponding enzymatic reactions.

-Thanks a lot for the suggestion. Information about the different dNKs in G. intestinalis with mammalian cells for comparison is included in the new figure (Fig. 1).

Line 132: Odd designation of supplementary figures, usually it is "Fig. S1" etc. The legend for Fig. S1 is not adequate, please add description of species and name of enzymes for all sequences shown. Also each sequences in alignment should start with number (a.a. number) as it is not clear if a full sequence is shown or not. Overall comment about the multiple sequence alignment (relevant to Fig. S1): with such a small number of sequences it is very hard to make any substantial predictions about conserved regions etc.

-Thanks for the suggestions. We have now included more sequences, sequence numbering, and description of species as well as enzyme names. Some other changes are also that we have now used the same G. intestinalis dAK sequence in the alignment as in the experiments (same strain and accession number), and that we have made a realignment using Clustal W instead of Clustal Omega (gives better alignment of the termini). The designation of supplementary figures is according to the style of PLoS journals.

Fig. 1 and elsewhere: I will prefer that all bar graphs show individual values + the error bar (if possible);

-We have now added individual values to the bar graphs.

I do not have any issues with X-ray data and cryo-EM studies (refinement statistics, particles classification etc).

**Referees cross-commenting**

I also agree with all the comments provided by Reviewer 2 and very pleased to see that we were very similar in our evaluations.

Reviewer #1 (Significance (Required)):

In their paper Ranjbarian and colleagues provide a tour de force at characterizing dAK from G. intestinalis using both enzymology and structural biology. G. intestinalis does not have RNR and therefore this organism relies on dAK which catalyzes formation of dAMP and ADP from deoxyadenosine and ATP (among other substrate pairs). The authors performed a terrific job at testing this reaction in depth using a recombinant dAK and a battery of various co-substrates (both natural as well as synthetic ones, Table 1). Extensive structural information on dAK was obtained using a combination of X-ray crystallography and cryo-EM. Overall, this work will be paramount aid in better understanding of the reaction mechanism, especially in the context of molecules which can be used as inhibitors of such a crucial enzyme (metabolic vulnerability for this parasite).

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

Summary:

Ranjbarian et al. investigated a non-TK1-Like deoxyribonucleoside kinase (dNK) found in the protozoan parasite Giardia intestinalis. They used enzyme kinetic assays on heterologously expressed Gi dNK in E. coli to determine which deoxyribonucleotides were most likely physiological substrates for the enzyme. Their characterization revealed that this Gi dNK has a strong affinity to deoxyadenosine. They further investigated the affinity and activity of the dNK on deoxyadenosine analogues, some of which have known pharmaceutical utility. Finally, using a combination of crystallography, cryo-EM, chromatography, and mass photometry, they reveal that unlike other dNKs, Gi dNK forms a tetramer. They characterize important regions required for tetramerization and postulate that this tetramerization evolved to provide Gi dNK with a heightened affinity for deoxyadenosine.

Major comments and questions:

• The claims in this manuscript are well-supported, and I found no major issues with experimental methods. • The authors provide a structure of tetrameric dNK and suggest that this tetramer leads to the increased affinity to substrate compared to non-giardia dNKs. They also show through mutations that removing the novel dimerization regions decreases substrate affinity by 100-fold. However, I was left unclear about why the tetramer would lead to such high affinity for substrate compared to two dimers. This is especially notable, since the authors state that there are no signs of cooperativity, which is a common way that oligomerization may lead to heightened affinity. If the authors have no current evidence explaining this, they can consider adding a short amount of discussion speculating on the mechanism and future directions of study. -Thanks for this suggestion. We have now added a section in the last paragraph of the discussion where we speculate on the subject.

Minor comments and questions:

• The authors state that dATP acts as a mixed inhibitor and not a simple competitive inhibitor, and that previous studies have shown that this is because the dNTP competes in two locations (line 163). Is it also possible that competitive inhibition + allosteric regulation could be causing this behavior instead? -It is true that this can be theoretically explained in many ways. In fact, many allosteric regulators affect both the Vmax and Km values. However, in all studied dNKs, the dNTP acts as a dual competitor and no proper allosteric regulation with a separate allosteric site has ever been observed so far. We have rephrased this part as following to make it clear: "Mixed inhibition is often the result of allosteric regulation but studies of other dNKs have shown that this is not the case [17]. Instead, the far-end dNTP product gives a dual inhibition where the deoxyribonucleoside moiety competes with the substrate and the phosphate groups mimic those of ATP but coming from the opposite direction."

• In the introduction (line 93), non-TK1-like dNKs are described as "not structurally related to TK1-like". This left me unclear, are they still interrelated among themselves? -We have added the following sentence for clarification: "The non-TK1-like dNKs are further subdivided into a monophyletic group of canonical non-TK1-like dNKs and a second group with thymidine kinases from Herpesviridae, which are structurally related to the canonical group but share very little amino acid sequence homology."

• I was left confused by lines 106-116 in the introduction, where the specificities of dNKs in giardia are discussed. This is touched upon again in the discussion, but it was not clear here that there are several deoxyribonucleotides unaccounted for. -We think this should be clear now with the added Fig. 1 where the dNKs are shown.

• When describing enzyme assays (Line 145), the authors say there is no salt dependence, but there looks to be MgCl2 always included in the assays (presumably for the ATP). -This is a good point and something we have overlooked when the sentence was written (Mg2+ is required). We have now corrected the sentence as follows: "Based on initial enzyme activity studies, it was confirmed that the assay did not have any specific requirements regarding K+, Na+, NH4+, acetate or reducing agents, and that it was linear with respect to time (S2 Fig)."

• I was confused by the y-axis of Fig 2. How is enzyme activity lower when dAdo is added? I think I read "enzyme activity" as total substrate depleted, when it is actually referring exclusively to the given non-dAdo substrate in each column. -This is a very good point that we seem to have overlooked. We have now adjusted the y-axis title to "Indicated enzyme activity" and added the following sentence to the figure legend: "The recorded enzyme activities are for the substrates indicated on the x-axis (excluding the activity with the competing substrate)."

• Lines 239 - 255 and Figure 3 were a little unclear to me. Specifically, I was having trouble following in the text which dimer is in the ASU, which is symmetry related, and matching those terms with which are canonical and non-canonical. -We agree with the reviewer and thank them for their comment. In order to improve the presentation of these results, we chose to extensively rearrange the figures and accompanying text. We now present the initial X-ray data together with the cryo-EM data in a new Figure 4 that focuses on the overall architecture of the tetramer. We realize that some of the nomenclature previously used in that figure was, as the reviewer pointed out, confusing and superfluous, and we have now simplified and unified it. The structural details of how the extended N- and C-termini interact with the neighboring subunit have been moved to the new figure 6 in order to present them just before the functional analyses of the consequences of truncating the termini. As a consequence of these changes to the figure layout, we made substantial changes in the organization of the text surrounding these figures, which also led to a clearer presentation. Since the changes to the figures and text are quite substantial, we would like to point out that they are only changes to the presentation, not to the data shown.

• The authors suggest that in the experiment shown in Figure S9 (Line 285), low activity may be caused by minor impurities. I'm not sure why impurities would lower activity significantly. Could there be other differences in experimental conditions that are at play instead? -The sentence refers to a side activity (dATP dephosphorylation) which is not the normal reaction of deoxyadenosine kinase. We have rephrased the sentence to make it clearer: "The dATP-dephosphorylating activity was several orders of magnitude lower than the regular dAK activity (to phosphorylate deoxyadenosine) and was possibly catalyzed by other enzymes present as minor impurities in the protein preparation."

• (Optional) From looking at the crystallography stats, I think the authors can potentially push the resolution more. At higher resolutions, Rmerge may become high, but depending on the data collection strategy, Eiger detectors can lead to high Rmerge just out of sheer data redundancy. Cc 1/2 can be a more useful metric in these contexts. -This is a good point and well spotted by the reviewer. Indeed, a CC1/2 of 0.802 suggests that the resolution can be pushed further. However, due to contaminating spots at higher resolutions the statistics significantly worsen when trying to push the resolution beyond 2.1 A, which is why we did not process the data to a higher resolution.

• For Figure S8, the Polder map feature in Phenix is another option for showing ligand occupancy in an unbiased way. Did the authors try this? -We want to thank the reviewer for suggesting this. We have calculated a polder map using the Polder map feature in Phenix and both the resulting map and correlation coefficients support the presence of a dADP in the active site of monomer I. We added a section to the relative paragraph to include these new findings: "To increase our confidence that dADP was correctly placed within active site I, we calculated a polder map for dADP to test whether the b-phosphate density is correctly attributed or if it rather belongs to the bulk solvent. The resulting polder map and statistics support the placement of dADP in active site I with correlation coefficients of CC1,2=0.7627, CC1,3=0.9424, and CC2,3=0.7423 suggesting that the density does belong to dADP as CC1,3 > CC1,2 and CC1,3 > CC2,3 (S8 Fig.)."

• It's disappointing that the tetramers show so much preferred orientation in the cryo-EM. With that said, while the nominal resolution is 4.8 Å, I think that with the streakiness the EM structure looks to have worse resolution than that. -We agree that the streakiness of the map is substantial. This is simply a result of the severe anisotropy of the map, which means that the resolution is probably worse than 4.8 Å in the "bad directions" of the map. The supplementary material (S9 Fig) clearly shows the preferred orientations leading to this problem. In the course of this study, we tried several methods to lessen the preferred orientation problem such as using graphene oxide-coated grids and collecting tilted data. However, when we got the crystal structure we saw no point in continuing these efforts. To address the comment of the reviewer, we extended the description of the EM map in the main text to say:

"Due to strong preferred orientations, it was not possible to get an isotropic, high-resolution 3D structure of dAK using cryo-EM. The resulting 3D map had a nominal resolution of 4.8 Å, but a clearly anisotropic appearance probably reflecting lower resolution in the poorly resolved direction (S9 Fig)."

**Referees cross-commenting**

Overall, I agree with Reviewer #1's evaluation, and don't have any further suggestions or thoughts at this time.

Reviewer #2 (Significance (Required)):

Medical relevance: G. intestinalis is a parasite that causes 190 million cases of giardiasis per year. While treatable, there is evidence that giardia are developing a resistance to the main treatment at the moment, metronidazole. Thus, the authors provide a compelling case for the medical relevance of their investigation of Gi dNK for further pharmaceutical development. They provide further evidence for this by showing that several deoxyadenosine analogs bind the dNK and inhibit giardia growth. This work represents a very useful first step into a potential avenue for medical development. It's important to note that clinical studies are not within the purview of this research. However, in the discussion, the authors provide several comments on the promise of this avenue for future research.

Conceptual, technical, and mechanistic relevance: Through biochemical and structural study, the authors provide a compelling framework to understand an enzyme that is very important to the unique lifestyle of giardia parasites. From an evolutionary standpoint, the authors provide insight into how giardia can survive even without major components of de novo DNA synthesis. The authors principally use well-established tools and techniques of the enzymology field. but do so to characterize a unique and previously uncharacterized enzyme system. This enzyme proves to be notable not just for its medical significance, but because it is unique among its family (non-TK1-like deoxynucleotide kinases) in its strong affinity for substrate and tetrameric quaternary structure. One relatively novel technique used in the study is mass photometry, which is a relatively new and exciting way to characterize native proteins at very low concentrations. Using this technique helps the authors overcome a common criticism of structural studies in which the high concentrations or crowding conditions of techniques like crystallography and cryo-EM may be inducing non-physiological oligomers.

In summary, this work represents a meaningful addition to the protein structure-function literature. While it will principally be of interest to basic/fundamental researchers who study the mechanistic detail of protein function and evolution, it also provides a foundation for future translational work and antiparasitic drug design.

Reviewer's background: I received my PhD in chemistry studying the structure and function of another enzyme key to DNA metabolism (except in giardia), ribonucleotide reductase. My background is in structural biology and biochemistry. I do not have sufficient expertise to comment on studies performed on G. intestinalis growth and susceptibility to deoxyadenosine analogs.

• *

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:

Ranjbarian et al. investigated a non-TK1-Like deoxyribonucleoside kinase (dNK) found in the protozoan parasite Giardia intestinalis. They used enzyme kinetic assays on heterologously ex-pressed Gi dNK in E. coli to determine which deoxyribonucleotides were most likely physiological substrates for the enzyme. Their characterization revealed that this Gi dNK has a strong affinity to deoxyadenosine. They further investigated the affinity and activity of the dNK on deoxyadenosine analogues, some of which have known pharmaceutical utility. Finally, using a combination of crys-tallography, cryo-EM, chromatography, and mass photometry, they reveal that unlike other dNKs, Gi dNK forms a tetramer. They characterize important regions required for tetramerization and pos-tulate that this tetramerization evolved to provide Gi dNK with a heightened affinity for deoxy-adenosine.

Major comments and questions:

• The claims in this manuscript are well-supported, and I found no major issues with experi-mental methods.
• The authors provide a structure of tetrameric dNK and suggest that this tetramer leads to the increased affinity to substrate compared to non-giardia dNKs. They also show through mu-tations that removing the novel dimerization regions decreases substrate affinity by 100-fold. However, I was left unclear about why the tetramer would lead to such high affinity for substrate compared to two dimers. This is especially notable, since the authors state that there are no signs of cooperativity, which is a common way that oligomerization may lead to heightened affinity. If the authors have no current evidence explaining this, they can con-sider adding a short amount of discussion speculating on the mechanism and future direc-tions of study.

Minor comments and questions:

• The authors state that dATP acts as a mixed inhibitor and not a simple competitive inhibitor, and that previous studies have shown that this is because the dNTP competes in two loca-tions (line 163). Is it also possible that competitive inhibition + allosteric regulation could be causing this behavior instead?
• In the introduction (line 93), non-TK1-like dNKs are described as "not structurally related to TK1-like". This left me unclear, are they still interrelated among themselves?
• I was left confused by lines 106-116 in the introduction, where the specificities of dNKs in giardia are discussed. This is touched upon again in the discussion, but it was not clear here that there are several deoxyribonucleotides unaccounted for.
• When describing enzyme assays (Line 145), the authors say there is no salt dependence, but there looks to be MgCl2 always included in the assays (presumably for the ATP).
• I was confused by the y-axis of Fig 2. How is enzyme activity lower when dAdo is added? I think I read "enzyme activity" as total substrate depleted, when it is actually referring ex-clusively to the given non-dAdo substrate in each column.
• Lines 239 - 255 and Figure 3 were a little unclear to me. Specifically, I was having trouble following in the text which dimer is in the ASU, which is symmetry related, and matching those terms with which are canonical and non-canonical.
• The authors suggest that in the experiment shown in Figure S9 (Line 285), low activity may be caused by minor impurities. I'm not sure why impurities would lower activity sig-nificantly. Could there be other differences in experimental conditions that are at play in-stead?
• (Optional) From looking at the crystallography stats, I think the authors can potentially push the resolution more. At higher resolutions, Rmerge may become high, but depending on the data collection strategy, Eiger detectors can lead to high Rmerge just out of sheer data redundancy. Cc 1/2 can be a more useful metric in these contexts.
• For Figure S8, the Polder map feature in Phenix are another option for showing ligand oc-cupancy in an unbiased way. Did the authors try this?
• It's disappointing that the tetramers show so much preferred orientation in the cryo-EM. With that said, while the nominal resolution is 4.8 Å, I think that with the streakiness the EM structure looks worse resolution than that.

Referees cross-commenting

Overall, I agree with Reviewer #1's evaluation, and don't have any further suggestions or thoughts at this time.

Significance

Medical relevance: G. intestinalis is a parasite that causes 190 million cases of giardiasis per year. While treatable, there is evidence that giardia are developing a resistance to the main treatment at the moment, metronidazole. Thus, the authors provide a compelling case for the medical relevance of their investigation of Gi dNK for further pharmaceutical development. They provide further evi-dence for this by showing that several deoxyadenosine analogs bind the dNK and inhibit giardia growth. This work represents a very useful first step into a potential avenue for medical develop-ment. It's important to note that clinical studies are not within the purview of this research. Howev-er, in the discussion, the authors provide several comments on the promise of this avenue for future research.

Conceptual, technical, and mechanistic relevance: Through biochemical and structural study, the authors provide a compelling framework to understand an enzyme that is very important to the unique lifestyle of giardia parasites. From an evolutionary standpoint, the authors provide insight into how giardia can survive even without major components of de novo DNA synthesis.

The authors principally use well-established tools and techniques of the enzymology field. but do so to characterize a unique and previously uncharacterized enzyme system. This enzyme proves to be notable not just for its medical significance, but because it is unique among its family (non-TK1-like deoxynucleotide kinases) in its strong affinity for substrate and tetrameric quater-nary structure. One relatively novel technique used in the study is mass photometry, which is a relatively new and exciting way to characterize native proteins at very low concentrations. Using this technique helps the authors overcome a common criticism of structural studies in which the high concentrations or crowding conditions of techniques like crystallography and cryo-EM may be inducing non-physiological oligomers.

In summary, this work represents a meaningful addition to the protein structure-function literature. While it will principally be of interest to basic/fundamental researchers who study the mechanistic detail of protein function and evolution, it also provides a foundation for future transla-tional work and antiparasitic drug design.

Reviewer's background: I received my PhD in chemistry studying the structure and function of another enzyme key to DNA metabolism (except in giardia), ribonucleotide reductase. My back-ground is in structural biology and biochemistry. I do not have sufficient expertise to comment on studies performed on G. intestinalis growth and susceptibility to deoxyadenosine analogs.

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 their paper Ranjbarian and colleagues provide a tour de force at characterizing dAK from G. intestinalis using both enzymology and structural biology. G. intestinalis does not have RNR and therefore this organism relies on dAK which catalyzes formation of dAMP and ADP from deoxyadenosine and ATP (among other substrate pairs). Authors performed a terrific job at testing this reaction in depth using a recombinant dAK and a battery of various co-substrates (both natural as well as synthetic ones, Table 1). Extensive structural information on dAK was obtained using a combination of X-ray crystallography and cryo-EM. Overall, this work will be paramount aid in better understanding of reaction mechanism, especially in the context of molecules which can be used as inhibitors of such crucial enzyme (metabolic vulnerability for this parasite).

This manuscript, in its current form does not require additional experiments but I would like to have a few aspects corrected/clarified, before it can be accepted for publication:

Line 30: "whereas the affinities for deoxyguanosine, deoxyinosine and deoxycytidine were 400-2000 times lower." Better not to use term "affinity" when KM or kcat/KM are implied (unless ITC was used to measure true Kds).

Line 31: "Deoxyadenosine analogues halogenated at the 2- and/or 2´-positions were also potent substrates, with comparable EC50 values as the main drug used today, metronidazole, but with the advantage of being usable on metronidazole-resistant parasites." Not sure this sentence is clear as written.

Line 55: "..G. intestinalis (synonymous to G. lamblia and G. duodenalis)..". Very nice that authors provide this information as it is usually a point of confusion i.e. multiple names for the same organism.

Line 61 and above as well: "Treatment regimes are mainly based on metronidazole and to a lesser extent other 5-nitroimidazoles...". MT is introduced a bit sporadically, and not completely clear which enzyme it inhibits and its mode of action." Common knowledge is that MT is known for its action in aerobic parasites/bacteria and known as Flagyl, where it is mode of action was linked to "activation" due to microaerophilic conditions. Mayve MT can be introduced after text starting from Line 71?

Line 70: what is "cyst-wall"?

Line 90: "The reaction is catalyzed by deoxyribonucleoside kinases (dNKs), which are.." I really do not like when in order to find a reaction which is catalyzed by an enzyme in a particular study one needs to dive into the literature, sometimes it requires a lot of time as in most of recent papers on the subject reactions catalyzed are not listed. Please add a Figure or a panel with reactions catalyzed by both dNKs families.

Line 96: "..was found to have a ~10-fold higher affinity to thymidine.." as I mentioned above I really do not like the usage of "affinity", when actually low KM is implied.

Line 113: "This does not match the current knowledge that there are three dNKs in total whereof one completely specific for thymidine. The lack of knowledge about these essential enzymes in the parasite has hampered the understanding of Giardia deoxyribonucleoside metabolism and hence its exploitation as a target for antiparasitic drugs." Very good rationale, as I mentioned above, I think a Figure needs to be introduced that depicts different enzymes involved in deoxyribonucleoside metabolism (both TK1 and non- TK1 members) in Giardia with clearly labeled all known paralogs and corresponding enzymatic reactions.

Line 132: Odd designation of supplementary figures, usually it is "Fig. S1" etc. The legend for Fig. S1 is not adequate, please add description of species and name of enzymes for all sequences shown. Also each sequences in alignment should start with number (a.a. number) as it is not clear if a full sequence is shown or not. Overall comment about the multiple sequence alignment (relevant to Fig. S1): with such a small number of sequences it is very hard to make any substantial predictions about conserved regions etc.

Fig. 1 and elsewhere: I will prefer that all bar graphs show individual values + the error bar (if possible);

I do not have any issues with X-ray data and cryo-EM studies (refinement statistics, particles classification etc).

Referees cross-commenting

I also agree with all the comments provided by Reviewer 2 and very pleased to see that we were very similar in our evaluations.

Significance

In their paper Ranjbarian and colleagues provide a tour de force at characterizing dAK from G. intestinalis using both enzymology and structural biology. G. intestinalis does not have RNR and therefore this organism relies on dAK which catalyzes formation of dAMP and ADP from deoxyadenosine and ATP (among other substrate pairs). Authors performed a terrific job at testing this reaction in depth using a recombinant dAK and a battery of various co-substrates (both natural as well as synthetic ones, Table 1). Extensive structural information on dAK was obtained using a combination of X-ray crystallography and cryo-EM. Overall, this work will be paramount aid in better understanding of reaction mechanism, especially in the context of molecules which can be used as inhibitors of such crucial enzyme (metabolic vulnerability for this parasite).

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

Learn more at Review Commons

---

Reply to the reviewers

Response and revision plan

Manuscript number: RC- 2024-02380

Corresponding author(s): Emma R Andersson

1. General Statements

We sincerely appreciate the thorough and positive review provided by all reviewers. Their comments have provided valuable suggestions to improve and enhance clarity of our study on the role of Jag1-mediated Notch signaling in cochlear development, and its implications for Alagille syndrome. Furthermore, their feedback has underscored the significance of our study in elucidating patterning and hearing deficits, and its relevance for therapeutic considerations*. *

2. Description of the planned revisions

Comment from BioRxiv

In addition to comments from appointed reviewers, Jaime García-Añoveros emailed us with a comment on our BioRxiv preprint. Professor García-Añoveros was interested in our finding thatTbx2 is expressed in OHC-like cells (Fig5), because his lab has shown that Tbx2 is an inner hair cell determinant (García-Añoveros et al., 2022). Fig 5 shows quantifications of Tbx2 RNAscope punctae in sections, showing that Tbx2 is expressed in Jag1Ndr/Ndr outer hair cell-like cells, in the inner hair cell compartment, at similar levels to that expressed by the extra inner hair cells also present in Jag1Ndr/Ndr mice. He suggested we perform RNAscope for Tbx2 on wholemount cochlear preparations, to confirm the Fig 5 data from cross sections. While we are confident of our quantifications, which were based on optical slice sections Reviewer comments

We have already implemented some of the reviewer suggestions, as detailed under point 3, and the list below is therefore discontinuously numbered.

Reviewer 1

*Comments regarding quality of images: the picture quality for Figure 4b is low, especially for F-actin staining. Please enhance the intensity. (check image). Fig. 1g, poor quality. The WT cochlea looks severely disorganized. (replace image) *

Response

Figure 4b and Fig1g images will be improved or replaced. We plan a more extensive analysis of the adult phenotype, to also address comment #1 from Reviewer 2 (described below in response to Reviewer 2, #1).

---

Reviewer 2

Fig1g shows a very abnormal cross section through the cochlear duct. There are no clearly visible Deiters' cells. Is this the case? Loss of outer hair cell function should only increase thresholds about 40dB, and there are increased thresholds reported here of 60+, despite remaining outer hair cells. This could be accounted for by the conduction defects, but also, there may be defects in the adult ear not observed earlier. Is there any inner hair cell loss? Deiter cell loss? Are inner and outer hair cell stereocilia normal? These may account for the severe hearing loss.

• *

Response

To further characterize the adult cochlear phenotype, we will quantify the number of IHCs, OHCs and SCs with immunohistological staining of cryosections from adult Jag1Ndr/Ndr mice, and address in the Discussion section how this phenotype relates to the observed hearing loss. Additionally, we plan to analyze ABR wave-I characteristics of existing recordings to further study auditory nerve fiber responses and IHC function.

We have added a discussion of the relative contribution of middle and inner ear defects to the overall hearing loss in the Discussion section (lines 385-399), to also address comment #3 from reviewer 1 (below in section 3 "revisions that have been already incorporated in transferred manuscript").

Reviewer 2

*What is the rationale for reporting differences in the p-value that are not significant at the adjusted p-value? Since these are whole genome analysis it is only appropriate to report significance by adjusted p-values. *

*One of the novel aspects of this study is the finding that Notch components are upregulated in the Jag1Ndr/Ndr mutants (although some of these results are not significant at the adjusted p value). Given the potential significance that these results would indicate (including c-inhibition), it would be important to confirm upregulation of key Notch components in situ using RNA-scope or immunohistochemistry. *

Response

We agree that multiple hypothesis testing should be corrected for (with adjusted p values), which we have done in all analyses. However, we considered it relevant to report enriched or depleted genes that reached a meaningful fold difference and p-value threshold, even though the adjusted p-value threshold was not met. Our hope was that this would provide transparency and allow for consideration of the different sample sizes (different abundance of specific cell types), allowing the reader to explore the data. For further transparency, a distinction in labelling of significant adj. p-values and p-values was previously made in the original manuscript.

We thank the reviewer for pointing out that the Notch target gene upregulation is an interesting and novel finding. We will perform RNAscope experiments to validate the upregulation of Notch components and target genes at P5, including Jag1, Jag2, Hes5, Nrarp, Tns1 and Cxcl12. Quantification of the RNA scope signal will also provide an alternative approach to testing whether the enrichment/upregulation of Notch target genes is statistically significant.

---

__Reviewer 1 __

Text and figure comments: Scale bar missing in Figure1b and Figure1h. Please mention the scale bar presented mm in the figure legends for Figure 2; Figure 3; SFigure 6.

Response

Scale bar information will be added to the specified figures.

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

Reviewer 1

Developmentally hair cells develop from the base to the apex starting from the IHC to OHC. The observation of the changes in HC pattern indicates the impact of Notch in timing and maturation status of HC differentiation. Likely by the time when OHCs are supposed to be developed, which is dictated by the suppression of IHC and the activation of OHC signals, due to the dysregulation of Jag1, the IHC signaling cannot be sufficiently suppressed, whereas the OHC signaling cannot be sufficiently activated. This has a positional effect as further it is from the IHCs, more mature OHC can develop. Could the authors dig deeper into the scRNAseq data to see if they can isolate the profile of extra IHCs in the JagNdr/Ndr mouse, to see if they can detect the expression of some OHC genes albeit at much lower levels?

Response

There were no significant gene expression differences between Jag1Ndr/Ndr and Jag1+/+ IHCs. As we expect the Jag1Ndr/Ndr IHC pool to contain similar numbers of de facto IHCs and ectopic IHCs, failure to detect any differences suggests that the ectopic IHCs are transcriptionally similar to de facto IHCs. To further address the ectopic IHC signature, we subsetted, renormalized and reclustered the Jag1Ndr/Ndr and Jag1+/+ IHCs. No Jag1Ndr/Ndr-specific clusters were identified in this analysis (new Supplementary Fig 4c). In addition, we analysed the expression of IHC- and OHC-specific markers to assess the faithfulness of Jag1Ndr/Ndr IHCs and OHCs. As reported in our original manuscript, Jag1Ndr/Ndr OHCs expressed lower levels of OHC markers. However, Jag1Ndr/Ndr IHCs were indistinguishable from *Jag1+/+ * IHCs (new Supplementary Fig 4b). These new analyses also address comment #2 by Reviewer 2 (see below).

As the reviewer pointed out that development of HCs occurs from base to apex, we have added a quantification of apex and base regions of the P5 phenotype to Sfig5 and described this data in the Results section (lines 230-231).

• *

Reviewer 1

It is difficult to dissect the contribution of middle ear malformation and inner ear defects to hearing loss in Alagille syndrome with the current model. For the development of any therapy, the two main factors have to be analyzed separately. One option is to generate an inner ear-specific JagNdr/Ndr model to bypass the middle ear issue, which can be evaluated for potential therapy. This part should be discussed.

Response

We agree that the relative contribution of middle and inner ear defects to hearing loss in a Jag1-compromised setting cannot be assessed with Jag1Ndr/Ndr mice. Generation of an inner-ear specific Jag1 Nodder model to bypass middle ear defects and address the relative contribution of middle and inner ear defects, would be technically challenging/impossible since the Nodder mouse model carries a single missense mutation in Jag1 and must be carefully maintained on a mixed genetic background to fully recapitulate Alagille syndrome. However, previous elegant work from other groups has dissected the function of Jag1 in supporting cells and neural crest, and how defects in each of these systems contribute to hearing loss. We therefore now comprehensively discuss this work by others (lines 385-399).

Reviewer 1

*In Figure 1, the author mentioned the major defects found in the vestibular system. Is there any difference in the vestibular system at the cellular level? Some evidence will be informative. *

Jag1Ndr/Ndr mice completely lack the posterior semicircular canal, which explains the head nodding behavior observed in our model, since the posterior semicircular canal detects head-tilting towards the shoulders. We have no data on the hair cells located in the saccule or utricle. Since the paper focusses on patterning and hearing, rather than balance, we consider further analysis of the vestibular system at cellular level outside of the scope of our paper.

---

Reviewer 2

From the UMAP plot in Fig 2b, it seems that the scRNA-seq data did not reveal any change in cell identities in the Jag1Ndr/Ndr ears. This result is not really discussed in the results or discussion-particularly why the OHC-like cells, extra IHCs, and absent Hensen's cells are not revealed in this analysis.

Response

In our scRNAseq dataset we were unable to identify, with certainty, an OHC-like population. After subsetting HCs, we did observe an additional OHC population exclusive to homozygous animals. However, after RNAscope validation, this population might have arisen from contamination with PCs. IHCs were transcriptionally similar between wildtype and homozygous animals, and we were unable to identify the ectopic IHCs. We additionally reported fewer to almost absent HeCs in the homozygous dataset. This data has been shown in the Results section (Fig2b) and in Supplementary Table 8 (number of cells per cell type) and has been discussed in the Discussion section. To further address the lack of separation of IHCs and ectopic IHCs, and failure to identify OHC-like cells, we have added additional panels assessing IHCs and OHC gene expression to SFigure4. This also addressed comments #2 addressed by Reviewer 1 (see above).

Reviewer 2

*It is difficult to know which cells are extra (+1), including inner hair cells. Since scRNAseq did not reveal a different gene signature for these 'extra' cells, it is more appropriate to just count them all together. *

Response

We have merged the quantification of IHCs and +1 IHCs to total IHCs in Fig4c. Separate original quantification of IHCs and +1 IHCs is reported in SFigure5, since the data presented in this way reflect a doubling of the IHC row.

Reviewer 2

Additionally, a previous report has suggested that JAG1 mediates cis-inhibition in the medial region of the cochlea. The data presented here do not show an upregulation of Notch signaling in the medial supporting cells, suggesting this is not the case. This should be discussed.

Response

It is indeed interesting to note that, although with comparable sample size for medial and lateral populations, upregulation of Notch activation is restricted to lateral SCs, and not, despite previous indications (Basch et al., 2016), observed in medial SC populations. We have discussed the possibility for cis-inhibition to a greater extent in the Discussion section (lines 310-311).

---

Reviewer 2 and Reviewer 3

*Pg 9 Discussion: The sentence: "The JAG1NDR missense mutant is expressed in vivo, and traffics normally, but does not bind or activate NOTCH1", is somewhat misleading because it suggests this allele has no function. Based on the milder ear phenotype to null alleles as well as survival suggests that this allele is hypomorphic. This should be clarified and discussed. *

• *

The authors should provide a more detailed description of the Nodder mice (the nature of the mutation and how it may effect Notch1 and Notch2 receptor activation) in the introduction.

Response

We now introduce the Nodder mouse model (Hansson et al., 2010) and signaling defects to a greater extent in the Introduction section (lines 66-68).

Reviewer 2

Pg 5 third paragraph, "Differential gene expression analysis identified 40 up- and 42-downregulated genes in Jag1Ndr/Ndr versus Jag1+/+ IPhCs, with pathway dysregulation similar to the pseudobulk analyses (Fig3c, Supp.Table 5,6)"-should be 40 downregulated and 42 upregulated. Similarly: Pg 6 second paragraph: Differential gene expression analysis identified 1 up and* 42-downregulated genes in Jag1Ndr/Ndr DCs versus Jag1+/+ DCs-should be 1 down and 42 up. *

Response

Thank you for catching our accidental inversion here. The text has been corrected accordingly.

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 1

To study how Jag1 insufficiency affects the development, the authors included the JagNdr/Ndr mouse model. To fully understand the characteristics of the Nodder mouse model, it's necessary to include the direct age-dependent comparison of the Jag1 level (by qPCR/and or Western blot) between Jag1+/+ v.s. from JagNdr/Ndr in Figure 1 at some selected stages to correlate the Jag1 insufficiency with the "Nodder" model. A spatial expression comparison of Jag1 between Jag1+/+ v.s. from JagNdr/Ndr from different the main age groups should be included in SFigure 2, together with Notch target genes.

The JAG1 Nodder mutation results in a hypomorphic ligand that is unable to bind and activate the Notch1 receptor (Hansson et al., 2010). The ligand itself, however, is still expressed, and its protein expression can even be upregulated in vivo (Hansson et al., 2010). Therefore, performing quantitative expression analysis of JAG1 expression (by qPCR or immunohistochemistry) would not provide insights into the levels of JAG1 activity. Instead, we show that there is decreased Notch target gene expression at the prosensory domain stage, as a proxy for Notch activation levels (SFigure2. A more detailed introduction of the model is provided in the Introduction (lines66-68), to also address a comment from Reviewer 2, #7 and Reviewer 3 comment #2.

Reviewer 3

The mutant form of Jagged1 in Nodder mice is trafficked to the cell surface, and while this mutant form of Jagged1 is incapable of activating the Notch1 receptor it may interact with "new" proteins, gaining new functions. My recommendation to the authors is to determine whether similar defects occur in conditional Jag1 knockout mice (increased Notch signaling in lateral supporting cells and presence of ectopic outer-hair cell like cells). The ability to disrupt Jag1 function at different stages of development may also help to determine why Jag1 deficiency renders some outer hair cells insensitive to Tbx2. If this is not possible due to time constrains, I would recommend a more in-depth discussion of the limitations of using Nodder mice.

---

Jag1 conditional knockout at various stages, has not been reported to result in ectopic OHC-like cells (Brooker et al., 2006; Chrysostomou et al., 2020; Gilels et al., 2022). However, two other Jag1 missense mutants display atypical hair cells in the IHC compartment, which could be the OHC-like cells we report here (Kiernan et al., 2001; Tsai et al., 2001). Taken together, these data would suggest that Jag1 loss of function in supporting cells is not sufficient to result in OHC-like cells, but that constitutive Jag1 insufficiency can drive OHC-like cell formation. We now cite these data and discuss possible interpretations, as suggested (lines 324-331).

References

Basch, M. L., Brown, R. M., Jen, H.-I., Semerci, F., Depreux, F., Edlund, R. K., Zhang, H., Norton, C. R., Gridley, T., Cole, S. E., Doetzlhofer, A., Maletic-Savatic, M., Segil, N., & Groves, A. K. (2016). Fine-tuning of Notch signaling sets the boundary of the organ of Corti and establishes sensory cell fates. ELife, 5, 841-850. https://doi.org/10.7554/eLife.19921

Brooker, R., Hozumi, K., & Lewis, J. (2006). Notch ligands with contrasting functions: Jagged1 and Delta1 in the mouse inner ear. Development, 133(7), 1277-1286. https://doi.org/10.1242/dev.02284

Chrysostomou, E., Zhou, L., Darcy, Y. L., Graves, K. A., Doetzlhofer, A., & Cox, B. C. (2020). The notch ligand jagged1 is required for the formation, maintenance, and survival of Hensen's cells in the mouse cochlea. Journal of Neuroscience, 40(49). https://doi.org/10.1523/JNEUROSCI.1192-20.2020

García-Añoveros, J., Clancy, J. C., Foo, C. Z., García-Gómez, I., Zhou, Y., Homma, K., Cheatham, M. A., & Duggan, A. (2022). Tbx2 is a master regulator of inner versus outer hair cell differentiation. Nature, 605(7909). https://doi.org/10.1038/s41586-022-04668-3

Gilels, F. A., Wang, J., Bullen, A., White, P. M., & Kiernan, A. E. (2022). Deletion of the Notch ligand Jagged1 during cochlear maturation leads to inner hair cell defects and hearing loss. Cell Death and Disease, 13(11). https://doi.org/10.1038/s41419-022-05380-w

Hansson, E. M., Lanner, F., Das, D., Mutvei, A., Marklund, U., Ericson, J., Farnebo, F., Stumm, G., Stenmark, H., Andersson, E. R., & Lendahl, U. (2010). Control of Notch-ligand endocytosis by ligand-receptor interaction. Journal of Cell Science, 123(Pt 17), 2931-2942. https://doi.org/10.1242/jcs.073239

Kiernan, A. E., Ahituv, N., Fuchs, H., Balling, R., Avraham, K. B., Steel, K. P., & Hrabé de Angelis, M. (2001). The Notch ligand Jagged1 is required for inner ear sensory development. Proceedings of the National Academy of Sciences of the United States of America, 98(7), 3873-3878. https://doi.org/10.1073/pnas.071496998

Tsai, H., Hardisty, R. E., Rhodes, C., Kiernan, A. E., Roby, P., Tymowska-Lalanne, Z., Mburu, P., Rastan, S., Hunter, A. J., Brown, S. D. M., & Steel, K. P. (2001). The mouse slalom mutant demonstrates a role for Jagged1 in neuroepithelial patterning in the organ of Corti. Hum Mol Genet, 10(5), 507-512. https://doi.org/10.1093/hmg/10.5.507

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:

Patients with Alagille syndrome have impaired Notch signaling (~94% with JAG1 mutations) resulting in sensorineural and conductive hearing loss. To gain a better understanding of the genesis of these functional defects, the authors conduct a detailed examination of a mouse model of Alagille syndrome, called Nodder mice (Jag1 Ndr/Ndr). Consistent with previously reported phenotypes for Jag1 mutant mice, the authors observed severe vestibular and auditory defects that were accompanied by semicircular canal abnormalities, and defects in the patterning of the auditory sensory epithelium (medial boundary defect causing duplication of inner hair cells and inner phalangeal cells and reduction in outer hair cells and lateral supporting cells). What makes this study stand out from previous studies is the elegant use of single cell RNA-sequencing technology. Their scRNA_seq data suggest that 1) Jag1 lowers Notch signaling in lateral supporting cells and 2) Jag1 regulates the gene expression of outer hair cells. Further marker analysis revealed ectopic outer hair cell-like cells in the medial compartment and pillar cell region of Jag1 Ndr/Ndr mice. Surprisingly the outer hair cell-like cells closest to inner hair cells expressed Tbx2. Previous studies have shown that ectopic activation of Tbx2 is sufficient to convert outer hair cells into inner hair cells, suggesting that Jag1 deficiency may render these outer hair cells insensitive to Tbx2.

Overall, this is a very well-designed and executed study. The main conclusions of the study are well supported by the presented data and both data and methods are presented in a clear and detailed manner. I have only few suggestions for improvement:

Major comment:

The mutant form of Jagged1 in Nodder mice is trafficked to the cell surface, and while this mutant form of Jagged1 is incapable of activating the Notch1 receptor it may interact with "new" proteins, gaining new functions. My recommendation to the authors is to determine whether similar defects occur in conditional Jag1 knockout mice (increased Notch signaling in lateral supporting cells and presence of ectopic outer-hair cell like cells). The ability to disrupt Jag1 function at different stages of development may also help to determine why Jag1 deficiency renders some outer hair cells insensitive to Tbx2. If this is not possible due to time constrains I would recommend a more in-depth discussion of the limitations of using Nodder mice.

Minor comment:

The authors should provide a more detailed description of the Nodder mice (the nature of the mutation and how it may effect Notch1 and Notch2 receptor activation) in the introduction.

Significance

Strength of the study: The establishment and in-depth characterization of a Jag1 homozygous mutant mouse model (Nodder mice) to study the effects of Alagille syndrome on the auditory and vestibular system. Another strength is the characterization of the cell-type specific effects of Jag1 mutations using single cell transcriptomics.

Limitation of the study: It is unclear if the missense mutation in Jag1 that is present in Nodder mice causes "only" a loss of function. It is possible that a mutant form of Jag1 protein gains new functions (see also major comment).

Advance: This study demonstrates a novel role for the Notch ligand Jag1 in repressing Notch activation in lateral supporting cells and uncovers an involvement for Jag1-activated Notch signaling in inner versus outer hair cell specification and positioning.

Audience: This study will be of interest for both clinical and basic scientist as it provides novel insights into how Alagille syndrome effects the auditory system and novel mechanistic insights into the complex and cell-type specific role Notch signaling.

My field of expertise is: inner ear/ cochlea development, Notch signaling, cell fate specification and differentiation of cochlear supporting cells.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Alagille's syndrome is a developmental disorder in which the vast majority of patients have heterozygote mutations in the gene for the Notch ligand Jagged1. Patients with Alagille's have developmental defects in the heart, liver, eye and ear. This manuscript describes a hypomorphic new allele of Jag1 (Ndr mutation), in which homozygotes animals survive, allowing analysis of the postnatal and adult inner ear. The authors show that Jag1Ndr/Ndr mutants demonstrate hearing loss, vestibular defects, and patterning defects in the cochlea. Similar to previous studies of Jag1 in the ear, the authors find a decrease of outer hair cells, an increase of inner hair cells, and misplaced outer hair cells in the inner hair cell region. The authors perform scRNA-seq and analyze transcriptional effects in different cell populations. Interestingly, the authors present data that shows that the Notch pathway is upregulated in supporting cells at postnatal day 5, suggesting Jag1 may be playing an inhibitory role during postnatal maturation. The authors also show intriguing expression data regarding the OHC-like cells in the OHC region. Overall the study is well-performed and provides new information regarding the hair cell and supporting cell patterning defects caused by a reduction of Jag1-Notch signaling. However, most of the histological analysis was performed during development, and thus the origin of the severe hearing loss is not completely clear.

Comments

Major:

Fig1g shows a very abnormal cross section through the cochlear duct. There are no clearly visible Deiters' cells. Is this the case? Loss of outer hair cell function should only increase thresholds about 40dB, and there are increased thresholds reported here of 60+, despite remaining outer hair cells. This could be accounted for by the conduction defects, but also, there may be defects in the adult ear not observed earlier. Is there any inner hair cell loss? Deiter cell loss? Are inner and outer hair cell stereocilia normal? These may account for the severe hearing loss.

From the UMAP plot in Fig 2b, it seems that the scRNA-seq data did not reveal any change in cell identities in the Jag1Ndr/Ndr ears. This result is not really discussed in the results or discussion-particularly why the OHC-like cells, extra IHCs, and absent Hensen's cells are not revealed in this analysis.

It is difficult to know which cells are extra (+1), including inner hair cells and inner phalangeal cells. Since scRNAseq did not reveal a different gene signature for these 'extra' cells, it is more appropriate to just count them all together.

What is the rationale for reporting differences in the p-value that are not significant at the adjusted p-value? Since these are whole genome analysis it is only appropriate to report significance by adjusted p-values.

One of the novel aspects of this study is the finding that Notch components are upregulated in the Jag1Ndr/Ndr mutants (although some of these results are not significant at the adjusted p value). Given the potential significance that these results would indicate (including c-inhibition), it would be important to confirm upregulation of key Notch components in situ using RNA-scope or immunohistochemistry.

Additionally, a previous report has suggested that JAG1 mediates cis-inhibition in the medial region of the cochlea. The data presented here do not show an upregulation of Notch signaling in the medial supporting cells, suggesting this is not the case. This should be discussed.

Pg 9 Discussion: The sentence: "The JAG1NDR missense mutant is expressed in vivo, and traffics normally, but does not bind or activate NOTCH1", is somewhat misleading because it suggests this allele has no function. Based on the milder ear phenotype to null alleles as well as survival suggests that this allele is hypomorphic. This should be clarified and discussed.

Minor:

Pg 5 third paragraph, "Differential gene expression analysis identified 40 up- and 42-downregulated genes in Jag1Ndr/Ndr versus Jag1+/+ IPhCs, with pathway dysregulation similar to the pseudobulk analyses (Fig3c, Supp.Table 5,6)"-should be 40 downregulated and 42 upregulated.

Similarly: Pg 6 second paragraph: Differential gene expression analysis identified 1 up and 42-downregulated genes in Jag1Ndr/Ndr DCs versus Jag1+/+ DCs-should be 1 down and 42 up

Significance

This study corroborates and extends previous studies of the role of JAG1 in the inner ear. The role of JAG1 in cochlear development is not well understood compared to other Notch ligands, because it is expressed in the supporting cells and not the hair cells. The single cell RNAseq analysis presented here sheds new light on how JAG1 may be functioning postnatally. In addition, because this is a novel allele of JAG1 in which the homozygotes survive, we can further understand how the phenotype may affect hearing and balance in Alagille syndrome patients. This study will be interesting to those who study developmental biology, Notch signaling and the inner ear.

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

Notch signaling regulates inner and middle ear morphogenesis and establishes a strict pattern of sensory cells in the organ of Corti in the mammalian cochlea. In the paper, the authors investigate the function of Jag1-mediated Notch activation in cochlear patterning and signaling using a novel Jag1 "Nodder" (Jag1Ndr/Ndr) mouse model of Alagille syndrome. In the transgenic mouse model, they found that the mice exhibited severe vestibular and auditory defects including an increase in ectopic inner hair cells, and a reduction in outer hair cells. By single-cell RNA study of the organ of Corti, the authors demonstrated a global dysregulation of genes associated with inner ear development and deafness. They observed that the role of Tbx2 in IHC specification is likely influenced by Notch signaling. This was a well-designed study with quality data that provided valuable information on the effect of dysregulated Jag1 on human Alagille syndrome. Given the recent success of a gene therapy clinical trial for human genetic hearing loss, this study helps to answer some key questions related to hearing, which should be of significance guiding future development of potential therapy.

To study how Jag1 insufficiency affects the development, the authors included the JagNdr/Ndr mouse model. To fully understand the characteristics of the Nodder mouse model, it's necessary to include the direct age-dependent comparison of the Jag1 level (by qPCR/and or Western blot) between Jag1+/+ v.s. from JagNdr/Ndr in Figure 1 at some selected stages to correlate the Jag1 insufficiency with the "Nodder" model. A spatial expression comparison of Jag1 between Jag1+/+ v.s. from JagNdr/Ndr from different the main age groups should be included in SFigure 2, together with Notch target genes.

Developmentally hair cells develop from the base to the apex starting from the IHC to OHC. The observation of the changes in HC pattern indicates the impact of Notch in timing and maturation status of HC differentiation. Likely by the time when OHCs are supposed to be developed, which is dictated by the suppression of IHC and the activation of OHC signals, due to the dysregulation of Jag1, the IHC signaling cannot be sufficiently suppressed, whereas the OHC signaling cannot be sufficiently activated. This has a positional effect as further it is from the IHCs, more mature OHC can develop. Could the authors dig deeper into the scRNAseq data to see if they can isolate the profile of extra IHCs in the JagNdr/Ndr mouse, to see if they can detect the expression of some OHC genes albeit at much lower levels?

It is difficult to dissect the contribution of middle ear malformation and inner ear defects to hearing loss in Alagille syndrome with the current model. For the development of any therapy, the two main factors have to be analyzed separately. One option is to generate an inner ear-specific JagNdr/Ndr model to bypass the middle ear issue, which can be evaluated for potential therapy. This part should be discussed.

In Figure 1, the author mentioned the major defects found in the vestibular system. Is there any difference in the vestibular system at the cellular level? Some evidence will be informative.

Scale bar missing in Figure1b and Figure1h.

The picture quality for Figure 4b is low, especially for F-actin staining. Please enhance the intensity.

Please mention the scale bar presented m in the figure legends for Figure 2; Figure 3; SFigure 6.

Fig. 1g, poor quality. The WT cochlea looks severely disorganized.

Significance

Notch signaling regulates inner and middle ear morphogenesis and establishes a strict pattern of sensory cells in the organ of Corti in the mammalian cochlea. In the paper, the authors investigate the function of Jag1-mediated Notch activation in cochlear patterning and signaling using a novel Jag1 "Nodder" (Jag1Ndr/Ndr) mouse model of Alagille syndrome. In the transgenic mouse model, they found that the mice exhibited severe vestibular and auditory defects including an increase in ectopic inner hair cells, and a reduction in outer hair cells. By single-cell RNA study of the organ of Corti, the authors demonstrated a global dysregulation of genes associated with inner ear development and deafness. They observed that the role of Tbx2 in IHC specification is likely influenced by Notch signaling. This was a well-designed study with quality data that provided valuable information on the effect of dysregulated Jag1 on human Alagille syndrome. Given the recent success of a gene therapy clinical trial for human genetic hearing loss, this study helps to answer some key questions related to hearing, which should be of significance guiding future development of potential therapy.

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

Learn more at Review Commons

---

Reply to the reviewers

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

Summary: This work focuses on two small molecule inhibitors of the Arp2/3 complex, CK-666 and CK-869. Previous studies have shown that although the Arp2/3 complex is well conserved in eukaryotes, the inhibitory effect of these molecules is highly species dependent. However, it has been unclear whether these drugs act equally well on Arp2/3 iso-complexes (complexes composed of subunit isoforms from the same species). This paper fills that gap. Using human Arp2/3 iso-complexes, it shows that the inhibitory effect of these two drugs depends on the subunit composition of the complex. In addition, this work shows that these drugs do not systematically and equally inhibit the ability of these Arp2/3 complexes to nucleate linear or branched filaments.

We thank the reviewer for their positive comments.

Major comments:

1/ Regarding the first part on vaccinia-induced actin polymerization The first paragraph of the Results section is difficult to follow for those who have not read the previous papers from this lab. I would recommend changing the text so that any reader can understand from the start the experimental system and the goal of the experiment.

As requested, we have expanded the first section to better allow the reader to understand vaccinia actin-based motility as a model system to understand Arp2/3 iso-complex function.

The data analysis of Figure 1C is not satisfactory. It is not very informative to statistically compare the effect of the two drugs at similar concentration. However, it is necessary to perform statistical tests to compare the different conditions with drug with the control condition (DMSO). By eye, I see a difference between DMSO and CK-666, so it is difficult to understand why the authors claim that CK-666 has no effect on actin polymerization.

We are claiming that CK-869 but not CK-666 fully inhibits the ability of Vaccinia virus to stimulate Arp2/3 dependent actin polymerisation. We agree that CK-666 partially inhibits Vaccinia induced actin polymerization and have changed the text accordingly to reflect this. In contrast to CK-869 this level of inhibition does change from that seen with 50 µM when CK-666 is increased up to 300 µM. We believe this partial ~30% inhibition reflects the impact of CK-666 inhibiting ArpC1A containing Arp2/3 from generating actin filaments. These inhibited ArpC1A containing Arp2/3 complexes are able to bind the VCA domain of N-WASP (see figure 4) which will block its interaction with ArpC1B containing complexes. We now have provided the requested statistical analysis between drug and DMSO and also retained our original statistical analysis between the drugs.

Images with CK-869 have a lower overall cortactin signal, which could indicate that immunolabeling was not very effective in this condition. This could affect the analysis of the data in Figure 1C.

In the figure we used cortactin as a marker for branched actin filaments to assess the impact of CK-666 and CK-869 on the ability of individual vaccinia viruses to induce actin polymerization rather than the extent of actin assembly. In general, CK-869 does not impact on cortactin signal, however, the differences the reviewer is referring to are probably due to cell-to-cell variability. Moreover, we have now provided the corresponding image of the actin visualized with phalloidin as supplementary figure 1. In these images the same virus induced actin structures are visible.

The authors mention that the exact levels of the 8 different Arp2/3 iso-complexes are not known in these HeLa cells, but it should be fairly easy (e.g. mass spectrometry) to quantify the expression level of ArpC1, ArpC5 and Arp3 in these cells and verify that it is consistent with the rest of the story.

This information about the expression level of ArpC1, ArpC5 and Arp3 in HeLa cells is also very important because a large community of researchers use CK-666 and HeLa cells. There are actually quite few papers that draw conclusions from the use of CK-666 in HeLa cells, and the authors should discuss the limitations of these studies much more clearly.

In Abella et al. NCB 2016 we quantified the amounts of ARPC1 and ARPC5 isoforms in our HeLa cell. ArpC1A is 0.3 {plus minus} 0.02 ng/µg cells; ArpC1B is 0.7 {plus minus} 0.05 ng/µg cells; ArpC5 is 0.46 {plus minus} 0.03 ng/µg cells; ArpC5L is 0.27 {plus minus} 0.03 ng/µg cells. Thus, ArpC1B is approximately twice that of ARPC1A which fits with the ~30 % level of inhibition we see with CK-666 in figure 1C. Unfortunately, we do not have a specific antibody against Arp3B, so have not been able to use the same approach to quantify the level of this isoform. However, Arp3B is 18.5-fold less abundant than Arp3 in HeLa cells according to Hein et al., 2015 (PMID: 26496610 DOI: 10.1016/j.cell.2015.09.053). In an early study (Kulak et al., 2014 PMID: 24487582 DOI: 10.1038/nmeth.2834, the same group reported that Arp3 was 61.5 X more abundant than Arp3B in HeLa cells. These two papers illustrate the difficulty in using mass spec to determine absolute protein concentrations, which is why we prefer quantitative western blotting as done in Abella et al., 2016.

2/ The pyrene assays are disappointing because they are performed with only one concentration of CK-666 and CK-869. This is especially true for the VCA data, where the effect of the drugs is not always "on"/"off" as naively presented in the text, but highly concentration dependent. The authors should definitely provide several drug concentrations for each condition, up to saturation levels, to provide a clear quantification of the drug concentrations needed to reach half inhibition.

Following the reviewer's advice, we now have performed the pyrene and TIRF assays in the presence of a range of drug concentrations (see individual figures). These new data have allowed us to calculate the half-maximal inhibitory concentration values (IC50) which strengthen our previous conclusions. CK-666 can prevent ArpC1A (IC50 = 20 µM) but not ArpC1B (IC50 undetectable) from generating branches. Meanwhile, CK-869 can inhibit both ArpC1 isoforms efficiently with IC50

3/ Similarly, the pull-down experiments performed at a single protein concentration are inconclusive. They cannot tell us whether the affinity of the Arp2/3 isoforms for these targets is altered in the presence of the small molecule inhibitors because we do not know the degree of saturation of the ligands. Given that some of the reported differences in inhibition of filament nucleation are modest, it is not possible at this stage to link these different data.

Following the reviewer's advice, we repeated the pull down Arp2/3 at a higher F-actin concentration. In the initial submission we said we used 7.5 µM F-actin, however, we discovered a miscalculation, so it was actually 3 µM, which would explain the lower levels of Arp2/3 co-pelleting. In Hetrick et al 2013 (PMID: 23623350 DOI: 10.1016/j.chembiol.2013.03.019), the binding of Arp2/3 to F actin reaches a plateau at 15 µM F-actin. We therefore used 15 µM F-actin for the additional pull down experiments (Figure 4). The new results with 15 µM F-actin agree with our previous observations at 3 µM F-actin concentration.

We do not feel it is necessary to repeat the pull down of Arp2/3 by GST-VCA at different concentrations. This is because Arp2/3 binds VCA with high affinity (0.9 µM) Marchand et al. NCB 2001 (PMID: 11146629 DOI: 10.1038/35050590). Thus in our initial experimental conditions (5 µM VCA), the binding is already saturated. In addition, we did not see a difference in binding between Arp2/3 iso-complexes.

Reviewer #1 (Significance (Required)):

The subunit composition of the Arp2/3 complex is cell-type dependent, so these data will be important for the many cell biologists using these molecules. In particular, it calls for caution in the use of these drugs and in the interpretation of the data.

The writing is very clear, but the manuscript seems quite rushed. Many experiments need to be analyzed in much more detail to clarify the conclusions.

We thank the reviewer for their positive comments and suggestions to improve our study.

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

The manuscript 'CK-666 and CK-869 differentially inhibit Arp2/3 iso-complexes' addresses how commonly used Arp2/3 complex inhibitors differentially inhibit Arp2/3 complex activity based on the subunit isoforms making up the Arp2/3 complex. This work directly tests how each inhibitor affects different iso-complexes, which may affect different cell types based on the predominant iso-complex present in the cell. The manuscript is well written, with experiments both in cell culture and with purified proteins in reconstitution and biochemical assays to establish that these small molecule inhibitors have different effects based on the iso-complex of Arp2/3 present. There are several points in the manuscript that if addressed would improve and support the conclusions presented.

We thank the reviewer for their comments and suggestions.

In Figure 1B, looking at the images of the CK-666 treated verses the DMSO, it looks like the actin structures in the DMSO-treated cells are potentially larger than those in the CK666 cells, but because only an inset of drug-treated is shown, and an inset of the DMSO-treated is not shown it is hard to compare. Are the size of the virus-associated structures affected in the CK-666 treated cells versus the DMSO-treated cells? This might indicate that CK-666 has some effect on actin polymerization, even if it is not as drastic as the CK-869.

The reviewer is right that actin tails are shorter in CK-666 treated cells. This is because CK-666 does partially inhibit actin polymerisation induced by the virus. In contrast to CK-869, this level of inhibition does change with increasing concentration of CK-666. We believe this partial inhibition reflects the impact of CK-666 inhibiting ArpC1A containing Arp2/3 from generating actin filaments. These inhibited complexes will bind the VCA domain of N-WASP (see figure 4) blocking its interaction with ArpC1B containing complexes.

In Figure 2 comparing the pyrene curves in figure 1A, it appears that CK-869 has a different effect on C1B/C5+VCA versus C1B/C5L+VCA (green curves as compared to no activation control, grey curves), but this is not commented on. Addressing the differing effects would strengthen the authors conclusions- namely, that CK-869 inhibits both iso-complexes better than CK-666, but there may be some differences on each isoform. It is unclear if the differences in the branching rate (Figure 2B) is also reflective of this. The authors should address these results.

This is a very good point. We have now performed more detailed analysis, measuring the branching rate of C1B/C5 and C1B/C5L complexes in TIRF assays in different concentrations of CK-869 (Supplementary figure 2B). By comparing the half-maximal inhibitory concentration values (IC50) of CK-869 on the two different complexes, we found CK-869 inhibits C1B/C5 slightly better than C1B/C5L (1.8 µM as compared to 3.6 µM) as the reviewer suggested.

For Figure 4, it is somewhat unexpected that inhibition of the Arp2/3 complex increases macrophage motility as compared to control, unless the reader is familiar with the 2017 Rotty et al paper. The manuscript may benefit from a sentence or two explaining this result in light of the findings of the 2017 Rotty paper beyond simply mentioning that the increase in motility is dependent on myosin II.

As requested, we have provided more information.

The Spin90 data looks good, clear, and consistent.

We thank reviewer for the positive comments.

In Figure 7, given that pyrene was used in all the previous assessments of drug treatment on arp2/3 isoforms, it seems appropriate for these assays to be performed for Arp3B/C1B/C5L in comparison with Arp3/C1B/C5L and between the different drug treatments. Likewise, this should be done for the Spin90 also. It is difficult to compare between the figures for Arp3b vs. Arp3C (Figures 2 and 3 vs. Figure 7), although this may require a repetition of data presented.

We have now provided quantification of the maximum actin polymerization rate induced by Arp3B/C1B/C5L complexes obtained in pyrene assembly assays over a range of drug concentrations (requested by reviewer 1) (Figure 6C). These new data confirm that Arp3B is not inhibited by CK-869. We did not feel it was necessary to perform a side-by-side comparison with Arp3/C1B/C5L complexes but have provided quantification of the branching rate of Arp3/C1B/C5L complexes over a range of drug concentrations using TIRF assays (see Figure 2D).

Minor issues: It would be helpful if the labels for what is labeled in the micrograph were on the images (Figure 1B, Figure 3B, Figure 7A).

We have provided the requested labels.

In Figure 1-B, the 200uM CK-869 cell image looks less representative of the data in Figure 1C than other cells in the figure. Perhaps there is higher background in this micrograph, but it might be clearer if a cell with similar background actin signal to the other CK-869 was used.

As we responded to reviewer 1: In the figure we used cortactin as a marker for branched actin filaments to assess the impact of CK-666 and CK-869 on the ability of individual vaccinia viruses to induce actin polymerization rather than the extent of actin assembly. In general, CK-869 does not impact on cortactin signal, however, the differences the reviewer is referring to are probably due to cell-to-cell variability. Moreover, we have now provided the corresponding image of the actin channel visualised with phalloidin as supplementary figure 1. In these images the same virus induced actin structures are visible.

Figure 4: Where is the mean thickness of the cell measured? In figure 4D, it would be helpful if the error bars could be in the color of the line, as it is hard to distinguish the range of the data for each condition because the error bars are overlapping and the same color for all.

We used the Phasefocus Livecyte to imagine and quantify the morphology and behaviour of live cells. The mean thickness of the cell is quantified from the whole cell area based on the method described in Marrison et al. Scientific reports 2013 (PMID: 23917865 DOI: 10.1038/srep02369). We have clarified this fact in the figure legend. We have also corrected the colour issue with the error bars.

Figure 5: In Figure 5A, the labeling of the gels (KDa, S and P) do not line up correctly. The legend for the quantification should indicate what bands were quantified- all the arp2/3 bands or just the isoforms? It is unclear what is being quantified in the graph in C. The pull-down results in C should be quantified via quantitative western blot if possible.

We have provided new F-actin pulldown gels and have made sure the labels are aligned. The level of Arp2/3 binding to F-actin was determined by quantifying the level of bound ArpC3. This subunit was chosen as it is well removed from the other bands on the gel. We have now also provided quantification of the VCA pulldowns assays as requested.

The statement in line 170- 'indicate' seems a bit strong based on the results presented. 'Suggests' might work better here.

We have changed the text as suggested by the reviewer.

Reviewer #2 (Significance (Required)):

It is of general interest to members of the actin field as well as cell-biologists who routinely use either CK-666 or CK-869 to inhibit Arp2/3 complex activity in cells, and specifically in mammalian cells.

We thank the reviewer for their positive comments and suggestions.

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

Summary: Cao et al combine in vitro and cellular work to show that neither of the two distinct and frequently used Arp2/3 inhibitors is truly pan-selective, at least when considering distinct classes of activators. Using in vitro assays, they show that CK-666 cannot inhibit ARPC1B iso-complexes when activated by class I nucleation promoting factors. Similarly, Arp2/3 complexes containing Arp3B are refractory to inhibition by CK-869. The latter is likely the result of substitutions at the inhibitor-binding site. They go on to show that these differences correlate with differential effects of CK-666 and -869 on Vaccinia tail formation and macrophage cell shape and motility at the cellular level.

Major comments:

Figure1: The authors state that "...even at 300 μM, the number of virus-induced actin polymerisation events were not diminished (Figure 1B, C)..." The figure shows that CK-666 does indeed not fully abolish cortactin colocalization. However, there seems to still be a significant effect that is not tested for. Statistical tests were only used to compare the two inhibitors at the same concentration. I suggest also testing for significant differences to the DMSO control and reporting p-values, because CK-666 seems to still have an effect. Along the same vein, it seems that valuating the fraction of virus with cortactin co-localization as the only metric for branched actin nucleation downplays the effects of CK-666. Can the authors consider additional other metrics such as the amount of polymerized actin in individual tails or the tail length, which were extensively used in previous publications?

This point was also raised by the two other reviewers (see above). We have now provided the requested statistical tests. In our study we used Vaccinia as a model to examine whether there were differences between the impact of CK-666 and CK-869 on Arp2/3 dependent actin polymerization in cells. This is clearly the case, so we focused on in vitro assays where we can do experiments with defined Arp2/3 iso-complexes to better understand what was going on. Given the complexity of cellular systems, we feel that additional analysis of the changes to the actin tails will not provide additional molecular insights, especially as the factors the determine actin tail lengths are still not fully understood.

Figure2/3: The authors claim that "...the ArpC5/ArpC5L isoforms are not differentially impacted by either CK-666 or CK-869..." I am not convinced that this conclusion can be drawn based on the data. Figure 2 shows that the inhibitory effect of CK-869 seems to be less pronounced for C5L-containing complexes (about 10-fold reduced branching rate) compare to C5-containing ones (about 100-fold reduction). This is in line with the pyrene assays, in which C5L-containing complexes (in contrast to C-5) appear to retain at least some activity. Differences should be quantified relative to the corresponding controls and then statistically tested for using appropriate tests.

This point was also raised by reviewer 2. We have now performed more detailed analysis, measuring the branching rate of C1B/C5 and C1B/C5L complexes in TIRF assays in different concentrations of CK-869 (Supplementary figure 2B). By comparing the half-maximal inhibitory concentration values (IC50) of CK-869 on the two different complexes, we found CK-869 inhibits C1B/C5 slightly better than C1B/C5L (1.8 µM as compared to 3.6 µM) as the reviewer suggested.

Figure 4: Cell metrics such as aspect ratio (A), thickness (A) and speed (C) are expressed as means from five independent experiments. It is not clear how many individual cells were scored per experiment per condition. Similarly, it is unclear at which time (or time window) after inhibitor addition these parameters were scored. Claiming that the authors "...observed that the morphology of macrophages treated with CK-869 changed significantly, with cells rounding up to become less spread..." is a slight over-interpretation, because these metrics have not been quantified in a time-resolved manner but only as a snapshot of the population mean.

We have now provided the number of cells analysed in the individual experiments in figure 4. All measurements were taken after incubating cells for 1 hour with the Arp2/3 inhibitors, which is commonly used for cell-based experiments. We have now also provided a movie (Phase and GFP-LifeAct) covering 5 hours immediately after treating cells with DMSO or 100 µM CK-666 / CK-869 for 1 hour showing that the cell morphology does not change during the imaging period.

Minor: Figure 2/3: In my opinion, separating the in vitro data for ARPC1A/B containing sub-complexes and starting with B does not work particularly well for the flow paper. The results for the C1A containing Arp2/3 complexes (Figure 3) essentially confirm that both inhibitors work at least on some, but not all iso-complexes, as they should. These experiments are -in a way- necessary controls for those shown in the previous figure (Figure 2). I would suggest merging the two figures and starting with the less surprising findings with 1A before showing differential inhibition on 1B.

We have now merge both figure 2 and 3, with some of the original pyrene panels being moved into supplemental figure 2. The new figure 2 also contains quantification of the branching rate in different drug concentrations.

Figure 2/3: Inhibitor concentrations should be stated in the figure and not only in the legend.

This information has been added to the figure.

Figure 4B: The macrophages shown for the CK-869 treatment appear less spread and more round already at t=0 (before inhibitor application), although this is hard to tell for the low contrast PC images. I would recommend showing either images of comparable contrast and cell spread area at t=0 or change to live cell marker and fluorescence imaging.

T=0 is at the point of live cell imaging of cells which have already been treated with the Arp2/3 inhibitors for 1 hour. Consequently, the cells will never appear spread in the CK-869 treated sample. We have provided the fluorescent channel (GFP-LifeAct) in the movie.

Figure 5A: Left and right sub-panels should contain clear labels on top indicating which iso-complexes are being examined (1A left, 1B right). Please also clearly state the total concentrations of actin and Arp2/3 complex used in the figure legend. The low fraction (Thank reviewer for this suggestion and the requested information is provided. The reviewer is totally correct as we found that there was calculation error and the final actin concentration was actually 3 µM and not 7.5 µM as we originally thought.

In Hetrick at al 2013 (PMID: 23623350 DOI: 10.1016/j.chembiol.2013.03.019), the binding of Arp2/3 to F actin reaches a plateau at 15 µM F-actin. We therefore used 15 µM F-actin for the additional pull down experiments requested by reviewer 1 (Figure 4). The new results with 15 µM F-actin agree with our previous observations at 3 µM F-actin concentration.

**Referee Cross-commenting**

The reviews appear to be quite consistent, highlighting several critical issues mentioned by multiple referees. While all referees appreciate the topic/focus of the manuscript, they criticize its preliminary nature.

I anticipate that this would lead to a "major revision" decision at a traditional journal. The numerous constructive comments should enable the authors to significantly enhance the paper if taken seriously.

All three reviewers had similar issues, which we believe we have now fully addressed.

Reviewer #3 (Significance (Required)):

Significance: Small molecule inhibitors such as CK-666 and -869 have been (and still are) widely utilized in the cytoskeleton community as straightforward tools to suppress Arp2/3 activity. However, the results presented here emphasize the need for caution in drawing simplistic conclusions. Hence, future interpretations must adopt a more nuanced perspective. The manuscript therefore makes an important, timely contribution and will be of great interest to a large community.

We thank the reviewer for their positive assessment.

In terms of its potential impact, it reminds me of the recent cautionary tale showing that small molecule formin inhibitors have significant off-target effects (Nishimura et al JCS 2021). However, it is crucial to note that isoform specificity differs from off-target effects, and this doesn't necessarily implicate CK-666 and -869 as inadequate inhibitors.

We agree with the reviewer that these are still useful inhibitors.

While the manuscript is technically sound, with carefully conducted experiments, the presentation and writing seem rushed at times, warranting improvement before publication. The points highlighted above are intended to enhance the overall quality.

The reviewer's points and comments have definitely helped improve the study.

Conceptually, one central weakness is that the reason for the differential inhibition remains ultimately unclear at least in some cases. Specifically, why ARPC1B complexes are refractory to CK-666 inhibition when activated by class I NPFs is not known. Similarly, why activation by different inputs (SPIN90 vs NPFs) is differentially sensitive to different inhibitors remains unclear. Addressing these gaps through additional experiments would strengthen the study. A more insightful discussion, drawing on existing structural and biochemical data, even if speculative, would also be helpful in this regard.

We agree that we still lack a full molecular understanding for the differences but feel that getting to that point will require a substantial amount of work and new Xtal structures that are beyond the scope of the current work. However, we have updated our discussion drawing on existing data as requested by the reviewer.

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:

Cao et al combine in vitro and cellular work to show that neither of the two distinct and frequently used Arp2/3 inhibitors is truly pan-selective, at least when considering distinct classes of activators. Using in vitro assays, they show that CK-666 cannot inhibit ARPC1B iso-complexes when activated by class I nucleation promoting factors. Similarly, Arp2/3 complexes containing Arp3B are refractory to inhibition by CK-869. The latter is likely the result of substitutions at the inhibitor-binding site. They go on to show that these differences correlate with differential effects of CK-666 and -869 on Vaccinia tail formation and macrophage cell shape and motility at the cellular level.

Major comments:

Figure1: The authors state that "...even at 300 μM, the number of virus-induced actin polymerisation events were not diminished (Figure 1B, C)..." The figure shows that CK-666 does indeed not fully abolish cortactin colocalization. However, there seems to still be a significant effect that is not tested for. Statistical tests were only used to compare the two inhibitors at the same concentration. I suggest also testing for significant differences to the DMSO control and reporting p-values, because CK-666 seems to still have an effect. Along the same vein, it seems that valuating the fraction of virus with cortactin co-localization as the only metric for branched actin nucleation downplays the effects of CK-666. Can the authors consider additional other metrics such as the amount of polymerized actin in individual tails or the tail length, which were extensively used in previous publications?

Figure2/3: The authors claim that "...the ArpC5/ArpC5L isoforms are not differentially impacted by either CK-666 or CK-869..." I am not convinced that this conclusion can be drawn based on the data. Figure 2 shows that the inhibitory effect of CK-869 seems to be less pronounced for C5L-containing complexes (about 10-fold reduced branching rate) compare to C5-containing ones (about 100-fold reduction). This is in line with the pyrene assays, in which C5L-containing complexes (in contrast to C-5) appear to retain at least some activity. Differences should be quantified relative to the corresponding controls and then statistically tested for using appropriate tests.

Figure 4: Cell metrics such as aspect ratio (A), thickness (A) and speed (C) are expressed as means from five independent experiments. It is not clear how many individual cells were scored per experiment per condition. Similarly, it is unclear at which time (or time window) after inhibitor addition these parameters were scored. Claiming that the authors "...observed that the morphology of macrophages treated with CK-869 changed significantly, with cells rounding up to become less spread..." is a slight over-interpretation, because these metrics have not been quantified in a time-resolved manner but only as a snapshot of the population mean.

Minor:

Figure 2/3: In my opinion, separating the in vitro data for ARPC1A/B containing sub-complexes and starting with B does not work particularly well for the flow paper. The results for the C1A containing Arp2/3 complexes (Figure 3) essentially confirm that both inhibitors work at least on some, but not all iso-complexes, as they should. These experiments are -in a way- necessary controls for those shown in the previous figure (Figure 2). I would suggest merging the two figures and starting with the less surprising findings with 1A before showing differential inhibition on 1B.

Figure 2/3: Inhibitor concentrations should be stated in the figure and not only in the legend. Figure 4B: The macrophages shown for the CK-869 treatment appear less spread and more round already at t=0 (before inhibitor application), although this is hard to tell for the low contrast PC images. I would recommend showing either images of comparable contrast and cell spread area at t=0 or change to live cell marker and fluorescence imaging.

Figure 5A: Left and right sub-panels should contain clear labels on top indicating which iso-complexes are being examined (1A left, 1B right). Please also clearly state the total concentrations of actin and Arp2/3 complex used in the figure legend. The low fraction (<20%) of Arp2/3 complex co-sedimenting with actin filaments is rather surprising considering the high concentrations used here. 7.5uM actin should be well above the KD for this interaction (compare to data of the Nolen lab such as Hetrick at al 2013). Please comment.

Referee Cross-commenting

The reviews appear to be quite consistent, highlighting several critical issues mentioned by multiple referees. While all referees appreciate the topic/focus of the manuscript, they criticize its preliminary nature.

I anticipate that this would lead to a "major revision" decision at a traditional journal. The numerous constructive comments should enable the authors to significantly enhance the paper if taken seriously.

Significance

Small molecule inhibitors such as CK-666 and -869 have been (and still are) widely utilized in the cytoskeleton community as straightforward tools to suppress Arp2/3 activity. However, the results presented here emphasize the need for caution in drawing simplistic conclusions. Hence, future interpretations must adopt a more nuanced perspective. The manuscript therefore makes an important, timely contribution and will be of great interest to a large community.

In terms of its potential impact, it reminds me of the recent cautionary tale showing that small molecule formin inhibitors have significant off-target effects (Nishimura et al JCS 2021). However, it is crucial to note that isoform specificity differs from off-target effects, and this doesn't necessarily implicate CK-666 and -869 as inadequate inhibitors.

While the manuscript is technically sound, with carefully conducted experiments, the presentation and writing seem rushed at times, warranting improvement before publication. The points highlighted above are intended to enhance the overall quality.

Conceptually, one central weakness is that the reason for the differential inhibition remains ultimately unclear at least in some cases. Specifically, why ARPC1B complexes are refractory to CK-666 inhibition when activated by class I NPFs is not known. Similarly, why activation by different inputs (SPIN90 vs NPFs) is differentially sensitive to different inhibitors remains unclear. Addressing these gaps through additional experiments would strengthen the study. A more insightful discussion, drawing on existing structural and biochemical data, even if speculative, would also be helpful in this regard.

Own expertise: cytoskeleton, actin, biochemistry, in vitro reconstitution, fluorescence microscopy, structural biology

Signed: Peter Bieling, MPI Dortmund

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 'CK-666 and CK-869 differentially inhibit Arp2/3 iso-complexes' addresses how commonly used Arp2/3 complex inhibitors differentially inhibit Arp2/3 complex activity based on the subunit isoforms making up the Arp2/3 complex. This work directly tests how each inhibitor affects different iso-complexes, which may affect different cell types based on the predominant iso-complex present in the cell. The manuscript is well written, with experiments both in cell culture and with purified proteins in reconstitution and biochemical assays to establish that these small molecule inhibitors have different effects based on the iso-complex of Arp2/3 present. There are several points in the manuscript that if addressed would improve and support the conclusions presented.

In Figure 1B, looking at the images of the CK-666 treated verses the DMSO, it looks like the actin structures in the DMSO-treated cells are potentially larger than those in the CK666 cells, but because only an inset of drug-treated is shown, and an inset of the DMSO-treated is not shown it is hard to compare. Are the size of the virus-associated structures affected in the CK-666 treated cells versus the DMSO-treated cells? This might indicate that CK-666 has some effect on actin polymerization, even if it is not as drastic as the CK-869.

In Figure 2 comparing the pyrene curves in figure 1A, it appears that CK-869 has a different effect on C1B/C5+VCA versus C1B/C5L+VCA (green curves as compared to no activation control, grey curves), but this is not commented on. Addressing the differing effects would strengthen the authors conclusions- namely, that CK-869 inhibits both iso-complexes better than CK-666, but there may be some differences on each isoform. It is unclear if the differences in the branching rate (Figure 2B) is also reflective of this. The authors should address these results.

For Figure 4, it is somewhat unexpected that inhibition of the Arp2/3 complex increases macrophage motility as compared to control, unless the reader is familiar with the 2017 Rotty et al paper. The manuscript may benefit from a sentence or two explaining this result in light of the findings of the 2017 Rotty paper beyond simply mentioning that the increase in motility is dependent on myosin II.

The Spin90 data looks good, clear, and consistent.

In Figure 7, given that pyrene was used in all the previous assessments of drug treatment on arp2/3 isoforms, it seems appropriate for these assays to be performed for Arp3B/C1B/C5L in comparison with Arp3/C1B/C5L and between the different drug treatments. Likewise, this should be done for the Spin90 also. It is difficult to compare between the figures for Arp3b vs. Arp3C (Figures 2 and 3 vs. Figure 7), although this may require a repetition of data presented.

Minor issues:

It would be helpful if the labels for what is labeled in the micrograph were on the images (Figure 1B, Figure 3B, Figure 7A). In Figure 1-B, the 200uM CK-869 cell image looks less representative of the data in Figure 1C than other cells in the figure. Perhaps there is higher background in this micrograph, but it might be clearer if a cell with similar background actin signal to the other CK-869 was used.

Figure 4: Where is the mean thickness of the cell measured? In figure 4D, it would be helpful if the error bars could be in the color of the line, as it is hard to distinguish the range of the data for each condition because the error bars are overlapping and the same color for all.

Figure 5: In Figure 5A, the labeling of the gels (KDa, S and P) do not line up correctly. The legend for the quantification should indicate what bands were quantified- all the arp2/3 bands or just the isoforms? It is unclear what is being quantified in the graph in C. The pull-down results in C should be quantified via quantitative western blot if possible.

The statement in line 170- 'indicate' seems a bit strong based on the results presented. 'Suggests' might work better here.

Significance

It is of general interest to members of the actin field as well as cell-biologists who routinely use either CK-666 or CK-869 to inhibit Arp2/3 complex activity in cells, and specifically in mammalian cells.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary:

This work focuses on two small molecule inhibitors of the Arp2/3 complex, CK-666 and CK-869. Previous studies have shown that although the Arp2/3 complex is well conserved in eukaryotes, the inhibitory effect of these molecules is highly species dependent. However, it has been unclear whether these drugs act equally well on Arp2/3 iso-complexes (complexes composed of subunit isoforms from the same species). This paper fills that gap. Using human Arp2/3 iso-complexes, it shows that the inhibitory effect of these two drugs depends on the subunit composition of the complex. In addition, this work shows that these drugs do not systematically and equally inhibit the ability of these Arp2/3 complexes to nucleate linear or branched filaments.

Major comments:

1. Regarding the first part on vaccinia-induced actin polymerization

The first paragraph of the Results section is difficult to follow for those who have not read the previous papers from this lab. I would recommend changing the text so that any reader can understand from the start the experimental system and the goal of the experiment.

The data analysis of Figure 1C is not satisfactory. It is not very informative to statistically compare the effect of the two drugs at similar concentration. However, it is necessary to perform statistical tests to compare the different conditions with drug with the control condition (DMSO). By eye, I see a difference between DMSO and CK-666, so it is difficult to understand why the authors claim that CK-666 has no effect on actin polymerization.

Images with CK-869 have a lower overall cortactin signal, which could indicate that immunolabeling was not very effective in this condition. This could affect the analysis of the data in Figure 1C.

The authors mention that the exact levels of the 8 different Arp2/3 iso-complexes are not known in these HeLa cells, but it should be fairly easy (e.g. mass spectrometry) to quantify the expression level of ArpC1, ArpC5 and Arp3 in these cells and verify that it is consistent with the rest of the story.

This information about the expression level of ArpC1, ArpC5 and Arp3 in HeLa cells is also very important because a large community of researchers use CK-666 and HeLa cells. There are actually quite few papers that draw conclusions from the use of CK-666 in HeLa cells, and the authors should discuss the limitations of these studies much more clearly. 2. The pyrene assays are disappointing because they are performed with only one concentration of CK-666 and CK-869. This is especially true for the VCA data, where the effect of the drugs is not always "on"/"off" as naively presented in the text, but highly concentration dependent. The authors should definitely provide several drug concentrations for each condition, up to saturation levels, to provide a clear quantification of the drug concentrations needed to reach half inhibition. 3. Similarly, the pull-down experiments performed at a single protein concentration are inconclusive. They cannot tell us whether the affinity of the Arp2/3 isoforms for these targets is altered in the presence of the small molecule inhibitors because we do not know the degree of saturation of the ligands. Given that some of the reported differences in inhibition of filament nucleation are modest, it is not possible at this stage to link these different data.

Significance

The subunit composition of the Arp2/3 complex is cell-type dependent, so these data will be important for the many cell biologists using these molecules. In particular, it calls for caution in the use of these drugs and in the interpretation of the data.

The writing is very clear, but the manuscript seems quite rushed. Many experiments need to be analyzed in much more detail to clarify the conclusions.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer_01

Major comments:

1. The authors cite that acetylated and tyrosinated microtubules have different spatial and compartmental distribution in dendrites and axons and investigate the distribution in the AIS of nonAcD cells and AcD cells, as well as the stem dendrites. However, they just show one example of two different cells (Figure 2D and E) without any statistical analysis. Either, they should remove this part or provide a thorough quantification. Reply: The spatial and compartmentalized distribution of stable and dynamic MTs in the dendrites and axons of nonAcD neurons has been extensively studied and reviewed (see Kapitein & Hoogenraad, 2011; Katrukha et al., 2021; Tas et al., 2017 for reference). However, the organization of the MT cytoskeleton in AcD neurons is still unknown. Here, we provide the very first evidence on the distribution of tyrosinated and acetylated MTs in AcD neurons, as well as data on MT orientations. We agree with the reviewer that to make our results on the spatial organization of these post-translational modifications in AcD neurons more complete, we need to provide a more thorough quantification analysis.

To achieve this, we plan to perform immunostainings on DIV10 neurons using antibodies against tyrosinated (tyr) and acetylated (ac-) tubulin to label dynamic and stable MTs, respectively. Subsequently, we will conduct high-resolution 3D confocal imaging and measure fluorescent intensity to illustrate the abundance and staining patterns of tyr- and ac- MTs in the axons and dendrites of AcD neurons. Since the spatial distribution of tyr- and ac-MTs is distinguishable with confocal microscopy, we will retain STED examples in the figures but conduct new analyses on confocal imaging data. We will measure the total fluorescent intensity of tyr- and ac- MTs in different compartments of AcD neurons and normalize it to the size of the measured area. We will then compare the normalized intensity values between the axons and dendrites of AcD neurons to examine whether there is a specific distribution pattern of stable and dynamic MTs. We will analyse at least 3 independent primary culture preparations with a minimum of 30 cells. Using the same dataset, we will also quantify the percentage of AcD neurons with ac-MTs specifically elongating into the axon compared to AcD.

The authors use EGFP-Rab3A vesicle to investigate anterograde transport at the axon and dendrites. They find a slightly faster transport of these vesicles at the AIS of AcD cells and conclude the axonal cargos in general are transported faster across the AIS in AcD cells. In my opinion, this generalization based on one type of vesicle is too farfetched.

Reply: The Rab3A protein is associated with pre-synaptic vesicles that are transported by KIF1A and KIF1Bβ, members of the kinesin-3 family, towards pre-synaptic buttons (see Guedes-Dias & Holzbaur, 2019; Niwa et al., 2008 for reference). Since KIF1A and KIF1Bβ are common motor proteins that mediate MT-based transport of different types of vesicles (e.g., synaptic vesicles and dense-core vesicles, see Carabalona et al., 2016; Helmer & Vallee, 2023 for reference), we reasoned that Rab3A should be a representative marker for an axonal cargo. However, this indeed does not rule out whether the faster trafficking effect we saw is specific to presynaptic vesicles, as different types of vesicles tend to recruit different modulators that could lead to different trafficking features.

To address this question, we will perform a live-imaging experiment including two additional organelle marker proteins, Neuropeptide Y (NPY) and Lysosome-associated membrane protein 1 (Lamp1). NPY is transported into the axon via KIF1A and KIF1Bβ-mediated dense-core vesicles (see Helmer & Vallee, 2023; Lipka et al., 2016 for reference). Lamp1 is associated with lysosomes and a range of endocytic organelles that recruit both kinesin-1 and kinesin-3, and are transported into both axons and dendrites (as reviewed in Cabukusta & Neefjes, 2018). By introducing two additional types of vesicles, we should be able to answer whether AcD neurons, in general, tend to transport cargoes into the axon faster than nonAcD neurons.

__Minor comments: __

In the introduction, the authors describe how synaptic inputs are received at the dendrites and propagated to the soma in the form of membrane depolarizations. They should add 'excitatory' to synaptic inputs or also describe the impact of inhibitory synaptic inputs at the dendrites.

In my opinion, Figure 2 could be presented in a slightly better way. The lower part of panel A better fits to panel B, which is next to the upper part of panel A. I understand that the authors systematically present their data first for nonAcD cells and then for AcD cells. However, in this special case it is a little bit more difficult to read the current figure in that order. The results displayed in Figure 4 are presented in a slightly confusing order. The authors jump from 4D to 4G, then to 4I and 4E, 4H, 4F. Similarly, 4M and N are addressed before 4O and P to finally get to 4K and L. It would be beneficial to present and address the data in a stringent way.

Reply: Thank you for the suggestions on how to improve the data representation in the figures. We will change Figures 2 and 4 and make adjustments in the text upon revision since we also plan to include additional data.

Reviewer_02

Major comments:

1. The authors suggest that there is reduced Na+ channel density at AcD AIS compared to other AIS arising from the cell body. This is not convincing. Immunostaining for Na+ channels is notoriously difficult and sensitive to fixation since the epitopes of the anti-Pan Nav antibodies are highly sensitive to fixation. In addition, this is based on immunofluorescence intensity quantification. Since the mechanism of localization is through binding to AnkG, the authors should also measure other AIS proteins like AnkG, b4 spectrin, and Nfasc. Do these change? If all uniformly change I would be much more inclined to accept the conclusion. If they do not change, it still doesn't rule out the concern about fixation conditions and slight differences in the cultures. The authors indicate there is about a 40% reduction in fluorescence intensity. That is quite large. This big difference should also be confirmed in brain sections. Reply: The potential fixation issue and antibody sensitivity on Na+ channel staining are indeed valid considerations, and we are aware of them. However, it should be noted that we used pan-Na+ channel antibodies that were previously characterised and widely used in literature (see Solé et al., 2019; Yang et al., 2020 for references). Furthermore, our samples underwent the same fixation and staining protocol, and comparable numbers of AcD and nonAcD neurons were imaged from the same preparation and coverslip for each experiment. Imaging settings were also kept constant. Any loss of Na+ channel staining at the AIS due to fixation should affect both neuron types and therefore our conclusion is justified. Nevertheless, the reviewer's point regarding other AIS components is valid and will be investigated further in the revised manuscript.

Following the reviewer's suggestion to further strengthen our conclusion, we will measure the intensity of AnkG, βIV-spectrin, and neurofascin in DIV21 AcD and nonAcD neurons. We will compare a minimum of 3 independent cultures, each containing at least 10 cells of each type per culture.

We agree with the reviewer that confirming observed differences in Na+ channel staining using brain slices would be beneficial. However, conducting such experiments presents several challenges. Firstly, one approach could involve immunostaining with antibodies against AIS marker AnkG, in combination with somatodendritic marker MAP2 and pan-Nav. However, this method lacks the advantage of clearly identifying neuronal morphology as seen in dissociated cultures, making the outcome unclear and difficult for analysis and interpretation. Alternatively, the use of Thy1-GFP rats, where a subset of neurons is labelled with GFP, could allow for morphological studies. Unfortunately, we do not have access to this rat line, and the process of importing it, obtaining permits, and establishing a colony is beyond the timeframe for manuscript revision. Additionally, while pan-Nav antibodies have shown reliability in dissociated cultures, their efficacy in tissue staining is less certain. We could provide example images upon request. Secondly, endogenously labelling of Na+ channels is another option, but remains a significant challenge. Recent developments in endogenous labelling, such as the CRISPR/Cas9-based method using pORANGE by Fréal et al. (Fréal et al., 2023), and the generation of Scn1a-GFP transgenic mice by Yamagata et al. (Yamagata et al., 2023), offer potential solutions. However, the labelling efficiency of pORANGE is uncertain, and both methods are time-consuming and cannot be completed within the three-month revision period.

As an alternative, we propose emphasising that our results are based on in vitro experiments and discussing the advantages and limitations of this approach in the discussion section.

The analysis of inhibitory synapse differences at the AIS are also not compelling - this is a limitation of the culture system. The authors have no control over the density of inhibitory neurons in the culture well. This interaction is not intrinsic to the AcD neuron, but rather a feature of neuron-neuron interactions which should only be modelled in the animal.

Reply: The reviewer is correct in pointing out that establishing inhibitory synapses at the AIS is not an intrinsic feature of AcD neurons; it depends on the network and should be modelled in animals. We will include this limitation of the cell culture model in the discussion section in the revised manuscript. We also understand the reviewer's concern that the lower amount of inhibitory synapses at AcD neuron AIS might be due to uneven density of inhibitory neurons between cultures. Nonetheless, assuming that the number of inhibitory neurons is constant between preparations, it is an interesting observation that AcD neurons form fewer inhibitory synapses at the AIS. This may be related to the features of the AIS and its morphology and should be further investigated.

To make our study more comprehensive and also address the reviewer's concern regarding the presence of inhibitory neurons, we will perform immunostainings in dissociated cultures (40.000 cells per 18 mm coverslip, same as in experiments with synapse quantification) with antibodies against pCaMKIIa, an excitatory neuron marker, and GAD1, a marker for inhibitory neurons. Then, we will quantify the density of inhibitory neurons in the culture. We will perform measurements from 3-6 independent cultures by analysing large fields of view in different areas of a coverslip (20-30 neurons per area) to determine if the density of inhibitory neurons varies between cultures as well as preparations. Furthermore, as also requested by reviewer 4, we will perform new immunostainings where pre- and post-synaptic markers (VGAT and Gephyrin) will be included in the same sample together with the AIS (AnkG or Neurofascin) and dendritic marker (MAP2). Synapses that contain pre- and post-synaptic components will be analysed and included in the revised version of the manuscript.

Finally, the major limitation of this study is that it is performed in vitro. Surprisingly, the authors actually argue this is a feature of their system. While it is true some of the questions can be addressed perfectly well in vitro, many cannot. In the first paragraph of the results the authors state an advantage of their system is that there are no microenvironments to influence the development of the AcDs. I'm afraid I view this as a drawback. The authors suggest this is an opportunity to examine intrinsic mechanisms of development - true, but it also foregoes the opportunity to determine if the outcomes are different from what occurs in vivo. To this point, the authors report that only 15-20% of the population of hippocampal neurons in culture are AcD neurons. But in their introduction they cite other literature indicating 50% of hippocampal neurons in vivo are AcD neurons - this suggests that the environment of the hippocampus in vivo influences whether a neuron becomes an AcD neuron or not.

Reply: The reviewer is right in pointing that the in vivo environment could indeed affect AcD neuron development, and we also find this to be a very interesting topic to investigate in the future. Even more intriguingly, as shown in a preprint by Lehmann et al. (doi: https://doi.org/10.1101/2023.07.31.551236), network activity stimulates neurons to acquire AcD morphology. While it is true that the impact of the microenvironment on AcD neuron development cannot be studied in dissociated cultures, our in vitro data undoubtedly support the fact that hippocampal neurons can intrinsically develop into AcD morphology independent of the in vivo environment. As also mentioned in the next point, our statement "...their development must be driven by genetically encoded factors rather than specific..." might sound too definitive and therefore eliminate possible effects from the microenvironment. We will revise this part. Although it is highly desirable to move cell biological studies from neuronal cell cultures to tissue, to date, it is still very challenging to perform many of experiments which we did in this study in slices or living animals due to a lack of appropriate technologies and tools. We are convinced that many basic biological questions can be and should be studied in simplified culturing models because they are truly fundamental, they should also be reproducible in these models.

To address the reviewer's question regarding the percentage difference between our data and the previous study by Thome et al. (2014), several factors should be considered. First, as noted by the reviewer, our results were obtained from an in vitro system, which is not directly comparable to the in vivo model system used in Thome et al.'s study (Thome et al., 2014). Second, the age of the neurons quantified in our developmental experiments is DIV5 and DIV7. This young age disparity could contribute to the percentage difference, as Thome et al. analyzed neurons from P28-35 adult animals, where 50% of the AcD neuron population was observed, specifically in the CA1 region. Third, it's important to note that in other hippocampal regions, the percentage of AcD neurons is lower (approximately 20-30%). Since our hippocampal primary cultures contain neurons from all hippocampal regions, this may have averaged out our quantification of AcD neuron percentage. Additionally, in the study by Benavides-Piccione et al. (Benavides-Piccione et al., 2020), they reported 20% AcD neurons in the CA1 region of hippocampi isolated from 8-week-old mouse pups, a number similar to what we observed in vitro. Interestingly, Thome et al. reported that in P8 pups, AcD neuron population in hippocampal CA1 region is 30%. This number increased to 50% in adult animals at age of P28-35, suggesting there is perhaps an age dependent increase of AcD neuron population. This could be an additional reason of why we only saw 15-20% of AcD neurons in our in vitro system, regardless of the in vivo environment.

In the revised version, we will clarify these points in the introduction and discussion sections. Additionally, we will quantify the proportion of AcD neurons in mature DIV21 dissociated hippocampal cultures and compare it to DIV7 cultures to assess whether there is an increase in the AcD population over time. We believe that this experiment, combined with the explanations provided above, will sufficiently address the reviewer's question. However, it is important to acknowledge that the establishment of neuronal networks in vitro differ from those in vivo. Therefore, there may be potential differences in the outcomes.

I appreciated the balanced discussion of whether this is a stochastic or genetically programmed process. This could have been emphasized earlier in the results since the authors invoke the concept that "...their development must be driven by genetically encoded factors rather than specific...". The authors have not shown this and cannot show it in this system. Indeed, as stated in point 4 above, I think their data argue against a simple genetic program.

Reply: As suggested by the reviewer and noted in point 4, we will revise the section on AcD neuron development in our manuscript to emphasize that hippocampal neurons may adopt AcD morphology through genetic or stochastic mechanisms. While we acknowledge that environmental and activity factors may also influence this process, particularly in mature neurons, our study focuses on developing neurons where genetic and stochastic factors are likely to be predominant. This conclusion is supported by the observation that neurons develop into AcD morphology in vitro, where environmental and activity patterns do not mimic those of in vivo systems.

Indeed, our current manuscript does not explore genetic factors involved in AcD neuron development. To address this question, one approach could be to label AIS markers endogenously in dissociated cultures using the PORANGE method (see Willems et al., 2020 for reference) or utilize AnkG-GFP transgenic mice (Fréal et al., 2023; Thome et al., 2023) along with a volume marker like mRuby or GFP. This would allow for the identification of AcD and nonAcD neurons in vivo and in vitro, followed by single-cell transcriptomics analysis to uncover potential genetic factors. Subsequently, candidate genes could be manipulated to demonstrate their essential role in AcD neuron development. However, such experiments require significant time and resources beyond the scope of our current revision timeframe. Nonetheless, this question presents an exciting direction for future research.

Reviewer 3

Major comments:

1. The authors classify neurons into axon-carrying dendrite (AcD) and non-AcD neurons by measuring the stem dendrite length (> 3 µm). I could not find the validity for this cut-off. The non-AcD neurons in Fig. 6B appear more AcD to this reviewer, and, in addition, other researchers have proposed a third category of 'shared root' neurons (doi: 10.7554/eLife.76101). For purposes of reproducibility and transparency, please provide first a comprehensive overview of the entire population of morphologies (i.e. all cells in control conditions). The distances from the soma could be plotted in histogram (etc.) and authors may want to think about independent supporting evidence for the cut-off to classify AcD and non-AcD neurons. Reply: Concerning the validity of AcD neuron classification, we did measure the length of the stem dendrite, as shown in Figure S4G, with an average distance of around 10 µm. However, we admit that this information is presented relatively late in the manuscript. To address the reviewer's criticism, in the revised version, we will include a supplementary figure displaying a gallery of representative images of both AcD and nonAcD neurons analyzed in our study (please refer to Hodapp et al., 2022; Fig S1 C&D; Fig S3 as an example). Given the sample size of AcD and nonAcD neurons in our study, including all images would result in a very large figure (for example, Figure 1: DIV5: 83 AcD neurons out of 427 cells, DIV7: 47 AcD neurons out of 387 cells). We will only show representative examples of AcD neurons in the gallery. Additionally, as suggested, we will plot the length of the stem dendrite (or axon distance) of AcD neurons as a histogram to demonstrate that the AcD neurons included in our study indeed have a stem dendrite longer than 3 µm. To further validate the used classification method, we will measure the diameter of the stem dendrite in all analyzed AcD neurons and then compare the distance between the soma and the start of the axon in each analyzed AcD neuron to the diameter of its stem dendrite. As described by Hodapp et al. (Hodapp et al., 2022; Fig S1A), AcD neurons are expected to have a stem dendrite longer than their diameter.

We have considered having independent evidence to support the classification of nonAcD and AcD neurons. However, the method used by Thome et al. and Wahle et al. for AcD and nonAcD neuron classification is well established and widely accepted (see Thome et al., 2014; Wahle et al., 2022 for references). Similar standards were also employed by Benavides-Piccione et al. (Benavides-Piccione et al., 2020). Introducing independent evidence could potentially raise further doubts, so we have chosen to maintain consistency with previous studies.

As for the "shared root" neurons described by Wahle et al., we did not analyze this category separately and included them in the nonAcD subtype. Nonetheless, it is an interesting direction to explore in the future. For completeness, we will discuss this point in the revised manuscript.

Related to point #1 the primary hippocampal neuron system is excellent for cell biological questions but comes with the drawback of imaginative morphologies including neurons with multiple axons and AISs. It is not mentioned here but literature indicates up to 20% of neurons have two axons (e.g. doi: 10.1007/s12264-017-0169-3, 10.1083/jcb.200707042). How did the authors classify the double axon cells? Since the main hypothesis is the existence of an intrinsic program for AcD neurons (p. 5 top), the two axons from one neuron should develop similarly. The authors can easily test this with the data.

Reply: We appreciate the reviewer's comment regarding the choice of the model system for this type of study. Indeed, as they pointed out, in primary cultures, some neurons develop more than one axon. Since we did not find any supporting evidence from the literature reporting that hippocampal neurons have multiple axons in vivo, we only analyzed neurons with one axon for both AcD and nonAcD neurons. We will clarify this in our method section of the revised manuscript.

Some interpretations about function are not correct and the authors should reconsider these. A role of cisternal organelles on neuronal excitability remains to be demonstrated (and see doi.org/10.1002/cne.21445 showing there is none). In addition, the statement that lower fluorescence intensity of Pan-Nav1 is indicating reduced excitability is flawed. Antibody staining does not scale linearly with voltage-gated sodium channel density and since the AIS of AcD neurons is further from the soma it is most likely smaller in diameter which may account for apparent fluorescent differences. For biophysical reasons (for details I refer to 10.3389/fncel.2019.00570, 10.1016/j.conb.2018.02.016 and 10.7554/eLife.53432) smaller diameter axons will be easier to depolarize by depolarizing voltage-gated channels or excitatory synapses. Finally, in AcD neurons the AIS distance from the soma poses all sorts of interesting cable properties with the soma and the local dendritic membrane and the electrotonic properties alone suffice to make these neurons more excitable.

Reply: The reviewer brings up very valid and important points that we will address in the revised manuscript. First, we will rephrase and adjust our interpretations regarding the functions of the cisternal organelle in the AIS. As also mentioned by reviewer #2, we are aware that antibody staining does not properly reflect Na+ channel density. As discussed above, we will also measure other AIS proteins that anchor Na+ channels to see if there are any correlations in fluorescence intensity between them and Nav1. We agree with the reviewer that AcD neuron's AIS could have a smaller diameter, resulting in fewer Na+ channels. Indirect evidence is already available in the study of Benavides-Piccione et al., showing a smaller axon diameter in AcD neurons compared to nonAcD neurons in both human and mouse brain sections (Figure S4). To test this in our model system, we propose to measure the AIS diameter in AcD neurons. If this is indeed the case, we will indicate it in our revised manuscript and edit the section on Na+ channels.

Exploring the biophysical properties of the AIS and axons of AcD neurons is indeed a highly interesting direction to pursue and is the project in its own. It would necessitate the use of computational modeling approaches, which require considerable time and resources that are not feasible within the timeframe of this revision.

Comparing AcD and non-AcD neurons for AIS plasticity is an excellent idea but the present statistical design is not suitable for answering this question. The authors should directly compare non-AcD and AcD neurons within a two-way ANOVA design, asking the question whether the independent variable axon type is significantly different and interacts with plasticity.

Related points: 'AIS distance' in Figure 7 seems to refer to something else than distance from soma (Figure 1). Please clarify. What were the absolute distances from the soma for the AcD neurons and was this dependent on treatment?

Reply: We appreciate reviewer's comment and in the revised version we will perform the analysis using two-way ANOVA.

Regarding the terminology and definitions used in our manuscript, the "AIS distance" refers to the measurement between the start of the AIS and the axon initiating point, as depicted in Figure S4 of the manuscript. We adopted this parameter from the previous study by Grubb et al. (Grubb & Burrone, 2010), ensuring consistency in our investigation of AIS plasticity. For AcD neurons, where the axon branches out from the dendrite, we defined the AIS distance as the length between the start of the AIS and the border of the stem dendrite, as illustrated in Figure S4B.

In Figure 1, the term "distance from soma" represents the length of stem dendrite and used for AcD and nonAcD neuron classification. As shown in Figure S4G, the absolute distance from the soma for AcD neurons is approximately 10 µm and remains consistent across treatments. We will explain these points more clearly in the revised manuscript.

Minor comments:

1. At p. 7 is stated that "The percentage of none-AcD forming collaterals at DIV1 is much lower than for AcD neurons" but statistical support is lacking. The conclusion in the next line is that "AcD neurons follow consensus development". That is puzzling given the difference just mentioned before. Please clarify. Reply: We will provide statistical support for comparing collateral formation between nonAcD and AcD neurons at DIV1.

Regarding the second point concerning consensus development, we were referring to the general developmental sequence of AcD neurons, as described by Dotti et al. (see Dotti et al., 1988 for reference), where neurons typically first establish an axon and then dendrites. This sequence is not necessary related to collateral formation, which indeed differs between nonAcD and AcD neurons. The ability to form collaterals may come from local differences in microtubule (MT) and actin dynamics at AcD neuron precursor axons, but it does not alter the fact that AcD neurons initially establish an axon and subsequently dendrites. We will clarify it in the revised manuscript.

A study not cited in this manuscript showed distinct dendritic morphologies (doi: 10.1073/pnas.1607548113) and AcD interneurons are different for their axonal arborization (doi: 10.1242/dev.202305). Differences in growth of branch arborization could hint to subtypes. Are the AcD and non-AcD neurons different in their adult morphology? A detailed account of the axonal and dendritic trees would strengthen the data.

Reply: Thank you for pointing this out. We will include this citation. In the study by Hodapp et al., it was shown that AcD and nonAcD neurons exhibit similar dendritic morphology and do not differ in spine density, number of dendritic branches, and total dendritic length. However, in hippocampal AcD neurons, the AcD occupies 35% of the total basal dendrite length, which is larger than basal dendrites in nonAcD neurons, suggesting that AcD neurons do possess specific features in their dendritic trees.

Regarding the axons of AcD neurons, there is currently no detailed study available, and it would be more appropriate to investigate neuronal connectivity through tracing studies in animals rather than in primary cultures. Therefore, this question falls outside the scope of the current manuscript.

Some key references are not included here, and a number of these are mentioned above. In the context of the detailed MT and Rab3A vesicle and cargo transport studies, please acknowledge some of the pioneering work of Alan Peters revealing the ultrastructure of axons emerging from dendrites. See Figs. 5-7 in Peters, Proskauer and Kaiserman-Abramof IR., J Cell Biol 39:604 (1968). What is the identity of the neurons? It makes a difference if the cells are interneurons or pyramidal neurons, CA1 or CA3-like. For plasticity experiments the authors uses cells as independent measurements, but this is inflating the power. How many cultures were used?

Reply: Thank you for pointing this out; we will include the suggested references in the revised manuscript. In our study, we focused on excitatory neurons from the hippocampus. We distinguished neuron types morphologically or with the inhibitory neuron marker GAD1. Identifying CA1, CA2, CA3, and DG subtypes in dissociated culture is more challenging, and this would be an interesting avenue to explore in an in vivo system. Here, we focused on fundamental cell biology aspects related to the AIS structure and its trafficking barrier function, which should be similar in all these neuron types. While there may be subtype-specific differences in AIS plasticity, investigating this is beyond the scope of our manuscript.

For the plasticity experiments, we used a total of 3 independent cultures, from which we collected a comparable number of neurons. In response to the reviewer's concern, we will also plot the mean of each culture to illustrate the variability of our data points.

Reviewer 4

Major comments:

1. A general limitation of this study is the low N for some critical experiments. In several experiments, individual cells become an N, therefore boosting the power of the analysis when in reality, due to the known heterogeneity of AIS length, position, and general cell morphology in vitro, the aim should be to compare means across animals / preparations, each consisting of a comparable number of individual cells. This is especially important for the analyses of COs, axo-axonic synapses and channel expression at the AIS. Reply: We would like to mention that this is a cell biological study where neurons are grown in dissociated cultures. To prepare one such culture, we typically use hippocampi from 6-8 E18 rat embryos, which are then mixed in one suspension before plating. The cells are then plated on coverslips in a 12-well plate format. When referring to replicates, for all experiments except for the longitudinal study of 5-day-long time-lapse imaging of developmental sequences (Figure 1), we used between 3 to 6 independent preparations. From each preparation, we took a comparable number of cells derived from 4-6 different coverslips. For each experiment, we measured more than a hundred cells, which is standard practice in the field. To address the issue with individual measurements, in the revised manuscript, we will additionally plot the means of each independent preparation.

Such critical parameters as e.g. synaptic innervation at the AIS are investigated in a way that does not support the clear statements given, e.g. "The AIS of AcD neurons receives fewer inhibitory inputs" (Highlights statement) or "AcD neurons have less inhibitory synapses at the AIS" (header of Fig. 6). The overall number of analyzed cells is low (3 and 4 preparations, respectively and approximately 50-cells for each marker). The combination of a pre- and postsynaptic marker for inhibitory / excitatory neurons is a solid decision, but the analysis is not done based on the close approximation of these markers, in 3D, along an AIS, but rather in maxIPs and without any regard of whether pre-and postsynaptic markers are actually close to each other not. The expression of these markers alone just points towards the epitopes being expressed, but are they localized to each other in such a manner that they could form bona fide synapses? The methods are not totally clear on the image depth (tile scans with 5 µm in z will not provide the detail of information to resolve synapses, so how did the authors address the subcellular analysis here and for the CO and VGSCs?). And generally, were Nyquist conditions taken into consideration throughout the study? This can be clarified in text and does not require additional experiments.

Reply: The overall number of cells for quantifying inhibitory synapses along the AIS was approximately 80 cells for each synaptic marker. To clarify this, we will indicate the number of cells in the figure legend of our revised manuscript and will additionally plot mean values across independent preparations.

In the current manuscript, our main goal was to provide an initial quantitative measurement of AIS features in AcD neurons to see if they differ from nonAcD neurons. Hence, maxIPs are sufficient for this purpose as they summarize the 3D information. To make our study more comprehensive, following the reviewer's suggestion, we will conduct additional experiments to co-label pre- and post-inhibitory synapses at the AIS with VGAT and gephyrin, respectively. Then, we will image samples in 3D to measure the density as well as the distance between pre- and post-synapses at the AIS of AcD neurons and compare them to nonAcD neurons.

The Nyquist condition was taken into consideration throughout the study. The pixel size of our data collection was 0.081 µm for the laser scanning microscope, as indicated in our methods section. Given the optical setup of our microscope and the fluorophores used to label target proteins (information available in the methods section of our manuscript), the acceptable Nyquist lateral sampling size (or pixel size, in other words) for confocal images is between 0.083 to 0.093 µm and 0.2 µm in the z-plane. In our data collection for laser scanning confocal images, the z-step size was 0.5 µm (see methods section of our manuscript), which is indeed undersampling the data. However, this should not significantly affect our analysis based on maxIPs. The new stainings with matched pre- and post-synaptic markers will be imaged with a smaller z-step (0.2 µm) and then reconstructed in 3D.

The chapter on AIS plasticity is certainly an interesting addition to the study, but is a bit superficial, yet reaches strong conclusions ("More importantly, it further indicates that the AIS of AcD neurons is insensitive to activity changes"). This is based on un-physiological concentrations of KCl, and certainly not on network manipulation that truly tests synaptic activity. It also comes back to the 1st point above. A suggestion would be to edit the conclusion.

Reply: KCl treatment globally depolarizes the membrane potential of neurons, leading to an increase in intracellular calcium via voltage-sensitive calcium channels as well as NMDA and AMPA receptors (Rienecker et al., 2020). This protocol has been used in several initial studies describing the plasticity of the AIS (see Evans et al., 2013, 2017; Grubb & Burrone, 2010; Jamann et al., 2021; Muir & Kittler, 2014; Wefelmeyer et al., 2015 for references). Moreover, as shown by Evans et al. and Grubb et al. (see Evans et al., 2013; Grubb & Burrone, 2010 for references), AIS plasticity is not abolished by TTX, which blocks Na+ channels, but is prevented by L-type calcium channel blockers. This suggests that the occurrence of AIS plasticity is independent of action potentials but more sensitive to calcium-related pathways downstream of membrane potential depolarization and post-synaptic activation. Hence, we believe our results are indicative of how the AIS would react when calcium signaling pathways are altered by activity levels. To address the reviewer's concern, we will focus our conclusion more on membrane potential depolarization and calcium signalling and edit out statements.

As discussed above in response to reviewer #3, the quantification of AIS plasticity includes 3 independent preparations, comprising approximately 200 neurons in total. To prevent inflation of statistical power in the analysis, we will also plot the means and standard error of the mean (SEM) for each independent experiment and assess whether any differences persist.

The rationale behind looking at the cisternal organelle (CO) in this study is outlined in the Introduction, where the authors state that "...... and is responsible for calcium handling". What is "calcium-handling" and where is the evidence cited? Furthermore, in the Results, they state that "...both compounds (VGSCs and COs) are critical for the AIS to regulate neuronal excitability". While this is the case for VGSCs, there is no conclusive evidence in the literature whether of not the CO is "critical" for neuronal excitability. In fact, a number of neurons have no CO in the AIS (as much as 50% of all AIS in mouse primary visual cortex for example do not express synpo at the AIS at all, Schlüter et al., 2017). The CO can therefore not be as critical for AP initiation as the authors state. Furthermore, the authors state that "AIS plasticity in excitatory neurons is triggered by calcium signaling". While certainly shown and adequately cited here, other factors (independent of calcium) can also play a role, therefore this statement is a bit absolute and should be edited accordingly.

Reply: Thank you for constructive editorial suggestions. Regarding the first question on calcium handling, we were referring to Ca2+ storage and release mechanisms. Benedeczky et al. already showed the existence of SERCA-type Ca2+ pumps at the membrane of the cisternal organelle (CO) to demonstrate the involvement of Ca2+ sequestering/storage by the CO at the AIS (Benedeczky et al., 1994). Although indirect, Sánchez-Ponce et al. showed the presence of IP3R, which promotes Ca2+ release from internal storage, at the AIS and partially colocalizes with synaptodin (Sánchez-Ponce et al., 2011). This is also the same case for the Ca2+-binding protein annexin 6. Together, this evidence indicates a putative role of the CO in regulating Ca2+ dynamics (storage/release) at the AIS. Since Ca2+ levels have a significant impact on action potential generation and timing at the AIS (see Bender & Trussell, 2009; Yu et al., 2010 for references), and therefore should be strictly regulated, it is likely that the CO at the AIS is important for regulating neuronal excitability by controlling Ca2+ dynamics. However, as mentioned by the reviewer, there are no conclusive pieces of evidence showing the relationship between the CO and neuron excitability regulation. We will edit our statement accordingly.

In contrast to the findings of Schlüter et al. (Schlüter et al., 2019), which were conducted in the mouse primary visual cortex, Sánchez-Ponce et al. showed that nearly 90% of hippocampal neurons contain synaptopodin, the CO marker protein, at the AIS. Furthermore, Schlüter et al. also demonstrated that in the other 50% of neurons containing COs at the AIS, the COs change size during visual deprivation, and their presence correlates with AIS length changes as well as eye-opening. These observations do suggest that COs are related to neuronal activity. However, this correlation and the formation of COs may be specific to neuro subtypes or require certain triggers. This is another interesting direction to explore, and we will include it in the discussion of the revised manuscript.

Regarding the last point on Ca2+ and AIS plasticity, we were not excluding other factors that could potentially participate in AIS plasticity and will also discuss it in the revised version.

The Introduction ends with the rationale of the study, namely that the authors seek to ....."provide a detailed characterization of the AIS, including its structural and functional properties....". Structure is investigated, but function is limited to the barrier function of the AIS. Since the authors provide no electrophysiology that would really dissect AIS function, I suggest to rephrase this part and focus on transport.

Reply: As suggested, we will certainly emphasize the cargo barrier function of the AIS in AcD neurons in our introduction. But we would like to keep the term "AIS function", because it has already been nicely demonstrated electrophysiologically by previous studies that the plasticity effect of the AIS is very important for maintaining cellular homeostasis.

The Discussion is more a list of future plans than a context to current data. The authors could move some of the new questions they identify into an "outlook" section at the end? Also, again have a critical look at the literature that is cited and which statements are accurate.

For example, the 2nd phrase in the Discussion states that is was shown that AcD neurons have a "role in memory consolidation", referenced to Hodapp et al., 2022. However, that paper does not provide direct evidence of such a role for AcD neurons. The statement "Collectively, our data provide new insights into the development of AcD neurons and demonstrate that there are differences in AIS functionality between AcD and nonAcD neurons", is not correct. AIS function was not investigated outside of the axonal barrier, and here, the AcD and nonAcD cells do not differ. Also, although the Discussion is geared towards excitatory / glutamatergic neurons, it has been shown by others that interneurons show an even stronger trend to exhibit AcD morphology (work by the Wahle lab and others). This is not clear from the current text (also compare "...AcD neurons being a different subtype if pyramidal neuron").

Further original publications should be included in the paragraph highlighting patch-clamp recordings (see above). In the same context, the statement "...showed that rapid AID plasticity occurs mainly in hippocampal dentate gyrus cells but not in principal excitatory neurons" is not accurate (see Kim, Kuba, Jamann and others). Generally, the Introduction and Discussion would benefit from a very clear distinction between studies done in vitro versus those done ex vivo or in vivo. This needs to be stated in the Abstract as well.

Methods: For the imaging of synapses, the CO and VGSCs, it is not clear to me from the methods whether Nyquist conditions were applied to produce data that can support the quantification of nanoscale structures. Basing the analysis and interpretation of channel expression on fluorescence intensity profiles is problematic (variance in staining quality from samples to sample, lack of an internal standard). This should be noted in the text. In the text, the first two references given for "Induction of plasticity" do not reference the correct papers.

Reply: Thank you for the valuable suggestions; we will incorporate them into the revised version of the manuscript. The structure will undoubtedly benefit from these improvements. We will also have a further look into our interpretation of the literatures as well as citations during our revision time frame.

Regarding methods, as stated in response to the second point raised by this reviewer, we ensured that the Nyquist condition was adhered to throughout the study. The pixel size, z-step size, and optical setup of the microscopes used were already indicated in our methods section. With respect to Na+ channel staining, we were indeed aware of the potential issues posed by the experimental setup, and we will explicitly mention this in our revised manuscript. Additionally, we plan to measure other AIS scaffolding and membrane proteins that anchor Na+ channels to assess for potential changes, which could indirectly support our Na+ channel staining results.

Finally, the text is lacking a discussion of limitations of the study, especially from a methodological point of view. In the Abstract/Summary already, the authors could point out that this is a pure in vitro study. Interestingly, to this day, AIS relocation during plasticity events has only been shown in cell culture systems, and not in vivo. Therefore, this needs to be put into context here - the chosen system is great for the type of imaging approach presented here, but may look at a type of AIS plasticity that is not seen in vivo.

Reply: These are very good points. We will include the limitations of the study in the discussion. Indeed, due to technical and methodological challenges, the relocation of the AIS has not yet been demonstrated using animal models. However, in the study by Wefelmeyer et al. (Wefelmeyer et al., 2015), a similar relocation of the AIS resulting from chronic stimulation was observed in hippocampal organotypic slices, and it was accompanied by reduced excitability of neurons. Furthermore, in the same study, neurons with axons/AIS originating from basal dendrites were also mentioned. However, the measurement of chronic AIS plasticity in their study was not performed based on different classes of neuron types. Hence, our work complements their results. Given that the network connectivity of organotypic slices is much closer to real physiological conditions, it is likely that similar plastic adaptations could occur in vivo.

__Minor comments __

1. How does intrinsic neuronal activity play into developmental programs in vitro? Electrical activity in maturing neurons is a major part of how networks are shaped, and cells differentiate. This is not genetically encoded per se, but has been shown to be a major driving force of neuronal development in vivo. Is this reflected in the culture setting in any way? And have the authors considered testing early changes in activity patterns in their cultures to see whether AcDs and nonAcDs develop in similar percentages? To clarify, I am not asking for additional experiments. Reply: It is indeed a valid point that activity can influence neuronal morphology. Lehmann et al. (pre-print, doi: https://doi.org/10.1101/2023.07.31.551236) have recently demonstrated that increased network activity leads to more excitatory principal neurons adopting AcD morphology. However, our developmental data were collected from DIV0 to DIV5, an age at which dissociated neurons do not yet form functional excitatory synapses. Therefore, it is highly unlikely that network activity plays a role in shaping AcD neuron development during this early stage.

The authors may want to add a bit of a technical discussion on the choice of KCl and TTX as triggers for plasticity, especially at the non-physiological concentrations offered here and elsewhere (15 mM KCl).

Reply: We appreciate the reviewer for pointing this out. We will add this in our revised manuscript.

Some key statements would benefit from citing the appropriate original literature (some examples would be the original work by Kole, Bender and Brette on the role of the AIS in AP initiation; original work by D'Este and Letterier on the dendritic and axonal scaffold using nanoscopy; work by Kim, Kuba and Jamann on AIS plasticity in vitro and in vivo that is critical for a more informed discussion of AIS plasticity here, and others)

Reply: These are very good points, we will make suggested edits in the revised version.

In the Introduction, the authors word their text explicitly for excitatory neurons. However, AIS plasticity has also been observed in interneurons (work by the Grubb lab for example), and axo-axonic synapses are in fact not all inhibitory - this is in important factor to consider given the embryonic state of the culture material. Does the DIV maturation reflect how axo-axonic synapses "switch" from excitatory to inhibitory in vivo (also see work of the Burrone lab)? Can the conclusions form the paper really be drawn based on this type of system?

Reply: The AIS plasticity was indeed also observed in inhibitory interneurons (see Chand et al., 2015 for reference) and show opposite phenotypes compared to excitatory neurons. Also related to major comment #5, we did take the potential influence of AcD interneurons on the outcome of AIS plasticity experiment into consideration. Therefore, we also did a control experiment where inhibitory interneurons were labelled with GAD1 after chronic KCl treatment and these neurons were excluded from the analysis. Consistently, we got the same results that excitatory AcD neurons do not undergo chronic AIS plasticity. We will include this data in our revised manuscript. Further, in our current manuscript, we decided to focus on excitatory AcD neurons not only because they are the major functional unit in neuronal circuits, but also because the majority of the electrophysiological features were studied in excitatory AcD neurons. But we agree with the reviewer that AcD interneuron is definitely an interesting subject for follow up research in the future.

As mentioned by the reviewer, Pan-Vazquez et al. (Pan-Vazquez et al., 2020) nicely showed that axo-axonic synapses made by GABAergic Chandelier cells (ChCs) depolarise neurons in brain slices obtained from P12-18 animals. But this effect is reversed in slices obtained from older animals (>>P40). Of note, their results were based on cortical neurons but not hippocampal neurons, hence cell type specificity should be considered. More importantly, previous study reported that this conversion or switch of GABAergic interneurons from excitatory to inhibitory occurs on hippocampal neurons in P12-13 animals (Leinekugel et al., 1995). In dissociated hippocampal neurons from E18 rat embryos, this switch of GABAergic interneurons takes place on DIV9-11 and completes on DIV19, which should have a comparable neuronal developmental stage as the P12-13 in in vivo system (see Ganguly et al., 2001 for reference). Therefore, the conclusion could be drawn in an in vitro system, but it certainly needs to be validated in in vivo system.

The authors state that "less COs account for higher intrinsic excitability". Why is that the case?

Reply: According to Yu et al. and Bender et al., Ca2+ transient at the AIS regulates the generation of action potentials (APs). For instance, reducing Ca2+ transient at the AIS by blocking Ca2+ channels with either mibefradil (a T-type Ca2+ channel antagonist) or Ni2+ (which blocks R- and T-type channels) decreased the number of spikelets evoked by EPSP-like current injection and delayed the timing of spike generation (please see Bender & Trussell, 2009 for details). Therefore, we speculate that Ca2+ transients are less affected when there are fewer cisternal organelles (COs) at the AIS, which could have a more direct impact on AP initiation. However, this is just our hypothesis, and there is indeed no direct evidence showing that COs regulate Ca2+ dynamics. We will discuss this in the revised manuscript.

Last but not least, some very recent studies on AcD biology (Stevens, Thome, Lehmann, Wahle) is available online also on preprint servers and may provide additional support for the current study.

Reply: We will check these pre-prints and include relevant information into the revised version.

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

Learn more at Review Commons

---

Referee #4

Evidence, reproducibility and clarity

Summary

Han and colleagues present cell biological data on the development of axon-carrying dendrite (AcD) neurons - cells, in which the axon emerges from a dendrite and not as is more common, from the soma. This particular morphological feature is not new; Cajal already described AcD neurons and since then, their existence was demonstrated in a number of CNS and PNS neurons and across species. However, more recent work from the rodent hippocampus ex vivo and in vivo has pointed towards an interesting functional consequence of AcD morphology, in that these cells can circumvent perisomatic inhibition during network oscillations. Han and colleagues therefore investigate one of the core questions that arises from these previous studies: How is the AcD configuration achieved during development?

The authors utilize an in vitro model (isolated hippocampal neurons from E18 rat) and investigate core parameters of axonal maturation, especially the cytoskeleton, membrane-associated proteins and intra-axonal calcium stores of the axon initial segment (AIS), where action potentials are generated and which contributes to neuronal polarity at different days in vitro (= maturation states). Using a combination of immunofluorescence, confocal, spinning disk and STED microscopy, and plasticity protocols, the authors present evidence indicating that during development in vitro, AcD neurons follow an intrinsically encoded developmental program, the AIS in nonAcD and AcD cells has comparable cytoskeletal features and retains cellular polarity in both configurations. Using culture conditions to elicit AIS plasticity, the authors then find differences in that AcD neurons do not seem to undergo AIS plasticity and generally show a reduced number of intra-axonal calcium stores. Also, AcD neurons are shown to have fewer axo-axonic synapses at the AIS than nonAcD neurons.

Major comments

This study aims at investigating an important question in AcD biology, and uses an easily-accessible model system (E18 rat derived hippocampal neurons in vitro). In this, the study follows previous work in vitro, and nicely reproduces some data, which is a strength of this current study in my opinion. That said, the system is, by nature, artificial and the emergence of axons in vitro often deviates from data obtained in vivo.

Throughout the manuscript, the authors often draw clear-cut conclusions which require a far more critical reflection of what their model can actually accomplish. Thus, a number of statements are not supported by the data (see below). The presentation of the data in the Supplements needs to reflect data distribution, which they currently do not. Likewise, showing S.E.M. instead of S.D. needs to be looked at critically. Otherwise, the data and methods are presented in such a way that they can be reproduced. The quality of the micrographs and videos is excellent and convey the main messages of the study in a very accessible way. I do not see the need for additional experiments, but would ask the authors to critically look at the following issues:

1. A general limitation of this study is the low N for some critical experiments. In several experiments, individual cells become an N, therefore boosting the power of the analysis when in reality, due to the known heterogeneity of AIS length, position, and general cell morphology in vitro, the aim should be to compare means across animals / preparations, each consisting of a comparable number of individual cells. This is especially important for the analyses of COs, axo-axonic synapses and channel expression at the AIS.
2. Such critical parameters as e.g. synaptic innervation at the AIS are investigated in a way that does not support the clear statements given, e.g. "The AIS of AcD neurons receives fewer inhibitory inputs" (Highlights statement) or "AcD neurons have less inhibitory synapses at the AIS" (header of Fig. 6). The overall number of analyzed cells is low (3 and 4 preparations, respectively and approximately 50-cells for each marker). The combination of a pre- and postsynaptic marker for inhibitory / excitatory neurons is a solid decision, but the analysis is not done based on the close approximation of these markers, in 3D, along an AIS, but rather in maxIPs and without any regard of whether pre-and postsynaptic markers are actually close to each other not. The expression of these markers alone just points towards the epitopes being expressed, but are they localized to each other in such a manner that they could form bona fida synapses? The methods are not totally clear on the image depth (tile scans with 5 µm in z will not provide the detail of information to resolve synapses, so how did the authors address the subcellular analysis here and for the CO and VGSCs?). And generally, were Nyquist conditions taken into consideration throughout the study? This can be clarified in text and does not require additional experiments.
3. The chapter on AIS plasticity is certainly an interesting addition to the study, but is a bit superficial, yet reaches strong conclusions ("More importantly, it further indicates that the AIS of AcD neurons is insensitive to activity changes"). This is based on unphysiological concentrations of KCl, and certainly not on network manipulation that truly tests synaptic activity. It also comes back to the 1st point above. A suggestion would be to edit the conclusion.
4. The rationale behind looking at the cisternal organelle (CO) in this study is outlined in the Introduction, where the authors state that "...... and is responsible for calcium-handling". What is "calcium-handling" and where is the evidence cited? Furthermore, in the Results, they state that "...both compounds (VGSCs and COs) are critical for the AIS to regulate neuronal excitability". While this is the case for VGSCs, there is no conclusive evidence in the literature whether of not the CO is "critical" for neuronal excitability. In fact, a number of neurons have no CO in the AIS (as much as 50% of all AIS in mouse primary visual cortex for example do not express synpo at the AIS at all, Schlüter et al., 2017). The CO can therefore not be as critical for AP initiation as the authors state. Furthermore, the authors state that "AIS plasticity in excitatory neurons is triggered by calcium signaling". While certainly shown and adequately cited here, other factors (independent of calcium) can also play a role, therefore this statement is a bit absolute and should be edited accordingly.
5. The Introduction ends with the rationale of the study, namely that the authors seek to ....."provide a detailed characterization of the AIS, including its structural and functional properties....". Structure is investigated, but function is limited to the barrier function of the AIS. Since the authors provide no electrophysiology that would really dissect AIS function, I suggest to rephrase this part and focus on transport.
6. The Discussion is more a list of future pans than a context to current data. The authors could move some of the new questions they identify into an "outlook" section at the end? Also, again have a critical look at the literature that is cited and which statements are accurate. For example, the 2nd phrase in the Discussion states that is was shown that AcD neurons have a "role in memory consolidation", referenced to Hodapp et al., 2022. However, that paper does not provide direct evidence of such a role for AcD neurons. The statement "Collectively, our data provide new insights into the development of AcD neurons and demonstrate that there are differences in AIS functionality between AcD and nonAcD neurons", is not correct. AIS function was not investigated outside of the axonal barrier, and here, the AcD and nonAcD cells do not differ. Also, although the Discussion is geared towards excitatory / glutamatergic neurons, it has been shown by others that interneurons show an even stronger trend to exhibit AcD morphology (work by the Wahle lab and others). This is not clear from the current text (also compare "...AcD neurons being a different subtype if pyramidal neuron"). Further original publications should be included in the paragraph highlighting patch-clamp recordings (see above). In the same context, the statement "...showed that rapid AID plasticity occurs mainly in hippocampal dentate gyrus cells but not in principal excitatory neurons" is not accurate (see Kim, Kuba, Jamann and others). Generally, the Introduction and Discussion would benefit from a very clear distinction between studies done in vitro versus those done ex vivo or in vivo. This needs to be stated in the Abstract as well.

Methods: For the imaging of synapses, the CO and VGSCs, it is not clear to me from the methods whether Nyquist conditions were applied to produce data that can support the quantification of nanoscale structures. Basing the analysis and interpretation of channel expression on fluorescence intensity profiles is problematic (variance in staining quality from samples to sample, lack of an internal standard). This should be noted in the text. In the text, the first two references given for "Induction of plasticity" do not reference the correct papers.

Finally, the text is lacking a discussion of limitations of the study, especially from a methodological point of view. In the Abstract/Summary already, the authors could point out that this is a pure in vitro study. Interestingly, to this day, AIS relocation during plasticity events has only been shown in cell culture systems, and not in vivo. Therefore, this needs to be put into context here - the chosen system is great for the type of imaging approach presented here, but may look at a type of AIS plasticity that is not seen in vivo.

Minor comments

1. How does intrinsic neuronal activity play into developmental programs in vitro? Electrical activity in maturing neurons is a major part of how networks are shaped, and cells differentiate. This is not genetically encoded per se, but has been shown to be a major driving force of neuronal development in vivo. Is this reflected in the culture setting in any way? And have the authors considered testing early changes in activity patterns in their cultures to see whether AcDs and nonAcDs develop in similar percentages? To clarify, I am not asking for additional experiments.
2. The authors may want to add a bit of a technical discussion on the choice of KCl and TTX as triggers for plasticity, especially at the non-physiological concentrations offered here and elsewhere (15 mM KCl).
3. Some key statements would benefit from citing the appropriate original literature (some examples would be the original work by Kole, Bender and Brette on the role of the AIS in AP initiation; original work by D'Este and Letterier on the dendritic and axonal scaffold using nanoscopy; work by Kim, Kuba and Jamann on AIS plasticity in vitro and in vivo that is critical for a more informed discussion of AIS plasticity here, and others)
4. In the Introduction, the authors word their text explicitly for excitatory neurons. However, AIS plasticity has also been observed in interneurons (work by the Grubb lab for example), and axo-axonic synapses are in fact not all inhibitory - this is in important factor to consider given the embryonic state of the culture material. Does the DIV maturation reflect how axo-axonic synapses "switch" from excitatory to inhibitory in vivo (also see work of the Burrone lab)? Can the conclusions form the paper really be drawn based on this type of system?
5. The second header in Results is not clearly formulated. What is meant by "consensus developmental sequence"?
6. The authors state that "less COs account for higher intrinsic excitability". Why is that the case?
7. Last but not least, some very recent studies on AcD biology (Stevens, Thome, Lehmann, Wahle) is available online also on preprint servers and may provide additional support for the current study.

Referees cross-commenting

In my opinion, the comments by the other three reviewers are clear, insightful and supportive of / complementary to my own. There are some strong leads within the revisions that the authors hopefully find helpful in the preparation of their final manuscript. I have no doubt that the final publication will be viewed as an important and significant finding in the field of axon onset biology.

Significance

The study by Han and colleagues addresses a timely and relevant question and provides excellent quality imaging data (fixed cells and live imaging), as well as convincing superresolution. The authors also provide a solid methods section that will aid others in repeating these experiments. The central question of how AcD neurons develop is of great interest and the study highlights novel findings especially regarding the detailed analysis of axonal and dendritic transport features in AcD cells. The authors also point out a number of questions that arise from their data and that can provide helpful insight for other researchers in the field. The study has limitations in that it is a pure in vitro study, and some data are based on a low sample number as well as superficial expression analysis (synapses, channels). A number of conclusions made by the authors are therefore not really supported by the current data. The discussion would benefit from a more detailed analysis of the current literature in the field and needs critical reflection on what the shown data really support in form of a section on "limitations".

The study is of broader interest for numerous research fields (developmental neurobiology, axon biology, AIS biology, neuronal plasticity). Given the focus that AcD neurons have received recently in the field of learning and memory consolidation, this study also provides interesting future questions for researchers with a background in network function and behavior.

Reviewer's expertise: Developmental neurobiology, axonal plasticity, AcD neuron morphology and development, in vivo rodent behavior, human slices, confocal and superresolution microscopy, patch-clamp electrophysiology

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 Han and colleagues report about structural and functional studies on the development of axons originating from dendrites. Leveraging the primary hippocampal neuron preparation, they investigated fundamental cell biological questions including microtubule organization, cargo transport and the early neurite development. I am impressed by the timelapse movies with live AIS labels providing, to the best of my knowledge, the first glance into the development of an axon emerging from a dendrite. The study is technically very good, a pleasure to read, and the results are well described. While conclusions about structure are well supported by their data, the claims about 'function' are weak and speculative. I have listed some issues and by improving clarity the study could become a valuable resource for the field.

1. The authors classify neurons into axon-carrying dendrite (AcD) and non-AcD neurons by measuring the stem dendrite length (> 3 µm). I could not find the validity for this cut-off. The non-AcD neurons in Fig. 6B appear more AcD to this reviewer, and, in addition, other researchers have proposed a third category of 'shared root' neurons (doi: 10.7554/eLife.76101). For purposes of reproducibility and transparency, please provide first a comprehensive overview of the entire population of morphologies (i.e. all cells in control conditions). The distances from the soma could be plotted in histogram (etc.) and authors may want to think about independent supporting evidence for the cut-off to classify AcD and non-AcD neurons.
2. Related to point #1 the primary hippocampal neuron system is excellent for cell biological questions but comes with the drawback of imaginative morphologies including neurons with multiple axons and AISs. It is not mentioned here but literature indicates up to 20% of neurons have two axons (e.g. doi: 10.1007/s12264-017-0169-3, 10.1083/jcb.200707042). How did the authors classify the double axon cells? Since the main hypothesis is the existence of an intrinsic program for AcD neurons (p. 5 top), the two axons from one neuron should develop similarly. The authors can easily test this with the data.
3. Some interpretations about function are not correct and the authors should reconsider these. A role of cisternal organelles on neuronal excitability remains to be demonstrated (and see doi.org/10.1002/cne.21445 showing there is none). In addition, the statement that lower fluorescence intensity of Pan-Nav1 is indicating reduced excitability is flawed. Antibody staining does not scale linearly with voltage-gated sodium channel density and since the AIS of AcD neurons is further from the soma it is most likely smaller in diameter which may account for apparent fluorescent differences. For biophysical reasons (for details I refer to 10.3389/fncel.2019.00570, 10.1016/j.conb.2018.02.016 and 10.7554/eLife.53432) smaller diameter axons will be easier to depolarize by depolarizing voltage-gated channels or excitatory synapses. Finally, in AcD neurons the AIS distance from the soma poses all sorts of interesting cable properties with the soma and the local dendritic membrane and the electrotonic properties alone suffice to make these neurons more excitable.
4. Comparing AcD and non-AcD neurons for AIS plasticity is an excellent idea but the present statistical design is not suitable for answering this question. The authors should directly compare non-AcD and AcD neurons within a two-way ANOVA design, asking the question whether the independent variable axon type is significantly different and interacts with plasticity. Related points: 'AIS distance' in Figure 7 seems to refer to something else than distance from soma (Figure 1). Please clarify. What were the absolute distances from the soma for the AcD neurons and was this dependent on treatment?

Minor comments

At p. 7 is stated that "The percentage of none-AcD forming collaterals at DIV1 is much lower than for AcD neurons" but statistical support is lacking. The conclusion in the next line is that "AcD neurons follow consensus development". That is puzzling given the difference just mentioned before. Please clarify. A study not cited in this manuscript showed distinct dendritic morphologies (doi: 10.1073/pnas.1607548113) and AcD interneurons are different for their axonal arborization (doi: 10.1242/dev.202305). Differences in growth of branch arborization could hint to subtypes. Are the AcD and non-AcD neurons different in their adult morphology? A detailed account of the axonal and dendritic trees would strengthen the data.

Some key references are not included here, and a number of these are mentioned above. In the context of the detailed MT and Rab3A vesicle and cargo transport studies, please acknowledge some of the pioneering work of Alan Peters revealing the ultrastructure of axons emerging from dendrites. See Figs. 5-7 in Peters, Proskauer and Kaiserman-Abramof IR., J Cell Biol 39:604 (1968).

What is the identity of the neurons? It makes a difference if the cells are interneurons or pyramidal neurons, CA1 or CA3-like.

For plasticity experiments the authors uses cells as independent measurements, but this is inflating the power. How many cultures were used?

Referees cross-commenting

When reading the other reviews I feel they are constructive and providing sufficient conceptual and technical insight to prepare a revision. Although some concerns are overlapping, with 4 independent review reports perhaps not all issues can be addressed within the estimated time frame of 3 months.

Significance

That axons originate from dendrites dates back to the 19th century drawings of Ramón y Cajal but today most textbook and schematic drawings of the neuron still show polarized axons and dendrites both emerging from the soma. Since a few years this specific morphological arrangement begins to receive attention and many fundamental cell biological questions remain to be answered. Leveraging the primary hippocampal neuron preparation, the authors use technically clever experiments to generate new insight into the microtubule organization, cargo transport and the early neurite development. The live imaging of fluorescent labelled axon initial segments is elegant, and an important conclusion is that the stem process, carrying dendrite and axon, grows at a later stage in development. Limitations of the primary neurons should be discussed, however, and the functional consequences of positioning axons on dendrites are not as simple as described by the authors. The study could become a valuable resource for those working in basic research, providing new technical directions.

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 Han et al. investigates the properties of AIS along axon carrying dendrites (AcDs). These are enigmatic structures with at present poorly defined features. Han et al. work to further characterize the nature of these AIS. Overall, the data are mostly compelling and the reveal that AIS at AcDs are mostly like those of AIS arising from the cell body. Many features were examined and shown not to differ. There were a few instances where the authors claim differences, and this reviewer is not convinced - see comments below. Overall, I think with bit more careful examination of the main differences this could be a nice descriptive paper reporting features of AIS along AcDs.

Major questions:

1. The authors suggest that there is reduced Na+ channel density at AcD AIS compared to other AIS arising from the cell body. This is not convincing. Immunostaining for Na+ channels is notoriously difficult and sensitive to fixation since the epitopes of the anti-Pan Nav antibodies are highly sensitive to fixation. In addition, this is based on immunofluorescence intensity quantification. Since the mechanism of localization is through binding to AnkG, the authors should also measure other AIS proteins like AnkG, b4 spectrin, and Nfasc. Do these change? If all uniformly change I would be much more inclined to accept the conclusion. If they do not change, it still doesn't rule out the concern about fixation conditions and slight differences in the cultures. The authors indicate there is about a 40% reduction in fluorescence intensity. That is quite large. This big difference should also be confirmed in brain sections.
2. The analysis of inhibitory synapse differences at the AIS are also not compelling - this is a limitation of the culture system. The authors have no control over the density of inhibitory neurons in the culture well. This interaction is not intrinsic to the AcD neuron, but rather a feature of neuron-neuron interactions which should only be modeled in the animal.
3. Finally, this reviewer is also skeptical of the chronic plasticity changes in AIS 'distance.' The authors claim their results are consistent with prior reports, but they see about a 1.5 um shift. Prior studies (Grubb et al.) report 15-17 um change - a full order of magnitude larger than what is reported here. The authors also show no differences in other previously described changes at the AIS. Together with the other results showing AcD neurons and non- AcD neuron AIS are mostly the same, the conclusion that one behaves differently is not compelling with the tiny shift reported.
4. Finally, the major limitation of this study is that it is performed in vitro. Surprisingly, the authors actually argue this is a feature of their system. While it is true some of the questions can be addressed perfectly well in vitro, many cannot. In the first paragraph of the results the authors state an advantage of their system is that there are no microenvironments to influence the development of the AcDs. I'm afraid I view this as a drawback. The authors suggest this is an opportunity to examine intrinsic mechanisms of development - true, but it also foregoes the opportunity to determine if the outcomes are different from what occurs in vivo. To this point, the authors report that only 15-20% of the population of hippocampal neurons in culture are AcD neurons. But in their introduction they cite other literature indicating 50% of hippocampal neurons in vivo are AcD neurons - this suggests that the environment of the hippocampus in vivo influences whether a neuron becomes an AcD neuron or not.
5. I appreciated the balanced discussion of whether this is a stochastic or genetically programmed process. This could have been emphasized earlier in the results since the authors invoke the concept that "...their development must be driven by genetically encoded factors rather than specific...". The authors have not shown this and cannot show it in this system. Indeed, as stated in point 4 above, I think their data argue against a simple genetic program.

Significance

interesting subject, timely as features of AIS are of great interest now - especially as a relatively new form of neuronal plasticity. Highly descriptive paper, but emphasizes in this reviewer's opinion that AcD neurons and non AcD neurons have AIS that are essentially the same.

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 manuscript by Han et al. describes the structural and functional differences between pyramidal cells in which the axon emanates from a basal dendrite (axon-carrying dendrite cell, AcD cell) and cells with a 'canonical', i.e. somatic origin of the axon (nonAcD cells). They investigate how pyramidal neurons develop into AcD or nonAcD cells during cell development and characterize the cytoskeletal architecture in the two cell classes. Additionally, they examine whether and how axon initial segments, the most important structure for action potential generation, change upon varying activity of the neuron.

The major claims of the paper are:

i) The formation into an AcD or nonAcD cell is intrinsically encoded by a developmental program.

ii) The cytoskeletal structure of AcD and nonAcD cells is similar. However, the stem dendrite inherent only to AcD cells is structurally more similar to an axon than to a dendrite

iii) Axon initial segments of AcD cells contain less cisternal organelles and show less homeostatic plasticity The authors make use of primary cell cultures from rat hippocampus which are a standard model to investigate developmental questions of single cells and neuronal networks. The manuscript is well structured and in general reads well and the data and conclusions are convincing. I have only a few major comments.

Major comments:

The authors cite that acetylated and tyrosinated microtubules have different spatial and compartmental distribution in dendrites and axons and investigate the distribution in the AIS of nonAcD cells and AcD cells, as well as the stem dendrites. However, they just show one example of two different cells (Figure 2D and E) without any statistical analysis. Either, they should remove this part or provide a thorough quantification. The authors use EGFP-Rab3A vesicle to investigate anterograde transport at the axon and dendrites. They find a slightly faster transport of these vesicles at the AIS of AcD cells and conclude the axonal cargos in general are transported faster across the AIS in AcD cells. In my opinion, this generalization based on one type of vesicle is too far-fetched. As stated above, the manuscript is well structured and generally reads well. However, throughout the text there are always small mistakes that should be corrected by careful proofreading. Examples are

Page 6, last paragraph: ...AcD neurons generated [a] collateral...

Page 24, last paragraph: ... line was then drew [drawn] along ...

Page 24, last paragraph: Neurons with ... was consider [were considered] as ...

Page 25, first paragraph: Antibodies ... was [were]

Page 41, (E) Percentage of AcD neurons [that] generate [a] collateral or bifurcate

Minor comments:

In the introduction, the authors describe how synaptic inputs are received at the dendrites and propagated to the soma in the form of membrane depolarizations. They should add 'excitatory' to synaptic inputs or also describe the impact of inhibitory synaptic inputs at the dendrites.

In my opinion, Figure 2 could be presented in a slightly better way. The lower part of panel A better fits to panel B, which is next to the upper part of panel A. I understand that the authors systematically present their data first for nonAcD cells and then for AcD cells. However, in this special case it is a little bit more difficult to read the current figure in that order.

The results displayed in Figure 4 are presented in a slightly confusing order. The authors jump from 4D to 4G, then to 4I and 4E, 4H, 4F. Similarly, 4M and N are addressed before 4O and P to finally get to 4K and L. It would be beneficial to present and address the data in a stringent way.

Significance

General assessment:

This study addresses a very important and timely question about structural and functional cell diversity of cortical pyramidal neurons. The specific function of AcD cells is currently mostly unknown, which is astonishing given their abundance of 15-50% of pyramidal neurons in cortical structures.

Advance:

This study presents a significant step forward in comprehending the structural and functional relationship of signal computation in single neurons.

Audience:

The study will be important for a wide readership working on very different levels including cellular, network, and behavioral neuroscience.

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

Learn more at Review Commons

---

Reply to the reviewers

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

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Angara et al describe a secreted Coxiella burnetii effector with an FFAT motif, which they name CbEPF1, that localizes to host cell LDs and is able to interact with VAP domain containing host proteins. Furthermore, this interaction is able to impact LD size. The study has several interesting findings that suggest that this protein can impact ER structuring around lipid droplets. The use of bacterial 2-hybrid approach to demonstrate binding between CbEPF1 and VAP domain containing proteins is convincing, further supported by co-ip experiments that go on to establish specificity of each of the FFAT motifs for distinct VAP domain containing proteins. However, the relationship between LD localization of the protein and ER-restructuring to the outcome of infection is not clear from the data presented here. In addition, there are many analyses that need to be performed for the data to convincingly support the claims made in the manuscript.

My major concerns are:

1. The authors claim that CbEPF1 localizes to lipid droplets in Coxiella burnetii infected epithelial cells. Towards this, the authors express CbEPF1-GFP in C. burnetii infected cells expressing BFP-KDEL. The data in Figure 1B and 1C indicate that the protein localizes to lipid droplets in oleic acid treated cells. It is interesting to note that without addition of oleic acid, the protein localizes to the ER. What is surprising is that the BFP-KDEL signal is also localizing to the LD surface (Figure 1B and 2A). While in this section it seems that the protein migrates from ER to LDs, in the later sections, similar data is used to make the claim that the protein induces ER apposition to the LD and localizes to regions of LD-ER contact. Therefore, it raises several questions about the localization of the protein: (i) is it an ER-localized protein that migrates to LDs. In that case, what features of the protein enable its LD binding? In addition, the authors must perform LD isolation to validate that the protein indeed localizes to lipid droplets. (ii) Is it an ER-localized protein, that like BFP-KDEL has the ability to localize to ER-LD contact sites, but remains on the ER membrane? Again, biochemical evidence supporting membrane specification is important to understand the localization of the protein. The authors have focussed only on the FFAT motifs of CbEPF1 and not described the overall domain analysis of the protein. It is therefore difficult to understand at this point how the protein localizes to the ER and the ER-LD junction or LD.
2. Quantitative image analysis:

(i) Authors must perform colocalization analyses to substantiate the claims for ER/LD localization. The authors refer to "extended ER-LD contacts" in figure 2B and the text. These data need to be supported with colocalization analysis between BFP-KDEL and the GFP channel.

(ii) Related to Figure 4A: Mander's Colocalization analyses with Costes correction are required to convincingly demonstrate that the dual FFAT motif is required for ER-LD contacts.

(iii) Related to Figure 4B: Please show the LD phenotype of untransfected, and CbEPF1-GFP transfected cells also. Can the authors provide a means to quantify the clustering of LDs.

(iv) Figure 5A and B. It is not clear from the figure legend whether the data are pooled from multiple experiments or a single experiment. Experimental replicates must be incorporated in the final analysis. 3. The data presented in this manuscript depends largely on overexpression of the protein in uninfected cells. Given that C. burnetii induces LD formation, there are three main areas that need clarity:

(i) What is the localization of the protein in infected cells without the addition of oleic acid under conditions where infection itself induces LD biogenesis.

(ii) What happens to ER-LD contacts upon infection with C. burnetii?

(iii) Does the presence of CbEPF1 play any role in infection induced LD biogenesis? This may be an optional experiment to undertake at this point as this would involve significant amount of time investment in generating bacterial strains.

Minor comments:

1. Line 36: "maintain" instead of "maintains"
2. The introduction cites mainly reviews with overlapping concepts but does not cite primary literature in the area of organelle-organelles contacts and inter-organelle communication (lines 39-40 and line 49). It would be good to cite key primary research articles in these areas.
3. There are some crucial references related to bacterial secreted effectors that target host lipid droplets, that are missing from the introduction. For example, Chlamydia trachomatis is known to secrete effectors that localize to host lipid droplets (PMID: 18591669). Legionella pneumophila secretes a small GTPase LegA15 to lipid droplets impacting vesicle secretion of the host cell (PMID:36525490).
4. For all figures, please show all individual channels in monochrome and a merge of BFP-KDEL+LD marker and merge of CbEPF1-GFP+LD marker and/or BFP-KDEL+CbEPF1 (wherever appropriate).

Significance

The study by Angara et al reports a dual FFAT motif containing protein of Coxiella burnetii that impacts ER-LD association. The strengths of the study lie in characterization of the FFAT motifs in VAP domain binding, and the role of these FFAT motifs in mediating ER-LD contacts. However, the claims made towards LD localization versus localization to ER-LD contact sites by this protein are not well supported by the data. In addition, the relevance of these findings in infected cells are not addressed in this study as the data presented here pertain to an over-expression system.

Bacterial proteins that exit the bacterial containing vacuole and impact organelles outside the endocytic compartments are fascinating as they have the potential to impact global processes such as transcription, metabolism, and protein secretion. Lipid droplet (LD) homeostasis is dysregulated in many bacterial infections and also plays a crucial role in host defense against infection. Therefore, the knowledge of bacterial effectors in this dysregulation will also potentially provide means of countering bacterial strategies to affect host LD homeostasis. Therefore, the findings presented by Angara et al will provide conceptual and mechanistic advances to the specialized audience in infection and immunity. However, I can foresee that the findings have the potential for interesting tools to be developed for organelle-organelle contacts to be studied, provided there is more clarity on how the protein localizes to the ER/ER-LD contacts/LDs.

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 is a strong manuscript about the existence of proteins coded by intracellular parasites (here Coxiella) that have evolved to parasitise the lipid transport machinery of their hosts. This is a first in that the parasite protein acts at a distance from the parasite itself, manipulating two of the host organelles - and not acting at their site of contact with PVs. There is considerable research into one protein and its effect when expressed by itself.

Despite all the advances there are a couple of areas where the manuscript can be improved, and a few extra fairly straightforward experiments added about the amphipathic helix. Even though these are unlikely change the overall message, they would make the story more complete.

Major points

More details are required about the amphipathic helix. Check that the AH does target LDs by expression of the AH alone in a GFP chimera +/- oleate and then mutagenesis. Also show the AH in a helical wheel projection (eg by Heliquest) and say if it aligns with similar AHs in homologs (see my point below)

Fig 1B: In infected cells, do the affected LDs tend to be close to the PVs?

Also in Fig 1B: highlight small KDEL+ve ER rings around LDs here. Study whether LDs have these in infected cells without the confounder (?artefact) of EPF1 over-expression

Fig 2A the ER looks quite different here from Fig 1B, even at t0. Grossly the strands are spaced wider apart. In detail there are no rings around LDs. Can the authors explain this? Which morphology is common, especially in cells early in infection without co-expressed protein?

Fig 6 & Line 237: "As the N-terminal region of CbEPF1 is undefined": I suggest that the authors could do more here. At minimum change model to highlight the strong probability that the N term is a globular domain that functions at the LD ER interface. (What are three other unidentified LD proteins? I suggest omitting them).

Although the Alphafold prediction for EPF1 is low confidence only, in a few minutes of BLAST searching I found the homolog A0A1J8NR10_9COXI (also FFAT+ve) which has a moderately confident structural prediction for its N terminus. This model has a quite large internal hydrophobic cavity, indicating lipid transfer capability and function similar to known LTPs. This means that action as a "tether" possibly results from experiments with viral promoters (see minor point on terminology).

Minor:

Fig 2B: add more arrowheads/arrows to fit legend (says they are both multiple)

FFAT selectivity for MOSPD2: say if this fits the di Mattia or (as appears likely) it extends the known differences between VAPA/B and MOSPD2. Also say if VAPA is expected to behave as VAPB

Explain how "Mutations in the CbEPF1 FFAT motif(s) did not influence CbEPF1-GFP localization to either the host ER (Supplementary Fig. 1)". In F3mt this shows that EPF1 has a way to target ER other than FFAT/VAP. Discuss if that is via AH insertion in ER.

Also, the (admittedly low) level of ER targeting is possibly slightly reduced by F3mt, as shown by greater GFP in the nucleus in the single cells shown. If this is a feature of the whole field of cells, it implies that the FFATs normally work with the AHs to target EPF1 to the ER.

"clustering" LDs w F3mt: could this indicate dimer formation by CbEPF1? Note: to me it appears wrong to describe fig 4A as showing ER exclusion. LD proximities to each other dominate. It's not 100% clear that LDs cluster as their proximities are not universal: "LD-LD interactions" may be (very) weak.

Fig 5: can levels of EPF1 here be compared to those in cells undergoing natural infection(approximate comparison by qPCR better than nothing if no antibodies are available)? Fig 5a: would it be possible to increase the number of cells counted to attempt to make the reduced number of LD in F3mt significant?

Minor

Line 226: no sequence homology: misses the point- there is the common feature of an AH

Issue to be discussed, as probably too difficult to experiment on: when EPF1 is on the ER does it engage vap only weakly (implying a means to mask its motifs), since if the interaction is strong vap is then unable to bind other partners?

Line 245: "MOSPD2, a sole VAP that is known to localize on LD surfaces" (worth citing Zouiouich again here). Do the cells/tissues infected by Coxiella express MOSPD2?

Line 259/260: this "suggestion" about cholesterol should be toned down. It is a speculation that could be tested in future, but the data here do not suggest it.

'Tether' this word implies more than just bridging but also a role in the physical formation of the contact. Since EPF1 most likely has an LTP domain, it seems linguistically confusing to refer to it as a tether, especially since the experiments that physically later LD-ER contact involve probable over-expression.

Discuss whether it is 100% certain that EPF1 is in the host cytosol or whether some experiment(s) at a future date (proteomic/western blotting) will be needed to make that conclusion 100% secure.

Referees cross-commenting

COMMENT 1

I realise that both reviewer 1 and reviewer 3 have considered this MS carefully, but I think that their reviews could be improved in some respects. I will add two comments, one for each of the other reviews.

Reviewer 1. The review poses multiple questions to the authors suggesting that answering these questions experimentally would strengthen the paper. Some of the points seem to misunderstand what is the accepted standard for membrane cell biological research into membrane contact sites. While it might be that the authors can rebut these points, I think it is preferable to use Cross Commenting as an opportunity to address these issues beforehand.

Major Comment 1: CbEPF1 and ER-LD contact

Looking at endogenous proteins: I wondered about the same point, but I concluded that this is not likely to be possible in the scope of this submission. If it were possible then I guess the authors would have attempted it. Looking on Google Scholar I could find no example of an endogenous Coxiella proteins being tagged in the bacterial genome. So the only way to find the portion is via an antibody. Assuming the authors do not have one, I do not think we should ask for one at this stage in the publication process.

Electron microscopy: the reviewer is incorrect to say that this is necessary. It may be the gold standard, but it is a huge amount of extra work. Furthermore it is not at all necessary when the protein in question localises clearly to the interface between organelles identified by confocal microscopy.

Can a specific CbEPF1 domain be identified? Here a Amphipathic Helix has been identified, but the lack of dissection of that region by the authors explains this question by the reviewer, which is also shared by Reviewer 3. I agree with the implication that more should be done to dissect that.

Major Comment 2: CbEPF1 FFAT motifs and VAP binding

Are the two FFAT motifs redundant or synergistic? I would say that the authors have addressed that to a reasonable extent

CbEPF1 binding specificity towards a VAP/MOSPD2 Ditto

Major Comment 3: LD clustering

Since this is an effect of mutated protein only, I think that the 3 questions posed at the end here need only be addressed in Discussion.

Major Comment 4: CbEPF1-mediated increase in LD number and size

less LD upon expression of F1mt or F2mt, compared to WT: this seems wrong. The numbers are the same. The comment about IF images are unjustified as they have been quantified and do show a difference. I agree that the biological relevance is unclear, and that this might be addressed. That would require making a mutant Coxiella strain. While that would make a big different to this work, my feeling is that this is well over a year's work.I would be guided by the authors on that and I would not suggest it as required for this MS.

De novo LD production at the ER is unlikely: This statement is ill-considered as the FFAT motifs ARE required (Fig 5). Furthermore, in all systems ever reported de novo LD production takes place at the ER, so any alternative would be quite extraordinary.

Altogether, strengthening this aspect of the study: In my view, this area does not need more work and it would not be constructive to ask for more.

Major Comment 5: Functional relevance

assessing the phenotype of a Coxiella CbEPF1 mutant I agree that this would be good, but it mightn't be feasible within the confines of this one paper. In the various projects that have made transposon mutants of Coxiella, has a strain been made that affects EPF1? If not, then the authors should state this and discuss it as work for the future. The reviewers cannot expect any experiments!

Is VAP required for Coxiella intracellular growth/vacuole maturation? On the surface this suggestion seems to offer an experimental route to understanding EPF1. However, VAP binds to >100 cellular proteins, many relating to lipids traffic and a considerable number of these already lcoalised to lipids droplets (ORP2, MIGA2, VPS13A/C). It is therefore unlikely that such an experiment would be interpretable, and I recommend that this request be reconsidered.

Are LD formation induced upon infection? Are ER-LD contact increased upon infection? These are very reasonable ideas and the results would be interesting additions to this paper.

COMMENT 2 I have given one set of comments already. Here are my comments for Reviewer 3.

The review makes a few assumptions that I question. While it might be that the authors can rebut these assumptions, I think it is preferable to use Cross Commenting as an opportunity to address these issues beforehand.

Major Point 1: What is surprising is that the BFP-KDEL signal is also localizing to the LD surface: "Surprising" is misguided, as it seems to deny the probability that there is a class of proteins that sit at organelle interfaces binding to both partners simultaneously. Maybe the reviewer means "significant" here, in which case I would agree.

The authors must perform LD isolation the reviewer is incorrect to say that this must be done. It is a huge amount of extra work. Furthermore it is not at all necessary when the protein in question localises clearly to the structures, and its may not even work as the protein may need a reasonably high general concentration to avoid gradual dissociation (wit any re-association) during organelle purification.

what features of the protein enable its LD binding? Here an Amphipathic Helix has been identified, but the lack of dissection of that region by the authors explains this question by the reviewer, which is also shared by Reviewer 1. I agree with the implication that more should be done to dissect that.

Major Point 2: Quantitative image analysis:

Mander's Colocalization analyses with Costes correction are required No. The images in Figure 4 speak for themselves.

Please show the LD phenotype of untransfected, and CbEPF1-GFP transfected cells also This s a good idea.

provide a means to quantify the clustering of LDs Unnecessary. Not all findings need to be quantified.

Major Point 3:

Data depends largely on overexpression of the protein in uninfected cells. I agree

What is the localization of the protein in infected cells? I wondered about the same point, but I concluded that this is not likely to be possible in the scope of this submission. If it were possible then I guess the authors would have attempted it. Looking on Google Scholar I could find no example of an endogenous Coxiella proteins being tagged in the bacterial genome. So the only way to find the portion is via an antibody. Assuming the authors do not have one, I do not think we should ask for one at this stage in the publication process.

What happens to ER-LD contacts upon infection with C. burnetii? This is a very valid question, and answering it would not only strengthen the manuscript but should be achievable in 1-3 months.

Significance

This work takes a reasonably big step towards uncovering how parasites have mimicked the molecular machinery of contact sites, here in the context of ER-LD interactions and tantalizingly suggestive of lipid transfer at that contact site (although hard to get strong evidence for that at this stage). This provides yet more evidence for the conservation and overall importance to cells of lipid transfer at contact sites, as well as reminding us of the ability of parasites to attack every aspect of cell 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 #1

Evidence, reproducibility and clarity

Summary:

The authors report a Coxiella burnetii effector protein, CbEPF1, which associate with lipid droplets (LD) at points of contact with the endoplasmic reticulum (ER). The presence of two FFAT motifs in CbEPF1 mediates CbEPF1 interaction with the ER protein VAP, LD-ER interaction, and an increase in LD size. Based on these results the authors put forward a model by which translocation of CbEPF1 into the host cell cytosol results in regulation of host cell lipid metabolism via the formation of ER-LD contact. The study relies heavily on fluorescence microscopy of ectopic (over)expression of CbEPF1 in eukaryotic cells.

Major comments:

CbEPF1 and ER-LD contact:

The immunofluorescence analysis of cell ectopically expressing CbEPF1-GFP provides convincing evidence that CbEPF1-GFP associates with LD and that CbEPF1-positive LD are positive for ER markers such as BFP-KDEL and VAPB (Fig 1&2). However, the study would benefit from looking at endogenous proteins to rule out any potential overexpression artefacts. Does endogenous VAP localize to CbEPF1-positive LD? Does CbEPF1 expressed from Coxiella (endogenous protein, or at least a tagged protein expressed from the bacteria) localize to LD? While the immunofluorescence images are of very high quality and convincing, electron microscopy is necessary to ascertain that membrane contacts between LD and the ER are induced upon CbEPF1 overexpression. CbEPF1 interaction with the ER is linked to the FFAT motifs (see below); what is known about CbEPF1 association with LD? Can a specific CbEPF1 domain be identified? In other words, does CbEPF1 contains 2 distinct membrane targeting domains that confer specificity to each of the contacting organelle (ER and LD in the present case) and thereby resemble other contact site localizing proteins.

CbEPF1 FFAT motifs and VAP binding:

The similarity of the sequence of the CbEPF1 FFAT motifs to the canonical sequence (Fig 1A), combined with data with alanine substitution mutation of the essential residue in position 2 of the FFAT motifs (Fig 3, 4), strongly support that CbEPF1 contains 2 functional FFAT motifs that confer VAP binding. FFAT motifs can mediate binding to VAPA, VAPB, and/or MOSPD2. Moreover, in addition to forming homodimer, VAPA, VAPB, and MOSPD2 can form heterodimers. What is the rationale for using VAPB (Fig 3CDE)? Do VAPA and/or MOSPD2 also yield positive results using the assays performed in Fig 3CDE, or is the CbEPF1-VAP interaction specific to VAPB? In the same line, in Fig 3E, is it possible that the CbEPF1-MOSPD2 is indirect and due to VAP- MOSPD2 interaction? Regarding the two FFAT motifs: are the two FFAT motifs redundant or synergistic? Although the data is not quantified, Figure 3E suggest a synergistic effect for VAP binding, however most IF data suggest redundancy. On the other end, the second FFAT motif seems necessary for MOSPD2 biding. Overall, clarifying CbEPF1 binding specificity towards a VAP/MOSPD2 and the role of each FFAT motif in this process could elevate the study by providing mechanistic insights into the hijacking of ER-LD contact sites by Coxiella and comparing and contrasting with the formation of ER-LD in naïve cells and/or the hijacking of VAP-dependent contact by other microbial pathogens.

CbEPF1-mediated LD clustering in the absence of VAP binding

A CbEPF1 FFAT motif mutant (F3mt) associates with LD that do not associate with the ER marker KDEL, and causes LD clustering (Fig 4). The authors speculate that the lack of LD-ER interaction results in LD-LD interaction, potentially via interaction of unidentified protein(s) on the LD surface. What is the biological relevance of the LD clustering phenotype? What is known about the role of LD clustering vs ER-LD contact, in the context of lipid metabolism? Could mechanistic characterization of this phenomenon provide insights in LD biology and/or the role of CbEPF1/ER-LD contacts in the context of Coxiella infection?

CbEPF1-mediated increase in LD number and size

CbEPF1-GFP overexpression result in an increase in the number of LD per cell independently of the FFAT motifs (Fig 5A). CbEPF1-GFP overexpression also result in an increase in LD diameter; however, this phenotype is dependent on wild-type FFAT motifs (Fig 5B). Quantification and corresponding statistical analysis support these conclusions. However, the representative images are not necessary in line with the bar graphs. For example, there appear to be less LD upon expression of F1mt or F2mt, compared to WT. Additionally, the increase in size is moderate and hard to appreciate in the IF images. It is also unclear, if/how an increase in LD number and/or size is biologically relevant in the context of Coxiella infection. Regarding potential mechanism(s), it is also unclear how CbEPF1 is promoting an increase in LD number. De novo LD production at the ER is unlikely given that the FFAT motifs, and therefore ER association, are not required. What would an alternative mechanism be, and can it be experimentally tested? Regarding the increase in LD size, the author suggest that the phenotype could be due to impaired lipid transfer from the ER to LD. This is an interesting model. Do the authors envision that CbEPF1 is a lipid transfer protein and/or act on ER-LD associated lipid transfer? Can either be experimentally tested? Altogether, strengthening this aspect of the study would clarify the proposed model and significantly increase the impact of the study.

Functional relevance:

One aspect that is not addressed in the study, is what are the benefit(s), if any, of CbEPF1 translocation into the host cytosol and targeting to ER-LD contact? This could be addressed by assessing the phenotype of a Coxiella CbEPF1 mutant? On the flip side, is VAP required for Coxiella intracellular growth/vacuole maturation? Another avenue would be to investigate if CbEPF1 affects the lipid composition of the CCV. The present study suggest that LD may be important for Coxiella intracellular life cycle. Are LD formation induced upon infection? Are ER-LD contact increased upon infection? One may also expect that inhibition of LD formation would affect bacterial replication and that stimulation may promote growth/vacuole maturation? Any experiments addressing the biological relevance of the present findings in the context of Coxiella infection would tremendously increase the impact of the study.

Minor comments:

None. The manuscript is well written and easy to follow.

Significance

Inter-organelle communication through the formation of membrane contact between two apposing organelles is critical to maintain host cell homeostasis via the transfer of small molecules such as lipid and ions. Bacterial and viral pathogens have been shown to manipulate proteins that localize to cellular membrane contact and to promote membrane contact between their intracellular vacuole and the ER. Both viral and bacterial proteins that contains FFAT motifs and interact with VAP have been described in the context of tethering of the ER to the membrane compartment in which the pathogen replicate. The present study stands out by the characterization of a FFAT motifs-containing bacterial effector protein that targets cellular contact, ER-LD more specifically. The significance is 2-fold. It expands the list of FFAT-motif containing proteins in pathogens, reinforcing the idea that VAP/membrane contact may be a universal mechanism of host-pathogen interaction, which will be of interest to those studying host-microbe interaction in basic or translational research settings. The study also has the potential to move forward research on LD biology and LD-ER contact, which will be of interest to cell biologists in general, and to the evergrowing membrane contact community, specifically.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

We thank the reviewers for their valuable comments, which definitely make our story stronger.

2. Description of the planned revisions

Reviewer 1

Comments:

No data are shown from the genome-wide screening approach, including the common regulators of KRAS and HRAS. Information about how imaging data were processed and analysed is missing. A final table of 8 selected factors with phosphatase activity is presented without providing further insight about the selection criteria and other factors.

This information will be included in the revised manuscript. In the subsequent characterization via image-based quantification of GFP-KRAS membrane localization, a Manders´ coefficient was calculated. A respective chapter in the methods section on how this was done is missing.

This information will be provided in the revised manuscript. I would be happy to see the following analyses to strengthen the dataset:

• Reconstitution experiments and further validation to show that it is dependent on the enzymatic activity of MTMRs.

MTMR3 knockdown (KD) cells will be rescued with wildtype (WT) MTMR3 or the phosphatase mutant MTMR3 (C413S, PMID: 11676921). MTMR4 KD cells will be rescued with WT MTMR4 or the phosphatase mutant MTMR4 (C407S, PMID: 20736309). In these cells, the PM localization of KRAS and PtdSer will be examined by confocal and electron microscopy. - Additive effect upon depletion of multiple MTMRs? Are they functionally co-operative?

MTMR3 and 4 KD cells will be rescued with WT MTMR4 and 3, respectively, and the PM localization of KRAS and PtdSer will be examined by confocal and electron microscopy. - Signalling analysis is very limited (Fig. 5). Do the authors detect any defects in K-RAS driven downstream signaling in these cells upon depletion of MTMRs.

Human pancreatic ductal adenocarcinoma (PDAC) cell lines that harbor oncogenic mutant KRAS and their growth is KRAS signaling-dependent (MiaPaCa2 and AsPC1), and PDAC cell line harboring WT KRAS and their growth is KRAS signaling-independent (BxPC3) will be infected with lentivirus expressing shRNAs against MTMR 2, 3, 4 or 7. Their growth (proliferation assay) and KRAS signaling (e.g. phosphorylated ERK and Akt by immunoblot) will be measure. Reviewer 2

Major comments

The unbiased siRNA screen used to identify proteins that impact KRAS membrane localization was a very nice approach to identify MTMR proteins. Although there is a clear phenotype of KRAS mislocalization associated with knockdown of the various MTMR proteins, the data provided does not prove a causational role for the MTMR proteins in maintaining PtdSer content, nor KRAS localization, at the PM. The current data does not provide a mechanism by which MTMR proteins are influencing this process, but rather speculates using existing literature that it is the loss in MTMR 3-phosphotase activity that leads to decreased PtdSer in the membrane. There is a series of conversions and exchanges that act upon PI3P (the substrate of MTMR proteins) and PI to generate PtdSer in the PM; thus, it is a dynamic process that is influenced by a variety of different proteins and transporters [3, 4, 5, 6]. To prove their single-protein-driven hypothesis, the authors should clone and express a mutant MTMR protein construct that contains an inactive phosphatase catalytic domain, to prove that it is indeed MTMR's generation of PI (which is further converted into PI4P) in the membrane that is responsible for maintaining PtdSer content and KRAS localization. Without this, there is not enough evidence to support this claim.

MTMR3 knockdown (KD) cells will be rescued with wildtype (WT) MTMR3 or the phosphatase mutant MTMR3 (C413S, PMID: 11676921). MTMR4 KD cells will be rescued with WT MTMR4 or the phosphatase mutant MTMR4 (C407S, PMID: 20736309). In these cells, the PM localization of KRAS and PtdSer will be examined by confocal and electron microscopy. In addition, the authors speculate that ORP5 is a critical intermediate in this process, and that the loss in PI4P/ORP5 at the PM following MTMR knockdown is responsible for the decrease in PtdSer at the PM. The authors should knockdown ORP5 in MTMR-wildtype cells, since it is downstream of their proposed mechanism, and see whether this leads to comparable reductions in PtdSer levels and KRAS mislocalization at the PM. This would confirm ORP5 as having a major role in this setting and would support the initial mechanistic hypothesis. These experiments are imperative to forming an appropriate conclusion, especially since some of their current data contradicts their mechanistic hypothesis: the authors identify a decrease in whole cell PtdSer content, not just PM PtdSer content, when MTMR proteins are knocked down. Based on this result, one would predict that a secondary or supporting mechanism must exist that contributes to a reduction in whole cell PtdSer content, which likely contributes to its loss at the PM as well. The authors describe in line 360 how "previous work has shown that PM PI4P depletion indirectly blocks PtdSer synthase 1 and 2 activities," to explain this reduction in total cell levels of PtdSer. The authors should look at PtdSer synthase 1 and 2 activities in the presence of MTMR knockdown, as the loss in PtdSer at the PM may rely more heavily on synthase activity than ORP-dependent transfer of PtdSer.

Investigating the PM localization of KRAS and PtdSer after silencing ORP5 in MTMR WT mammalian cell lines has been published (PMID: 31451509 and 34903667). In these studies, silencing ORP5 1) reduces the levels of PtdSer and KRAS from the plasma membrane (PM), 2) reduces KRAS signal output, 3) blocks the growth of KRAS-dependent PDAC in vitro and in vivo. These studies have been appropriately cited in our manuscript in lines 82 and 277. Although the c. Elegans model that was used to investigate downstream let-60 (RAS ortholog) activity through a multi-vulva phenotype is quite intriguing, it is more critical to assess downstream RAS pathway activation, especially in the human colorectal adenocarcinoma or the human mammary gland ductal carcinoma cell lines. Not only would this line of questioning provide a higher significance and increase the clinical applicability of these findings, but it is also crucial to support the author's claim that MTMR knockdown can influence mutant KRAS activity. Although small changes in KRAS localization to the PM can have significant effects on downstream signaling, these effects need to be measured and confirmed in this setting. The authors should perform western blots to assess the activation of both the PI3K and MAPK pathway in the MTMR knockdown cell lines.

Human pancreatic ductal adenocarcinoma (PDAC) cell lines that harbor oncogenic mutant KRAS and their growth is KRAS signaling-dependent (MiaPaCa2 and AsPC1), and PDAC cell line harboring WT KRAS and their growth is KRAS signaling-independent (BxPC3) will be infected with lentivirus expressing shRNAs against MTMR 2, 3, 4 or 7. Their growth (proliferation assay) and KRAS signaling (e.g. phosphorylated ERK and Akt by immunoblot) will be measure. In addition to this, it might be important to know whether there are any changes in the levels of the KRAS protein itself, as recycling/transport pathways may be impacted by its lack of recruitment to the plasma membrane.

Total KRAS protein expression will be measured in MTMR KD cell lines. Finally, the authors show that proliferation is inhibited by MTMR knockdown as a readout of RAS activity. The authors should also assess the levels of cell death, as the inhibition of mutant KRAS in cancer cells would likely lead to cell death. The authors do not describe why reducing any one of the MTMR proteins alone is sufficient to deplete the PM of PtdSer. This sort of discussion is important for understanding compensatory or regulatory mechanisms in place between the MTMR proteins, as this may influence PtdSer levels at the PM. For example, it has been shown that MTMR2 can stabilize MTMR13 on membranes. Do the levels, stability, or localization of the other MTMR proteins change when one specific MTMR is knocked down? Is this why we see an effect on PtdSer in any one of the knockdowns? The authors should at the very least provide western blots for each of the MTMR proteins discussed in the presence of each individual MTMR knockdown.

MTMR3 knockdown (KD) cells will be rescued with WT MTMR3 or the phosphatase mutant MTMR3 (C413S, PMID: 11676921). MTMR4 KD cells will be rescued with WT MTMR4 or the phosphatase mutant MTMR4 (C407S, PMID: 20736309). In these cells, the PM localization of KRAS and PtdSer will be examined by confocal and electron microscopy. In addition, we will measure endogenous MTMR 2/3/4/7 proteins levels in the presence of each individual MTMR KD by immunoblotting. In addition to the above experiments, the MTMR hairpins should be expressed in a secondary or tertiary cell line to prove that these events are not specific to the current model used. Since their current human mammary gland ductal carcinoma cell line overexpresses a mutant KRAS-GFP construct, perhaps doing similar experiments in a cancer cell line that already expresses an endogenous mutant KRAS might provide a better model.

Human pancreatic ductal adenocarcinoma (PDAC) cell lines that harbor oncogenic mutant KRAS and their growth is KRAS signaling-dependent (MiaPaCa2 and AsPC1), and PDAC cell line harboring WT KRAS and their growth is KRAS signaling-independent (BxPC3) will be infected with lentivirus expressing shRNAs against MTMR 2, 3, 4 or 7. Their growth (proliferation assay) and KRAS signaling (e.g. phosphorylated ERK and Akt by immunoblot) will be measure. Although this protein would not include a GFP-tag, other ways of visualizing its localization at the PM (such as immunofluorescent staining) could be used to confirm its localization there.

The anti-KRAS antibody for IF has not been reported to my knowledge. In addition, the effects on downstream RAS signaling could be measured through western blot of PI3K and MAPK pathways.

Human pancreatic ductal adenocarcinoma (PDAC) cell lines that harbor oncogenic mutant KRAS and their growth is KRAS signaling-dependent (MiaPaCa2 and AsPC1), and PDAC cell line harboring WT KRAS and their growth is KRAS signaling-independent (BxPC3) will be infected with lentivirus expressing shRNAs against MTMR 2, 3, 4 or 7. Their growth (proliferation assay) and KRAS signaling (e.g. phosphorylated ERK and Akt by immunoblot) will be measure. Supplemental Figure 4 is incorrectly referred to in the text as Supplemental Figure 3 (line 257-258). The text reads, "Confocal microscopy further demonstrates that HRASG12V cellular localization is not disrupted after silencing MTMR 2/3/4/7 (Fig. S3)" but Figure S3 is an EM image of PM basal sheets from T47D cells expressing GFP-KRASG12V. Supplemental Figure 4 shows that mutant HRAS is unaffected by the various MTMR knockdowns.

They will be labeled correctly in the revised manuscript. Since the authors show decreased proliferation in mutant KRAS cells following MTMR knockdown, the authors should also investigate any changes to proliferation rates in mutant HRAS cell lines following MTMR knockdown. This data is necessary to prove that MTMR-driven changes in downstream RAS signaling are specific to mutant KRAS and not mutant HRAS.

Cell proliferation assay will be performed using MTMR 2/3/4/7-silenced T47D cell lines stably expressing oncogenic mutant HRAS (HRASG12V) to address this questions. It may also be important for the authors to also show any effects on wildtype RAS localization to the PM when MTMR-2,-3,-4, and -7 are knocked down, to show whether this is a oncoprotein-specific event.

Cells expressing the truncated mutant KRAS, which contains the minimal membrane anchor and does not have G-domain will be infected with lentivirus expressing shRNA against MTMR 2/3/4/7, and their localization will be examined. The representative images chosen for Figure 4 diminish the reliability of the data, as it is difficult to see a visible change in the PI3P probe between the control and MTMR knockdown cells in these images. Since the authors rely on the Mander's coefficient and the number of gold particles throughout much of the paper, having the same conclusion quantitatively but not qualitatively for these assays is confusing. Perhaps the authors should elaborate on whether MTMR knockdown has a stronger effect on PtSer and KRAS PM presence than PI3P PM presence.

We will include the discussion in the revised manuscript. They should also describe their method for identifying early endosomes, since they switch back and forth between describing the content of the PM and of early endosomes, such as in Figure 1 and Figure 4.

We will include the information in the revised manuscript. Minor comments:

An additional experiment that may add another layer of clinical applicability would be the use of an MTMR inhibitor in this cell line, to see whether similar effects can be achieved pharmacologically [7]. This would provoke other researchers to investigate MTMR inhibitors in vitro and in vivo to assess the effect on mutant KRAS cancers.

• This is an important point, but while vanadate, a general phospho-tyrosine phosphatase (PTP) inhibitor, has been reported to inhibit myotubulin, a family member of MTMR (PMID: 8995372 and 1943774), there are no commercially available MTMR-specific inhibitors. Using vanadate to inhibit MTMR proteins will produce non-specific effects by blocking other PTPs. The inclusion of cell lines that express KRAS proteins of different mutational statuses would be extremely interesting, as KRAS' orientation within the plasma membrane has been shown to be altered by these mutations. This fact should potentially be considered when choosing a secondary or tertiary cell line to do additional experiments in, but it is not necessary for the authors to elaborate on how MTMR proteins may impact different KRAS mutants for the scope of this project.

For the aforementioned experiments using human KRAS-dependent and -independent PDAC cell lines, we will use MiaPaCa2 (KRASG12C) and AsPC1 (KRASG12D). Reviewer #3

*Major comments: *

One of the two main manuscript claims indicates that KRAS12V "function" is impaired upon MTMR knockdown. While this is an obvious phenotype expected by mislocalizing KRAS from the inner PM it is not sufficiently demonstrated in the current version of the manuscript. Western blots of at least MAPK and PI3K signalling following MTMR knockdown in KRAS-dependent cell lines should be included. In addition to the T47D cells used in the manuscript, it would be ideal to include a KRAS-mutant cell line from tumour types where KRAS mutations are more frequent that in breast.

• Human pancreatic ductal adenocarcinoma (PDAC) cell lines that harbor oncogenic mutant KRAS and their growth is KRAS signaling-dependent (MiaPaCa2 and AsPC1), and PDAC cell line harboring WT KRAS and their growth is KRAS signaling-independent (BxPC3) will be infected with lentivirus expressing shRNAs against MTMR 2, 3, 4 or 7. Their growth (proliferation assay) and KRAS signaling (e.g. phosphorylated ERK and Akt by immunoblot) will be measure. Since the MTMR dependent phenotypes are mutant-KRAS specific it would be interesting to study the resulting phenotypes in HRAS-mutant cell line.

Cell proliferation assay will be performed using MTMR 2/3/4/7-silenced T47D cell lines stably expressing oncogenic mutant HRAS (HRASG12V) to address these questions.

**Referee cross-commenting**

After reading the reviews of my colleagues I think there is a clear agreement on the need to further substantiate that KRAS membrane mis-localization is indeed affecting oncogenic output. The use of other KRAS addicted and non-addicted models would further enhance this analysis.

Likewise, the other two reviewers request experimental evidences to validate the role of MTMR enzymatic activity in the process. This is a pertinent request that I failed to put forward. Suggestions include the use of reconstitution experiments catalytically dead mutants. Also, the use of MTMR small molecule inhibitors is proposed. If those exist with sufficient specificity this would indeed be appropriate to perform.

Experiments addressing these comments have been described above.

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

N/A

• *

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. *

Reviewer 2

R2 suggests to investigate PtdSer synthase 1 and 2 activities in presence of MTMR knockdown, as the loss in PtdSer at the PM may rely more heavily on synthase activity than ORP-dependent transfer of PtdSer.

Although it is intriguing to examine the effect of MTMR loss on the activities of PtdSer synthase 1 and 2, our lab does not have resources/techniques to carry out the experiment. * *

The results of this paper rely heavily on one experimental technique, which is calculating a Mander's coefficient and counting the co-localization of the probe of interest with the CellMask stain of the plasma membrane. How this coefficient is derived is explained in appropriate detail in the methods section of this manuscript; however, a secondary route of identifying these changes in membrane constituents would greatly enhance the paper's conclusions. This would eliminate any doubt surrounding the accuracy of the technique, since so much of the data relies on one experimental output.

In addition to Manders' coefficient for examining the colocalization of KRAS and LactC2 (the PtdSer probe) to propose KRAS/PS redistribution to endomembranes after MTMR loss. To complement this, we also performed quantitative EM to demonstrate the PM depletion of KRAS and PtdSer from the inner PM leaflet. We believe these two techniques would appropriate to investigate KRAS/PtdSer PM depletion and cellular re-distribution. * *

Reviewer 3

To further support the conclusions, oncogenic signalling should be studied in the C.elegans model by immunofluorescence of immunohistochemistry. Furthermore, although not strictly required to support the author's claims, it would be interesting to elucidate whether the inhibition of the multivulva phenotype upon MTMR knockdown in vivo results as a consequence of cell death.

Our collaborator for C. elegans study does not have resources to carry out the proposed IF and IHC experiment. Instead, we will measure KRAS signaling (e.g. phosphorylated ERK and Akt by immunoblot) and the growth of KRAS-dependent PDAC after MTMR loss. These experiments would be more clinically and physiologically relevant.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

In recent years, various genetic and pharmacological studies have clearly demonstrated a causal relationship between phosphatidylserine (PtdSer) distribution at the plasma membrane (PM) and KRAS clustering at the inner leaflet. In this manuscript, Henkels and colleagues have performed a high-content genome wide siRNA screen in search of hits that upon knockdown would result in membrane mislocalization of an exogenous KRAS12V-GFP fusion while not affecting HRASV12 membrane distribution. The results identified 4 of the 14 human members of the myotubularin-related (MTMR) protein family. Individual knockdown of MTMR2, 3, 4 and 7 resulted in specific relocalization of KRAS12V (and not of HRAS12V) from the cell membrane to endomembranes assessed both by confocal and electron microscopy. The MTMRs display enzymatic activity (3-phosphatase activity towards PI3P and PI(3,5)P2) controlling membrane trafficking. Given the known dependency of KRAS PM clustering on PtdSer, the authors showed that KRAS12V mislocalization upon MTMR depletion is due to an overall reduction of PtdSer accompanied by depletion of inner PM PtdSer. Using lipid-specific probes the authors went on to demonstrate that this phenotype occurs as a combined reduction of inner PM PI4P levels and concomitant elevation of PM PI3P. As expected, KRAS12V inner PM mislocalization affected oncogenic function. This is shown by a 50% reduction in cell proliferation of KRAS-transformed T47D (human mammary gland ductal carcinoma) cell line. More convincingly, si-RNA mediated depletion of the C.elegans MTMR3 and 7 orthologs potently reduces a KRAS-dependent multivulva phenotype.

Overall, I find that the experimental part of the manuscript is satisfactory. Yet, the overall conclusion is that inactivation of a subset of MTMR phosphatases reduces KRAS PM localization and KRAS signalling. While changes of KRAS inner PM are well documented, there is not a single experiment demonstrating that this results in reduced oncogenic output. This needs to be further documented if a mention to KRAS function is included in the title.

Major comments:

1. One of the two main manuscript claims indicates that KRAS12V "function" is impaired upon MTMR knockdown. While this is an obvious phenotype expected by mislocalizing KRAS from the inner PM it is not sufficiently demonstrated in the current version of the manuscript. Western blots of at least MAPK and PI3K signalling following MTMR knockdown in KRAS-dependent cell lines should be included. In addition to the T47D cells used in the manuscript, it would be ideal to include a KRAS-mutant cell line from tumour types where KRAS mutations are more frequent that in breast.
2. Since the MTMR dependent phenotypes are mutant-KRAS specific it would be interesting to study the resulting phenotypes in HRAS-mutant cell line.
3. To further support the conclusions, oncogenic signalling should be studied in the C.elegans model by immunofluorescence of immunohistochemistry. Furthermore, although not strictly required to support the author's claims, it would be interesting to elucidate whether the inhibition of the multivulva phenotype upon MTMR knockdown in vivo results as a consequence of cell death.

Referee cross-commenting

After reading the reviews of my colleagues I think there is a clear agreement on the need to further substantiate that KRAS membrane mis-localization is indeed affecting oncogenic output. The use of other KRAS addicted and non-addicted models would further enhance this analysis. Likewise, the other two reviewers request experimental evidences to validate the role of MTMR enzymatic activity in the process. This is a pertinent request that I failed to put forward. Suggestions include the use of reconstitution experiments catalytically dead mutants. Also, the use of MTMR small molecule inhibitors is proposed. If those exist with sufficient specificity this would indeed be appropriate to perform.

Significance

In spite of its importance for oncogenic function, KRAS cellular trafficking remains one of the least studied processes. As such, reports like the current work are important to increase our biological knowledge. Furthermore, this increased biological understanding could identify vulnerabilities with future therapeutic potential.

It is known, mainly from previous work of one of the co-authors (John Hancock), that a PtdSer interplay with oxysterol-binding protein related proteins ORP5 and 8 regulate KRAS membrane distribution. The current study describes a further layer of control depending on MTMR phosphatases. In my opinion the cellular phenotypes are properly addressed, but not the phenotypic consequences on KRAS-signalling.

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:

Henkels et al. propose the role of myotubularin-related proteins in promoting KRAS4B localization to the plasma membrane. Their data shows that shRNA-mediated knockdown of myotubularin-related proteins -2, -3, -4, or -7 led to measurable changes in RAS localization and in plasma membrane (PM) content. More specifically, knockdown of any one of these MTMR proteins led to a decrease in PI4P levels in the PM, an increase in PI3P content in the PM, a decrease in phosphatidyl serine (PtdSer) in the PM/whole cell, and a decrease in mutant KRAS localization to the PM. Their data also shows a decreased presence of ORP5 at the PM, a protein which is responsible for the exchange of PI4P in the plasma membrane for PtdSer in the endoplasmic reticulum. These results are somewhat predictable and are supported by the existing literature, as MTMR proteins are known to exhibit 3-phosphotase activity towards PI3P to generate PI (a precursor to PI4P), PI4P is known to recruit ORP5, and ORP5 is known to contribute to PtdSer content in the membrane [1, 2]. Regardless, the authors find that the individual knockdown of MTMR proteins is sufficient to cause measurable changes in PM content and mislocalization of mutant KRAS4B. Thus, despite the fact that many proteins are involved in regulating PM content, such as PI4KA, PtdSer synthase 1 and 2, Nir2/3, and PITPs, Henkels et al. speculate that MTMR proteins are the primary regulators of PtdSer PM levels [3, 4, 5, 6]. The authors propose that the loss of function in any one of these MTMR proteins alone is sufficient to cause significant changes in PM content through an ORP5-dependent process, and that this ultimately leads to a decrease in mutant KRAS signaling.

Major comments:

The unbiased siRNA screen used to identify proteins that impact KRAS membrane localization was a very nice approach to identify MTMR proteins. Although there is a clear phenotype of KRAS mislocalization associated with knockdown of the various MTMR proteins, the data provided does not prove a causational role for the MTMR proteins in maintaining PtdSer content, nor KRAS localization, at the PM. The current data does not provide a mechanism by which MTMR proteins are influencing this process, but rather speculates using existing literature that it is the loss in MTMR 3-phosphotase activity that leads to decreased PtdSer in the membrane. There is a series of conversions and exchanges that act upon PI3P (the substrate of MTMR proteins) and PI to generate PtdSer in the PM; thus, it is a dynamic process that is influenced by a variety of different proteins and transporters [3, 4, 5, 6]. To prove their single-protein-driven hypothesis, the authors should clone and express a mutant MTMR protein construct that contains an inactive phosphatase catalytic domain, to prove that it is indeed MTMR's generation of PI (which is further converted into PI4P) in the membrane that is responsible for maintaining PtdSer content and KRAS localization. Without this, there is not enough evidence to support this claim. In addition, the authors speculate that ORP5 is a critical intermediate in this process, and that the loss in PI4P/ORP5 at the PM following MTMR knockdown is responsible for the decrease in PtdSer at the PM. The authors should knockdown ORP5 in MTMR-wildtype cells, since it is downstream of their proposed mechanism, and see whether this leads to comparable reductions in PtdSer levels and KRAS mislocalization at the PM. This would confirm ORP5 as having a major role in this setting and would support the initial mechanistic hypothesis. These experiments are imperative to forming an appropriate conclusion, especially since some of their current data contradicts their mechanistic hypothesis: the authors identify a decrease in whole cell PtdSer content, not just PM PtdSer content, when MTMR proteins are knocked down. Based on this result, one would predict that a secondary or supporting mechanism must exist that contributes to a reduction in whole cell PtdSer content, which likely contributes to its loss at the PM as well. The authors describe in line 360 how "previous work has shown that PM PI4P depletion indirectly blocks PtdSer synthase 1 and 2 activities," to explain this reduction in total cell levels of PtdSer. The authors should look at PtdSer synthase 1 and 2 activities in the presence of MTMR knockdown, as the loss in PtdSer at the PM may rely more heavily on synthase activity than ORP-dependent transfer of PtdSer. Although the c. Elegans model that was used to investigate downstream let-60 (RAS ortholog) activity through a multi-vulva phenotype is quite intriguing, it is more critical to assess downstream RAS pathway activation, especially in the human colorectal adenocarcinoma or the human mammary gland ductal carcinoma cell lines. Not only would this line of questioning provide a higher significance and increase the clinical applicability of these findings, but it is also crucial to support the author's claim that MTMR knockdown can influence mutant KRAS activity. Although small changes in KRAS localization to the PM can have significant effects on downstream signaling, these effects need to be measured and confirmed in this setting. The authors should perform western blots to assess the activation of both the PI3K and MAPK pathway in the MTMR knockdown cell lines. In addition to this, it might be important to know whether there are any changes in the levels of the KRAS protein itself, as recycling/transport pathways may be impacted by its lack of recruitment to the plasma membrane. Finally, the authors show that proliferation is inhibited by MTMR knockdown as a readout of RAS activity. The authors should also assess the levels of cell death, as the inhibition of mutant KRAS in cancer cells would likely lead to cell death. The authors do not describe why reducing any one of the MTMR proteins alone is sufficient to deplete the PM of PtdSer. This sort of discussion is important for understanding compensatory or regulatory mechanisms in place between the MTMR proteins, as this may influence PtdSer levels at the PM. For example, it has been shown that MTMR2 can stabilize MTMR13 on membranes. Do the levels, stability, or localization of the other MTMR proteins change when one specific MTMR is knocked down? Is this why we see an effect on PtdSer in any one of the knockdowns? The authors should at the very least provide western blots for each of the MTMR proteins discussed in the presence of each individual MTMR knockdown.<br /> In addition to the above experiments, the MTMR hairpins should be expressed in a secondary or tertiary cell line to prove that these events are not specific to the current model used. Since their current human mammary gland ductal carcinoma cell line overexpresses a mutant KRAS-GFP construct, perhaps doing similar experiments in a cancer cell line that already expresses an endogenous mutant KRAS might provide a better model. Although this protein would not include a GFP-tag, other ways of visualizing its localization at the PM (such as immunofluorescent staining) could be used to confirm its localization there. In addition, the effects on downstream RAS signaling could be measured through western blot of PI3K and MAPK pathways. Supplemental Figure 4 is incorrectly referred to in the text as Supplemental Figure 3 (line 257-258). The text reads, "Confocal microscopy further demonstrates that HRASG12V cellular localization is not disrupted after silencing MTMR 2/3/4/7 (Fig. S3)" but Figure S3 is an EM image of PM basal sheets from T47D cells expressing GFP-KRASG12V. Supplemental Figure 4 shows that mutant HRAS is unaffected by the various MTMR knockdowns. Since the authors show decreased proliferation in mutant KRAS cells following MTMR knockdown, the authors should also investigate any changes to proliferation rates in mutant HRAS cell lines following MTMR knockdown. This data is necessary to prove that MTMR-driven changes in downstream RAS signaling are specific to mutant KRAS and not mutant HRAS. It may also be important for the authors to also show any effects on wildtype RAS localization to the PM when MTMR-2,-3,-4, and -7 are knocked down, to show whether this is a oncoprotein-specific event. <br /> The representative images chosen for Figure 4 diminish the reliability of the data, as it is difficult to see a visible change in the PI3P probe between the control and MTMR knockdown cells in these images. Since the authors rely on the Mander's coefficient and the number of gold particles throughout much of the paper, having the same conclusion quantitatively but not qualitatively for these assays is confusing. Perhaps the authors should elaborate on whether MTMR knockdown has a stronger effect on PtSer and KRAS PM presence than PI3P PM presence. They should also describe their method for identifying early endosomes, since they switch back and forth between describing the content of the PM and of early endosomes, such as in Figure 1 and Figure 4.

Minor comments:

An additional experiment that may add another layer of clinical applicability would be the use of an MTMR inhibitor in this cell line, to see whether similar effects can be achieved pharmacologically [7]. This would provoke other researchers to investigate MTMR inhibitors in vitro and in vivo to assess the effect on mutant KRAS cancers.

The inclusion of cell lines that express KRAS proteins of different mutational statuses would be extremely interesting, as KRAS' orientation within the plasma membrane has been shown to be altered by these mutations. This fact should potentially be considered when choosing a secondary or tertiary cell line to do additional experiments in, but it is not necessary for the authors to elaborate on how MTMR proteins may impact different KRAS mutants for the scope of this project.

The results of this paper rely heavily on one experimental technique, which is calculating a Mander's coefficient and counting the co-localization of the probe of interest with the CellMask stain of the plasma membrane. How this coefficient is derived is explained in appropriate detail in the methods section of this manuscript; however, a secondary route of identifying these changes in membrane constituents would greatly enhance the paper's conclusions. This would eliminate any doubt surrounding the accuracy of the technique, since so much of the data relies on one experimental output.

References

1. Clague MJ, Lorenzo O. The myotubularin family of lipid phosphatases. Traffic. (12):1063-9 (2005).
2. Chung J, Torta F, Masai K, Lucast L, Czapla H, Tanner LB, Narayanaswamy P, Wenk MR, Nakatsu F, De Camilli P. PI4P/phosphatidylserine countertransport at ORP5- and ORP8-mediated ER-plasma membrane contacts. Science. 349(6246):428-32 (2015).
3. Kim YJ, Guzman-Hernandez ML, Wisniewski E, Balla T. Phosphatidylinositol-Phosphatidic Acid Exchange by Nir2 at ER-PM Contact Sites Maintains Phosphoinositide Signaling Competence. Dev Cell 33: 549-561 (2015).
4. Balla A, Balla T. Phosphatidylinositol 4-kinases: Old enzymes with emerging functions. Trends Cell Biol 16, 351-361 (2006).
5. Arikketh D, Nelson R, Vance JE. Defining the importance of phosphatidylserine synthase-1 (PSS1): unexpected viability of PSS1-deficient mice. J Biol Chem. 283(19):12888-97 (2008).
6. Cockcroft S. The diverse functions of phosphatidylinositol transfer proteins. Curr Top Microbiol Immunol. 362:185-208 (2012).
7. Taylor GS, Maehama T, Dixon JE. Myotubularin, a protein tyrosine phosphatase mutated in myotubular myopathy, dephosphorylates the lipid second messenger, phosphatidylinositol 3-phosphate. Proc Natl Acad Sci U S A. 1;97(16):8910-5 (2000).

Significance

The significance of this paper lies in providing the field with an additional regulator of KRAS localization at the PM, as this is localization is critical to KRAS function. Despite three decades worth of understanding and even successfully blocking KRAS membrane localization in vitro, no KRAS-membrane-localization inhibitors have been approved for the clinic. Thus, there is still room in the field for the development of a safe therapeutic target that can effectively block this process. There is a consensus in the literature that PtdSer is critical for KRAS anchoring to the membrane, and this paper describes how MTMR proteins may impact the supply of PtdSer to the PM. Since this work is done in a cancer background by utilizing a mutant KRAS construct (KRASG12V), this work would be interesting to many cancer researchers that are attempting to target mutant KRAS. This paper would also be interesting to researchers who investigate mechanisms of PM maintenance.

Our lab studies RAS signaling in tumorigenesis. The authors are clear in their explanations of the mechanisms of PM maintenance and PM components relevant to this study.

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 their manuscript with the title "Myotubularin-related proteins regulate KRAS function by controlling plasma membrane levels of polyphosphoinositides and phosphatidylserine", Henkels and colleagues describe the identification and characterization of factors modulating the plasma-membrane localization of KRASG12V, required for its activation. In an siRNA-based screening approach, they identify four members of the myotubularin-related (MTMR) protein family, namely MTMR 2, 3, 4 and 7, when downregulated result in impaired localization of KRAS to the plasma membrane.

Validation was performed via confocal microscopy in cells overexpressing GFP-RASG12V, stained with the membrane marker CellMask and with gold labelling-Electron microscopy. Co-expression of GFP-LactC2, a well-established marker for PtdSer, and subsequent EM analysis revealed a reduction of PtdSer at the PM upon MTMR depletion. This observation was further validated by whole cell lipidomics, showing a significant reduction in total cellular PtdSer content in MTMR 2/3/4/7 KD conditions.Reduction of PI4P at the PM of PI4P and increase in overall (and PM) levels of PI3P - measured by overexpression of fluorescently tagged marker proteins for the respective phospholipid as well as EM.inally, to investigate the effect of MTMR knockdown on RAS signalling, the authors used transformed T47D cells as well as a C. elegans model system. In both systems, RAS signalling was found to be impaired upon MTMR depletion.

Overall, the authors convincingly present MTMR proteins as regulators of KRAS plasma membrane localization. Upon depletion of MTMR 2,3,4 and 7, they see KRAS mis-localizing away from the PM and KRAS signalling being disrupted in cell culture and C. elegans model systems. The data are well presented and of high quality. Electron microscopy after immunogold labelling was used to provide quantitative data. The study can be further strengthened by uncovering the role of MTMR in KRAS driven pathobiology.

Please find some minor comments below

Comments:

• no data are shown from the genome-wide screening approach, including the common regulators of KRAS and HRAS. Information about how imaging data were processed and analysed is missing. A final table of 8 selected factors with phosphatase activity is presented without providing further insight about the selection criteria and other factors.
• in the subsequent characterization via image-based quantification of GFP-KRAS membrane localization, a Manders´ coefficient was calculated. A respective chapter in the methods section on how this was done is missing.

I would be happy to see the following analyses to strengthen the dataset:

• reconstitution experiments and further validation to show that it is dependent on the enzymatic activity of MTMRs
• additive effect upon depletion of multiple MTMRs? Are they functionally co-operative?
• signalling analysis is very limited (Fig. 5). Do the authors detect any defects in K-RAS driven downstream signaling in these cells upon depletion of MTMRs.

Significance

Very interesting and potentially important study. But needs further evidence on how MTMRs regulate pathobiology. Does MTMR depletion inhibit KRAS driven downs stream events needs to investigated here.

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

Learn more at Review Commons

---

Reply to the reviewers

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

__Summary __

Geng et al. explore the molecular mechanisms underlying the role of KIF1C in RNA transport, focusing on how it interacts with RNA. KIF1C is shown to form dynamic puncta when overexpressed in COS-7 cells that do not appear to colocalise with organelle markers. An IDR in the tail of the kinesin is necessary and sufficient for the formation of these structures and FRAP experiments show that they can exchange their contents with proteins in the cytosol and that their formation can can be reversibly modulated by hypotonic shock, consistent with LLPS. In vitro, the IDR and flanking regions can undergo phase separation at physiologically relevant concentrations and salt conditions. In cells, KIF1C puncta enrich for RNAs and support their transport, and depletion of RNA modulates KIF1C LLPS properties. A model is proposed whereby KIF1C mediated RNA transport to the cell periphery promotes the formation of a protein-RNA condensate that may act to fine tune local RNA activity.

__Major comments __

In general, the claims made here are well-supported by the data. However, I think that some exploration of the extent of LLPS at different KIF1C expression levels in cells is important but missing. The authors carefully estimate the endogenous concentration of KIF1C in COS-7 cells (at around 25 nm), but is isn't clear how this compares to that observed in transient transfection experiments. Although this is partly addressed in the vitro assays, I am still left with some questions over the extent of this phenomenon in a cellular context. Can the authors provide some experimental evidence to support the proposition that LLPS occurs (perhaps in a more localised fashion?, as Fig.9) at lower KIF1C expression levels? One way to address this might be a GFP-knock-in (although how feasible this is may depend on the genomic context), alternatively, the authors could generate cell lines that express KIF1C-GFP from a very weak promoter, demonstrate LLPS using their established assays, show that this is comparable to endogenous expression.

Response: We thank the reviewer for this suggestion. We have carried out additional experiments to explore the extent of KIF1C LLPS at endogenous levels. We used antibody against KIF1C to stain WT and KIF1C knockout (KO) cells. Although the antibody shows a high background of non-specific signal in the cytoplasm and nucleoplasm of both WT and KO cells, we were able to observe small puncta of KIF1C at the periphery of WT but not KO cells (new Figure 8). This finding supports our hypothesis that endogenous KIF1C undergoes LLPS upon reaching a high local concentration at the periphery of cells. Two lines of evidence support that these puncta of endogenous KIF1F protein are RNA-containing biomolecular condensates formed by LLPS (new Figure 8). First, these small puncta of endogenous KIF1C incorporate RAB13 mRNA, suggesting that they are RNA granules. Second, the puncta do not form in cells stably expressing KIF1C DIDR at near-endogenous levels.

Minor comments

Lines 107-109 and Figure 1B on localisation of other kinesin-3s. The authors state that they localise to certain organelle but don't show co-staining for those organelles.

Response: The localization of other kinesin-3s to certain organelles has been shown in the cited literature. In response to the reviewer's request, we now verify these findings by staining cells expressing the other kinesin-3s for specific organelles (new Figure S1 A).

Lines 172-183 and Figure 3. Evidence is provided through FRAP experiments that KIF1C puncta exchange with the cytosolic pool. However, the extent of recovery appears to saturate at Response: We agree that the data suggest the existence of an immobile pool of KIF1C within the condensates. We have added this information to the main text (lines 178-182). We note that these findings are consistent with recent studies demonstrating membrane-less organelles with at least partially solid-like properties, including nucleoli and stress granules as well as microtubule associated proteins (see references, reviewed in Van Treeck & Parker 2019).

Line 238 - Fig. S5C is cited as data on endogenous concentration of KIF1C - this should be Fig. S6C.

Response: Thank you. We have corrected this (now Fig S8 C).

Line 331-332 - I did not fully follow the logic here the RNAse A injection experiment supports the idea that KIF1C interaction with RNA is sequence selective. Could the authors expand on this.

Response: We thank the reviewer for this comment. We have rewritten the text (lines 235-238, 246-248).

__Reviewer #1 (Significance (Required)): __

This study introduces a new and exciting concept to motor protein biology: that some cytoskeletal motors and motor-cargo complexes can undergo phase separation, and that this is important for their function. The experiments are logical, progressive, and form a clear and compelling case. The main limitation is that demonstration of LLPS in cells is limited to over-expressed protein. Some exploration/demonstration of LLPS properties of KIF1C in cells at near to endogenous expression levels would enhance the study.

The work should be of interest to a broad range of readers, from the cytoskeletal motor community, those interested in mRNA regulation, as well as scientists studying phase separation more generally.

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

This paper investigates mRNA transport by the kinesin Kif1C and tests the hypothesis that liquid condensation of the disordered C terminal region is important for mRNA recruitment. It is based on prior work from other labs showing that Kif1C recruits and transports a set of mRNAs to the periphery of cells. The mechanism of the KifC1-mRNA interaction was not investigated in the prior work, so the proposal that a liquid condensate is involved is novel. It is also topical, since there is intense current interest in transport and regulation of mRNAs by condensate-mediated mechanisms. The most useful part of this paper to the field may be the identification of IDR2 as required for mRNA binding in Fig 7.

__Major comments __

A major concern is reliance on expression of tagged KifC1 in Cos cells in several figures. The expression level in these experimental probably far exceeds normal, though this comparison is not reported. It is possibly justified to use over-expression to reveal a condensate mechanism, but it is concerning and the authors needs to strongly qualify their conclusions. One way to moderate this concern would be to examine condensation as a function of expression level.

Response: We thank the reviewer for this suggestion. We have carried out additional experiments to explore the extent of KIF1C LLPS at endogenous levels. We used antibody against KIF1C to stain WT and KIF1C knockout (KO) cells. Although the antibody shows a high background of non-specific signal in the cytoplasm and nucleoplasm of both WT and KO cells, we were able to observe small puncta of KIF1C at the periphery of WT but not KO cells (new Figure 8). This finding supports our hypothesis that endogenous KIF1C undergoes LLPS upon reaching a high local concentration at the periphery of cells. Two lines of evidence support that these puncta of endogenous KIF1F protein are RNA-containing biomolecular condensates formed by LLPS (new Figure 8). First, these small puncta of endogenous KIF1C incorporate RAB13 mRNA, suggesting that they are RNA granules. Second, the puncta do not form in cells stably expressing KIF1C DIDR at near-endogenous levels.

Another significant concern is that the biochemical reconstitution figure tests protein alone, not protein + RNA. Disordered RNA binding proteins usually phase separate better in the presence of RNA. The best reconstitution papers evaluate specificity of RNA recruitment to condensates. Specificity testing in a reconstituted system may not be required for a first paper, but testing the effect of some kind of RNA seems important.

Response: The purified CC4+IDR and IDR constructs form condensates at low mM concentrations and in the absence of RNA or crowding agents, thus we did not test whether they would phase separate better in the presence of RNA. In response to the reviewer's comments, we now evaluate the specificity of RNA recruitment to the KIF1C condensates. We utilized the purified CC4+IDR protein and added the same GU-rich and polyA RNAs used in cells (now Fig 4 B) at different concentrations. Interestingly, there is selective incorporation of GU-rich oligos in condensates at low RNA concentrations, incorporation of both RNAs into condensates at medium concentrations, and an inhibition of condensate formation at high RNA concentrations (new Fig 7 E,F).

A final concern is that the specificity of mRNA recruitment to Kif1C puncta in cells is not critically evaluated. Among endogenous mRNAs, only one (Rab13) is tested. The paper would be stronger with a second positive mRNA and a negative control mRNA.

Response: We have now tested whether the specificity of mRNA recruitment to KIF1C puncta applies to additional mRNAs. We carried out single-molecule FISH (smFISH) experiments for two additional mRNAs. Based on the literature showing KIF1C-dependent localization of specific RNAs, we chose NET1 as a second positive mRNA and CAM1 as a negative control mRNA (Pichon et al., 2021). We first show that NET1 mRNA is mislocalized in KIF1C KO cells whereas CAM1 mRNA is not (new Fig S7 C,D). We then rescued the KO cells with FL or DIDR constructs and show that the FL protein rescues NET1 mRNA localization to the cell periphery whereas the DIDR construct does not (new Fig S7 E,F).

__Reviewer #2 (Significance (Required)): __

The mechanism of the KifC1-mRNA interaction was not investigated in the prior work, so the proposal that a liquid condensate is involved in novel. It is also topical, since there is intense current interest in transport and regulation of mRNAs by condensate-mediated mechanisms. The most useful part of this paper to the field may be the identification of IDR2 as required for mRNA binding.

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

KIF1C is a member of the kinesin-3 family, which is responsible for fast organelle transport in cells. The cargos of KIFIC are diverse, such as Golgi apparatus, Rab6 vesicles, exon junction complex (EJC), integrins, and RNA. Mutations in the KIF1C coding sequence leads to neurodegenerative diseases, such as hereditary spastic paraparesis (HSP). In addition, as an RNA transporter, KIF1C transports various types of mRNAs (e. g., APC-dependent mRNAs, KIF1C's own mRNA) along the microtubules and clusters them to cytoplasmic protrusions to fulfill certain biological functions.

In the current manuscript, Gen et.al., investigated the intracellular behaviors of the kinesin-3 member KIF1C. The study revealed that the KIF1C can form dynamic condensates both in cells and in vitro via an unstructured domain within the tail of the motor. KIF1C was found to also interact with synthesized RNA and other RNA granules in cells. In addition, the authors also show the KIFIC participates intracellular transport of endogenous mRNA, Rab13mRNA, identified a 47aa fragment in the KIF1C's IDR is critical for the KIF1C- Rab13mRNA interaction. Finally, as well as other prion-like proteins, the PPLS of KIF1C is buffered by the non-specific RNA pool in the cytoplasm.

In summary, this is an interesting work in the field, and reveals novel results about the mechanisms of motor protein transport that will be broadly interesting. The assays are generally well performed, and the results and discussion are well described, but some descriptions in the article should be more rigorous and objective. The article is very long, and I think it would benefit from streamlining and reducing the number of figures to make it more accessible for non-specialists in the field.

Here are some concerns:

__Major: __

Fig. 1A shows the domain organization of all kinesin-3 members, but Figure 1B only represents KIF1Bβ, KIF13B and KIF16B as controls. Generally, the KIF1Bα has the highest sequence similarity with KIF1C in kinesin-3 family (very high sequence similarity before aa 992 in KIF1C, which locates in IDR, probably contains IDR2a from Fig. S10A). In addition, both KIF1C and KIF1Bα contain a PLD from the prediction in this paper (Figure S2C). Although the authors show the phenotype of KIF1Bα in the Fig. S9, it might be better to put some descriptions up front, as readers may consider why the authors did not use KIF1Bα as a control. Actually, I kept thinking about this concern before I got to the discussion.

Response: We thank the reviewer for this suggestion. We have moved the descriptions of KIF1Ba phenotypes to earlier in the manuscript. We show that KIF1Ba forms puncta in cells but unlike KIF1C, the KIF1Ba puncta do not colocalize with known RNA granules P-bodies or stress granules (now in Fig S5 B,C). We show that, unlike KIF1C, the KIF1Ba puncta do not incorporate GU-rich or polyA RNA (now in Fig S6 B).

It would be better if the authors can combine the Fig. 2B and 2C, since the article did not mention Fig. 2B at all. In addition, Fig. S3 does not help this article too much. Probably it would be better if the authors could take the ΔIDR-mNG data from the Fig. S3 and put into the Fig. 2. as a negative control, especially for Fig. 2D an 2F. As for whether the phenotype of the ΔIDR-mNG construct is "similar to a constitutively active KIF1C construct containing only the motor domain (amino acids 1-348) (Fig. S3 C)", I do not think it is important here, since in this part, the authors are aiming to confirm the IDR is critical for KIF1C phase separation.

Response: We have combined Figures 2B and 2C as suggested. We prefer to leave Figure S3 intact since, as the reviewer mentioned, the article is already long and these data are not critical for the story.

The description "the condensate properties can be modulated by adjacent coiled-coil segments" in the abstract and the sentence "However, the coiled-coil segments in the stalk domain appear to facilitate puncta formation as the addition of increasing amounts of coiled coil resulted in increased KIF1C enrichment in puncta as compared to the IDR alone" in the article are not accurate, since there is no direct evidence in this manuscript that shows that. In Fig. 2D, as well as Fig. 6A and Fig. S7, it is manifest at a glance there are lots of IDR-mNG localized in nucleus, which decreases the concentration of this construct in cytoplasm which in turn may lower its capability to form puncta. This is important, as the results in Fig.4 show that the concentration of protein directly affects the formation of phase separated puncta in cells. From my view, the words "modulate", "tune" ... usually describe active processes, and these words may be confusing unless there are enough evidence support direct regulation. But the data presented in this article suggests to us that it is likely a passive process, such as the coiled coil region preventing the CC4-IDR construct from entering the nucleus (Fig. 2D, Fig. 6A and Fig. S7). Moreover, CC4 does affect the critical concentration of IDR in vitro (Fig. 5E), but that could be attributed to the coiled coil domain increasing its solubility. I like the word "influence" used in a subtitle in the discussion portion.

Response: We have removed this from the text.

In addition, the in vitro study of this paper in Fig. 5 did not show any significant difference of the puncta formation between IDR-mNG and CC4 - IDR-mNG (Diameter: 0.43 {plus minus} 0.22 μm (mean {plus minus} STD) for IDR-mNG vs 0.48 {plus minus} 0.27 μm (mean {plus minus} STD) for CC4-IDR-mNG. Roundness: No value was show in the article). So, a stricter assay or a more accurate description is required here to avoid any misleading to the readers.

Response: We now include p values showing that the differences in diameter and roundness are statistically significant (data moved to Fig S8 B).

The description for Fig. 5 "At 2 uM protein concentration and 100 mM NaCl, the KIF1C(IDR) droplets were smaller [diameter 0.43 {plus minus} 0.22 μm (mean {plus minus} STD)] than KIF1C(CC4+IDR) droplets [0.48 {plus minus} 0.27 μm (mean {plus minus} STD)] (Fig. 5 C)" does not appear accurate as well, since there is no significant difference between the value 0.43 {plus minus} 0.22 μm and the value 0.48 {plus minus} 0.27 μm, so it should not be descripted as "smaller". In addition, the article mentioned that "The KIF1C(IDR) puncta were also less round than those of KIF1C(CC4+IDR) (Fig. 5 C)", but there is no corresponding value from the quantification show the KIF1C(IDR) is less round.

Response: We now include p values showing that the differences in diameter and roundness are statistically significant (data moved to Fig S8 B).

The description in sentence "We thus tested whether ... LLPS is mutually exclusive (Fig. S5 A)" may not be accurate. Results in Fig. S5 only show there is no direct interaction between KIF1C and CLIP-170 or these two proteins do not colocalize. The words "mutually exclusive" means two proteins competent each other in the same location from my understanding.

Response: We have replaced the words "mutually exclusive" with "no colocalization" (line 204).

In addition, is it necessary to put Fig. S5 into this article? Since from my side, it does not help too much for the whole story. In cells, kinesin motors are autoinhibited in the cytoplasm. For this KIF1C, most of motors appear autoinhibited as well, even when the authors removed the IDR based on Fig. S3C (ΔIDR-mNG vs. MD-mNG). In this case, it is hard to investigate the potential interaction between the KIF1C (or its ΔIDR mutant) with the microtubules or with the tubulin due to the autoinhibition of constructs used in Fig. S5. It would be better to use other active versions of KIF1C, such as ΔP (Soppina et. al., PNAS, 2014) or other mutants (Ren et. al., PNAS, 2018; Wang et. al., Nat. Commun., 2022) if the authors want to show this part in the article.

Response: We agree that this data is not essential for the story, however, it may be of interest and benefit to others in the field studying LLPS of microtubule-associated proteins and we prefer to leave Figure S5 (now Figure S4) in the supplementary information.

The conclusion "This result suggests that the IDR- driven LLPS of KIF1C does not depend on mRNA incorporation, but is strongly affected by it" may not be accurate, there is no direct evidence that shows that mRNA, at least Rab13mRNA incorporation strongly affects the IDR- driven LLPS of KIF1C. Perhaps a knock out of Rab13mRNA would alter the formation of condensates, which would support a direct effect on LLPS.

Response: We have changed the text (line 306).

In addition, the sentence "These results also show that the LLPS is resistant to truncations of large portions of IDR" may not accurate, from my view, except IDR2a, the rest of the IDR may not participate or contribute too much to the formation of puncta, but that doesn't mean LLPS is resistant to the truncation of these portions in IDR, these are different logics. The quantification from Fig. 7E also show there is no significant difference between the ST and truncations except ΔIDR2 and ΔIDR2a in statistics, such as ST (21.8 {plus minus} 12.0 puncta per cell, 2.06 {plus minus} 0.83 μm diameter), ΔPLD (20.1 {plus minus} 13.7 puncta per cell, 1.62 {plus minus} 0.94 μm diameter), ΔIDR1 (23.1 {plus minus} 14.3 puncta per cell, 2.13 {plus minus} 1.04 μm diameter), ΔIDR3 (18.4 {plus minus} 8.1 puncta per cell, 1.71 {plus minus} 0.98 μm diameter).

Response: We have changed the text (line 307).

I am not sure I agree with the author's interpretation of their FRAP data in Fig. 3. It appears to me that there is a large immobile population of molecules, as the bleached areas recover less than 50% of their initial intensity. However, the authors conclude that there is rapid exchange of molecules in the puncta. The authors need to further analyze and discuss both the exchange rate of the population of molecules that exchange, but also the fraction of apparently immobile molecules that do not recover in their experiments. These data appear to suggest that a large percentage of the molecules in the KIF1C puncta in fact do not exchange with the cytoplasm and undermine their argument for a liquid-like phase of the puncta.

Response: We agree that the data suggest the existence of an immobile pool of KIF1C within the condensates. We have added this information to the main text (lines 178-182). We note that these findings are consistent with recent studies demonstrating membrane-less organelles with at least partially solid-like properties, including nucleoli and stress granules as well as microtubule associated proteins (see references, reviewed in Van Treeck & Parker 2019).

__Minor: __

As mentioned above, Fig. 2 F needs a negative control, since the values of FL and IDR are lower than other constructs, maybe use the Δ IDR-mNG protein is better. In addition, from my view, the lower value of IDR construct does not represent this construct has lower capability to form puncta, but more likely because most of this protein localizes in nucleus, thus dramatically lowering the cytoplasmic concentration.

Response: We have changed the text as suggested (lines 152-154).

Fig. 6A probably need a negative control as well, maybe use the same construct ΔIDR in Fig. S7 is better.

Response: We have now included KIF1Ba as a negative control (Fig S6 B).

Although I guess the reason for using hTERT-RPE1 cells in Rab13mRNA rescue assay (Fig. 6D-G) probably is easier to get KIF1C knock out cells (if I am correct), it would be better if there is a brief introduction for the reason to use hTERT-RPE1 here, since all previous assay in the article used COS-7 cells.

Response: You are correct and we have added text introducing the use of hTERT-RPE1 cells (line 269).

Is there any specific reason to use the construct ST in Fig. 7? Since in Fig. 6, the authors used FL-length KIFIC, if the authors want to avoid any effects caused by motor domain, the construct CC4-IRD also could be a simpler candidate.

Response: No specific reason other than to be consistent as most experiments that we carried out in cells used the ST construct (e.g. FRAP assay in Fig 3, hypotonic assay in Fig 3, RNaseA injection in Fig 4, RNA incorporation in Fig 4). (Note that Fig 7 is now Fig 6).

This article is a great case for motor-cargo interaction, since the RNA binding site of KIF1C is within its tail domain. This left me curious about if the interaction between the KIF1C and the membrane-less RNA granule is sufficient to release the KIF1C motor from autoinhibition? I guess the binding of RNA is not enough to release the KIF1C from autoinhibition. From Fig. S3C and Fig. 6D, seems the motor still in autoinhibition, even remove the Rab13mRNA binding region.

Response: We believe the question of whether the RNA binding relieves autoinhibition of KIF1C is beyond the scope of this manuscript and we plan to address this in the future with recombinant full-length KIF1C and RAB13 mRNAs.

There are some grammar mistakes, e.g., There should be a "is" between "IDR" and "critical" in the title "A subregion of the KIF1C IDR critical for enrichment of Rab13mRNA in condensates".

Response: Thank you. We have corrected this (line 289).

There should be a definition for the full names of the abbreviate "RBD" mentioned in the article although the readers may guess that is an RNA binding domain, if possible, it would be better but not necessary if the authors could show the residues or the region in IDR.

Response: RBD is defined at the beginning to the section "KIF1C condensates display properties of RNA granules" (line 219) but in response to the reviewer's comment, we now include this definition a second time in the Discussion section (line 420).

In the results (line 126), the authors refer to the KIF1C IDR without first defining this region in the introduction. I would re-word this sentence for clarity by first defining what an IDR is and how it's assessed in the current study.

Response: The IDR is defined at the end of the Introduction (lines 94-95).

What is the significance of the roundness measurement in Fig. 5? This should be described for the reader.

Response: Roundness refers to the shape of the droplet and this is now included in the text (line 323, data moved to Fig S8 B).

The authors state several times that this is the first kinesin shown to undergo LLPS. However, is this true? What about the recent work showing that the yeast Tea2 kinesin undergoes LLPS with other +TIP components (Maan et al. NCB 2023).

Response: We thank the reviewer for this comment. The recent work from the Dogterom lab (Maan et al., 2023) demonstrates that the end binding (EB) protein Mal3 forms condensates alone and with the kinesin-7 family member Tea2 and its cargo Tip1 for enrichment at microtubule plus ends. The authors show images of Mal 3 droplets and the requirement of the IDR domain and the crowding agent polyethylene glycol for droplet formation. The authors state that "Tea2 and Tip1 formed condensates under similar crowding conditions and concentrations on their own (Extended Data Fig. 5)." However, Extended Data Fig 5 reports on the fluorescence intensity of Mal3-EGFP colocalizing with Tea2 or Tip1. No images of Tea2-only droplets are shown and no information is provided on the Tea2 and/or PEG concentrations required for droplet formation or the liquid nature of Tea2 droplets. Thus, we do not feel comfortable stating that Tea2 on its own undergoes LLPS. We do reference the Maan et al., 2023 work in the Discussion listing microtubule-associated proteins shown to undergo LLPS (line 403) and when comparing the mM concentrations of KIF1C required for LLPS to the mM concentrations of these other microtubule-associated proteins (line 417).

The authors don't discuss KIF5A, but their analysis reveals it also contains a low complexity region that may undergo LLPS (Fig. S2D). This would fit with recent reports that KIF5A tends to oligomerize more than other KIF5 isoforms, and that mutations in KIF5A that impact the tail domain may lead to aberrant oligomerization. I feel that it would be useful to the field for the authors to discuss these results in light of their own.

Response: We thank the reviewer for this suggestion. Although it is intriguing that KIF5A is predicted to contain an IDR, there is, however, no data to suggest that KIF5A undergoes LLPS. Rather, the current literature suggests that KIF5A undergoes higher-order oligomerization and accumulation at the cell periphery, especially for the isoform lacking exon 27 (Nakano et al., 2022, Baron et al., 2022, Pant et al., 2023, Soustelle et al., 2023). It thus does not seem prudent for us to speculate on whether or not KIF5A undergoes LLPS.

__Reviewer #3 (Significance (Required)): __

The study is novel and interesting and will be impactful for the cytoskeletal and RNA biology communities. The experiments are of high quality and controls are appropriate. The finding that motor proteins can participate in LLPS will be of high interest for a variety of fields and provides a very interesting advance over current knowledge in the 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 #3

Evidence, reproducibility and clarity

KIF1C is a member of the kinesin-3 family, which is responsible for fast organelle transport in cells. The cargos of KIFIC are diverse, such as Golgi apparatus, Rab6 vesicles, exon junction complex (EJC), integrins, and RNA. Mutations in the KIF1C coding sequence leads to neurodegenerative diseases, such as hereditary spastic paraparesis (HSP). In addition, as an RNA transporter, KIF1C transports various types of mRNAs (e. g., APC-dependent mRNAs, KIF1C's own mRNA) along the microtubules and clusters them to cytoplasmic protrusions to fulfill certain biological functions.

In the current manuscript, Gen et.al., investigated the intracellular behaviors of the kinesin-3 member KIF1C. The study revealed that the KIF1C can form dynamic condensates both in cells and in vitro via an unstructured domain within the tail of the motor. KIF1C was found to also interact with synthesized RNA and other RNA granules in cells. In addition, the authors also show the KIFIC participates intracellular transport of endogenous mRNA, Rab13mRNA, identified a 47aa fragment in the KIF1C's IDR is critical for the KIF1C- Rab13mRNA interaction. Finally, as well as other prion-like proteins, the PPLS of KIF1C is buffered by the non-specific RNA pool in the cytoplasm.

In summary, this is an interesting work in the field, and reveals novel results about the mechanisms of motor protein transport that will be broadly interesting. The assays are generally well performed, and the results and discussion are well described, but some descriptions in the article should be more rigorous and objective. The article is very long, and I think it would benefit from streamlining and reducing the number of figures to make it more accessible for non-specialists in the field.

Here are some concerns:

Major:

1. Fig. 1A shows the domain organization of all kinesin-3 members, but Figure 1B only represents KIF1Bβ, KIF13B and KIF16B as controls. Generally, the KIF1Bα has the highest sequence similarity with KIF1C in kinesin-3 family (very high sequence similarity before aa 992 in KIF1C, which locates in IDR, probably contains IDR2a from Fig. S10A). In addition, both KIF1C and KIF1Bα contain a PLD from the prediction in this paper (Figure S2C). Although the authors show the phenotype of KIF1Bα in the Fig. S9, it might be better to put some descriptions up front, as readers may consider why the authors did not use KIF1Bα as a control. Actually, I kept thinking about this concern before I got to the discussion.
2. It would be better if the authors can combine the Fig. 2B and 2C, since the article did not mention Fig. 2B at all. In addition, Fig. S3 does not help this article too much. Probably it would be better if the authors could take the ΔIDR-mNG data from the Fig. S3 and put into the Fig. 2. as a negative control, especially for Fig. 2D an 2F. As for whether the phenotype of the ΔIDR-mNG construct is "similar to a constitutively active KIF1C construct containing only the motor domain (amino acids 1-348) (Fig. S3 C)", I do not think it is important here, since in this part, the authors are aiming to confirm the IDR is critical for KIF1C phase separation.
3. The description "the condensate properties can be modulated by adjacent coiled-coil segments" in the abstract and the sentence "However, the coiled-coil segments in the stalk domain appear to facilitate puncta formation as the addition of increasing amounts of coiled coil resulted in increased KIF1C enrichment in puncta as compared to the IDR alone" in the article are not accurate, since there is no direct evidence in this manuscript that shows that. In Fig. 2D, as well as Fig. 6A and Fig. S7, it is manifest at a glance there are lots of IDR-mNG localized in nucleus, which decreases the concentration of this construct in cytoplasm which in turn may lower its capability to form puncta. This is important, as the results in Fig.4 show that the concentration of protein directly affects the formation of phase separated puncta in cells. From my view, the words "modulate", "tune" ... usually describe active processes, and these words may be confusing unless there are enough evidence support direct regulation. But the data presented in this article suggests to us that it is likely a passive process, such as the coiled coil region preventing the CC4-IDR construct from entering the nucleus (Fig. 2D, Fig. 6A and Fig. S7). Moreover, CC4 does affect the critical concentration of IDR in vitro (Fig. 5E), but that could be attributed to the coiled coil domain increasing its solubility. I like the word "influence" used in a subtitle in the discussion portion.

In addition, the in vitro study of this paper in Fig. 5 did not show any significant difference of the puncta formation between IDR-mNG and CC4 - IDR-mNG (Diameter: 0.43 {plus minus} 0.22 μm (mean {plus minus} STD) for IDR-mNG vs 0.48 {plus minus} 0.27 μm (mean {plus minus} STD) for CC4-IDR-mNG. Roundness: No value was show in the article). So, a stricter assay or a more accurate description is required here to avoid any misleading to the readers.

The description for Fig. 5 "At 2 uM protein concentration and 100 mM NaCl, the KIF1C(IDR) droplets were smaller [diameter 0.43 {plus minus} 0.22 μm (mean {plus minus} STD)] than KIF1C(CC4+IDR) droplets [0.48 {plus minus} 0.27 μm (mean {plus minus} STD)] (Fig. 5 C)" does not appear accurate as well, since there is no significant difference between the value 0.43 {plus minus} 0.22 μm and the value 0.48 {plus minus} 0.27 μm, so it should not be descripted as "smaller". In addition, the article mentioned that "The KIF1C(IDR) puncta were also less round than those of KIF1C(CC4+IDR) (Fig. 5 C)", but there is no corresponding value from the quantification show the KIF1C(IDR) is less round. 4. The description in sentence "We thus tested whether ... LLPS is mutually exclusive (Fig. S5 A)" may not be accurate. Results in Fig. S5 only show there is no direct interaction between KIF1C and CLIP-170 or these two proteins do not colocalize. The words "mutually exclusive" means two proteins competent each other in the same location from my understanding.

In addition, is it necessary to put Fig. S5 into this article? Since from my side, it does not help too much for the whole story. In cells, kinesin motors are autoinhibited in the cytoplasm. For this KIF1C, most of motors appear autoinhibited as well, even when the authors removed the IDR based on Fig. S3C (ΔIDR-mNG vs. MD-mNG). In this case, it is hard to investigate the potential interaction between the KIF1C (or its ΔIDR mutant) with the microtubules or with the tubulin due to the autoinhibition of constructs used in Fig. S5. It would be better to use other active versions of KIF1C, such as ΔP (Soppina et. al., PNAS, 2014) or other mutants (Ren et. al., PNAS, 2018; Wang et. al., Nat. Commun., 2022) if the authors want to show this part in the article. 5. The conclusion "This result suggests that the IDR- driven LLPS of KIF1C does not depend on mRNA incorporation, but is strongly affected by it" may not be accurate, there is no direct evidence that shows that mRNA, at least Rab13mRNA incorporation strongly affects the IDR- driven LLPS of KIF1C. Perhaps a knock out of Rab13mRNA would alter the formation of condensates, which would support a direct effect on LLPS.

In addition, the sentence "These results also show that the LLPS is resistant to truncations of large portions of IDR" may not accurate, from my view, except IDR2a, the rest of the IDR may not participate or contribute too much to the formation of puncta, but that doesn't mean LLPS is resistant to the truncation of these portions in IDR, these are different logics. The quantification from Fig. 7E also show there is no significant difference between the ST and truncations except ΔIDR2 and ΔIDR2a in statistics, such as ST (21.8 {plus minus} 12.0 puncta per cell, 2.06 {plus minus} 0.83 μm diameter), ΔPLD (20.1 {plus minus} 13.7 puncta per cell, 1.62 {plus minus} 0.94 μm diameter), ΔIDR1 (23.1 {plus minus} 14.3 puncta per cell, 2.13 {plus minus} 1.04 μm diameter), ΔIDR3 (18.4 {plus minus} 8.1 puncta per cell, 1.71 {plus minus} 0.98 μm diameter). 6. I am not sure I agree with the author's interpretation of their FRAP data in Fig. 3. It appears to me that there is a large immobile population of molecules, as the bleached areas recover less than 50% of their initial intensity. However, the authors conclude that there is rapid exchange of molecules in the puncta. The authors need to further analyze and discuss both the exchange rate of the population of molecules that exchange, but also the fraction of apparently immobile molecules that do not recover in their experiments. These data appear to suggest that a large percentage of the molecules in the KIF1C puncta in fact do not exchange with the cytoplasm and undermine their argument for a liquid-like phase of the puncta.

Minor:

1. As mentioned above, Fig. 2 F needs a negative control, since the values of FL and IDR are lower than other constructs, maybe use the Δ IDR-mNG protein is better. In addition, from my view, the lower value of IDR construct does not represent this construct has lower capability to form puncta, but more likely because most of this protein localizes in nucleus, thus dramatically lowering the cytoplasmic concentration.
2. Fig. 6A probably need a negative control as well, maybe use the same construct ΔIDR in Fig. S7 is better.
3. Although I guess the reason for using hTERT-RPE1 cells in Rab13mRNA rescue assay (Fig. 6D-G) probably is easier to get KIF1C knock out cells (if I am correct), it would be better if there is a brief introduction for the reason to use hTERT-RPE1 here, since all previous assay in the article used COS-7 cells.
4. Is there any specific reason to use the construct ST in Fig. 7? Since in Fig. 6, the authors used FL-length KIFIC, if the authors want to avoid any effects caused by motor domain, the construct CC4-IRD also could be a simpler candidate.
5. This article is a great case for motor-cargo interaction, since the RNA binding site of KIF1C is within its tail domain. This left me curious about if the interaction between the KIF1C and the membrane-less RNA granule is sufficient to release the KIF1C motor from autoinhibition? I guess the binding of RNA is not enough to release the KIF1C from autoinhibition. From Fig. S3C and Fig. 6D, seems the motor still in autoinhibition, even remove the Rab13mRNA binding region.
6. There are some grammar mistakes, e.g., There should be a "is" between "IDR" and "critical" in the title "A subregion of the KIF1C IDR critical for enrichment of Rab13mRNA in condensates".
7. There should be a definition for the full names of the abbreviate "RBD" mentioned in the article although the readers may guess that is an RNA binding domain, if possible, it would be better but not necessary if the authors could show the residues or the region in IDR.
8. In the results (line 126), the authors refer to the KIF1C IDR without first defining this region in the introduction. I would re-word this sentence for clarity by first defining what an IDR is and how it's assessed in the current study.
9. What is the significance of the roundness measurement in Fig. 5? This should be described for the reader.
10. The authors state several times that this is the first kinesin shown to undergo LLPS. However, is this true? What about the recent work showing that the yeast Tea2 kinesin undergoes LLPS with other +TIP components (Maan et al. NCB 2023).
11. The authors don't discuss KIF5A, but their analysis reveals it also contains a low complexity region that may undergo LLPS (Fig. S2D). This would fit with recent reports that KIF5A tends to oligomerize more than other KIF5 isoforms, and that mutations in KIF5A that impact the tail domain may lead to aberrant oligomerization. I feel that it would be useful to the field for the authors to discuss these results in light of their own.

Significance

The study is novel and interesting and will be impactful for the cytoskeletal and RNA biology communities. The experiments are of high quality and controls are appropriate. The finding that motor proteins can participate in LLPS will be of high interest for a variety of fields and provides a very interesting advance over current knowledge in the 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

This paper investigates mRNA transport by the kinesin Kif1C and tests the hypothesis that liquid condensation of the disordered C terminal region is important for mRNA recruitment. It is based on prior work from other labs showing that Kif1C recruits and transports a set of mRNAs to the periphery of cells. The mechanism of the KifC1-mRNA interaction was not investigated in the prior work, so the proposal that a liquid condensate is involved in novel. It is also topical, since there is intense current interest in transport and regulation of mRNAs by condensate-mediated mechanisms. The most useful part of this paper to the field may be the identification of IDR2 as required for mRNA binding in Fig 7

A major concern is reliance on expression of tagged KifC1 in Cos cells in several figures. The expression level in these experimental probably far exceeds normal, though this comparison is not reported. It is possibly justified to use over-expression to reveal a condensate mechanism, but it is concerning and the authors needs to strongly qualify their conclusions. One way to moderate this concern would be to examine condensation as a function of expression level.

Another significant concern is that the biochemical reconstitution figure tests protein alone, not protein + RNA. Disordered RNA binding proteins usually phase separate better in the presence of RNA. The best reconstitution papers evaluate specificity of RNA recruitment to condensates. Specificity testing in a reconstituted system may not be required for a first paper, but testing the effect of some kind of RNA seems important.

A final concern is that the specificity of mRNA recruitment to Kif1C puncta in cells is not critically evaluated. Among endogenous mRNAs, only one (Rab13) is tested. The paper would be stronger with a second positive mRNA and a negative control mRNA.

Significance

The mechanism of the KifC1-mRNA interaction was not investigated in the prior work, so the proposal that a liquid condensate is involved in novel. It is also topical, since there is intense current interest in transport and regulation of mRNAs by condensate-mediated mechanisms. The most useful part of this paper to the field may be the identification of IDR2 as required for mRNA binding.

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

Geng et al. explore the molecular mechanisms underlying the role of KIF1C in RNA transport, focusing on how it interacts with RNA. KIF1C is shown to form dynamic puncta when overexpressed in COS-7 cells that do not appear to colocalise with organelle markers. An IDR in the tail of the kinesin is necessary and sufficient for the formation of these structures and FRAP experiments show that they can exchange their contents with proteins in the cytosol and that their formation can can be reversibly modulated by hypotonic shock, consistent with LLPS. In vitro, the IDR and flanking regions can undergo phase separation at physiologically relevant concentrations and salt conditions. In cells, KIF1C puncta enrich for RNAs and support their transport, and depletion of RNA modulates KIF1C LLPS properties. A model is proposed whereby KIF1C mediated RNA transport to the cell periphery promotes the formation of a protein-RNA condensate that may act to fine tune local RNA activity.

Major comments

In general, the claims made her are well-supported by the data. However, I think that some exploration of the extent of LLPS at different KIF1C expression levels in cells is important but missing. The authors carefully estimate the endogenous concentration of KIF1C in COS-7 cells (at around 25 nm), but is isn't clear how this compares to that observed in transient transfection experiments. Although this is partly addressed in the vitro assays, I am still left with some questions over the extent of this phenomenon in a cellular context. Can the authors provide some experimental evidence to support the proposition that LLPS occurs (perhaps in a more localised fashion?, as Fig.9) at lower KIF1C expression levels? One way to address this might be a GFP-knock-in (although how feasible this is may depend on the genomic context), alternatively, the authors could generate cell lines that express KIF1C-GFP from a very weak promoter, demonstrate LLPS using their established assays, show that this is comparable to endogenous expression.

Minor comments

Lines 107-109 and Figure 1B on localisation of other kinesin-3s. The authors state that they localise to certain organelle but don't show co-staining for those organelles.

Lines 172-183 and Figure 3. Evidence is provided through FRAP experiments that KIF1C puncta exchange with the cytosolic pool. However, the extent of recovery appears to saturate at <40%. Does this suggest the existence of an immobile pool of KIF1C within these structures?

Line 238 - Fig. S5C is cited as data on endogenous concentration of KIF1C - this should be Fig. S6C.

Line 331-332 - I did not fully follow the logical here the RNAse A injection experiment supports the idea that KIF1C interaction with RNA is sequence selective. Could the authors expand on this.

Significance

This study introduces a new and exciting concept to motor protein biology: that some cytoskeletal motors and motor-cargo complexes can undergo phase separation, and that this is important for their function. The experiments are logical, progressive, and form a clear and compelling case. The main limitation is that demonstration of LLPS in cells is limited to over-expressed protein. Some exploration/demonstration of LLPS properties of KIF1C in cells at near to endogenous expression levels would enhance the study.

The work should be of interest to a broad range of readers, from the cytoskeletal motor community, those interested in mRNA regulation, as well as scientists studying phase separation more generally.

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

Learn more at Review Commons

---

Reply to the reviewers

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

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

Learn more at Review Commons

---

Reply to the reviewers

Manuscript number: RC-2023-01938R

Corresponding author(s): Ilan, Davis

1. General Statements

We thank all four reviewers for their helpful and constructive comments. We have gone through each and every comment and proposed how we would address each point raised by the reviewers. We are confident our proposed revisions are feasible within a reasonable and expected time frame. Some of the comments regarding minor typo/aesthetics and extra references have already been addressed in the transferred manuscript. The changes are highlighted in yellow in the transferred manuscript.

2. Description of the planned revisions

Reviewer #1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Major points:

1. The presented work itself (Figures 1-4) does not need significant adjustments prior to publication, in my view, with only a few points to address. However, the work in Figure 5- doesn't really support the claims the authors make on its own, and would require some additional experiments or at the very least discussion of the caveats to its current form.

We thank the reviewer for these comments and will follow the reviewer’s suggestion by discussing the caveats regarding the interpretation of Figure 5. We will also add to the discussion to suggest future research approaches beyond the scope of this manuscript that would address the functional importance of localised mRNA translation. We will briefly mention in the discussion methods such as the quantification of the mRNA foci and the disruption of the mRNA localisation signals to disrupt localised translation and the use of techniques such as Sun-Tag (Tanenbaum et al, 2014) and FLARIM (Richer et al, 2021) to visualise local translation directly.

---

Tanenbaum et al, 2014 DOI: 10.1016/j.cell.2014.09.039

Richer et al, 2021 DOI: 10.1101/2021.08.13.456301

* __ Localized glia transcripts, are they "glial/CNS/PNS" significant or are they similar to other known datasets of protrusion transcriptomes? The authors compared their 4801 "total" localized to a local transcriptome dataset from the Chekulaeva lab finding that a significant fraction are localized in both. As the authors note, this is in good agreement with a recent paper from the Talifarro lab showing conservation of localization of mRNAs across different cell types. What the authors haven't done here, is further test this by looking at other non-neuronal projection transcriptomic datasets (for example Mardakheh Developmental Cell 2015, among others). If the predicted glia-localized processes are similar to non-neuronal processes transcriptomes, this would further strengthen this claim and rule out some level of CNS/PNS derived linage driving the similarities between glia and neuronal localized transcripts. __*

This is a good point and we thank the review for pointing out this interesting cancer data set. We will do as the reviewer suggests and intersect our data with Mardakheh Dev Cell 2015 to test the further generality of localisation in neurons and glia, in other cell types. Specifically, we plan to intersect both glial (this study) and neuronal (von Kuegelgen & Chekulaeva, 2020) dataset with protrusive breast cancer cells (Mardakeh et al, 2015).

• *

von Kuegelgen & Chekulaeva, 2020 DOI: 10.1002/wrna.1590

Mardakeh et al, 2015 DOI: 10.1016/j.devcel.2015.10.005

* __ The presentation/discussion around Figure 3 is a bit weaker than other parts of the manuscript, and it doesn't really contribute to the story in its current form. Notably there is no discussion about the significance of glia in neurological disorders until the very end of the manuscript (page 21), meaning when its first brought up.. it just sits there as a one off side point. The authors might consider strengthening/tightening up the discussion here, if they really want to keep it as a solo main figure rather than integrating it somewhere else/putting it into supplemental. In my view, Figures 2 & 3 should be merged into something a bit more streamlined. __*

This is a good point. We plan to strengthen the presentation of Figure 3 and discussion of the significance of glia in neurological disorders by adding a description of the Figure in the Results section and highlighting the significance of glia in nervous system disorders in the Discussion section.

* __ Why aren't there more examples of different mRNAs in Figure 4? Seems a waste to kick them all to supplemental. __*

We agree that it could be helpful to show different expression patterns in the main figure. To address this point we will add Pdi (Fig. S4D), which shows mRNA expression in both the glia and the surrounding muscle cell. This pattern is in contrast to Gs2, which is highly specific to glial cells. We will also note that although pdi mRNA is present in both the glia and muscle, Pdi protein is only abundant in the glia, suggesting that translation of pdi mRNA to protein is regulated in a cell-specific manner.

The plasticity experiments, while creative, I think need to be approached far more cautiously in their interpretation. Given that the siRNAs will completely deplete these mRNAs- it really needs to be stressed any/all of the effects seen could just be the result of "defective" or "altered" states in this glial population- which has spill over effects on plasticity in at the NMJ. Without directly visualizing if these mRNAs are locally translated in these processes and assessing if their translation is modulated by their plasticity paradigm, all these experiments can say is that these RNAs are needed in glia to modulate ghost bouton formation in axons. This represents the weakest part of this manuscript, and the part that I feel does not actually backup the claims currently being made. Without any experiments to A. quantify how much of these transcripts are localized vs in the cell body of these glia, B. visualize/quantify the translation of these mRNAs during baseline and during plasticity; the authors cannot use these data to claim that localized mRNAs are required for synaptic plasticity.

We are grateful to the reviewer for pointing out that we were not precise enough in defining our interpretation of the structural plasticity assay. We did not intend to claim that our results show that local translation of these transcripts is necessary for plasticity, only that these transcripts are localized and are required in the glia for plasticity in the adjacent neuron (in which the transcript levels are not disrupted in the experiment). Definitively proving that these transcripts are required locally and translated in response to synaptic activity would require genetic/chemical perturbations and imaging assays that would require a year or more to complete, so are beyond the scope of this manuscript. To address this point, we will clarify that the results do not show that localized transcripts are required, only that the transcripts are required somewhere specifically in the glial cell (without affecting the neuron level), and we can indeed show in an independent experiment that there are localized transcripts.

Reviewer #2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Major points:

1. * __ The authors analyse the 1700 shortlisted genes for Gene Ontology and associations with austism spectrum disorder, leading to interesting results. However, it is not clear to what extent the enrichments they observe are driven by their presumptive localization or if the associations are driven to a significant extent by the presence of these genes in the selected cell types in the Fly Cell Atlas. One way to address this would be to perform the GO and SFARI analysis on genes that are expressed in the same cells in the Fly Cell Atlas but were not shortlisted from the mammalian cell datasets - the results could then be compared to those obtained with the 1700 localized transcripts. __* This is a fair point raised by the reviewer as genes involved in neurological disease such as Autism Spectrum Disorder may be enriched in CNS/PNS cell types. We will follow the reviewer’s suggestion to perform GO and SFARI gene enrichment analysis in genes that were not shortlisted for presumptive glial localisation.

Although the authors attempt to justify its inclusion, I'm not convinced why it was important to use the whole cell transcriptome of perisynaptic Schwann cells as part of the selection process for localizing transcripts. Including this dataset may reduce the power of the pipeline by including mRNAs that are not localized to protrusions. How many of the shortlisted 1700 genes, and how many of the 11 glial localized mRNAs in Table 5, would be lost if the whole cell transcriptome were excluded. More generally, what is the distribution of the 11 validated localizing transcripts in each dataset in Table 4? This information might be valuable for determining which dataset(s), if any, has the best predictive power in this context.

We thank the reviewer for raising this point, which we will address with further analysis and adding to the discussion. We propose to address the criticism by running our analysis pipeline without the inclusion of the dataset using Perisynaptic Schwann Cells (PSCs) and then intersect with the PSCs-expressed genes, since their functional similarity with polarised Drosophila glial cells is highly relevant. We also agree with the reviewer that it would be a useful control for us to assess the ‘predictive power’ of each glial dataset by calculating their contribution to the shortlisted 1,700 glial localised transcripts and to the 11 experimentally validated transcripts via in situ hybridisation. To address this point, we plan to add this information in the revised manuscript.

* __ Did the authors check if any of the RNAi constructs are reducing levels of the target mRNA or protein? Doing so would strengthen the confidence in these important results significantly. In any case, the authors should also mention the caveat of potential off-target effects of RNAi. __*

We thank the reviewer for their useful comment and agree that the extent to which the RNAi expression reduces the levels of mRNA is not specifically known. We will add a FISH experiment on lac, pdi and gs2 RNAi showing very strong reduction in mRNA levels. We will also add an explanation of the caveats of the use of the RNAi system to the discussion.

Methods: what is the justification for assuming that if the RNAi cross caused embryonic or larval lethality then the 'next most suitable' RNAi line is reporting on a phenotype specific to the gene. If the authors want to claim the effect is associated with different degrees of knockdown they should show this experimentally. An alternative explanation is that the line used for phenotypic analysis in glia is associated with an off-target effect.

We thank the reviewer for this comment. We agree that off target effects cannot in principle be completely ruled out without considerable additional experimental analysis beyond the scope of this manuscript. To address the criticism we will remove the expression data of the lines that cause lethality and revise the discussion to explain that the level of knockdown in each line is unknown, and would require further experimental exploration.

Minor points:

1. It would be helpful to have in the Introduction (rather than the Results, as is currently the case) an operational definition of mRNA localization in the context of the study. And is it known whether or not localization in protrusions is the norm in mammalian glia or the Drosophila larval glia? I ask because it may be that almost all mRNAs diffuse into the protrusion, so this is not a selective process. One interesting approach to test this idea might be to test if the 1700 shortlisted transcripts have a significant underrepresentation of 'housekeeping' functions. We thank the reviewer for this excellent suggestion. To address the comment, we will move our explanation of the operational definition of mRNA localization to the Introduction. We will also perform enrichment analysis of housekeeping genes within 1,700 shortlisted transcripts compared to the transcriptome background, as the reviewer suggested.

Reviewer #3 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Major points:

1. The authors have pooled data from different studies across different type of glial cells performed from in vitro to in vivo. While pooling datasets may reveal common transcripts enriched in processes, this may not be the best approach considering these are completely different types of glial cells with distinct function in neuronal physiology. We thank the reviewer for highlighting the need for us to further justify why we pooled datasets. We will revise the manuscript to better emphasise that the overarching goal of our study was to try to discern a common set of localised transcripts shared between the cells. The problem with analysing and comparing individual data sets is that much of the variation may be due to differences in the methods used and amount of material, rather than differences in the type of cells used. We will revise the discussion to make this point and plan to explain that our approach corresponds well with a previous publication pooling localised mRNA datasets in neurons (von Kugelgen & Chekulaeva 2021).

von Kuegelgen & Chekulaeva, 2020 DOI: 10.1002/wrna.1590

It is important to note the limitations of the study. For example, DeSeq2 is biased for highly expressed transcripts. How robust was the prediction for low abundance transcripts?

The presented 1,700 transcripts were shortlisted based on their presence and expression level (TPM) in glial protrusions rather than their relative enrichment. Nevertheless, the reviewer makes a valid criticism of our use of DESeq2, where we compared enriched transcripts in glial and neuronal protrusions in Figure 1D. To address this point we will discuss this caveat in the relevant section.

The issue raised regarding low abundance transcript prediction raises an important question: does the likelihood of localisation to cell extremities correlate with mRNA abundance? We have already partially addressed this point, since our analysis of the fraction of localised transcripts per expression level quantiles shows only limited correlation. To address this comment, we will add these results in the revised manuscript as a supplementary figure.

The authors identify 1,700 transcripts that they classify as "predicted to be present" in the projections of the Drosophila PNS glia. This was based on the comparison to all the mammalian glial transcripts. Since the authors have access to a transcriptomic study from Perisynaptic Schwann cells (PSCs), the nonmyelinating glia associated with the NMJ isolated from mice; it would be more convincing to then validate the extent of overlap between Drosophila peripheral glial with the mammalian PSCs. This may reveal conserved features of localized transcripts in the PNS, particularly associated with the NMJ function.

Thank you for the valuable suggestion. A similar point was also raised by __[Reviewer #2 - Major point 2] __to re-run our pipeline excluding the PSCs dataset and intersect with the PSC transcriptome post-hoc. Please see the above section for our detailed response.

Fig 2: What is the extent of overlap between the translating fractions versus the localized fraction? It will be informative to perform the functional annotation of the translating glial transcripts as identified from Fig 1D.

This is an interesting question. To address this point, we plan to: (i) compare transcripts that are translated vs. localised in glial protrusions, and (ii) perform functional annotation enrichment analysis on the translated fraction of genes.

"We conclude predicted group of 1,700 are highly likely to be peripherally localized in Drosophila cytoplasmic glial projections". To validate their predictions, the authors test some of these candidates in only one glial cell type. It might be worthy to extend this for other differentially expressed genes localized in another glial type as well.

The presented in vivo analyses made use of the repo-GAL4 driver, which is active in all glial subtypes, including subperineurial, perineurial and wrapping glia that make distal projection to the larval neuromuscular junction. We agree that subtype-specific analysis would be highly informative, but we believe this is outside the scope of the current work where we aimed to identify conserved localised transcriptomes across all glial subtypes. Nevertheless, to address the comment, we plan to further clarify our use of pan-glial repo-GAL4 driver in the Results and Method section of the revised manuscript.

Figure 5: The authors perform KD of candidate transcripts to test the effect on synapse formation. However, these are KD with RNAi that spans across the entire cell. To make the claim about the importance of "target" RNA localization in glia stronger, ideally, they should disrupt the enrichment specifically in the glial protusions and test the impact on bouton formation. Do these three RNAs have any putative localization elements?

We agree with the review, that we would ideally test the effect of disruption of mRNA localization (and therefore localised translation). However, we feel these experiments are beyond the scope of this current study, as they will require a long road of defining localisation signals that are small enough to disrupt without affecting other functions. To address this comment we will revise the Discussion section to mention those difficulties explicitly, and clarify the limitations of the approach used in our study for greater transparency.

Reviewer #4 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Major points:

1. The authors use FISH to validate the glial expression of their target genes, though these experiments are not quantified, and no controls are shown. The authors should provide a supplemental figure with "no probe" controls, and/or validate the specificity of the probe via glial knockdown of the target gene (see point 2). Furthermore, these data should be quantified (e.g. number of puncta colocalized with NMJ glia membrans). Thank you for requesting further information regarding the YFP smFISH probes. We have validated the specificity and sensitivity of the YFP probe in our recent publication (Titlow et al, 2023, Figure 1 and S1). Specifically, we demonstrated the lack of YFP probe signal from wild-type untagged biosamples and showed colocalization of YFP spots with additional probes targeting the endogenous exon of the transcript. Nevertheless, we will address this comment by adding control image panels of smFISH in wild-type (OrR) neuromuscular junction preparations.

Titlow et al, 2023 DOI: 10.1083/jcb.202205129

For the most part, the authors only use one RNAi line for their functional studies, and they only show data for one line, even if multiple were used. To rule out potential false negatives, the authors should leverage their FISH probes to show the efficacy of their knockdowns in glia. This would serve the dual purpose of validating the new probes (see point 1).

Thank you for the suggestion. This point was also raised by [Reviewer #2 - Major point 3]. Please see above for our detailed response.

In Figure 5 E, given the severe reduction in size in the stimulated Pdi KD animals, the authors should show images of the unstimulated nerve as well. Do the nerve terminals actually shrink in size in these animals following stimulation, rather than expand? The NMJ looks substantially smaller than a normal L3 NMJ, though their quantification of neurite size in F suggests they're normal until stimulation.

We share the same interpretation of the data with the reviewer that the neurite area is reduced post-potassium stimulation in pdi knockdown animals. We will follow the reviewer’s suggestion and add an image showing unstimulated neuromuscular junctions.

Minor points:

The authors claim that there is an enrichment of ASD-related genes in their final list of ~1400 genes that are enriched in glial processes. It is well-appreciated that synaptically-localized mRNAs are generally linked to ASDs. Can the authors comment on whether the transcripts localized to glial processes are even more linked to ASDs and neurological disorders than transcripts known to be localized to neuronal processes?

This is an interesting point. To address the comment, we will add a comparison of the degree of enrichment of ASD-related genes in neurite vs. glial protrusions in the revised manuscript.

__*

*__

• *

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

Reviewer #1

---

1. The use of blue/green or blue/green/magenta is difficult to resolve in some places. Swapping blue for cyan would greatly aid in visualizing their data.
2. *

This comment is much appreciated. We have swapped blue for cyan in Figures 4 and S4. We have also changed Figure S1 to increase contrast and visibility as per reviewer’s comment.

Make the colouring/formatting of the tables more consistent, its distracting when its constantly changing (also there is no need for a blue background.. just use a basic white table).

This comment is much appreciated. We have applied a consistent colour palette to the Tables without background colourings and made the formatting uniform.

• *

Reviewer #2

• *

Introduction: 'Asymmetric mRNA localization is likely to be as important in glia, as it is in neurons,...'. Remove commas

Thank you for pointing this mistake out. We have made the corresponding edits.

• *

Reviewer #3

RNA localization in oligodendrocytes has been well studied and characterized. The authors should cite and discuss those papers (PMID: 18442491; PMID: 9281585).

We thank the reviewer for this useful suggestion. We have added these references to the paper.

• *

• *

Reviewer #4

• *

• In Figure 5D, the authors should include a label to indicate that these images are from an unstimulated condition. We thank the reviewer for pointing this out. We have added the label as requested.

The authors are missing a number of key citations for studies that have explored the functional significance of mRNA trafficking in glia, and those that have validated activity-dependent translation:

- ____https://pubmed.ncbi.nlm.nih.gov/18490510____/

-____https://pubmed.ncbi.nlm.nih.gov/7691830____/

-____https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001053

-____https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7450274____/

-_https://pubmed.ncbi.nlm.nih.gov/36261025_*/

*__

We thank the reviewer for the comment. We have added these references to the text.

• *

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

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

Learn more at Review Commons

---

Reply to the reviewers

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

Our point-by-point response contains figures that we could not manage to upload in this text box.

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

Learn more at Review Commons

---

Reply to the reviewers

The authors do not wish to provide a response at this time, please see our Revision Plan

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer #1

In this paper, authors report that radiation, acidic pH, hypoxia, and drugs that interfere with lipid synthesis, all of which affect lipid droplets (LD), also affect the production of small extracellular vesicles (sEVs). In addition, they also report that LD content and sEV secretion are also modulated in CR-CSCs. Authors conclude that sEV formation and secretion is directly linked to LDs, and that their studies may open the way to new clinical perspectives. However, some important issues need to be addressed before the paper can be considered for publication.

My _main concern is that the notion that LDs and sEVs are linked remains vague. Do cells contain more LDs and secrete more sEVs because these two pathways are selectively up-regulated via some mechanism____s_ that controls both pathways in a concerted manner? Or do cells with more LDs and more sEVs also contain more of everything, perhaps as a result of metabolic activity? __

We appreciate the Reviewer's observations. Indeed, this comment represents the main pillar of the entire manuscript. We have attempted to uncover the molecular mechanism behind this novel and intriguing organelle connection. First of all, we have adapted the manuscript emphasizing that the LD – sEV connection might be direct or indirect. Our omic data suggested that some proteins belonging to the RAB family, mainly Rab18, Rab7a and Rab5c, could play a pivotal role in the LDs-sEVs axis. To strengthen those results, we have performed additional experiments by silencing the expression of the three candidate Rabs. Rab5c seems to be a good candidate to modulate the LD-sEV connection. We believe that Rab5c is not the only contributor to the LD-sEV connection but is part of a whole set of different elements that regulate this axis. However, it is quite challenging to rule out other molecular candidates as co-contributors to this phenomenon, especially when considering cellular metabolic pathways.

We recognize that external stimuli, such as radiation, pH, and lipid-interfering drugs, may exert their effects on other cellular organelles, even though we have strived to analyze each individual phenomenon rigorously. We are confident that our work lays the foundation for further research in the field.

__A direct corollary of this issue is whether increased sEV secretion reflects more endosomes and lysosomes (e.g. LysoTracker-positive compartments) or whether sEV secretion is selectively up-regulated. __

Thanks to the Reviewer’ suggestion, we have analyzed both the lysosome and endosome contents in our experimental cell systems. These data are now included in the manuscript in Figure S8. We have observed that it is unlikely that lysosomes are directly involved in the LD – sEV connection. However, the expression of Rab7a, a regulator of the late endosomal pathway, correlated with the LD content of the cells and their sEV release. Therefore, the endosomal pathway might be a good candidate to contribute to this LD – sEV connection.

__At one point, authors argue that cells that secrete more sEVs also contain more MVBs, but this issue remains elusive. To what extent is the increase in LDs and sEVS correlated in particular with an increase in endosome-lysosomes, and ER-Golgi (LDs originate from the ER)? __

We thank the Reviewer for this comment. We agree that the analyses of sEVs secreted in the media might not reflect the MVB content in the cells. However, two experiments, one on Panc01 cells and another one on MCF7 cells, showed that the number of MVBs, assessed by confocal microscopy using CD63 staining (MCF7) or CD63 and Alix plasmids (PANC-01), was directly correlated with the number of released sEVs in the media (Figure Fig S3C and 4J).

In addition, we included additional experiments assessing the lysosome content in HT29 LDHigh and LDLowcells. Hereby, we confirmed that HT29 LDHigh cells showed a higher LD content than HT29 LDLow cells. Inversely, by studying the lysotracker area per cell, we showed that HT29 LDLow population has a higher lysosomal content as compared to their counterpart, HT29 LDHigh cells (test = Wilcoxon rank sum test with continuity correction_ W = 85127, p-value = 7.255e-07 for LDs and W = 49321, p-value = 1.14e-11 for Lysotracker). However, we could not demonstrate a clear correlation between the number of LDs in the cell and the lysotracker signal.

Finally, we have also studied the expression of GM130, a Golgi-shaping protein (Ref. 1) and Rab7, a late-endocytic protein (Fig S8C). While the expression of Rab7 (endosome) seemed to correlate with the LD and sEV contents, the expression of GM130 (Golgi) gave back no coherent results. Indeed, it was inversely correlated to the LD and sEV amount, in accordance with what was already reported elsewhere (Ref 2 and 3)

• Nakamura N. Emerging new roles of GM130, a cis-Golgi matrix protein, in higher order cell functions. J Pharmacol Sci. (2010) 112:255–64. Doi: 10.1254/jphs.09R03CR
• Lydia-Ann L.S. Harris, James R. Skinner, Trevor M. Shew, Nada A. Abumrad, Nathan E. Wolins. Monoacylglycerol disrupts Golgi structure and perilipin 2 association with lipid droplets.Doi.org/10.1101/2021.07.09.451829
• Alvin Kamili, Nuruliza Roslan, Sarah Frost, Laurence C. Cantrill, Dongwei Wang, Austin Della-Franca, Robert K. Bright, Guy E. Groblewski, Beate K. Straub, Andrew J. Hoy, Yuyan Chen, Jennifer A. Byrne; TPD52 expression increases neutral lipid storage within cultured cells. J Cell Sci 1 September 2015; 128 (17): 3223–3238. Doi: 10.1242/jcs.167692

  Authors conclude that the data with lipid inhibitors strengthen the connection between LDs and sEVs (Fig 2 and S2). However, is this regulation selective, or does it merely reflect the general effect of these inhibitors on membrane-related processes? The same comment applies to the role of iron metabolism after knockdown of ferritin heavy chain (Fig 3 and S3), acidic pH and X-ray radiation (Fig 4 and S4)____.

We thank the Reviewer for the interesting observation. As previously mentioned, we cannot rule out other potential contributors to the LDs-sEVs connection upon lipid inhibitor treatments and/or the others external stimuli applied to our cell systems.

The data presented in this manuscript merely represent a novel and unexplored (at least so far) organelle connection, direct or indirect, with a broad clinical implication. As the membrane-related processes (such as Endosomes, Golgi apparatus, Exosome (sEV) pathway, Lysosomes and Autophagosome) are all interconnected, in our opinion, it might be quite challenging to make such a definitive statement.

Such assertion would require extensive further investigation to relate each organelle to the LDs and/or sEVs. However, with our research, we hope to open the door to a new era of investigations regarding the sEV – LDs connection.

OTHER COMMENTS

1) Which cell line is used for sEV analysis (markers vs contaminants (Fig S1B)? In any case, the data should be shown for both cell types.

Our method to isolate sEVs is a standardized method that was already published by our group and collaborators in 2020 (M. Bordas, et al., Optimized Protocol for Isolation of Small Extracellular Vesicles from Human and Murine Lymphoid Tissues. Int J Mol Sci (2020) https:/doi.org/10.3390/ijms21155586.). This protocol was validated on human and mouse tissues, much more complex samples than cell culture supernatant.

Figure S1C was modified, as requested by the Reviewer, including new data for HT29, Panc01 and MCF7 cell lines to broaden the panel. Those results confirmed the good purity of sEV samples isolated from cell culture supernatant.

2) The Tsg101 blot is not impressive (Fig S1B): the difference between cells and sEVs is not easy to see. It would be nice if blots were quantified.

Indeed, the signal obtained for TSG101 for sEVs derived from Panc01 cell line is quite weak. It is important to remember that not all sEV markers are highly expressed in all cell lines and their derived sEVs. Some cell line-derived sEVs show a low or high expression of the diverse sEV markers. To answer the Reviewer #1’s comment, we quantified the expression of TSG101 in Panc01-derived sEVs. The quantification showed that TSG101 is 6.8 times more expressed on Panc01-dervied sEVs as compared to the cell line. However, since the expression is quite low, this quantification should be taken with some caution.

In light of the Reviewer ‘comment, we have performed the Western Blot analysis on other cell lines (HT29 and MCF7), and we have replaced TSG101 marker with CD9 marker (Figure S1C).

3) From Fig 1B it cannot be concluded that the size of sEVs ranges from 30 to 200nm: the micrograph only shows a few structures.

We appreciate the Reviewer's comment and have attempted to provide more clarity. Firstly, we want to highlight that TEM micrographs of sEVs typically show the donut shape, a unique feature of sEVs imaged with TEM, as well as a size range. In Figure 1B micrograph, the sEV size is approximately 100 nm. The size distribution of LoVo and HT29-derived sEVs can be observed from the NTA size measurements in Figure S1B. Indeed, the peak size is 148 nm for LoVo-derived sEVs and 135 nm for HT29, which aligns with the sEV sizes presented in Figure 1B. We have also included multiple micrographs here under. As the number of Supplementary Figures is already large, we have decided to not include those micrographs in the manuscript. The average size of LoVo-derived sEVs, based on TEM micrograph analysis, was 94 ± 41.10 nm, while the average size of HT29-derived sEVs was 76.41 ± 44.22 nm. The size discrepancy between the two methods (NTA versus TEM) can be ascribed to the dehydration step required for TEM, which results in a reduction of the actual sEV size.

4) HT29 cells contain far more LDs than LoVo cells (Fig 1A). Similarly, sEV proteins (CD63, CD81, CD9, Hsc-70) are more abundant in HT29 sEV____s____ than in LoVo sEVs (Fig 1D). However, the sEV preparation from HT29 cells contains only approx. 50% more total protein than LoVo sEVs (Fig S1D-E). Are sEVs prepared from LoVo cells far more contaminated with cell debris etc.. than sEV fractions from HT29 cells?

We are confident that our EV isolation method allows us to achieve high yield and excellent purity. It is possible that a lower number of sEVs in samples may lead to increased protein contamination during ultracentrifugation. However, size exclusion chromatography should minimize this protein contamination. It is important to note that the NTA method is significantly more sensitive and accurate than Qubit protein quantification. Consequently, protein concentration and particle concentration should not be directly compared.

5) LD staining should be shown for the corresponding populations of cells with high/low CD63 (Fig 1E). Cells in culture can be somewhat heterogeneous, but the difference between low and high CD63 is quite extreme (Fig 1E). Is such high heterogeneity also observed with other proteins of the endocytic and biosynthetic pathways? Authors conclude that cells containing high CD63 levels also contain more MVBs (Fig 1E): are all late endosomal proteins (e.g. LAMP1, RAB7) upregulated in cells with high CD63?

We thank the Reviewer for this comment, and we totally agree with the Reviewer that it would be better to have the LD and CD63 staining on the same images. Unfortunately, the staining for CD63 on LD540-sorted HT29 cells requires a permeabilization step that interferes with the cellular lipid part and could therefore negatively affect the LD imaging by confocal microscopy. To prove that the HT29 LDHigh and HT29 LDLowcontain high and low LD amount respectively, we sorted HT29 cells based on the LD content and, soon after, we observed them at the confocal microscopy. We thus added new images in Figure S1F, corresponding to the LD fluorescence detection. The readers will also appreciate the explanation regarding the inability of observing both LDs and CD63 staining on the same confocal images under the line 165 – 166:

“As the staining for CD63 required a permeabilization step, and therefore lipid digestion, it was not possible to assess both LDs and CD+MVBs on the same micrographs “.

In addition, we have added confocal images representing HT29 cells sorted based on their LD content and stained with Hoechst and Lysotracker. A quantification of the Lysotracker fluorescence per cell and the correlation with the number of LDs can also be appreciated in Figure S8A-B.

Finally, we performed Western Blot analysis to examine Rab7a expression under various conditions described in our manuscript (Figure S8C). In general, Rab7 expression corresponded with LD content, indicating that cells with high LD content exhibited higher Rab7 expression, while cells with low LD amount showed lower Rab7 expression, except for Triacsin-C. The Reviewer can now appreciate the quantification in the graphs provided below (not included in the manuscript).

Regarding the heterogeneity of LDs, CD63+MVBs, or lysotracker among the cell population, we have indeed noticed heterogeneity observable in these three types of staining in HT29, particularly in the HT29 LDHighpopulation.

6) Inhibitors of lipid synthesis reduce LD formation (Fig 2B), sEV production and CD63 / CD81/ CD9 secretion (Fig2C-D, Fig S2B). Are the cellular levels of these (and other endosomal) proteins also reduced after inhibitor treatment? Does the stimulation of LD formation with oleic acid also stimulate CD63 synthesis and sEV production?

We thank the Reviewer for this very interesting comment. To answer this question, we have added a supplementary figure (Figure S2A, S2B) showing the cellular expression of CD63 upon LD inhibition or stimulation.

During the planning of our experiments, we discussed about the possibility of using oleic acid to induce the formation of Lipid Droplets, which was ultimately not done. This is because the use of oleic acid would have more strongly stimulated the triglyceride pathway, as extensively discussed elsewhere (Mejhert N. et al., The lipid droplet knowledge portal: a resource for systematic analyses of lipid droplet biology, Developmental Cell, 2022). Since Lipid Droplets are made by cholesterol esters and triglycerides, we preferred to use other stimuli (hypoxia, radiation), all of them already discussed in literature, to induce both pathways simultaneously, resulting in the Lipid Droplet formation/induction.

7) It seems that pH and irradiation increase sEV markers far more significantly (Fig 4 B-C and Fig S4A-E) than FTH1 depletion decreases sEV markers (Fig 3 D-E). In fact, authors mention that they cannot exclude a contamination of sEVs with small apoptotic / autophagic vesicles after irradiation (Fig 4). To facilitate comparison, it would be nice to also show the number of secreted particles per cell (like after FTH1 depletion Fig 3D), as well as the distribution of possible contaminants (e.g. Fig S1). Also, authors state that the increase in the number CD63+ MVBs after irradiation is shown, but this is not the case.

We apologize to the Reviewer because, in fact, one figure was missing (Figure 4). We have rectified this by increasing the quality of Figure 4 and have added representative images for each acquisition of the number of MVBs, either positive for CD63 or Alix, in transfected Panc01 cells X-ray irradiated (8 Gy) or not (0Gy). In addition, a similar experiment was performed in MCF7 cells transduced with shRNA or shFTH1. CD63+ MVBs were assessed in both cell line and the number of CD63+ puncta (MVBs) were quantified by ImageJ. The results, although not significative, illustrated a trend for MCF7 shFTH1 to contain less CD63+ MVBs than MCF7 shRNA. Furthermore, the quantification of sEVs released in the conditioned media was performed in three independent experiments and demonstrated that significantly less particles (sEVs) were released by MCF7 shFTH1 than MCF7 shRNA.

8) Are the proteomic data (Fig 6) with LDlow and LDhigh cells obtained after cell sorting, as in Fig 1E? Did authors compare the proteome of LoVo and HT29 sEVs? How do the protein profiles (in particular proteins involved in lipid metabolism) obtained under different conditions compare with each other, in particular after irradiation (Fig4N) and knockdown of ferritin heavy chain (Fig 3, Fig S3)? It would also be interesting to compare these data with the data obtained in CR-CSCs culture under hypoxia (Fig S5). Are common proteins involved in sEV production and LD biosynthesis identified in the analysis of these biological processes? Is there a common set of proteins/genes revealed by this analysis, which may potentially control sEV production and LD biosynthesis?

We thank the reviewer for this interesting comment.

Proteomic analyses have been performed on the following conditions:

• Panc01 (0 Gy – 6 Gy – 8 Gy) for sEV samples
• MCF7 (shFTH1 and MCF7 shRNA)
• MCF7 (0 Gy and 6 Gy)
• MCF7 (Normoxia and Hypoxia)
• H460 (0 Gy and 6 Gy)

• H460 (Normoxia and Hypoxia) RNA sequencing was performed on the following conditions:

• CR-CSCs (#4, #8, #21) Based on all those data, we have analyzed the sEV pathway and how this pathway was modulated in the conditions with high LD content and low LD content. We therefore came up with several proteins, presented in Figure S7. Based on this analysis, we have decided to further investigate the role of RAB18, RAB5c and RAB7a in the connection between LDs and sEVs. Those additional results can be found in Figure 6 and Figure S7A (originally Figure 6). We have found that RAB5c, but not RAB7a or RAB18, seems to be a good candidate to intervene in the LD – sEV connection.

Minor comments

1) Some parts of the text are still a bit rough, and should be read and corrected carefully. For example: i) isn't it obvious that a common source of lipids builds up the membrane of sEVS, much like any other membrane (line 90, p.2); ii) what does this sentence mean: "LD have been considered as mere fat storage organelles for a long time, although important evidence could be traced back to the early 1960's". Important evidence for what? iii) why is the acronym AdExo used? iv) (line 138) the text should probably be "sEVs released during 72h were studied" and not "released sEVs were studied ... 72 h after seeding".

We apologize to the Reviewer if some parts of the paper were a bit rough. We have re-read the entire manuscript and corrected all the parts that needed revision work.

2) The captions are far too small in most figures and diagrams (for example ____X____ and Y axis in Fig 1C-D, text in Fig 1E; Fig S1; Fig 3C proteins in the heatmap).

We agree with the Reviewer. All images and their captions were properly revised.

3) The color code for LoVO and HT29 cells is reversed in Fig S1D-E

The mistake was corrected.

4) In Fig 1D, I cannot see CD81 in the LoVo blot.

In the image below, it is possible to see the LoVo blot.

5) Wording is not adequate in following sentences: "62.7% of proteins related to the exosomal pathway are downregulated in MCF7 shFTH1 cells" (line 233) and a few lines below: ".. the expression of almost all exosomal markers was downregulated in MCF7 shFTH1 cells" (line 239). Does 62.7% represent all proteins?

We apologize to the reviewer for the mistake. We rephrased this sentence.

6) In Fig 3E authors compare sEV markers secreted by cells treated with shFTH1 or control shRNA. The Anx5 and CD63 blots are not very convincing (quantification would be helpful).

We apologize to the Reviewer for this issue. These Western Blot analyses were performed only once, therefore a quantification in the manuscript would not be relevant. However, we report here the results of the quantification. The expression of Annexin V was 1.58 times higher in MCF7 shRNA than MCF7 shFTH1, while the expression of CD63 was 1.34 time higher in MCF shRNA as compared to MCF7 shFTH1.

7) The micrographs in Fig 4L are too small: gold particles cannot be seen, even in the high magnification views.

We thank the Reviewer for her/his comment. We have moved the micrograph and the quantification histogram to the Figure S6. Now, it is possible to discriminate easily gold nanoparticles.

8) The micrographs showing ALIX and CD63 (Fig 4J) in irradiated and unirradiated Panc01 cells should be shown for comparison.

We followed the Reviewer’ suggestion as it is possible to note in the Figure below.

Reviewer #2

This manuscript describes a relationship between lipid droplet presence in cells and small EV secretion. First, correlations are done between number of lipid droplets and numbers of EVs secreted. Then chemical inhibitors of lipid droplet biosynthesis pathways were shown to reduce small EV secretion. Then various processes known to target lipid droplets, including iron metabolism, irradiation, hypoxia, low pH are used to show concordant effects on lipid droplets and small EV secretion. Proteomic analysis of EVs and cells subjected to some of the treatments are also performed. Overall, it is an interesting line of investigation and the data overall seem solid. Several flaws exist, which can probably be fixed. These include the use of different cell lines for different experiments. It makes it a bit difficult to connect everything together. It could be fixed by adding some extra cell lines to some experiments - for example taking the MCF7 and Panc-01 cells for which proteomics was performed and redoing some of the correlative and causative experiments from Figs 1 and 2.

We appreciate the Reviewer's insightful observation. Following her/his suggestion, we have conducted additional experiments on MCF7, H460 and PANC-01 cell lines to enhance data consistency and facilitate a smoother transition between different sections of the paper.

It also would be good to have some more direct evidence of the connection between lipid droplets and EV secretion - one could argue that this was already done in Fig 2 with the chemical inhibitors, I wonder if there is a genetic way to do it too?

We totally agree with the Reviewer. Indeed, starting from our proteomic data we highlighted some genes belonging to the RAB family as potential candidates to interfere with the LD – sEV connection. The Reviewer can now appreciate in Figure 6 and Figure S7, the results from the additional experiments we carried out on RAB5c, RAB7a and RAB18 silencing in HT29 cells. The former Figure 6 has been moved in the Supplementary part (Figure S7).

Some tightening up of the writing (especially the Discussion) and the resolution of the figures would also improve the manuscript.

We apologize to the Reviewer for this issue. We have now re-prepared all Figures by increasing their resolution, as well as reviewing the entire manuscript with the aim of making the reading smoother and simpler.

__Overall, it is a nice piece of work but there are many minor things to be fixed. __

__Specific Comments: __

The sentence in the Introduction: "The non-endosomal pathway generates sEVs devoid of CD63, CD81 and CD9 or sEVs enriched in ECM and serum-derived factors (7)." is not well-supported and should be removed. The idea that you can classify membrane of origin based on markers has not been proven, but rather assumed.

We agree with the Reviewer. We have rephrased the sentence.

We thank the Reviewer for this comment. In response to this, we have generated correlation graphs for several of our experiments:

• HT29 (CTL – Triacsin-C - PF-06424439) in Figure 2E
• PANC-01 (CTL – 2 – 4 – 6 – 8 Gy) in Figure 4K
• CR-CSCs (#4, #8, #21) in Figure 5E

  __The Method used for EV purification should be stated in the Results rather than referring to a reference and a Supplemental Figure (S1A) that is too low of a resolution to see. __

Our method to isolate sEVs is a standardized methods that was already published by our group and collaborators in 2020 (M. Bordas, et al., Optimized Protocol for Isolation of Small Extracellular Vesicles from Human and Murine Lymphoid Tissues. Int J Mol Sci (2020) https:/doi.org/10.3390/ijms21155586.). This protocol was validated on human and mouse tissues, much more complex samples than cell culture supernatant.

In regard to the Reviewer’s comment, we have added a better description of the protocol in the Results part, referring to the Material and Method. For this reason, we decided to keep the sEV protocol in the SI section. We apologize for the low quality of the Figure S1. In agreement with the Reviewer suggestion, we have modified the image by increasing its quality.

__Fig 1B would be better to have an image in which the EVs are not aggregated. __

We thank the Reviewer for this comment and have modified the Figure accordingly.

__Fig 3 is interesting but jumps cell lines. For better continuity, some of the experiments from Figs 1 and 2 should be repeated in the MCF7 cells to connect with the proteomics. __

In agreement with the Reviewer’ comment, we decided to perform additional experiment on MCF7, using Triacsin-C. The Reviewer can now appreciate the results in Figure 2F, Figure 2G and Figure S2E.

__Fig 3C is too low resolution to read, please export at higher resolution. __

We are sorry for the low-quality Figure. We have modified the image accordingly.

__Please provide all the raw proteomics data as a supplementary spreadsheet____. __

We have provided all the raw data regarding our proteomic analyses.

__Fig 4 panels are low resolution __

We apologize for the low-resolution Figure. We have modified the figure by increasing the quality.

Fig 4 again adds new cell lines with H460 and Panc-01

We thank the reviewer for this comment. In this regard, we have performed additional experiment:

• Western Blot: comparison cellular and exosomal markers (Figure S1C)
• MCF7 (CTL - Triacsin) (Figure 2F, Figure 2G and Figure S2E)
• Western Blot: analysis of RAB7a, GM130

  __The images corresponding to 4J should be shown in a Supp Figure somewhere __

We thank the reviewer for pointing out this oversight. We have added the confocal images corresponding to the Figure 4J below the quantification.

The statements: "In addition, the exosomal nature of Panc01-derived vesicles was demonstrated by an analysis of CD63+ or Alix+ multivesicular bodies (MVBs) in unirradiated (0 Gy) or irradiated (8 Gy) pancreatic cancer cells (Fig 4J). Moreover, we confirmed a clear correlation between cellular LD content and sEV biogenesis, as represented in Fig 4K." are overly conclusive. For 4J, one can make a statement about the MVBs but not the EVs as that's not what was measured there. Likewise for 4K, what was measured was how many EVs were released not how many were formed. While the data are suggestive of alteration of exosome biogenesis, they are not conclusive.

We agree with the reviewer and have performed the necessary changes in the manuscript. The reviewer can see the changes under the lines 282 – 284:

“In addition, the analysis of CD63+ or Alix+ multivesicular bodies (MVBs) in unirradiated (0 Gy) or irradiated (8 Gy) pancreatic cancer cells revealed an increased number of MVBs after irradiation (Fig____ure 4J).”

__Western blot is always capitalized by convention - Western not western. __

We have corrected it accordingly.

__Fig 5A is too small and low resolution - suggest eliminating and just put info in methods. __

We are sorry for the low-resolution image. We have followed the Reviewer suggestion. The graphical method has been now moved to the Supplementary Figure S6.

Fig 5G, many of the genes shown are frequently EV cargoes but most not involved in exosome biogenesis - not sure where the label of Exosome pathway came from but it is not very compelling. Only ANXA2, Arf6, and Rab5C seem related and they are barely elevated.

We completely agree with the Reviewer's comment. As a result, we have revised the heatmap title to "Exosomal Cargoes and Pathways" instead of "Exosomal Pathway".

__Most main figures and all supplementary figures are extremely low res - please fix. __

We are very sorry for the low-quality figures. We have revised all Figures (main text and SI) by increasing their quality.

__Fig 6 is first mentioned in the Discussion - it should be described in the Results before that (or alternatively removed). __

We agree with the Reviewer. Our initial idea was to mention perspectives of analyses that could be carried ulteriorly. Nevertheless, we have performed additional experiments in order to get insight on the mechanism involved in the LD – sEV connection. Indeed, based on our proteomic data, we have analyzed the sEV pathway and how this pathway was modulated in the conditions with high LD content and low LD content. We therefore came up with several proteins, presented in Figure S7A (originally Figure 6). Based on this analysis, we have decided to further investigate the role of RAB18, RAB5c and RAB7a in the connection between LDs and sEVs. Those additional results can be found in Figure 6 and Figure S7 in the Results section. We have found that RAB5c, but not RAB7a or RAB18, seems to be a good candidate to intervene in the LD – sEV connection.

Table S1, also first mentioned in the Discussion, is missing. Either describe in the Results section or remove the callout to it.

Our apologies for that. The Table S1 has been now mentioned in the Results section and has been properly uploaded.

__The discussion is too dense with too many trains of thought, often many different directions in the same paragraph. It needs to be streamlined, with a central thought for each paragraph and good transitions between the paragraphs. __

We apologize to the Reviewer if the Discussion part was a bit confusing. We rewrote the paragraph, streamlining it and making the transitions between its paragraphs smoother.

Reviewer #2 (Significance (Required)):

__ Strengths of this manuscript are the interesting connection between lipid droplets and exosomes and the number of experiments to address it. __

__ Limitations: use of different cell lines for different figures, overall descriptive nature with regard to direct demonstration of connection to lipid droplets -- it's kind of done in Fig 2, but could be possibly bolstered. __

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

Learn more at Review Commons

---

Reply to the reviewers

We thank the reviewers for their thoughtful comments. We were delighted that the reviewers found our manuscript and results “solid”, “important”, “well-written”, “thoughtful”, “critical addition to the literature”, that the “design of (these) experiments is high in quality” and “conclusions are convincing and the experiments are well executed”.

We were thrilled the reviewers appreciated that “this manuscript provides solutions to technical limitations to observe mRNA in vivo” by approaching such limitations “in a thoughtful study where many of the salient features of MS2 epitope tagging are systematically measured” and “it provides a greatly improved tool to track mRNA by live imaging” that “also alerts of experimental noise that can be found and can be specific for each gene/transcript”

We will address all the concerns raised by the reviewers. Most of the comments concern text edits. In addition, we will add the following to the Results section:

1. Quantitation of observed phenotypes in Figures 1C-D and 2C-D;

2. Quantitation of cytoplasmic transcripts in Figure 1G-L.

Quantitation will be performed as previously done in Tocchini et al., 2021.

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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements [optional]

We thank the Reviewers for their helpful and constructive comments. In response to these suggestions we have performed new experiments and amended the manuscript, as we describe in our detailed response below.

2. Point-by-point description of the revisions

Reviewer #1

1. The Reviewer notes that while our analysis of centrosome size was comprehensive, we provided no analysis of centrosomal MTs, pointing out that while centrosome size declines as the embryos enter mitosis, the ability of centrosomes to organise MTs might not. This is a good point, and we now provide an analysis of centrosomal-MT behaviour (Figure 2). We find that there is a dramatic decline in centrosomal MT fluorescence at NEB, although the pattern of centrosomal MT recruitment prior to NEB is surprisingly complex.

The Reviewer questions how PCM client proteins can be recruited in different ways by the same Cdk/Cyclin oscillator. We apologise for not explaining this properly. It is widely accepted that Cdk/Cyclins drive cell cycle progression, in part, by phosphorylating different substrates at different activity thresholds (e.g. Coudreuse and Nurse, Nature, 2010; Swaffer et al., Cell, 2016). Moreover, it is also clear that Cdk/Cyclins can phosphorylate the same protein at different sites at different activity thresholds (e.g. Koivomagi et al., Nature, 2011; Asafa et al., Curr. Biol., 2022; Ord et al., Nat. Struct. Mol. Biol., 2019). Thus, we hypothesise that rising Cdk/Cyclin cell cycle oscillator (CCO) activity phosphorylates multiple proteins at different times and/or at different sites to generate the complicated kinetics of centrosome growth. We now explain this point more clearly throughout the manuscript.

The Reviewer is puzzled as to how we conclude that Cdk/Cyclins phosphorylate Spd-2 and Cnn at all the potential Cdk/Cyclin phosphorylation sites we mutate in our study. The Reviewer is right that we cannot make this conclusion, and we did not intend to make this claim. As we now clarify (p11, para.1), although it is unclear if Cdk/Cyclins phosphorylate Spd-2 or Cnn on all, some, or none of these sites, if either protein can be phosphorylated by Cdk/Cyclins, then these mutants should not be able to be phosphorylated in this way—allowing us to address the potential significance of any such phosphorylation. We now also note that several of these sites have been shown to be phosphorylated in embryos in Mass Spectroscopy screens (Figure S6).

The Reviewer highlights differences in how Spd-2 and Cnn help recruit γ-tubulin to centrosomes (Figure 6). They ask for a more detailed description, and are puzzled as to how this is compatible with direct regulation by a single oscillator. We now explain our thinking on this important point in much more detail. It appears that Spd-2 helps recruit γ-tubulin throughout S-phase, while Cnn has a more prominent role in late S-phase (Figure 6). This is consistent with our overall hypothesis of CCO regulation, as we postulate that low-level CCO activity promotes the Spd-2/γ-tubulin interaction in early S-phase, while higher CCO activity promotes the Cnn/γ-tubulin interaction in late-S-phase, potentially explaining the increase in the rate of γ-tubulin (but not γ-TuRC) recruitment we observe at this point (see minor comment #1, below, for an explanation of the various γ-tubulin complexes in flies). This is consistent with recent literature showing that CCO activity promotes γ-tubulin (but not γ-TuRC) recruitment by Cnn/SPD-5 in worms and flies (Ohta et al., 2021; Tovey et al., 2021).

The Reviewer was not convinced by our model (Figure 8, now Figure 9), raising two major concerns. First, they were unsure how a single oscillator could generate different patterns of protein recruitment. We addressed this in point #2 and #4, above, where we explain how different thresholds of CCO activity trigger different events, so there is no expectation that we should observe steady changes in recruitment over time as CCO activity rises. Second, they questioned how modest levels of Cdk/Cyclin activity can promote recruitment, while high levels of activity can inhibit recruitment. In point #1, above, we cite several examples where such positive and negative regulation by different Cdk/Cyclin activity levels have been described. We also now explain throughout the manuscript why this hypothesis provides a plausible explanation for our results: with moderate CCO activity promoting Spd-2-dependent PCM-client recruitment in early S-phase; higher CCO activity promoting a decrease in Spd-2 recruitment in mid-late-S-phase (so centrosomal Spd-2 levels decline); and even higher levels of CCO activity leading to a decrease in the interactions between the client proteins and the Spd-2/Cnn scaffold as the embryos enter mitosis (so the client proteins are rapidly released from the centrosome).

The Reviewer also raised the important point here that our model does not explain why the mutant forms of Spd-2 and Cnn accumulate to higher levels at the start of S-phase, and not just at the end of S-phase/entry into mitosis. We apologise for not explaining this properly. The accumulation of the mutant proteins (particularly Spd-2, Figure 5C) in early-S-phase occurs because the excess mutant protein that accumulates at centrosomes in late-S-phase/mitosis is not removed properly from centrosomes during mitosis (presumably because there is insufficient time). Thus, centrosomes still have too much mutant Spd-2 at the start of the next S-phase. We show this in Reviewer Figure 1 (attached to this letter), which tracks Spd-2 behaviour further into mitosis, and now explain this in more detail in the text (p12, para.1).

The Reviewer questions how the CCO can both induce centrosome growth and also switch it off, as it is unclear how an oscillator that only phosphorylates sites to decrease centrosome binding could also promote growth. They ask if we can identify and mutate any Cdk/Cyclin sites in centrosome proteins that promote centrosome recruitment. As we now clarify, we did not intend to claim that the CCO only phosphorylates sites that decrease the centrosome binding of proteins, although we do hypothesise that such phosphorylation is important for switching off centrosome growth in mitosis. In addition, we hypothesise that moderate levels of CCO initially promote centrosome growth, and our data suggests that the CCO does this, at least in part, by promoting Polo recruitment (Figure 8). We speculate that the CCO phosphorylates specific Polo-box-binding sites in Ana1 and Spd-2, the main proteins that recruit Polo to centrioles. We agree that identifying these sites is an important next step, but it is complicated as our studies indicate that multiple sites contribute in a complex manner. Importantly, it is well established that the CCO triggers centrosome growth as cells prepare to enter mitosis, so our hypothesis that moderate levels of CCO activity initiate centrosome growth is not new or controversial.

Minor Comments

1. The reviewer asks how we explain the different incorporation profiles we observe for the different subunits of the γ-tubulin ring complex. We apologise for not discussing this point. In flies there is a “core” γ-tubulin-small complex (γ-TuSC) and a larger γ-tubulin-ring complex (γ-TuRC) that contains the Grip71, Grip75 and Grip128 subunits we analyse here (Oegema et al., JCB, 1999). The γ-TuSC functions independently of the γ-TuRC so γ-tubulin and γ-TuRC components can behave differently.

The Reviewer questions why we claim an “inverse-linear” relationship between S-phase length and the centrosome growth rate when the relationship is not linear (Figure 3, now Figure S3). I was originally confused by this as well but, mathematically, a linear relationship means y is proportional to x, whereas an inverse-linear relationship means y is proportional to 1/x. Thus, an inverse-linear relationship between x and y does not plot as a straight line, but rather as the curves we show on the graphs. We now explain this in text (p9, para.2).

Reviewer #2

This Reviewer found the manuscript hard to follow, so we are very grateful that they took the time to try to understand it. We agree that the subject matter is complicated, and that our presentation was not always helpful. The Reviewer’s comments have been very useful in helping us to identify (and hopefully improve) areas of particular difficulty.

Major points:

1. The Reviewer highlights that the two experimental approaches underpinning our main conclusions are problematic: (1) Experiments with mutants of Spd-2 and Cnn that theoretically cannot be phosphorylated by Cdk/Cyclins are hard to interpret as these mutations may have other effects; (2) It is unclear whether reducing Cyclin B levels reduces peak CDK activity or simply slows the time it takes to reach peak levels. They suggest a more direct test of our model would be to analyse PCM recruitment in embryos arrested in S-phase or mitosis. (1) We agree that the mutations designed to prevent Cdk/Cyclin phosphorylation could perturb function in other ways, but this is true for any such mutation, and there are many papers that infer a function for Cdk/Cyclin phosphorylation from such experiments. Importantly, the centrosomal accumulation of the phospho-null mutants actually slightly increases compared to WT (Figure 5C and I), and we now show that the centrosomal accumulation of a phosphomimicking Spd-2-Cdk20E mutant slightly decreases (Figure S8). We now acknowledge the potential caveat of a non-specific perturbation of protein function, but feel that the reciprocal behaviour of the phospho-null and phospho-mimicking mutants somewhat mitigates this concern (p12, para.2). (2) Fortunately, and as we now clarify, it has recently been shown that reducing Cyclin levels does not reduce peak Cdk activity, but rather slows the time it takes to reach peak activity (Figure 2A, Hayden et al., Curr. Biol., 2022). Thus, the cyclin half-dose experiments provide an excellent alternative test of our hypothesis as they show that the WT proteins can exhibit similar behaviour to the mutants if the rate of Cdk/Cyclin activation is slowed. We feel the evidence supporting our hypothesis is strong enough that it warrants serious consideration. The suggestion to look at PCM recruitment in embryos arrested in either S-phase or M-phase is a good one, but these experiments produce complicated data. In M-phase arrested embryos, for example, Cnn levels continue to rise (see Figure 1G, Conduit et al., Dev. Cell, 2014), but the other PCM proteins do not (unpublished); in S-phase arrested embryos (arrested by mitotic cyclin depletion) centrosomes continue to duplicate, but now do so asynchronously, greatly complicating the analysis (McCleland and O’Farrell, Curr. Biol.., 2008; Aydogan et al., Cell, 2020). The centrosomes that don’t duplicate, however, reach a constant steady-state size (where the rate of centrosome protein addition is balanced by the rate of loss). These observations are consistent with our recent mathematical modelling of mitotic PCM assembly (Wong et al., 2022) if we additionally account for cell cycle regulation (which was not considered in our original model). We believe such analyses are beyond the scope of the current paper and we plan to publish a second paper incorporating our new hypothesis into our mathematical modelling.

The Reviewer questions whether our methods accurately measure centrosomal protein accumulation, pointing out that γ-tubulin and Grip128 occupy different centrosomal areas—which should not be possible if they are part of the same complex. They suspect that our use of different transgenes with different promotors could explain these differences. As we should have described (see point #1 in our response to the minor comments of Reviewer #1), γ-tubulin exists in two complexes in flies, only one of which contains Grip128, so γ-tubulin and Grip128 exhibit different localisations. Moreover, as we now show (Figure S2), using different promotors does not seem to make a difference to overall recruitment kinetics. Thus, we are confident that our methods measure centrosome protein recruitment dynamics accurately.

The Reviewer is concerned that our measurements of centrosome size based on fluorescence intensity (Figure 1) and centrosomal area (Figure S1) do not always match. They suggest a potential reason for this is that proteins are not uniformly distributed within centrosomes, and this may impact our ability to measure protein accumulation based on 2D projections (noting, for example, that Polo and Spd-2 are concentrated at centrioles and in the PCM, potentially explaining the different shape of their growth curves compared to the client proteins). When the centrosome-fluorescence-intensity and centrosome-area recruitment profiles of a protein do not match, the average “centrosome-density” of that protein must be changing over time. In some cases, we understand why density changes. Cnn, for example, stops flaring outwards on the centrosomal MTs during mitosis so its centrosomal area decreases even as its fluorescence intensity increases (leading to an increase in its centrosomal-density). We agree (and now discuss—p19, para.3) that the prominent accumulation of Spd-2 and Polo at centrioles could help to explain why Spd-2 and Polo accumulation dynamics differ from the client proteins.

Other points:

The Reviewer suggests it would be good to know how much Polo at the centrosome is active____. We agree, but although commercial antibodies against PLK1 phosphorylated in its activation loop work in cultured fly cells, we cannot get them to work in embryos. Moreover, the recruitment of Polo/PLK1 to its site of action by its Polo-Box Domain is sufficient to partially activate the kinase independently of phosphorylation (Xu et al., NSMB, 2013). Thus, it seems likely that all the Polo/PLK1 recruited to centrosomes will be at least partially activated, even if it is not necessarily phosphorylated on its activation loop.

The Reviewer asks if it is clear that less Spd-2 and Cnn are recruited to centrosomes in the half gene-dosage embryos. We apologise for not mentioning that this is indeed the case. We showed this previously for Cnn (Conduit et al., Curr. Biol., 2010) and we now state that this is also the case for Spd-2. We do not show the Spd-2 data as we plan to publish a comprehensive dose-response curve of Spd-2 (and Cnn) recruitment in our next modelling paper.

Would it not be relevant to examine Polo ½ dosage embryos? We do have this data (Reviewer Figure 2), attached to this letter, but it is quite complicated to interpret (as we explain in the legend). We feel it would be more appropriate to include this in our next modelling paper where we can properly explain the behaviours we observe. Publishing this data here would distract from our main message without changing any of our conclusions.

The Reviewer asks why the non-phosphorylatable Spd-2 protein is also present at higher levels on centrosomes at the start of S-phase (not just the end of S-phase). This was also raised by Reviewer #1 (point #5), so please see the second paragraph of our response there.

Minor/Discussion Points:

We thank the Reviewer for highlighting that absolute and relative centrosome size control are different things and we have amended the manuscript accordingly.

The Reviewer questions whether it is accurate to describe Spd-2 and Polo as scaffold proteins, noting that only Cnn has been shown to have scaffolding properties. There is strong evidence that Spd-2 has Cnn-independent scaffolding properties in flies (e.g. Conduit et al., eLife, 2014), but this is a fair point for Polo. We think it is justified to separate Polo from other client proteins as Polo is essential for scaffold assembly, whereas other client proteins are not. We now define our scaffold/client terminology to avoid confusion (p4, para.3).

The Reviewer highlights several points related to differences in recruitment kinetics (also touched on in points #2 and #3, above), noting we don’t discuss properly the idea of two different modes of PCM recruitment. These are all good points, largely addressed in our response to points #2 and #3, above. We now discuss much more prominently the two different modes of client protein recruitment throughout the manuscript.

As we now clarify, in all our experiments we use centrosome separation and nuclear envelope breakdown (NEB) to define the start and end of S-phase, respectively.

The Reviewer quotes the landmark Woodruff paper (Cell, 2017) as showing that the ability to concentrate client proteins (including ZYG-9, the worm homologue of Msps) is an intrinsic property of the PCM scaffold, so how do we explain that Msps departs prior to NEB while Cnn continues to accumulate? It is indeed a striking observation of our study that all PCM client proteins (not just Msps) start to leave the centrosome prior to NEB, even as Cnn levels continue to accumulate. Our hypothesis is that this ‘leaving’ event is triggered by a threshold level of Cdk/Cyclin activity—explaining why these client proteins all start to leave the PCM at the same time (just prior to NEB) irrespective of nuclear cycle length. This is not incompatible with the Woodruff paper, which did not attempt to reconstitute any potential regulation by Cdk/Cyclins in their in vitro studies.

The Reviewer questions why Spd-2 that cannot be phosphorylated by Cdk/Cyclins (Spd-2-Cdk20A) accumulates abnormally at centrosomes in late S-phase, yet γ-tubulin (which is recruited by Spd-2) seems to leave centrosomes more slowly in the presence of the mutant protein. As we now explain more clearly, there is no contradiction here. Spd-2-Cdk20A accumulates to abnormally high levels in late-S-phase/early mitosis (Figure 5C), and this reduces the γ-tubulin dissociation rate, as we would predict (Figure 7B, right most graph). It does not “prevent” dissociation, however, (as the Reviewer seems to suggest it should?), but this is probably because these experiments have to be performed in the presence of large amounts of the WT Spd-2 (Figure 5A).

The referencing error has been corrected.

The Reviewer asks why in Figure 1 not all of the centrosome proteins could be followed for the full time period (as we mention in the legend, but do not explain). There are different reasons for different proteins: (1) Polo cannot be followed in mitosis as it binds to the kinetochores, making it impossible to accurately track centrosomes (so the data for mitosis is missing for Polo); (2) Cnn exhibits extensive flaring at the end of mitosis/early S-phase (Megraw et al., JCS, 1999), so we cannot track individual separating centrosomes labelled with NG-Cnn in early S-phase until they have moved sufficiently far-apart (so the early S-phase time-points are missing for Cnn); (3) In addition, several of the client proteins bind to the mitotic spindle, so although we can still track and measure the centrosomes in late mitosis in the graphs, we don’t show pictures of these late mitosis centrosomes in the montage in Figure 1A as the images look a bit odd. We now explain these reasons in the Materials and Methods.

We now indicate that nuclear cycle 12 (NC12) is being analysed in Figures 4-8.

The reviewer questions why we don’t show the decrease rate for γ-tubulin in Figure 6 (the Spd-2 and Cnn half-dose experiments), when we do show it in Figure 7 (the Spd-2 and Cnn Cdk-mutant experiments), suspecting that it is slowed in both cases. The reviewer is correct and we now show this data for both sets of experiments.

We have corrected the labelling error in Figure S1.

The Reviewer suggest moving some of the data from the main Figures, and the entirety of Figures 2 and 3 to the Supplemental Information. We understand this point, and agree that the amount of data presented in Figures 1-3 is somewhat overwhelming. We have played around with the Figures a lot—in particular trying to show a few examples of the data and moving the rest to Supplementary—but it is hard to pick a “typical” example, and the power of comparing the behaviour of so many different centrosome proteins is somewhat lost. We have tidied up several Figures and, as a compromise, we keep Figure 2 (now Figure 3) in the main text, but have moved Figure 3 to Supplementary (now Figure S5).

The Reviewer suggests that we should repeat the analysis of Spd-2, Polo and Cnn dynamics that we show here, as we already presented this data in a previous publication (Wong et al., EMBO. J, 2022). We understand this point, but feel this would be a less accurate comparison, as essentially all of the data shown in Figure 1 was obtained several years ago during a contiguous ~6month period. Since then, the lasers and software on our microscope system have been updated, so it would probably be less fair of a comparison to obtain new data for a subset of these proteins (and it seems overkill to perform the entire analysis again). We clearly state that this data has been presented previously, so we hope the Reviewer will agree that it is acceptable to present it again here so readers can more easily compare the data.

Reviewer #3

This Reviewer is broadly supportive of the manuscript, but to publish in a prestigious journal they think additional experimental evidence will be required to support our hypothesis.

The Reviewer notes that our only evidence that Cdk/Cyclins directly phosphorylate Spd-2 comes from our analysis of the Spd-2-Cdk20A mutant, as the effect of reducing Cyclin B dosage on WT Spd-2 behaviour is very modest. They request that we analyse the behaviour of a Spd-2-Cdk20E phospho-mimicking mutant. The effect of halving the dose of Cyclin B on Spd-2 behaviour is modest, but this is what we would predict as all we are doing in this experiment is slowing S-phase by ~15%, so Spd-2 should accumulate at centrosomes for a slightly longer time and to a slightly higher level (as we observe, Figure 5E). A great advantage of the early fly embryo system is that we can compare the behaviour of many hundreds of centrosomes, so even subtle differences like this are usually meaningful. To illustrate this point, we have now repeated the Spd-2 analysis in WT and CycB1/2 embryos (but now using a CRISPR/Cas9 Spd-2-NG knock-in line) and we see the same subtle differences (Figure S9). In addition, as requested, we have now analysed the behaviour of a Spd-2Cdk20E mutant protein using an mRNA injection assay (as it would have taken too long to generate and test new transgenic lines). In this assay we injected embryos with mRNA encoding either WT Spd-2-GFP, Spd-2-Cdk20A-GFP or Spd-2-Cdk20E-GFP. The mRNA is quickly translated, and we computationally measured the fluorescence intensity of the centrosomes in mid-S-phase (i.e. at the Spd-2 peak) (Figure S8). This analysis confirms that Cdk20A accumulates to slightly higher levels, and reveals that Cdk20E accumulates to slightly lower levels, than the WT protein. Together, these new experiments strongly support our original conclusions.

The Reviewer notes that we propose that the CCO initially promotes centrosome growth by stimulating Polo recruitment to centrosomes, but states that we only provide indirect evidence for this by showing that centrosomal Polo levels are strongly reduced in Cyclin B half-dose embryos. They suggest we determine Spd-2 levels in Polo half-dose embryos, and/or the centrosome levels of mutant forms of Spd-2 that cannot be phosphorylated by Polo. We believe the Cyclin B half-dose experiment provide direct support for our hypothesis that Cdk/Cyclin activity influences Polo recruitment (Figure 8), although, clearly, we have not identified the mechanism. We do, however, suggest a plausible mechanism: Ana1 and Spd-2 are largely responsible for recruiting Polo to centrosomes, and we have previously shown that several of the potential phosphorylation sites in these proteins that help recruit Polo to centrosomes are Cdk/Cyclin or Polo phosphorylation sites (Alvarez-Rodrigo et al., eLife, 2020 and JCS, 2021; Wong et al., EMBO J., 2022). We are currently testing this hypothesis, but progress is slow as it is clear that multiple sites in both proteins can influence this process.

As the Reviewer requests, we have now also examined how Spd-2 and Cnn behave in Polo half-dose embryos (Reviewer Figure 2, attached to this letter). As we describe in the Figure legend, this data is informative, but is complicated. With relatively minor, but mechanistically important, tweaks to our previous mathematical modelling we can explain these behaviours, but introducing such a significant mathematical modelling element would be beyond the scope of this paper. As described above, these findings will form the basis of a follow-up paper that is more mathematically oriented.

It is a great idea to look at mutant forms of Spd-2 that cannot be phosphorylated by Polo, but the consensus Polo phosphorylation site (N/D/E-X-S, with the N/D/E at -2 and the S at 0 being preferences, rather than a strict rule) is less well-defined than the consensus Cdk/Cyclin phosphorylation site (where the Pro at -1 is essentially invariant). Thus, we cannot accurately predict which sites would need to be mutated to generate such a mutant.

The Reviewer requests that we analyse the behaviour of TACC in embryos expressing the Spd-2-Cdk20A and Cnn-Cdk6A (as we do in Figure 7 for γ-tubulin). This is a reasonable request, but we prefer not to show this data as we have recently identified an interesting interaction between TACC, Spd-2 and Aurora A that will be the subject of another paper we hope to submit shortly. This data is hard to interpret without explaining these interactions properly, which is beyond the scope of the current manuscript.

We hope the Reviewers will agree that these changes have improved the manuscript substantially, and that it is now suitable for publication. We would like to thank them again for taking the time to read this rather complicated paper so thoroughly.

We look forward to hearing from you.

Yours sincerely,

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

Learn more at Review Commons

---

Reply to the reviewers

A complete response to all the reviewers´ comments and suggestions has been uploaded as a separate file.

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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements [optional]

We thank the reviewers for their appreciation of the interest, novelty and quality of our study, and their useful feedback to improve its presentation.

We have revised the manuscript addressing all the points they made, as detailed below, section by section, following the organization in the reviews. The corresponding changes are highlighted in yellow (new text) or crossed out (deleted text) in our revised manuscript.

In case it is useful for the editor to check how each individual point was addressed, we also have extracted from the reviews each individual reviewer’s comment and our direct response, listed as bullet points at the end of this text.

2. Point-by-point description of the revisions

I - General criticisms

Reviewer #1: My main criticism is unfortunately inherent to the approach: comparative studies are absolutely critical, but they can only provide a very sparse sampling of diversity. Fortunately, thanks to high-throughput sequencing, bioinformatic analyses can now be performed on a large number of species, but experimental validation is typically restricted to two or three species. The consequence of this for the present manuscript is that while the functional conservation of the Gwl site is convincingly shown, the exact mechanisms responsible for the reduced effect of PKA phosphorylation remain relatively vaguely defined. Indeed, in their Discussion the authors list a number of experimental approaches to address this - but I understand that these would all involve substantial efforts to address. In particular, testing chimeric constructs around the consensus PKA site and from multiple species could be very informative.

We completely agree with the reviewer that comparative approaches are critical to understanding biological mechanisms, and are excited by the increasing possibilities to perform not only sequence and descriptive comparisons but functional studies across a range of emerging model organisms. We hope that more and more researchers in cell and molecular biology will profit from experimental tools and techniques now available in such species, and to pioneer new ones. Of course, and he/she rightly points out, conclusions are currently limited by the number of species studied, but comparisons between two judiciously chosen species can already be very informative. Thus, in our study, the use of Xenopus and Clytia allowed us to make significant progress towards our main objective of understanding the cAMP-PKA paradox in the control of oocyte maturation; specifically by showing both that PKA phosphorylation of Clytia ARPP19 is lower in efficiency and that the phosphorylated protein has a lower effect on oocyte maturation than the Xenopus protein. As the reviewer points out, unravelling the exact mechanisms underlying these differences will require a large amount of additional work and is beyond the scope of the current study. Actually, we have embarked on several series of experiments to this end using some of the approaches listed in the Discussion. Specifically, we are testing the biochemical and functional properties of chimeric constructs containing the consensus PKA site from various species. This is a substantial undertaking which will require one to two years to complete, but is already giving some very interesting findings.

Reviewer #1: The figures and text could be slightly condensed down to about 6 figures.

We have reduced the number of figure panels but we prefer to maintain the number of figures, because the experimental data presented in them is essential to the interpretation of our results and the overall conclusions of the article. If the journal editor would like us to reduce the number of figures, we could do this by displacing Figure 4 and some panels of other figures (to then fuse some of them) to supplementary material, but this would be a pity.

______________II - Abstract__

As recommended by Reviewer #2, we have reworked the Abstract to make it more accessible to new readers, attempting to bring out more clearly and simply the main results and conclusions of the study. We correspondingly simplified and shortened the title of the article. Changes: Page 2.

____________III- Introduction points__

Reviewer #2: I believe that it would be interesting to include some time-references when introducing the prophase arrest of Clytia and Xenopus oocytes. How long is prophase arrest in Xenopus compared to Clytia or other organisms? How can this affect the prophase arrest mechanisms? It seems that the prophase arrest in Xenopus oocytes is found to be significantly more prolonged compared to Clytia and various other organisms, and also meiotic maturation proceeds much more rapidly in Clytia than in Xenopus. This should be indicated in the introduction with a short introduction of why, and not others, were these species chosen for this study.

Differences in timing of oocyte prophase arrest and in maturation kinetics across animals are indeed highly relevant in relation to the underlying biochemical mechanisms. Unfortunately, not enough information is currently available concerning the duration of the successive phases of oocyte prophase arrest across species to make any meaningful correlations with PKA regulation of maturation initiation. We have nevertheless expanded the Introduction to cover this issue as follows:

• We start the introduction by mentioning how the length of the prophase arrest varies across species. __Changes: Page 3, lines 5-11. __
• We have added examples of species which likely have similar durations of prophase arrest but show cAMP-stimulated vs cAMP-inhibited release. Changes:____ Page 4, lines 28-35.
• We have specified the temporal differences in meiotic maturation in Xenopus (3-7 hrs) and Clytia (10-15 min). Changes: Page 5, lines 32-33. Reviewer #2: why, and not others, were these species [Xenopus, Clytia] chosen for this study. A brief justification is included in lines 1-page 5 "..a laboratory model hydrozoan species well suited to oogenesis studies", but it does not explain why this and not other hydrozoan species like Hydra, that has also been used for meiosis studies.

As requested by Reviewer #2, fuller details are now included about the advantages of Clytia compared to other hydrozoan species, citing several articles and recent reviews here and also in the Discussion. Changes: Page 5, lines 21-32 & 37-39.

Hydra is a classic cnidarian experimental species and has proved an extremely useful model for regeneration and body patterning, but is not suitable for experimental studies on oocyte maturation because spawning is hard to control and fully-grown oocytes cannot easily be obtained, manipulated or observed. In contrast many hydromedusae (including Clytia, Cytaeis, and Cladonema) have daily dark/light induced spawning and accessible gonads, so provide great material for studying oogenesis and maturation. Of these, Clytia has currently by far the most advanced molecular and experimental tools.

Reviewer #2: The proteins MAPK is not introduced properly, as it is first mentioned in the results section in line 12. Given the importance of the results provided with it, it should be presented in the introduction prior to the results section.

As requested by Reviewer #2, the involvement of MAPK activation during Xenopus oocyte meiotic maturation is now introduced, explaining how its phosphorylation serves as a marker of Cdk1 activation. Changes: Page 5, lines 1-5.

Reviewer #2: *These sentences need a more elaborate explanation: Page 4 Lines 16-17 "... no role for cAMP has been detected in meiotic resumption, which is mediated by distinct signaling pathways" Which pathways? *

We now give the example of the well-characterized pathway Gbg-PI3K pathway for oocyte maturation initiation in the starfish. Changes: Page 4, lines 1-15.

Reviewer #2: Page 4 line 34-39. Introduction indicates that the phosphorylation of ARPP19 on S67 by Gwl is a poorly understood molecular signaling cascade (line 34). However, the positive role of ARPP19 on Cdk1 activation, through the S67 phosphorylation by Gwl, appears to be widespread across all eukaryotic mitotic and meiotic divisions studied (lines 36-37). These two sentences seem a little contradictory. If the general pathway has been identified but the signaling cascade is still not well described, please indicate that in a clearer way.

We apologise that the wording we used was not clear and implied that the mechanisms of PP2A inhibition by Gwl-phosphorylated ARPP19 were poorly understood. On the contrary, they are very well studied. The part that remains mysterious concerns the upstream mechanisms. We have reworded the paragraph to make this point unambiguous. Changes: Page 5, lines 1-8.

______________IV - Results__

Reviewer #2: The text of the results is generally well described; however, all the sections start with a long introductory paragraph. I believe this facilitates the contextualization of the experiments, but please try to summarize when possible. For example, in page 5 lines 12-25, or page 7 lines 30-37, are all introduction information.

As requested by Reviewer #2, we have shortened or removed the introductory passages of the Results section paragraphs, which were redundant with the information given in the introduction. We did not restrict to the two examples cited by the reviewer, but have shortened all the Results passages that repeat information already provided in the Introduction. Changes: Page 7, lines 3-4 & 14-16 & 36-37 - Page 8, lines 12-15 - Page 8, lines 37-40 & Page 9, lines 1-6.

Reviewer #2: Page 7, Lines 14-19 present a general conclusion of the findings explained in lines 20-27. I think these results are important and they should be explained better, in my opinion they are slightly poorly described.

We have followed the reviewer's recommendation. The explanation of the experiments and the results are more detailed and the paragraph ends with a general conclusion which came too early in the previous version. Changes: Page 8, lines 22-24 & 32-34.

Reviewer #2: Page 8, lines 16-17: "It was not possible to increase injection volumes or protein concentrations without inducing high levels of non-specific toxicity". What are the non-specific toxicity effects? How was this addressed? What fundaments this conclusion?

Clytia oocytes are relatively fragile. Sensitivity of oocytes to injection varies between batches, while in general increasing injection volumes or protein concentrations increases the levels of lysis observed. We do not know exactly what causes this but lysis can happen either immediately following injection or during the natural exaggerated cortical contraction waves that accompany meiotic maturation, suggesting that it relates to mechanical trauma. We have expanded this paragraph and the legend of Fig. 3C to explain these injection experiments more fully in the text and to clarify these issues. Changes: Page 9, lines 16-29 - Page 32, lines 34-41 & Page 33, lines 1-11 - Supplementary Table 1.

Same paragraph: Lines 25-27 of page 8. Text reads, "These results suggest that PP2A inhibition is not sufficient to induce oocyte maturation in Clytia, although we cannot rule out that the quantity of OA or Gwl thiophosphorylated ARPP proteins delivered was insufficient to trigger GVBD.". Please provide evidence if higher concentrations of OA or Gwl were tested to state this conclusion.

As explained above, we could not increase the concentrations of ARPP19 protein beyond 4mg/ml. It is important to note that at the same concentration, both Clytia and Xenopus proteins induce activation of Cdk1 and GVBD in the Xenopus oocyte.

Concerning OA, it is well documented in many systems including Xenopus, starfish and mouse oocytes as well as mammalian cell cultures, that high concentrations lead to cell lysis/apoptosis as a result of a massive deregulation of protein phosphorylation (Goris et al, 1989; Rime & Ozon, 1990; Alexandre et al, 1991; Boe et al, 1991; Gehringer, 2004; Maton el al, 2005; Kleppe et al, 2015). Specific tests in Xenopus oocytes, have shown that injecting 50 nl of 1 or 2 mM OA specifically inhibits PP2A, while injecting 5 mM also targets PP1 and higher OA concentrations inhibit all phosphatases. For these reasons, we did not increase OA concentrations over 2 mM. When injected in Xenopus oocyte at 1 or 2 mM, OA induces Cdk1 activation, GVBD but then the cell dies because PP2A has multiple substrates essential for cell life. When injected at 2 mM in Clytia oocytes, OA does not induce Cdk1 activation nor GVBD but promotes cell lysis. This supports the conclusion that 2 mM OA is sufficient to inhibit PP2A (and possibly other phosphatases) but that PP2A inhibition is not sufficient to induce oocyte maturation in Clytia.

We have reworded the relevant text to make these points clearer. The previous statement that “we cannot rule out that the quantity of OA or Gwl thiophosphorylated ARPP proteins delivered was insufficient to trigger GVBD” has been removed because it was unnecessarily cautious in the context of the literature cited above, as now fully explained. Changes: Page 9, lines 31-35 - Page 32, lines 34-41 & Page 33, lines 1-11 - Supplementary Table 1.

References: Alexandre et al, 1991, doi: 10.1242/dev.112.4.971; Boe et al, 1991, doi: 10.1016/0014-4827(91)90523-w; Gehringer, 2004, doi: 10.1016/s0014-5793(03)01447-9; Goris et al, 1989, doi: 10.1016/0014-5793(89)80198-x; Kleppe et al, 2015, doi: 10.3390/md13106505; Maton el al, 2005, doi: 10.1242/jcs.02370; Rime & Ozon, 1990, doi: 10.1016/0012-1606(90)90106-s

Reviewer #2: Lines 12-13: the sentence "This in vitro assay thus places S81 as the sole residue in ClyARPP19 for phosphorylation by PKA." is overstated. As not all residues had been tested, please indicate that "it is likely that" or "among the residues tested", in contrast to "the sole residue in ClyARPP19".

We realise that we had not explained clearly enough how the thiophosphorylation assay works. In this assay, γ-S-ATP will be incorporated into any amino acid of ClyARPP19 phosphorylatable by PKA. The observed thiophosphorylation of the wild-type protein, demonstrates that one or more residues are phosphorylated by PKA. This thiophosphorylation was completely prevented by mutation of a single residue, S81. This experiment thus shows that S81 is entirely responsible for phosphorylation by PKA in this assay. We have rewritten this section more clearly. Changes: Page 10, lines 18-28.

______________V - Figures and text related to the figures__

Figure 1A

Reviewer #2: *Why is mouse not included in Figure 1A? Although it might be very similar to human, given that mouse is the species that is most commonly use as a mammalian model, I believe it could be included. However, this is optional upon decision by the authors. *

We have replaced the human sequence in Figure 1A with the mouse sequence as suggested. The sequences of each of the mouse and human ENSA/ARPP19 proteins are indeed virtually identical across mammals. Changes: Fig. 1A.

Figure 1C

Reviewer #2: *There should be a better explanation in the text of the results sections for the image included in in Fig1 C. Note that Clytia is not a commonly used species, therefore images should be properly explained for general readers. Please indicate in the text that ClyARPP19 mRNA is expressed in previtellogenic oocytes and not in vitellogenic, plus any additional information needed to understand the image. In addition, the detection of ARPP19 in the nerve rings is intriguing. This is mentioned in the discussion section, any idea of its function there? Please include some additional information or additional references, if they exist. *

We have expanded the explanations of Fig. 1C in the text and in the figure legend. We have also added cartoons to the figure to help readers understand the organisation of the Clytia jellyfish and gonad. As now explained, ClyARPP19 mRNA is detected in oocytes at all stages, but the signal is much stronger in pre-vitellogenic oocytes because all cytoplasmic components including mRNAs are significantly diluted by high quantity of yolk proteins as the oocytes grow to full size. Changes: page 7, line 40 & page 8, lines 1-9 - Fig. 1C - Legend page 31, lines 19-31.

Nothing is known about the function of ARPP19 in the Clytia nervous system. The only data linking ARPP19 and the nervous system concerns mammalian ARPP16, an alternatively spliced variant of ARPP19. ARPP16 is highly expressed in medium spiny neurons of the striatum and likely mediates effects of the neurotransmitter dopamine acting on these cells (Andrade et al, 2017; Musante et al, 2017). This point is included in the Discussion in relation to the hypothesis that PKA phosphorylation of ARPP19 proteins in animals first arose in the nervous system and only later was coopted into oocyte maturation initiation. Changes: page 16, lines 12-13 & 17-20 - page 19, lines 6-9.

Figure 2A

Reviewer #1: Fig. 2A (and similar plots in subsequent figures): is it really necessary to cut the x axis? Would it be possible to indicate the number of oocytes for each experiment (maybe in the legend in brackets)?

As requested by reviewer #1, the x-axis is no longer cut. The number of oocytes for each experiment is now provided in the legend of Fig. 2A and in similar plots of Fig. 5A and 5D. Changes: Fig. 2A - Legends page 31, lines 37-38 (Fig. 2A), page 33, line 25 (Fig. 5A) - page 33, line 34 (Fig. 5D).

Figure 2D-E (as well as Figure 6C-D and Figure 8B-C)

Reviewer #1: *Fig. 2D (and all similar plots below): I am lacking the discrete data points that were measured. Without these it is impossible to evaluate the fits. The half-times shown in 2E are somewhat redundant, and the information could be combined on a single plot. *

We added all the data points to the concerned plots: 2D, 6C and 8B. As recommended by reviewer #1, we combined on a single plot the phosphorylation levels and the half-times. 2D-E => 2D, 6C-D => 6C and 8B-C => 8B. Changes: Figs 2D, 6C and 8B - Legends page 32, lines 9-14 (Fig. 2D), page 34, lines 24-30 (Fig. 6C) - page 35, lines 13-18 (Fig. 8B).

Figure 3A and 3B

Reviewer #1: Fig. 3: why is the blot for PKA substrates cut into 3 pieces? It would be clearer to show the entire membrane.

In western blot experiments using Clytia oocytes, the amount of material was limited so the membranes were cut into three parts. The central part was incubated sequentially in distinct antibodies. We finally incubated all three parts of the membrane with the anti-phospho-PKA substrate antibody to reveal the full spectrum of proteins recognized by this antibody. The 3 pieces in Fig. 3A therefore together make up the same original membrane. We had separated them on the figure to make it clear that the membrane had been cut. In the new presentation, the 3 pieces are shown next to each other, making it clear that all the membrane is present, with dotted lines indicating the cut zone as explained in the legend. Changes: Fig. 3A and 3B - Legend page 32, lines 22-25 (Fig. 3A), lines 30-33 (Fig. 3B) - Page 24, lines 3-6 (Methods).

Figure 3C

Reviewer #2: Fig. 3C needs a better explanation in the text. The way these graphs are presented is somehow confusing. The meaning of the dots is not self-explanted in the graph, and it seems that each experiment was done independently but then the complete set of results is presented. Legend says that "each dot represents one experiment" but this is difficult to read as in every analysis the figure also indicates the average and the total number of oocytes. If authors wish so, they can keep the figure as it is, but then please explain this graph better in the text, and please include statistical analysis. These results are very robust, but a comparison between the number of oocytes that go through spontaneous GVBD of lysis in the different conditions will benefit their understanding.

This figure is intended to provide an overview of all the Clytia oocyte injection experiments that we performed, for which full details are given in Supplementary Table 1. Since these experiments were not equivalent in terms of exact timing and types of observation (or films) made and oocyte sensitivity to injection -as ascertained by buffer injections-, it is not justified to make statistical comparisons between groups. We apologise that the presentation was misleading in this respect and hope that the new version is easier to understand. We removed from the figure the average percentage of maturation for each condition between experiments to avoid any misunderstanding of the nature of the data, and rather represent the values of each experiment independently. We also now explain the data included in the figure fully in the text and figure legend. Changes: Page 9, lines 16-39 - Fig. 3C and Supplementary Table 1 - Legend page 32, lines 34-41 & page 33, lines 1-11.

Reviewer #2: Also, please provide in the text a plausible explanation for the cause of oocyte lysis for all experimental conditions (Fig 3C). Given that in the control experiments with buffer this effect is also observed in some oocytes, please explain if this is caused by a mechanical disruption of the oocyte during the injection. In contrast, okadaic acid induces the lysis in all the 14/14 oocytes analyzed, is this due also to the mechanical approach? Or is there other reason more related to the PP2A inhibition? Please explain.

These points are treated above in the response to this reviewer concerning the Results section.

Figure 5

Reviewer #2: In Figure 5 D-F, cited in page 9 lines 35-35. Can you provide an explanation of why the time course of meiosis resumption was delayed?

The binding partners/effectors of XeARPP19-S109D that are involved in maintaining the prophase arrest have not yet been identified. The most probable explanation of the delay in meiotic maturation induced by ClyARPP19-S109D is that Clytia protein recognizes less efficiently these unknown ARPP19 effectors that mediate the prophase arrest. As a result, maturation would be delayed, but not blocked. This explanation was provided in the Discussion (page 17, lines 14-17) and is now mentioned in the Results section. Changes: page 11, lines 16-19.

______________VI - Discussion__

Reviewer #2: Although it presents highly interesting suggestions, discussion may border on being overly speculative, especially from line 37 of page 15 till the end.

We agree and have reduced the speculation in this part of the discussion, in particular regrouping and reformulating ideas about evolutionary scenarios in a single paragraph. Changes: page 17, lines 37-41 - page 18, lines 1-41 - page 19, lines 1-18.

SUMMARY - ____Point by point responses to individual reviewers’ comments in their order of appearance.

Reviewer 1

• The figures and text could be slightly condensed down to about 6 figures. We have reduced the number of figure panels but we prefer to maintain the number of figures, because the experimental data presented in them is essential to the interpretation of our results and the overall conclusions of the article. If the journal editor would like us to reduce the number of figures, we could do this by displacing Figure 4 and some panels of other figures (to then fuse some of them) to supplementary material, but this would be a pity.

• The exact mechanisms responsible for the reduced effect of PKA phosphorylation remain relatively vaguely defined. Indeed, in their Discussion the authors list a number of experimental approaches to address this - but I understand that these would all involve substantial efforts to address. In particular, testing chimeric constructs around the consensus PKA site and from multiple species could be very informative. As the reviewer points out, unravelling these exact mechanisms will require a large amount of additional work and is beyond the scope of the current study.

• 2A (and similar plots in subsequent figures): is it really necessary to cut the x axis? Would it be possible to indicate the number of oocytes for each experiment (maybe in the legend in brackets)? Fig. 2A has been changed in line with the reviewer's request (as well as similar plots in Fig. 5A and 5D). Changes: Fig. 2A - Legends page 31, lines 37-38 (Fig. 2A), page 33, line 25 (Fig. 5A) - page 33, line 34 (Fig. 5D).

• 2D (and all similar plots below): I am lacking the discrete data points that were measured. Without these it is impossible to evaluate the fits. The half-times shown in 2E are somewhat redundant, and the information could be combined on a single plot. Fig. 2D has been changed in line with the reviewer's request (as well as similar plots in Figs 6C-D and 8B-C). Changes: Fig. 2D, 6C and 8B - Legends page 32, lines 9-14 (Fig. 2D), page 34, lines 24-30 (Fig. 6C) - page 35, lines 13-18 (Fig. 8B).

• 3: why is the blot for PKA substrates cut into 3 pieces? It would be clearer to show the entire membrane. In western blot experiments using Clytia oocytes, the amount of material was limited so the membranes were cut into three parts. The central part was incubated sequentially in distinct antibodies. We finally incubated all three parts of the membrane with the anti-phospho-PKA substrate antibody to reveal the full spectrum of proteins recognized by this antibody. The 3 pieces in Fig. 3A therefore together make up the same original membrane. In the new presentation, the 3 pieces are shown next to each other, making it clear that all the membrane is present, with dotted lines indicating the cut zone as explained in the legend. Changes: Fig. 3A and 3B - Legend page 32, lines 22-25 (Fig. 3A), lines 30-33 (Fig. 3B) - Page 24, lines 3-6 (Methods).

Reviewer 2

• Abstract needs to be simplified if wants to reach a broader range of readers. We have reworked the Abstract to make it more accessible to new readers. Changes: Page 2.

• It would be interesting to include some time-references when introducing the prophase arrest of Clytia and Xenopus oocytes. This should be indicated in the introduction with a short introduction of why, and not others, were these species chosen for this study. We have expanded the Introduction to cover the issue of time-references. Fuller details are now included about the advantages of Clytia compared to other hydrozoan species. Changes: Page 3, lines 5-11, page 4, lines 28-35, page 5, lines 32-33, page 5, lines 21-32 & 37-39.

• The proteins MAPK is not introduced properly, as it is first mentioned in the results section. The involvement of MAPK activation during Xenopus oocyte meiotic maturation is now introduced. Changes: Page 5, lines 1-5.

• Page 4 Lines 16-17 "... no role for cAMP has been detected in meiotic resumption, which is mediated by distinct signaling pathways" Which pathways? We now give the example of the well-characterized pathway Gbg-PI3K pathway for oocyte maturation in starfish, also mentioning that in many species the pathways are still unknown. Changes: Page 4, lines 1-15.

• Page 4 line 34-39. Introduction indicates that the phosphorylation of ARPP19 on S67 by Gwl is a poorly understood molecular signaling cascade (line 34). However, the positive role of ARPP19 on Cdk1 activation, through the S67 phosphorylation by Gwl, appears to be widespread across all eukaryotic mitotic and meiotic divisions studied (lines 36-37). These two sentences seem a little contradictory. The mechanisms of PP2A inhibition by Gwl-phosphorylated ARPP19 are very well studied. The part that remains mysterious concerns the upstream mechanisms. We have reworded the paragraph to make this point unambiguous. Changes: Page 5, lines 1-8.

• Why is mouse not included in Figure 1A? We have replaced the human sequence in Figure 1A with the mouse sequence. Changes: Fig. 1A.

• 1C: There should be a better explanation in the text of the results sections for the image included in in Fig1 C. Please indicate in the text that ClyARPP19 mRNA is expressed in previtellogenic oocytes and not in vitellogenic. We have expanded the explanations of Fig. 1C in the text. We have also added cartoons to the figure to help readers understand the organisation of the Clytia jellyfish and gonad. As now explained, ClyARPP19 mRNA is detected in oocytes at all stages, but the signal is much stronger in pre-vitellogenic oocytes because all cytoplasmic components are significantly diluted by high quantity of yolk proteins. Changes: page 7, line 40 & page 8, lines 1-9 - Fig. 1C - Legend page 31, lines 19-31.

• In addition, the detection of ARPP19 in the nerve rings is intriguing. Any idea of its function there? The only data linking ARPP19 and the nervous system concerns a mammalian variant of ARPP19 that is highly expressed in the striatum. This point is included in the Discussion. __Changes: __page 16, lines 12-13 & 17-20 - page 19, lines 6-9.

• Figure 3C. The way these graphs are presented is somehow confusing. If authors wish so, they can keep the figure as it is, but then Also, please provide in the text a plausible explanation for the cause of oocyte lysis for all experimental conditions. please explain this graph better in the text, and please include statistical analysis. This figure is intended to provide an overview of all the Clytia oocyte injection experiments, for which full details are given in Supplementary Table 1. We have modified the figure and now clarified this fully in the text and figure legend. Clytia oocytes are relatively fragile. Sensitivity of oocytes to injection varies between batches, while in general increasing injection volumes or protein concentrations increases the levels of lysis observed. We do not know exactly what causes this but it probably relates to mechanical trauma. We now explain these injection experiments more fully in the text. Changes: Page 9, lines 16-39 - Fig. 3C and Supplementary Table 1 - Legend page 32, lines 34-41 & page 33, lines 1-11.

• In Figure 5 D-F, cited in page 9 lines 35-35. Can you provide an explanation of why the time course of meiosis resumption was delayed? The most probable explanation is that Clytia protein recognizes less efficiently the unknown ARPP19 effectors that mediate the prophase arrest in Xenopus. This explanation is provided in the Results section. Changes: page 11, line 16-19.

• All the sections start with a long introductory paragraph. I believe this facilitates the contextualization of the experiments, but please try to summarize when possible. As requested, we have shortened or removed the introductory passages of the Results section paragraphs, which were redundant with the information given in the introduction. Changes: Page 7, lines 3-4 & 14-16 & 36-37 - Page 8, lines 12-15 - Page 8, lines 37-40 & Page 9, lines 1-6.

• Page 7, Lines 14-19 present a general conclusion of the findings explained in lines 20-27. I think these results are important and they should be explained better, in my opinion they are slightly poorly described. The explanation of the experiments and the results are now more detailed and the paragraph ends with a general conclusion which came too early in the previous version. Changes: Page 8, lines 22-24 & 32-34.

• Page 8, lines 16-17: "It was not possible to increase injection volumes or protein concentrations without inducing high levels of non-specific toxicity". What are the non-specific toxicity effects? How was this addressed? What fundaments this conclusion? As explained above, increasing injection volumes or protein concentrations increases the levels of lysis observed due probably to mechanical trauma. But it is important to note that at the same concentration, both Clytia and Xenopus proteins induce activation of Cdk1 and GVBD in the Xenopus oocyte. Changes: Page 9, lines 16-29 - Page 32, lines 34-41 & Page 33, lines 1-11 - Supplementary Table 1.

• Lines 25-27 of page 8. "These results suggest that PP2A inhibition is not sufficient to induce oocyte maturation in Clytia, although we cannot rule out that the quantity of OA or Gwl thiophosphorylated ARPP proteins delivered was insufficient to trigger GVBD." Please provide evidence if higher concentrations of OA or Gwl were tested to state this conclusion. High OA concentrations lead to cell lysis/apoptosis as a result of a massive deregulation of protein phosphorylation. For these reasons, we cannot increase OA concentrations over 2 µM. When injected in Xenopus oocyte at 1 or 2 µM, OA induces Cdk1 activation, but then the cell dies because PP2A has multiple substrates essential for cell life. When injected at 2 µM in Clytia oocytes, OA does not induce Cdk1 activation but promotes cell lysis. This supports the conclusion that 2 µM OA is sufficient to inhibit PP2A but that PP2A inhibition is not sufficient to induce oocyte maturation in Clytia. We have reworded the relevant text. Changes: Page 9, lines 31-35 - Page 32, lines 34-41 & Page 33, lines 1-11 - Supplementary Table 1.

• Lines 12-13: the sentence "This in vitro assay thus places S81 as the sole residue in ClyARPP19 for phosphorylation by PKA." is overstated. As not all residues had been tested, please indicate that "it is likely that" or "among the residues tested", in contrast to "the sole residue in ClyARPP19". The observed thiophosphorylation of the wild-type protein demonstrates that one or more residues are phosphorylated by PKA. This thiophosphorylation was completely prevented by mutation of a single residue, S81. This experiment thus shows that S81 is entirely responsible for phosphorylation by PKA in this assay. We have rewritten this section more clearly. Changes: Page 10, lines 18-28.

• Some parts of the discussion are a bit speculative. We have reduced the speculation in this part of the discussion, in particular regrouping and reformulating ideas about evolutionary scenarios into a single paragraph. Changes: page 17, lines 37-41 - page 18, lines 1-41 - page 19, lines 1-18.

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

Learn more at Review Commons

---

Reply to the reviewers

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

*REVIEW COMMENT *

*The article titled "The tRNA thiolation-mediated translational control is essential for plant immunity" by Zheng et al. highlights the critical role of tRNA thiolation in Arabidopsis plant immunity through comprehensive analysis, including genetics, transcriptional, translational, and proteomic approaches. Through their investigation, the authors identified a cbp mutant, resulting in the knockout of ROL5, and discovered that ROL5 and CTU2 form a complex responsible for catalyzing the mcm5s2U modification, which plays a pivotal role in immune regulation. The findings from this study unveil a novel regulatory mechanism for plant defense. Undoubtedly, this discovery is innovative and holds significant potential impact. However, before considering publication, it is necessary for the authors to address the various questions raised in the manuscript concerning the experiments and analysis to ensure the reliability of the study's conclusions. *

__Response: Thank you very much for your support and suggestions! __

*Here is Comments: *

*Line 64-65: *

*The author mentioned that 'While NPR1 is a positive regulator of SA signaling, NPR3 and NPR4 are negative regulators.' However, several recent discoveries are suggesting that it may not be a definitive fact that NPR3 and NPR4 are negative regulators. Therefore, I recommend the authors to review this section in light of the findings from recent papers and make necessary edits to reflect the most current understanding. *

__Response: Thank you for your feedback. Since we mainly focused on NPR1 in this study, we removed this sentence to avoid confusion. We provided additional information about NPR1 in the Introduction section to emphasize the importance of NPR1 (Line 64-68). __

*Line 182- & Figure 4: *

*The author conducted RNA-seq, Ribo-seq, and proteome analysis. Describing the analysis as "transcriptional and translational" using RNA-seq and proteome data seems not entirely accurate. Proteome data compared with RNA-seq not only reflects translational changes but may also encompass post-translational regulations that contribute to the observed differences. To maintain precision, the title of this section should either be modified to "transcriptional and protein analysis" or, alternatively, compare RNA-seq and Ribo-seq data to demonstrate the transcriptional and translational changes more explicitly. *

__Responses: Thank you for your suggestions. We agree with you and thus change the description accordingly throughout the manuscript. __

*Line 229-235 and Figure 5C: *

*The interpretation of Figure 5C's polysome profiling results is inconclusive. There does not seem to be a noticeable difference in polysomal fractions between the cab mutant and CAM. The observed differences in the overlay of multiple polysome fractions between cab and COM could be primarily influenced by baseline variations rather than a significant decrease in the polynomial fractions in cpg. Therefore, it is necessary to carefully review other relevant papers that discuss polysome fraction data and their analysis. By doing so, the authors can make the appropriate corrections to ensure accurate interpretations. *

__Responses: Thank you for your comments. We agree that the difference between cgb and COM was not dramatic visually. This is a common feature of ____polysome profiling assay (e.g. Extended Data Fig. 1f in Nature 545: 487–490; Fig. 1c in Nature Plants, 9: 289–301). In our case, the difference between polysome fractions was unlikely due to the baseline variation for two reasons. First, baseline variation affects monosome and polysome fractions in the same way. However, our results showed the monosome fraction of cgb is higher than that of COM, whereas the polysome fraction of cgb is lower than that of COM. Second, this result was repeatedly detected. For better visualization, we adjusted the scale of Y axis in the revised manuscript (Figure 5D). __

*Line 482 Ion Leakage assay: *

I could not find the ion leakage assay in this manuscript, so I wonder why it is mentioned.

__Response: We are sorry for the mistake. The Ion leakage data were included in previous visions of the manuscript. We removed the data but forgot to remove the corresponding method in the present version. __

*Materials and Methods: *

*To enhance the reproducibility of the study, the authors should provide a more detailed description of the materials and methods, especially for critical experiments like the Yeast-two-hybrid assays. Clear documentation of specific reagents, strains, and protocols used, along with information on controls, will bolster the validity of the results and facilitate future research in this area. *

__Response: Thank you for your suggestions. We provided more details in the methods. For y____east two-hybrid assays, the vector information was included in “Vector constructions” section. __

*Minor Point: *

Line 61: There is a space between ')' and '.', which needs to be edited.

Response: The space was deleted.

*Reviewer #1 (Significance (Required)): *

*This study holds significant importance within the field of plant immunity research. The authors have made valuable contributions through their comprehensive analysis, encompassing genetics, transcriptional, translational, and proteomic approaches, to elucidate the critical role of tRNA thiolation in plant immunity. One of the major strengths of this study lies in its ability to shed light on a previously unknown regulatory mechanism for plant defense. By identifying the cbp mutant and investigating the role of ROL5 and CTU2 in catalyzing the mcm5s2U modification, the authors have unveiled a novel aspect of plant immune regulation. This innovative discovery provides a deeper understanding of the intricate molecular processes governing immunity in plants. *

*Moreover, the study's findings are not limited to the immediate field of plant immunity but also have broader implications for the scientific community. By employing diverse methodologies, the authors have demonstrated how tRNA thiolation exerts control over both transcriptional and translational reprogramming, revealing intricate links between these processes. This integrative approach sets a precedent for future research in the field of plant molecular biology and opens up new avenues for investigating other aspects of immune regulation. *

In terms of its relevance, the study's findings have the potential to captivate researchers across various disciplines, such as plant biology, molecular genetics, and translational research. The insights gained from this study may inspire researchers to explore further the role of tRNA in other regulation.

Response: Thank you very much for your positive comments and support!

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

The authors presented an intriguing and previously unknown mechanism that the tRNA mcm5s2U modification regulates plant immunity through the SA signaling pathway, specifically by controlling NPR1 translation. The manuscript was well-written and logically structured, allowing for a clear understanding of the research. The authors provided strong and persuasive data to support their key claims. However, further improvement is required to strengthen the conclusion that mcm5s2U regulates plant immunity by controlling NPR1 translation.

__Response: Thank you very much for your positive comments and support! __

Major comments:

• NPR1 translation should be examined to verify the Mass Spec (Figure 5B) and polysome profiling data (Figure 5D) by checking the NPR1 protein and mRNA level using antibodies and qPCR, respectively, in the cgb mutant background to establish a concrete confirmation of CGB regulation in NPR1 translation. * Response: This is a very constructive suggestion. We performed these experiments and found that the transcription levels of NPR1 were similar between COM and cgb both before and after ____Psm_ES4326 infection (Figure S2), _which is consistent with RNA-Seq data____. Consistent with the Mass Spec and polysome profiling data, _the NPR1 protein level was much higher in COM than that in cgb(Figure 5C) after _Psm____ ES4326 infection. Together, these data further supported our conclusion that translation of NPR1 is impaired in cgb. __

• Analyzing the genetic epistasis of CGB and NPR1 to check if CGB regulates plant immunity through the NPR1-dependent SA signal pathway. If the authors' claim is valid, I would expect no addictive effect on bacterial growth in the cgb/npr1 double mutant compared to the single mutants. Due to the broad impact of CGB on plant signaling (Figures 4E and 4F), the SA protection assay, which concentrates on the SA signal pathway, needs to be tested in WT, cgb and npr1 plants as an alternative assay to the genetic epistasis analysis. I expect that the SA-mediated protection is also compromised in cgb mutant background.*

__Response: Thank you for your suggestions. We did examine the growth of Psm ES4326 in the cgb npr1double mutant and found that cgb npr1 was significantly more susceptible than npr1 and cgb (Figure below). Although the additive effects were observed, this result was not against our conclusion for the following reasons. First, the translation of NPR1 was reduced rather than completely blocked in cgb. In other words, NPR1 still has some function in cgb. But in the cgb npr1 double mutant, the function of NPR1 is completely abolished, which explains why cgb npr1 was more susceptible than cgb. Second, in addition to NPR1, some other immune regulators (such as PAD4, EDS5, and SAG101) were also compromised in cgb(Figure 5B), which explained why cgb npr1 was more susceptible than npr1. Since the result of the genetic analysis was not intuitive, we decided not to include it in the manuscript. __

__SA signaling is known to regulate both basal resistance and systemic acquired resistance (SA-mediated protection). We have shown that cgb is defective in the defect of basal resistance, which cgb is sufficient to support our conclusion that the tRNA thiolation is essential for plant immunity. We agree that it is expected that the SA-mediated protection is also compromised in cgb. We will test this in the future study. __

• Could the authors comment on why using COM instead of WT as a control to perform the majority of the experiments? __Response: Thank you for your comments. In addition to ROL5, the cgb mutant may have other mutations compared with WT.COM is a complementation line in the cgb background. Therefore, the genetic background between COM and cgb may be more similar than that of WT and cgb*. __

• In Figure 5E, why does ACTIN2 have an enhanced translation while NPR1 shows a compromised one in cgb mutant? How does the mcm5s2U distinguish NPR1 and ACTIN2 codons? Does mcm5s2U modification have both positive and negative roles in regulating protein translation? __Response: Thank you for raising this question. As previously reported, _loss of the mcm5s2U modification causes ribosome pausing at AAA and CAA codons. Therefore, the translation of the mRNAs with more _GAA/CAA/AAA codons (called s2 codon) is likely to be affected more dramatically in cgb*. We have analyzed the percentage of s2 codon at whole-genome level (Figure below). The average percentage is 8.5%, while NPR1 contains 10.1% s2 codon and actin contains only 4.5% s2 codon. When fewer ribosomes are used for translation of the mRNAs with high s2 codon percentage, more ribosomes are available for translation of the mRNAs with low s2 codon percentage, which may account for the enhanced translation efficiency. To focus on NPR1 and to avoid confusion, we removed the ACTIN data in the revised manuscript. __

• Specify the protein amount used for the in vitro pull-down assay and agrobacteria concentration used for the tobacco Co-IP assay in the protocol section.*

Response: We added this information in Method section in the revised manuscript.

• 1. Delete the SA quantification and Ion leakage assay in the protocol, which are not used in the study.*

__Response: We are sorry for the mistake. The ____SA quantification and ion leakage data were included in previous visions of the manuscript. We removed the data but forgot to remove the corresponding method in the present version. We deleted them in the revised manuscript. __

• The strain Pst DC3000 avrRPT2 was not used in this study. Please remove it.*

Response: We are sorry for the mistake. ____The strain Pst DC3000 avrRPT2 was used for ion leakage assay in previous visions of the manuscript. We deleted it in the revised manuscript.

• In Figure 5F, did the 59 genes tested overlap with the 366 attenuated proteins in the cgb mutant? Were the 59 genes translationally regulated?*

__Response: Thank you for your suggestion. Venn diagram analysis revealed that 12 genes (about 20%) are also attenuated proteins, suggesting that ____the mcm5s2U modification regulates the translation of some SA-responsive genes. __

Reviewer #2 (Significance (Required)):

The authors' study is significant as it establishes the first connection between tRNA mcm5s2U modification and plant immunity, specifically by regulating NPR1 protein translation. This research expands our understanding of the biological role of tRNA mcm5s2U modification and highlights the importance of translational control in plant immunity. It is likely to captivate scientists working in this field.

Response: Thank you very much for your positive comments and support!

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

In this manuscript, the authors identified a cgb mutant that carries a mutation in the ROL5 gene Both the cgb mutant and the newly created rol5-c mutant are susceptible to the bacterial pathogen Psm. The authors showed that ROL5 interacts with CTU2, the Arabidopsis homologous protein of the yeast tRNA thiolation enzyme NCS2. A ctu2-1 mutant is also susceptible to Psm, suggesting the tRNA thiolation may play a role in plant immunity. Indeed, tRNA mcm5S2U levels are undetectable in rol5-c and ctu2-1 mutants. The authors found that the cgb mutation significantly attenuated basal and Psm-induced transcriptome and proteome changes. Furthermore, it was found that the translation efficiency of a group of SA signaling-related proteins including NPR1 is compromised.

The manuscript provides solid evidence for the involvement of ROL5 and CTU2 in plant immunity using the rol5 and ctu2 mutants. The authors may consider the following suggestions and comments to improve the manuscript.

Response: Thank you very much for your support and suggestions!

• The function of the Elongator complex in tRNA modification/thiolation has been extensively studied. In Arabidopsis Elongator mutants, mcm5S2U levels are very low, similar to the levels in the rol5 and ctu2 mutants (Mehlgarten et al., 2010, Mol Microbiology, 76, 1082-1094; Leitner et al., 2015 Cell Rep). In elp mutants, the PIN protein levels are reduced without reduced mRNA levels (Leitner et al., 2015), indicating that Elongator-mediated tRNA modification is involved in translation regulation. The Elongator complex plays an important role in plant immunity, though the reduced mcm5S2U levels in elp mutants were not proposed as the exclusive cause of the immune phenotypes. In fact, it would be difficult to establish a cause-effect relationship between tRNA modification and immunity. These results should be discussed in the manuscript.* Response: Thank you very much for your insightful comment on the role of the ELP complex in tRNA modification and plant immunity. We added a paragraph ____discussing the ELP complex in the revised manuscript (Line 280-295).

__In addition to tRNA modification, the ELP complex has several other distinct activities including histone acetylation, α-tubulin acetylation, and DNA demethylation. Therefore, it is difficult to dissect which activity of the ELP complex contributes to plant immunity. However, the only known activity of ROL5 and CTU2 is to catalyze _tRNA thiolation. Considering that the elp, rol5, and ctu2 mutants are all defective in tRNA thiolation, it is likely the _tRNA modification activity of the ELP complex underlies its function in plant immunity. __

• The interaction between CTU2 and ROL5 in Y2H has previously been reported (Philipp et al., 2014). The same report also showed reduced tRNA thiolation in the ctu2-2 mutant using polyacrylamide gel. These results should be mentioned/discussed in the manuscript.*

__Response: Thank you for pointing them out. We added this information in the revised version (Line 146-147). __

• tRNA modification unlikely plays a unique role in plant immunity. It can be inferred that mutations affecting tRNA modification (rol5, ctu2, elp, etc.) would delay both internal and external stimulus-induced signaling including immune signaling.*

Response: We agree with you that tRNA modification has other roles in addition to plant immunity. In the Discussion section, we have mentioned that “it was found that tRNA thiolation is required for heat stress tolerance ____(Xu et al., 2020)____. ……It will also be interesting to test whether tRNA thiolation is required for responses to other stresses such as drought, salinity, and cold.” (Line276-279).

• It would be interesting to conduct statistical analyses on the genetic codons used in the CDSs whose translation was attenuated as described in the manuscript. Do these genes including NPR1 use more than average levels of AAA, CAA, and GAA codons? If not, why their translation is impaired?*

__Response: Thank you for your suggestion. We called _GAA/CAA/AAA codons s2 codon. We have analyzed the percentage of s2 codon at whole-genome level (Figure below). NPR1 does contain more s2 codon (10.1%) than the average level (8.5%). We are preparing another manuscript, which will report the relationship between _s2 codon and translation. __

**Referees cross-commenting**

It is important to put current research in the context of available knowledge in the field. The digram in Figure 3C shows that the Elongator complex functions upstream of ROL5 & CTU2 in modifying tRNA. The function of Elongator in plant immunity has been well established. The similarities and differences should be discussed. Additionally, it may no be a good idea to claim that the results are novel.

__Response: Thank you for your comments. We added a paragraph ____discussing the ELP complex in the revised manuscript (Line 280-295). The ELP complex catalyzes the cm5U modification, which is the precursor of mcm5s2U catalyzed by ROL5 and CTU2. In addition to tRNA modification, the ELP complex has several other distinct activities including histone acetylation, α-tubulin acetylation, and DNA demethylation. Therefore, it is difficult to dissect which activity of the ELP complex contributes to plant immunity. However, the only known activity of ROL5 and CTU2 is to catalyze tRNA thiolation. Considering that the elp, rol5, and ctu2 mutants are all defective in tRNA thiolation, it is likely the tRNA modification activity of the ELP complex underlies its function in plant immunity. Therefore, our study improved our understanding of the ELP complex in plant immunity. We have deleted the words “new” and “novel” throughout the manuscript. __

Reviewer #3 (Significance (Required)):

*The manuscript provides solid evidence for the involvement of ROL5 and CTU2 in plant immunity. However, the authors did not acknowledge the existing results about the Elongator complex that functions in the same pathway in modifying tRNA. The involvement of Elongator in plant immunity has been well established. The cause-effect relationship between tRNA modification and plant immunity is difficult to demonstrate. *

Response: We think that t____he cause-effect relationship between the activities of the ELP complex and plant immunity is difficult to demonstrate because the ELP complex has several distinct activities other than tRNA modification. However, since the only known activity of ROL5 and CTU2 is to catalyze tRNA thiolation, the cause-effect relationship between tRNA thiolation and plant immunity is clear, which indicated that ____the ____tRNA modification activity of the ELP complex contributes to plant immunity.

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

Learn more at Review Commons

---

Reply to the reviewers

*We appreciate the valuable suggestions and the overall highly positive review of our manuscript. We have now included many suggestions provided by the reviewers, which have made our manuscript much stronger and more rigorous. One reviewer acknowledged, “This study uncovers sex-dependent mechanisms underlying cold sensitivity between male and female mice. The detailed IHC analysis of MHCII expression in DRG neurons is a clear strength of this study and supports flow cytometry results as well as existing literature. The specificity of MHCII expression on small diameter is well characterized and supported by conditional knockout mouse models of MHCII in TRVPV1-lineage neurons.” *

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, Whitaker EE and co-authors implicate MHCII expression in DRG neurons in the resolution of pain following paclitaxel treatment. The authors demonstrate that CD4 T cells closely interact with DRG neurons, which also express MHCII proteins. They further characterize neuronal MHCII expression in naïve and paclitaxel treated mice in small diameter TRPV1+ neurons. Utilizing genetic animal models with MHCII knockout in TRPV1-lineage neurons, the authors highlight that loss of MHCII in TRPV1 neurons exaggerates cold sensitivity in naïve male mice, and in both sexes following paclitaxel treatment.

Major concerns:

The most pressing concern regarding this study is a lack of a vehicle control group. It is not appropriate to be comparing paclitaxel treated mice to naïve mice. Please include a vehicle treatment (cremophor:ethanol 1:1 diluted 1:3 in PBS) group for all experiments involving paclitaxel. This would also improve statistics as unpaired T tests comparing naïve vs paclitaxel is not convincing.

Figure 1A only includes representative images of a small number of T cells in presumable contact with DRG neurons in female Day 14 paclitaxel mice, but does not include images from other groups. Similarly, B-D show a single CD4+ T cell in contact with DRG neurons in Day 14 paclitaxel and naïve female mice. Please include quantification of the frequency of CD4+ T cells interacting with DRG neurons in the different experimental groups utilized in this study.

Please include entire blot for Figure 2A (or at least more of the blot). There is plenty of space in the figure and as it currently appears is not free from apparent manipulation.

The authors conclude that MHCII helps to suppress chemotherapy-induced peripheral neuropathy, resolving cold allodynia following paclitaxel treatment. To support this conclusion, I think it is necessary to include a time-course experiment highlighting whether cKO of MHCII in TRPV1 neurons indeed increases the duration for cold hypersensitivity to resolve following paclitaxel treatment.

Minor concerns:

The graphical abstract is misleading. The authors suggest paclitaxel is acting exclusively via TLR4 and that signaling is resolved at Day 14 which their data does not support. Please adjust to reflect findings from the experiments included in this study.

Figure 4 and 6 MHCII labelling is oversaturated in most of the images, creating a blurry hue in the representative images. This should be fixed

The effects of the PTX cHET group are very mild in both the male and female cohorts, and specific to 1 trial. I believe these assessments were conducted at Day 6 post injection. Why was this timepoint chosen considering differences in MHCII expression in small neurons was only present at Day 14 relative to naïve? The statistical analysis should also have been a mixed-effects repeated measures between groups ANOVA.

Significance

This study uncovers sex-dependent mechanisms underlying cold sensitivity between male and female mice. The detailed IHC analysis of MHCII expression in DRG neurons is a clear strength of this study, and supports flow cytometry results as well as existing literature. The specificity of MHCII expression on small diameter is well characterized and supported by conditional knockout mouse models of MHCII in TRVPV1-lineage neurons. The clear limitations of this study is the lack of a vehicle control group and limited behavioral analysis. They undermine the conclusions made by the author, and in extension, the significance of this study.

This study adds to the understanding of neuro-immune signaling in peripheral neuropathic pain. As far as I am aware, this is the first study to investigate MHCII expression in DRGs in relation to development of chemotherapy-induced peripheral neuropathy. Thus this study provides an incremental advance in neuroimmune mechanisms contributing to the development of chemotherapy-induced peripheral neuropathy in mice.

This study would be of interest to basic researchers interested in neuropathic pain, with particularly researchers with a focus on neuroimmunology and chemotherapy-induced peripheral neuropathy models. The sex differences observed in naïve mice would also be of interest to basic researchers within the wider pain field. Given the preliminary nature of the findings, I do not think this would be of interest to broader neuroimmunology or clinical audiences.

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

Learn more at Review Commons

---

Reply to the reviewers

Dear Editor and Reviewers,

*We thank you for the thorough and detailed examination of our preprint and providing the very valuable comments that helped to even better present and interpret our data. *

*Thank you in particular for appreciating the extensive set of microscopic techniques that we have combined to study in a unique manner the characteristics and functionalities of FIT nuclear bodies in living plant cells. *

We prepared a revised preprint in which we address all reviewer comments. Our revision includes a NEW experiment (in four repetitions) that addresses one comment made by the reviewers with regard to the effects of the environmental FIT NB-inducing situation:

• NEW Supplemental Figures S6 and S7: Analysis of previously reported intron retention splicing variants of Fe deficiency genes FIT, BHLH039, IRT1, FRO2 in new gene expression experiments (Four independent repetitions of the experiments with three biological replicates of each sample – white/blue light treatment, sufficient and deficient iron supply). In the following, please find our detailed response to all reviewer comments.

With these changes, we hope that our peer-reviewed preprint can receive a positive vote,

We are looking forward to your response,

Sincerely

Petra Bauer and Ksenia Trofimov on behalf of all authors

Comments to the reviews:

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

In this paper entitled " FER-LIKE IRON DEFICIENCY-INDUCED TRANSCRIPTION FACTOR (FIT) accumulates in homo- and heterodimeric complexes in dynamic and inducible nuclear condensates associated with speckle components", Trofimov and colleagues describe for the first time the function of FIT in nuclear bodies. By an impressive set of microscopies technics they assess FIT localization in nuclear bodies and its dynamics. Finally, they reveal their importance in controlling iron deficiency pathway. The manuscript is well written and fully understandable. Nonetheless, at it stands the manuscript present some weakness by the lack of quantification for co-localization and absence controls making hard to follow authors claim. Moreover, to substantially improve the manuscript the authors need to provide more proof of concepts in A. thaliana as all the nice molecular and cellular mechanism is only provided in N. bentamiana. Finally, some key conclusions in the paper are not fully supported by the data. Please see below:

Main comments:

1) For colocalization analysis, the author should provide semi-quantitative data counting the number of times by eyes they observed no, partial or full co-localization and indicate on how many nucleus they used.

Authors:

We have added the information in the Materials and Method section, lines 731-734:

In total, 3-4 differently aged leaves of 2 plants were infiltrated and used for imaging. One infiltrated leaf with homogenous presence of one or two fluorescence proteins was selected, depending on the aim of the experiment, and ca. 30 cells were observed. Images are taken from 3-4 cells, one representative image is shown.

In all analyzed cases, except in the case of colocalization of FIT and PIF4 fusion proteins, the ca. 30 cells had the same localization and/or colocalization patterns. This information has also been added in the figure legends. Each experiment was repeated at least 2-3 times, or as indicated in the figure legend.

2) Do semi-quantitative co-localization analysis by eyes, on FIT NB with known NB makers in the A. thaliana root. For now, all the nicely described molecular mechanism is shown in N. benthamiana which makes this story a bit weak since all the iron transcriptional machinery is localized in the root to activate IRT1.

Authors:

The described approach has been very optimal, and we were able to screen co-localizing marker proteins in FIT NBs in N. benthamiana to better identify the nature of FIT NBs. This has been successful as we were able to associate FIT NBs with speckles. The N. benthamiana system allowed optimal microscopic observation of fluorescence proteins and quantification of FIT NB characteristics in contrast to the root hair zone of Arabidopsis where Fe uptake takes place. FIT is expressed at a low level in roots and also in leaves, whereby fluorescence protein expression levels are insufficient for the here-presented microscopic studies. The tobacco infiltration system is also well established to study FIT-bHLH039 protein interaction and nuclear body markers. We discuss this point in the discussion, see line 489-500.

3) The authors need to provide data clearly showing that the blue light induce NB in A. thaliana and N. benthamiana.

Authors:

*For tobacco, see Figure 1B (t = 0, 5 min) and Supplemental Movies S1. For Arabidopsis, please see Figure 1A (t = 0, 90 and 120 min) and Supplemental Figure S1A. We provide an additional image of pFIT:cFIT-GFP Arabidopsis control plants, showing that NB formation is not detected in plants that were grown in white light and not exposed to blue light before inspection (Supplemental Figure S1B). We state, that upon blue light exposure, plants had FIT NBs in at least 3-10 nuclei of 20 examined nuclei in the root epidermis in the root hair zone (in three independent experiments with three independent plants). White-light-treated plants showed no NB formation unless an additional exposure to blue light was provided (in three independent experiments, three independent plants per experiment and with 15 examined nuclei per plant). *

4) Direct conclusion in the manuscript:

• Line 170: At this point of the paper the author cannot claim that the formation of FIT condensates in the nucleus is due to the light as it might be indirectly linked to cell death induced by photodamaging the cell using a 488 lasers for several minutes. This is true especially with the ELYRA PS which has strong lasers made for super resolution and that Cell death is now liked to iron homeostasis. The same experiment might be done using a spinning disc or if the authors present the data of the blue light experiment mentioned above this assumption might be discarded. Alternatively, the author can use PI staining to assess cell viability after several minutes under 488nm laser.

Authors:

As stated in our response to comment 3, we have included now a white light control to show that FIT NB formation is not occurring under the normal white light conditions. Since the formation of FIT NBs is a dynamic and reversible process (Figure 1A), it indicates that the cells are still viable, and that cell death is not the reason for FIT NB formation.

• Line 273: I don't agree with the first part of the authors conclusion, saying that "wild-type FIT had better capacities to localize to NBs than mutant FITmSS271AA, presumably due its IDRSer271/272 at the C-terminus. This is not supported by the data. In order to make such a claim the author need to compare the FA of FIT WT with FITmSS271AA by statistical analysis. Nonetheless, the value seems to be identical on the graphs. The main differences that I observed here are, 1) NP value for FITmSS271AA seems to be lower compared to FIT-WT, suggesting that the Serine might be important to regulate protein homedimerization partitioning between the NP and the NB. 2) To me, something very interesting that the author did not mention is the way the FA of FITmSS271AA in the NB and NP is behaving with high variability. The FA of those is widely spread ranging from 0.30 to 0.13 compared to the FIT-WT. To me it seems that according to the results that the Serine 271/272 are required to stabilize FIT homodimerization. This would not only explain the delay to form the condensate but also the decreased number and size observed for FITmSS271AA compared to FIT-WT. As the homodimerization occurs with high variability in FITmSS271AA, there is less chance that the protein will meet therefore decreasing the time to homodimerize and form/aggregate NB.

Authors:

We fully agree. We meant to describe this result it in a similar way and thank you for help in formulating this point even better. Rephrasing might make it better clear that the IDRSer271/272 is important for a proper NB localization, lines 272-278:

“Also, the FA values did not differ between NBs and NP for the mutant protein and did not show a clear separation in homodimerizing/non-dimerizing regions (Figure 3D) as seen for FIT-GFP (Figure 3C). Both NB and NP regions showed that homodimers occurred very variably in FITmSS271AA-GFP.

In summary, wild-type FIT could be partitioned properly between NBs and NP compared to FITmSS271AA mutant and rather form homodimers, presumably due its IDRSer271/272 at the C-terminus.”

• Line 301: According to my previous comment (line 273), here it seems that the Serine 271/272 are required only for proper partitioning of the heterodimer FIT/BHLH039 between the NP and NB but not for the stability of the heterodimer formation. However, it might be great if the author would count the number of BHLH039 condensates in both version FITmSS271AA and FIT-WT. To my opinion, they would observe less BHLH039 condensate because the homodimer of FITmSS271AA is less likely to occur because of instability.

Authors:

bHLH039 alone localizes primarily to the cytoplasm and not the nucleus, and the presence of FIT is crucial for bHLH039 nuclear localization (Trofimov et al., 2019). Moreover, bHLH039 interaction with FIT depends on SS271AA (Gratz et al., 2019). We therefore did not consider this experiment for the manuscript and did not acquire such data, as we did not expect to achieve major new information.

5) To wrap up the story about the requirements of NB in mediating iron acquisition under different light regimes, provide data for IRT1/FRO2 expression levels in fit background complemented with FITmSS271AA plants. I know that this experiment is particularly lengthy, but it would provide much more to this nice story.

Authors:

Data for expression of IRT1 and FRO2 in FITmSS271AA/fit-3 transgenic Arabidopsis plants are provided in Gratz et al. (2019). To address the comment, we did here a NEW experiment. We provide gene expression data on FIT, BHLH039, IRT1 and FRO2 splicing variants (previously reported intron retention) to explore the possibility of differential splicing alterations under blue light (NEW Supplemental Figure S6 and S7, lines 454-466). Very interestingly, this experiment confirms that blue light affects gene expression differently from white light in the short-term NB-inducing condition and that blue light can enhance the expression of Fe deficiency genes despite of the short 1.5 to 2 h treatment. Another interesting aspect was that the published intron retention was also detected. A significant difference in intron retention depending on iron supply versus deficiency and blue/white light was not observed, as the pattern of expression of transcripts with respective intron retentions sites was the same as the one of total transcripts mostly spliced.

Minor comments

In general, I would suggest the author to avoid abbreviation, it gets really confusing especially with small abbreviation as NB, NP, PB, FA.

Authors:

*We would like to keep the used abbreviations as they are utilized very often in our work and, in our eyes, facilitate the understanding. *

Line 106: What does IDR mean?

Authors:

Explanation of the abbreviation was added to the text, lines 105-108:

“Intrinsically disordered regions (IDRs) are flexible protein regions that allow conformational changes, and thus various interactions, leading to the required multivalency of a protein for condensate formation (Tarczewska and Greb-Markiewicz, 2019; Emenecker et al., 2020).”

Line 163-164: provide data or cite a figure properly for blue light induction.

Authors:

We have removed this statement from the description, as we provide a white light control now, lines 157-158:

“When whole seedlings were exposed to 488 nm laser light for several minutes, FIT became re-localized at the subnuclear level.”

Line 188: Provide Figure ref.

Authors:

Figure reference was added to the text, lines 184-185:

“As in Arabidopsis, FIT-GFP localized initially in uniform manner to the entire nucleus (t=0) of N. benthamiana leaf epidermis cells (Figure 1B).”

Line 194: the conclusion is too strong. The authors conclude that the condensate they observed are NB based on the fact the same procedure to induce NB has been used in other study which is not convincing. Co-localization analysis with NB markers need to be done to support such a claim. At this step of the study, the author may want to talk about condensate in the nucleus which might correspond to NB. Please do so for the following paragraph in the manuscript until colocalization analysis has not been provided. Alternatively provide the co-localization analysis at this step in the paper.

Authors:

We agree. We changed the text in two positions.

Lines 176-178: “Since we had previously established a reliable plant cell assay for studying FIT functionality, we adapted it to study the characteristics of the prospective FIT NBs (Gratz et al., 2019, 2020; Trofimov et al., 2019).”

Lines 192-193: “We deduced that the spots of FIT-GFP signal were indeed very likely NBs (for this reason hereafter termed FIT NBs).”

Line 214: In order to assess the photo bleaching due to the FRAP experiment the quantification of the "recovery" needs to be provided in an unbleached area. This might explain why FIT recover up to 80% in the condensate. Moreover, the author conclude that the recovery is high however it's tricky to assess since no comparison is made with a negative/positive control.

Authors:

In the FRAP analysis, an unbleached area is taken into account and used for normalization.

We reformulated the description of Figure 1F, lines 212-214:

“According to relative fluorescence intensity the fluorescence signal recovered rapidly within FIT NBs (Figure 1F), and the calculated mobile fraction of the NB protein was on average 80% (Figure 1G).”

Line 220-227: The conclusion it's too strong as I mentioned previously the author cannot claim that the condensate are NBs at this step of the study. They observed nuclear condensates that behave like NB when looking at the way to induce them, their shape, and the recovery. And please include a control.

Authors:

Please see the reformulated sentences and our response above.

Lines 176-178: “Since we had previously established a reliable plant cell assay for studying FIT functionality, we adapted it to study the characteristics of the prospective FIT NBs (Gratz et al., 2019, 2020; Trofimov et al., 2019).”

Lines 192-193: “We deduced that the spots of FIT-GFP signal were indeed very likely NBs (for this reason hereafter termed FIT NBs).”

Line 239: It's unappropriated to give the conclusion before the evidence.

Authors:

Thank you. We removed the conclusion.

Line 240: Figure 2A, provide images of FIT-G at 15min in order to compare. And the quantification needs to be provided at 5 minutes and 15 minutes for both FIT-G WT and FIT-mSS271AA-G counting the number of condensates in the nucleus. Especially because the rest of the study is depending on these time points.

Authors:

*This information is provided in the Supplemental Movie S1C. *

Line 241: the author say that the formation of condensate starts after 5 minutes (line 190) here (line 241) the author claim that it starts after 1 minutes. Please clarify.

Authors:

In line 190 we described that FIT NB formation occurs after the excitation and is fully visible after 5 min. In line 241 we stated that the formation starts in the first minutes after excitation, which describes the same time frame. We rephrased the respective sentences.

Lines 185-188: “A short duration of 1 min 488 nm laser light excitation induced the formation of FIT-GFP signals in discrete spots inside the nucleus, which became fully visible after only five minutes (t=5; Figure 1B and Supplemental Movie S1A).”

Lines 239-242: “While FIT-GFP NB formation started in the first minutes after excitation and was fully present after 5 min (Supplemental Movie S1A), FITmSS271AA-GFP NB formation occurred earliest 10 min after excitation and was fully visible after 15 min (Supplemental Movie S1C).”

Line 254: Not sure what the authors claim "not only for interaction but also for FIT NB formation ". To me, the IDR is predicted to be perturbed by modeling when the serines are mutated therefore the IDR might be important to form condensates in the nucleus. Please clarify.

Authors:

The formation of nuclear bodies is slow for FITmSS271AA as seen in Figure 2. Previously, we showed that FITmSS271AA homodimerizes less (Gratz et al., 2019.) Therefore, the said IDR is important for both processes, NB formation and homodimerization. We have added this information to make the point clear, lines 253-255:

“This underlined the significance of the Ser271/272 site, not only for interaction (Gratz et al., 2019) but also for FIT NB formation (Figure 2).”

Line 255: It's not clear why the author test if the FIT homodimerization is preferentially associated with condensate in the nucleus.

Authors:

We test this because both homo- and heterodimerization of bHLH TFs are generally important for the activity of TFs, and we unraveled the connection between protein interaction and NB formation. We state this in lines 228-232.

Line 269-272: It's not clear to what the authors are referring to.

Authors:

We are describing the homodimeric behavior of FIT and FITmSS271AA assessed by homo-FRET measurements that are introduced in the previous paragraph, lines 256-268.

Line 309: This colocalization part should be presented before line 194.

Authors:

We find it convincing to first examine and characterize the process underlying FIT NB formation, then studying a possible function of NBs. The colocalization analysis is part of a functional analysis of NBs. We thank the reviewer for the hint that colocalization also confirms that indeed the nuclear FIT spots are NBs. We will take this point and discuss it, lines 516-522:

“Additionally, the partial and full colocalization of FIT NBs with various previously reported NB markers confirm that FIT indeed accumulates in and forms NBs. Since several of NB body markers are also behaving in a dynamic manner, this corroborates the formation of dynamic FIT NBs affected by environmental signals.”

“In conclusion, the properties of liquid condensation and colocalization with NB markers, along with the findings that it occurred irrespective of the fluorescence protein tag preferentially with wild-type FIT, allowed us to coin the term of ‘FIT NBs’.”

Line 328: add the ref to figure, please.

Authors:

Figure reference was added to the text, lines 330-332:

“The second type (type II) of NB markers were partially colocalized with FIT-GFP. This included the speckle components ARGININE/SERINE-RICH45-mRFP (SR45) and the serine/arginine-rich matrix protein SRm102-mRFP (Figure 5).”

Line 334: It seems that the size of the SR45 has an anormal very large diameter between 4 and 6 µm. In general a speckle measure about 2-3µm in diameter. Can the author make sure that this structure is not due to overexpression in N. benthamiana or make sure to not oversaturate the image.

Authors:

Thank you for this hint. Indeed, there are reports that SR45 is a dynamic component inside cells. It can redistribute depending on environmental conditions and associate into larger speckles depending on the nuclear activity status (Ali et al., 2003). We include this reference and refer to it in the discussion, lines 557-564:

“Interestingly, typical FIT NB formation did not occur in the presence of PB markers, indicating that they must have had a strong effect on recruiting FIT. This is interesting because the partially colocalizing SR45, PIF3 and PIF4 are also dynamic NB components. Active transcription processes and environmental stimuli affect the sizes and numbers of SR45 speckles and PB (Ali et al., 2003; Legris et al., 2016; Meyer, 2020). This may indicate that, similarly, environmental signals might have affected the colocalization with FIT and resulting NB structures in our experiments. Another factor of interference might also be the level of expression.”

Line 335: It seems that the colocalization is partial only partial after induction of NB. The FIT NB colocalize around SR45. But it's hard to tell because the images are saturated therefore creating some false overlapping region.

Authors:

The localization of FIT with SR45 is partial and occurs only after FIT has undergone condensation, see lines 335-338.

Line 344-345: It's unappropriated to give the conclusion before the evidence.

Authors:

We explain at an earlier paragraph that we will show three different types of colocalization and introduce the respective colocalization types within separate paragraphs accordingly, see lines 314-321.

Line 353: increase the contrast in the image of t=5 for UAP56H2 since it's hard to assess the colocalization.

Authors:

This is done as noted in the figure legend of Figure 6.

Line 381-382: "In general" does not sound scientific avoid this kind of wording and describe precisely your findings.

Authors:

We rephrased the sentence, line 387-388:

Localization of single expressed PIF3-mCherry remained unchanged at t=0 and t=15 (Supplemental Figure S5A).

Line 384-385: Provide the data and the reference to the figure.

Authors:

We apologize for the misunderstanding and rephrased the sentence, line 389-391:

After 488 nm excitation, FIT-GFP accumulated and finally colocalized with the large PIF3-mCherry PB at t=15, while the typical FIT NBs did not appear (Figure 7A)

Line 386: The structure in which FIT-G is present in the Figure 7A t=15 is not alike the once already observed along the paper. This could be explained by over-expression in N. benthamiana. Please explain.

Authors:

Thank you for the hint. We discuss this in the discussion part, see lines 555-568.

Line 393: Explain and provide data why the morphology of PIF4/FIT NB do not correspond to the normal morphology.

Authors:

Thank you for the valuable hints. Several reasons may account for this and we provide explanations in the discussion, see lines 555-568.

Line 396-398: It seems also from the data that co-expression of PIF4 of PIF3 will affect the portioning of FIT between the NP and the NB.

Authors:

We can assume that residual nucleoplasm is depleted from protein during NB formation. This is likely true for all assessed colocalization experiments. We discuss this in lines 492-494.

The discussion is particularly lengthy it might be great to reduce the size and focus on the main findings.

Authors:

*We shortened the discussion. *

**Referees cross-commenting**

All good for me, I think that the comments/suggestions from Reviewer #2 are valid and fair. If they are addressed they will improve considerably the manuscript.

Reviewer #1 (Significance (Required)):

This manuscript is describing an unprecedent very precise cellular and molecular mechanism in nutrition throughout a large set of microscopies technics. Formation of nuclear bodies and their role are still largely unexplored in this context. Therefore, this study sheds light on the functional role of this membrane less compartment and will be appreciated by a large audience. However, the fine characterization is only made using transient expression in N. Bentamiana and only few proofs of concept are provided in A. thaliana stable line.

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

The manuscript of Trofimov et al shows that FIT undergoes light-induced, reversible condensation and localizes to nuclear bodies (NBs), likely via liquid-liquid phase separation and light conditions plays important role in activity of FIT. Overall, manuscript is well written, authors have done a great job by doing many detailed and in-depth experiments to support their findings and conclusions.

However, I have a number of questions/comments regarding the data presented and there are still some issues that authors should take into account.

Major points/comments:

1) Authors only focused on blue light conditions. Is there any specific reason for selecting only blue light and not others (red light or far red)?

Authors:

There are two main reasons: First, in a preliminary study (not shown) blue light resulted in the formation of the highest numbers of NBs. Second, iron reductase activity assays and gene expression analysis under different light conditions showed a promoting effect under blue light, but not red light or dark red light (Figure 9). This indicated to us, that blue light might activate FIT, and that active FIT may be related to FIT NBs.

2) Fig. 3C and D: as GFP and GFP-GFP constructs are used as a reference, why not taking the measurements for them at two different time points for example t=0 and t=5 0r t=15???

Authors:

Free GFP and GFP-GFP dimers are standard controls for homo-FRET that serve to delimit the range for the measurements.

3) Line 27-271: Acc to the figure 3d, for the Fluorescence anisotropy measurement of NBs appears to be less. Please explain.

Authors:

FA in NBs with FITmSS271AA is variable and the value is lower than that of whole nucleus but not significantly different compared with that in nucleoplasm. We describe the results of Figure 3D in lines 272-275.

4) Figure 4: For the negative controls, data is shown at only t=0, data should be shown at t=5 also to prove that there is no decrease in fluorescence in these negative controls when they are expressed alone without bhlh39 as there is no acceptor in this case.

Authors:

Neither for FIT/bHLH039 nor the FITmSS271AA/bHLH039 pair, there is a significant decrease in the fluorescence lifetime values between t=0 and t=5/15. FIT-G is a control to delimit the range. The interesting experiment is to compare the protein pairs of interest between the different nuclear locations at t=5/15.

5) Line 300-301: In Figure 4D and 4E. Fluorescence lifetime of G measurement at t=0 seems very similar for both FIT-G as well as FITmSS but if we look at the values of t=0 for FIT-G+bhlh039 it is greater than 2.5 and for FITmSS271AA-G+bhlh039 it is less which suggests more heterodimeric complexes to be formed in FITmSS271AA-G+bhlh039. Similar pattern is observed for NBs and NPs, according to the figure 4d and E.

Therefore, heterodimeric complexes accumulated more in case of FITmSS271AA-G+bhlh039 as compared to FIT-G+bhlh039 (if we compare measurement values of Fluorescence lifetime of G of FITmSS271AA-G+bhlh039 with FIT-G+bhlh039).

Please comment and elaborate about this further.

Authors:

These conclusions are not valid as the experiments cannot be conducted in parallel. Since the experiments had to be performed on different days due to the duration of measurements including new calibrations of the system, we cannot compare the absolute fluorescence lifetimes between the two sets.

6) Figure 4: For the negative controls, data is shown at only t=0, data should be shown at t=5 also to prove that there is no decrease in fluorescence in these negative controls when they are expressed alone without bhlh39 as there is no acceptor in this case.

Authors:

Please see our response to your comment 4).

7) Line 439-400: As iron uptake genes (FRO2 and IRT1) are more induced in WT under blue light conditions and FRO2 is less induced in case of red-light conditions. So, what happens to Fe content of WT grown under blue light or red light as compared to WT grown under white light. Perls/PerlsDAb staining of WT roots under different light conditions will add more information to this.

Authors:

We focused on the relatively short-term effects of blue light on signaling of nuclear events that could be related to FIT activity directly, particularly gene expression and iron reductase activity as consequence of FRO2 expression. These are both rapid changes that occur in the roots and can be measured. We suspect that iron re-localization and Fe uptake also occur, however, in our experience differences in metal contents will not be directly significant when applying the standard methods like ICP-MS or PERLs staining.

Minor comments:

Line 75-76: Rephrase the sentence

Authors:

We rephrased the sentence, lines 73-74:

“As sessile organisms, plants adjust to an ever-changing environment and acclimate rapidly. They also control the amount of micronutrients they take up.”

Line 119: Rephrase the sentence

Authors:

We rephrased the sentence, line 118-119:

“Various NBs are found. Plants and animals share several of them, e.g. the nucleolus, Cajal bodies, and speckles.”

Line 235-236: rephrase the sentence

Authors:

We rephrased the sentence, line 232-234:

“In the work of Gratz et al. (2019), the hosphor-mimicking FITmS272E protein did not show significant changes in its behavior compared to wild-type FIT.”

Line 444: Correct the sentence “Fe deficiency versus sufficiency”

Authors:

We corrected that, line 449-451:

“In both, the far-red light and darkness situations, FIT was induced under iron deficiency versus sufficiency, while on the other side, BHLH039, FRO2 and IRT1 were not induced at all in these light conditions (Figure 9I-P).”

**Referees cross-commenting**

I agree with R1 suggestions/comments and i think manuscript quality will be much better if authors carry out the experiments suggested by R1. I believe this will also strengthen their conclusions.

Reviewer #2 (Significance (Required)):

Overall, manuscript is well written, authors have done a nice job by doing several key experiments to support their findings and conclusions. However, the results and manuscript can be improved further by addressing some question raised here. This study is interesting for basic scientists which unravels the crosstalk of light signaling in nutrient signaling pathways.

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

Learn more at Review Commons

---

Reply to the reviewers

We thank all three reviewers for their positive comments on the value of our work and for and for their helpful suggestions.

*Reviewer #1 (Evidence, reproducibility and clarity:

This manuscript reports the development of a new bright fluorescent reporter that allows to label neighbouring cells. The system is based on upon secretion and uptake of s36GFP, a positively supercharged fluorescent protein. The authors also develop a Halo tag that will allow for straight forward colour exchange, as well as a customisable single plasmid construct with modular components.

There are some minor suggestions that the authors may want to consider: 1) The authors conclude that "PUFFIN labelling is transferred rapidly between cells within minutes". They report in their time lapse experiments (Figure 2A,C) that sGFP can be detected within neighbours of secretors after 30 minutes when the cells are plated in a 50:1 non-labelled/secretor cell ratio, whereas it can be detected after 15 minutes when the cells are plated in a 1:9 ratio. Is there any synergistic effect on the signal when the proportion of secretors is increased or is this difference just because there is more signal, making it easier to visualise. *

We have addressed this point with new experiments (new data shown in Figure 2E and Supp Figure S2A,B). This makes it clear that labelling can indeed be detected earlier when the proportion of secretors is higher. This is likely to be because higher secretor:acceptor ratios result in stronger labelling, which in turn makes it easier to detect labelled neighbours at very early time points - even within as early as 15 minutes. We also confirm that, even when secretors are very sparse (1:50 ratio), label becomes detectable in neighbours within 60 minutes.

1. *Is there any reason why the main Figure legends lack a title, but the supplementary figures have one? 2. In Figure 3, it may be helpful to label each option as A, B, C.. 3. In Figure 4E, the legends + JF646 and -JF646 may be switched around. Shouldn't the orange box should be (+) and the grey box should be (-)?

   *

We have modified / corrected the labelling as suggested and added titles to the main figure legends.

*Reviewer #1 (Significance):

This is a very valuable tool to address how cells change the behaviour of those in their environment. It will be very valuable for those interested in cell non-autonomous effects within a cell population or tissue. It will be especially valuable for live cell imaging; pulse chase experiments as well as omics approaches to understand cell behaviour in niches. *

We thank this reviewer for their positive comments on the value of our work.

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

The authors describe a new method, Positive Ultra-bright Fluorescent Fusion For Identifying Neighbours (PUFFFIN), to label with Fluorescent Proteins, neighboring cells. In brief, specific cells that express a nuclear mCherry are engineered to secrete a supercharged fluorescent protein (36GFP) fused to the ultra-bright green-fluorescent mNeonGreen (mNG) (s36GFP). Neighboring cells uptake s36GFP and can be easily visualized. The authors added the human serum albumin signal peptide which is efficiently cleaved to create s36GFP. The PUFFFIN system can also be customized for color-of-choice labelling using HaloTags. A shortcoming of the paper is that it is a method paper established in tissue culture cells with no biological applications. A test of the system in an in vivo model would improve the study. The authors should at least describe specific examples of how the method can be used to answer biological questions. *

We agree that the paper would be improved by demonstrating a biological application of our system. We are currently working on experiments to address a biological question, and will be submitting a revised manuscript containing these data.

*Reviewer #2 (Significance (Required)):

This straightforward and elegant approach is an improvement of current methods that are based on synthetic receptor-ligand interactions as it does not require genetic modification of both 'sender' cells and 'responding' cells. The approach should prove to be an effective and flexible tool for illuminating cellular neighborhoods. An interesting potential application of the method is to effectively deliver proteins fused to s36GFP.

A shortcoming of the paper is that it is a method paper established in tissue culture cells with no biological applications. A test of the system in an in vivo model would improve the study. The authors should at least describe specific examples of how the method can be used to answer biological questions.*

We thank this reviewer for their positive comments on the value of our work. We agree that the paper would be improved by demonstrating a biological application of our system. We are currently working on experiments to address a biological question, and will be submitting a revised manuscript containing these data

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

In this manuscript, the authors introduced a novel cell-neighbor-labeling system named PUFFFIN. PUFFFIN, as well as 'PUFFHalo', offers an elegantly simple method for distinguishing between secretors and receivers, providing users with a versatile tool to label proximate neighbors through the uptake of s36GFP, subsequently permitting their isolation via FACS for subsequent analysis. In addition, this system could be very useful considering of its customizability by exchanging elements, such as tissue-specific promotors, color-of-choice (HaloTag), and genes of interest to cater to the diverse requirements of secretors. Overall, this system is well-designed and characterized, and the claims in this study are mostly supported by the data. However, this neighbor-labeling approach is not efficiently used to obtain biological insights. The following comments are intended to enhance the overall quality of the study:

Major comments: *

1. * In Vidio1, it appears that certain nuclear mCherry+ cells did not secrete s36GFP-mNG during 19hrs recording window. However, in Figure1D and E, these GFP-mCherry+ cells were reported as having a 0% occurrence. This may be the result of either a delay in GFP secretion, or possible mCherry leakiness in unmodified cells. Please provide clarification. *

There is indeed one mCherry+ cell in video 1 that fails to generate s36GFP-mNG signal. This cell, unlike most other cells in the movie, fails to divide or actively migrate during the 19h recording period, but instead is being passively “pushed around” by surrounding cells, and therefore looks to us very much like a dead or dying cell (levels of cell death to tend to be slightly higher than usual during live imaging). We have looked through our other videos and identified only one other example of an mCherry+ GFP-negative cell: this cell is clearly dying because the nucleus disintegrates over the course of the movie.

We considered the possibility that some proportion of secretors may fail to generate signal even if they are healthy. We examined all our FACS analysis data. We detected at most 0.15% of such ‘failed secretors’, and most usually none. We conclude that any mCherry+ GFP- cells exist at extremely low frequencies and/or tend to be dying cells. Either way, they are very unlikely to interfere with interpretation of experimental data.

*Additionally, including representative images of the co-culture experiment in Figure 1.E would enhance the presentation of the data. *

These data have now been added to Supplemental Figure S1 C

*Since the authors mention that s36GFP-mNG labeling was not detectable beyond four cell diameters, it would be helpful to include statistical data regarding the average distances or cell layers that GFP can travel, thus describing the permeation and labeling limit of s36GFP-mNG, adjacent to Figure2C. *

We’ve now quantified the data and provide this information in a new panel (Figure 2D).

*Please comment on the application prospect of this system utilizing in vivo. In addition, comment should be made on the difference of PUFFFIN system and recent reported CILP (PNAS 2023). *

We have added discussion on prospects for using the system in vivo (new text lines 65-67). We have also described the CILP system in the revised introduction, explaining that it is an inducible version of the Cherry Niche system that we describe in our introduction (new text lines 291-294).

*Minor comments: 1. Please include the percentage of GFP+ and GFP- cells in Figure2.D, similar to what is provided in Figure S1.B. *

This is a great suggestion so we have decided to add this information to all flow cytometry histograms within the paper, Figure 2D.

*The '+' and '-' marks in Figure3.E appears to be mismatched with the results, please double-check and correct. *

This has now been corrected.

*I am curious about the interactions between secretors and 'receivers.' As the authors claim 'unbiased labeling' with this system, it's important to investigate whether the uptake abilities of receivers vary among different cell types. In other words, does the system exhibit cell-type preferences among receiver cells? This question could be optionally addressed through co-culture experiments involving secretors, receiver type A, and receiver type B. *

We will perform additional experiments to address the reviewer’s question by directly comparing labelling efficiency across different receiver cell-types.

Reviewer #3 (Significance (Required)):

*This study reported a simple and sensitive system for labeling neighboring cells in vitro, which can be customized by replacing exchangeable components for customized need. With promising application in vitro, this system could be further developed and tested in vivo. Fluorescent protein labeling in neighboring cells has been a topic of study recently, and this manuscript introduced a new tool that is added to such resources, offering a user-friendly and customizable alternative. Overall, this system will be of interest to researchers working on neighbor-cell labeling and study of cell-cell communications. *

We thank this reviewer for their positive comments on the value of our 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 #3

Evidence, reproducibility and clarity

In this manuscript, the authors introduced a novel cell-neighbor-labeling system named PUFFFIN. PUFFFIN, as well as 'PUFFHalo', offers an elegantly simple method for distinguishing between secretors and receivers, providing users with a versatile tool to label proximate neighbors through the uptake of s36GFP, subsequently permitting their isolation via FACS for subsequent analysis. In addition, this system could be very useful considering of its customizability by exchanging elements, such as tissue-specific promotors, color-of-choice (HaloTag), and genes of interest to cater to the diverse requirements of secretors. Overall, this system is well-designed and characterized, and the claims in this study are mostly supported by the data. However, this neighbor-labeling approach is not efficiently used to obtain biological insights. The following comments are intended to enhance the overall quality of the study:

Major comments:

1. In Vedio1, it appears that certain nuclear mCherry+ cells did not secrete s36GFP-mNG during 19hrs recording window. However, in Figure1D and E, these GFP-mCherry+ cells were reported as having a 0% occurrence. This may be the result of either a delay in GFP secretion, or possible mCherry leakiness in unmodified cells. Please provide clarification. Additionally, including representative images of the co-culture experiment in Figure 1.E would enhance the presentation of the data.
2. Since the authors mention that s36GFP-mNG labeling was not detectable beyond four cell diameters, it would be helpful to include statistical data regarding the average distances or cell layers that GFP can travel, thus describing the permeation and labeling limit of s36GFP-mNG, adjacent to Figure2C.
3. Please comment on the application prospect of this system utilizing in vivo. In addition, comment should be made on the difference of PUFFFIN system and recent reported CILP (PNAS 2023).

Minor comments:

1. Please include the percentage of GFP+ and GFP- cells in Figure2.D, similar to what is provided in Figure S1.B.
2. The '+' and '-' marks in Figure3.E appears to be mismatched with the results, please double-check and correct.
3. I am curious about the interactions between secretors and 'receivers.' As the authors claim 'unbiased labeling' with this system, it's important to investigate whether the uptake abilities of receivers vary among different cell types. In other words, does the system exhibit cell-type preferences among receiver cells? This question could be optionally addressed through co-culture experiments involving secretors, receiver type A, and receiver type B.

Significance

This study reported a simple and sensitive system for labeling neighboring cells in vitro, which can be customized by replacing exchangeable components for customized need. With promising application in vitro, this system could be further developed and tested in vivo. Fluorescent protein labeling in neighboring cells has been a topic of study recently, and this manuscript introduced a new tool that is added to such resources, offering a user-friendly and customizable alternative. Overall, this system will be of interest to researchers working on neighbor-cell labeling and study of cell-cell communications.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

The authors describe a new method, Positive Ultra-bright Fluorescent Fusion For Identifying Neighbours (PUFFFIN), to label with Fluorescent Proteins, neighboring cells. In brief, specific cells that express a nuclear mCherry are engineered to secrete a supercharged fluorescent protein (36GFP) fused to the ultra-bright green-fluorescent mNeonGreen (mNG) (s36GFP). Neighboring cells uptake s36GFP and can be easily visualized. The authors added the human serum albumin signal peptide which is efficiently cleaved to create s36GFP. The PUFFFIN system can also be customized for color-of-choice labelling using HaloTags.

A shortcoming of the paper is that it is a method paper established in tissue culture cells with no biological applications. A test of the system in an in vivo model would improve the study. The authors should at least describe specific examples of how the method can be used to answer biological questions.

Significance

This straightforward and elegant approach is an improvement of current methods that are based on synthetic receptor-ligand interactions as it does not require genetic modification of both 'sender' cells and 'responding' cells. The approach should prove to be an effective and flexible tool for illuminating cellular neighborhoods. An interesting potential application of the method is to effectively deliver proteins fused to s36GFP.

A shortcoming of the paper is that it is a method paper established in tissue culture cells with no biological applications. A test of the system in an in vivo model would improve the study. The authors should at least describe specific examples of how the method can be used to answer biological questions.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

This manuscript reports the development of a new bright fluorescent reporter that allows to label neighbouring cells. The system is based on upon secretion and uptake of s36GFP, a positively supercharged fluorescent protein. The authors also develop a Halo tag that will allow for straight forward colour exchange, as well as a customisable single plasmid construct with modular components.

There are some minor suggestions that the authors may want to consider: 1. The authors conclude that "PUFFIN labelling is transferred rapidly between cells within minutes". They report in their time lapse experiments (Figure 2A,C) that sGFP can be detected within neighbours of secretors after 30 minutes when the cells are plated in a 50:1 non-labelled/secretor cell ratio, whereas it can be detected after 15 minutes when the cells are plated in a 1:9 ratio. Is there any synergistic effect on the signal when the proportion of secretors is increased or is this difference just because there is more signal, making it easier to visualise. 2. Is there any reason why the main Figure legends lack a title, but the supplementary figures have one? 3. In Figure 3, it may be helpful to label each option as A, B, C.. 4. In Figure 4E, the legends + JF646 and -JF646 may be switched around. Shouldn't the orange box should be (+) and the grey box should be (-)?

Significance

This is a very valuable tool to address how cells change the behaviour of those in their environment. It will be very valuable for those interested in cell non-autonomous effects within a cell population or tissue. It will be especially valuable for live cell imaging; pulse chase experiments as well as omics approaches to understand cell behaviour in niches.

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer #1

Evidence, reproducibility and clarity

In this manuscript, Hoskins et al describe analyses of the effects of sequence variation on RNA levels, protein levels, and ribosome loading for the COMT gene. They use multiple experimental approaches to assay these levels and report on how sequence differences affect expression. Overall, the paper is interesting in that it presents a very deep dive into the effects of sequence variation on gene expression, including in coding sequences. However, there are some issues with the polysome loading assay technique and there are substantial issues with the figure presentation, which is often confusing.

__Response: __Thanks for the positive assessment of our manuscript and the constructive feedback regarding the issues with the figure presentation. We have addressed all of these below and they have significantly improved the clarity.

• Major comments:*

• 1) Figures:*

• --Fig 1C needs a cartoon description to show where the UTRs are. Y-axis should say "Ribo-seq CPM"*

__Response: __Fig 1C now includes a schematic and the y-axis is updated. Locations of the uORFs are also now included in Fig 1A.

• --Sup Fig 1A confusing, what is "start" what is the point of this panel?*

__Response: __We apologize for the confusing labeling of the panels in Sup Fig 1. “Start” refers to the MB-COMT start codon. We removed this annotation as it is irrelevant to the figure. We included Supplementary Figure 1A to show RNA probing data for the entire transcript. Figure 1A and B only show the regions that encompass the variants assayed in our study.

• --Sup Fig 1B what is PCBP del?*

Response: “PCBP del” refers to deletion of PCBP1/PCBP2 RNA binding protein motifs. The legend now specifies this.

• --Sup Fig 1C what is "uORF B restore"? The description in the figure legend is not interpretable. Draw diagrams of the mutations that tell the reader what was assayed and why it was assayed. Why are there multiplication factors listed (e.g. 1.33X)? The data are depicted on a log scale, which makes it difficult to appreciate the fold-effects of the mutations (e.g. does uORFA mutation increase expression 1.5-fold?). Please calculate median expression values and report them on a bar graph or something like that so readers can interpret the results.*

Response: “uORF B restore” refers to restoration of the endogenous uORF B frame with a silent variant in the Flag tag of the transgene. The multiplication factors listed were the fold change in median fluorescence between each mutant and the template (wild-type) transgene. We retained the figures as they show the raw distribution of fluorescence in each cell line, but in response to the reviewer’s suggestion we included a new figure displaying the effects as a bar graph (Supplementary Figure 1E).

• --Fig 2A. It's hard to understand the cartoon diagram of the expression reporter construct. Why is +Dox shown here? Does that induce transcription?*

__Response: __The reviewer is correct. “+Dox” indicated addition of Doxycycline to induce transcription before the data collection step. We agree that there may have been too much detail in this diagram and have now removed this for simplicity and indicated this in the Methods section.

• --Fig 2B. What's on the x-axis? is it Log2(RNA/gDNA) from sequencing? is it Log2 or Log10 or Ln?*

__Response: __Variant effects in each figure were derived from ALDEx2 analysis, which reports effect size as the median standardized difference between groups. The effect size is not directly interpretable as a log fold change; it takes into account the difference between groups as well as the dispersion. This analysis strategy has been previously demonstrated for analysis of SELEX experiments (Fernandes et al. 2014), which are used to select small populations of cells with specific phenotypes.

ALDEx2 is a robust and principled choice for the analysis of count-compositional datasets, particularly after selection (e.g. sorted cell populations or low-input RNA fractions arising from polysome profiling). While we understand that this choice leads to less easily interpretable effect sizes, the mathematical advantages make ALDEx2 a more appropriate choice for this type of data. In the past, we had used other methods to analyze log frequencies (limma, a frequency based normalization-dependent analysis, as previously employed in Hoskins et al. 2023. Genome Biology) that directly reported fold changes. In our experience, the ALDEx2-derived effect sizes are well-correlated with those estimates (Pearson correlation 0.93 for variants significant at a FDR

• --Fig 2C. What's on the y-axis (same question). I think it's LogX(mutant/wt)RNA level?*

__Response: __For consistency with other figures, we replaced Figure 2C to report the effect size statistic as described above.

• --Fig 2D. What's on the y-axis now? Fold-difference (not log transformed)?*

__Response: __Please see our response above.

• --Fig 2E. The scale bar is flipped vs. normal convention. This is also log transformed, but it's not labeled. Please label as log(whatever) and put the negative values on the left side of the bar (red on the left, blue on the right).*

__Response: __Thanks for the suggestion, we have now updated the scale bar.

--Fig 2F y-axis should say Ribo-seq CPM.

__Response: __Done

• --Fig 3A - please separate the graphs more. Did you sort cells from ROI2 into populations, or just cells from ROI1?*

__Response: __Thanks for the suggestion, we now separate the graphs further. Cells were sorted for both ROI 1 and ROI 2 libraries.

• --Fig3C-F What's the "effect size" mean on these graphs?*

__Response: __Please see the response above regarding the effect size estimate from ALDEx2.

• --Fig3D It looks like the colors have switched for positive / negative "effects" on the heat map*

• compared to Figure 2E. Please define what "median effect" means and be consistent with*

• comparison to figure 2E.*

__Response: __We intentionally inverted colors for Figure 3. The rationale is that a variant causing low protein abundance corresponds to enrichment in P3 compared to gDNA, as opposed to depletion in P3. On the other hand, for effects on RNA abundance and ribosome load, a variant leading to low abundance for these measures is depleted.

• --Figure 4 what does effect size mean, what's the log-transformed scale (log2, 10, etc) same issues from earlier figures.*

__Response: __Please see response above.

• --Figure 5 "effect size"*

__Response: __The same definition of effect size was used with the exception that effect sizes are multiplied by -1 so that color schemes are consistent for deleterious effects.

• 2) "Codon stability" should always be "Codon Stability Coefficient", maybe use "CSC". Otherwise it's confusing.*

__Response: __Thanks for the suggestion. This has been updated throughout the manuscript.

3) Flow cytometry section talks about "RNA fluorescence", which is confusing. You need to explain that it's IRES-driven mCherry as a proxy for the level of RNA first. It would also help to state explicitly that you sorted the cells into four populations, and define them all first before describing the results.

__Response: __We apologize for the use of imprecise language with respect to this reporter. We revised the text to emphasize that mCherry is a proxy for RNA abundance and described the populations first as suggested.

4) What are DeMask scores? How are they related to conservation or amino acid properties? If you define these, you can help the reader interpret the result.

__Response: __Thanks for the suggestion. We now include a conceptual interpretation of the DeMask score in the relevant section. We also include a comparison to a recent large language model for variant effect prediction (ESM1b, Brandes et al. 2023) which is now reported in Supplementary Figure 5C.

5) There are several issues with the Polysome gradient fractionation. The gradients did not separate 40S, 60S, and monosomal fractions, so it's hard to tell how many ribosomes correspond to each peak on the gradient graph in Figure S5. This is probably because the authors used a 20-50% gradient instead of a lower percentage on top. More significantly, variations in the coding region of COMT are likely affecting the polysome association in ways the authors didn't consider. Nonsense codons will simply make the orf a lot shorter, hence fewer ribosomes. This may have nothing to do with NMD. Silent and missense variants may have unpredictable effects because they may make translation faster (fewer ribosomes) or slower (more ribosomes) on the reporter. This could lead to more ribosomes with less protein or fewer ribosomes with more protein. The reporter RNA also has an IRES loading mCherry on it, which probably helps blunt or dampen the effects of the COMT sequence variants on polysome location distribution. Overall, the design of the polysome assay is probably very limited in power to detect changes in ribosome loading (four fractions, limited separation by 20-50 gradient, IRES loading, etc). This is partially addressed in the limitations section, but these issues could be discussed in more detail.

__Response: __Given high polysomal association of endogenous COMT and our COMT transgene (Supplementary Figure 2B, Supplementary Figure 5B-C), we chose a 20-50% sucrose gradient to better resolve changes in ribosome load among heavy polysomes.

We thank the reviewer for offering another valid explanation regarding the depletion of nonsensense variants. We have now included a sentence in the discussion to indicate lower ribosome load for nonsense variants may be due to a shorter ORF as opposed to NMD. We further include the potential limitation of the assay due to the presence of the IRES-mCherry.

We agree that variants may have unpredictable effects due to effects on the dynamics of translation elongation. To address this potential limitation, we attempted to devise a selective ribosome profiling strategy by immunoprecipitating N-terminal Flag tagged peptides to enrich ribosomes translating COMT. However, we were unable to achieve significant enrichment, limiting our ability to measure variant effects on elongation in a high-throughput manner.

Significance

The study is novel in that it assays both 5' UTR and a wide range of protein coding sequence variants for effects on RNA and protein levels from a clinically important gene, COMT. The manuscript reports that most protein coding variants have modest effects on RNA levels, and that the minority of variants that do affect RNA levels are not predictable due to their affect on codon usage. The work also determines the distribution of effects of variants on protein levels, finding a variety of effects on expression. Interestingly, the authors found SNPs that affect ribosome loading generally affect RNA structure of the COMT coding region, rather than affecting codon usage.

This should appeal to many different communities of biologists - gene expression experts, geneticists, and clinical neurobiologists who focus on COMT. So there is a potential for fairly broad interest. The main limitations to the work are in a lack of clarity in the figures and perhaps in the underdeveloped nature of the discussion section. The discussion section reports new results (SNP associations that affect expression). These would make more sense in the results section, such that the discussion could do a better job relating the impact of sequence variants on expression levels to prior work to highlight the novelty.

__Response: __We thank reviewer #1 for their positive assessment of the broad significance of our study. We also thank them for constructive suggestions that led to increased clarity in the presentation. We have moved the analysis of gnomAD variants to the Results section and expanded the discussion.

Reviewer #2

Evidence, reproducibility and clarity

Summary:

Hoskins and colleagues expressed a reporter containing all silent, missense, and nonsense codons at 58 amino acid positions in the human COMT gene in HEK293T cells and measured levels of DNA, bulk RNA, and pooled polysomal mRNA. They included a C-terminal translational GFP fusion and a downstream transcriptional mCherry fusion in the reporter in order to also bin variants by their relative protein and mRNA levels by flow cytometry. They hypothesized that RNA structure, in-part by mediating uORF translation, influences COMT gene expression. The authors conclude by identifying previously-uncharacterized COMT variants that, in this reporter system, affect RNA abundance and ribosome load. We generally found the results of this paper convincing and clear. We do not have major comments, but have many minor comments that we hope the authors can address. These comments mostly deal with clarification on analysis metrics and giving recommendations on data presentation.

__Response: __Thanks for highlighting the strengths of our study and the constructive suggestions to improve the presentation.

Minor comments:

In Figure 2C, the vertical axis reads "Median between-group difference". How was this metric calculated and normalized? We also agree that nonsense mutations having consistently-detrimental effects on RNA abundance is reassuring, but recommend more explanation as to why the difference in the effects of silence and missense mutations between regions may be biologically relevant.

__Response: __Variant effects in each figure derive from ALDEx2 analysis, which reports effect size as the median standardized difference between groups. In particular, to avoid any distributional assumptions for standardization, ALDEx2 uses a permutation based non-parametric estimate of dispersion. The effect size is not directly interpretable as a log fold change; it takes into account the difference between groups as well as the max dispersion of the groups. We have now provided explicit references to the specific R functions that were used to calculate the effect size.

ALDEx2 is robust for analysis of count-compositional datasets, particularly after selection and bottlenecking (e.g. sorted cell populations or low-input RNA fractions arising from polysome profiling). While we have used other methods to analyze log frequencies (limma, a frequency based normalization-dependent analysis, as previously employed in Hoskins et al. 2023. Genome Biology), we opted for the less-interpretable but more robust ALDEx2 analysis to report variant effects between varying nucleic acid inputs.

We currently lack a mechanistic interpretation for the difference in RNA abundance effects between ROI 1 and 2. However, we observed consistent results using a different analysis framework, which makes use of variant frequencies (as in Hoskins et al. 2023 Genome Biology) instead of the centered log ratios used in ALDEx2 analysis, further supporting a biological difference between the two.

In Figure 3, we believe that the authors are claiming that lower RNA abundance causes lower protein abundance in some variants. However, this data only reports on protein abundance relative to transcript abundance, not absolute protein abundance. We think the claim should be revised to (1) clarify that the authors are measuring protein per mRNA, and (2) express that lower mRNA amounts are more likely to co-occur with lower protein amounts, but that this data does not support any causative model.

__Response: __Thanks for the suggestion. We have now included an explicit description of the experimental design in the results section and noted that we are unable to assign protein abundance effects to underlying RNA abundance effects. In the current setup, we did not sort cells based on the ratio of moxGFP/mCherry fluorescence (protein per mRNA), but rather we defined gates based on the 2D plot of moxGFP versus mCherry. This is explicitly marked in Figure 3A.

On page 9, the authors claim that their data supports a model that rs4633 increases RNA

abundance, leading to higher COMT expression. Can the authors rule out a model whereby rs4633 facilitates translation initiation, as suggested by Tsao et al. 2011, leading to both an increase in mRNA and protein abundance?

__Response: __Thanks for this question and opportunity to clarify. We have now added a sentence to the Discussion and included the following paragraph in the Supplementary Note:

“Importantly, our study does not rule out a model where rs4633 facilitates translation initiation. Nevertheless, our data suggest a potential concurrent mechanism where rs4633 leads to higher protein abundance in human cell lines and in an in vitro translation assay (Tsao et al. 2011) by increasing RNA abundance. We note that Tsao et al did not directly measure RNA abundance in their study. In Supplementary Figure 3A of Nackley et al 2006, the APS haplotype containing rs4633 C>T showed slightly higher total RNA abundance compared to the LPS haplotype (in our study, the wild-type template). However, this was not statistically significant and was only observed for the S-COMT isoform. It is possible that our observations are compatible with the conclusions in Tsao et al. 2011. For example, increased translation of rs4633 C>T may lead to stabilization of the RNA.”

The paper references "effect size" at multiple points (e.g. "polysome effect size") but we could not find this term explicitly defined (for example: for the polysome effect size, were RNA counts for each polysome fraction divided by the relative abundance of that RNA in total RNA?)

__Response: __We apologize for this confusion. Please see our response above. We have also stated the definition of effect size explicitly in the revised manuscript.

Could you elaborate on how you define "protein abundance and "effect size: in Figure 5G? How is enrichment in P3 or P1 calculated?

__Response: __Effect size is defined as described above. Enrichment in P3 or P1 is calculated with respect to the abundance in gDNA (unsorted cells).

Were 3396 variants considered for all readouts in this paper? How many of these variants were present in each ROI? It may be worth clarifying sample sizes.

__Response: __Thanks for the suggestion. The reviewer is correct: 3396 variants were present in all biological replicates and all readouts (after excluding polysome metafractions 1 and 2 and flow cytometry population 4). The Methods were updated to include all readouts that were dropped. The number of variants in each ROI are now included in this section of the main text.

How did Twist generate these mutagenized sequences? We assumed that they used error-prone PCR due to the mention of multiple nucleotide polymorphisms, but couldn't find an explicit answer.

__Response: __Twist generates these mutagenized inserts using degenerate primers. This allows all alternate codons to be assayed (all silent, missense changes). This is now noted in the Methods.

https://www.twistbioscience.com/resources/technical-note/solid-phase-dna-synthesis-allows-tight-control-combinatorial-library

In the methods, it may be worth elaborating on the composition of the HsCD00617865 plasmid. For example: this COMT reporter is under the control of a constitutively-expressed T7 promoter, correct?

__Response: __The HsCD00617865 plasmid was only used as a template for PCR amplification and generation of the transgene. The transgene is cloned into a vector containing attB sites for recombination into the landing pad cell line (Matreyek et al 2020). Transcription is induced by Doxycycline from the landing pad locus. Plasmid maps used for transfection into the landing pad line are now included in the GitHub repository.

In Supplementary Figures 4 and 5, it would be helpful to explicitly say that you are reporting Pearson correlations between biological replicates.

__Response: __Thanks for the suggestion. The legends have been updated accordingly.

"After summarizing biological replicates (N=4) for each readout...": how did the authors summarize biological replicates? Were counts averaged?

__Response: __Biological replicates were summarized using the median. This is now clarified in the Methods.

The authors used pairwise correlations between flow cytometry fractions, polysome fractions, and total RNA/gDNA as indications of data quality. Do the authors expect for these counts to be strongly correlated? We would not necessarily expect to see a strong correlation between ribosome load and RNA/gDNA.

__Response: __We used replicate correlation as an indicator of data quality. Our readouts of ribosome load reflect the abundance of a variant in a particular polysome fraction. Given that variants that are highly abundant in the RNA pool will on average be more highly represented in polysome fractions, we would expect a correlation between the abundance of a variant in total RNA and in polysome fractions.

The authors may need to check that their standard deviations on fold changes are properly reported.

__Response: __iIn the Figures and the main text, we specified the confidence intervals as calculated by ALDEx2 method instead of reporting standard deviations on fold changes,. Specifically, the confidence intervals were determined by Monte Carlo methods that produce a posterior probability distribution of the observed data given repeated sampling. Variants in which the confidence intervals do not cross 0 are considered true discoveries (section 5.4.1 of the ALDEx2 vignette on Bioconductor).

https://www.bioconductor.org/packages/devel/bioc/vignettes/ALDEx2/inst/doc/ALDEx2_vignette.html#541_The_effect_confidence_interval

We would expect standard deviation bounds to be symmetric for log fold changes, but not on unlogged fold changes - for example see page 8, for the sentence "our point estimate for nonsense variant effects on COMT RNA abundance was approximately a two-fold decrease relative to the gDNA frequency (fold change of 0.43 +/- 0.13; mean +/- standard deviation; Methods)."

__Response: __Thanks for the suggestion. To avoid any confusion about the symmetry, we replaced the +/- notation, and explicitly noted the mean and standard deviation. To help the reader gain an intuition of the magnitude of variant effects, we conducted a frequency based normalization-dependent analysis using limma (as previously employed in Hoskins et al. 2023. Genome Biology). We now report a fold change (unlogged) for RNA abundance compared to gDNA abundance. The point estimate is the mean and s.d. across all nonsense variants.

On page 10, the authors say that their data suggests that hydrophobicity in the early coding region of COMT may be important for COMT folding. If this is the case, would we expect to see this effect in flow cytometry data (which is affected by protein degradation) and not polysome profiling (which is unaffected by post-translational protein degradation)?

__Response: __We apologize as we are uncertain about the reviewer’s intended question. The section that refers to the importance of hydrophobicity indeed refers to the flow cytometry data. While there are specific instances in which the amino acid properties encoded by the mRNA influences translation dynamics, these are not universally true. Consequently, we did not expect these impacts to be observed at the level of polysome profiling.

We believe that we would have some trouble replicating the analysis from this paper from the raw data, given that the bulk of the analysis on GitHub is presented as a single R Markdown file, with references to local files to which we do not have access. We recommend that the authors add additional documentation to their repository to facilitate re-analysis.

__Response: __Thanks for the opportunity to address this issue of critical importance. To facilitate replication, we have now deposited all analysis files to Zenodo and refactored the code to enable replication by simply running a markdown file.

In Figure 1B, indicating that more signal indicates less structure (in the legend or the figure itself) may assist readers who are unfamiliar with DMS-seq.

__Response: __Thanks for the suggestion. This is now updated.

Figure 1C does a great job presenting evidence for the translation of uORFs, but does not seem to flow with the overall argument of the paper, so may fit better in the supplement.

__Response: __We considered this suggestion, and opted for keeping its placement as it gives evidence that our transgene is translated primarily as the MB-COMT isoform. This ensures that, for variants upstream of the S-COMT isoform, we can assay effects on ribosome load that are tied to mechanisms of translation elongation and codon stability.

We believe there is a typo in the Figure 1 legend that should read "K562" instead of "H562".

__Response: __Thank you, this was indeed a typo.

You also gated to separate into P1-P4, correct? Can you also show the bounds of that gating

strategy in Figure 3A?

__Response: __This has been updated. We also added the gating strategy in response to comments from reviewer #1.

We find Figure 3F very compelling. Do you have any theories as to why mutating I59-H66 to

nonpolar, uncharged residues leads to increased COMT expression?

__Response: __We do not have any theories for why this may be. However, we noted that with the exception of V63, residues I59-H66 are not evolutionarily constrained (based on DeMask entropy values). This suggests mutational tolerance for nonpolar, uncharged residues in this region (with the exception of V63 and H66; see Figure 3D).

There appears to be a non-negligible proportion of di- and tri- nucleotide polymorphisms in Supplementary Figure 4. Were these excluded in downstream analyses?

__Response: __These variants are expected from the Twist mutagenesis strategy and included in analysis. We believe they are at lower frequency compared to SNPs due to less favorable annealing of the degenerate primers.

A minor typo in the discussion reads "fluoresce".

__Response: __Done

Significance

Describe the nature and significance of the advance (e.g. conceptual, technical, clinical) for the field.

This work investigated the regulatory effects of thousands of coding variants in the COMT gene, focusing on two regions with clinical significance, by using high-throughput reporter assays. The results from this will be useful for clinical scientists interested in understanding the impacts of COMT mutations and be a useful framework for other systems/computational biologists to understand the impacts of coding mutations across different levels of regulatory function. Mutations in protein regions, if having a function, are classically known to interfere with protein function. There are fewer large-scale efforts to understand the impacts of coding mutations affecting expression through potentially changing of RNA structure or codon optimization - this work has contributed towards that frontier.

Place the work in the context of the existing literature (provide references, where appropriate). This is (as far as I am aware) the first paper that has integrated high-throughput screens massively parallel reporter assays from RNA degradation, ribosomal load, and flow cytometry. Previous papers have tended to measure on expression regulation on only one dimension (i.e. Greisemer et al. 2023 on RNA degradation, Sample et al. 2019 on ribosomal load, and de Boer at al. 2020 on protein expression).

__Response: __Thanks for highlighting the novelty of our approach compared to existing strategies in the literature.

State what audience might be interested in and influenced by the reported findings.

Clinicians/researchers interested in COMT, computational biologists, geneticists and potentially structural biologists interested in understanding the consequences of amino acid mutations on RNA/protein expression

__Response: __Thanks for noting the broad significance of our study.

Define your field of expertise with a few keywords to help the authors contextualize your point of view. Indicate if there are any parts of the paper that you do not have sufficient expertise to evaluate.

Genomics, Massively parallel reporter assays, High-throughput regulatory screens.

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

*This manuscript reports on transcript sequence variants that affect expression of the gene COMT. Targeted analysis of SNPs identifies 5' UTR variants that affect COMT, leading to the identification of translated uORFs. Common coding sequence SNPs do not affect COMT expression, however. Massively parallel analyses of mRNA abundance, protein abundance, and translation are combined to look more broadly at coding sequence variants. These analyses focus on regions of predicted structure in the COMT transcript. Both silent and missense mutations that increase mRNA abundance are identified. Protein abundance is then measured and many missense mutations are found to change protein levels. To address translation directly, analysis of polysome loading is performed and significant differences are identified, although technical challenges limit data quality in these experiments. These different experiments are then analyzed jointly to classify mutation effects and identify a class of silent mutations with expression effects, leading to a proposal that these act through structure. *

*The joint, integrative analysis of COMT variants through a range of methods allows clearer insights into interconnected post-transcriptional effects. The massively parallel experiments generate high-quality data, although targeted validation of key results would strengthen the work. The findings advance our understanding of silent variant effects, which remains an open question, and technical innovations could find broader applications. *

__Response: __Thanks for the positive assessment of the quality of the data generated and the potential for the broader application of the technical innovations.

*I do have concerns with the present version of this work. *

1. • There is no validation presented for high-throughput experimental data. I would say that validating the effects of M152T and V63V variants from Figure 2B would substantially strengthen the work and support key conclusions. * __Response: __Our experiments collectively enabled nearly 10,000 measurements of variant effect (summed over three layers of gene expression). The goal of our study was to identify broad mechanisms of variant effect. While we are excited about the specific variants uncovered, targeted experimental methods for validating changes to RNA abundance, such as RT-qPCR, are unlikely to be sufficiently sensitive. For example, RNA abundance effects in our study had a median effect size of 1.47 for variants up in RNA, and 0.4 for variants down in RNA. This likely corresponds to less than one Ct difference between the variant and the reference allele. Indeed, previous studies such as Findlay et al., 2018 Nature that reported similar effect sizes (FGF7 and FOS, respectively (Figure 4B).

Thus, for time and cost concerns, we respectfully suggest that targeted experiments involving V63V and M152T are beyond the scope of our study. Nevertheless, to further strengthen our conclusions, we have computationally confirmed our findings using a different analysis framework. We found 75/76 of the variants significant by ALDEx2 analysis were also significant by limma analysis (a frequency based normalization-dependent analysis, as previously employed in Hoskins et al. 2023. Genome Biology) using the same FDR (0.1).

• In the fluorescent reporter scheme, it seems that variants reducing mRNA abundance should be enriched in the "P2" gate region relative to "P1", as they would have lower mRNA abundance and correspondingly lower protein abundance. However, this analysis is not performed, and instead P1 and P3 are compared (Figure 3G), which would seem to focus on protein-level effects. *

__Response: __Our initial hesitation in comparing P2 to P1 is that the P2 population may be enriched for cells that underwent inefficient induction of transcription with Doxycycline. Hence technical factors as opposed to the effect of the variants may dominate this comparison. In response to the reviewer’s comments, we carried out the suggested analysis (new Supplementary Figure 5B). We found that variants that are down in RNA are enriched in P2 relative to P1 as expected. This is now noted in the Results section.

• In general the work classifies variants in several different ways and it would help to be a little clearer in naming these classes. For instance, in describing the FACS-based analysis of variant expression it is written, "protein fluorescence conditioned on RNA fluorescence" which is confusing at best-it's a fluorescence-based measurement that is used indirectly to measure COMT reporter abundance. *

__Response: __Thanks for the suggestion. We agree that our initial word-choice was imprecise. We rewrote this section to indicate mCherry fluorescence is an indirect proxy for RNA abundance.

• Likewise, the populations with shifted GFP/mCherry ratio in this assay are described as "uncorrelated" populations, which is opaque and somewhat inaccurate-there seems to be a correlation in this group but at a different ratio. *

__Response: __We have revised the language in the manuscript. We opted for “low or high RNA/protein abundance” to indicate the relationship between GFP and mCherry fluorescence in populations P3 and P4.

• In the same way, "deleterious variants" is used to describe protein abundance changes, but this term implies a fitness effect and is not very specific. *

__Response: __We apologize for the confusing word choice. We did away with this term in favor of “variants with low protein abundance”.

• In discussing the effects of missense COMT variants on protein levels, there is an implicit assumption that degradation of mis-folded protein (or perhaps properly-folded protein with excess hydrophobic exposure?) explains these effects. This is plausible, but it would help to lay out this reasoning more clearly. *

__Response: __Thanks for the suggestion. We have added a sentence at the end of the section that specifies this assumption and cites a recent study reporting that rare missense variants in COMT may be misfolded and degraded by the proteasome (Larsen et al. 2023).

• It is written that,"In line with codon stability as a predictor of translational efficiency (Presnyak et al., 2015), variants with low codon optimality were depleted from polysomes compared to variants with optimal codons". However, this mis-states the conclusions of the cited study, which notes, "Importantly, under normal conditions the ribosome occupancy of the HIS3 opt and non-opt constructs was determined to be similar (Fig. 6B)". *

__Response: __We apologize for mis-stating the conclusions of Presnyak et al. 2015. We have now revisited the relevant literature to more accurately place our conclusions in the context of literature. While Presnyak et al. and several other studies (Bazzini et al., 2016; Mauger et al., 2019) have clearly linked the association between codon choice and mRNA stability. We now reference Mauger et al. 2019 who used elegant experiments to demonstrate that mRNA secondary structure is a driver of increased protein production and synergizes with codon optimality (Figure 5B). Their results further support the role of codon optimality on RNA stability while providing evidence of additive impact on translation efficiency.

• It is written that, "One intriguing possibility is to develop multiplexed assays of variant effect on RNA folding, using mutational profiling RNA probing methods (Weng et al., 2020; Zubradt et al., 2017)." How would this differ from the "Mutate and Map" approach in doi://10.1038/nchem.1176 and subsequent work from the same group? *

__Response: __Thanks for pointing out the more recent work following the initial papers in 2010-2011. We have missed the work from the Das lab that extended the Mutate and Map approach to utilize mutational profiling (Cheng and Kladwang et al., 2017). We updated our Discussion to indicate that the proposed assay has been pioneered and is a viable approach for high-throughput determination of variant effects on RNA folding.

Because mutational profiling methods leverage reverse transcriptase readthrough and mismatch incorporation, they enable deeper and more uniform coverage of sequencing reads, particularly for longer transcripts. A key design principle of the proposed assay is to mutagenize only certain types of variants in the library such that they do not overlap RT mismatch signatures arising from the RNA probing reagent/RT enzyme. For example, readthrough of DMS base adducts largely generates A>N or C>N mismatches, so a variant library would be designed to only contain variants at G or T bases. This ensures variants in the library can be differentiated from signals of the RNA probing method.

***Referees cross-commenting** *

*I generally agree with the other reviewers and found that many small points on the figures were confusing, and in some cases the values being computed and displayed were under-specified. *

*I agree with Reviewer 1 that the polysome fractionation probably has limited power due to experimental design, and that the interpretation of changed ribosome loading is subtle. *

__Response: __In response to these helpful comments, we have clarified the points highlighted by the reviewers and expanded the limitations section related to the ribosome loading assay. Thanks for these constructive suggestions to strengthen our study.

*Reviewer #3 (Significance (Required)): *

*The joint, integrative analysis of COMT variants through a range of methods allows clearer insights into interconnected post-transcriptional effects. The massively parallel experiments generate high-quality data, although targeted validation of key results would strengthen the work. The findings advance our understanding of silent variant effects, which remains an open question, and technical innovations could find broader applications. *

__Response: __Thanks for pointing out the high-quality of the generated data and the broad significance of our study. The goal of our study was to identify broad mechanisms of variant effect instead of focusing on differential expression for any specific variants.

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 reports on transcript sequence variants that affect expression of the gene COMT. Targeted analysis of SNPs identifies 5' UTR variants that affect COMT, leading to the identification of translated uORFs. Common coding sequence SNPs do not affect COMT expression, however. Massively parallel analyses of mRNA abundance, protein abundance, and translation are combined to look more broadly at coding sequence variants. These analyses focus on regions of predicted structure in the COMT transcript. Both silent and missense mutations that increase mRNA abundance are identified. Protein abundance is then measured and many missense mutations are found to change protein levels. To address translation directly, analysis of polysome loading is performed and significant differences are identified, although technical challenges limit data quality in these experiments. These different experiments are then analyzed jointly to classify mutation effects and identify a class of silent mutations with expression effects, leading to a proposal that these act through structure.

The joint, integrative analysis of COMT variants through a range of methods allows clearer insights into interconnected post-transcriptional effects. The massively parallel experiments generate high-quality data, although targeted validation of key results would strengthen the work. The findings advance our understanding of silent variant effects, which remains an open question, and technical innovations could find broader applications.

I do have concerns with the present version of this work.

1. There is no validation presented for high-throughput experimental data. I would say that validating the effects of M152T and V63V variants from Figure 2B would substantially strengthen the work and support key conclusions.
2. In the fluorescent reporter scheme, it seems that variants reducing mRNA abundance should be enriched in the "P2" gate region relative to "P1", as they would have lower mRNA abundance and correspondingly lower protein abundance. However, this analysis is not performed, and instead P1 and P3 are compared (Figure 3G), which would seem to focus on protein-level effects.
3. In general the work classifies variants in several different ways and it would help to be a little clearer in naming these classes. For instance, in describing the FACS-based analysis of variant expression it is written, "protein fluorescence conditioned on RNA fluorescence" which is confusing at best-it's a fluorescence-based measurement that is used indirectly to measure COMT reporter abundance.
4. Likewise, the populations with shifted GFP/mCherry ratio in this assay are described as "uncorrelated" populations, which is opaque and somewhat inaccurate-there seems to be a correlation in this group but at a different ratio.
5. In the same way, "deleterious variants" is used to describe protein abundance changes, but this term implies a fitness effect and is not very specific.
6. In discussing the effects of missense COMT variants on protein levels, there is an implicit assumption that degradation of mis-folded protein (or perhaps properly-folded protein with excess hydrophobic exposure?) explains these effects. This is plausible, but it would help to lay out this reasoning more clearly.
7. It is written that,"In line with codon stability as a predictor of translational efficiency (Presnyak et al., 2015), variants with low codon optimality were depleted from polysomes compared to variants with optimal codons". However, this mis-states the conclusions of the cited study, which notes, "Importantly, under normal conditions the ribosome occupancy of the HIS3 opt and non-opt constructs was determined to be similar (Fig. 6B)".
8. It is written that, "One intriguing possibility is to develop multiplexed assays of variant effect on RNA folding, using mutational profiling RNA probing methods (Weng et al., 2020; Zubradt et al., 2017)." How would this differ from the "Mutate and Map" approach in doi://10.1038/nchem.1176 and subsequent work from the same group?

Referees cross-commenting

I generally agree with the other reviewers and found that many small points on the figures were confusing, and in some cases the values being computed and displayed were under-specified.

I agree with Reviewer 1 that the polysome fractionation probably has limited power due to experimental design, and that the interpretation of changed ribosome loading is subtle.

Significance

The joint, integrative analysis of COMT variants through a range of methods allows clearer insights into interconnected post-transcriptional effects. The massively parallel experiments generate high-quality data, although targeted validation of key results would strengthen the work. The findings advance our understanding of silent variant effects, which remains an open question, and technical innovations could find broader applications.

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:

Hoskins and colleagues expressed a reporter containing all silent, missense, and nonsense codons at 58 amino acid positions in the human COMT gene in HEK293T cells and measured levels of DNA, bulk RNA, and pooled polysomal mRNA. They included a C-terminal translational GFP fusion and a downstream transcriptional mCherry fusion in the reporter in order to also bin variants by their relative protein and mRNA levels by flow cytometry. They hypothesized that RNA structure, in-part by mediating uORF translation, influences COMT gene expression. The authors conclude by identifying previously-uncharacterized COMT variants that, in this reporter system, affect RNA abundance and ribosome load.

We generally found the results of this paper convincing and clear. We do not have major comments, but have many minor comments that we hope the authors can address. These comments mostly deal with clarification on analysis metrics and giving recommendations on data presentation.

Minor comments:

In Figure 2C, the vertical axis reads "Median between-group difference". How was this metric calculated and normalized? We also agree that nonsense mutations having consistently-detrimental effects on RNA abundance is reassuring, but recommend more explanation as to why the difference in the effects of silence and missense mutations between regions may be biologically relevant.

In Figure 3, we believe that the authors are claiming that lower RNA abundance causes lower protein abundance in some variants. However, this data only reports on protein abundance relative to transcript abundance, not absolute protein abundance. We think the claim should be revised to (1) clarify that the authors are measuring protein per mRNA, and (2) express that lower mRNA amounts are more likely to co-occur with lower protein amounts, but that this data does not support any causative model.

On page 9, the authors claim that their data supports a model that rs4633 increases RNA abundance, leading to higher COMT expression. Can the authors rule out a model whereby rs4633 facilitates translation initiation, as suggested by Tsao et al. 2011, leading to both an increase in mRNA and protein abundance?

The paper references "effect size" at multiple points (e.g. "polysome effect size") but we could not find this term explicitly defined (for example: for the polysome effect size, were RNA counts for each polysome fraction divided by the relative abundance of that RNA in total RNA?)

Could you elaborate on how you define "protein abundance and "effect size: in Figure 5G? How is enrichment in P3 or P1 calculated?

Were 3396 variants considered for all readouts in this paper? How many of these variants were present in each ROI? It may be worth clarifying sample sizes.

How did Twist generate these mutagenized sequences? We assumed that they used error-prone PCR due to the mention of multiple nucleotide polymorphisms, but couldn't find an explicit answer.

In the methods, it may be worth elaborating on the composition of the HsCD00617865 plasmid. For example: this COMT reporter is under the control of a constitutively-expressed T7 promoter, correct?

In Supplementary Figures 4 and 5, it would be helpful to explicitly say that you are reporting Pearson correlations between biological replicates.

"After summarizing biological replicates (N=4) for each readout...": how did the authors summarize biological replicates? Were counts averaged?

The authors used pairwise correlations between flow cytometry fractions, polysome fractions, and total RNA/gDNA as indications of data quality. Do the authors expect for these counts to be strongly correlated? We would not necessarily expect to see a strong correlation between ribosome load and RNA/gDNA.

The authors may need to check that their standard deviations on fold changes are properly reported. We would expect standard deviation bounds to be symmetric for log fold changes, but not on unlogged fold changes - for example see page 8, for the sentence "our point estimate for nonsense variant effects on COMT RNA abundance was approximately a two-fold decrease relative to the gDNA frequency (fold change of 0.43 +/- 0.13; mean +/- standard deviation; Methods)."

On page 10, the authors say that their data suggests that hydrophobicity in the early coding region of COMT may be important for COMT folding. If this is the case, would we expect to see this effect in flow cytometry data (which is affected by protein degradation) and not polysome profiling (which is unaffected by post-translational protein degradation)?

We believe that we would have some trouble replicating the analysis from this paper from the raw data, given that the bulk of the analysis on GitHub is presented as a single R Markdown file, with references to local files to which we do not have access. We recommend that the authors add additional documentation to their repository to facilitate re-analysis.

In Figure 1B, indicating that more signal indicates less structure (in the legend or the figure itself) may assist readers who are unfamiliar with DMS-seq.

Figure 1C does a great job presenting evidence for the translation of uORFs, but does not seem to flow with the overall argument of the paper, so may fit better in the supplement.

We believe there is a typo in the Figure 1 legend that should read "K562" instead of "H562".

You also gated to separate into P1-P4, correct? Can you also show the bounds of that gating strategy in Figure 3A?

We find Figure 3F very compelling. Do you have any theories as to why mutating I59-H66 to nonpolar, uncharged residues leads to increased COMT expression? There appears to be a non-negligible proportion of di- and tri- nucleotide polymorphisms in Supplementary Figure 4. Were these excluded in downstream analyses?

A minor typo in the discussion reads "fluoresce".

Significance

Describe the nature and significance of the advance (e.g. conceptual, technical, clinical) for the field.

This work investigated the regulatory effects of thousands of coding variants in the COMT gene, focusing on two regions with clinical significance, by using high-throughput reporter assays. The results from this will be useful for clinical scientists interested in understanding the impacts of COMT mutations and be a useful framework for other systems/computational biologists to understand the impacts of coding mutations across different levels of regulatory function. Mutations in protein regions, if having a function, are classically known to interfere with protein function. There are fewer large-scale efforts to understand the impacts of coding mutations affecting expression through potentially changing of RNA structure or codon optimization - this work has contributed towards that frontier.

Place the work in the context of the existing literature (provide references, where appropriate).

This is (as far as I am aware) the first paper that has integrated high-throughput screens massively parallel reporter assays from RNA degradation, ribosomal load, and flow cytometry. Previous papers have tended to measure on expression regulation on only one dimension (i.e. Greisemer et al. 2023 on RNA degradation, Sample et al. 2019 on ribosomal load, and de Boer at al. 2020 on protein expression).

State what audience might be interested in and influenced by the reported findings.

Clinicians/researchers interested in COMT, computational biologists, geneticists and potentially structural biologists interested in understanding the consequences of amino acid mutations on RNA/protein expression

Define your field of expertise with a few keywords to help the authors contextualize your point of view. Indicate if there are any parts of the paper that you do not have sufficient expertise to evaluate.

Genomics, Massively parallel reporter assays, High-throughput regulatory screens.

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, Hoskins et al describe analyses of the effects of sequence variation on RNA levels, protein levels, and ribosome loading for the COMT gene. They use multiple experimental approaches to assay these levels and report on how sequence differences affect expression. Overall, the paper is interesting in that it presents a very deep dive into the effects of sequence variation on gene expression, including in coding sequences. However, there are some issues with the polysome loading assay technique and there are substantial issues with the figure presentation, which is often confusing.

Major comments:

1. Figures: Fig 1C needs a cartoon description to show where the UTRs are. Y-axis should say "Ribo-seq CPM"

Sup Fig 1A confusing, what is "start" what is the point of this panel?

Sup Fig 1B what is PCBP del?

Sup Fig 1C what is "uORF B restore"? The description in the figure legend is not interpretable. Draw diagrams of the mutations that tell the reader what was assayed and why it was assayed. Why are there multiplication factors listed (e.g. 1.33X)? The data are depicted on a log scale, which makes it difficult to appreciate the fold-effects of the mutations (e.g. does uORFA mutation increase expression 1.5-fold?). Please calculate median expression values and report them on a bar graph or something like that so readers can interpret the results.

Fig 2A. It's hard to understand the cartoon diagram of the expression reporter construct. Why is +Dox shown here? Does that induce transcription?

Fig 2B. What's on the x-axis? is it Log2(RNA/gDNA) from sequencing? is it Log2 or Log10 or Ln?

Fig 2C. What's on the y-axis (same question). I think it's LogX(mutant/wt)RNA level?

Fig 2D. What's on the y-axis now? Fold-difference (not log transformed)?

Fig 2E. The scale bar is flipped vs. normal convention. This is also log transformed, but it's not labeled. Please label as log(whatever) and put the negative values on the left side of the bar (red on the left, blue on the right).

Fig 2F y-axis should say Ribo-seq CPM.

Fig 3A - please separate the graphs more. Did you sort cells from ROI2 into populations, or just cells from ROI1?

Fig3C-F What's the "effect size" mean on these graphs?

Fig3D It looks like the colors have switched for positive / negative "effects" on the heat map compared to Figure 2E. Please define what "median effect" means and be consistent with comparison to figure 2E.

Figure 4 what does effect size mean, what's the log-transformed scale (log2, 10, etc) same issues from earlier figures.

Figure 5 "effect size" 2. "Codon stability" should always be "Codon Stability Coefficient", maybe use "CSC". Otherwise it's confusing. 3. Flow cytometry section talks about "RNA fluorescence", which is confusing. You need to explain that it's IRES-driven mCherry as a proxy for the level of RNA first. It would also help to state explicitly that you sorted the cells into four populations, and define them all first before describing the results. 4. What are DeMask scores? How are they related to conservation or amino acid properties? If you define these, you can help the reader interpret the result. 5. There are several issues with the Polysome gradient fractionation. The gradients did not separate 40S, 60S, and monosomal fractions, so it's hard to tell how many ribosomes correspond to each peak on the gradient graph in Figure S5. This is probably because the authors used a 20-50% gradient instead of a lower percentage on top. More significantly, variations in the coding region of COMT are likely affecting the polysome association in ways the authors didn't consider. Nonsense codons will simply make the orf a lot shorter, hence fewer ribosomes. This may have nothing to do with NMD. Silent and missense variants may have unpredictable effects because they may make translation faster (fewer ribosomes) or slower (more ribosomes) on the reporter. This could lead to more ribosomes with less protein or fewer ribosomes with more protein. The reporter RNA also has an IRES loading mCherry on it, which probably helps blunt or dampen the effects of the COMT sequence variants on polysome location distribution. Overall, the design of the polysome assay is probably very limited in power to detect changes in ribosome loading (four fractions, limited separation by 20-50 gradient, IRES loading, etc). This is partially addressed in the limitations section, but these issues could be discussed in more detail.

Referees cross-commenting

I generally agree with the other reviews. Reviewer 2 asks a lot of clarifying questions, which is in line with my comments and suggestions to clarify the presentation of results in the figures. Reviewer 3 has some similar comments and also asks for validation of a few of the MPRA results, which I agree would strengthen the manuscript.

Significance

The study is novel in that it assays both 5' UTR and a wide range of protein coding sequence variants for effects on RNA and protein levels from a clinically important gene, COMT. The manuscript reports that most protein coding variants have modest effects on RNA levels, and that the minority of variants that do affect RNA levels are not predictable due to their affect on codon usage. The work also determines the distribution of effects of variants on protein levels, finding a variety of effects on expression. Interestingly, the authors found SNPs that affect ribosome loading generally affect RNA structure of the COMT coding region, rather than affecting codon usage.

This should appeal to many different communities of biologists - gene expression experts, geneticists, and clinical neurobiologists who focus on COMT. So there is a potential for fairly broad interest. The main limitations to the work are in a lack of clarity in the figures and perhaps in the underdeveloped nature of the discussion section. The discussion section reports new results (SNP associations that affect expression). These would make more sense in the results section, such that the discussion could do a better job relating the impact of sequence variants on expression levels to prior work to highlight the novelty.

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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements

We would like to thank the reviewers for their critical input on the manuscript and we are glad that, overall, they recognize that the extensive analysis of the endochondral perinatal bone we describe in this manuscript can constitute a useful resource both for the bone development and hematopoietic fields. Their input has allowed us to revise the manuscript such that it is much improved in our opinion. In this section, we wish to comment on the main common aspects raised by the reviewers, while specific point-by-point responses are provided below.

Fist, we are aware of the lack of functional assays mentioned by the reviewers, a limitation we explicitly mentioned in our original manuscript. While this is certainly a direction we will take in the future, we consider that such experiments are out of the scope and intentions of our study, given the magnitude of the resources and time they require (e.g. generation of new mouse alleles for cell fate tracking or selective ablation of specific populations, cell transplants into immunocompromised newborns, etc.). As stated in our original manuscript, this study is meant to be a resource that provides new findings and hypotheses that might be relevant for more specialized groups to functionally evaluate (e.g. teams working on thymus seeding progenitors, on adipogenesis or on immune tolerance in newborns, to name a few). As such, we believe our work has an intrinsic value. In fact, this is the first study with single cell resolution that not only compares bone populations before and after birth and with the adult tissue, but also one of the few in which all cell compartments (mesenchymal, endothelial and hematopoietic) are considered. Our manuscript hence brings a new layer of analysis not available in more directed studies, such as those based on flow cytometry (FC), in which not all populations are detected, either by lineage fraction discrimination or due to the lack of surface markers with validated antibodies for FC. This is relevant as our study identifies several new cluster-specific genetic markers and reveals their dynamic/changing expression (perinatal vs adult), or identifies that loci previously targeted for lineage tracing studies are not cluster-specific, which in our view will be useful for the interpretation of previous reports.

The other major point brought up in the reviewers’ reports is that our analysis would be nicely complemented by the spatial localization in the perinatal bone of the various populations we describe in our study. We also agree with the reviewers on this point, which we had considered, but for which we found severe technical limitations. Spatial transcriptomics with cellular resolution would be the ideal method to address this aspect, and we tested two different methods on our samples and under several fixation and permeabilization conditions. Unfortunately, and in contrast to brain tissue used as control, these attempts have been unsuccessful in consistently detecting even ubiquitous transcripts in perinatal bone samples. As spatial transcriptomics is a technology in constant development and several new platforms and approaches are becoming available, we expect that one or several of these various methods, at the moment mostly optimized for soft tissues, will be eventually set-up for mRNA detection with true cellular resolution in perinatal and adult bone samples.

Finally, immunofluorescence (IF) against specific markers is not a suitable approach in this case to unequivocally localize related cell populations such as the ones we describe (e.g. fibroblastic clusters). While flow cytometry has the unique advantage of performing lineage exclusion using cocktails of antibodies conjugated to the same fluorophore to label populations of cells which are not the aim of the study (e.g. hematopoietic and endothelial cells can be excluded by the use of TER119 plus CD45 and CD31, respectively), IF would require the availability of multiple specific antibodies, each conjugated to a different fluorophore, which are not available. In this regard, we would also like to point out that several studies that report the localization of specific cell populations in the bone have done so by taking advantage of genetic reporters (e.g. knock-in alleles encoding intracellular GFP or RFP). As previously mentioned, we consider that the generation of such new genetic tools is out of the scope of this manuscript.

1. Point-by-point description of the revisions

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

* In this new manuscript, Rueda and colleagues present an extensive bioinformatics analysis of single cell transcriptomic data obtained for mouse endochondral bone cell populations before and after birth. They describe gene markers of mesenchymal and hematopoietic cells pointing to differences with adult bone populations, and they use gene ontology and trajectory analyses to infer possible roles of these cells in the developing bone. The data could provide a valuable resource for further understanding endochondral bone development and the changes driving this process in peri-natal stages. However, they are also significant weaknesses. *

* A major weakness is that the scRNA-seq data lack validation through other techniques and functional assays. Namely, in situ data are missing to locate the various cell populations in the developing bones, especially the different types of fibroblastic cells identified by the authors. Such data would go a long way to understand the possible functions of the cell populations. Although the authors tried to complement their data with a review of the literature, most of the conclusions remain purely speculative and not sufficiently supported by scientific and statistic rigor. This makes the Results section more like a discussion than a description of the results. For instance, the authors proposed important regulatory functions for the fibroblastic clusters, but there is no data supporting this other than broad GO terms associated with genes expressed in these cells. Related to this point, the title of the manuscript does not accurately reflect the content of the study**. *

We thank the reviewer for the critical evaluation of our study and for considering it of potential interest for the field, and we have revised the manuscript to take into account his/her comments. We agree with the reviewer that including data to localize in situ the different cell populations would be highly informative. In fact, we had already attempted to perform these experiments using one of the most validated approaches, in situ sequencing (ISS). Despite assaying several fixation and permeabilization conditions, we could not reliably detect even ubiquitously expressed genes in all cells in PN1 bone sections. After discussing with a number of providers that have recently launched instruments capable of performing spatial transcriptomics technology, they all agreed that bone tissue is generally difficult to use for spatial transcriptomics technology. In summary, this data suggests that further optimization of ISS or of alternative spatial transcriptomics approaches will be needed in the future to robustly detect transcripts in bone sections with cellular resolution so as to localize in situ the various cell populations we describe in our study.

Finally, and given our attempt to interpret our analysis of the scRNAseq data in the context of the vast literature that considers both the mesenchymal and the hematopoietic compartments, we agree with the reviewer on the speculative nature of some our conclusions that he/she mentions at the end of the paragraph, an aspect also brought up by the other reviewers. Hence, starting with the title (being now “The cellular landscape of the endochondral bone during the transition to extrauterine life”), we have systematically modified such statements throughout the text to accurately make this distinction.

Other points: 1. The authors missed to report in the Results section which skeletal elements they used for their analyses and which skeletal elements were used for the adult dataset that they compared their data with. Differences in skeletal elements and in the ways whereby these samples were collected and processed could explain differences detected in the two types of datasets. Also, the sex and age of the samples for the adult dataset should be reported.

We now state also in the Results section that we collected forelimb long bones (excluding the handplate) for perinatal stages. In addition, we also indicate that the benchmark study by Baccin et al. used adult bone samples of mixed origin (femurs, tibiae, hips and spines) from 8-12 weeks old females. We agree with the reviewer that both this difference, as well as those related to the extraction protocols, might contribute to some of the variability we report. We now mention both these possibilities in the Discussion.

• It is unclear whether PN1 is the day that mice are born (classically referred to as P0) or the next day.*

As the reviewer indicates, P0 is the day of birth, and PN1 is the following day, which is the stage we chose for analysis. We have now indicated this clearly in the Materials and Methods section.

• It is unclear whether the cells obtained for each biological replicate were pooled for the scRNA-seq assays or were treated individually. It is thus unclear how reproducible the data are.*

In order to capture biological variability, each sample represented pooled littermates (5 fetuses for E18.5 and 4 pups at PN1), and processed as a single scRNA-seq library per stage to minimize technical variation. As our samples contained individuals from both sexes, already indicated in the original manuscript, we have now deconvoluted our datasets and computed male/female cell clustering so as to capture biological variability in duplicates (except for the sex, which is not considered as highly relevant at these stages). We assigned a “female” or “male” sex to a cell if it had at least one transcript read from a female or male specific transcript, respectively. If cells had at least one transcript read from both male and female specific genes, the cell was tagged as “undetermined”. Cells without any sex-specific transcript reads were tagged as “NA”. For the E18.5 sample we identified 21% female cells, 42% male cells, 3.7% undetermined and 33.3% NA cells. For the PN1 sample we identified 42.3% female cells, 28.1% male cells, 4.4% undetermined and 25.2% NA cells. This analysis, now shown in the new Fig. S2 and explained in Materials and Methods, reveals that all mesenchymal, hematopoietic and endothelial clusters are detected in both biological replicates. Finally, the changes we highlighted in the manuscript in the mesenchymal compartment between E18.5 and PN1 (TC, SPF and AFP) are maintained independently if the cells are processed as a single pool per stage or separated according to sex.

• It is not clear in the gating strategy chosen for the flow cytometry as shown in Fig. 1A why the green gate containing cells expressing high levels of CD9, CD140 and CD31 has been extended in between the purple and orange gates containing CD140 and CD31 negative cells.*

While the option mentioned by the reviewer is certainly plausible, this would have diluted the number of hematopoietic cells with intermediate CD9 levels present in our datasets. As our aim was to make sure even less abundant populations from all compartments would be captured in the scRNAseq libraries, we selected the sorting strategy depicted in Fig. 1A.

• Are cells from all the sequenced samples homogenously distributed in the scRNA-seq clusters? Authors should provide this information and add statistic when they describe changes in the amount of cells per cluster between E18.5 and PN1 stages.*

As mentioned in comment 3, we have now deconvoluted the datasets according to sex, which shows all clusters are represented in both biological duplicates and that overall follow similar trends in the E18.5 and PN1 samples.

• On the basis of what markers the AFP population has been called adipogenic? Authors present Ptch2 and Notch3 as markers of this cluster, but not adipogenic progenitor genes.*

Fig. 1C represents differentially expressed markers between clusters, which is why we chose these two representative markers for the AFP population. AFP cells also express adipogenic genes such as Pparg, Lpl or Gas6, although not exclusively. Cluster annotation is based on their molecular signature per se, GO and SCENIC analysis, which identified adipogenic regulons as active in the AFP cluster (see Fig. S10).

• Authors claim that there is a good correlation between OsC and osteo-CAR clusters. However, OsC cells do not express Cxcl12 and other typical CAR cell markers.*

We thank the reviewer for raising this very important point, as both our study and other recent ones (Liu et al., 2022, doi.org/10.1038/s41467-022-28775-x; Kara et al., 2023, doi:10.1016/j.devcel.2023.02.003), show that the most representative genes that historically define CARs (e.g. Cxcl12-high and LepR) are still not expressed at these stages, which indicates that these cells are not yet present at perinatal stages. Accordingly, we did not annotate any perinatal cluster as CAR cells. However, we did observe that other genes such as Runx2, Sp7, Spp1 or Alpl define populations belonging to the OsC cluster that map to the same integrated coordinates as the adult osteo-CAR cluster defined by Baccin et al. (Fig. 2C, bottom panel and Fig. S3, bottom panels). These observations stress the importance of performing ontogenic analysis for each marker defining specific populations, and that data obtained from adult tissue cannot be extrapolated to perinatal stages. We have also corrected the figure legend, which was certainly confusing in this respect.

• In Figure 6 expression of PaS cell markers should be shown for both adult and perinatal populations. Additionally, have the authors tested that the sorted cells in panel C have the same progenitor properties as the PaS cells?*

As requested by the reviewer, we have added the expression of PaS cell markers in adults to Fig. 6 (new panels in Fig. 6B). We are certainly considering exploring in the future the progenitor properties of the sorted cells in comparison to PaS, but these in vivo experiments will require extensive experimentation such as kidney subcapsular transplants in newborns in an immunocompromised background. We consider that these complex in vivo experiments are out of the scope of this manuscript, conceived as a resource paper.

Reviewer #1 (Significance (Required)):

* *As indicated in the comments for the authors, the new scRNA-seq data could become a useful resource for subsequent studies, but they are at present insufficient to represent a significant scientific advancement. The main concern is that new cell populations appear to have been identified by the authors, but a number of questions were not answered such as regarding their actual location in the skeletal elements, their origins, their fates and their functions. Generating such data would require a major amount of effort and require substantial revision of the manuscript.

Our study uncovers, in an unbiased and unsupervised manner, the heterogeneity of the entire perinatal bone with cellular resolution. As the reviewer points out, addressing the origin, fates and functions of the various cell clusters we describe would require a major financial effort and years to be completed. We consider that those aims are well beyond the aims of our manuscript, which is intended as a resource for the large scientific community in the field.

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

* **In the study by Rueda et al. the authors use single cell RNA-sequencing to investigate differences in cellular composition bones/bone marrow between late gestational stage mouse embryos and their perinatal counterparts. The authors describe specific differences in the relative abundance of putative cell types and use established bioinformatic tools to infer interactions as well as molecular mechanisms determining specific functions. The employed methods are well described and the results are presented in a very clear and understandable manner. Despite that, the findings do not provide any substantial knowledge advance but are rather confirming work of published literature while supplementing available single cell RNA-sequencing datasets of mouse bones at adult ages. As such, this work provides an interesting resource but does not report novel biology.

Major comment: The authors explore the interesting transition of embryonic to perinatal bone/bone marrow using single RNA-sequencing. This fills a gap for the field of bone and hematopoietic researchers. There is little to criticize about the presented data. However, while it provides a nice resource, the knowledge gained is incremental. As acknowledged by the authors themselves, their study lacks functional validation of any findings made or conclusions drawn from bioinformatic tools in this manuscript. They use published work to validate their findings but do not go beyond that to confirm putative new biology. Some examples are listed in the minor comments.*

We thank the reviewer for the overall positive comments on our manuscript as a resource study and his/her critical input that we have taken into account when preparing the revised version of the manuscript. Despite the lack of functional validation (already discussed in the General Statements section), we feel that our molecular analysis does provide new valuable insight into the biology of the perinatal bone. For instance, this is the first report that categorizes the heterogeneity of all perinatal bone populations with single cell resolution, and the first that explores the cellular changes in the bone that accompany birth. It also provides an important resource for the generation of more specific genetic models for cell fate tracking or for the interpretation of previous results. Finally, while it is only an inference, our interactome analysis predicts interactions between specific mesenchymal and hematopoietic populations, opening new possibilities for specialists in the specific fields to functionally address in a directed manner (e.g. interactions between the mesenchymal compartment and the Eo/Bas or the ICL-TSP2 subpopulations, which, to the best of our knowledge, have not been previously postulated).

*Minor comments: *

1. * Remark: 10X Chromium does not provide whole transcriptomic coverage but rather captures the most highly abundant transcripts without for example being able to distinguish alternatively spliced gene variants. Based on that, interpretation of gene expression, or the absence of a gene in the dataset, should be interpreted carefully.*

The reviewer is correct, as the method used only captures the 3’UTR of each transcript. We have therefore added a sentence in Materials and Methods to address the limitation of the method. Still, our approach is widely used in the field, as it allows capturing several thousand cells and one facilitating the direct comparison with other datasets, as we ourselves did when integrating the adult dataset from Baccin et al.

• The fact that bone marrow adipose tissue begins to accumulate after birth is well known. It is therefore not surprising that adipogenic progenitor populations start to accumulate perinatally (established by studies cited by the authors). Thus, these results only confirm the validity of the dataset. This represents an example on how the majority of findings have been presented here.*

We fully agree that some of our results confirm, at the single cell level, knowledge previously stablished with other methods. However, and continuing with the case of adipose tissue mentioned by the reviewer, the analysis of our datasets with unbiased tools allowed the identification of fibroblastic populations, such as AFP or GFP, which shown by GO terms and, most importantly, by highly-relevant regulons identified by SCENIC, to be potentially associated with thermogenesis and brown fat differentiation. As far as we know, the specific transcriptional regulators involved in brown fat differentiation in the bone are poorly defined. In addition, adipogenesis is not the only aspect we highlight, being other novel association the putative interaction between fibroblastic mesenchymal populations and Eo/Bas and ILC-ISLP2 hematopoietic cells. These are just two examples of relevant aspects uncovered by that our holistic analysis of all bone population, and that might be further explored by specialized groups in the respective fields.

• Given that the authors do not provide functional validation of putative new molecular interactions (by CellPhoneDB) their conclusions should be presented in a more tempered manner and acknowledged as inference rather than fact.*

We agree with the reviewer and accordingly, we have tuned-down several statements throughout the text (see also response to Reviewer #1).

• Similarly, the authors claim "...we identified Ptx4 as a novel tenogenic-specific gene...". This is too strong a conclusion as this has not been functionally validated. It should at least be tested by immuno-(co)-staining.*

Being Ptx4 a secreted molecule, it would be very difficult to reliably assign the signal to a specific population by co-immunolocalization with bona fide tenogenic markers such as Scx or Tnmd. Besides, when pointing out Ptx4 specific expression in the tenogenic branch of the TC cluster, we intended to suggest the potential use of this locus for the generation of novel genetic tools. We have reformulated this sentence to clearly indicate this and avoid claiming that Ptx4 is a novel tenogenic marker.

• The authors identify "uncommitted clusters" as mesenchymal progenitor populations without actual showing that they are even related by lineage. This is a general pitfall in analyzing single cell RNA-sequencing data and making trajectory/pseudotime inferences. It is now well-established that the mesenchymal compartment is highly heterogeneous and composed of multiple distinct cellular lineages. Trajectory inference tools such as PHATE do not distinguish those different mesenchymal lineages. As such, the presented results cannot be considered valid unless there is proper functional validation.*

We agree with the reviewer on the limitation of this type of analysis, such as its inability to resolve phenotypic convergence (e.g. the case for osteoblasts generated from reprogrammed hypertrophic chondrocytes or from perichondrial cells). We have therefore removed the PHATE data from the manuscript.

• The description of PaS being mainly associated with compact bone is neither correct nor supported by cited studies. The authors show potential additional markers to target Pas in mice, but fail to validate their point that these markers could be used in human tissue as well.*

We thank the reviewer for pointing this out and we apologize for our incorrect wording. What we intended to mean is that PaS cells can only be efficiently extracted by enzymatic treatment of the bone fraction after bone marrow aspiration in adults. We have now corrected these instances in the revised manuscript.

Concerning the validation in human samples of the proposed additional markers for the PaS population, we agree that this is an important point, but one that would require the processing of fetal/newborn human bone tissue for FC, which is beyond our capacities and the scope of the current manuscript

Figure 1c: the legend for dotsize is off scale.

We thank the reviewer for spotting this mistake, which inadvertedly happened during figure assembly and is now corrected.

Reviewer #2 (Significance (Required)):

• Strength:
• thorough analysis of single cell RNA-sequencing datasets including integration of published work
• good writing and figure presentation
• dataset fills gap for the field as the presented ages have not been published

Limitations: - lacking functional validation - lack of new biology - mostly confirmation of known facts

Advance: - knowledge gain is incremental - good resource

Audience: - fills a gap in the bone and hematopoietic research field as a resource

My expertise: Skeletal stem cell lineage biology, single cell RNA-sequencing of bone cell populations.*

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

*__

Summary The authors present data on a very interesting model, mouse bone just before and after birth. In this timeframe, the organism has to adapt from a buoyant, nurtured environment to stronger gravitational forces acting upon the skeletal structure, changed oxygen uptake, changed demands to the immune system and its development, and an overall changed metabolism. The authors introduce these changes and their importance in a clear, easy-to read introduction, and this clear structure and language continue throughout the manuscript. Comparing scRNA-seq of bone E18.5 and adult stage, comparable findings by Liu Y. et al., (https://doi.org/10.1038/s41467-022-28775-x) have been previously shown. However, this manuscript showed additional postnatal day 1 (P1) data. All computational analyses are well done, for the most part well described, and, notably, the integration of previously published data allows us to put the results of this study into context and compare them to the adult situation. Data sharing is not optimal, but it is already very good. The only downside is that most of the computational analyses are done at a very limited level of depth and merely provide initial insights and an overview of the data presented.

We thank the reviewer for the overall positive view of our manuscript and his/her critical comments which we have tried to address in our revisions. We wish to apologize for our omission in citing the Liu et al. study, which is now corrected in the revised text. In this respect, we would like to point out that the Liu et al. study is mostly centered in the endothelial compartment, whereas our work is more focused on the mesenchymal populations. Hence, both studies are complementary. Of note, Liu et al. were able to detect Wnt2 expression in E18.5 endothelial cells using targeted single-cell RNA-seq for a panel of specific genes, while in our data, more focused on the mesenchymal compartment, Wnt2 expression maps mostly to the SFP fibroblastic cluster, with low expression in few endothelial cells. In our view, this apparent discrepancy is not such, but the result of different strategies of sorting and enrichment, and illustrates the need of having complementary studies and datasets (e.g the SFP populations may also be an additional source of Wnt2 to promote hematopoietic stem and progenitor cell proliferation, as reported by Liu et al.).

As for the limited depth on our analysis mentioned by the reviewer, we would like to point out that we made a major effort to put our observations in the context of the vast literature on both the mesenchymal and hematopoietic compartments, which forced us to synthetize in the main text. When possible, we added additional data as part of the supplementary information (e.g. full CellPhoneDB inferred interactions as Excel tables).

*Further comments will be given in bullet-point form, split by their impact on the overall message of the manuscript. *

* Major • The authors provide FACS results for the cell clusters and types they defined from the scRNA-Seq data but do not provide any results on where and in which cellular contexts these cells are found in the bone and whether their spatial proximity and proteome (via staining, for example) make it likely or unlikely to see the cell-cell communication suggested in their CellPhoneDB analysis. The authors should either provide such results or adjust sentences as follows to not overstate their results: "These analyses also unveiled the complex MC-HC connectome, in particular the abundant interactions of fibroblastic SFP, AFP, CLFP, and GFP populations with HPC, and quite outstandingly, with the ILC-TSP2 and Eo/Bas clusters."*

We agree with the reviewer but, as previously commented in the General Statements and in our response to Reviewer #1, spatial localization of all populations is technically challenging in the bone and the methods we have tested fall short for the precise and reliable localization of specific bone cell populations with cellular resolution. Following his/her suggestion, we have systematically edited the text so as to not overstate any message stemming from our expression analysis.

• To address the issue of lineage commitment, the authors could offer some functional assessments between E18.5 and P1 or Adult bone BMSCs stromal cells subset that were sorted using FACS (Fig. 6).*

As commented in our response to Reviewer #2, and given the complex lineage relations in the bone, addressing this point would require extensive in vivo experimentation through transplant surgery in immunocompromised newborns or genetic analysis using novel mouse alleles, both of which we consider out of the scope of our study, conceived as a resource. Of note, in Fig. 6 we did not analyze by FC any adult population, but in the revised Fig. 6A, we now provide the expression of all markers in perinatal and adult datasets. In relation with lineage relations, we have also removed the PHATE analysis from the revised manuscript, as suggested by Reviewer #2.

Minor • The visualization of UMAP embeddings is very inconsistent across the manuscript and misleading or irritating in some cases. For example, in Figure 1b, the separation of the background grid is not clearly visible between E18 and PN1. In other figures in the same manuscript, borders around the figures solve this issue. Additionally, axes are either missing or unlabeled, whereas for UMAP embeddings, irrelevant axis tick labels and grid lines are present in most figures. It would benefit the overall flow and visualization of the manuscript if UMAP figures were more consistent.

We apologize for these inconsistencies and we thank the reviewer for pointing them out. We have now separated both panels in Fig. 1B and added borders so as to separate both UMAP plots. We have also added the missing labeling of axes throughout the manuscript so as to make all figures more consistent. We have chosen to keep both grid lines and tick labels as they help in the comparison of Harmony-integrated datasets.

• For Figure 1b specifically, it might also make sense to outline the main cell populations in both UMAPs, as in Figure 2a.*

Agreed and done.

• Average gene expression cannot take on values below 0, as that is the lower bound for expression counts. Figure 1c seems to show the colorbar dropping below 0 though. This might just be a problem of confusing color bar label placement, but it should be addressed. It should also be assured that, indeed, there has not been a mix-up and expression values are limited to >=0.*

We should have explained this better. These dotplots in Fig.1 and the heatmap in Fig. S1 use the normalized and scaled expression value (mean=0; standard deviation=1), which means that it might be negative expression values. These instances are interpreted as genes in which the expression levels are lower than the mean expression level in the dataset and facilitate the visualization of differential gene expression in the different clusters. We have now indicated this clearly in the figure legends.

• For the PHATE analysis, was there any batch correction applied to address potential batch effects between the E18 and PN1 datasets?*

We have removed from the manuscript all PHATE analysis. Still, as we use Harmony integration as a batch-correction tool, we now describe it now in detail in the Materials and Methods section.

• Figure 4a: From the text, it is clear that CellPhoneDB was used to calculate significant interactions between cell types. However, it is not clear which threshold (even if default) was used to determine what constitutes a significant interaction.*

We apologize for this omission. We have indicated both in the revised figure legend and in Materials and Methods the threshold (p-value ≤0.05; as calculated by CellPhoneDB) that was used to represent all significant interactions and shown in Fig. 4A.

• Figure 4b: It is unclear why a collection of chord representations was chosen here, as chord diagrams of this kind generally do not provide any useful additional information apart from an interaction being found to be significant (by a certain threshold) between two cell types. Lacking are generally more interesting parameters, such as the interaction score of such interactions or the expression of the involved ligands and receptors, in comparison to other cell types, where the respective interaction was not predicted to be significant. In this particular case, it is also unclear what is encoded by the width of the respective arrows. This should be made clear. Additionally, a suggestion could be to either present this information in an array of two DotPlots, one for ligands and receptors, respectively, or to encode additional information in, for example, the arrow or connector width, with the connector encoding the mean ligand expression and the arrow head encoding the mean receptor expression in the chord diagram.*

We initially chose to use chord plots as we thought it would be a visual way to represent significant interactions but, as the reviewer points out, they do not provide any additional information. In the line with the reviewer’s suggestion, we have substituted all Fig. 4B chord representations for bubble plots in which are both encoded the mean scaled expression of the ligand/receptor pair (the output of the CellPhoneDB tool) and the mean percentage of cells in the clusters expressing the corresponding molecules. We believe that this modification makes this figure more informative and visually easier to interpret.

• The authors do not mention how many genes were used as marker genes for GFP, SFP, etc. for the GO term enrichment analysis. This number (if low), the significance cut-offs, and the method used to determine DEGs could potentially have an impact on the GO enrichment results. The authors should therefore, already in the main manuscript text, mention the number of genes used for each of these cell subtypes and the method used to determine them. The text mentions cellranger, but the underlying methodology is not mentioned.*

In the revised manuscript, we have included how differential gene expression between clusters was calculated (DEGs were obtained using the FindAllMarkers() function in Seurat, using the default parameters -by default Seurat uses the Wilcoxon Rank Sum test for statistical testing) and the genes used for GO analysis (DEGs were filtered to include genes with an adjusted p-value ≤0.005; gene lists provided as new Supplementary Table 1). The resulting number of genes used for GO analysis at E18.5/PN1 was 218/280 (AFP), 455/480 (CLFP), 185/234 (GFP) and 436/305 (SFP). Retrieved GO terms were filtered by a ratio fold of enriched/expected ≥ 2 and manually curated.

Reviewer #3 (Significance (Required)):

* This study and single cell RNA-sequencing data further analyze the distinctions between the neonatal and adult stages of hematopoietic cells and bone stromal cells. This study also demonstrated the cellular heterogeneity of hematopoietic and bone stromal cells, as well as how cellular cross-talk supports osteogenic and hematopoietic cells. This sequencing data will be useful in the future to comprehend how the bone and marrow adapt to a stronger gravitational force operating on the skeletal structure, as well as to changed oxygen consumption, requirements for the development of the immune system, and an overall altered metabolism.*

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary

The authors present data on a very interesting model, mouse bone just before and after birth. In this timeframe, the organism has to adapt from a buoyant, nurtured environment to stronger gravitational forces acting upon the skeletal structure, changed oxygen uptake, changed demands to the immune system and its development, and an overall changed metabolism. The authors introduce these changes and their importance in a clear, easy-to read introduction, and this clear structure and language continue throughout the manuscript. Comparing scRNA-seq of bone E18.5 and adult stage, comparable findings by Liu Y. et al., (https://doi.org/10.1038/s41467-022-28775-x) have been previously shown. However, this manuscript showed additional postnatal day 1 (P1) data.

All computational analyses are well done, for the most part well described, and, notably, the integration of previously published data allows us to put the results of this study into context and compare them to the adult situation. Data sharing is not optimal, but it is already very good. The only downside is that most of the computational analyses are done at a very limited level of depth and merely provide initial insights and an overview of the data presented. Further comments will be given in bullet-point form, split by their impact on the overall message of the manuscript.

Major

• The authors provide FACS results for the cell clusters and types they defined from the scRNA-Seq data but do not provide any results on where and in which cellular contexts these cells are found in the bone and whether their spatial proximity and proteome (via staining, for example) make it likely or unlikely to see the cell-cell communication suggested in their CellPhoneDB analysis. The authors should either provide such results or adjust sentences as follows to not overstate their results:
  • "These analyses also unveiled the complex MC-HC12 connectome, in particular the abundant interactions of fibroblastic SFP, AFP, CLFP13, and GFP populations with HPC, and quite outstandingly, with the ILC-TSP2 and 14 Eo/Bas clusters."
• To address the issue of lineage commitment, the authors could offer some functional assessments between E18.5 and P1 or Adult bone BMSCs stromal cells subset that were sorted using FACS (Fig. 6).

Minor

• The visualization of UMAP embeddings is very inconsistent across the manuscript and misleading or irritating in some cases. For example, in Figure 1b, the separation of the background grid is not clearly visible between E18 and PN1. In other figures in the same manuscript, borders around the figures solve this issue. Additionally, axes are either missing or unlabeled, whereas for UMAP embeddings, irrelevant axis tick labels and grid lines are present in most figures. It would benefit the overall flow and visualization of the manuscript if UMAP figures were more consistent.
• For Figure 1b specifically, it might also make sense to outline the main cell populations in both UMAPs, as in Figure 2a.
• Average gene expression cannot take on values below 0, as that is the lower bound for expression counts. Figure 1c seems to show the colorbar dropping below 0 though. This might just be a problem of confusing color bar label placement, but it should be addressed. It should also be assured that, indeed, there has not been a mix-up and expression values are limited to >=0.
• For the PHATE analysis, was there any batch correction applied to address potential batch effects between the E18 and PN1 datasets?
• Figure 4a: From the text, it is clear that CellPhoneDB was used to calculate significant interactions between cell types. However, it is not clear which threshold (even if default) was used to determine what constitutes a significant interaction.
• Figure 4b: It is unclear why a collection of chord representations was chosen here, as chord diagrams of this kind generally do not provide any useful additional information apart from an interaction being found to be significant (by a certain threshold) between two cell types. Lacking are generally more interesting parameters, such as the interaction score of such interactions or the expression of the involved ligands and receptors, in comparison to other cell types, where the respective interaction was not predicted to be significant. In this particular case, it is also unclear what is encoded by the width of the respective arrows. This should be made clear. Additionally, a suggestion could be to either present this information in an array of two DotPlots, one for ligands and receptors, respectively, or to encode additional information in, for example, the arrow or connector width, with the connector encoding the mean ligand expression and the arrow head encoding the mean receptor expression in the chord diagram.
• The authors do not mention how many genes were used as marker genes for GFP, SFP, etc. for the GO term enrichment analysis. This number (if low), the significance cut-offs, and the method used to determine DEGs could potentially have an impact on the GO enrichment results. The authors should therefore, already in the main manuscript text, mention the number of genes used for each of these cell subtypes and the method used to determine them. The text mentions cellranger, but the underlying methodology is not mentioned.

Significance

This study and single cell RNA-sequencing data further analyze the distinctions between the neonatal and adult stages of hematopoietic cells and bone stromal cells. This study also demonstrated the cellular heterogeneity of hematopoietic and bone stromal cells, as well as how cellular cross-talk supports osteogenic and hematopoietic cells. This sequencing data will be useful in the future to comprehend how the bone and marrow adapt to a stronger gravitational force operating on the skeletal structure, as well as to changed oxygen consumption, requirements for the development of the immune system, and an overall altered metabolism.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

In the study by Rueda et al. the authors use single cell RNA-sequencing to investigate differences in cellular composition bones/bone marrow between late gestational stage mouse embryos and their perinatal counterparts. The authors describe specific differences in the relative abundance of putative cell types and use established bioinformatic tools to infer interactions as well as molecular mechanisms determining specific functions. The employed methods are well described and the results are presented in a very clear and understandable manner. Despite that, the findings do not provide any substantial knowledge advance but are rather confirming work of published literature while supplementing available single cell RNA-sequencing datasets of mouse bones at adult ages. As such, this work provides an interesting resource but does not report novel biology.

Major comment: The authors explore the interesting transition of embryonic to perinatal bone/bone marrow using single RNA-sequencing. This fills a gap for the field of bone and hematopoietic researchers. There is little to criticize about the presented data. However, while it provides a nice resource, the knowledge gained is incremental. As acknowledged by the authors themselves, their study lacks functional validation of any findings made or conclusions drawn from bioinformatic tools in this manuscript. They use published work to validate their findings but do not go beyond that to confirm putative new biology. Some examples are listed in the minor comments.

Minor comments:

1. Remark: 10X Chromium does not provide whole transcriptomic coverage but rather captures the most highly abundant transcripts without for example being able to distinguish alternatively spliced gene variants. Based on that, interpretation of gene expression, or the absence of a gene in the dataset, should be interpreted carefully.
2. The fact that bone marrow adipose tissue begins to accumulate after birth is well known. It is therefore not surprising that adipogenic progenitor populations start to accumulate perinatally (established by studies cited by the authors). Thus, these results only confirm the validity of the dataset. This represents an example on how the majority of findings have been presented here.
3. Given that the authors do not provide functional validation of putative new molecular interactions (by CellPhoneDB) their conclusions should be presented in a more tempered manner and acknowledged as inference rather than fact.
4. Similarly, the authors claim "...we identified Ptx4 as a novel tenogenic-specific gene...". This is too strong a conclusion as this has not been functionally validated. It should at least be tested by immuno-(co)-staining.
5. The authors identify "uncommitted clusters" as mesenchymal progenitor populations without actual showing that they are even related by lineage. This is a general pitfall in analyzing single cell RNA-sequencing data and making trajectory/pseudotime inferences. It is now well-established that the mesenchymal compartment is highly heterogeneous and composed of multiple distinct cellular lineages. Trajectory inference tools such as PHATE do not distinguish those different mesenchymal lineages. As such, the presented results cannot be considered valid unless there is proper functional validation.
6. The description of Pas being mainly associated with compact bone is neither correct nor supported by cited studies. The authors show potential additional markers to target Pas in mice, but fail to validate their point that these markers could be used in human tissue as well.
7. Figure 1c: the legend for dotsize is off scale.

Significance

Strength:

• thorough analysis of single cell RNA-sequencing datasets including integration of published work
• good writing and figure presentation
• dataset fills gap for the field as the presented ages have not been published

Limitations:

• lacking functional validation
• lack of new biology - mostly confirmation of known facts

Advance:

• knowledge gain is incremental
• good resource

Audience:

• fills a gap in the bone and hematopoietic research field as a resource

My expertise: Skeletal stem cell lineage biology, single cell RNA-sequencing of bone cell populations

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 new manuscript, Rueda and colleagues present an extensive bioinformatics analysis of single cell transcriptomic data obtained for mouse endochondral bone cell populations before and after birth. They describe gene markers of mesenchymal and hematopoietic cells pointing to differences with adult bone populations, and they use gene ontology and trajectory analyses to infer possible roles of these cells in the developing bone. The data could provide a valuable resource for further understanding endochondral bone development and the changes driving this process in peri-natal stages. However, they are also significant weaknesses.

A major weakness is that the scRNA-seq data lack validation through other techniques and functional assays. Namely, in situ data are missing to locate the various cell populations in the developing bones, especially the different types of fibroblastic cells identified by the authors. Such data would go a long way to understand the possible functions of the cell populations. Although the authors tried to complement their data with a review of the literature, most of the conclusions remain purely speculative and not sufficiently supported by scientific and statistic rigor. This makes the Results section more like a discussion than a description of the results. For instance, the authors proposed important regulatory functions for the fibroblastic clusters, but there is no data supporting this other than broad GO terms associated with genes expressed in these cells. Related to this point, the title of the manuscript does not accurately reflect the content of the study.

Other points:

1. The authors missed to report in the Results section which skeletal elements they used for their analyses and which skeletal elements were used for the adult dataset that they compared their data with. Differences in skeletal elements and in the ways whereby these samples were collected and processed could explain differences detected in the two types of datasets. Also, the sex and age of the samples for the adult dataset should be reported.
2. It is unclear whether PN1 is the day that mice are born (classically referred to as P0) or the next day.
3. It is unclear whether the cells obtained for each biological replicate were pooled for the scRNA-seq assays or were treated individually. It is thus unclear how reproducible the data are.
4. It is not clear in the gating strategy chosen for the flow cytometry as shown in Fig. 1A why the green gate containing cells expressing high levels of CD9, CD140 and CD31 has been extended in between the purple and orange gates containing CD140 and CD31 negative cells.
5. Are cells from all the sequenced samples homogenously distributed in the scRNA-seq clusters? Authors should provide this information and add statistic when they describe changes in the amount of cells per cluster between E18.5 and PN1 stages.
6. On the basis of what markers the AFP population has been called adipogenic? Authors present Ptch2 and Notch3 as markers of this cluster, but not adipogenic progenitor genes.
7. Authors claim that there is a good correlation between OsC and osteo-CAR clusters. However, OsC cells do not express Cxcl12 and other typical CAR cell markers.
8. In Figure 6 expression of PaS cell markers should be shown for both adult and perinatal populations. Additionally, have the authors tested that the sorted cells in panel C have the same progenitor properties as the PaS cells?

Significance

As indicated in the comments for the authors, the new scRNA-seq data could become a useful resource for subsequent studies, but they are at present insufficient to represent a significant scientific advancement. The main concern is that new cell populations appear to have been identified by the authors, but a number of questions were not answered such as regarding their actual location in the skeletal elements, their origins, their fates and their functions. Generating such data would require a major amount of effort and require substantial revision of the manuscript.