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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements [optional]

We thank all the reviewers for their constructive and critical comments. We provide a point-by-point response to the reviewers' comments, as detailed below. By responding to them, we believe that our revised manuscript will significantly improve so that it will be of interest for researchers in the field of cell biology, signaling pathways, physiology and nutrition.

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity):

Summary: The manuscript by Yusuke Toyoda and co-workers describes that the phosphorylation of the a-arrestin Aly3 downstream of TORC2 and GAD8 (AKT) negatively regulates endocytosis of the hexose transporter Ght5 in S.pombe under glucose limiting growth conditions.

To arrive at these conclusions, the researchers define a set of redundant c-terminal phosphorylation sites in Aly3 that are downstream by GAD8. Phosphorylation of these sites reduces Ght5 ubiquitination and endocytosis. For ubiquitination, Aly3 interacts with the ubiquitin ligases Pub1/3.

We thank the reviewer for his/her time and reporting advantages and issues of this study.

Major points:

Figure 3B: it would be interesting to compare Aly3 migration pattern (and hence potential phosphorylation) under glucose replete or limiting growth conditions. Can the authors provide direct evidence that Aly3 phosphorylation changes in response to glucose availability? Also please explain the 'smear' in lanes aly3(4th Ala), aly3(4th Ala, A584S), aly3(4th Ala, A586T).

While it is an interesting possibility that the Aly3 migration pattern changes in response to glucose concentrations in medium, we think that this is unlikely and that examining this possibility is beyond the scope of this study. Because a phospho-proteomics study reported by Dr. Paul Nurse's lab showed Tor1-dependent phosphorylation of Aly3 at S584 under high glucose (2%) conditions (Mak et al, EMBO J, 2021), the Aly3 phosphorylation (migration) pattern is likely to be constant regardless of glucose conditions. Glucose conditions affect the mRNA and protein levels of Ght5, but supposedly not its endocytosis to vacuoles (Saitoh et al, Mol Biol Cell, 2015; Toyoda et al, J Cell Sci, 2021).

As for the smear in Aly3(4th A), Aly3(4th A;A584S), Aly3(4th A; A586T), we suspect that some posttranslational modification occurs on these mutant Aly3 proteins, but the identity of the modification is unclear. We did not mention the smear signals in the original manuscript, because the presence or absence of the smear did not necessarily correlate with cell proliferation in low glucose and thus vacuolar localization of Ght5, which is the main topic of this study. In the revised manuscript, we will mention this point more clearly.

Figure 4: Ght5 localization should be analyzed + / - thiamine and in media with different glucose levels. Also, a co-localization with a vacuolar marker (FM4-64) would be nice (but not necessary). Ideally, the authors should add WB analysis of Ght5 turnover to complement the imaging data. Also, would it be possible to measure directly the effects on glucose uptake (using eg: 2-NBDG).

In this revision, we plan to observe Ght5 localization under the conditions indicated by the reviewer (+/- thiamine and high/low glucose levels) to unambiguously show that the vacuolar localization of Ght5 occurs in a manner dependent solely on expression of the mutant Aly3 protein.

We thank the reviewer for the suggestion of co-staining with FM4-64. Indeed, because we previously reported that the cytoplasmic Ght5 signals were surrounded by FM4-64 signals in the TORC2-deficient tor1Δ mutant cells (Toyoda et al, J Cell Sci, 2021), the cytoplasmic Ght5-GFP signals in Figure 4 are very likely to co-localize with vacuoles. We will modify the text to clarify this point.

As suggested, we plan to add Western blot analysis of Ght5 turnover in Aly3-expressing cells, to complement the imaging data (Figure 4) in the revised manuscript. Persistent appearance of GFP in Western blot would be a good support for vacuolar transport of Ght5-GFP.

While regulation of glucose uptake is an important issue, measurement of Ght5-dependent glucose uptake using 2-NBDG was very difficult in our hands. Another reviewer (Reviewer #2) also mentioned the difficulty of this measurement in the Referees cross-commenting section.

Figure 5: Given the localization of Ght5 shown in Figure 4, I'm surprised that it is possible in to detect full length Ght5, and its ubiquitination in the phospho-mutants of Aly3. I expected that the majority of Ght5 would be constitutively degraded, and that one would need to prevent endocytosis and/or vacuolar degradation to detect full length Ght5 and ubiquitination. Please explain the discrepancy. Also it seems that the quantification in B was performed on a single experiment.

As the aim of Figure 5 is to compare the ubiquitinated species of Ght5 among the samples expressing different species of Aly3, the loading amount of each sample was adjusted so that the abundance of immunoprecipitated Ght5 is same across them. Therefore, as the reviewer points out, before the adjustment, abundance of the full-length Ght5 might be different in these samples. In the revised manuscript, we will add explanation on this point; why the anti-GFP blot of Figure 5A has the similar intensities in those samples.

In the revised manuscript, we will add two additional replicates of the same experiment as Figure 5 in Supplementary material to show reproducibility of the result.

Figure 6: Which PPxY motif of Aly3 is used for interaction with Pub1/3 and does their interaction depend on (de)phosphorylation?

In the revised manuscript, we will discuss that "both PY motifs of Aly3 might be required for full interaction with Pub1/3," by citing the following published knowledge:

(a) Mutation of both PPxY motif of budding yeast Rod1 and Rog3 (Aly3 homologs) diminished their interaction with the ubiquitin ligase Rsp5 (Andoh et al, FEBS Lett, 2002).

(b) Mutating either one of two PPxY motifs of budding yeast Cvs7/Art1 greatly decreased interaction with WW domain, and mutating both abolished the interaction (Lin et al, Cell, 2008).

Our preliminary results indicated that Pub3 interacted with Aly3, Aly3(4th A) and phospho-mimetic Aly3(4th D), and thus suggested that the Aly3-Pub1/3 interaction does not depend on the phosphorylation status of Aly3. Consistently, budding yeast Rod1 reportedly interacts with Rsp5 regardless of its phosphorylation status (e.g. Becuwe et al, J Cell Biol, 2012). While we have partially mentioned this point in the original manuscript (L499-503), we will discuss this point more clearly in the revised manuscript.

Reviewer #1 (Significance):

The results are well presented and clear cut (with few exceptions, please see major points). They provide further evidence that metabolic cues instruct the phosphorylation of a-arrestins. Phosphorylation then negatively regulates a-arrestin function in selective endocytosis and is essential to adjust nutrient uptake across the plasma membrane to the given biological context.

We thank the reviewer for finding significance of our study. We believe that adding new results of the requested experiments and responding to the raised comments will clarify the significance of our revised manuscript.

Reviewer #2 (Evidence, reproducibility and clarity):

**Summary / background. This paper focuses on the regulation of endocytosis of the hexose transporter, Ght5, in S. pombe by nutrient limitation through the arrestin-like protein Aly3. Ght5 is induced when glucose is limiting and is required for growth and proliferation in these conditions. ght5+ encodes the only high-affinity glc transporter from fission yeast. ght5+ is induced in low glucose conditions at the transcriptional level and is translocated to the plasma membrane to allow glc import. Ght5 is targeted to the vacuole in conditions of N limitation. Mutations in the TORC2 pathway lead to the same process, thus preventing growth on low glucose medium, as shown in the gad8ts mutant, mutated for the Gad8 kinase acting downstream of TORC2. Previously, the authors demonstrated that the vacuolar delivery of Ght5 in the gad8ts mutant is suppressed by mutation of the arrestin-like protein Aly3. Arrestin-like proteins are in charge of recognising and ubiquitinating plasma membrane proteins to direct their vacuolar targeting by the endocytosis pathway. This suggested that Aly3 is hyperactive in TORC2 mutants, and accordingly, Ght5 ubiquitination was increased in gad8ts.

**Overall statement This study aims at deepening our understanding of the regulation of endocytosis by signalling pathways through arrestin-like proteins. Ght5 is a nice model to study a physiological regulation, and the authors have a great set of tools at hand. However, I think the conclusions are not always rigorous and the conclusions are sometimes far-reaching. The main problem is that much of the conclusions concern a potential phosphorylation of Aly3 which is not experimentally addressed. An additional issue is the fact that they look at Ght5 ubiquitination by co-immunoprecipitation in native conditions (or at least, it seems to me) which cannot be conclusive. Overall, I think some experiments should be performed to address (at least) these 2 points before the manuscript can be published, see detailed comments below.

We thank the reviewer for pointing both advantages and issues of our manuscript.

We admit that phosphorylation of Aly3 was not experimentally shown in our manuscript, although its phosphorylation has already been shown in phospho-proteomic studies by other groups. For this issue, we plan to add an experiment and modify the text, as explained below.

The other major issue raised by this reviewer is that detection of Ght5 ubiquitination by immunoprecipitation in a native condition cannot be conclusive. Although we noticed that many studies perform affinity purification after denaturing and precipitating proteins with TCA or acetone to detect ubiquitination of the affinity-purified protein (e.g. Lin et al, Cell, 2008), we disagree with this opinion of the reviewer #2. In a review article describing methods to study ubiquitination by immunoblotting (Emmerich and Cohen, Biochem Biophys Res Comm, 2015), affinity purification of the protein of interest in a native condition is mentioned as one major choice. Moreover, a denaturing condition was not applicable to detect ubiquitinated Ght5 because the Ght5 protein that is once denatured and precipitated with TCA cannot be re-solubilized for immune-purification and -blotting. As the reviewer points out, a pitfall of detection of ubiquitinated Ght5 in a native condition is the presence of co-immunoprecipitated proteins. In our previous study (Toyoda et al, J Cell Sci, 2021), we purified GFP-tagged Ght5 and showed that a 110 kDa band detected in an anti-Ub immunoblot was also recognized by an anti-GFP antibody, confirming that the detected 110 kDa band corresponded to an ubiquitinated species of Ght5, but not a co-immunoprecipitated protein. Similarly, in the revised manuscript, we will add a panel of high-contrast (over-exposed) anti-GFP immunoblot, in which the indicated 110 kDa band was clearly detected by an anti-GFP antibody, in Figure 5A.

We appreciate these issues raised by the reviewer #2. By responding to them, we believe that conclusions of our study will be more rigorous and undoubtful in the revised manuscript.

**Major statements and criticism.

*Fig 1. Based on the hypothesis that TORC2-mediated phosphorylation regulate Ght5 endocytosis, the authors first considered a possible phosphorylation of Ght5. They mutagenised 11 **possible** phosphorylation sites on the Ct of Ght5, but none affected the growth on low glucose in the absence of thiamine, suggesting that they don't contribute to the observed TORC2-mediated regulation. However, I disagree with the statement that "phosphorylation of Ght5 is dispensable for cell proliferation in low glucose", given that the authors do not show 1- that Ght5 is phosphorylated and 2-that this is abolished by these mutations. They should either provide data on this or tone down and say that these residues are not involved in the regulation, without implying phosphorylation which is not proven.

Although we did not experimentally test whether these 11 residues of Ght5 was phosphorylated in our hand, these residues have been shown to be phosphorylated in phospho-proteomics studies by other groups (Kettenbach et al, Mol Cell Proteomics, 2015; Swaffer et al, Cell Rep, 2018; Tay et al, Cell Rep, 2019; Halova et al, Open Biol, 2021; Mak et al, EMBO J, 2021). In the revised manuscript, we plan to be more precise by replacing this conclusion with the following statement: "11 Ser/Thr residues of Ght5, which are reportedly phosphorylated, are not essential for cell proliferation in low glucose."

In the presence of Thiamine (Supp fig 1), it seems that the ST/A mutant grows better in low glucose, and this is not explained nor commented. Since the transporter is not expressed, could the authors provide an explanation to this? If the promoter is leaky and some ght5-ST/A is expressed, it may be more stable and allow better growth than the WT, which would tend to indicate that impairing phosphorylation prevents endocytosis (which is classical for many transporters, see the body of work on CK1-mediated phosphorylation of transporters). Have the authors tried to decrease glc concentration lower than 0.14% in the absence of thiamine to see if this also true when the transporters is strongly expressed? (OPTIONAL)

Improved growth of Ght5(ST11A)-expressing cells in the presence of thiamine was mentioned in the legend of Supplementary Figure 1A. In the revised manuscript, we will mention this observation also in the main text for better description of the results.

Adding thiamine to medium does not completely shut off transcription from the nmt1 promoter but allows some transcription, as previously reported (Maundrell, J Biol Chem, 1990; Forsburg, Nuc Acid Res, 1993). In the revised manuscript, we will mention this "leakiness" of the nmt1 promoter and, by citing the suggested studies, will discuss a possibility that the ST11A mutations might prevent endocytosis of Ght5 and consequently promote cell proliferation in low glucose conditions.

We found that, in the absence of thiamine, cells expressing ght5+ and ght5(ST11A) proliferated to the comparable extent on medium containing 0.08% glucose. This result will be added to the revised manuscript.

*Fig 2. The authors then follow the hypothesis that TORC2 exerts its Ght5-dependent regulation through the phosphorylation of Aly3. They mutagenised 18 **possible** phosphorylation sites on Aly3. This led to a strong defect in growth in low-glc medium. Mutation of the possible Gad8 site (S460) did not recapitulate this phenotype, suggesting that it is not sufficient, however, mutations of 4 ST residues in a CT cluster (582-586) mimicked the full 18ST/A mutation, suggesting these are the important residues for Ght5 endocytosis.

We thank the reviewer for appreciating the results in Fig. 2. As we explain below, we plan to perform an additional experiment to show that the Aly3 C-terminus is phosphorylated. With this result, our model will gain another experimental support.

*Fig 3A. Further dissection did not allow to pinpoint this regulation to a specific residue, beyond the dispensability of the T586 residue. Fig 3B. The authors look at the effects of mutation of Aly3 on these sites at the protein level. They had to develop an antibody because HA-epitope tagging did not lead to a functional protein (Supp fig 2). Whereas I agree that the mutations causing a phenotype lead to a change in the migration pattern, I disagree with the statement that "This observation indicated that slower migrating bands were phosphorylated species of Aly3" (p.9 l.271). First, lack of phosphorylation usually causes a slower mobility on gel, which is not clear to spot here. Second, a smear appears on top of the mutated proteins (eg. 4th Ala) which is possibly caused by another modification. There are many precedents in the literature about arrestins being ubiquitinated when they are not phosphorylated (see the work on Bul1, Rod1, Csr2 in baker's yeast from various labs). My gut feeling is that lack of phosphorylation unleashes Aly3 ubiquitination leading to change in pattern. All in all, it is impossible to state about the phosphorylation of a protein without addressing its phosphorylation properly by phosphatase treatment + change in migration, or MS/MS. Thus, whereas the data looks promising, this hypothesis that Aly3 is phosphorylated at the indicated sites is not properly demonstrated.

We disagree with the reviewer's opinion that a lack of phosphorylation usually causes slower mobility on gel. There are many examples in which phosphorylation causes slower mobility on gel, including budding yeast Rod1 (Alvaro et al, Genetics, 2016), and mammalian TXNIP (Wu et al, Mol Cell, 2013). In the revised manuscript, we will cite these reports to support our interpretation that the slower migrating bands are likely phosphorylated species of Aly3 (L270-271).

Smear-like signals in Aly3(4th Ala), Aly3(4th A;A584S) and Aly3(4th A;A586T) might result from some modification, but identity of the modification is unknown. As the reviewer #2 mentioned, phosphorylation on Aly3 might negatively regulate another modification. The precedent studies revealed that budding yeast Rod1 and Rog3 arrestins tend to be ubiquitinated in snf1/AMPK-deficient cells (Becuwe et al, J Cell Biol, 2012; O'Donnell et al, Mol Cell Biol, 2015), and that Bul1 arrestin is dephosphorylated and ubiquitinated in budding yeast cells deficient in Npr1 kinase (Merhi and Andre, Mol Cell Biol, 2012). Also, budding yeast Csr2 arrestin is deubiquitinated and phosphorylated upon glucose replenishment, while non-phosphorylated Csr2 is ubiquitinated and activated by Rsp5 (Hovsepian et al, J Cell Biol, 2012). While the smear-like signals are interesting, we noticed that the smear-like signals did not necessarily correlate with cell proliferation defects in low glucose. We therefore think that clarifying the identity of the smear-like signals is beyond the scope of this study. We will discuss the smear-like signals only briefly in the revised manuscript, and would address this issue in our future work, hopefully.

While the 4 S/T residues at the C-terminus of Aly3 as well as the other 14 S/T residues have been already shown to be phosphorylated in the precedent studies (Kettenbach et al, Mol Cell Proteomics, 2015; Tay et al, Cell Rep, 2019; Halova et al, Open Biol, 2021), we will confirm that the slower migrating Aly3 is indeed phosphorylated by phosphatase treatment in the revised manuscript. This planned experiment will further strengthen our study and support our conclusion and model.

*Fig 4. The authors now look at the functional consequences of these mutations on ALy3 on Ght5 localisation. The data clearly shows that mutation of the 4 identified S/T residues (Aly3-4th A) causes aberrant localisation of the transporter to the vacuole, likely to cause the observed growth defect on low glucose. There is a nice correlation between the vacuolar localisation and growth in low-glucose for the various aly3 mutants. (A final proof could be to express this in the context of an endocytic mutant, which should restore membrane localisation and suppress the aly3-4thA phenotype - OPTIONAL). However, I still disagree with the statement that "These results indicate that phosphorylation of Aly3 at the C-terminal 582nd, 584th, and/or 585th serine residues is required for cell-surface localization of Ght5." given that phosphorylation was not properly demonstrated.

While phosphorylation of the 582nd, 584th and/or 585th serine residues of Aly3 is not experimentally demonstrated in our hands, they have been shown to be phosphorylated in phospho-proteomics studies by other groups (Kettenbach et al, Mol Cell Proteomics, 2015; Tay et al, Cell Rep, 2019; Halova et al, Open Biol, 2021; Mak et al, EMBO J, 2021). Among them, the 584th serine residue (S584) was reported to be phosphorylated in a TORC2-dependent manner (Mak et al, EMBO J, 2021), consistent with our model. To explicitly demonstrate that S584 is phosphorylated, we plan to make a strain expressing a mutant Aly3 protein in which all the possible phosphorylation sites except S584 are replaced with alanine, namely Aly3(ST17A;S584). Hopefully, we can properly show the phosphorylation of S584 by measuring the mobility of the Aly3(ST17A;S584) on gel with/without phosphatase treatment or gad8 mutation.

We thank the reviewer for suggestion of the experiment using an endocytic mutant. Previously we reported that vacuolar localization of Ght5 in gad8 mutant cells was suppressed by mutations in not only aly3 but also genes encoding ESCRT complexes (Toyoda et al, J Cell Sci, 2021). We therefore think that in cells expressing Aly3(ST18A) or Aly3(4th Ala), Ght5 is subject to endocytosis and ensuing selective transport to vacuoles via endosome-localized ESCRT complexes. We will discuss this point in the revised manuscript.

*Fig 5. Here, the authors question the role of Aly3 mutations on Ght5 ubiquitination. They immunoprecipitate Ght5 and address its ubiquitination status in various Aly3 mutants. The data is encouraging for a role in Aly3 phosphorylation (?) in the negative control of Ght5 ubiquitination. My main problem with this experiment is that it seems that Ght5 immunoprecipitations were made in non-denaturing conditions, which leads to the question of what is the anti-ubiquitin revealing here (Ght5 or a co-immunoprecipitated protein, for example Aly3 itself, or the Pub ligases, or an unknown protein). It seems that this protocol was previously used in their previous paper, but I stand by my conclusion that ubiquitination of a given protein can only be looked in denaturing conditions. The experiments should be repeated in buffers classical for the study of protein ubiquitination to be able to conclude unambiguously that we are looking at Ght5 ubiquitination itself, especially in the absence of a non-ubiquitinable form of Ght5 as a negative control. Could the authors comment on the fact that S-A or S-D mutations display the same phenotype regarding the possible Ght5 ubiquitination?

As mentioned above, immunoprecipitation of Ght5 in denaturating conditions is not feasible. Ght5 can be affinity-purified only in a non-denaturing condition. In addition, affinity purification in a native condition is considered as a major choice to examine its ubiquitination according to a literature by Emmerich and Cohen (Emmerich and Cohen, Biochem Biophys Res Comm, 2015). A drawback of native condition is, as the reviewer points out, that the affinity-purified fraction might include non-bait (non-Ght5) proteins. The 110 kDa band indicated by an arrow in Fig. 5A was confirmed to be Ght5, not a non-bait protein, as a band at the identical position was detected in the immunoblot with anti-GFP antibody. Because this band in the anti-GFP immunoblot was too faint to be visible in Fig. 5A of the original manuscript, we will add an additional panel showing the contrast-enhanced anti-GFP immunoblot in which the 110 kDa band is clearly visible.

As for the result that "S-A or S-D mutations display the same phenotype regarding the possible Ght5 ubiquitination," we are afraid that the reviewer #2 misunderstood the labels of the samples. We apologize for confusing notational system of the sample name. Full description of samples is as follows; In Aly3(4th A), all of S582, S584, S585 and T586 are replaced with A; In Aly3(4th A;A584S), S582, S585 and T586 are replaced with A, whereas S584 remains intact; In Aly3(4th A;A584D), S582, S585 and T586 are replaced with A, and S584 is replaced with phospho-mimetic D. Because cells expressing Aly3(4th A;A584S) and Aly3(4th A;A584D) exhibited similarly low levels of Ght5 ubiquitination, we speculated that phosphorylation at S584 of Aly3 negatively regulates ubiquitination of Ght5.

In the revised manuscript, we plan to add a table showing amino acid sequence of each species of Aly3 (just like Figure 3A) to avoid confusion.

*Fig 6. The authors want to document the model whereby Aly3 may interact with some of the Nedd4 ligases (Pub1/2/3) to mediate its Ght5-ubiquitination function. They actually use the Aly3-4thA mutant, it should have been better with the WT protein. But the results indicate a clear interaction with at least Pub1 and Pub3. By the way, are the Pub1/2/3 fusions functional? Nedd4 proteins are notoriously affected in their function by C-terminal tagging and are usually tagged at their N-terminus (See Dunn et al. J Cell Biol 2004).

We plan to test whether Pub1-myc is functional by comparing proliferation of the Pub1-myc-expressing strain and pub1Δ strain, as pub1Δ cells reportedly show proliferation defects at a high temperature (Tamai and Shimoda, J Cell Sci, 2002). As deletion of pub2 or pub3 reportedly exhibited no obvious defects (Tamai and Shimoda, J Cell Sci, 2002; Hayles et al, Open Biol, 2013), it is not easy to assess functionality of the myc-tagged genes.

Please note that C-terminally tagged Pub1/2/3 proteins have been widely used in studies with fission yeast. Both Pub1-HA and non-tagged Pub1 were reported to be ubiquitinated (Nefsky and Beach, EMBO J, 1996; Strachan et al, J Cell Sci, 2023). Pub1-GFP, which complemented the high temperature sensitivity of pub1Δ, localized to cell surface and cytoplasmic bodies (Tamai and Shimoda, J Cell Sci, 2002). Pub2-GFP, overexpression of which arrested cell growth just like overexpression of non-tagged Pub2, localized to cell surface, and consistently Pub2-HA was detected in membrane-enriched pellet fractions after ultracentrifugation (Tamai and Shimoda, J Cell Sci, 2002). They also reported ubiquitin conjugation of the HECT domain of Pub2 fused with myc epitope at its C-terminus. Pub3-GFP localized to cell surface (Matsuyama et al, Nat Biotech, 2006).

Regardless of functionality of the myc-tagged Pub1/2/3, we believe that results of this experiment (Figure 6) support our model, because the aim of this experiment, which is to identify the HECT-type and WW-domain containing ubiquitin ligase(s) that interact with Aly3, is irrelevant to functionality of the myc-tagged Pub proteins.

*Fig 7. The authors want to provide genetic interaction between the Pub ligases and the growth defects in low glc due to alterations in Ght5 trafficking. It is unclear how the gad8ts pub1∆ mutant was generated since it doesn't seem to grow on regular glc concentration (Supp fig 5), could the authors provide some information about this? It is also not clear whether it can be stated thatches mutant is "more sensitive" to glc depletion because of the low level of growth to begin with (even at 3%). Altogether, the data show that deletion of pub3+ is able to suppress the growth defect of the gad8ts mutant on low glc medium, suggesting it is the relevant ligase for Ght5 endocytosis. This is confirmed by microscopy observations of Ght5 localisation. However, I would again tone down the main conclusion, which I feel is far-reaching: "Combined with physical interaction data, these results strongly suggest that Aly3 recruits Pub3, but not Pub2, for ubiquitination of Ght5." Work on Rsp5 in baker's yeast has shown that Rsp5 function goes beyond cargo ubiquitination, including ubiquitination of arrestins (which is often required for their function as mentioned in the introduction) or other endocytic proteins (epsins, amphyphysin etc). I agree that the data are compatible with this model but there are other possible explanations. Anything that would block endocytosis would supposedly suppress the gad8ts phenotype.

gad8ts pub1Δ was produced at 26 {degree sign}C, a permissive temperature of the gad8ts mutant. While this is described in the Methods section of the original manuscript, we will mention this more clearly in the Results section of the revised manuscript.

We did not conclude low glucose sensitivity of gad8ts pub1Δ cells in the indicated part (L376-377). Rather, we compared proliferation of gad8ts single mutant and pub1Δ single mutant cells in low glucose, and we found that the pub1Δ single mutant exhibited the higher sensitivity. In the revised manuscript we will correct the text to clarify that we compared proliferation of two single mutants (but not gad8ts pub1Δ mutant).

We agree with the opinion that the recruited Pub3 may ubiquitinate proteins other than Ght5. In the revised manuscript, we will correct our conclusion of the Figure 7 experiment (L388-390), not to limit the possible ubiquitination target(s) to Ght5.

In a genetic screen, we found that mutations in aly3+ and genes encoding ESCRT complexes suppressed low-glucose sensitivity and vacuolar transport of Ght5 of gad8ts mutant cells (Toyoda et al, J Cell Sci, 2021). This finding appears consistent with the reviewer's opinion that blocking endocytosis would supposedly suppress the gad8ts phenotype. We will mention this point in the revised manuscript.

*Discussion Some analogy with the regulation of the Bul arrestins by TORC1/Npr1 and PP2A/Sit4 could be mentioned (Mehri et al. 2012), at the discretion of the authors. The possibility that phosphorylation may neutralise a basic patch on Aly3 Ct, possibly involved in electrostatic interactions with Ght5 is very interesting. Regarding the effect of the mutations on Aly3 localisation (p.15 l.498), did the authors tag Aly3 with GFP? There are examples where proteins tagged with HA are not functional whereas tagging with GFP does not alter their function (eg. Rod1, Laussel et al. 2022) - and here Supp Fig 2 only relates to HA-tagging. Proof of a change in Aly3 localisation upon mutation would definitely be a plus (OPTIONAL).

We thank the reviewer for the suggestion of a reference. In the revised manuscript, we will cite the indicated report in the corresponding part for an additional support of TORC1-mediated control of Aly3 (de)phosphorylation.

While examining localization of Aly3 by GFP-tagging is interesting, we do not believe that it is necessary in this study. We would like to produce Aly3-GFP and to examine its functionality and localization in our future study. We thank the reviewer's insightful suggestion.

**Minor comments.

*Introduction: - I believe the text corresponding to the work on TXNIP is incorrect (p.5 l.127). TXNIP is degraded after its phosphorylation, not "rectracted" from the surface.

In the revised manuscript, we will correct the text accordingly.

• For the sake of completion, the authors could add other references concerning the regulation of Rod1 in budding yeast such as Becuwe et al. 2012 J Cell Biol and O'Donnell et al. 2015 Mol Cell Biol, in addition to Llopis-Torregrosa et al. 2016.

In the revised manuscript, we will add the suggested references and correct the text in the corresponding part of the Introduction (L123-138).

• Other examples of the requirement for arrestin ubiquitination beyond Art1 (p.5 l.136-137) are listed in the ref cited: Kahlhofer et al. 2021.

We will cite the indicated review to navigate readers for more examples of arrestin ubiquitination (and transporter ubiquitination).

*Figures: In general, I think it would be clearer if the authors showed on the figures that the background strain in which the XXX gene is added (or its mutant forms) is a xxx∆ strain.

We will modify the figures to clearly show the genetic background of the strains used.

**Referees cross-commenting**

Cross review of Reviewer 1 - *I don't believe that the authors "define a set of redundant c-terminal phosphorylation sites in Aly3", because phosphorylation is not proven. *I thinks the points raised for Fig 3B are valid but the authors should focus on making their story conclusive before expanding to other data (except for the explanation of the smear, see my review). Also, I don't think 2NBDG actually works to measure Glc uptake. * same for Fig 6 - not sure the interaction site mapping between Aly3 and Pubs would bring much value since there are more urgent things to do to make the story solid.

As mentioned above, we will experimentally show phosphorylation of the Aly3 C-terminus in the revised manuscript. Such experiments would make our story more solid and conclusive. We truly appreciate the comments and suggestions.

We agree with the comments on difficulty of measuring glucose uptake using 2-NBDG. In fact, we tried and failed measuring Ght5-mediated glucose uptake using 2-NBDG.

---

Cros review of Reviewer 3 - we have many overlaps, so briefly : *I agree that the bibliography is incomplete (mentioned in my review) *I agree that there is no demonstration of the phospho-status of Aly3, and it is a problem *I agree that the results can be better quantified, esp. in the light of the points raised by this referee concerning the variability of expression of ST18A Other specific comments : *I agree that the statement that dephosphorylation activates alpha-arresting should be toned down - this was observed in several instances but there are examples of arrestin-mediated endocytosis which does not require their prior dephosphorylation. *I fully agree that efforts could be made regarding the classification/nomenclature of arrestins in S. pombe, this had escaped my attention

As detailed in the individual point raised by the reviewers, we will add the suggested references and accordingly correct the text in the revised manuscript.

In addition to experimentally showing Aly3 phosphorylation, we will quantify the immunoblot result.

Our statement that dephosphorylation activates alpha-arrestins might be too generalized. We will mention reports in which arrestin-mediated endocytosis does not require prior dephosphorylation (e.g. O'Donnell et al, Mol Biol Cell, 2010; Gournas et al, Mol Biol Cell, 2017; Savocco et al, PLoS Biol, 2019), and modify the text precisely.

Reviewer #2 (Significance):

*strengths and limitations This study aims at deepening our understanding of the regulation of endocytosis by signalling pathways through arrestin-like proteins in S. pombe. Ght5 is a nice model to study a physiological regulation, and the authors have a great set of tools at hand, including the discovery of Aly3 as the main arrestin for this regulation, and a signalling pathway (TORC2/Gad8) acting upstream. The main question is now to understand at the mechanistic level how TORC2 signaling impinges on the regulation of this arrestin.

Overall, the authors nicely demonstrate that C-terminal Ser/Thr residues are crucial for the function of Aly3 in Ght5 endocytosis. They propose a model whereby Aly3 phosphorylation by an unknownn kinase inhibits its function on Ght5 ubiquitination, which would favour its endocytosis. However, I think the conclusions are not always rigorous and the conclusions are sometimes far-reaching. The main problem is that much of the conclusions concern a potential phosphorylation of Aly3 which is not experimentally addressed. An additional issue is the fact that they look at Ght5 ubiquitination by co-immunoprecipitation in native conditions (or at least, it seems to me) which cannot be conclusive. Overall, I think some experiments should be performed to address (at least) these 2 points before the manuscript can be published, see detailed comments above.

*Advance

This study, if completed carefully, would provide among the first examples of mapping of phosphorylation sites on arrestins, which are usually phosphorylated at many sites and are thus difficult to study. Few studies went down to this level in this respect (see Ivshov et al. eLife 2020). There are no changes in paradigms or new conceptual insights, but this work is a nice example of the conservation of these regulatory mechanisms.

We appreciate that this study is highly evaluated by this reviewer. We understand the main problems raised by the reviewer, and as we detailed above, we plan to perform an experiment and make explanation to respond to the problems. With the raised issues answered, we believe that conclusions of the revised manuscript will be more rigorous.

Our study reveals mechanisms regulating vacuolar transport of the Ght5 hexose transporter via the TORC2 pathway in fission yeast. The serine residues at the Aly3 C-terminus (582nd, 584th and 585th serine residues), which are probably phosphorylated in a manner dependent on the TORC2 pathway, are required for sustained Ght5 localization to cell surface and cellular adaptation to low glucose. To our knowledge, there is no such study, and thus we think that this study is novel. By responding to the reviewers' comments and adding new data as explained above, the revised manuscript will be able to present novelty of our study more clearly. Comparison of our study in fission yeast to related studies in other model organisms may reveal the conservation and diversity of these regulatory mechanisms.

*Audience Should be of interest for people studying basic research in the field of cell biology, signalling pathways, transporter regulation by physiology. Reviewer background is on the regulation of transporter endocytosis by signalling pathways and arrestin-like proteins.

Reviewer #3 (Evidence, reproducibility and clarity): (Authors' response in blue)

In this manuscript, the authors work to address how phospho-regulation of a-arrestin Aly3 in S. pombe regulates the glucose transporter Ght5. The authors use a series of phospho-mutants in Aly3 and assess function of these mutants using growth assays and localization of Ght5. My main concerns with the manuscript are that 1) there is a lack of appreciation for the similar work that has been done in S. cerevisiae to define a-arrestin phospho-regulation, which is evidenced by the severe lack of referencing throughout the document, 2) the sites mutated on Aly3 are not demonstrated to change phospho-status of Aly3 and so all interpretations of these mutants need to be better contextualized and 3) almost none of the findings are quantified (imaging or immunoblots) making it difficult to assess the rigor of the outcomes. More detailed comments are provided below.

We thank the reviewer for thorough reading of the manuscript and the detailed comments. As explained below, we will respond to the points raised by the reviewer and accordingly modify the manuscript.

Minor Comments

Immunoblotting or immunostaining to define the levels and localization of phospho-mutants - In Figure 1, an immunoblot or immunostaining to define the abundance/localization of WT Ght5 vs its ST11A mutant would be appreciated. It is very difficult to know if ST11A is as functional as WT or not without an assessment of the levels and localization of the WT and mutant proteins to accompany the spot assays. Perhaps a version of Ght5 that is a phospho-mimetic would be more useful here as well since that version should not be dephosphorylated and then presumably would be internalized and not allow for growth on low glucose medium.

We plan to add fluorescence microscopy data of WT Ght5 and Ght5(ST11A) in the revised manuscript, to compare the localization and abundance of these two Ght5 species. In our preliminary observation, those of two Ght5 species seemed to be indistinguishable.

We'd like to emphasize that the primary aim of this study is to reveal mechanisms regulating Ght5 localization and consequently ensuring cell proliferation in low glucose. While analyzing a phospho-mimetic Ght5 mutant (e.g. Ght5(ST11D)) is interesting in terms of understanding of the nature of Ght5, we believe that such an analysis is out of the scope on this study. As Ght5(ST11A)-expressing cells proliferated comparably to Ght5(WT)-expressing cells and WT and ST11A Ght5 indistinguishably localize on the cell surface, phosphorylation of the ST residues of Ght5 is not likely to be the primary mechanism regulating Ght5 localization and function. We would like to assess a phospho-mimetic Ght5 mutant protein in our future studies.

For the Aly3 mutants where the abundance of Aly3 appears lower via immunoblotting (i.e., 4thA-A582S or S582A) how is the near perfect functional readout explained when the levels of the protein are much lower than WT? For the ST18A mutant, this is a particularly important point since the authors indicate on lines 194-197 that based on the functional data for ST18A, some of these ST residues are needed for phospho-regulation of Aly3. However, in Figure 3B the authors clearly show that there is very little ST18A protein in cells, and so these mutations have impacted Aly3 stability, which may or may not be linked to its phospho-status. The authors should be upfront about this finding on lines 194-197 and should not present this phospho-model as the only reason for why ST18A may not be functional. On lines 265-276 for the authors indicate that ST18A is expressed equivalently to WT Aly3, which is just not the case in Figure 3B. Perhaps quantification of replicate data would help clarify this issue. Further, if the authors wish to conclude that the upper MW bands in Figure 3B are due to phosphorylation, perhaps they should perform phosphatase treatments of their extracts to collapse these bands. However, most certainly the overall abundance of the single band for ST18A is reduced compared to the total bands of WT Aly3.

We disagree with the opinion that the levels of the mutant Aly3 are much lower than WT. For semi-quantitative measurement of the protein abundance, 2-fold dilution series of the WT Aly3 sample were loaded in the leftmost 3 lanes of Figure 3B. Although the levels of Aly3(4th A;A582S), Aly3(S582A) and Aly3(ST18A) were lower than that of WT Aly3, those are 50% or more of the WT, judging from the intensities of the serially-diluted WT samples. To clearly show that the expression of these Aly3 proteins is within comparable levels, we plan to add a column chart of the quantified expression levels and to mention abundances of the Aly3 proteins more quantitatively in the revised text. We do not think that replicate data (of Western blots as in Figure 3B) helps clarify this issue, because nmt1 promoter-driven gene transcription is induced with a small variation (Forsburg, Nuc Acid Res, 1993). We will cite this report and mention this point in the revised text.

We are afraid that this reviewer seems to consider that Aly3(ST18A) is not functional, but it is not a case and we do not intend to claim so. While deletion of aly3 did not interfere with cell proliferation in low glucose (see vector controls in Figures 2B, 2C and 3A, -Thiamine), expression of the ST18A mutant clearly hinders cell proliferation in low glucose, indicating that the ST18A performs dominant negative function to inhibit cell proliferation. That is, even though the expression level and/or stability of the ST18A is reduced, it is still sufficiently abundant to perform the dominant negative function. We propose the phospho-model not due to dysfunctionality of ST18A, but its dominant negative functionality. The 18 S/T residues of Aly3, which are shown to be phosphorylated in precedent phospho-proteomics studies, seem to be required to down-regulate Aly3's function to inhibit cell proliferation in low glucose. We apologize for this confusion, and we will modify the text and figures to clarify these points in the revised manuscripts.

To obtain an experimental support for our description that the slower migrating bands in Figure 3B are due to phosphorylation, we plan to perform a phosphatase treatment experiment as suggested.

Figure 2A - how do the phosphorylation sites identified in Aly3 compare to those identified in Rod1 from S. cerevisiae? See PMID 26920760 or SGD for more information. I am confused as to why the Aly3 protein has an arrowhead at the C-terminus. What does this denote?

We will mention reported phosphorylation sites of Aly3 and budding yeast Rod1/Art4 in the revised manuscript, by referring to the indicated report and database. It should be noted that similarity between amino acid sequences of Aly3 and S. cerevisiae Rod1 is not so high and limited in Arrestin-N and -C domains. The C-terminal half of Aly3, in which most of the potential phosphorylation sites are found, is not similar to Rod1. Thus, these sites are unlikely to be conserved between them.

An arrowhead indicates the direction of transcription (from N to C-terminus). We will describe it explicitly in the revised figure legend.

Figure 2 - The WT and Aly3-ST18A are expressed in S. pombe from a non-endogenous locus under the control of the Nmt1 promoter. However, are these mutants present in cells that contain WT copies of Aly3 at other genomic loci? If so, this would surely muddy the interpretations of these data as a- and b-arrestins are capable of multimerizing and the effect of multimerization on their activities can vary.

As mentioned in L188, an aly3 deletion mutant strain (aly3Δ) was used as a host, and thus all strains harboring an nmt1-driven aly3 gene lack the endogenous aly3 gene. We will add an illustration clearly showing that the host strain lacks the endogenous aly3+ gene and modify the legend of Figure 2.

Functional readouts for Aly3 using Ght5 localization - The reduced surface levels of Ght5 does correspond to the spot assay growth in low glucose for the various Aly3 mutants used. However, it would be useful if these assays incorporated an endocytosis inhibitor to help prevent the activities of these Aly3 plasmids to see if the transporter is retained at the PM. At the end of these mutational analyses, the authors conclude that phosphorylation of Aly3 at any of 3 sites is required for Ght5 trafficking to the vacuole in low glucose, however no experiment is done to demonstrate that these sites are phosphorylated residues. A phosphatase assay would be useful to help demonstrate that the modifications in 3B really are phosphorylation and a quantification of the phosphorylated bands in these WBs would also be useful to solidify the statement made on lines 306-309.

We thank the reviewer for suggestion of the experiment using an endocytosis inhibitor. Previously we reported that vacuolar localization of Ght5 in gad8ts mutant cells was suppressed by mutations in not only aly3 but also genes encoding ESCRT complexes (Toyoda et al, J Cell Sci, 2021). We therefore think that, in cells expressing Aly3(ST18A) or Aly3(4th Ala), Ght5 is subject to endocytosis and subsequent selective transport to vacuoles via ESCRT complexes. We will mention these previous findings in the revised manuscript.

As mentioned in responses to the comments above and other reviewer's, we will perform a phosphatase treatment experiment and its quantification in the revised manuscript. Here, we'd like to emphasize that these 3 sites have been shown to be phosphorylated in phospho-proteomic studies by other researchers (Kettenbach et al, Mol Cell Proteomics, 2015; Tay et al, Cell Rep, 2019; Halova et al, Open Biol, 2021; Mak et al, EMBO J, 2021), although we do not show it directly in this study.

Phosphorylation assessments - in general, it would be good to not only build the non-phosphorylatable versions of Aly3 but also the phospho-mimetic forms.

We produced a phospho-mimetic mutant Aly3 (i.e. Aly3(4th A;A584D)), and showed the result in Figure 5A; cells expressing Aly3(4th A;A584D) exhibited a low ubiquitination of Ght5, similarly to Aly3(WT)- and Aly3(4th A;A584S)-expressing cells. According to our experiences, replacing S/T with D/E does not necessarily mimic phosphorylation. Thus, we do not believe that systematic production of phospho-mimetic Aly3 mutants would help achieve the aim of this study.

Pub1, 2, and 3 - It would be helpful if the authors indicated what genes Pubs 1-3 correspond to in S. cerevisiae, where Rsp5 is the predominant Ub ligase interacting with a-arrestins. Is there no ortholog of Rsp5 in S. pombe?

Pub1, Pub2 and Pub3 are regarded as orthologs of budding yeast Rsp5, according to the fission yeast database PomBase. We will perform a homology search for these E3 proteins, and based on the result, we will add a description in the revised manuscript.

Pub-Aly3 interactions - could the authors please comment on the reason why so very little Aly3 is copurified with Pub1 or Pub2? Can any clear conclusion be drawn about pub2 given how very little Pub2 is present in the IPs? Based on my understanding of these data I do not think that this can be cleanly interpreted. What is is the identity of the ~50kDa MW band in Figure 6 in the upper MYC detection panel?

We do not have an accurate answer for the result that a small amount of Aly3 is copurified with Pub1 or Pub3. The Pub1/3-Aly3 interaction may be weak or transient. We will discuss this point in the revised manuscript.

Regarding whether Aly3 interacts with Pub2, we agree with the reviewer. As described in the Results (L360-362), we could not conclude anything about Aly3-Pub2 interaction by this immunoprecipitation experiment alone. On the other hand, the genetic interaction experiment (Figure 7A) suggests that pub2+ is not involved in defects caused by the gad8ts mutation (while pub3+ and aly3+ are). By this experiment, we think that Pub2 is not a partner of Aly3.

In the revised manuscript, we will describe that Pub2 is not a partner of Aly3 in a paragraph describing the Figure 7A experiment.

Because the 50 kDa band found in the IP fraction of all the samples appears even in "beads only" (Figure 6), those are supposedly derived from mouse IgG dissociated from the beads used for immunoprecipitation. We will mention this in the legend of Figure 6.

Phosphorylation and ubiquitination of a-arrestins - The paragraph from lines 123-138 is very superficial in addressing what is known about phosphorylation and ubiquitination of a-arrestins. The way this section is written, it feels misleading to the reader as it omits many of the details for regulation that would help place the current study in context. The discussion of Rod1 phosphorylation by AMPK for example, which is directly relevant to this study, is underdeveloped. I would recommend splitting this into two paragraphs and providing a more in depth, and accurate, view of the literature on this topic, with a focus on the regulation that is relevant for the ortholog of Aly3 in S. cerevisiae. For example, Rod1 phosphorylation by AMPK is greatly expanded upon in the following papers (PMID 22249293 and 25547292) and AMPK regulation of C-tail phosphorylation of a-arrestins is defined further in PMID 26920760. These references are each particularly important to compare with the current findings presented in this manuscript. Torc2 regulation ofa-arrestins is also reviewed in PMID 36149412 and references therein should be considered.

Because the primary aim of this study is to reveal mechanisms regulating Ght5 localization in fission yeast, but not to dissect modification and regulation of α-arrestins, we decided not to get into the details of phosphorylation and ubiquitination of α-arrestins. Furthermore, although budding yeast Rod1 and Rog3 are found to be downstream of the TORC2-Ypk1 signaling in the context of internalization of the Ste2 pheromone receptor, it is not clear whether TORC2-Ypk1 signaling also regulate α-arrestin-mediated internalization of hexose transporters in budding yeast. For these reasons, we focused on limited literatures essential for interpretation of the results and omitted many references describing the details of α-arrestin regulation. However, as this reviewer commented, we realize that our decision makes the discussion superficial and misleading to the reader. We sincerely apologize for this inconvenience.

In the revised manuscript, we will reorganize the paragraphs in the discussion and include the suggested references. Regarding budding yeast Rod1, we will cite the study reporting Ypk1-mediated phosphorylation on Rod1 in mating pheromone response via regulation of Ste2 endocytosis (Alvaro et al, Genetics, 2016). We will also mention other reports (Becuwe et al, J Cell Biol, 2012; O'Donnell et al, Mol Cell Biol, 2015) about AMPK-dependent phosphorylation of Rod1 in the corresponding part (e.g. L129-130). In addition, we will mention that Aly2, Rod1 and Rog3 α-arrestins were found downstream of the TORC2-Ypk1 signaling (Muir et al, eLife, 2014; Thorner, Biochem J, 2022).

As a further detailed example, there is far more work done on ubiquitination of a-arrestins in S. cerevisiae than the single citation provided by the authors on line 137. The way this section is written it feels misleading. Considerable effort has been spent on defining how mono- and poly-ubiquitination regulate a-arrestins and the authors should consider the data provided in the following citations and revise the two sentences they provide in this introduction to better reflect the breadth of our understanding rather than simply indicate that the 'mechanisms that regulate functions of a-arrestisn are not fully understood'. (PMIDs 23824189; 22249293; 17028178; 28298493)

Ubiquitination of α-arrestin itself is not the topic of this study, and physiological consequences of ubiquitination of Aly3 remain unknown. Because of these reasons, we did not describe the details of ubiquitination of α-arrestins in the original manuscript. However, we never intend to mislead the reader, and thus to avoid it, we will revise the indicated sentences and cite the suggested literatures (O'Donnell et al, J Biol Chem, 2013; Becuwe et al, J Cell Biol, 2012; Kee et al, J Biol Chem, 2006; Ho et al, Mol Biol Cell, 2017) in the revised manuscript.

Context of the findings and lack of citations - The referencing in this manuscript is very poor as many of the key papers that report analogous findings in the budding yeast Saccharomyces cerevisiae are not cited. This oversight in citing the appropriate literature must be remedied before this manuscript can be considered further for publication. Examples of these omissions occur at the following places:

We will modify the text and carefully cite more literatures describing analogous finding in budding yeast and other organisms in the revised manuscript. We appreciate the insightful suggestions by this reviewer. It should be noted, however, that it is not evident whether budding yeast Rod1 and Rog3 are orthologous to fission yeast Aly3. Although Rod1 and Aly3 share overlapping roles, amino acid sequence similarity of them is not high and limited only in domains which are generally conserved among α-arrestin-family proteins.

Line 90 - The Puca and Brou citations is one example of this but the first examples come from Daniela Rotin's work looking at Rsp5 interactions in budding yeast, which is where the association between HECT-domain Ub ligases and a-arrestins is also documented by Scott Emr and Hugh Pelham's labs. Here are some PMID numbers to improve the citations of this section (PMID 17551511; 18976803; 19912579) and each of these references long predates the Puca and Brou publication.

In the revised manuscript, we will improve the citations by including the suggested studies (Gupta et al, Mol Syst Biol, 2007; Lin et al, Cell, 2008; Nikko and Pelham, Traffic, 2009).

Lines 123-126 - Phosphorylation can also increase vacuole-dependent degradation of alpha-arrestins as demonstrated in PMID 35454122. The interaction with 14-3-3 proteins that is driven by phosphorylation of a-arrestins was first demonstrated by the Leon group in PMID 22249293). Lines 129-132 - Here again the Leon reference that helps demonstrate the 14-3-3 inhibition of Rod1 is lacking (PMID 22249293).

We will cite the suggested studies in description of these topics (Bowman et al, Biomolecules, 2022; Becuwe et al, J Cell Biol, 2012).

Lines 130-132 - Please include references for the statement that dephosphorylation activates a-arrestin activity. There are no citations on this statement and there are many to choose from and I would urge the authors to cite the primary literature on these points.

We will cite studies for the statement "Conversely, dephosphorylation is thought to activate α-arrestins and to promote selective endocytosis of transporter proteins" (L130-132).

These are just a few examples from the Introduction, but the Discussion is similarly wrought with issues in referencing and framing the experimental results within the context of the larger field, including what is known about Rod1/Rog3 regulation in S. cerevisiae. For example, the Llopis-Torregrosa et al reference and statement on lines 508-510 is incorrect. There are other phosphorylation sites defined in the C-terminus of Rod1, as described in Alvaro et al. PMID: 26920760.

We will carefully correct Discussion by citing the suggested references (e.g. Alvaro et al, Genetics, 2016) and framing the obtained results within the context of the larger field.

Of note, a combination of α-arrestin, upstream kinase(s) and distinct phosphorylation sites appears to determine the target transporter (Kahlhofer et al, Biol Cell, 2021; Thorner, Biochem J, 2022), and it has not been explicitly proved that TORC2-Ypk1 signaling also regulate α-arrestin-mediated internalization of hexose transporters in budding yeast. For these reasons, we stated "S. cerevisiae Rod1 and Rog3 are phosphorylated solely by Snf1p/AMPK" in the context of internalization of hexose transporters. We will also discuss this point in the revised manuscript.

Minor Comments Clarification needed - Lines 107-121 - The relationship between the S. pombe arrestins and those in other organisms is somewhat unclear. Frist, all the arrestins in humans and S. cerevisiae can be sorted into the alpha, beta and Vps26 classes. However, the authors indicate that the S. pombe genome has 11 arrestin-like proteins but only 4 of these are a-arrestins. What classes do the other 7 arrestins belong to? It would be appreciated if this point was clarified.

To our knowledge, fission yeast arrestins are not well classified yet. We will perform a phylogenetic tree analysis to classify them, and modify the description of the indicated part accordingly. We will also cite our previous report (Toyoda et al, J Cell Sci, 2021), in which the overall protein structure and domains of 11 fission yeast arrestin-like proteins were reported.

Next, for the 4 a-arrestins identified in S. pombe the authors indicate that Aly3 is the homolog of Rod1/Art4 and Rog3/Art7 from S. cerevisiae. What is the relationship of Rod1 in S. pombe to Rod1 in S. cerevisiae? Are these also homologs? You can see how the nomenclature is confusing and, given the functional overlap of S. cerevisiae Rod1/Rog3 proteins it is important to know if Aly3 is the only version of these a-arrestins or if there is an additional counterpart in S. pombe. This point becomes somewhat more confusing when on lines 134-136 the authors talk about Arn1/Any1 as an arrestin related protein in S. pombe yet this protein was not included on the list of a-arrestins in the preceding section. What class of arrestin is this protein?

According to PomBase, both Aly3 and Rod1 are assigned as the orthologue of budding yeast Rod1 and Rog3. However, as mentioned in responses above, it is unclear whether Aly3 is really orthologous to budding yeast Rod1/Rod3. In the revised manuscript, we will perform a homology search for these 4 proteins, and add information on how much these arrestins share homology.

Arn1/Any1 is regarded as a β-arrestin (Nakase et al, J Cell Sci, 2013). We will also mention this in the revised manuscript.

Alpha-arrestin homology - On lines 127-129 the authors indicate that TXNIP is the mammalian homolog of Aly3. To my knowledge, there are no evolutionary analyses that can draw these lines of homology between the a-arrestins in humans and those in yeasts. It would be appreciated if the authors could cite the work that leads to this conclusion or revise the sentence to more accurately reflect what is known on this topic. It certainly appears that, given their functional overlap in regulating glucose transporters, Txnip and Rod1/Rog3 in humans and S. cerevisiae are functionally connected. I urge the authors to use more caution when describing this protein family.

Among human α-arrestins, ARRDC2 (22%) but not TXNIP (20%) has the highest amino acid identity to Aly3 (Toyoda et al, J Cell Sci, 2021). However, as TXNIP has been reported to regulate endocytosis of hexose transporters, GLUT1 and 4 (Wu et al, Mol Cell, 2013; Waldhart et al, Cell Rep, 2017), we think that TXNIP and Aly3 share physiological roles. We will revise the sentence (L127-129) more accurately.

Text editing - The text could use editing as there are awkward and grammatically incorrect sentences in several places. Here are a few examples to help the authors:

Please note that the original manuscript is edited by a professional editor, who is a native English (American) speaker and has edited thousands of research papers, before initial submission. We will ask an editor to check the revised draft again before submission.

Lines 57-60 - the protein is not expressed over the entire cell surface, but is localized to the entire cell surface.

We will correct this wording.

Lines 80-83 - this sentence is very confusing

We will correct this part by changing the phrase "Unlike TORC1," into a clause.

Line 86 - Is there more than one gene encoding Aly3 in S. pombe?

No, there is only one gene encoding Aly3. We will correct this part so as to avoid being misunderstood.

Line 88, 109, - these sentences need to start with a capitol so either capitalize the A in arrestin or write out Alpha with a capitol A.

We will correct the sentence as suggested.

Lines 145-148 - unclear as written

We will clarify the meaning of the sentence by changing the voice.

Line 224 - why are these amino acids being referred to as hydroxylated? Perhaps hydroxyl-containing amino acids or 18 amino acids with hydroxyl side chains would be better choices?

We will correct the word as suggested.

Line 300 - very confusing sentence structure

We will correct this part by simplifying the structure of the sentence.

And elsewhere....

We will carefully check the revised text before submission.

Reviewer #3 (Significance):

The authors provide some information as to the residues needed in the Aly3 C-tail for Ght5 trafficking in S. Pombe. These results are not places in the context of similar phosphor-regulatory work done for a-arrestins in S. cerevisiae, and this is needed for appreciation of the significance of the study.

Overall, it appears that the model put forth is very similar to the one already proposed in S. cerevisiae where phosphorylation impedes a-arrestin-mediated trafficking of glucose transporters. It is interesting to see this similarity hold in S. Pombe, but it does not dramatically alter our appreciation of a-arrestin biology.

The significance of the findings are somewhat underscored by the fact that very little quantification of data are presented, making the rigor of the work difficult to assess.

We thank the reviewer for careful reading and evaluation of our study. As the reviewer states, the results are not placed in the context of similar phospho-regulatory works done for α-arrestins in S. cerevisiae. This may partly come from the fact that it remains unclear whether internalization of hexose transporters is regulated by TORC2-dependent phosphorylation in S. cerevisiae. We believe that our study is novel and significant for this reason. By performing the additional experiments/quantification and revising the text as suggested by the reviewers, the manuscript will be further strengthened, and we will be able to clearly conclude that TORC2-dependent phosphorylation of Aly3 regulates localization of the Ght5 hexose transporter and cellular responses to glucose shortage stress.

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 work to address how phospho-regulation of a-arrestin Aly3 in S. pombe regulates the glucose transporter Ght5. The authors use a series of phospho-mutants in Aly3 and assess function of these mutants using growth assays and localization of Ght5. My main concerns with the manuscript are that 1) there is a lack of appreciation for the similar work that has been done in S. cerevisiae to define a-arrestin phospho-regulation, which is evidenced by the severe lack of referencing throughout the document, 2) the sites mutated on Aly3 are not demonstrated to change phospho-status of Aly3 and so all interpretations of these mutants need to be better contextualized and 3) almost none of the findings are quantified (imaging or immunoblots) making it difficult to assess the rigor of the outcomes. More detailed comments are provided below.

Minor Comments

Immunoblotting or immunostaining to define the levels and localization of phospho-mutants - In Figure 1, an immunoblot or immunostaining to define the abundance/localization of WT Ght5 vs its ST11A mutant would be appreciated. It is very difficult to know if ST11A is as functional as WT or not without an assessment of the levels and localization of the WT and mutant proteins to accompany the spot assays. Perhaps a version of Ght5 that is a phospho-mimetic would be more useful here as well since that version should not be dephosphorylated and then presumably would be internalized and not allow for growth on low glucose medium.

For the Aly3 mutants where the abundance of Aly3 appears lower via immunoblotting (i.e., 4thA-A582S or S582A) how is the near perfect functional readout explained when the levels of the protein are much lower than WT? For the ST18A mutant, this is a particularly important point since the authors indicate on lines 194-197 that based on the functional data for ST18A, some of these ST residues are needed for phospho-regulation of Aly3. However, in Figure 3B the authors clearly show that there is very little ST18A protein in cells, and so these mutations have impacted Aly3 stability, which may or may not be linked to its phospho-status. The authors should be upfront about this finding on lines 194-197 and should not present this phospho-model as the only reason for why ST18A may not be functional. On lines 265-276 for the authors indicate that ST18A is expressed equivalently to WT Aly3, which is just not the case in Figure 3B. Perhaps quantification of replicate data would help clarify this issue. Further, if the authors wish to conclude that the upper MW bands in Figure 3B are due to phosphorylation, perhaps they should perform phosphatase treatments of their extracts to collapse these bands. However, most certainly the overall abundance of the single band for ST18A is reduced compared to the total bands of WT Aly3.

Figure 2A - how do the phosphorylation sites identified in Aly3 compare to those identified in Rod1 from S. cerevisiae? See PMID 26920760 or SGD for more information. I am confused as to why the Aly3 protein has an arrowhead at the C-terminus. What does this denote?

Figure 2 - The WT and Aly3-ST18A are expressed in S. pombe from a non-endogenous locus under the control of the Nmt1 promoter. However, are these mutants present in cells that contain WT copies of Aly3 at other genomic loci? If so, this would surely muddy the interpretations of these data as - and -arrestins are capable of multimerizing and the effect of multimerization on their activities can vary.

Functional readouts for Aly3 using Ght5 localization - The reduced surface levels of Ght5 does correspond to the spot assay growth in low glucose for the various Aly3 mutants used. However, it would be useful if these assays incorporated an endocytosis inhibitor to help prevent the activities of these Aly3 plasmids to see if the transporter is retained at the PM. At the end of these mutational analyses, the authors conclude that phosphorylation of Aly3 at any of 3 sites is required for Ght5 trafficking to the vacuole in low glucose, however no experiment is done to demonstrate that these sites are phosphorylated residues. A phosphatase assay would be useful to help demonstrate that the modifications in 3B really are phosphorylation and a quantification of the phosphorylated bands in these WBs would also be useful to solidify the statement made on lines 306-309.

Phosphorylation assessments - in general, it would be good to not only build the non-phosphorylatable versions of Aly3 but also the phospho-mimetic forms.

Pub1, 2, and 3 - It would be helpful if the authors indicated what genes Pubs 1-3 correspond to in S. cerevisiae, where Rsp5 is the predominant Ub ligase interacting with -arrestins. Is there no ortholog of Rsp5 in S. pombe?

Pub-Aly3 interactions - could the authors please comment on the reason why so very little Aly3 is copurified with Pub1 or Pub2? Can any clear conclusion be drawn about pub2 given how very little Pub2 is present in the IPs? Based on my understanding of these data I do not think that this can be cleanly interpreted. What is is the identity of the ~50kDa MW band in Figure 6 in the upper MYC detection panel?

Phosphorylation and ubiquitination of -arrestins - The paragraph from lines 123-138 is very superficial in addressing what is known about phosphorylation and ubiquitination of a-arrestins. The way this section is written, it feels misleading to the reader as it omits many of the details for regulation that would help place the current study in context. The discussion of Rod1 phosphorylation by AMPK for example, which is directly relevant to this study, is underdeveloped. I would recommend splitting this into two paragraphs and providing a more in depth, and accurate, view of the literature on this topic, with a focus on the regulation that is relevant for the ortholog of Aly3 in S. cerevisiae. For example, Rod1 phosphorylation by AMPK is greatly expanded upon in the following papers (PMID 22249293 and 25547292) and AMPK regulation of C-tail phosphorylation of -arrestins is defined further in PMID 26920760. These references are each particularly important to compare with the current findings presented in this manuscript. Torc2 regulation of-arrestins is also reviewed in PMID 36149412 and references therein should be considered. As a further detailed example, there is far more work done on ubiquitination of -arrestins in S. cerevisiae than the single citation provided by the authors on line 137. The way this section is written it feels misleading. Considerable effort has been spent on defining how mono- and poly-ubiquitination regulate -arrestins and the authors should consider the data provided in the following citations and revise the two sentences they provide in this introduction to better reflect the breadth of our understanding rather than simply indicate that the 'mechanisms that regulate functions of -arrestisn are not fully understood'. (PMIDs 23824189; 22249293; 17028178; 28298493)

Context of the findings and lack of citations - The referencing in this manuscript is very poor as many of the key papers that report analogous findings in the budding yeast Saccharomyces cerevisiae are not cited. This oversight in citing the appropriate literature must be remedied before this manuscript can be considered further for publication. Examples of these omissions occur at the following places:

Line 90 - The Puca and Brou citations is one example of this but the first examples come from Daniela Rotin's work looking at Rsp5 interactions in budding yeast, which is where the association between HECT-domain Ub ligases and -arrestins is also documented by Scott Emr and Hugh Pelham's labs. Here are some PMID numbers to improve the citations of this section (PMID 17551511; 18976803; 19912579) and each of these references long predates the Puca and Brou publication.

Lines 123-126 - Phosphorylation can also increase vacuole-dependent degradation of alpha-arrestins as demonstrated in PMID 35454122. The interaction with 14-3-3 proteins that is driven by phosphorylation of -arrestins was first demonstrated by the Leon group in PMID 22249293).

Lines 129-132 - Here again the Leon reference that helps demonstrate the 14-3-3 inhibition of Rod1 is lacking (PMID 22249293).

Lines 130-132 - Please include references for the statement that dephosphorylation activates -arrestin activity. There are no citations on this statement and there are many to choose from and I would urge the authors to cite the primary literature on these points.

These are just a few examples from the Introduction, but the Discussion is similarly wrought with issues in referencing and framing the experimental results within the context of the larger field, including what is known about Rod1/Rog3 regulation in S. cerevisiae. For example, the Llopis-Torregrosa et al reference and statement on lines 508-510 is incorrect. There are other phosphorylation sites defined in the C-terminus of Rod1, as described in Alvaro et al. PMID: 26920760.

Minor Comments

Clarification needed - Lines 107-121 - The relationship between the S. pombe arrestins and those in other organisms is somewhat unclear. Frist, all the arrestins in humans and S. cerevisiae can be sorted into the alpha, beta and Vps26 classes. However, the authors indicate that the S. pombe genome has 11 arrestin-like proteins but only 4 of these are -arrestins. What classes do the other 7 arrestins belong to? It would be appreciated if this point was clarified. Next, for the 4 -arrestins identified in S. pombe the authors indicate that Aly3 is the homolog of Rod1/Art4 and Rog3/Art7 from S. cerevisiae. What is the relationship of Rod1 in S. pombe to Rod1 in S. cerevisiae? Are these also homologs? You can see how the nomenclature is confusing and, given the functional overlap of S. cerevisiae Rod1/Rog3 proteins it is important to know if Aly3 is the only version of these -arrestins or if there is an additional counterpart in S. pombe.

This point becomes somewhat more confusing when on lines 134-136 the authors talk about Arn1/Any1 as an arrestin related protein in S. pombe yet this protein was not included on the list of -arrestins in the preceding section. What class of arrestin is this protein?

Alpha-arrestin homology - On lines 127-129 the authors indicate that TXNIP is the mammalian homolog of Aly3. To my knowledge, there are no evolutionary analyses that can draw these lines of homology between the -arrestins in humans and those in yeasts. It would be appreciated if the authors could cite the work that leads to this conclusion or revise the sentence to more accurately reflect what is known on this topic. It certainly appears that, given their functional overlap in regulating glucose transporters, Txnip and Rod1/Rog3 in humans and S. cerevisiae are functionally connected. I urge the authors to use more caution when describing this protein family.

Text editing - The text could use editing as there are awkward and grammatically incorrect sentences in several places. Here are a few examples to help the authors:

Lines 57-60 - the protein is not expressed over the entire cell surface, but is localized to the entire cell surface.

Lines 80-83 - this sentence is very confusing

Line 86 - Is there more than one gene encoding Aly3 in S. pombe?

Line 88, 109, - these sentences need to start with a capitol so either capitalize the A in arrestin or write out Alpha with a capitol A.

Lines 145-148 - unclear as written

Line 224 - why are these amino acids being referred to as hydroxylated? Perhaps hydroxyl-containing amino acids or 18 amino acids with hydroxyl side chains would be better choices?

Line 300 - very confusing sentence structure

And elsewhere....

Significance

The authors provide some information as to the residues needed in the Aly3 C-tail for Ght5 trafficking in S. Pombe. These results are not places in the context of similar phosphor-regulatory work done for a-arrestins in S. cerevisiae, and this is needed for appreciation of the significance of the study.

Overall, it appears that the model put forth is very similar to the one already proposed in S. cerevisiae where phosphorylation impedes a-arrestin-mediated trafficking of glucose transporters. It is interesting to see this similarity hold in S. Pombe, but it does not dramatically alter our appreciation of a-arrestin biology.

The significance of the findings are somewhat underscored by the fact that very little quantification of data are presented, making the rigor of the work difficult to assess.

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/background.

This paper focuses on the regulation of endocytosis of the hexose transporter, Ght5, in S. pombe by nutrient limitation through the arrestin-like protein Aly3. Ght5 is induced when glucose is limiting and is required for growth and proliferation in these conditions. ght5+ encodes the only high-affinity glc transporter from fission yeast. ght5+ is induced in low glucose conditions at the transcriptional level and is translocated to the plasma membrane to allow glc import. Ght5 is targeted to the vacuole in conditions of N limitation. Mutations in the TORC2 pathway lead to the same process, thus preventing growth on low glucose medium, as shown in the gad8ts mutant, mutated for the Gad8 kinase acting downstream of TORC2. Previously, the authors demonstrated that the vacuolar delivery of Ght5 in the gad8ts mutant is suppressed by mutation of the arrestin-like protein Aly3. Arrestin-like proteins are in charge of recognising and ubiquitinating plasma membrane proteins to direct their vacuolar targeting by the endocytosis pathway. This suggested that Aly3 is hyperactive in TORC2 mutants, and accordingly, Ght5 ubiquitination was increased in gad8ts.

Overall statement

This study aims at deepening our understanding of the regulation of endocytosis by signalling pathways through arrestin-like proteins. Ght5 is a nice model to study a physiological regulation, and the authors have a great set of tools at hand. However, I think the conclusions are not always rigorous and the conclusions are sometimes far-reaching. The main problem is that much of the conclusions concern a potential phosphorylation of Aly3 which is not experimentally addressed. An additional issue is the fact that they look at Ght5 ubiquitination by co-immunoprecipitation in native conditions (or at least, it seems to me) which cannot be conclusive. Overall, I think some experiments should be performed to address (at least) these 2 points before the manuscript can be published, see detailed comments below.

Major statements and criticism.

• Fig 1. Based on the hypothesis that TORC2-mediated phosphorylation regulate Ght5 endocytosis, the authors first considered a possible phosphorylation of Ght5. They mutagenised 11 possible phosphorylation sites on the Ct of Ght5, but none affected the growth on low glucose in the absence of thiamine, suggesting that they don't contribute to the observed TORC2-mediated regulation. However, I disagree with the statement that "phosphorylation of Ght5 is dispensable for cell proliferation in low glucose", given that the authors do not show 1- that Ght5 is phosphorylated and 2-that this is abolished by these mutations. They should either provide data on this or tone down and say that these residues are not involved in the regulation, without implying phosphorylation which is not proven. In the presence of Thiamine (Supp fig 1), it seems that the ST/A mutant grows better in low glucose, and this is not explained nor commented. Since the transporter is not expressed, could the authors provide an explanation to this? If the promoter is leaky and some ght5-ST/A is expressed, it may be more stable and allow better growth than the WT, which would tend to indicate that impairing phosphorylation prevents endocytosis (which is classical for many transporters, see the body of work on CK1-mediated phosphorylation of transporters). Have the authors tried to decrease glc concentration lower than 0.14% in the absence of thiamine to see if this also true when the transporters is strongly expressed? (OPTIONAL)
• Fig 2. The authors then follow the hypothesis that TORC2 exerts its Ght5-dependent regulation through the phosphorylation of Aly3. They mutagenised 18 possible phosphorylation sites on Aly3. This led to a strong defect in growth in low-glc medium. Mutation of the possible Gad8 site (S460) did not recapitulate this phenotype, suggesting that it is not sufficient, however, mutations of 4 ST residues in a CT cluster (582-586) mimicked the full 18ST/A mutation, suggesting these are the important residues for Ght5 endocytosis.
• Fig 3A. Further dissection did not allow to pinpoint this regulation to a specific residue, beyond the dispensability of the T586 residue. Fig 3B. The authors look at the effects of mutation of Aly3 on these sites at the protein level. They had to develop an antibody because HA-epitope tagging did not lead to a functional protein (Supp fig 2). Whereas I agree that the mutations causing a phenotype lead to a change in the migration pattern, I disagree with the statement that "This observation indicated that slower migrating bands were phosphorylated species of Aly3" (p.9 l.271). First, lack of phosphorylation usually causes a slower mobility on gel, which is not clear to spot here. Second, a smear appears on top of the mutated proteins (eg. 4th Ala) which is possibly caused by another modification. There are many precedents in the literature about arrestins being ubiquitinated when they are not phosphorylated (see the work on Bul1, Rod1, Csr2 in baker's yeast from various labs). My gut feeling is that lack of phosphorylation unleashes Aly3 ubiquitination leading to change in pattern. All in all, it is impossible to state about the phosphorylation of a protein without addressing its phosphorylation properly by phosphatase treatment + change in migration, or MS/MS. Thus, whereas the data looks promising, this hypothesis that Aly3 is phosphorylated at the indicated sites is not properly demonstrated.
• Fig 4. The authors now look at the functional consequences of these mutations on ALy3 on Ght5 localisation. The data clearly shows that mutation of the 4 identified S/T residues (Aly3-4th A) causes aberrant localisation of the transporter to the vacuole, likely to cause the observed growth defect on low glucose. There is a nice correlation between the vacuolar localisation and growth in low-glucose for the various aly3 mutants. (A final proof could be to express this in the context of an endocytic mutant, which should restore membrane localisation and suppress the aly3-4thA phenotype - OPTIONAL). However, I still disagree with the statement that "These results indicate that phosphorylation of Aly3 at the C-terminal 582nd, 584th, and/or 585th serine residues is required for cell-surface localization of Ght5." given that phosphorylation was not properly demonstrated.
• Fig 5. Here, the authors question the role of Aly3 mutations on Ght5 ubiquitination. They immunoprecipitate Ght5 and address its ubiquitination status in various Aly3 mutants. The data is encouraging for a role in Aly3 phosphorylation (?) in the negative control of Ght5 ubiquitination. My main problem with this experiment is that it seems that Ght5 immunoprecipitations were made in non-denaturing conditions, which leads to the question of what is the anti-ubiquitin revealing here (Ght5 or a co-immunoprecipitated protein, for example Aly3 itself, or the Pub ligases, or an unknown protein). It seems that this protocol was previously used in their previous paper, but I stand by my conclusion that ubiquitination of a given protein can only be looked in denaturing conditions. The experiments should be repeated in buffers classical for the study of protein ubiquitination to be able to conclude unambiguously that we are looking at Ght5 ubiquitination itself, especially in the absence of a non-ubiquitinable form of Ght5 as a negative control. Could the authors comment on the fact that S-A or S-D mutations display the same phenotype regarding the possible Ght5 ubiquitination?
• Fig 6. The authors want to document the model whereby Aly3 may interact with some of the Nedd4 ligases (Pub1/2/3) to mediate its Ght5-ubiquitination function. They actually use the Aly3-4thA mutant, it should have been better with the WT protein. But the results indicate a clear interaction with at least Pub1 and Pub3. By the way, are the Pub1/2/3 fusions functional? Nedd4 proteins are notoriously affected in their function by C-terminal tagging and are usually tagged at their N-terminus (See Dunn et al. J Cell Biol 2004).
• Fig 7. The authors want to provide genetic interaction between the Pub ligases and the growth defects in low glc due to alterations in Ght5 trafficking. It is unclear how the gad8ts pub1∆ mutant was generated since it doesn't seem to grow on regular glc concentration (Supp fig 5), could the authors provide some information about this? It is also not clear whether it can be stated thatches mutant is "more sensitive" to glc depletion because of the low level of growth to begin with (even at 3%). Altogether, the data show that deletion of pub3+ is able to suppress the growth defect of the gad8ts mutant on low glc medium, suggesting it is the relevant ligase for Ght5 endocytosis. This is confirmed by microscopy observations of Ght5 localisation. However, I would again tone down the main conclusion, which I feel is far-reaching: "Combined with physical interaction data, these results strongly suggest that Aly3 recruits Pub3, but not Pub2, for ubiquitination of Ght5." Work on Rsp5 in baker's yeast has shown that Rsp5 function goes beyond cargo ubiquitination, including ubiquitination of arrestins (which is often required for their function as mentioned in the introduction) or other endocytic proteins (epsins, amphyphysin etc). I agree that the data are compatible with this model but there are other possible explanations. Anything that would block endocytosis would supposedly suppress the gad8ts phenotype.

Discussion

Some analogy with the regulation of the Bul arrestins by TORC1/Npr1 and PP2A/Sit4 could be mentioned (Mehri et al. 2012), at the discretion of the authors. The possibility that phosphorylation may neutralise a basic patch on Aly3 Ct, possibly involved in electrostatic interactions with Ght5 is very interesting. Regarding the effect of the mutations on Aly3 localisation (p.15 l.498), did the authors tag Aly3 with GFP? There are examples where proteins tagged with HA are not functional whereas tagging with GFP does not alter their function (eg. Rod1, Laussel et al. 2022) - and here Supp Fig 2 only relates to HA-tagging. Proof of a change in Aly3 localisation upon mutation would definitely be a plus (OPTIONAL).

Minor comments.

Introduction:

• I believe the text corresponding to the work on TXNIP is incorrect (p.5 l.127). TXNIP is degraded after its phosphorylation, not "rectracted" from the surface.
• For the sake of completion, the authors could add other references concerning the regulation of Rod1 in budding yeast such as Becuwe et al. 2012 J Cell Biol and O'Donnell et al. 2015 Mol Cell Biol, in addition to Llopis-Torregrosa et al. 2016.
• Other examples of the requirement for arrestin ubiquitination beyond Art1 (p.5 l.136-137) are listed in the ref cited: Kahlhofer et al. 2021.

Figures: In general, I think it would be clearer if the authors showed on the figures that the background strain in which the XXX gene is added (or its mutant forms) is a xxx∆ strain.

Referees cross-commenting

Cross review of Reviewer 1

• I don't believe that the authors "define a set of redundant c-terminal phosphorylation sites in Aly3", because phosphorylation is not proven.
• I thinks the points raised for Fig 3B are valid but the authors should focus on making their story conclusive before expanding to other data (except for the explanation of the smear, see my review). Also, I don't think 2NBDG actually works to measure Glc uptake.
• same for Fig 6 - not sure the interaction site mapping between Aly3 and Pubs would bring much value since there are more urgent things to do to make the story solid.

Cros review of Reviewer 3 - we have many overlaps, so briefly :

• I agree that the bibliography is incomplete (mentioned in my review)
• I agree that there is no demonstration of the phospho-status of Aly3, and it is a problem
• I agree that the results can be better quantified, esp. in the light of the points raised by this referee concerning the variability of expression of ST18A

Other specific comments :

• I agree that the statement that dephosphorylation activates alpha-arresting should be toned down - this was observed in several instances but there are examples of arrestin-mediated endocytosis which does not require their prior dephosphorylation.
• I fully agree that efforts could be made regarding the classification/nomenclature of arrestins in S. pombe, this had escaped my attention

Significance

strengths and limitations

This study aims at deepening our understanding of the regulation of endocytosis by signalling pathways through arrestin-like proteins in S. pombe. Ght5 is a nice model to study a physiological regulation, and the authors have a great set of tools at hand, including the discovery of Aly3 as the main arrestin for this regulation, and a signalling pathway (TORC2/Gad8) acting upstream. The main question is now to understand at the mechanistic level how TORC2 signaling impinges on the regulation of this arrestin.

Overall, the authors nicely demonstrate that C-terminal Ser/Thr residues are crucial for the function of Aly3 in Ght5 endocytosis. They propose a model whereby Aly3 phosphorylation by an unknownn kinase inhibits its function on Ght5 ubiquitination, which would favour its endocytosis. However, I think the conclusions are not always rigorous and the conclusions are sometimes far-reaching. The main problem is that much of the conclusions concern a potential phosphorylation of Aly3 which is not experimentally addressed. An additional issue is the fact that they look at Ght5 ubiquitination by co-immunoprecipitation in native conditions (or at least, it seems to me) which cannot be conclusive. Overall, I think some experiments should be performed to address (at least) these 2 points before the manuscript can be published, see detailed comments above.

Advance

This study, if completed carefully, would provide among the first examples of mapping of phosphorylation sites on arrestins, which are usually phosphorylated at many sites and are thus difficult to study. Few studies went down to this level in this respect (see Ivshov et al. eLife 2020). There are no changes in paradigms or new conceptual insights, but this work is a nice example of the conservation of these regulatory mechanisms.

Audience

Should be of interest for people studying basic research in the field of cell biology, signalling pathways, transporter regulation by physiology. Reviewer background is on the regulation of transporter endocytosis by signalling pathways and arrestin-like proteins.

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 Yusuke Toyoda and co-workers describes that the phosphorylation of the a-arrestin Aly3 downstream of TORC2 and GAD8 (AKT) negatively regulates endocytosis of the hexose transporter Ght5 in S.pombe under glucose limiting growth conditions.

To arrive at these conclusions, the researchers define a set of redundant c-terminal phosphorylation sites in Aly3 that are downstream by GAD8. Phosphorylation of these sites reduces Ght5 ubiquitination and endocytosis. For ubiquitination, Aly3 interacts with the ubiquitin ligases Pub1/3.

Major points:

Figure 3B: it would be interesting to compare Aly3 migration pattern (and hence potential phosphorylation) under glucose replete or limiting growth conditions. Can the authors provide direct evidence that Aly3 phosphorylation changes in response to glucose availability? Also please explain the 'smear' in lanes aly3(4th Ala), aly3(4th Ala, A584S), aly3(4th Ala, A586T).

Figure 4: Ght5 localization should be analyzed + / - thiamine and in media with different glucose levels. Also, a co-localization with a vacuolar marker (FM4-64) would be nice (but not necessary). Ideally, the authors should add WB analysis of Ght5 turnover to complement the imaging data. Also, would it be possible to measure directly the effects on glucose uptake (using eg: 2-NBDG).

Figure 5: Given the localization of Ght5 shown in Figure 4, I'm surprised that it is possible in to detect full length Ght5, and its ubiquitination in the phospho-mutants of Aly3. I expected that the majority of Ght5 would be constitutively degraded, and that one would need to prevent endocytosis and/or vacuolar degradation to detect full length Ght5 and ubiquitination. Please explain the discrepancy. Also it seems that the quantification in B was performed on a single experiment.

Figure 6: Which PPxY motif of Aly3 is used for interaction with Pub1/3 and does their interaction depend on (de)phosphorylation?

Significance

The results are well presented and clear cut (with few exceptions, please see major points). They provide further evidence that metabolic cues instruct the phosphorylation of a-arrestins. Phosphorylation then negatively regulates a-arrestin function in selective endocytosis and is essential to adjust nutrient uptake across the plasma membrane to the given biological context.

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

i) "Enhancers dependent on TPR during senescence are enriched for binding sites of inflammatory transcription factors". *Proximity to genes does not confirm an enhancer role for that gene, although Tasdemir et al., 2016 suggested this. At that time, HI-C and Hi-CHiP techniques were not well-established. Nowadays, without combining HI-C and H3K27ac ChIP, Hi-ChIP alone cannot definitively identify actual enhancer regions. If we repeatedly use the Tasdemir et al., 2016 map, we risk incorrect mapping of enhancers of SASP. The authors should either use other public Hi-C databases to map the enhancer of SASP or temper their conclusions about enhancers. Otherwise, this could set a precedent for the SASP enhancer region that might not be entirely accurate. *

The enhancer mapping for SASP is outdated, as advancements in Hi-C have significantly developed this area. Therefore, the claimed enhancers of SASP may not be accurate.

__Response: __We agree with the reviewer that enhancers are not easy to define, or to pair with their target gene(s). Indeed, we would argue that even combined HI-C and H3K27ac does not define enhancers or enhancer-gene pairs and that the gold-standard evidence for an enhancer is genetics – does its deletion/mutation abrogate gene activation. We would also point out that we did not actually use the Tasdemir data to call enhancers. In response to the reviewer’s comment, we will temper our terminology and now refer to our inter-and intra-genic ATAC-seq peaks only as “putative enhancers”.

ii) “Many of these include putative enhancers located close to key SASP genes, such as IL1B and IL8 (Figure 1D).” I have the same concern as mentioned above (i). However, I am interested in knowing the other key SASP genes where DNA is accessible near the genes. A supplementary table listing key SASP genes along with their distances to the TSS and affected by TPR knock-down would be helpful.

__Response: __We thank the reviewer for this suggestion. We will provide tables listing the TPR dependent, senescent specific ATAC-seq peaks that are close to genes associated with the ‘positive regulation of inflammatory response’, ‘cytokine activity’ and ‘cytokine receptor binding’ gene ontology terms which were significant in the GREAT analysis, and which includes many SASP genes. We will also provide distances of these regions from the associated genes.

iii) "As we previously reported, knockdown of TPR (siTPR) in RAS cells blocks SAHF formation, but it also results in reduced nuclear localisation (decreased nucleocytoplasmic ratio) of NF-κB, consistent with decreased NF-κB activation (Figure 2A and B, Figure S2A)." TPR is required for CCF, SASP, and SAHF. The relationship between CCF and SASP is well established, but the relationship between SAHF and CCF/SASP remains elusive. Both SAHF and CCF are enriched with heterochromatin markers, suggesting that CCF might originate from SAHF. However, this has not been confirmed. Do the authors think that SAHF is a prerequisite for CCF in the OIS model, or is it an independent event?

Response: __We agree with the reviewer that CCFs likely originate from SAHF. Whilst we cannot definitively prove thisin our ER-Ras OIS model, in the revised manuscript we intend to further investigate the relationship between SAHF and CCF by knocking down HMGA1 during RAS-induced senescence. Like TPR, HMGA1 depletion is known to lead to loss of SAHF (Narita et al., Cell, 2006) but, unlike TPR, HMGA1 is a chromatin protein enriched on heterochromatin itself. We will assess whether loss of HMGA1 also abrogates CCF formation.__

iv) The authors suggested that "it is plausible that the decrease in CCFs produced during the early phases of OIS upon TPR knockdown may be caused by an increase in the stability of the nuclear periphery due to the heterochromatin that remains there when SAHF are not formed." I do not completely agree with this explanation because CCF starts forming at day 3-4 but culminates at later time points. According to Figure 5A, only 5-6% of cells are positive for CCFs on day 5. What happens on day 8? By day 8, the percentage of CCF-positive cells could be 20-25%, or the number of CCFs per cell might be 0.2-0.3. If TPR is not required for CCF formation at this stage, then linking CCF to SASP at day 8 becomes critical. This suggests that another mechanism might be driving SASP expression and that TPR could be regulating downstream signaling of CCF. It is possible that changes in nuclear pore density affect the localization of cGAS from the nucleus to the cytoplasm.

Response: __In our hands and using this IMR90 ER-RAS system, CCF formation decreases later in senescence (d8 - only 2% of cells) hence our focus on early timepoints after oncogenic RAS activation. At later timepoints, cGAS activation is also mediated by retrotransposons (de Cecco et al., Nature, 2019; Liu et al., Cell, 2023), as well as leakage of mitochondrial DNA (Victorelli et al., Nature, 2023; Chen et al., Nat. Comms, 2024), and so it is difficult to disentangle the net contribution of these three inputs.__

v) Additionally, the authors did not address what happens in the later stages of CCF formation in the absence of TPR. If TPR is not required for CCF formation at later stages, it fails to explain the downstream processes at these time points adequately. This suggests that TPR may also have another mechanism of SASP regulation independent of CCF formation.

__Response: __In our cellular system CCFs precede the SASP - CCFs are already present at day 3 but SASP factors are not secreted until day 5. However, CCFs are not necessarily required for maintenance of the SASP. Once initiated the SASP is maintained by cytokine feedback loops.

…………

Reviewer #2:

1. The claim that TPR knockdown does not affect NFkappaB nuclear translocation indeed stands, but it would be nice if the authors also compared data across conditions in Fig. 2F, i.e. siCTRL+Ras CM versus siTPR+Ras CM in RAS cells and provided a p-value as it seems to me that there is some dampening of translocation intensity, which is clearly not the case for STOP cells. The authors focus on this for d3 and d5, but it seems to be also the case for later time points.

__Response: __As basal NF-κB translocation is lower in RAS cells on TPR knockdown, we would expect a dampening in NF-κB translocation between siCTRL+RAS CM and siTPR+Ras CM regardless of whether there is a transportation defect. Consistent with this, the p-value for this comparison is significant, but we did not show it because it is not important in considering whether NF-κB nuclear translocation is impeded by TPR knockdown, which is the focus here. We will add a table with median nuclear:cytoplasmic NF-κB ratios and 95% confidence intervals to make the changes in basal level (treatment with STOP CM) clearer.

Also, a comment based on literature or from the authors previous work on TPR, on the extent to which the structural integrity of the nuclear basket is at all affected upon TPR depletion would be helpful for data interpretation.

__Response: __In the revised manuscript we will refer to the literature showing that TPR is the final component added to the nuclear pore and that its absence does not affect localisation of NUP153 to the nuclear basket (Hase and Cordes., Mol. Biol. Cell 2003; Aksenova et al., Nat Comms, 2020).

Magnification of representative cells per each condition in Fig. 2E would be welcome.

__Response: __We will provide a revised figure 2E with the magnifications as requested.

Regarding the data in Figs 3 and S3: I am a bit confused about how the obviously decreased NFkappaB nuclear signal (e.g., in Fig. 3D) does not translate into a skewed N/C ratio (e.g., in Fig. 3C)? The western blots indicate that overall NFkappaB levels remain essentially unchanged? Am I missing something?

Response: __As stated in the Methods section, we used a 50-pixel expansion of the detected nuclear area as our cytoplasmic area in the analysis (see image below). This was because we found detecting and segmenting the whole cytoplasmic area in the NF-κB channel to be unreliable. At day 3 and 5, the decrease in NF-κB nuclear signal in RAS cells on TPR knockdown was accompanied by a decrease in signal in the portion of the cytoplasm closest to the nucleus. This led to no change in the nuclear:cytoplasmic ratio. We believe the redistribution of NF-κB closer to the nucleus in the RAS siCTRL sample indicates early activation and will make this clearer in the revised text. We will also quantify the NF-κB western blots (see point 5), to help clarification of this issue.____ __

Also, along these lines, d8 western blots seem to portray an overall drop in NFkappaB levels. Is this indeed so? Can the authors maybe quantify their blots' replicates and provide a box plot and statistical testing?

Response: __We will provide quantification for the NF-κB western blots, though box plots would not be appropriate as we only have two replicates.__

Regarding the ATAC-seq data from d3, I think it could be mined a bit more. For example, compare to d8 (which the authors have apparently done, but don't present in detail) and discuss which are these early regions that also become accessible by d3 and what kind of genes and motifs are associated with them. Moreover, the focus in Fig. S3E is on ATAC sites shared with d8; how about d3-specific ones? How many of these are there (if any) and how might they be affected?

__Response: __As shown in Table S2, TPR knockdown did not cause any changes in chromatin accessibility at day 3, so there are no day 3 specific TPR dependent peaks. We will edit the text to make this clearer. We will carry out motif analysis and GREAT analysis on the day 3 peaks that become accessible in RAS cells but are not accessible in STOP (RAS-specific peaks).

I trust that the authors quantified their STING blots for the conclusions they present, but since it is difficult to assess these confidently by eye, again, some quantification plots would be welcome in Figs 4C,D and S4D,E.

__Response: __We will provide quantification for the STING western blots.

As controls for Fig. 5, it would be interesting to see if active histone readouts also mark CCFs in this system.

__Response: __Ivanov et al., J. Cell Biol., 2013 showed the absence of H3K9 acetylation from chromatin in CCFs. Further exploration of the types of chromatin/sequences in CCFs is outside the scope of our current manuscript.

*The POM121 channel in Fig. 5C appears to have some small signal foci in the cytoplasm; could these be small CCFs? More generally, the authors focus on these large blobs that only appear in

__Response: __The small signal foci the reviewer is highlighting are background from the POM121 antibody staining rather than CCFs – they do not show DAPI staining, and similar foci are evident in non-senescent cells where CCFs are generally not present. Our unpublished data (see response to Reviewer 1, point iv) from day 8 cells shows that only ~2% of senescent cells are positive for CCF regardless of TPR knockdown, which is a similar number to that observed in non-senescent cells at earlier timepoints. Thus, in our hands CCF formation occurs earlier, triggering the SASP, rather than at day 8 when the SASP is already established and reinforced through positive feedback cytokine signalling.

I wonder if there is a simple experiment the authors could do to test if this mechanism is only linked to senescence, specifically oncogene-induced senescence? I don't think this is needed to support the conclusions drawn here, but it could significantly broaden the scope of their discovery of, for example, this was true in other senescence models or during proinflammatory activation in general?

__Response: __These are interesting suggestions, but setting up, characterising and quantifying other senescence models will take a substantial amount of time that would be outside the scope of our current manuscript.

………….

Reviewer #3

1. The study uses a single cell strain IMR90 undergoing a single form of senescence, induced by activated Ras. To show the generalizability of the finding, the authors are advised to inhibit TPR in other forms of senescence in addition to IMR90. For example, IR or etoposide induces greater amount of CCF than in OIS of IMR90. BJ, MEFs, and ARPE-19 senescence also show prominent CCF.

__Response: __These are interesting suggestions, but as we responded to reviewer 2, setting up, characterising and quantifying other senescence models will take a substantial amount of time that would be outside the scope of our current manuscript.

To convincing show the CCF pathway is involved, the authors need to measure the activity of cGAS-STING pathway. Including cGAMP ELISA will be informative.

__Response: __We thank the reviewer for this suggestion, and we will try to include this assay in our revised manuscript.

The authors used conditioned media to show that TPR KD does not directly affect NFkB nuclear translocation. While this is helpful, conditions other than senescence will be more direct. For example, TNFa treatment or poly I:C transfection induces efficient NFkB nuclear translocation in IMR90 cells.

__Response: __This experiment (Fig. 2EF) was designed to simply show that knocking down TPR does not impair the ability of activated NFkB to enter the nucleus, it is not about senescence per se. Indeed, this is why we included the addition of SASP (RAS) conditioned media to non-senescence STOP cells in Fig. 2. We do not think investigating other methods of activating NFkB would add more to the question of whether TPR loss abrogates NFkB nuclear import.

Fig. 4C and Fig. S4D are identical.

Response: Though these STING immunoblots look similar; in fact they are not identical. Below we attach the raw original image in which both biological replicates (Fig 4C and S4D) for Day 3 were run on the same gel as proof of this claim.

Figure legend for Fig. S4F is mislabeled.

__Response: __We will correct this.

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

Bartlett et al reported that knockdown of TPR inhibits CCF and NFkB during oncogene-induced senescence in IMR90 cells. The manuscript is well-written, and the results are clear and convincing. I have the following suggestions.

1. The study uses a single cell strain IMR90 undergoing a single form of senescence, induced by activated Ras. To show the generalizability of the finding, the authors are advised to inhibit TPR in other forms of senescence in addition to IMR90. For example, IR or etoposide induces greater amount of CCF than in OIS of IMR90. BJ, MEFs, and ARPE-19 senescence also show prominent CCF.
2. To convincing show the CCF pathway is involved, the authors need to measure the activity of cGAS-STING pathway. Including cGAMP ELISA will be informative.
3. The authors used conditioned media to show that TPR KD does not directly affect NFkB nuclear translocation. While this is helpful, conditions other than senescence will be more direct. For example, TNFa treatment or poly I:C transfection induces efficient NFkB nuclear translocation in IMR90 cells.

Minor:

1. Fig. 4C and Fig. S4D are identical.
2. Figure legend for Fig. S4F is mislabeled.

Significance

This study builds on the group's prior publication that knockdown of TPR inhibits SAHF and SASP. The current study finds that the underlying mechanism is via CCF-STING-NFkB pathway. Overall, the study broadens our understanding of CCF and SASP in senescence, and will be of general interest to the senescence 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

The manuscript by Bartlett et al. revisits the role of the nuclear pore component, TPR, in OIS to uncover its contribution to NFkappaB activation magnitude via the control of chromatin fragment release from the senescent nucleus. The flow of the manuscript is very good and the conclusions are supported by clear experimental evidence. This is an overall mature manuscript, and I offer below some comments that might help the authors streamline their message even more:

• The claim that TPR knockdown does not affect NFkappaB nuclear translocation indeed stands, but it would be nice if the authors also compared data across conditions in Fig. 2F, i.e. siCTRL+Ras CM versus siTPR+Ras CM in RAS cells and provided a p-value as it seems to me that there is some dampening of translocation intensity, which is clearly not the case for STOP cells. The authors focus on this for d3 and d5, but it seems to be also the case for later time points.
• Also, a comment based on literature or from the authors previous work on TPR, on the extent to which the structural integrity of the nuclear basket is at all affected upon TPR depletion would be helpful for data interpretation.
• Magnification of representative cells per each condition in Fig. 2E would be welcome.
• Regarding the data in Figs 3 and S3: I am a bit confused about how the obviously decreased NFkappaB nuclear signal (e.g., in Fig. 3D) does not translate into a skewed N/C ratio (e.g., in Fig. 3C)? The western blots indicate that overall NFkappaB levels remain essentially unchanged? Am I missing something?
• Also, along these lines, d8 western blots seem to portray an overall drop in NFkappaB levels. Is this indeed so? Can the authors maybe quantify their blots' replicates and provide a box plot and statistical testing?
• Regarding the ATAC-seq data from d3, I think it could be mined a bit more. For example, compare to d8 (which the authors have apparently done, but don't present in detail) and discuss which are these early regions that also become accessible by d3 and what kind of genes and motifs are associated with them. Moreover, the focus in Fig. S3E is on ATAC sites shared with d8; how about d3-specific ones? How many of these are there (if any) and how might they be affected?
• I trust that the authors quantified their STING blots for the conclusions they present, but since it is difficult to assess these confidently by eye, again, some quantification plots would be welcome in Figs 4C,D and S4D,E.
• As controls for Fig. 5, it would be interesting to see if active histone readouts also mark CCFs in this system.
• The POM121 channel in Fig. 5C appears to have some small signal foci in the cytoplasm; could these be small CCFs? More generally, the authors focus on these large blobs that only appear in <6% of cells in d3 and d5. Does this increase by d8? What is the effect of TPR knockdown on CCF numbers at that later time point?
• I wonder if there is a simple experiment the authors could do to test if this mechanism is only linked to senescence, specifically oncogene-induced senescence? I don't think this is needed to support the conclusions drawn here, but it could significantly broaden the scope of their discovery of, for example, this was true in other senescence models or during proinflammatory activation in general?

I typically disclose my identity to the authors: A. Papantonis

Significance

This is a very clearly written and well-controlled study that addresses a gap in knowledge from the previous work on TPR in senescence. It also brings about a perhaps unexpected effect of a nuclear pore component on NFkappaB signaling that might not necessarily be senescence-specific. As such, I think that the study would be of interest to both the senescence community and to researchers studying inflammatory responses, especially those driven by TNFalpha or IL1A/B.

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

DNA damage triggers senescence, inducing chromatin reorganization and SASP activation. The authors previously demonstrated that the TPR nucleoprotein at nuclear pores is crucial for both SAHF formation and SASP activation during senescence. Here they also showed that TPR is required for the formation of cytoplasmic chromatin fragments (CCF), which activate cGAS-STING-TBK1-NF-kB signaling to express SASP. While the mechanistic regulation of CCF formation by TPR remains unclear, their study provides compelling evidence of downstream processes involving CCF. This study offers new insights into CCF formation, suggesting a promising direction for further research. I endorse the manuscript; however, there are several concerns that need addressing before acceptance.

i) "Enhancers dependent on TPR during senescence are enriched for binding sites of inflammatory transcription factors".

Proximity to genes does not confirm an enhancer role for that gene, although Tasdemir et al., 2016 suggested this. At that time, HI-C and Hi-CHiP techniques were not well-established. Nowadays, without combining HI-C and H3K27ac ChIP, Hi-ChIP alone cannot definitively identify actual enhancer regions. If we repeatedly use the Tasdemir et al., 2016 map, we risk incorrect mapping of enhancers of SASP. The authors should either use other public Hi-C databases to map the enhancer of SASP or temper their conclusions about enhancers. Otherwise, this could set a precedent for the SASP enhancer region that might not be entirely accurate.

ii) Many of these include putative enhancers located close to key SASP genes, such as IL1B and IL8 (Figure 1D).

I have the same concern as mentioned earlier about enhancers. However, I am interested in knowing the other key SASP genes where DNA is accessible near the genes. A supplementary table listing key SASP genes along with their distances to the TSS and affected by TPR knock-down would be helpful.

iii) "As we previously reported, knockdown of TPR (siTPR) in RAS cells blocks SAHF formation, but it also results in reduced nuclear localisation (decreased nucleocytoplasmic ratio) of NF-κB, consistent with decreased NF-κB activation (Figure 2A and B, Figure S2A)." TPR is required for CCF, SASP, and SAHF. The relationship between CCF and SASP is well established, but the relationship between SAHF and CCF/SASP remains elusive. Both SAHF and CCF are enriched with heterochromatin markers, suggesting that CCF might originate from SAHF. However, this has not been confirmed. Do the authors think that SAHF is a prerequisite for CCF in the OIS model, or is it an independent event?

iv) The authors suggested that "it is plausible that the decrease in CCFs produced during the early phases of OIS upon TPR knockdown may be caused by an increase in the stability of the nuclear periphery due to the heterochromatin that remains there when SAHF are not formed." I do not completely agree with this explanation because CCF starts forming at day 3-4 but culminates at later time points. According to Figure 5A, only 5-6% of cells are positive for CCFs on day 5. What happens on day 8? By day 8, the percentage of CCF-positive cells could be 20-25%, or the number of CCFs per cell might be 0.2-0.3. If TPR is not required for CCF formation at this stage, then linking CCF to SASP at day 8 becomes critical. This suggests that another mechanism might be driving SASP expression and that TPR could be regulating downstream signaling of CCF. It is possible that changes in nuclear pore density affect the localization of cGAS from the nucleus to the cytoplasm.

Significance

The authors previously demonstrated that the TPR nucleoprotein at nuclear pores is crucial for both SAHF formation and SASP activation during senescence. Here they also showed that TPR is required for the formation of cytoplasmic chromatin fragments (CCF), which activate cGAS-STING-TBK1-NF-kB signaling to express SASP. While the mechanistic regulation of CCF formation by TPR remains unclear, their study provides compelling evidence of downstream processes involving CCF. This study offers new insights into CCF formation, suggesting a promising direction for further research.

However, there are some limitations to this study. The enhancer mapping for SASP is outdated, as advancements in Hi-C have significantly developed this area. Therefore, the claimed enhancers of SASP may not be accurate. Additionally, the authors did not address what happens in the later stages of CCF formation in the absence of TPR. If TPR is not required for CCF formation at later stages, it fails to explain the downstream processes at these time points adequately. This suggests that TPR may also have another mechanism of SASP regulation independent of CCF formation.

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 *

1. The authors conclude that RFP-Ac expression is restricted to emerging SOPs and surroundings cells at 18h APF, indicating that Ac is activated later than Sc. Can the authors provide images for RFP-Ac expression at 10h and 16h APF similar to GFP-Sc as shown in their figures. Do the SOPs that contain high levels of both Ac and Sc (as some SOPs have Sc expression but not Ac) undergo fate divergence and SB faster than the SOPs containing higher levels of only Sc?

We are now showing the expression pattern of GFP-SC and RFP-Ac/GFP-Ac in fixed samples stained also for E-cad at 13h, 16h and 18h APF (Fig 1I-K' and Fig S1E-G'). Ac and Sc were found to be activated around the same time. However, Ac appeared to accumulate at lower levels than Sc prior to SOP selection in the central domain of the ADHN (Fig 1J-K'). We also confirmed that Ac was more strongly expressed in SOPs. Additionally, SOPs appeared to accumulate both Ac and Sc, i.e. SOPs with high levels of GFP-Sc also showed a strong RFP-Ac signal (Fig S1H-H'). Finally, since RFP-Ac was not detectable in living pupae, possibly due to the rapid turn-over of Ac and the slow maturation of RFP, we could not study more precisely the relative dynamics of Ac and Sc. For the same reason, we could not address whether the rate of fate divergence (measured using GFP-Sc) varied with the level of Ac.

2. It would be interesting to see the spatial and temporal dynamics of Ac and Sc in Notch mutants or even Notch dynamics in Sc and Ac mutants to better understand the progression fate divergence and its effect on lateral inhibition in real time.

Following the reviewer's suggestion, we examined the expression pattern of NRE-deGFP, a Notch activity reporter, in ac sc double mutant pupae at 16h and 24h APF (Fig S3A-D). This showed that the initial pattern of NRE-deGFP at 16h APF (signal detected in posterior ADHN cells as well as in the ADHN) did not depend on Ac and Sc. By contrast, the second phase of NRE-deGFP expression (in cells of the proneural ADHN domain, around emerging SOPs) was found to depend on the activity of Ac and Sc. Thus, strong Notch activation observed in cells surrounding emerging SOPs was found to depend on the activity of Ac-Sc, presumably because Ac and Sc are required for SOP specification and SOPs produce Delta, serving as the local source to activate Notch (see also our response to reviewer 3, point #6). Thus, since NRE-deGFP was not up-regulated in the proneural ADHN domain of sc10-1 ac3 mutant pupae, a quantitative analysis of the dynamics of NRE-deGFP may not be informative.

The reviewer also suggested us to study the dynamics of GFP-Sc in Notch mutants. One can easily predict that most Notch mutant cells would accumulate GFP-Sc, as observed in the notum (PMID: 28386027). Therefore, analysis of fate symmetry breaking is unlikely to be useful in that context. Likewise, a FDI analysis would not be relevant. From a technical point of view, live imaging of GFP-Sc would have to be performed in Notch mutant clones. This is because RNAi against Notch (strong 10xUAS-Notch hp2 construct, PMID: 19487563) driven by escargot-Gal4 to knock down Notch in larval histoblasts only led to a partial loss of Notch function (our unpublished data). Generation of Notch mutant clones in the abdomen would require constructing appropriate GFP-Sc Notch FRT recombinant chromosome as well as generating a new FRT GFP-Sc chromosome with an infrared marker (not currently available) to compare the relative dynamics of GFP-Sc in wild-type and mutant cells. In sum, this proposed experiment would take a significant amount of time and is unlikely to shed new light. Given that this experiment is not essential to support the claims of the paper and that it is not clear to us what would be learnt from this experiment, we opted for not performing this experiment.

Minor comments * 1. In figure 1F and F', the authors mention GFP-Sc is not expressed prior to 14h, however, there is still GFP signal detected in their imaging. Can the authors comment what would be the cause of this GFP signal or was it due to non-specific background signal during their imaging analysis?*

We thank the referee for raising this issue. Yes, a strong autofluorescence signal was detected prior to the onset of GFP-Sc expression. We provide below the results of our analysis of the autofluorescence signal (Fig R1) relative to the nuclear signal (Fig R2), and how normalization of the signal was used to measure the specific GFP-Sc signal.

Analysis of the autofluorescence signal over time

To estimate the autofluorescence signal, we measured the average intensity of the signal acquired in the GFP channel for each frame and plotted these values over time. The results are shown in Fig R1 below:

*Fig R1: temporal profile of the autofluorescence signal *

Each measurement corresponds to the average intensity measured in the GFP channel over the entire field at each z-section and for each time point. Mean and SD values of measured are shown over time in black and grey, respectively. Time is in frame number (dt is 2.5 min). The data shown above corresponds to movie 1 (see also Fig 2).

This plot indicates that the autofluorescence signal was progressively bleached. We therefore excluded from our analysis the first 50 time points when the autofluorescence signal was initially strong. No nuclear GFP-Sc signal was detectable in these first 50 frames in the cells of the central area of the ADHN which are studied here (see Fig 2A', t=1:12, time frame #29).

While revising the manuscript, we realized that t=0 corresponded to two distinct time points in the first version of our manuscript: it corresponded to the onset of imaging in Fig 2A-D', and to t=2:08 (time frame #51) in all other figures showing data following removal of the first 50 time points. We have now fixed this issue and are presenting all data with t=0 corresponding to the onset of imaging.

Analysis of the nuclear fluorescence signal over time

To detect the nuclear GFP-Sc signal, we measured the average intensity of the signal acquired in the GFP channel (raw intensity values corresponding to the sum of the GFP-Sc and autofluorescence signals) in segmented nuclei (in 3D, within the entire z-stack). These values were plotted over time (pink curve in Fig R2 below; the autofluorescence is plotted in black, as in Fig R1, for the sake of comparison). This showed that the intensity of the signal measured in nuclei was initially identical to the mean intensity measured across the entire field of view, indicative of autofluorescence only. A specific increase in signal intensity in nuclei (relative to the entire field of view) was detectable after 2h of imaging (time frame 48 in Fig R1; dt is 2.5 min). Importantly, mean intensity values of the autofluorescence signal appeared to be approximately 10-fold stronger than the mean intensity associated with the nuclear GFP-Sc signal.

Fig R2: temporal profile of the GFP-Sc signal

*The plot in pink corresponds to the average intensity in the GFP channel (raw intensity values corresponding to the GFP-Sc and/or autofluorescence signals) per nucleus (within the entire z-stack) for each time point. Mean and SD values measured in each nucleus are shown over time (in pink; these data correspond to movie 1; shown also in Fig 3). This plot (pink) should be compared with the plot shown in Fig R1 (also in black in Fig R2). The intensity difference between the pink and black curves was attributed to the specific GFP-Sc signal. *

Signal normalization and analysis of the GFP-Sc signal

In our study, we normalized the GFP-Sc signal by dividing the averaged value measured in each single nucleus (data corresponding to the pink curve in Fig R2) by the mean value of the signal measured at the same time point in the same channel in the entire image stack (data corresponding to the black curve in Fig R1/R2). Given the low intensity of the GFP-Sc signal, and the small number of pixels corresponding to Scute-expressing nuclei over the entire field of view, this value should closely reflect the autofluorescence noise. Thus, the background autofluorescence signal should be close to 1. This was experimentally verified by measuring the normalized intensity values of the PDHN nuclei that did not express Scute: a mean intensity value of 0.96 +/- 0.10 was measured (at time frame #51; see Fig R1 below). In contrast, the normalized GFP-Sc values measured several hours before SB were found to be close to 1.1 (see Fig 3D). Whether these values reflect very low levels of nuclear GFP-Sc that cannot be detected visually or result from imperfect normalization of the signal remain unclear. Given the intensity and non-uniformity of the autofluorescence signal, we cannot exclude the latter. For this reason, we chose to not over-interpret the initial low intensity values of GFP-Sc.

In the materials and methods, the authors mention that prior to imaging the larvae and pupae are grown at 18, 21 or 25{degree sign}C. Is there a reason why the larvae and pupae are grown at different temperatures for different experiments? Can the authors specify (i.e. in the figure legends) in which experiments different temperatures were used?

Larvae and pupae were grown at different temperatures for convenience, i.e. to adapt the time interval between staging at 0h APF and mounting for live imaging. Indeed, it is much easier to obtain 10-14h APF pupae by collecting staged pupae at 0h APF the day before and incubating them overnight at lower temperature to slow-down development. However, all live imaging experiments were performed at 23-25{degree sign}C, and we have no reason to think that this prior incubation would affect the process studied here.

The citations need to have a better format as they show up as each citation within a single bracket which makes it a little hard to read when multiple references are cited in a single sentence. fixed

In the abstract, the sentence 'Unexpectedly, we observed at low frequency (10%) pairs of cells that are in direct contact at the time of SB'. SB should be replaced with "Symmetry breaking", as it appeared for the first time in the manuscript and should be written out in full. fixed

Throughout the manuscript there are instances where the abbreviations are written in full with the abbreviation in brackets after they have already been introduced in the introduction which can be changed to just the abbreviation itself. fixed

In the discussion on page 11, 'our observation...', our needs to be changed to Our. fixed

7. It would be nice to have arrow heads or dotted lines around the cells or areas on interest in both, all the figures and movies, so that it will be easier to follow the results. The videos have a lot of background due to fragmented apoptotic nuclei, etc. as mentioned by the authors, hence arrow heads or dotted lines would bring viewers focus on the areas of interest.

fixed (see for instance Fig 1D, Fig 2A, Fig 5B, Fig 7A, Fig S3D, etc...)

8. It would be helpful to have anterior - posterior axis (i.e. with an arrow) shown on top of all the figures.

In our earlier version, we indicated that 'In this and all other figures, dorsal is up and anterior is left' in the legend of Fig 1B. We have now moved this sentence at the end of Fig 1 to have it more apparent. Additionally, the AP axis is now clearly indicated in Fig 1C. We believe that it is not necessary to repeat this orientation in all figures.

Scale bars are missing in all figures, videos, and figure legends. Added

Only movies 1 and 3 are referenced in the text. All movies are now referenced in the text

Keeping the colors in the movies and figures consistent and same would be helpful. For example, Movie 2 Histone3.3-mIFP marker is in blue but in figure 3 it is in magenta. fixed (H3.3-mIFP in magenta in this movie, now numbered 3)

As mentioned above, it would be helpful if the authors have arrow heads or dotted lines around the cells or areas of interest in both the figures and movies for better representation of their data. For example, movie 1 shows a larger area of imaging than shown in figure 2A, which makes it hard to follow the cells of interest in the movie.

An additional movie corresponding to the SOP shown in Fig 2A is now provided (new movie 2).

--

Reviewer #2

1. Despite "symmetry breaking" being the main focus of the paper, in the Introduction, the authors do not explain what this term means and do not provide any description of this process. This is a critical point that makes understanding of the goals of the paper difficult. Therefore, the authors are encouraged to provide more information and a clear description of this term/phenomenon. We thank the reviewer for this suggestion, we are now stating in the introduction what symmetry breaking means in the context of lateral inhibition: 'To describe and study the process of SOP selection, we studied fate SB. The latter refers to the transition point when one cell, the future SOP, starts to stably accumulate a higher level of GFP-Sc relative to its immediate neighbors.'

The role of Achaete in the story is not clear. Even though both factors are required for SOP determination, the authors mainly focus on Scute, so it is not very clear what the role of Achaete in this process is, if there is any. As shown in the paper, Achaete is expressed later when heterogeneity is promoting cell fate divergence. Is Achaete maybe contributing to cell heterogeneity/ cell fate divergence?

We thank the reviewer for raising this point. We now show in Fig S1A-D that abdominal bristles develop in a protein null allele of sc (scM6 ) as well as in an ac mutant corresponding to a 45 kb deletion that removes ac but not sc (PMID: 16216235)). Together with our analysis of sc10-1 ac3 __mutant flies, we can now conclude that __Sc and Ac act redundantly for SOP specification in the pupal abdomen. We have also further studied the expression of Ac relative to Sc and E(spl)HLH-m3 (see our response above to point #1 of reviewer 1). We fully agree with the reviewer that cell-to-cell variations in Ac expression might contribute to proneural heterogeneity and SB. This is now briefly discussed.

Minor points: * * 1. Symmetry Breaking (SB) should be abbreviated in the Abstract. The authors initially use the full term without abbreviation, and only on page 5, the abbreviation is finally defined; however, it should be introduced much earlier.

fixed

The second-to-last sentence in the abstract, "These lateral inhibition defects were correlated via cellular rearrangements," is unclear regarding what defects the authors are referring to.

This sentence was rewritten: 'Live imaging showed that these patterning defects were corrected via cellular rearrangements associated with global tissue fluidity, not via cell fate change.'

For clarity, being more specific in the text in regards to description of the figure panels would be beneficial (e.g. page 3 Fig 1C-E); referring to C-E together makes it hard to understand what does each panel shows.

fixed

In many instances, the movies are not properly referenced (e.g. on page 5, third row simply states "movies"), making it difficult to discern which movie should be checked. On page 8, when authors refer to movie 3, they likely meant movie 5.

fixed

Figure S1 requires some corrections.

We thank the reviewer for helping us improve the presentation of our results.

The authors use the short name "scute" initially and then switch to the shortened version "sc'.

fixed

Additionally, the nlsRFP (blue) is difficult to see; adjusting the levels or changing colors/showing separate channels may improve visibility.

The authors mention clone borders, but none are shown. It would greatly help to outline the borders in all figures.

The ubiquitous nlsRFP marker is now shown in magenta in Fig S1I that now shows only 2 channels to outline the ADHN (white dotted line) and the clones (yellow dotted lines).

We also outlined the clone borders in Fig 4C,C'.

Genotypes of the samples should be indicated, and clarification is needed regarding what "n" represents (number of cells, clones, or flies).

The genotype studied in Fig S1 and Fig 4 (which is the only complex genotype studied here) is now indicated in the Methods section. We have clarified what the different 'n' meant, in Fig 4 (see text) and elsewhere (see legend of Fig S2 for instance).

What do the arrows in the panel B show?

Thanks for pointing this out. The arrows in Fig S1I' indicate Cut/Hnt-positive cells (SOPs) within the clones (as now explained in the legend).

It is also recommended to display important channels as separate black and white images.

Separate channels are now shown in Fig S1 and S3.

Additionally, the use of RNAi against GFP instead of RNAi against scute should be justified; using RNAi GFP as the genotype on the graph could be interpreted as a control genotype rather than downregulation of scute.

A RNAi construct against GFP was used because this construct was known to very efficient and specific. Indeed, a strong knock-down of GFP-Sc was obtained by this approach (see Fig 4C'). We did not test sc RNAi constructs in the context of GFP-Sc. To avoid confusion, we are now indicating Sc downregulation (gfp RNAi) in Fig 4C'.

In the Figure 2 Legend, the authors use "std" as an abbreviation to define standard deviation. Typically, this is abbreviated as SD.

fixed

In Figure 4E, the authors do not explain on why there are points on the x-axis that correspond to a decimal number of cells.

Since heterogeneity was calculated over a 20 min interval, we likewise calculated the number of neighbors over the same time interval. Thus, the number of neighbors for each SOP corresponds to an averaged value calculated over this time interval. This is now explained in the legend.

--

Reviewer #3

1. First and foremost, the authors should state in the first paragraph of the Results that scGFP is a CRISPR knockin and thus it's the only source of Sc protein in the animals imaged (this is stated only in the Methods section). Thanks for this comment, we agree that this is one of the strengths of our work that we should emphasize. We now state in the results section: 'GFP-Sc is produced from the endogenous locus such that all Sc molecules produced in these pupae are GFP-tagged'

The magnitude of the Sc increase should be commented on. Based on the intensity and FDI plots in Fig. 3B-E, an increase of 15-17% in the amount of Sc is suggested (the FDI plateaus at 0.08, which gives 1.08/0.92 = 1.17x the level of Sc in the SOP vs the surrounding cells). However, in the stills shown in Fig. 2BCD and in Fig. 3A, the intensity differential between SOPs and neighbors seems at least 100% (ie at least double the intensity, which would yield an FDI of >1/3 =0.33). Why is this high contrast never seen in the quantitative measurements?

Thanks for asking about the fold change of GFP-Sc levels in SOPs, from SB to its plateau. This fold change can be seen in Fig 3D: the normalized value of GFP-Sc is 1.12 at SB, and 1.26 three hours after SB (when the FDI plateaus), indicative of a 2.2x fold increase of GFP-Sc in SOPs (0.26/0.12= 2.2, following background subtraction; see our detailed response to reviewer #1, minor point 1, about background signal analysis and normalization of the signal). This fold-change value is now indicated in the legend of Fig 3D. Obviously, this fold-change value is highly sensitive to signal normalization. Since the autofluorescence signal was stronger than the GFP-Sc signal (see Fig R2 above) and varied over time (due to bleaching; see Figs R1 and R2 above), we feel that this fold-change value should be taken with a grain of salt.

From Fig. 2A-D it appears that the ScGFP fluorescence intensity is at the same level or weaker than nearby autofluorescence. Please state (1) how you confirmed that the histoblast nest has lower autofluorescence than the larval epidermis and (2) how you corrected for histoblast nest autofluorescence in your quantifications.

As detailed above (our response to reviewer #1, minor point 1), the specific GFP-Sc signal is ten-times lower than the autofluorescence signal. We did not compare the autofluorescence signal produce by larval and imaginal cells (but note that larval epidermal cells had a stronger autofluorescence signal; see the yellow dots in Fig 2A). Normalization of the signal to correct for autofluorescence was explained in the Methods (and is also detailed above in our response to reviewer #1, minor point 1).

The paradoxical result of Fig. S1B should be discussed. On the one hand it is stated that "Ac and Sc specify the fate of the Sensory Organ Precursor cells (SOPs)" (p.2) and on the other S1B shows SOP specification in the absence of Sc. Are the SOPs shown in Fig S1B rare exceptions? Do the authors believe that these rare exceptions are there because of inefficient RNAi (since in comparison with S1A, in the null condition almost no SOPs should be formed)? Or they are the SOPs in RNAi clones as rare as the occasional bristles in S1A?

We do not see the result of Fig S1B as paradoxical but interpreted this result assuming that Ac and Sc were redundant for SOP determination. We now provide clear genetic evidence in support of this view (see our response above to reviewer #2, point 2). Otherwise, we found that RNAi is efficient (see loss of the GFP signal in clone in Fig. 4C'). In adult males, the density of bristles appeared to be quite normal over clonal patches of gfp RNAi cells (not shown), consistent with Ac being redundant with Sc

One figure that is not straightforward to interpret is Fig. 4E. It plots ScGFP heterogeneity vs. number of RNAi neighbors. Each point in the plot must be an individual SOP (165 total). Therefore, its neighbors (the x-axis) should take integral (not decimal) values. How can a single SOP have a decimal number of RNAi neighbors, especially since heterogeneity was sampled over a 10min time-window, when not much cell rearrangement can take place? Please explain.

Since heterogeneity was calculated over a 20 min interval, we likewise calculated the number of neighbors over the same time interval. Thus, the number of neighbors for each SOP corresponds to an averaged value calculated over this time interval. This is now explained in the legend: 'Note that the number of neighbors was likewise calculated over this time interval, and the resulting number of neighbors may not take an integral value.'

I found the discussion of the Notch reporter dynamics (Fig. 7) confusing in several places. * * (6a) Whereas it's clear that there is plenty of Notch signaling going on before SBN, the authors repeatedly imply that Notch signaling starts after SBN. For example, in the Results (p.9) they state "Thus, this quantitative approach failed to detect a phase of reciprocal Notch signaling during which proneural cluster cells would both send and receive a Delta-Notch signal prior to SOP emergence." The fact that the NRE-deGFP gave a robust signal before the start of the movies clearly means that mutual inhibition was going on for quite some time before SB. In fact, an FDI of 0 for >4h prior to SBN (Fig. 7G) means exactly this: that the level of Notch response among the cluster cells is equivalent ("mutual inhibition" lasts for at least 4h before SBN). (6b) In the first paragraph of this section (p.8) they comment that the pre-existence of Notch signaling is unexpected - why? I interpret it to simply be mutual inhibition (see above). Then they go on to quantitate the average Notch response intensity over the entire posterior ADHN (please define the borders the "posterior" ADHN). I question the informational value of this analysis (averaging over a large region), when Notch signaling is known to have intense local cell-to-cell variability (also evident in the stills shown in Fig. 7A,B,C).

We apologize for not describing well enough the data shown in Fig 7E, and for not explaining clearly our interpretation of the NRE-deGFP signal.

While the observation of a strong NRE-deGFP signal indeed indicates that Notch signaling had been active prior to the time of observation (in this sense, Notch is indeed active long before SBN), this does not necessarily imply that Notch is still active at that time. This is because the deGFP protein produced by the NRE-deGFP reporter is stable relative to the time scale of the studied process. Its measured half-life in S2 cells cultured at 25{degree sign}C is 2h (PMID: 31140975). Based on this data, the NRE-deGFP signal is likely to remain detectable several hours after Notch signaling has been switched off. If the rate of production of deGFP is lower than its rate of degradation, then the NRE-deGFP signal is expected to progressively decay over time. We believe that this is what we observed in our movies: while a strong signal was detected over the posterior half of the ADHN at 14-15h APF, this signal decreased over time (Fig 7D). To interpret the temporal dynamics of NRE-deGFP signal in terms of instantaneous Notch activity, we examined the Rate of Change (ROC): an increase of the NRE-deGFP signal over time (positive ROC) would indicate that Notch activity is increasing (more precisely, the production rate of deGFP is higher than its rate of degradation), whereas a decrease (negative ROC) indicates that Notch becomes less active (or inactive if the rate of decrease approximates the decay rate of the deGFP protein). Our data shown in Fig 7D showed that the NRE-deGFP signal (measured in the area indicated with a dotted line in Fig 7A,B; this area was defined by the initial pattern of NRE-deGFP) decreased over time (negative ROC) between t=1 and t=6.5h. We therefore conclude that Notch signaling is decreasing to reach a minimum at t=~3.5h, indicating that the level of Notch activity is at its lowest around the time of SB. At this minimum, the decay rate corresponds to a protein half-life of 4.4h, which is not so different from the measured half-time of deGFP in S2 cells (particularly if one assumes a 1.4x difference between the decay rates measured at 22 and 25{degree sign}C, based on the known temperature-dependent speed of development). This is why we conclude that Notch signaling is very low at this stage. Additionally, no NRE-deGFP signal was detected before t=4:30h (movie 7) in the initially NRE-deGFP negative cells (located anterior to the area indicated with a dotted line in Fig 7A). This indicated that Notch was activated late in this area. Together, our observations are not consistent with the view that Notch mediates a strong mutual inhibition signal over a prolonged time interval prior to SB.

To further study the pattern of Notch activity, we have monitored over time the accumulation pattern of GFP-tagged E(spl)m3-HLH (GFP-m3) (PMID: 31375669) in fixed sample (Fig S3F-G'). This confirmed that Notch was active in posterior ADHN cells and in the PDHN prior to 14h APF, i.e. prior to the onset of Ac and Sc, and that Notch activation extended to the central ADHN domain at 17-18h APF (Fig S3E-E' and G-G', and Fig 7I-I''), coinciding with SOP emergence.

Otherwise, the reviewer is correct when stating that a FDI value close to 0 indicates that the level of measured fluorescence among the different cells of the considered cluster is similar. Such a FDI value would be measured if cells did not express NRE-deGFP or had decreasing but similar levels of NRE-deGFP. This FDI value does not, per se, imply that Notch is active.

And then they move on to a (much more informative) cell-by-cell analysis, without even changing paragraphs, making it hard for the reader to follow. (6c) The conclusion at the end of the second paragraph (p. 9) "It also showed that SB was detected soon after the onset of Notch-mediated inhibitory signaling." is nowhere supported by data. If I understand well, SB refers to Sc and "the onset of Notch-mediated inhibitory signaling" refers to SBN (which is the onset of ASYMMETRY in Notch signaling, not the onset of Notch signaling, which has been going on for hours earlier). I don't see any data comparing SB with SBN. In fact, this is an important question to address (see below - comment 10).

We apologize for the lack of clarity in our writing, we meant: "It also showed that SBN was detected soon after the onset of Notch-mediated inhibitory signaling."

Yes, SBN refers to the onset of asymmetry in Notch signaling, as measured using NRE-deGFP. As explained above (but see also our response to point #7 below), our data do not provide evidence for a detectable Notch signal prior to SBN.

We agree that comparing SB and SBN would be nice. Unfortunately, our current tools do not permit a detailed comparison (see our detailed response below, point #10).

Mutual inhibition amongst neighboring cells has been proposed to involve (besides mutual Notch signaling) an increase in Sc levels in 2-3 cells in a cluster before the singularization of a single SOP. The authors seem rather biased against such a transient Sc hike based on their results in Fig. 2D, where the neighboring cells stay at rather constant basal Sc levels for several hours, while the Sc SB event happens. However, looking at an individual SOP in Fig. 2B, I do detect a mild hike in the pink curve right around SB in the blue curve. Could the average result from 160 SOPs (in Fig 2D) simply blur such transient Sc hikes, if they happen with different kinetics for different SOPs? Couldn't the 10% of SOP twins (shown in Fig. 6) represent a special case of this transient "subcluster" Sc hike? I would appreciate some discussion on this point. [Whether Sc is transiently upregulated or not, however, does not change my firm conclusion - from the data presented - that Notch-mediated mutual inhibition has been going on long before SBN.]

First, our data are consistent with the notion that a few proneural cells progressively accumulate higher level of Scute prior to SB (as proposed above). Indeed, the moderate increase in both GFP-Sc levels and coefficient of variation values (GFP-Sc heterogeneity) seen prior to SB correspond to what the reviewer has in mind (higher levels of GFP-Sc in a few proneural cluster cells). We also appreciate the reviewer's comment about the plot shown in Fig 2D. However, we strongly feel that our quantitative analysis of a large dataset is a strength. Thus, we do not find useful to discretize a continuous process by introducing the notion of 'subclusters' of 2-3 cells. Likewise, we believe that it is more informative to focus our analysis on the entire dataset using average and SD values and do not wish to base our interpretation of the process based on selected tracks (the one shown in Fig 2B only served as an illustration of how we performed our analysis and has no interpretation value).

The reviewer also states that "mutual inhibition amongst neighboring cells has been proposed to involve an increase in Sc levels in 2-3 cells in a cluster before the singularization of a single SOP". Since there is no published description of the pattern of accumulation of Scute in abdominal histoblats (to our best knowledge), we hypothesize that this statement applies to the proneural clusters in the developing wing disc. This is because the accumulation pattern of Sc has been studied in detail in that context by the Modollel and Carroll labs (PMID: 2044965, PMID: 2044964). However, their description of the accumulation pattern of Scute (in fixed samples, using anti-Sc antibodies) did not refer to sub-clusters of 2-3 cells. We would appreciate if the reviewer could direct us to the relevant published observation.

Finally, we are not sure to follow the reviewer when she/he firmly concluded from our data that Notch-mediated mutual inhibition has been going on long before SBN. Instead, our data clearly showed that the ADHN region that produced SOPs exhibited two distinct NRE-deGFP patterns, with Notch signaling being active prior to imaging (i.e. prior to 14h APF) and decreasing to reach a minimum of Notch activation around 17h APF (i.e. around the time of SB, as determined by GFP-Sc imaging) in the posterior area of the ADHN.

Thus, our data do not show that mutual inhibition does not take place in this tissue but rather imply that the phase of mutual inhibition (or competition) must be relatively short, or transient, and that competition amongst proneural cluster cells operate at low Notch and Sc levels (probably contrary to what many people have in mind).

Some minor points: * * 8. Please change Cad-GFP to Ecad-GFP or shg-GFP, as Cad misdirects to caudal.

Thanks, changed into Ecad-GFP and Ecad-mKate

What is c in "(x,y,z,c,t) movies"? (a fifth independent variable?)

c stands for channel. This is relatively standard nomenclature.

The authors show that Sc displays a SB event leading to FDI of 0.08 and the Notch reporter displays another SB (SBN) leading to a much more pronounced FDI of -0.2. Are these two events (the hike of Sc levels and the plummeting of Notch signal) contemporaneous or does one precede the other? Having both tagged with GFP makes it impossible to image simultaneously, but the authors could register each reporter's dynamics relative to the time of SOP division (as done in Fig. 5C) to get a sense of their relative order.

We do agree with the reviewer that it would be nice to be able to align in time these two data sets. Unfortunately, the temporal correlation between SB and the SOP division is too variable (4.7 +/- 1.1) to confidently align these two datasets using this event as a time reference. New tools are needed (see our response to point #11 below).

Where in the above timeline is the SOP fate definitively adopted? neur-nlsGFP, Ac-RFP, m3Cherry and Sens detection in Figs. 1 and 7 give us a rough idea that these other markers appear around the time of Sc FDI peaking, around 3h after the initial SB. But this is not presented in an organized fashion - the reader collects this information sporadically. A reanalysis of the already existing data attempting to place these various markers in an integrated timeline would be of great importance in understanding the details of this cell fate specification process. Which is the earliest SB event? sc, neur or Notch? How long does it take from that early SB until definitive SOP markers (Sens) first appear?

We agree with the reviewer, it would be interesting to extend the approach reported here for Scute to characterize SB and rate of FDI for other key factors governing the selection of SOPs. As pointed out by the reviewer (point #10 above), it would also be important to register in time these various events. Unfortunately, the maturation time of RFP, mCherry, FP670, etc... appeared to be too slow relative to the rapid turnover of the Ac, Sc and E(spl)-HLH factors prevented us from performing two-color imaging. Hence, current tools do not permit to determine which is the earliest SB event.

More genetic perturbations could be performed to solidify the model of cell-cell communication during lateral inhibition. Two obvious ones come to mind: (a) How would the Sc-GFP dynamics change in a Notch-RNAi background? (b) How would the NRE-deGFP dynamics change in a sc-RNAi background?

See our detailed response to reviewer #1, major point #2.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Phan et al present their work on the dynamics of Sc accumulation and Notch signaling in the dorsal abdomen of Drosophila. They use live imaging to provide more detailed knowledge on a well-studied phenomenon of lateral inhibition, where proneural proteins (like Sc) promote a certain fate (like SOP) in groups of equivalent cells. The selection of individual SOPs among these equivalent cells depends on Delta-Notch signaling, which allows neighboring cells to communicate. Cells that accumulate higher levels of Sc acquire an increased ability to inhibit their neighbors and will ultimately become SOPs. This paradigm seems to be used time and again in arthropods and vertebrates alike and its central importance is witnessed by the fact that it has been under intense study over the last 3 decades. The present manuscript adds the temporal dimension to one instance that deploys this proneural-Notch interplay. The authors show that the timecourse from equivalence to singularized SOPs takes 3-8 hours (from 13 to 21 h APF) and is visible as an increase in Sc-GFP levels in one cell of the cluster. They calculated precisely the onset of this increase (which they term "symmetry breaking") and showed that Sc levels plateau approx. 3 h later, although some SOPs are faster than others in reaching that plateau. Sc increase is accompanied by Notch reporter decrease. The apical area of the cells does not seem to bias the level of Notch signaling/ Sc accumulation. What does seem to speed up the process is pre-existing heterogeneity in Sc (and Notch?) levels. Interestingly, when this process of lateral inhibition fails to singularize a single cell (resulting in two adjacent cells with high Sc levels), these two cells move apart by cellular rearrangements. During the 3h that the SOP upregulates its Sc levels after SB, its neighboring cells stay at relatively constant baseline Sc levels and only afterwards do they start reducing their own Sc-GFP. The authors have taken the trouble to collect live data for >100 SOPs in different experimental settings, so there is no doubt about their reproducibility and statistical robustness. In general, the figures are clear and self-explanatory. I found it hard to follow the text, however, and I have some suggestions for its improvement.

1. First and foremost, the authors should state in the first paragraph of the Results that scGFP is a CRISPR knockin and thus it's the only source of Sc protein in the animals imaged (this is stated only in the Methods section).
2. The magnitude of the Sc increase should be commented on. Based on the intensity and FDI plots in Fig. 3B-E, an increase of 15-17% in the amount of Sc is suggested (the FDI plateaus at 0.08, which gives 1.08/0.92 = 1.17x the level of Sc in the SOP vs the surrounding cells). However, in the stills shown in Fig. 2BCD and in Fig. 3A, the intensity differential between SOPs and neighbors seems at least 100% (ie at least double the intensity, which would yield an FDI of >1/3 =0.33). Why is this high contrast never seen in the quantitative measurements?
3. From Fig. 2A-D it appears that the ScGFP fluorescence intensity is at the same level or weaker than nearby autofluorescence. Please state (1) how you confirmed that the histoblast nest has lower autofluorescence than the larval epidermis and (2) how you corrected for histoblast nest autofluorescence in your quantifications.
4. The paradoxical result of Fig. S1B should be discussed. On the one hand it is stated that "Ac and Sc specify the fate of the Sensory Organ Precursor cells (SOPs)" (p.2) and on the other S1B shows SOP specification in the absence of Sc. Are the SOPs shown in Fig S1B rare exceptions? Do the authors believe that these rare exceptions are there because of inefficient RNAi (since in comparison with S1A, in the null condition almost no SOPs should be formed)? Or they are the SOPs in RNAi clones as rare as the occasional bristles in S1A?
5. One figure that is not straightforward to interpret is Fig. 4E. It plots ScGFP heterogeneity vs. number of RNAi neighbors. Each point in the plot must be an individual SOP (165 total). Therefore, its neighbors (the x-axis) should take integral (not decimal) values. How can a single SOP have a decimal number of RNAi neighbors, especially since heterogeneity was sampled over a 10min time-window, when not much cell rearrangement can take place? Please explain.
6. I found the discussion of the Notch reporter dynamics (Fig. 7) confusing in several places. (6a) Whereas it's clear that there is plenty of Notch signaling going on before SBN, the authors repeatedly imply that Notch signaling starts after SBN. For example, in the Results (p.9) they state "Thus, this quantitative approach failed to detect a phase of reciprocal Notch signaling during which proneural cluster cells would both send and receive a Delta-Notch signal prior to SOP emergence." The fact that the NRE-deGFP gave a robust signal before the start of the movies clearly means that mutual inhibition was going on for quite some time before SB. In fact, an FDI of 0 for >4h prior to SBN (Fig. 7G) means exactly this: that the level of Notch response among the cluster cells is equivalent ("mutual inhibition" lasts for at least 4h before SBN). (6b) In the first paragraph of this section (p.8) they comment that the pre-existence of Notch signaling is unexpected - why? I interpret it to simply be mutual inhibition (see above). Then they go on to quantitate the average Notch response intensity over the entire posterior ADHN (please define the borders the "posterior" ADHN). I question the informational value of this analysis (averaging over a large region), when Notch signaling is known to have intense local cell-to-cell variability (also evident in the stills shown in Fig. 7A,B,C). And then they move on to a (much more informative) cell-by-cell analysis, without even changing paragraphs, making it hard for the reader to follow. (6c) The conclusion at the end of the second paragraph (p. 9) "It also showed that SB was detected soon after the onset of Notch-mediated inhibitory signaling." is nowhere supported by data. If I understand well, SB refers to Sc and "the onset of Notch-mediated inhibitory signaling" refers to SBN (which is the onset of ASYMMETRY in Notch signaling, not the onset of Notch signaling, which has been going on for hours earlier). I don't see any data comparing SB with SBN. In fact, this is an important question to address (see below - comment 10).
7. Mutual inhibition amongst neighboring cells has been proposed to involve (besides mutual Notch signaling) an increase in Sc levels in 2-3 cells in a cluster before the singularization of a single SOP. The authors seem rather biased against such a transient Sc hike based on their results in Fig. 2D, where the neighboring cells stay at rather constant basal Sc levels for several hours, while the Sc SB event happens. However, looking at an individual SOP in Fig. 2B, I do detect a mild hike in the pink curve right around SB in the blue curve. Could the average result from 160 SOPs (in Fig 2D) simply blur such transient Sc hikes, if they happen with different kinetics for different SOPs? Couldn't the 10% of SOP twins (shown in Fig. 6) represent a special case of this transient "subcluster" Sc hike? I would appreciate some discussion on this point. [Whether Sc is transiently upregulated or not, however, does not change my firm conclusion - from the data presented - that Notch-mediated mutual inhibition has been going on long before SBN.] Some minor points:
8. Please change Cad-GFP to Ecad-GFP or shg-GFP, as Cad misdirects to caudal.
9. What is c in "(x,y,z,c,t) movies"? (a fifth independent variable?)

Significance

The mechanism of proneural-Notch interplay is evolutionarily conserved, so this study of its temporal dynamics is valuable and will interest a broad audience in the field of animal developmental biology. The rich data collected by the authors should contain enough information to make a big contribution to the field, but the presentation in the manuscript stops a little short of that. The fact that Sc expression is highly dynamic was already known - now we have quantitative measurements of this variability. Same holds for Notch signaling. The authors should try to integrate their data better to make a complete timeline of events that leads to SOP specification, using the panoply of fluorescent markers at their disposal.

1. The authors show that Sc displays a SB event leading to FDI of 0.08 and the Notch reporter displays another SB (SBN) leading to a much more pronounced FDI of -0.2. Are these two events (the hike of Sc levels and the plummeting of Notch signal) contemporaneous or does one precede the other? Having both tagged with GFP makes it impossible to image simultaneously, but the authors could register each reporter's dynamics relative to the time of SOP division (as done in Fig. 5C) to get a sense of their relative order.
2. Where in the above timeline is the SOP fate definitively adopted? neur-nlsGFP, Ac-RFP, m3Cherry and Sens detection in Figs. 1 and 7 give us a rough idea that these other markers appear around the time of Sc FDI peaking, around 3h after the initial SB. But this is not presented in an organized fashion - the reader collects this information sporadically. A reanalysis of the already existing data attempting to place these various markers in an integrated timeline would be of great importance in understanding the details of this cell fate specification process. Which is the earliest SB event? sc, neur or Notch? How long does it take from that early SB until definitive SOP markers (Sens) first appear?
3. More genetic perturbations could be performed to solidify the model of cell-cell communication during lateral inhibition. Two obvious ones come to mind: (a) How would the Sc-GFP dynamics change in a Notch-RNAi background? (b) How would the NRE-deGFP dynamics change in a sc-RNAi background?

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

Understanding cell fate is crucial for various biological processes, including development, tissue regeneration, and disease progression. In this manuscript, colleagues in the team of François Schweisguth provide high-quality live cell observations demonstrating how SOP (Sensory Organ Precursor) cells are determined from a cluster of proneuronal cells via symmetry breaking (SB), which is key in the process of lateral inhibition. Although the process of lateral inhibition has been proposed long ago, no studies have been conducted to demonstrate it using live imaging. Using Drosophila abdominal epidermis as a model, the authors showed how the levels of the main neuronal determinants are expressed during SOP cell determination. The team uses a tagged version of one of the cell fate determinants, Scute, to follow the dynamics of this process, which is then further supported by genetic experiments showing that cell-to-cell variations in Scute expression levels promote fate divergence and patterning. This paper provides a new perspective on how dynamic expression of proneural factors determines cell fate acquisition from the equivalent population of cells. The data is presented very clearly, and the methods are adequately detailed, with suitable experiments and statistical analysis, as well as convincing key conclusions.

Major points:

1. Despite "symmetry breaking" being the main focus of the paper, in the Introduction, the authors do not explain what this term means and do not provide any description of this process. This is a critical point that makes understanding of the goals of the paper difficult. Therefore, the authors are encouraged to provide more information and a clear description of this term/phenomenon.
2. The role of Achaete in the story is not clear. Even though both factors are required for SOP determination, the authors mainly focus on Scute, so it is not very clear what the role of Achaete in this process is, if there is any. As shown in the paper, Achaete is expressed later when heterogeneity is promoting cell fate divergence. Is Achaete maybe contributing to cell heterogeneity/ cell fate divergence? Minor points:
3. Symmetry Breaking (SB) should be abbreviated in the Abstract. The authors initially use the full term without abbreviation, and only on page 5, the abbreviation is finally defined; however, it should be introduced much earlier.
4. The second-to-last sentence in the abstract, "These lateral inhibition defects were correlated via cellular rearrangements," is unclear regarding what defects the authors are referring to.
5. For clarity, being more specific in the text in regards to description of the figure panels would be beneficial (e.g. page 3 Fig 1C-E); referring to C-E together makes it hard to understand what does each panel shows.
6. In many instances, the movies are not properly referenced (e.g. on page 5, third row simply states "movies"), making it difficult to discern which movie should be checked. On page 8, when authors refer to movie 3, they likely meant movie 5.
7. Figure S1 requires some corrections. The authors use the short name "scute" initially and then switch to the shortened version "sc'. Additionally, the nlsRFP (blue) is difficult to see; adjusting the levels or changing colors/showing separate channels may improve visibility. The authors mention clone borders, but none are shown. It would greatly help to outline the borders in all figures. Genotypes of the samples should be indicated, and clarification is needed regarding what "n" represents (number of cells, clones, or flies). What do the arrows in the panel B show? It is also recommended to display important channels as separate black and white images. Additionally, the use of RNAi against GFP instead of RNAi against scute should be justified; using RNAi GFP as the genotype on the graph could be interpreted as a control genotype rather than downregulation of scute.
8. In the Figure 2 Legend, the authors use "std" as an abbreviation to define standard deviation. Typically, this is abbreviated as SD.
9. In Figure 4E, the authors do not explain on why there are points on the x-axis that correspond to a decimal number of cells.

Reviewer Cross-Commenting

I fully agree with the comments provided by the other reviewers, most of which were complementary and overlapping with mine. Their comments are well-reasoned and highlight certain aspects that I overlooked. I agree that conducting additional genetic analyses would enhance our understanding of the progression fate divergence and its impact on lateral inhibition in real-time. Specifically, exploring the spatial and temporal dynamics of Ac and Sc in Notch mutants, as well as Notch dynamics in Sc and Ac mutants, could strengthen the proposed model of cell-cell communication during lateral inhibition.

Significance

Understanding cell fate is crucial for various biological processes, including development, tissue regeneration, and disease progression. In this manuscript, colleagues in the team of François Schweisguth provide high-quality live cell observations demonstrating how SOP (Sensory Organ Precursor) cells are determined from a cluster of proneuronal cells via symmetry breaking (SB), which is key in the process of lateral inhibition. Although the process of lateral inhibition has been proposed long ago, no studies have been conducted to demonstrate it using live imaging. Using Drosophila abdominal epidermis as a model, the authors showed how the levels of the main neuronal determinants are expressed during SOP cell determination. The team uses a tagged version of one of the cell fate determinants, Scute, to follow the dynamics of this process, which is then further supported by genetic experiments showing that cell-to-cell variations in Scute expression levels promote fate divergence and patterning. This paper provides a new perspective on how dynamic expression of proneural factors determines cell fate acquisition from the equivalent population of cells. The data is presented very clearly, and the methods are adequately detailed, with suitable experiments and statistical analysis, as well as convincing key conclusions.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary

In this manuscript, the authors addressed when and how fate symmetry breaking occurs during lateral inhibition using live imaging of Drosophila pupal abdomen as a model system. Using quantitative live imaging of a GFP tagged version of the proneural gene scute (sc), the authors demonstrated that GFP-Sc expression appears first along the posterior margin of the anterior dorsal histoblast nest (ADHN) and later in the central region of ADHN, suggesting that posterior cues regulate early proneural expression in this region, which is consistent with previous findings. By tracking the temporal expression of GFP-Sc in the sensory organ precursor cells (SOPs), the authors further showed that SOPs emerge within a 2 hour time frame around 17h APF in the ADHN. Moreover, the presumptive SOP and its surrounding histoblasts showed a weak and slowly increasing GFP-Sc signal until the presumptive SOP showed a rapid increase in GFP-Sc accumulation, whereas GFP-Sc levels remained relatively constant in non-selected histoblasts. Interestingly, using symmetry breaking as a reference point, the authors found that the onset of fate divergence took place at low levels of Sc, soon after the onset of proneural gene expression and was not preceded by a prolonged phase of Sc accumulation. The authors also demonstrated that lateral inhibition in the pupal abdomen failed to single out SOPs in around 10% of the cell clusters that are in direct contact with each other during symmetry breaking and that pattern refinement involving cellular rearrangements were required to correct these defects. By manipulating the heterogeneity of Sc in genetically mosaic pupae, the authors successfully showed that increasing the heterogeneity of Sc positively correlated with an increased rate of fate divergence, suggesting that cell-to-cell variations in Sc levels promote fate divergence during lateral inhibition. Although earlier modelling suggested that differences in apical cell area may serve as a possible source of bias for Notch-based decisions, the authors found no significant difference in the apical area of the presumptive SOP compared to the mean area of its neighbouring cells, suggesting that apical cell shape does not bias Notch-mediated cell fate decisions in this developmental context. Finally, examination of Notch activity dynamics using a destabilized GFP expressed downstream of a Notch-Responsive Element as well as analysing the expression of the E(spl)m3 Notch target gene, the authors demonstrate that Notch activity was minimal around the time of SOP emergence as E(spl)m3 was detected after the onset of Sc expression but not prior to SOP emergence, indicating that SOP selection in the abdominal epidermis took place at low levels of Notch signalling.

Major comments:

• Are the key conclusions convincing?

Yes, the key conclusions are convincing with quantitative live imaging and proper explanations on how the analyses were carried out. - Should the authors qualify some of their claims as preliminary or speculative, or remove them altogether?

No.

• Would additional experiments be essential to support the claims of the paper? Request additional experiments only where necessary for the paper as it is, and do not ask authors to open new lines of experimentation.
• The authors conclude that RFP-Ac expression is restricted to emerging SOPs and surroundings cells at 18h APF, indicating that Ac is activated later than Sc. Can the authors provide images for RFP-Ac expression at 10h and 16h APF similar to GFP-Sc as shown in their figures. Do the SOPs that contain high levels of both Ac and Sc (as some SOPs have Sc expression but not Ac) undergo fate divergence and SB faster than the SOPs containing higher levels of only Sc?
• It would be interesting to see the spatial and temporal dynamics of Ac and Sc in Notch mutants or even Notch dynamics in Sc and Ac mutants to better understand the progression fate divergence and its effect on lateral inhibition in real time.
• Are the suggested experiments realistic in terms of time and resources? It would help if you could add an estimated cost and time investment for substantial experiments.

Yes, I believe the suggested experiments are realistic in terms of time and resources, with an estimation of 3-4 months to complete the experiments. - Are the data and the methods presented in such a way that they can be reproduced?

Yes. - Are the experiments adequately replicated and statistical analysis adequate?

The experiments are straight forward and were performed with a good number of n values for their live imaging and supported by quantifications.

Minor comments:

• Specific experimental issues that are easily addressable.
• In figure 1F and F', the authors mention GFP-Sc is not expressed prior to 14h, however, there is still GFP signal detected in their imaging. Can the authors comment what would be the cause of this GFP signal or was it due to non-specific background signal during their imaging analysis?
• In the materials and methods, the authors mention that prior to imaging the larvae and pupae are grown at 18, 21 or 25C. Is there a reason why the larvae and pupae are grown at different temperatures for different experiments? Can the authors specify (i.e. in the figure legends) in which experiments different temperatures were used?
• The citations need to have a better format as they show up as each citation within a single bracket which makes it a little hard to read when multiple references are cited in a single sentence.
• In the abstract, the sentence 'Unexpectedly, we observed at low frequency (10%) pairs of cells that are in direct contact at the time of SB'. SB should be replaced with "Symmetry breaking", as it appeared for the first time in the manuscript and should be written out in full.
• Throughout the manuscript there are instances where the abbreviations are written in full with the abbreviation in brackets after they have already been introduced in the introduction which can be changed to just the abbreviation itself.
• In the discussion on page 11, 'our observation...', our needs to be changed to Our.
• Are prior studies referenced appropriately?

Yes. - Are the text and figures clear and accurate? 1. It would be nice to have arrow heads or dotted lines around the cells or areas on interest in both, all the figures and movies, so that it will be easier to follow the results. The videos have a lot of background due to fragmented apoptotic nuclei, etc. as mentioned by the authors, hence arrow heads or dotted lines would bring viewers focus on the areas of interest. 2. It would be helpful to have anterior - posterior axis (i.e. with an arrow) shown on top of all the figures. 3. Scale bars are missing in all figures, videos, and figure legends. 4. Only movies 1 and 3 are referenced in the text. 5. Keeping the colors in the movies and figures consistent and same would be helpful. For example, Movie 2 Histone3.3-mIFP marker is in blue but in figure 3 it is in magenta. - Do you have suggestions that would help the authors improve the presentation of their data and conclusions?

As mentioned above, it would be helpful if the authors have arrow heads or dotted lines around the cells or areas of interest in both the figures and movies for better representation of their data. For example, movie 1 shows a larger area of imaging than shown in figure 2A, which makes it hard to follow the cells of interest in the movie.

Significance

Since the dynamics of fate acquisition has mostly been studied in fixed samples in Drosophila, this is an interesting study to understand the spatial and temporal dynamics of Notch signalling as well as proneural activity during lateral inhibition using Drosophila pupal abdomen as a model. Using quantitative live imaging the authors provide key evidence on how and when SB occurs during lateral inhibition, providing experimental support for the intracellular feedback model. Most importantly, they show that fate selection occurred early and was not preceded by a detectable phase of mutual inhibition, but instead, the initial bias in Sc expression and heterogeneity might play a significant role in in SB.

• Place the work in the context of the existing literature (provide references, where appropriate).

So far, most work in this field has focused on the dynamics of fate acquisition using fixed samples and mathematical remodelling, with some live imaging analysis (Skeath and Carol, 1991; Collier et al, 1996; Castro et al, 2005; Corson et al, 2017; Couturier et al, 2019; Troost et al, 2023). Here, considering previous literature, the authors move one step forward, using quantitative live imaging of proneural factors Scute to determine fate SB and monitor fate divergence during lateral inhibition. This study though not entirely conceptually novel provides important new insights into SB and fate divergence in the pupal abdomen, wherein symmetry breaking occurred at low Sc levels and that fate divergence was not preceded by a prolonged phase of low or intermediate level of Sc accumulation. Furthermore, the relative size of the apical area did not influence this fate choice but cell-to-cell variations in Sc levels promoted fate divergence, thereby providing experimental support for the intercellular negative feedback loop model. - State what audience might be interested in and influenced by the reported findings.

Developmental and cell biologists. - 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.

Stem cells, developmental biology.

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

Learn more at Review Commons

---

Reply to the reviewers

Response to reviewers

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 Hazari et al reports on an important new function of the UPR IRE1 in liver fibrosis. The results are clearly and logically described. It shows that the genetic ablation of IRE1 in mice prevents liver fibrosis through retention of collagen in the ER and its degradation hence preventing secretion and accumulation in liver parenchyma. This effect was reversed was P4HB (a multifunctional enzyme that belongs to the protein disulfide isomerase) showing that IRE1 control P4HB.

As such the paper is both scientifically and medically relevant.

There is in my opinion a conceptual issue the paper does not directly address nor resolve.

A clear functional distinction between IRE1 and XBP1 is not made nor attempted. This comes across as an unresolved issue. In the Introduction the authors cite ref 29 that provided evidence for a role of RIDD in alleviating hepatic cytotoxicity. The paper is based on chronic liver toxicity by CCL4 to identify the protective role of IRE1 and acute CCL4 toxicity to identify P4HB. IRE1 KO demonstrates that IRE1 controls collagen metabolism (degradation vs. secretion). However, many considerations involve XBP1 (see Fig. 8 as a example). Yet the paper concludes by saying "we propose that pharmacological inhibition of IRE1 activation..." Granted that IRE1 inhibition would definitely cause an attenuation of XBP1 splicing, I see a clear distinction between IRE1 and XBP1 still necessary to back the conclusion that inhibition of IRE1 and not XBP1 is the therapeutic modality one should develop.

It is clear that IRE1 controls XBP1, but it also controls RIDD. Both independently control the fate of multiple downstream genes and also miRNAs in the case of RIDD. Because RIDD may offer some advantages in attenuating liver pathology, forfeiting some benefits IRE1 can offer via RIDD in order to blunt XBP1 may not be the optimal solution. Therefore, I suggest to complement the present study with experiments that target XBP1 specifically bypassing IRE1. Experiments of deletion (siRNA or Cre) as well as specific activation of XBP1 for which there exist commercially available molecules (e.g., IAX4 is a direct activator of XBP1 without UPR that also does not induce XBP1s-independent IRE1 signaling such as RIDD or JNK phosphorylation) will permit to better differentiate the role of XBP1 from IRE1 in P4HB regulation and collagen degradation vs. secretion in hepatic stellate cells and deposition in liver.

Throughout the paper the authors refer to XBP1 and even center the Discussion on XBP1 more than on IRE1. Since determining the precise mechanism in this particular instance is critical to future treatments to prevent liver fibrosis these additional experiments should be performed.

Minor point

Fig. 4D. The label must have been erroneously copied and pasted. There is not way to distinguish what is different in lane 1-2 from lane 3-4. The legend is not helpful.

Fig. S4B. Same problem.

Significance

This is an excellent paper with profound new medical implications with conceptual advances for the treatment of NASH.

The limitations have been underscored in comments to authors.

The audience remains specialized for the time being.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

The current study aimed to investigate the possible contribution of the unfolded protein response (UPR), the main adaptive pathway that monitors and adjusts the protein production capacity at the endoplasmic reticulum (ER), to collagen biogenesis and liver disease. The authors targeted the ER sensor, inositol requiring transmembrane kinase/endoribonuclease 1 alpha (IRE1), using the IRE1Lox/Lox, Mx1Cre/- mouse strain where Cre is induced with 3 poly I:C injections. After confirmation of depleting IRE1 in the liver, they challenged mice with a high dose of carbon tetrachloride (CCl4). Phenotype analysis revealed the deleterious role of IRE1 in the liver in acute liver damage. Then, the authors determined the biological consequence of IRE1 deletion on the progression of experimental liver fibrosis. Consistently, IRE1 plays a promotive role in the CCl4-induced liver fibrosis mouse model since the results suggest that IRE1 knockout mice have reduced collagen deposition and AST/ALT activity (two major liver damage enzymes). To determine the potential mechanisms underlying the protective effects of IRE1 deficiency, the authors did mass spectrometry-based quantitative proteomics of liver samples from the experimental mice. Through bioinformatic analysis, they determined that protein and mRNA expression levels of P4HB decreased in IRE1 deficiency mice. Besides, in vitro and ex vivo studies suggested that IRE1 deficiency downregulated collagen generation. They further investigated the impact of IRE1 deficiency in liver steatosis in a mouse model fed a high-fat diet (HFD). Similarly, IRE1 deficiency suppressed liver steatosis and fibrosis in this mouse model. To clarify how IRE1 modulates collagen expression, the authors generated IRE1 knockout cell (KO) lines. They found that IRE1 KO cells may cause misfolding and intracellular accumulation of collagen inside the cells, reducing its secretion. Finally, they suggested the correlation of IRE1 associated gene (XBP1) and PH4B but not PH4A1 in human patients with non-alcoholic steatohepatitis (NASH). Overall, the current study may provide a potential molecular linkage between IRE1 and collagen in chronic liver disease progression. However, the inconsistent data presentation, poor data quality, and lack of suitable animal models make the current manuscript's impact on the field of chronic liver disease low.

Major concern:

1. The authors suggested that P450 levels are not changed in the liver samples of IRE1 KO mice compared to the wildtype (WT) ones. It has been demonstrated that IRE1 activation reduced P450 expression levels (PMID: 22291093). Can the authors explain the inconsistent findings?
2. Why did the authors challenge mice with CCl4 for 12 weeks? The CCl4-induced liver fibrosis model would have a severe fibrotic liver phenotype at 8 weeks. Did the author check if IRE1 deficient mice have less liver injury than wildtype (WT) mice at early time points? Also, could you check the liver fibrosis phenotype using another typical liver fibrosis mouse model? The HFD-feeding mouse model would induce liver fibrosis after challenging mice for 50 weeks (PMID: 37270060).
3. What are the mouse numbers, age, and sex in the CCl4 studies? In Fig. 2, several figures have inconsistent mouse numbers for the data presentation.
4. There are many downregulated and upregulated target genes. What is the rationale for focusing on downregulated P4HB in IRE1 deficient mice, given that IRE1 has an RNase domain functioning in posttranscriptional regulation? In this case, how does IRE1 depletion upregulate P4HB?
5. The author suggested that TGFb1 would promote collagen generation, but targeting IRE1 would reduce P4HB and the TGFb1-mediated collagen generation. The Fig. 4D does not support the statement. However, the authors' data demonstrated that TGFb1 cannot promote IRE1 and P4HB expression (Fig. 4G). If TGFb1 cannot promote collagen generation through activating IRE1 and P4HB, how can targeting IRE1 reduce TGFb1-induced collagen (Fig. 4H)?
6. The authors stated that hepatic Ern1 deficiency suppresses the progression of liver steatosis. In fact, it has been reproducibly demonstrated that hepatic IRE1 deletion promotes hepatic steatosis progression (PMID: 21407177 and PMID: 29764990), contradicting the authors' findings. Besides, the authors did not deplete IRE1 specifically in the liver. Therefore, they made a misleading conclusion without solid evidence. The authors basically depleted IRE1 in the whole body of these experimental mice upon poly I:C injection. The authors must not conclude that hepatic Ern1 deficiency suppresses the progression of liver steatosis without considering the contributions of other organs in the phenotypes that you observed. Also, the HFD-treated mice would develop liver fibrosis 50 weeks post-feeding (PMID: 37270060). It is unclear what other treatments the authors used to accelerate the fibrotic liver phenotype, as shown in Fig. 5D. The authors should show body weight and liver weight over body weight in the result section. Besides, hepatic cholesterol, serum triglyceride, and cholesterol need to be measured in these mice to clarify how deleting IRE1 in the whole body can suppress hepatic steatosis, but liver-specific deletion of IRE1 promotes fatty liver. Without clarifying this issue, it is unclear how hepatic IRE1 deficiency can reduce steatosis and liver fibrosis.
7. The authors suggested that ablation of IRE1 expression increased the levels of intracellular GFP-collagen as compared with control cells (Fig. 6C). How did the authors quantify the results? It is not clear if KO really increased the intracellular collagen levels. As the authors showed in Fig. 6C, WT-NT, and WT-GFP-collagen-untreated have no overlap of green fluorescence. However, KO-NT and KO-GFP-collagen-untreated still have an overlap of green fluorescence, indicating that some cells are not GFP-positive. In this case, how could authors conclude that IRE1-KO cells have a more than 2-fold increase of green fluorescence change compared to WT? Besides, Fig. 6F suggested that secreted collagens increased in KO cells, contradicting the authors' previous data in Fig. 2, 4, and 5. Why did you use U2OS, Hepa1-6, and Huh7 in these studies? Should the collagen be secreted by hepatic stellate cells?
8. In Fig. 7, the authors suggested that IRE1 KO promotes the levels of collagen inside cells using the whole cell lysate. Interestingly, they indicated that IRE1 deficiency suppressed TGFb1-induced collagen production using whole cell lysates (Fig. 4D). It is really confusing if IRE1 KO promotes or suppresses collagen production or secretion. Also, Fig. 7C did not support that IRE1-KO reduced collagen secretion. Besides, what cells did the authors use for these studies? Are they hepatic stellate cells?
9. It is interesting to see the positive correlation between XBP1 and P4HB mRNA expression. However, it is still unclear if IRE1 deficiency could downregulate P4HB mRNA expression, given its RNase function. Thus, it would be essential to determine how IRE1 regulates P4HB expression before analyzing the correlation using human datasets. Besides, Fig. 8D did not suggest that XBP1 expression levels are really correlated with chronic liver disease progression, given that its correlation scores with AST and ALT are 0 and -0.01, respectively.

Significance

The current study aimed to investigate the possible contribution of the unfolded protein response (UPR), the main adaptive pathway that monitors and adjusts the protein production capacity at the endoplasmic reticulum (ER), to collagen biogenesis and liver disease. The authors targeted the ER sensor, inositol requiring transmembrane kinase/endoribonuclease 1 alpha (IRE1), using the IRE1Lox/Lox, Mx1Cre/- mouse strain where Cre is induced with 3 poly I:C injections. After confirmation of depleting IRE1 in the liver, they challenged mice with a high dose of carbon tetrachloride (CCl4). Phenotype analysis revealed the deleterious role of IRE1 in the liver in acute liver damage. Then, the authors determined the biological consequence of IRE1 deletion on the progression of experimental liver fibrosis. Consistently, IRE1 plays a promotive role in the CCl4-induced liver fibrosis mouse model since the results suggest that IRE1 knockout mice have reduced collagen deposition and AST/ALT activity (two major liver damage enzymes). To determine the potential mechanisms underlying the protective effects of IRE1 deficiency, the authors did mass spectrometry-based quantitative proteomics of liver samples from the experimental mice. Through bioinformatic analysis, they determined that protein and mRNA expression levels of P4HB decreased in IRE1 deficiency mice. Besides, in vitro and ex vivo studies suggested that IRE1 deficiency downregulated collagen. They further investigated the impact of IRE1 deficiency in liver steatosis in a mouse model fed a high-fat diet (HFD). Similarly, IRE1 deficiency suppressed liver steatosis and fibrosis in this mouse model. To clarify how IRE1 modulates collagen expression, the authors generated IRE1 knockout cell (KO) lines. They found that IRE1 KO cells may cause misfolding and intracellular accumulation of collagen inside the cells, reducing its secretion. Finally, they suggested the correlation of IRE1 associated gene (XBP1) and PH4B but not PH4A1 in human patients with non-alcoholic steatohepatitis (NASH). Overall, the current study may provide a potential molecular linkage between IRE1 and collagen in chronic liver disease progression. However, the inconsistent data presentation, poor data quality, and lack of suitable animal models make the current manuscript's impact on the field of chronic liver disease low.

Major concern:

1. The authors suggested that P450 levels are not changed in the liver samples of IRE1 KO mice compared to the wildtype (WT) ones. It has been demonstrated that IRE1 activation reduced P450 expression levels (PMID: 22291093). Can the authors explain the inconsistent findings?
2. Why did the authors challenge mice with CCl4 for 12 weeks? The CCl4-induced liver fibrosis model would have a severe fibrotic liver phenotype post the 8-week CCl4 challenge. Did the author check if IRE1 deficient mice have less liver injury than wildtype (WT) mice at early time points? Also, could you check the liver fibrosis phenotype using another typical liver fibrosis mouse model? The HFD-feeding mouse model would induce liver fibrosis after challenging mice for 50 weeks (PMID: 37270060).
3. What are the mouse numbers, age, and sex in the CCl4 studies? In Fig. 2, several figures have inconsistent mouse numbers for the data presentation.
4. There are many downregulated and upregulated target genes. What is the rationale for focusing on downregulated P4HB in IRE1 deficient mice, given that IRE1 has an RNase domain functioning in posttranscriptional regulation? In this case, how does IRE1 depletion upregulate P4HB mRNA expression?
5. The author suggested that TGFb1 would promote collagen generation, but targeting IRE1 would reduce P4HB and the TGFb1-mediated collagen generation. The Fig. 4D does not support the statement. However, the authors' data demonstrated that TGFb1 cannot promote IRE1 and P4HB expression (Fig. 4G). If TGFb1 cannot promote collagen generation through activating IRE1 and P4HB, how can targeting IRE1 reduce TGFb1-induced collagen (Fig. 4H)?
6. The authors stated that hepatic Ern1 deficiency suppresses the progression of liver steatosis. In fact, it has been reproducibly demonstrated that hepatic IRE1 deletion promotes hepatic steatosis progression (PMID: 21407177 and PMID: 29764990), contradicting the authors' findings. Besides, the authors did not deplete IRE1 specifically in the liver. Therefore, they made a misleading conclusion without solid evidence. The authors basically depleted IRE1 in the whole body of these experimental mice upon poly I:C injection. The authors must not conclude that hepatic Ern1 deficiency suppresses the progression of liver steatosis without considering the contributions of other organs in the phenotypes that they observed. Also, the HFD-treated mice would develop liver fibrosis 50 weeks post-feeding (PMID: 37270060). It is unclear what other treatments the authors used to accelerate the fibrotic liver phenotype, as shown in Fig. 5D. The authors should show body weight and liver weight over body weight in the result section. Besides, hepatic cholesterol, serum triglyceride, and cholesterol need to be measured in these mice to clarify how deleting IRE1 in the whole body can suppress hepatic steatosis, but liver-specific deletion of IRE1 promotes fatty liver. Without clarifying this issue, it is unclear how hepatic IRE1 deficiency can reduce steatosis and liver fibrosis.
7. The authors suggested that ablation of IRE1 expression increased the levels of intracellular GFP-collagen as compared with control cells (Fig. 6C). How did the authors quantify the results? It is not clear if KO really increased the intracellular collagen levels. As the authors showed in Fig. 6C, WT-NT and WT-GFP-collagen-untreated have no overlap of green fluorescence. However, KO-NT and KO-GFP-collagen-untreated still have an overlap of green fluorescence, indicating that some cells are not GFP-positive. In this case, how could authors conclude that IRE1-KO cells have a more than 2-fold increase of green fluorescence change compared to WT? Besides, Fig. 6F suggested that secreted collagens increased in KO cells, contradicting the authors' previous data in Fig. 2, 4, and 5. Why did you use U2OS, Hepa1-6, and Huh7 in these studies? Should the collagen be secreted by hepatic stellate cells?
8. In Fig. 7, the authors suggested that IRE1 KO promotes the levels of collagen inside cells using the whole cell lysate. Interestingly, they indicated that IRE1 deficiency suppressed TGFb1-induced collagen production using whole cell lysates (Fig. 4D). It is really confusing if IRE1 KO promotes or suppresses collagen production or secretion. Also, Fig. 7C did not support that IRE1-KO reduced collagen secretion. Besides, what cells did the authors use for these studies? Are they hepatic stellate cells?
9. It is interesting to see the positive correlation between XBP1 and P4HB mRNA expression. However, it is still unclear if IRE1 deficiency could downregulate P4HB mRNA expression, given its RNase function. Thus, it would be essential to determine how IRE1 regulates P4HB expression before analyzing the correlation using human datasets. Besides, Fig. 8D did not suggest that XBP1 expression levels are correlated with chronic liver disease progression, given that its correlation scores with AST and ALT are 0 and -0.01, respectively.

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 taking the time to read and comprehensively evaluate our manuscript. We are pleased that, overall, they recognize the quality of our data and that it supports our conclusions. We are grateful for their comments, insights and advice and have revised the manuscript accordingly as described in the point-by-point response below. We believe that the revised manuscript is substantially improved by some experimental additions, additional replicates, improved analysis and increased clarity. Some key enhancements are as follows:

Previously we had found increased expression of the WNT pathway following CHRDL2 treatment, using RNA seq. We have now demonstrated this experimentally using the cellular levels and localisation of β-catenin. Previously we had shown that overexpression of CHRDL2 increased resistance to common chemotherapy treatments, as well as irradiation in colorectal cell lines. We have now shown that cells surviving treatment show a further reduction SMAD1/5/8 phosphorylation indicating a selection for CHRLD2 high cells during the treatment. We have also demonstrated a decrease in chemotherapy sensitivity in intestinal organoids treated with secreted forms of CHRDL2.

1. Point-by-point description of the revisions

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

Reviewer #1

__Evidence, reproducibility and clarity __

Clarkson and Lewis present data suggesting that overexpression of Chordin like 2 (CHRDL2) can affect colorectal cancer cell responses to chemotherapy agents, possibly by modulating stem-cell like pathways. I have the following comments:

1. Fig. 1J-it is standard to show the images of cell migration-this is important here, given the modest effect of CHRDL2 overexpression here.

We have now included 3 replicate control and CHRDL2 overexpressing cell images in this figure panel to support the quantification in the graph.

Fig. 2A-the very small error bars for most of the data on the curves suggests these are n=1 experiments with multiple technical replicates to generate the error bars. Please clarify. The legend says n=3 with ANOVA analysis but no significance detected. Please clarify.

All experiments in this figure were done with 5 technical replicates per experiment, this was replicated at least three times to give n=3 biological replicates. The error bars represent the standard error of the mean of these 3 biological replicates as stated in the legend. Some data points showed very little data variation, hence the small error bars. Raw data is available if requested.

1. Fig. 2B-given the overlapping error bars here, how can there be a pWe have removed this representation of the data as it combined many different experiments with variable cell types and chemotherapeutics and it was difficult to carry out meaningful statistics. An overview of the data can be better seen in table form as shown in the revised figure 2B.

Fig. 2G-did the authors try to estimate the concentration of CHRDL2 in the conditioned medium? Which cell line was used to generate this CM?

Conditioned media was harvested from the matching transgenic cell lines with inducible CHRDL2. eg RKO cells were treated with media collected from doxycycline induced transgenic RKO cells whereas CaCO2 cells were treated with media from CaCO2 cells. The concentration of doxycycline was represented by ++ for 10ug/ml, the same notation we have used for directly induced cells treated with 10ug/ml dox.

We did not try to quantify the absolute concentration of CHRDL2 but we have shown the relative amount on a Western blot normalised with a ponceau stain (quantification now included in supplementary figure 1).

We have clarified our description of this experiment, inserting the following statement, "Conditioned media was harvested from corresponding cell lines with the inducible CHRDL2 transgene and the parental control cells. Induction of CHRDL2 to generate conditioned media was carried out using the same concentration and duration of doxycycline treatment as the cells in figure 2A. "

Fig. 5-what is the potential mechanism for gene expression changes in response to CHRDL2 overexpression? Is it all due to BMP inhibition? More mechanistic detail would be welcome here.

We have suggested other pathways involved in these functional effects based on our RNA seq data but at the moment it is not possible to say whether any changes are independent of BMP signaling. CHRDL2 is relatively understudied and as yet there is not much literature supporting BMP independent actions of CHRDL2. However, we have added some discussion and reference to an article suggesting interactions between CHRLD2 and YAP (Wang et al., 2022) including the following statement on page 17: "While the changes in BMP and WNT signaling shown in our GSEA analysis suggest that the effects of CHRDL2 in our system work directly through inhibition of BMP, it is not possible to rule out that some pathways are affected by BMP independent actions of CHRLD2. Indeed, Wang et al, suggest that CHRDL2 can directly alter phosphorylation and activity of YAP in gastric cancer cell lines, which merits further exploration (Wang et al., 2022)"

Significance

Unclear whether genetically engineered inducible overexpression has any real physiological significance but we all use cell models so this is OK.

---

Reviewer #2

__Evidence, reproducibility and clarity __

Summary: In the manuscript entitled "BMP antagonist CHRDL2 enhances the cancer stem-cell phenotype and increases chemotherapy resistance in Colorectal Cancer" the authors demonstrated that Chordin-like 2 (CHDRL2), a secreted BMP antagonist, promotes a chemo-resistant colorectal cancer stem cell phenotype through the inhibition of BMP signaling. The authors took advantage of both 2D engineered colorectal cancer (CRC) cells and healthy murine 3D organoid systems. Specifically, the authors showed a decreased proliferation rate and reduced clonogenic capability upon overexpression of CHRDL2 in established human colon cancer cell lines. Subsequently, they identified a chemo-resistant phenotype upon standard therapies (5FU, Oxaliplatin and Irinotecan) in CHDRL2 overexpressing cells by performing MTS assay. The authors showed that this chemo-resistant phenotype is associated with ATM and RAD21 activation, supporting an induction of DNA damage signaling pathway. Of note, the authors assessed that the exposure of 3D murine organoid to CHRDL2 resulted in a stem-like phenotype induction accompanied by a reduction of the differentiated counterpart. From RNA-seq data analysis emerged the upregulation of genes associated to stemness and DNA repair pathways in CHRDL2 overexpressing cells.

Major comments: 1. In the first paragraph of the result section authors assessed that "Colorectal adenocarcinoma cell lines were deliberately chosen to encompass a range of CHRDL2 expression levels and genetic mutations", without showing qRT-PCR or WB data on the differential expression levels of CHRDL2 in a panel of immortalized CRC cell lines. Authors should include these data to better support their choice.

*We have now included some qRT-PCR in supplementary figure 1 alongside a table of some of the key driver mutations in each cell line. Western blotting of these cells shows only a very low concentration of CHRDL2 protein. As shown in figure 1B in the control columns, no significant protein expression is observed in any line. *

In Figure 1F, authors described a reduction of cell proliferation in CRC cell lines expressing high levels of CHRDL2 only under low glucose conditions. Why did the authors perform the assay under these conditions? They should better argue this aspect and validated the role of CHRDL2 in metabolism rewiring by performing additional in vitro assays.

We have removed this aspect of the paper as it does not add significantly to our overall conclusions and we can clearly see the effects of CHRDL2 overexpression under standard growth conditions (Figure 1G).

The authors should evaluate the role of CHRDL2 in promoting a stem-like phenotype in human colon cancer stem cells freshly isolated from patients and characterized.

We would very much like to do experiments such as this but it is beyond the scope of this study and will be included in upcoming grant proposals.

In order to confirm the data obtained on 3D murine organoids system obtained from normal Intestinal Stem Cells, authors should investigate the stemness induction, driven by CHRDL2, also in human intestinal organoids.

Experiments using human intestinal organoids are currently planned and ethical approval applications and grant proposals are underway for future experiments of this nature.

The authors should evaluate the oncogenic role of CHRDL2, through the maintenance of stemness, by performing orthotopic or subcutaneous experiments in vivo model.

Similarly, this is not possible for this manuscript but is planned for the future alongside a transgenic mouse model of inducible CHRDL2 overexpression in the intestine.

BMPs proteins are part of a very broad protein family. In the introduction section, authors should indicate the specific BMP protein on which CHRDL2 exerts its inhibitory function. Moreover, they should have assessed BMP protein levels in CACO2, LS180, COLO320 and RKO cell lines.

We have clarified the interactions between CHRDL2 and specific BMPs in the introduction. We have not specifically assessed the BMP protein levels in our cells however we have now included an analysis of expression data from the Cancer Cell Line Encyclopedia in supplementary figure 1 C.

In first panel, the authors should quantify the secreted levels of CHRDL2 in the media of overexpressing CHRDL2 cell lines.

We have done this using the ponceau staining as a loading control and the results are displayed (supplementary figure 1).

In Figure 2D the authors should use the appropriate controls and describe this with more details in results section.

In this figure we have used Hoechst staining followed by FACs analysis to identify the cell cycle profile of our CHRDL2 treated cells. We have improved the description of this in the methods section. Appropriate controls for staining, both negative and positive, are used when setting up the analysis for this experiment. The cell cycle profile is calculated using the Novocyte in house software. We have now included the histogram plots in the main figure to clarify these data in figure 2D.

In Figure 3A, the authors should have performed the assay by choosing IC50.

*We attempted these experiments with the IC50 levels, however the high amount of cell death and frequency of apoptotic cells meant that clear images were difficult to obtain. We therefore reduced the concentrations and still had very measurable effects. *

In Supplementary Fig. 4A-B. the results are unclear. The control cell lines are already chemo resistant.

Again, we used IC25 levels of the drugs so that our cells were damaged but still live throughout the experiment. This has been explained on page 10.

The authors should review and add statistical analysis in both main and supplementary figures.

*We have now added additional details about statistical analysis throughout the figures, legend and main text, showing all significance levels as well as non-significance for each data set. * Minor comments: 1. The quality of immunofluorescence and WB images should be implemented, and in the immunofluorescence panels scale bars should be added.

We have added or improved scale bars on each immunofluorescence image. Western blot images have been improved.

In the graphical abstract authors reported that CHRDL2 overexpression increase WNT and EMT pathways, without performing any specific assay to demonstrate this. Authors should correct and graphically improve the graphical abstract.

*This is a good point and we have now carried out Beta-catenin immunofluorescence as a measure of WNT signaling on both our cancer cell lines - showing an increase in nuclear beta-catenin (figure 1J and K), and our organoids - showing an increase in overall levels and cytoplasmic staining (Figure 4 F). In terms of EMT markers we have carried out immunofluorescence on IQGAP1 (Figure 1K). IQGAP1 is significantly upregulated in CHRDL2 cells, reflecting its role in reduced cell adhesion and increased migration. This correlates with our data showing increased cellular migration as well as the increase in EMT related transcription in our RNAseq data. *

The term "significantly" in the discussion section is inappropriately referred to data showed in the histogram in Figure 1J. Moreover, in Figure 1Jthe authors should delete from the y-axis the term "corrected".

We have changed significantly to substantially

The term "significant" in discussion is inappropriately referred to BMI1 expression level if compared to the histogram in Figure 4G.

We have changed significantly to "a trend to increase"

In Figure 2C the authors should add the unit of measurement (fold over control) in the table.

We have done this

In Figure 4E the authors should add the figure legend reporting OLFM4 protein.

We have done this

The authors should include few sentences summarizing the findings at the end of each paragraph.

We have added short summaries at the start or end of each section to improve the flow of the results section.

Significance

General assessment: Overall, the work is aimed to elucidate the role of CHRDL2 already considered a poor prognosis biomarker involved in the promotion of CRC (PMID: 28009989), in promoting stem-like properties. The authors elucidated new additional insights into the molecular mechanisms regulating stemness phenotype induced by the BMP antagonist CHRDL2 in CRC. The authors include in the study a large amount of data, which only partially support their hypothesis. However, this manuscript lacks organization and coherence, making it challenging to follow and read. Numerous concerns need to be addressed, along with some sentences to rephrase in the result and discussion sections.

Advance: The manuscript reported some functional insights on the role of CHRDL2 in colorectal cancer, but additional data should be added to support authors 'conclusions.

Audience: The manuscript is suggested for basic research scientists.

__Reviewer #3 (Evidence, reproducibility and clarity (Required)):____ __ Summary BMP antagonist CHRDL2 enhances the cancer stem-cell phenotype and increases chemotherapy resistance in Colorectal Cancer Eloise Clarkson et al. The manuscript explored the function of CHRDL2, a BMP antagonist, on colorectal cancer (CRC). The authors found that CHRDL2 overexpression can enhance the survival of CRC cells during chemotherapy and irradiation treatment with elevated levels of stem-cell markers and reduced differentiation. Further RNA-seq analysis revealed that CHRDL2 increased the expression of stem-cell markers, WNT signaling and other well-established cancer-associated pathways. Overall, the manuscript is well-written and presented. I have some suggestions:

Major points:

1. The authors assert BMP antagonism was demonstrated by assessing levels of phosphorylated SMAD1/5 (Figure 1G). However, the immunoblotting assay only depicted P-SMAD 1/5 levels and B-ACTIN as internal control. It's suggested to include total-SMAD1/5 immunoblotting as an internal control to further support the claim of BMP antagonism.

The reviewer is correct that this is the best control. Western blotting has now been performed with total SMAD1 protein expression used as an internal control and this is shown in Figure 1D and Supplementary figure 1F

The authors argue that CHRDL2 overexpression reduced the proliferation of CRC cell lines, as evidenced by cell proliferation assays. However, from Figure 1E, the reduction in proliferation appears insignificant. It would be beneficial to perform one-way ANOVA tests on each time point for CHRDL2+ and CHRDL2++ with Control in Figure 1E to ascertain significance.

*We now have repeated this experiment to reduce variability and have also provided two-way ANOVA analysis between Control and CHRDL2+ and Control and CHRDL2++. One-way ANOVA at timepoint 96hr also provided with details in the figure legend. *

The findings indicating that overexpressing CHRDL2 can confer resistance to chemotherapy in CRC cells (Figure 2A-C) are noteworthy. To deepen the understanding of BMP signaling in cancer stemness and the molecular underpinning of CHRDL2 antagonism, additional western blot assays on P-SMAD1/5 with CHRDL2 overexpression and drug treatment are recommended.

*Western blotting of P-SMAD1/5 upon cells treated with IC50 5FU has now been performed in figure 2C (in the same experiment as the revised panels in figure 1D). The data suggest that CHRDL2 overexpressing cells able to survive chemotherapy have higher levels of P-SMAD1/5 reduction compared to that of untreated cells, strongly suggesting that chemotherapy treatment acts to select the cells with the highest CHRDL2 expression. We thank reviewer 3 for this suggested experiment and have included further discussion on this on page 9. *

The assertion that extrinsic CHRDL2 addition diminishes differentiation and enhances stem-cell numbers in an intestinal organoid model is intriguing. As BMP signaling inhibition contributes to intestinal cell stemness, incorporating additional layers for BMP antagonism of CHRDL2 on intestinal organoids through immunoblotting or real-time quantitative PCR for treated organoids would augment the conclusions.

As stated in the response to reviewer 2, we have investigated Beta-catenin in our organoids following CHRDL2 treatment using immunofluorescence and find that the levels are increased with the staining shifting from the membrane to the cytoplasm and nucleus (Figure 4F).

The authors claim CHRDL2 overexpression can decrease BMP signaling based on GSEA analysis (Figure 5E). However, the GSEA results did not demonstrate the downregulation of BMP signaling. Reanalysis of this GSEA analysis is warranted.

*We agree with this point and have changed the description of this result since the gene set covers both positive and negative regulators of the BMP pathway. We cannot conclusively say from this RNAseq data set that BMP signaling is "downregulated", however since SMAD phosphorylation is increased and nuclear beta-catenin is increased, overall we suggest that the changes we see are likely to represent the effects of decreased BMP signaling along with increased WNT signaling. *

Minor Points:

6.Provide the threshold/cutoff values chosen for differential expressed genes (DEGs) in CHRDL2+ and CHRDL2++ RNA-seq compared with control cells. Explain the minimal overlap between CHRDL2 LOW and CHRDL2 HIGH DEGs. Consider presenting all DEGs in CHRDL2 LOW and CHRDL2 HIGH compared with control cells in one gene expression heatmap for better visualization.

We have now provided the cutoff values for the DEGs in the legend for figure 5 (PThe minimal overlap of DEGs in the low and high expressing cells is an interesting point. We hypothesize that this may be related to the different effects of intermediate vs high levels of WNT signaling that occurs in colon cancer cells, frequently discussed in the literature as the "Just right hypothesis" (Lamlum et al. 1999, Albuquerque et al., 2002, Lewis et al., 2010). However, we haven't included this in the discussion as it merits further exploration. However, we have mainly focused on specific genes that are modified in both data sets, which are more likely to be the direct result of CHRDL2 modification. *

After DEGs analysis, perform Gene Ontology (GO) analysis on these DEGs to further investigate possible gene functions rather than selectively discussing some genes, enhancing understanding of CHRDL2 functions in CRC cells.

We have carried out this analysis using a variety of tools and have now included a Gene Ontology Panther analysis as supplementary figure 7. We have included a comment on this in the text on page 14 saying "Gene ontology analysis supports these findings with enrichment in biological processes such as cellular adhesion, apoptosis and differentiation. "

Conduct similar experiments in both 2D culture and organoid systems, if feasible, to provide more comprehensive insights into CHRDL2's oncogenic roles in CRC tumor progression.

*We have now performed chemotherapy treatment on our organoid systems, and have found that organoids with extrinsic CHRDL2 addition have a higher survival rate after chemotherapy compared to a control (Figure 4H and I). *

Label significance (*, **, ***, and n.s.) for every CRC cell line treated with CHRDL2 in Figure 2D, 2F, 2J, 4G, 5D, and 5F.

We have done this

Label the antibodies with different colors used for immunofluorescence in the figure text in Figure 4E.

We have done this

* * Include replicate dots for the Control group in the bar plots in Figure 1F and 2B.

We have done this

* * Add scale bars in Figure 3A and correct similar issues in other figures if applicable.

We have done this

* *13.Correct grammar and punctuation mistakes throughout the manuscript. For example:

We have done this and further proofread our revised manuscript

Page 7: "As seen in Figure 1J, CHRDL2 overexpression significantly increased the number of migrated cells (P *We have now added additional details about statistical analysis throughout the figures, legend and main text, showing all significance levels as well as non-significance for each data set. * Reviewer #3 (Significance (Required)):

The current study presents compelling evidence demonstrating that BMP signaling antagonist CHRDL2 enhances colon stem cell survival in colorectal cancer cell lines and organoid models. Further validation through CRC mouse models could offer invaluable insights into the clinical relevance and therapeutic implications of CHRDL2 in colorectal cancer.

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

BMP antagonist CHRDL2 enhances the cancer stem-cell phenotype and increases chemotherapy resistance in Colorectal Cancer Eloise Clarkson et al. The manuscript explored the function of CHRDL2, a BMP antagonist, on colorectal cancer (CRC). The authors found that CHRDL2 overexpression can enhance the survival of CRC cells during chemotherapy and irradiation treatment with elevated levels of stem-cell markers and reduced differentiation. Further RNA-seq analysis revealed that CHRDL2 increased the expression of stem-cell markers, WNT signaling and other well-established cancer-associated pathways. Overall, the manuscript is well-written and presented. I have some suggestions:

Major points:

1. The authors assert BMP antagonism was demonstrated by assessing levels of phosphorylated SMAD1/5 (Figure 1G). However, the immunoblotting assay only depicted P-SMAD 1/5 levels and B-ACTIN as internal control. It's suggested to include total-SMAD1/5 immunoblotting as an internal control to further support the claim of BMP antagonism.
2. The authors argue that CHRDL2 overexpression reduced the proliferation of CRC cell lines, as evidenced by cell proliferation assays. However, from Figure 1E, the reduction in proliferation appears insignificant. It would be beneficial to perform one-way ANOVA tests on each time point for CHRDL2+ and CHRDL2++ with Control in Figure 1E to ascertain significance.
3. The findings indicating that overexpressing CHRDL2 can confer resistance to chemotherapy in CRC cells (Figure 2A-C) are noteworthy. To deepen the understanding of BMP signaling in cancer stemness and the molecular underpinning of CHRDL2 antagonism, additional western blot assays on P-SMAD1/5 with CHRDL2 overexpression and drug treatment are recommended.
4. The assertion that extrinsic CHRDL2 addition diminishes differentiation and enhances stem-cell numbers in an intestinal organoid model is intriguing. As BMP signaling inhibition contributes to intestinal cell stemness, incorporating additional layers for BMP antagonism of CHRDL2 on intestinal organoids through immunoblotting or real-time quantitative PCR for treated organoids would augment the conclusions.
5. The authors claim CHRDL2 overexpression can decrease BMP signaling based on GSEA analysis (Figure 5E). However, the GSEA results did not demonstrate the downregulation of BMP signaling. Reanalysis of this GSEA analysis is warranted.

Minor Points:

6.Provide the threshold/cutoff values chosen for differential expressed genes (DEGs) in CHRDL2+ and CHRDL2++ RNA-seq compared with control cells. Explain the minimal overlap between CHRDL2 LOW and CHRDL2 HIGH DEGs. Consider presenting all DEGs in CHRDL2 LOW and CHRDL2 HIGH compared with control cells in one gene expression heatmap for better visualization. 7. After DEGs analysis, perform Gene Ontology (GO) analysis on these DEGs to further investigate possible gene functions rather than selectively discussing some genes, enhancing understanding of CHRDL2 functions in CRC cells. 8. Conduct similar experiments in both 2D culture and organoid systems, if feasible, to provide more comprehensive insights into CHRDL2's oncogenic roles in CRC tumor progression. 9. Label significance (, , **, and n.s.) for every CRC cell line treated with CHRDL2 in Figure 2D, 2F, 2J, 4G, 5D, and 5F. 10. Label the antibodies with different colors used for immunofluorescence in the figure text in Figure 4E. 11. Include replicate dots for the Control group in the bar plots in Figure 1F and 2B. 12. Add scale bars in Figure 3A and correct similar issues in other figures if applicable. 13.Correct grammar and punctuation mistakes throughout the manuscript. For example:

Page 7: "As seen in Figure 1J, CHRDL2 overexpression significantly increased the number of migrated cells (P

Page 7: "As seen in Figure 1J, CHRDL2 overexpression significantly increased the number of migrated cells (P

Page 7: "As seen in Figure 1J, CHRDL2 overexpression significantly increased the number of migrated cells (P < 0.0449)," suggesting increased migratory ability, a hallmark of cancer stem cells."

Page 8: "CHRDL2 overexpression resulted in an approximate twofold increase in IC50 values compared to control cells (P < 0.001)."

Page 10: "As seen in Figure 4B, upon the" should be corrected to "Figure 4B."

1. Specify the statistical methods or estimates used for determining statistical significance.

Significance

The current study presents compelling evidence demonstrating that BMP signaling antagonist CHRDL2 enhances colon stem cell survival in colorectal cancer cell lines and organoid models. Further validation through CRC mouse models could offer invaluable insights into the clinical relevance and therapeutic implications of CHRDL2 in colorectal cancer.

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 the manuscript entitled "BMP antagonist CHRDL2 enhances the cancer stem-cell phenotype and increases chemotherapy resistance in Colorectal Cancer" the authors demonstrated that Chordin-like 2 (CHDRL2), a secreted BMP antagonist, promotes a chemo-resistant colorectal cancer stem cell phenotype through the inhibition of BMP signaling. The authors took advantage of both 2D engineered colorectal cancer (CRC) cells and healthy murine 3D organoid systems. Specifically, the authors showed a decreased proliferation rate and reduced clonogenic capability upon overexpression of CHRDL2 in established human colon cancer cell lines. Subsequently, they identified a chemo-resistant phenotype upon standard therapies (5FU, Oxaliplatin and Irinotecan) in CHDRL2 overexpressing cells by performing MTS assay. The authors showed that this chemo-resistant phenotype is associated with ATM and RAD21 activation, supporting an induction of DNA damage signaling pathway. Of note, the authors assessed that the exposure of 3D murine organoid to CHRDL2 resulted in a stem-like phenotype induction accompanied by a reduction of the differentiated counterpart. From RNA-seq data analysis emerged the upregulation of genes associated to stemness and DNA repair pathways in CHRDL2 overexpressing cells.

Major comments:

1. In the first paragraph of the result section authors assessed that "Colorectal adenocarcinoma cell lines were deliberately chosen to encompass a range of CHRDL2 expression levels and genetic mutations", without showing qRT-PCR or WB data on the differential expression levels of CHRDL2 in a panel of immortalized CRC cell lines. Authors should include these data to better support their choice.
2. In Figure 1F, authors described a reduction of cell proliferation in CRC cell lines expressing high levels of CHRDL2 only under low glucose conditions. Why did the authors perform the assay under these conditions? They should better argue this aspect and validated the role of CHRDL2 in metabolism rewiring by performing additional in vitro assays.
3. The authors should evaluate the role of CHRDL2 in promoting a stem-like phenotype in human colon cancer stem cells freshly isolated from patients and characterized.
4. In order to confirm the data obtained on 3D murine organoids system obtained from normal Intestinal Stem Cells, authors should investigate the stemness induction, driven by CHRDL2, also in human intestinal organoids.
5. The authors should evaluate the oncogenic role of CHRDL2, through the maintenance of stemness, by performing orthotopic or subcutaneous experiments in vivo model.
6. BMPs proteins are part of a very broad protein family. In the introduction section, authors should indicate the specific BMP protein on which CHRDL2 exerts its inhibitory function. Moreover, they should have assessed BMP protein levels in CACO2, LS180, COLO320 and RKO cell lines.
7. In first panel, the authors should quantify the secreted levels of CHRDL2 in the media of overexpressing CHRDL2 cell lines.
8. In Figure 2D the authors should use the appropriate controls and describe this with more details in results section.
9. In Figure 3A, the authors should have performed the assay by choosing IC50.
10. In Supplementary Fig. 4A-B. the results are unclear. The control cell lines are already chemoresistant.
11. The authors should review and add statistical analysis in both main and supplementary figures.

Minor comments:

1. The quality of immunofluorescence and WB images should be implemented, and in the immunofluorescence panels scale bars should be added.
2. In the graphical abstract authors reported that CHRDL2 overexpression increase WNT and EMT pathways, without performing any specific assay to demonstrate this. Authors should correct and graphically improve the graphical abstract.
3. The term "significantly" in the discussion section is inappropriately referred to data showed in the histogram in Figure 1J. Moreover, in Figure 1Jthe authors should delete from the y-axis the term "corrected".
4. The term "significant" in discussion is inappropriately referred to BMI1 expression level if compared to the histogram in Figure 4G.
5. In Figure 2C the authors should add the unit of measurement (fold over control) in the table.
6. In Figure 4E the authors should add the figure legend reporting OLFM4 protein.
7. The authors should include few sentences summarizing the findings at the end of each paragraph.

Significance

General assessment:

Overall, the work is aimed to elucidate the role of CHRDL2 already considered a poor prognosis biomarker involved in the promotion of CRC (PMID: 28009989), in promoting stem-like properties. The authors elucidated new additional insights into the molecular mechanisms regulating stemness phenotype induced by the BMP antagonist CHRDL2 in CRC. The authors include in the study a large amount of data, which only partially support their hypothesis. However, this manuscript lacks organization and coherence, making it challenging to follow and read. Numerous concerns need to be addressed, along with some sentences to rephrase in the result and discussion sections.

Advance: The manuscript reported some functional insights on the role of CHRDL2 in colorectal cancer, but additional data should be added to support authors 'conclusions.

Audience: The manuscript is suggested for basic research scientists.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Clarkson and Lewis present data suggesting that overexpression of Chordin like 2 (CHRDL2) can affect colorectal cancer cell responses to chemotherapy agents, possibly by modulating stem-cell like pathways. I have the following comments:

1. Fig. 1J-it is standard to show the images of cell migration-this is important here, given the modest effect of CHRDL2 overexpression here.
2. Fig. 2A-the very small error bars for most of the data on the curves suggests these are n=1 experiments with multiple technical replicates to generate the error bars. Please clarify. The legend says n=3 with ANOVA analysis but no significance detected. Please clarify.
3. Fig. 2B-given the overlapping error bars here, how can there be a p<0.01 between the groups?
4. Fig. 2G-did the authors try to estimate the concentration of CHRDL2 in the conditioned medium? Which cell line was used to generate this CM?
5. Fig. 5-what is the potential mechanism for gene expression changes in response to CHRDL2 overexpression? Is it all due to BMP inhibition? More mechanistic detail would be welcome here.

Significance

Unclear whether genetically engineered inducible overexpression has any real physiological significance but we all use cell models so this is OK.

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

The authors investigate the role of the AMP-activated protein kinase (AMPK) in mitochondrial dysfunction. Using HEK293T cells as model system they induce expression of either the wild-type or a dominant-negative variant of the mitochondrial DNA polymerase, which results in depletion of mtDNA and a decreased mitochondrial membrane potential. Using different time points of Pol induction they correlate the mitochondrial defects with activation of AMPK and make the interesting observation that only the mitochondrial associated fraction of AMPK becomes activated at an early stage of mitochondrial dysfunction. The authors then apply a known AMP activator (A-769662) and assess its impact on mtDNA levels and respiratory chain subunit steady state levels. Finally, they compare the findings using the HEK cells with patient derived fibroblasts, which show the same response to the activator.

Regarding the so far provided data I have the following concerns:

• I do not agree with the statement that the mechanisms in the HEK and patient cells are different. First of all, there is no analysis of the mechanism in the HEK cells nor the patient fibroblasts. Secondly, the control cell line (Pol WT overexpression) is also showing a decrease in mitochondrial membrane potential, but no change in mtDNA - which is fully reflecting the observation of the patient cells.
• Figure 1H-J: the authors claim that CIV activity is decreasing. However, CIV is virtually absent in these samples and therefore the statement that CIV activity is decreased is not correct.
• It is not clear why the authors used the dominant negative D1135A variant in the HEK system and not the most common dominant patient mutations (of which they in the end use the patient fibroblasts).
• Supplemental Figure S1A/B: Which time point of induction is shown here?
• The rho zero cell line and the control UV treatment are not described in the materials and methods section.

Significance

The observation that mitochondrial-associated AMPK reacts much earlier than the global AMPK pool to the mitochondrial dysfunction is interesting. The other observations described in the manuscript were rather to be expected given previous publications. Overall, the study primarily provides descriptive findings, which, in my view, seem preliminary at this stage and requires significant revisions for it to be truly valuable to the scientific community.

At least some molecular insight into the regulation of the different cellular AMPK pools or detailed analysis on how the activator impacts on mtDNA or the general mechanism that results in stabilization of the mitochondrial membrane potential are necessary to provide sufficient novel findings for publication. This additional analysis would be necessary to strengthen the study's conclusions and broaden its relevance to a larger readership.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

The authors aimed to investigate the functional consequences of AMPK agonist, A-769662, in improving the cellular energetic response to mitochondrial DNA depletion. AMP kinase plays a crucial role in switching the metabolic programming in the cells upon energetic stresses. It drives the activation of transcription factors of the mitochondrial genome as well as nuclear genes important for mitochondrial biogenesis (Herzig and Shaw., 2018, Bonekamp et al., 2021). Previous studies have shown that triggering the AMPK cascade has positive outcomes in mitochondrial disease models (Peralta et a., 2016, Moore et al., 2020). However, the mechanistic basis of their impact on mitochondrial function, especially mtDNA is not known.

Brief Summary:

In this study, the authors characterize the dynamics of mitochondrial dysfunction in response to severe mtDNA depletion, using a cell model and report that A-769662, a non-AMP mimetic AMPK agonist maintains cellular energy homeostasis by stimulating the AMPK cascade to restore membrane potential in mtDNA depleted cells. The positive effect was observed only in patient-derived, mtDNA-depleted cells and absent in control cells, suggesting that A-769662 mediates mitochondrial activity and cellular function via partially different mechanisms.

Major Results:

• The authors used a previously established inducible cell line to transiently express a dominant negative mutant of the mitochondrial DNA polymerase, Poly, as a model system for mtDNA depletion. Compared to the wildtype cells, mtDNA copy number, membrane potential, assembly and levels of respiratory chain complexes I, III and IV were severely reduced in the PolyD1135A cells (Figure 1A-K).
• The authors checked the relative ratio of AMP/ADP and ATP in the PolyD1135A cells and found that ADP levels were elevated, as expected (Figure 2A). Consequently, the levels of phosphorylated AMPK in/at (?) the mitochondria were also increased, in good correlation with the depletion of mtDNA levels and respiratory complexes after 3 days of induction (Figure 2B-E). Activation of the pAMPK pathway was further confirmed by the increased levels of AMPK substrate ACC. However, metabolite levels and cell cycle profiles are mildly altered in the PolyD1135A cells, suggesting that the activation of mitochondrial AMPK is an early response to mtDNA dysfunction.
• Chemical activation of AMPK by agonist A-769662 had a sustained positive effect on the membrane potential in both induced and uninduced cells. This was specific to AMPK signalling as evidenced by no change in membrane potential in cells transfected with AMPK siRNA (Figure 3A-C). However, the A-769662 treatment was insufficient to rescue the proliferation defects in the PolyD1135A mutant cells (Figure 3D), suggesting that the growth defect is independent of AMPK activation.
• A-769662 treatment for 48 or 72h in PolyD1135A improved mtDNA copy number and respiratory complex subunit expression, while having no effect on control cells (Figure 4A-G). Upon A-769662 treatment, patient-derived mutant cell lines showed no change in mtDNA levels (Figure 4H). Interestingly, membrane potential was enhanced in both control and patient cell lines (Figure 4I), suggesting an overall activation of mitochondrial function and not a specific response to restoring mtDNA.

Taken together the manuscript by Carvalho et al., proposes a stimulatory effect of AMP agonist, A-769662 on mtDNA depletion. However, since the model is not consistent in different model systems, the authors should provide stronger evidence for the utility of A-769662 as a therapeutic possibility for mtDNA disorders. Moreover, some mechanistical molecular insights into these largely descriptive results must be presented in a revised version. What drives AMPK localization to mitochondria? Is this kinase imported? Is it just attached? What regulates the distribution of AMPK between their different locations?

Significance

Major points:

• The effect of A-769662 on mtDNA levels in the FlipIn-TRex cell line, harboring a severe mutation and clinically-relevant patient-derived cell lines are not comparable (Figure4A and 4H), suggesting that the amelioration of mitochondrial defect is probably dependent on the extent of mtDNA damage. The FlipIn system shows a loss of almost 85% of mtDNA on day 3 of induction (Figure 1A) whereas the patient-derived cells retain almost 60% of the mtDNA (Figure 4H). The authors argue that the two systems are not comparable. A good control would be to check an inducible FlipIn-TRex cell line with the same patient-derived mutations or alter the induction system, with a shorter induction time or reduced concentration of doxycycline, to have comparable levels of mtDNA depletion.
• A more thorough investigation of A-769662 in different cell models of mtDNA dysfunction, possibly different disease-specific mutations in Poly, or cell types which contain AMPK complexes with -subunits (irresponsive to A-769662 stimulation) will be needed to claim its therapeutic merit.
• To substantiate the claim that the downstream effect of A-769662 treatment is dependent on the metabolic context of the cell, it would be necessary to test the levels of crucial metabolites like mtDNA transcripts, ATP, NADH, in addition to dNTPs tested in the study. It would also help to compare the levels of these metabolites in of PolyD1135A cells, grown in galactose medium.
• In Figure 2C, TFAM is nearly absent in the mitochondrial fraction of PolyD1135A cells, since Poly dysfunction triggers a reduced expression of the transcription machinery. Considering the mtDNA level upregulation is mild after if A-769662 treatment (about 8%, Figure 4A), it would be worthwhile to check if A-769662 could alter transcript levels of mtDNA and/or expression of mitochondrial transcription factors TFAM and TF2B.

Minor points:

• The data from Figure 1 conclusively show defects in mtDNA. It would be necessary to compare OCR, ROS and cellular ATP levels to demonstrate the extent of mitochondrial dysfunction in this model due to mtDNA depletion.
• To further delineate the differences in mitochondrial bioenergetics in PolyD1135A , cells should be grown in galactose media and probed for respiratory fitness.
• The study uses 100uM of A-769662 in the cell assays (Figures 3 and 4) and 200uM to test the activation of AMPK substrate ACC (Figure S2A-C). The authors should explain if a dose-dependent study was performed and how the concentration of A-769662 to be used was determined.

A-769662 is a known AMPK activator with possible therapeutic effects in metabolic disorders, type II diabetes (Cool et al., 2006, Görransson et al., 2007): However, due to the wide range of effects it may have, it would be necessary to get to the molecular basis of how A-769662 targets mtDNA depletion. This study is a nice starting point to further probe into the benefits of A-769662, however it is not (yet) conclusive and definitely needs the clarification of the underlying molecular mechanisms.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

The manuscript by Carvalho and colleagues addresses how an increase in AMPK signaling, triggered by the agonist A769662, contributes to ameliorate cellular phenotypes caused by mtDNA depletion. The authors build on previously established cellular models of mtDNA depletion, and use subcellular fractions in an attempt to distinguish pools of AMPK associated with different organelles. The data is very clearly presented, and the western blot data shown is of high quality. The authors also use cells from two patients with mtDNA depletion, in which they stimulate AMPK using A769662. The overall conclusion is that AMPK stimulation in the cell models with mtDNA depletion is advantageous to minimize the disease-related phenotypes.

There is one fundamental weakness: there are several intracellular AMPK pools described, the major ones being in the cytoplasm, or associated with mitochondria, or with lysosomes, or in the nucleus. However, and importantly, while the authors convincingly show that there is no cytoplasm in their mitochondrial fractions, they do not control for the presence of lysosomal proteins. For the conclusions to be valid, it is absolutely essential to distinguish the effects of mitochondria-associated AMPK from lysosome-associated AMPK, because they may have different effectors and because they are activated by different mechanisms. Furthermore, the authors do not show what happens to AMPK in the patient cells, and this would be very informative. Finally, it would be important to put these findings in the context of other studies on AMPK signaling in response to other mitochondrial perturbations and that find AMPK to be down-regulated in chronic mitochondrial respiratory chain deficiency, as well as to more carefully reference the different intracellular AMPK pools. These studies might help to strengthen the discussion, given that the authors find AMPK signaling to be increased but show that increasing it pharmacologically has benefits - is A769662 activating different pools of AMPK?

Significance

The manuscript addresses an important question: how does the metabolic hub AMPK contribute to the phenotypes observed in chronic mitochondrial malfunction. Most of the studies on AMPK and mitochondrial malfunction focus on acute effects, which are not a model for mitochondrial diseases. Therefore, focusing on chronic effects is important to understand the long term consequences on AMPK signaling and its downstream signaling. Furthermore, if AMPK reactivation is beneficial (as this study proposes and other studies have also shown in other models of chronic mitochondrial malfunction), then this can become a new important therapeutic strategy.

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 thank the reviewers for the overwhelmingly positive feedback on our initial submission. *

• *

Reviewer 1: “Overall this is a very simple, although extensive and excellent, study analyzing a wide range of data form (sic) patients with bronchoalveolar lavage and epithelial cell samples, human epithelial cell cultures after infection with a range of respiratory viruses as well as the development of a 3D in silico analysis of potential protease and serpin interactions.”

Reviewer 2: “Overall, this paper will be interesting to a specialized audience that is interested in SERPIN function. The SERPIN expression data during viral infection, discovery of CTSL as a target of PAI-1, and evidence that PAI-1 can inhibit SARS-CoV-2 replication, will move that field forward.”

No new experiments were requested, but some were either suggested or explicitly marked optional. We thus focused the initial 4-week-revision on performing new experiments aimed to enhance our study’s significance and impact by validating the heart of our study: the data from the in-silico docking screen.

Reviewer 1: “Further analysis of the detected proteases that are reported here to bind to PAI-1 would be of great interest.”

Reviewer 2: “It is exciting that they predicted CTSL as a target of PAI-1, but it is not obvious that this is a generalizable approach without further hypothesis testing.”

• *

Thus, we performed additional protease activity assays to validate SERPIN-protease pairs from the in-silico-screen. The results elevate our study above the proof-of-principle state. Beyond their described roles in infectious disease, the two SERPINs that are now tested in more detail (SERPINB2, plasminogen activator inhibitor 2 and SERPINE1, plasminogen activator inhibitor 1) also play critical roles in cancer, neurodegeneration, aging, and cardiovascular disease. (Bouton et al., EMBO Mol Med 2023 Vol. 15 No. 6; Zhang et al., EMBO Mol Med 2023 Vol. 15 No. 9; Bode et al., EMBO Journal 1986 Vol. 5 No. 10; Uhl et al., EMBO Mol Med 2021 Vol. 13 No. 6). Given these multifaceted roles, we anticipate that our discovery of new SERPIN-protease binders and non-binders will advance various areas of human disease driven by SERPIN biology.

2. Description of the planned revisions

*We believe that the planned revisions outlined below can be finalized within 1-2 months. *

• *

Reviewer 1: “Further analysis of the detected proteases that are reported here to bind to PAI-1 would be of great interest.”

*At the time of the 30-day revision, recombinant SERPINB1 (LEI) and SERPING1 (C1-INH) were still backordered with an estimated shipping date the week of resubmission. Once delivered, we will perform protease activity assays with LEI or C1-INH and uPA, TMPRSS2, Cathepsin L, and Cathepsin B to bring up the number of validated SERPIN:protease interactions from 8 to 16. *

---

Reviewer 2, major points:

9) Further, to strengthen the conclusions of this data the authors should include additional controls. One would be to use trixplanin as they did in previous panels to show that PAI-1 is necessary. Further, if the authors generate mutant PAI-1 that is unable to inhibit TMPRSS2 (see comment 11 below), they could also use this as a control to show the necessity of functional PAI-1.

*We agree that these optional experiments would increase rigor. We generated plasmids containing mutated PAI-1 that we can use in spike cleavage assays as suggested and can perform this experiment. *

*We can unfortunately not use triplaxinin on cells, as our preliminary data show that it is quite cytotoxic at the concentrations required to inhibit PAI-1. *

10) For Figures 4I-J, is it possible to also blot for S1 cleavage? If possible, this optional data would be helpful to understand whether the entire cleavage process is disrupted or only S2 to S2' especially given that visually it appears as if the full length is more depleted in the condition with PAI-1 suggesting that it is cleaving spike better into S1 and S2. Could also suggest that the dynamics of cleavage are shifted rather than impaired?

*S1 cleavage is shown indirectly in (now) Figure 5f,g – the main product of S1 cleavage is the fragment annotated as S2. Due to high levels of endogenous furin in BHK cells, this cleavage always occurs in this experimental setting. It is true that we have not shown the effects of PAI-1 inhibits on S1 cleavage– we can include that control in the above optional experiment (point 9). We do not expect PAI-1 to have an effect on S1 cleavage, as it is well-established that it does not inhibit furin. *

• *

Reviewer 2, minor points

7) As a supplemental figure, can the authors show a complex blot (similar to Figure 4F) for CTSB to show that is does not complex with PAI-1.

*Purified active CTSB is not commercially available, but we can attempt to perform gel shift analysis on the samples from the in vitro protease assay. Due to the presence of proteinaceous substrate in these samples, we have previously observed lot of background on the gel, but we can attempt it and include it in a revised manuscript if reviewer/editor find it useful. *

*

• *

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

Reviewer 1:

Summary:

---

The authors do not reference the prior work which has examined cross class serpins, viral and mammalian, - this should be noted as alternative protease targets are known.

*Thank you – please see our response in point 1 below. *

• *

The bronchiolar lavage analysis is excellent but cannot differentiate epithelial cell and associated immune cells and their roles in the response.

We apologize for not making this clear – scRNAseq can indeed differentiate between different cell types using cell-type specific expression markers for each individual cell. This is how we were able to retrieve expression data specific for individual cell types. The reviewer is correct in that an expression analysis cannot show the role of individual cell types in the antiviral response. However, as epithelial cells are the primary cell type infected by SARS-CoV-2 gene expression patterns in these epithelial cells may show us cell-intrinsic effectors that are upregulated in response to viral infection. We now revised language in this paragraph to make this clearer (lines 156-162).

• *

PAI-1 does not seem to be present in the bronchoalveolar lavage samples.

We do not know if PAI-1 is present, as we did not analyze protein levels in these samples. The gene expression data suggests that SERPINE1, the gene encoding PAI-1, is expressed at low levels in the epithelial cell subset at baseline, and expressed at slightly higher levels in individuals with severe COVID-19 (Figure 1c). This is consistent with previously published data on SERPINE1 gene expression upon viral infection (Dittmann et al., Cell, 2015).

• *

Further discussion of prior work with cross class serpins and also the limitations of the in-silico analyses and the lavage specimens should be provided.

Prior work evidencing other cross class serpin protease targets as well as limitations related to the analyses as discussed in the critiques above should be noted and the abstract and title could better describe and define the studies as performed.

*Thank you for raising these important points. For cross-class SERPINs, please see our response to point 1. The limitations of in silico analyses are discussed in-depth in a paragraph of the discussion (lines 608-631). We also discuss discrepancies observed between SERPIN expression in lavage specimens and in HAEC – please advise whether this is sufficient or needs bolstering (lines 546-564). We revised both title and abstract to better describe and define the studies as performed. *

• *

These correlations between changes in serpin and protease expression with viral infections and potential new interactions for serpins with previously non identified proteases is of clear interest. This shows an excellent correlation but as with big data sets this does not provide a true cause and effect - rather providing new potential directions for analysis of these interactions in viral infections in lung epithelium and this is valuable as a basis for ongoing studies.

*We are in agreement with the lack of cause and effect –to our knowledge, we make no such claim from the gene expression data. We state that we used the expression data to guide the selection of SERPINs for our in-silico screen (lines 317-319). We then validated select data from our in-silico screen in vitro, which provides true cause and effect (Figures 4 and 5). *

• *

Major points:

• Cross class serpin interactions are known and have been reported for at least two viral serpins Serp-1 and CrmA - both of which bind cysteine proteases as well as serine proteases as well as the mammalian SCCA serpins. *Thank you for bringing these two examples to our attention – we added them to the discussion (lines 648-652). We now also emphasized throughout the manuscript that the novelty of our findings is in PAI-1 cross-class inhibition, specifically, which has not been previously reported despite PAI-1 being an extremely well-studied SERPIN. *

*We also would like to mention that in our opinion the scientific advance provided by our in-silico screen is not limited to the identification of new PAI-1 targets, but also provides a birds-eye view on SERPIN selectivity in a specific proteolytic landscape. For example, to our knowledge, it was unknown that SERPINB1 is promiscuous and that SERPINC1 is more selective, which our docking predicted. It was unknown that most TMPRSSs are unlikely SERPIN targets and that those that are SERPIN targets need to be in their active state to bind. The unsupervised clustering in Figure 4b (both on the SERPIN and on the protease side) predicts such unrecognized patterns in SERPIN selectivity. *

• *

• The protease targets are reported to vary when interacting with glycosaminoglycans such as heparan sulfate - PAI-1 inhibits thrombin in the presence of heparin - thus while a canonical serpin suicide inhibition is considered specific - it can vary. This is noted in the discussion Yes, we agree (lines 608-610).

• What is the potential impact of the noted interactions of PAI-1 with other proteases such as cathepsin - PAI-1 is considered to have predominately extracellular functions, but prior work indicates internalization of PAI-1 when bound to the uPA/uPAR complex with alterations in intra cellular activation This is correct and PAI-1 internalization is cited and mentioned in discussion (lines 620-624). We now also added data on SARS-CoV-2 variant Omicron BA.1, which predominantly uses CTSL for maturation, and we show is also inhibited by PAI-1 (new Figure 5).

• *

• This is supported by basic in vivo and in vitro serpin and protease interactions that are demonstrated confirming in silico analyses, eg. gel shift analyses or even Mass spectrometry analysis particularly for PAI-1 Yes, this is the data shown in Figure 4. We now also added protease activity assays for other SERPIN-protease pairs, thereby elevating our study above the proof-of-principle state. *This was also a suggestion raised by reviewer 2. *

• *

• Per the authors "To date, three SERPINs have been studied in the context of innate antiviral defense: PAI- 1 (encoded by SERPINE1) against influenza viruses encoding hemagglutinin H1 and SARS-CoV-2, by impeding the proteolytic maturation of H1 or spike, respectively19,20; alpha-1-antitrypsin (encoded by SERPINA1) and antithrombin (encoded by SERPINC1) against SARS-CoV-2, likely through the inhibition of TMPRSS2, by reducing maturation of spike, although direct inhibition of TMPRSS2 by either SERPIN was not shown". This is partially complete however other serpins such as C1Inh and one virus derived serpin that have been analyzed for efficacy in treating SARS Thank you for mentioning this, we added the information to the introduction *(lines 106-111). *

• *

• While TMPRSS2 is indeed a serine protease - Beneficial effects of some serpins may be due to modulation of the immune response as opposed to selective anti-viral responses. The immune / cytokine storm and coagulopathies (with clotting and even hemorrhage) seen in the excess inflammatory response that causes respiratory vascular leak and severe viral sepsis. PAI-1 targets tPA and uPA - uPA has marked proinflammatory actions when bound to the uPA receptor (uPAR) and can activate growth factors and MMPs which can enhance immune cell invasion - PAI-1 binds to the uPA / uPAR complex which can thus also alter inflammatory cell responses and cell activation when internalized. Thank you for bringing up this point. The role of SERPINs in inflammation and anti-viral immune responses is indeed well-established. While our study focuses on cell-intrinsic antiviral roles of SERPINs by shutting down pro-viral proteases, which is much less established, we now added this to the results section for clarification (line 153-156).

• The RCL does in general incorporate P4 to P4' but can vary from this specific P4 to P4' sequence *Yes, we agree. *

• *

• How accurately does in silico protease serpin analysis predict real interactions? - this should be discussed as HADDOCK may have some limitations - This is outside my field of expertise We added an in-depth paragraph on how HADDOCK operates to the results section to help readers not familiar with the technique (lines 248-290). We discuss the limitations of HADDOCK in depth in the discussion section *(lines 608-631)– please advise whether this needs additional information. *

*We argue that, with the limitations stated in the discussion, our in-silico method predicts interactions well, as shown by the correct prediction of known binders and non-binders, as well as of new binders (PAI-1 to *active* TMPRSS2 and CTSL) and a new non-binder (CTSB). *

*As with any screening method, results require validation via another method, which we performed for select SERPINs and proteases. In fact, the revised manuscript now features in vitro validation of 8 SERPIN-protease pairs (Figure 4a, b), with 8 additional planned (see “planned revisions” section). *

• *

• The data from a published study examining bronchoalveolar lavage fluid single cell transcriptional analysis from patients with and without COVID - mild and severe - and with comparison to patients without COVID does demonstrate altered protease and serpin activity - but does not indicate specific interactions *We agree with this statement partially. We disagree in that the data does not demonstrate altered protease and SERPIN activity; it instead demonstrates changes in gene expression levels. We agree in that this does indeed not indicate specific interactions. *

• What is the significance for changes in gene expression in epithelial cells versus macrophage T and B cells looks - This looks like a small change like a small change in the mean values Figure 1b *We performed additional statistical analyses on the Figure 1 data – please refer to Reviewer 2 point 1. *

• *

• Of interest - is the brocholaveolar lavage fluid likely to contain both epithelial cells as well as immune response macrophage, T cells and NK cells etc - one assumes single cells were identified and isolated- Is this defined? Apologies if this was unclear. Yes, the BALF contains all of these cell types. We now added some sentences to the results section explaining scRNAseq and analyses in more detail *(lines 147-162). *

• *

• The known previously reported target proteases for PAI-1 should be noted Agreed; it is noted in the results section where we first speak about PAI-1 target specificity (line 379-382).

SERPINE1 is not noted in figure 1 - this is PAI-1 - but is seen in the HAEC infection model data

SERPINE1 is indeed not significantly upregulated in Figure 1, but is significantly upregulated in HAEC upon infection with Reovirus and parainfluenzavirus 3, and upon IFN stimulation (new Supplemental Tables S1 and S2). The possible reasons for discrepancies between the BALF and HAEC data are discussed in lines 546-564.

• “To overcome this limitation, we developed a computational method to predict 3D interactions between SERPINs and proteases, simulating the binding process depicted in Supplemental Figure 1a. Specifically, we employed High Ambiguity Driven protein- protein Docking (HADDOCK), a tool that predicts complex structures, integrating experimental and computational data35,36." This analysis looks to be extensive however this is a correlation - not a true analysis of cause and effect. We agree on the first point – to our knowledge, our study provides the most extensive SERPIN target discovery process (testing 480 SERPIN-protease interactions). We disagree on the point that our results provide a mere correlation. If you will, we performed a computer-modeled interaction experiment that yields predicted binding energies between each SERPIN with each tested protease. We added a paragraph on how HADDOCK operates to the results section to help readers unfamiliar with the technique. As with any screening method, results need to be validated with another method, which we did for select SERPINs and proteases (Figure 4a, b). This does however have the potential to identify significant interactions We certainly agree on this point. * In future it might be of interest to assess PAI-1 given to infected cultures to assess viral replication and titers or perhaps examine a knock out cell model? We did exactly the former in Figure 4 (now 5). *

• *

• As PAI-1 was identified as having new cathepsin protease binding in addition to TMPRSS2 - the authors did demonstrate inhibition of the new targets on fluorometric analysis and also demonstrated interaction by gel shift - This is excellent *Thank you. *

• *

• The title and the abstract could be better written and more clearly indicate the extent of the analyses performed and the discovery of alternate protease targets for PAI-1 We modified both title and abstract.

• *

• Was the SARS CoV2 lung epithelial cell culture analysis performed in BSL3? Yes. All SARS-CoV-2 infection experiments were performed in a BSL3 environment. We added this information throughout the Methods section, and also generated a new Methods section on Biohazards (lines 779-797).

__Minor critiques __

1) Results section heading "SERPINs are differentially expressed individuals with COVID-19 and in response to respiratory virus infection in a model of the human airway epithelium." The word in needs to be inserted between expressed and individuals *Thank you for catching this – we fixed the sentence (lines 128-129). *

*

• *

Reviewer 2:

Major points:

---

1) The rigor of the results presented in Figure 1 are unclear. For the COVID-19 analyses (Figure 1), only one dataset is used, and no statistical analyses are performed to determine to what degree any of the changes they observe are significant relative to variation in the dataset. This makes it difficult to determine how much can be extrapolated from these data.

We agree that performing statistics on the BALF dataset would be ideal. However, the BALF contains only two non-infected individuals (intubated gun-shot victims), limiting our possibilities for statistical analysis.

*For Figure 1b, we overcame this limitation by adding statistical analysis of upregulated expression values between cell types (i.e. by analyzing differences of upregulation of given SERPIN in epithelial cells compared to macrophages; Supplemental Table S1). We also performed statistical analysis on upregulation for individual SERPINs compared to housekeeping gene B2M (Supplemental Table S1). This revealed that SERPINs statistically significantly upregulated in severe COVID-19 in most cell types, including epithelial cells, in which SERPIN function has not been broadly studied. Upregulation was not statistically significant in mild COVID-19 samples, likely due to the n=3 (as compared to n=6 in the severe COVID-19 group). *

*As for analysis of Figure 1c, we could theoretically perform analysis of differential levels between mild and severe COVID-19, but this is not the question we are trying to answer. The question is whether epithelial cells express SERPINs and proteases, and whether there is an upregulation of either in infected individuals. We now state the limitation of lacking statistical power in the figure legend and the text (lines 176-177). *

2) Similarly, the qPCR data presented in Figure 2 are presented with no statistical analyses. Results should not only be presented with fold change but also p-values that are adjusted for multiple testing.

*We now present p-values in Supplemental Table S2. Of note, data obtained with the experimental system of polarized airway epithelial cultures, differentiated over several weeks, tends to be noisier than that obtained with cell lines. Despite this, a number of SERPINs reach statistical significance. *

• *

3) How is the dotted line drawn in Figure 3C and D? It would appear there is very little in terms of HADDOCK score to distinguish a predicted "binder" from "non-binder". Also, they later show that CTSB is non inhibited, and yet in Figure 3C it is below the dotted line. Can the authors more clearly delineate how one might use their dataset shown in Figure 3B to accurately predict targets of SERPINs?

This is a valid point. We added a more in-depth description to the results section on how we define “binders” and “non-binders” *(lines 324-331 and Figure 3 legend). We added raw data graph with the thresholds in Supplemental Figure 3d. We further added and defined a threshold line to the PAI-1:CTSs graph (Figure 3c). It is now evident that CTSL, A, F, K score as high-confidence “binders”, while CTSB and others do not. We also added the normalization process and the visual assessment of top-scoring complexes to the in silico docking screen schematic in Figure 3a and the respective figure legend to guide readers. *

4) Based on this, it would be preferable for the authors to tone down their claims about the broad applicability of this approach to predict SERPIN-protease interactions. It is true that they have used it to accurately predict PAI-1-CTSL interactions, but to make such a broad claim about the generalizable nature of this approach would require testing several more SERPIN-protease pairs (both binders and non-binders) to clearly define the scores and parameters that can used to robustly predict interactions.

We thank the reviewer for this criticism. We now address this in the text as outlined in our response to point 3 above. As with any screening method, the results require to be validated via an alternative approach, which we did in the initial submission for TMPRSS2 and CTSL as binders and CTSB as a non-binder. The revised manuscript now features additional in vitro validation of binders and non-binders for a total of 8 SERPIN-protease combinations (Figure 4a, b), which were all correctly predicted by our in-silico method. Two more SERPINs will be added in the final revision (see “planned revisions” section). Our study provides ample data for future studies validating additional predicted pairs and characterizing their biological function, in infectious disease and beyond.

5) In Figure 3D, the authors mutate all eight modeled RCL residues to alanine to create a LOF mutant that has a higher HADDOCK score. Single residue mutations would be more convincing for their model, and would be more informative in terms of their predicted models of interactions.

*We now performed the docking with the single mutant, please see new Figure 3c. *

• *

7) Further, in Figure 4G lanes 2-4, the PAI-1 band at ~38kDa is not present. Can the authors explain this?

*This is likely because CTSL digests PAI-1 working at its optimum pH (aka “the protease wins”). We removed the panel from the manuscript. *

9) In Figure 4I, the authors claim that the addition of PAI-1 is inhibiting cleavage of the SARS-CoV-2 spike protein (S2) based on densitometry quantifications. However, it is unclear how the authors are normalizing their data, nor whether the experiments (and therefore quantification) are from a single experiment or multiple replicates. Could the authors explain the quantification further and provide replicate information (including statistical support) if those experiments were performed?

Thank you for pointing this out. An explanation has now been added to the Figure 5 legend.

__Minor comments: __

1) The authors speculate about SERPINA1 regulation during viral infection, suggesting an active process of "viral evasion". However, it would appear that even upon interferon treatment in Figure 2C, SERPINA1 expression is decreased. Based on that, the authors should soften their claims about the cause of downregulation of SERPINA1.

Thank you for pointing this out – we softened the language on this point (*lines 225-228). *

2) In Figure 2C, do the authors have an explanation or hypothesis for why SERPINE1 is less upregulated at 72hrs when compared to 24hr infection of SARS-CoV-2?

*We can only speculate on this point. It is possible that one or several of the SARS-CoV-2 accessory proteins modulate SERPINE1 expression in a time-dependent manner. *

3) Can the authors demonstrate how the docking structure of the TMPRSS2 zymogen differs from the active version (especially zooming in on the interface of PAI-1 and the protease)? This could be supplemental data but can the authors show a panel like that in Figure 3F to show how the interface between PAI-1 and TMPRSS2 zymogen looks. Does the inactive TMPRSS2 not interface well with the RCL? Or what is leading to the decreased HADDOCK score?

We added an extensive paragraph on how HADDOCK operates to the results section to introduce how the HADDOCK score is calculated *(lines 248-290). We also added a visual of the top-scoring docking complex of PAI-1 and the TMPRSS2 zymogen (Figure 3d) to illustrate the differences in binding. *

4) In methods, uPA fluorometric protease assay information is missing. Please add this information.

Thank you for catching this – we added the information (line 890).

5) It is a bit confusing that Figure 4K is the quantification of assays shown in Figure 4A-C, rather than quantification of any of the intervening figure panels. It might be clearer to move this quantification next to 4A-C so that it is clearer.

*Thank you for the suggestion – Figure 4 has been restructured. *

6) In Figure 4H, the authors show that addition of recombinant PAI-1 decreases the number of SARS-CoV-2 nucleoprotein positive cells. Have the authors examined whether this decreases the viral titers as well?

*Yes, this is now part of the (new) Figure 5. *

8) In the text, the authors suggest that PAI-1 inhibition of CTSL is surprising/novel. The authors should reconsider phrasing this since there are several other SERPINs that have been shown to inhibit other cathepsins, making this appear less surprising than the authors are suggesting.

*Thank you for pointing this out. We have now clarified throughout the manuscript that while other SERPINs indeed are known to inhibit cathepsins, this had not been previously shown for the extremely well-studied SERPIN PAI-1 with over 15,000 pubmed entries. We also added the implications of this PAI-1-specific finding to the discussion section. *

__Significance: __

The claim of novelty about TMPRSS2 is confusing. In their previous paper (reference 19) they show that PAI-1 inhibits TMPRSS2 activity. These data are clearly shown in Figure 4C & 4D of that paper and are summarized in their sentence in the discussion: "Here, we find three new PAI-1 protease targets: human tryptase (tryptase Clara; club cell secretory protein), HAT, and TMPRSS2 ...". In this current paper, although they characterize the PAI-1-TMPRSS2 interaction in more detail than in their previous paper, they have truly only discovered one new target for PAI-1, which is CTSL.

Thank you for pointing this out – we softened language on the novelty of TMPRSS2 as a PAI-1 target *throughout the manuscript. We further clarify that the novelty is that TMPRSS2 has to be in its active form to be inhibited by PAI-1, which was previously unknown (lines 392, 432). The revised manuscript now also provides validation of total 8 predicted binders and non-binders for 2 (Figure 4 b,c), with 8 more pending (see “planned revisions” section). As those two (future four) SERPINs have various roles in cancer, cardiovascular disease, neurodegeneration, and immunity, our findings have impact beyond their antiviral potential, thereby increasing the overall significance of the manuscript. *

• *

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

Reviewer 1 major point:

---

The more common names for the SERPINS as detected in COVID alveolar lavage samples would be helpful in figure 1 - and specifically labelling PAI-1 as this is a focus for this study - together with the known SERPIN nomenclature or under abbreviations - For example SERPINB2 is PAI-2 and SERPING1 is C1INH and SERPINA1 is alpha 1 antitrypsin *Thank you for this suggestion. We tried to keep the SERPIN nomenclature consistent throughout the manuscript, in that the SERPIN genes are referred to by their gene name (i.e., SERPINE1), while the proteins are referred to by their protein name (i.e., PAI-1). Editor and/or Reviewer 1, please advise whether this is acceptable or should be changed. We also added the protein corresponding names in the figure legend. *

Why does supplemental figure 2 show SERPINB1 and not PAI-1. *We performed this computer-modeled experiment (docking SERPINs to known binders and known non-binders) for each SERPIN tested in the study. This was needed to obtain thresholds to define likely binders and likely non-binders. We chose to show SERPINB1 in this supplemental figure because it is well-described with regards to binders and non-binders (the latter, as “negative result”, is not always published for a given SERPIN). We also did not want to narrow the study immediately to PAI-1, as we believe our screen is a generalizable method and our findings are valid beyond PAI-1. We can easily show any other SERPIN here - editor and/or Reviewer 1, please advise. *

Reviewer 2 major point:

6) Figures 4F and 4G are rather confusing. First, in Figure 4F, amount of PAI-1 in lane 1 is not the same as in the lanes with CTSL. The biggest concern with this is that there is a second, higher MW band that is present in lane 1 (also in Figure 4G lane 1) that runs near the band in lanes 2&3 that is marked as the PAI-1-CTSL complex. Although it does appear that the band in lane 1 and lanes 2&3 are slightly different sizes, it is hard to say that conclusively when the amounts of PAI-1 are different. Can the authors repeat this assay to load consistent amounts PAI-1 across all conditions and even potentially separate the top bands to more convincingly show that the band in lanes 2&3 is not in the PAI-1 alone control?

*The upper band is an impurity that disappears upon addition of a protease to the reaction. We confirmed that this band is neither PAI-1 nor CTSL via western blot with PAI-1- or CTSL-specific antibodies. Should reviewer 2 and/or the editor feel that we should repeat the experiment with more loading in the first lane, we can certainly do so. Please advise. *

8) The authors show that exogenous PAI-1 can inhibit SARS-CoV-2 in a multicycle infection in Figure 4H. However, this could be acting at multiple points during the viral infection cycle. A clearer virology experiment to support their model would be to perform single-cycle infections to show that the virus fails to productively infect the cell. For instance, have the authors attempted a high MOI, single-cycle infection to see whether they can detect uncleaved spike protein to show inhibition of cleavage? Or show that no early products of viral infection are produced? While this type of experiment is optional in that it is not required to support the claim that PAI-1 inhibits multicycle SARS-CoV-2 infection, it would support the conclusion that PAI-1 is inhibiting viral entry.

*We agree with the reviewer. We did expand on the virology by using now two strains of SARS-CoV-2 with different proteolytic needs, ancestral WA-1 and Omicron BA.1. We also performed titer analysis (all in Figure 5). *

*However, the other suggested experiments would represent a substantial amount of work in a BSL3 environment. We thus would prefer not do these experiments (as the reviewer states, it is optional), and instead tone down the manuscript to make clear we make no claims on viral entry. *

• *

Reviewer 2 minor point:

---

11) One (optional) way to extend these data and support their molecular model would be to mutate residues in PAI-1 that they predict are important for protease inhibition. As their source of PAI-1 currently is commercial, this would require purification of WT and variant PAI-1, which is clearly an undertaking. However, these data would strongly support their modeling and the importance of these residues in engaging with the proteases and springing the mousetrap for their in-vitro/in-vivo experiments (as suggested by data shown in Figure 3F and explained in text). Further, the authors can use these mutants to do some of the functional experiments in Figure 4 as a negative control, and potentially even separate the role of PAI-1 in inhibition of CTSL and TMPRSS2 in terms of SARS-CoV-2 inhibition.

*We agree that these (optional) experiments would be beautiful and are indeed part of future studies on the subject. We feel that they exceed the scope of this current manuscript. *

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

Rodriguez Galvan et al. use a combined computational and functional approach to identify a novel target for the protease inhibitor, PAI-1 (SERPINE1), and show that exogenous PAI-1 can inhibit SARS-CoV-2 replication. They first use a COVID-19 dataset to identify SERPINs that are differentially expression in individuals with mild and severe COVID-19. They further use experimental infections of a model of human airway epithelium to identify SERPINs that are upregulated in response to several viruses, as well as treatment with interferon. Using this panel of SERPINs and a panel of host proteases, they use computational docking to predict SERPINs that may inhibit human proteases that may be relevant for viral infection. Using these predictions, they show that PAI-1 inhibits TMPRSS2 (previously shown) and CTSL (newly shown in this study), two proteases with relevance for SARS-CoV-2 infection. They finally show that extracellular addition of PAI-1 inhibits multicycle replication of SARS-CoV-2.

Major comments:

1) The rigor of the results presented in Figure 1 are unclear. For the COVID-19 analyses (Figure 1), only one dataset is used, and no statistical analyses are performed to determine to what degree any of the changes they observe are significant relative to variation in the dataset. This makes it difficult to determine how much can be extrapolated from these data. 2) Similarly, the qPCR data presented in Figure 2 are presented with no statistical analyses. Results should not only be presented with fold change but also p-values that are adjusted for multiple testing. 3) How is the dotted line drawn in Figure 3C and D? It would appear there is very little in terms of HADDOCK score to distinguish a predicted "binder" from "non-binder". Also, they later show that CTSB is non inhibited, and yet in Figure 3C it is below the dotted line. Can the authors more clearly delineate how one might use their dataset shown in Figure 3B to accurately predict targets of SERPINs? 4) Based on this, it would be preferable for the authors to tone down their claims about the broad applicability of this approach to predict SERPIN-protease interactions. It is true that they have used it to accurately predict PAI-1-CTSL interactions, but to make such a broad claim about the generalizable nature of this approach would require testing several more SERPIN-protease pairs (both binders and non-binders) to clearly define the scores and parameters that can used to robustly predict interactions. 5) In Figure 3D, the authors mutate all eight modeled RCL residues to alanine to create a LOF mutant that has a higher HADDOCK score. Single residue mutations would be more convincing for their model, and would be more informative in terms of their predicted models of interactions. 6) Figures 4F and 4G are rather confusing. First, in Figure 4F, amount of PAI-1 in lane 1 is not the same as in the lanes with CTSL. The biggest concern with this is that there is a second, higher MW band that is present in lane 1 (also in Figure 4G lane 1) that runs near the band in lanes 2&3 that is marked as the PAI-1-CTSL complex. Although it does appear that the band in lane 1 and lanes 2&3 are slightly different sizes, it is hard to say that conclusively when the amounts of PAI-1 are different. Can the authors repeat this assay to load consistent amounts PAI-1 across all conditions and even potentially separate the top bands to more convincingly show that the band in lanes 2&3 is not in the PAI-1 alone control? 7) Further, in Figure 4G lanes 2-4, the PAI-1 band at ~38kDa is not present. Can the authors explain this? 8) The authors show that exogenous PAI-1 can inhibit SARS-CoV-2 in a multicycle infection in Figure 4H. However, this could be acting at multiple points during the viral infection cycle. A clearer virology experiment to support their model would be to perform single-cycle infections to show that the virus fails to productively infect the cell. For instance, have the authors attempted a high MOI, single-cycle infection to see whether they can detect uncleaved spike protein to show inhibition of cleavage? Or show that no early products of viral infection are produced? While this type of experiment is optional in that it is not required to support the claim that PAI-1 inhibits multicycle SARS-CoV-2 infection, it would support the conclusion that PAI-1 is inhibiting viral entry. 9) In Figure 4I, the authors claim that the addition of PAI-1 is inhibiting cleavage of the SARS-CoV-2 spike protein (S2) based on densitometry quantifications. However, it is unclear how the authors are normalizing their data, nor whether the experiments (and therefore quantification) are from a single experiment or multiple replicates. Could the authors explain the quantification further and provide replicate information (including statistical support) if those experiments were performed? Further, to strengthen the conclusions of this data the authors should include additional controls. One would be to use trixplanin as they did in previous panels to show that PAI-1 is necessary. Further, if the authors generate mutant PAI-1 that is unable to inhibit TMPRSS2 (see comment 11 below), they could also use this as a control to show the necessity of functional PAI-1. 10) For Figures 4I-J, is it possible to also blot for S1 cleavage? If possible, this optional data would be helpful to understand whether the entire cleavage process is disrupted or only S2 to S2' especially given that visually it appears as if the full length is more depleted in the condition with PAI-1 suggesting that it is cleaving spike better into S1 and S2. Could also suggest that the dynamics of cleavage are shifted rather than impaired? 11) One (optional) way to extend these data and support their molecular model would be to mutate residues in PAI-1 that they predict are important for protease inhibition. As their source of PAI-1 currently is commercial, this would require purification of WT and variant PAI-1, which is clearly an undertaking. However, these data would strongly support their modeling and the importance of these residues in engaging with the proteases and springing the mousetrap for their in-vitro/in-vivo experiments (as suggested by data shown in Figure 3F and explained in text). Further, the authors can use these mutants to do some of the functional experiments in Figure 4 as a negative control, and potentially even separate the role of PAI-1 in inhibition of CTSL and TMPRSS2 in terms of SARS-CoV-2 inhibition.

Minor comments:

1) The authors speculate about SERPINA1 regulation during viral infection, suggesting an active process of "viral evasion". However, it would appear that even upon interferon treatment in Figure 2C, SERPINA1 expression is decreased. Based on that, the authors should soften their claims about the cause of downregulation of SERPINA1. 2) In Figure 2C, do the authors have an explanation or hypothesis for why SERPINE1 is less upregulated at 72hrs when compared to 24hr infection of SARS-CoV-2? 3) Can the authors demonstrate how the docking structure of the TMPRSS2 zymogen differs from the active version (especially zooming in on the interface of PAI-1 and the protease)? This could be supplemental data but can the authors show a panel like that in Figure 3F to show how the interface between PAI-1 and TMPRSS2 zymogen looks. Does the inactive TMPRSS2 not interface well with the RCL? Or what is leading to the decreased HADDOCK score? 4) In methods, uPA fluorometric protease assay information is missing. Please add this information. 5) It is a bit confusing that Figure 4K is the quantification of assays shown in Figure 4A-C, rather than quantification of any of the intervening figure panels. It might be clearer to move this quantification next to 4A-C so that it is clearer. 6) In Figure 4H, the authors show that addition of recombinant PAI-1 decreases the number of SARS-CoV-2 nucleoprotein positive cells. Have the authors examined whether this decreases the viral titers as well? 7) As a supplemental figure, can the authors show a complex blot (similar to Figure 4F) for CTSB to show that is does not complex with PAI-1. 8) In the text, the authors suggest that PAI-1 inhibition of CTSL is surprising/novel. The authors should reconsider phrasing this since there are several other SERPINs that have been shown to inhibit other cathepsins, making this appear less surprising than the authors are suggesting.

Significance

Assessment and impact:

This paper brings attention to the potential role of SERPINs in viral pathogenesis. The datasets shown in Figure 1 and 2, with the statistical caveats described above, are interesting demonstrations of the regulation of SERPINS during viral infection. In particular, the comparison of different viruses, and viruses compared to interferon alone, in Figure 2B is intriguing. These data are the strongest points of the paper.

The impact of the computational modeling is difficult to assess. While they have used this dataset to predict one novel interaction (CTSL) with PAI-1, the generalizable nature of this approach to broadly predict SERPIN-protease interactions is unclear since they have not tested or validated any other SERPIN-protease pairs. One major concern is the one raised in Major comments 3&4 above, which is that the score difference between a "non-binder" (CTSB) and a "binder" (uPa) is very small. It is exciting that they predicted CTSL as a target of PAI-1, but it is not obvious that this is a generalizable approach without further hypothesis testing.

The claim of novelty about TMPRSS2 is confusing. In their previous paper (reference 19) they show that PAI-1 inhibits TMPRSS2 activity. These data are clearly shown in Figure 4C & 4D of that paper and are summarized in their sentence in the discussion: "Here, we find three new PAI-1 protease targets: human tryptase (tryptase Clara; club cell secretory protein), HAT, and TMPRSS2 ...". In this current paper, although they characterize the PAI-1-TMPRSS2 interaction in more detail than in their previous paper, they have truly only discovered one new target for PAI-1, which is CTSL.

Finally, the data on SARS-CoV-2 are intriguing and contribute to an emerging field on antiviral SERPINs. This reveals an additional virus that is inhibited by PAI-1, to add to their previous discoveries (reference 19) of influenza virus and Sendai virus inhibition by PAI-1. Future virology experiments, and experiments with mutants that ideally separate the ability of PAI-1 to inhibit TMPRSS2 versus CTSL, will further reveal the step of viral replication that is inhibited, and reveal the contribution of inhibition of TMPRSS2, CTSL, or any other PAI-1 targets, on SARS-CoV-2 replication.

Audience: Overall, this paper will be interesting to a specialized audience that is interested in SERPIN function. The SERPIN expression data during viral infection, discovery of CTSL as a target of PAI-1, and evidence that PAI-1 can inhibit SARS-CoV-2 replication, will move that field forward.

Field of expertise: Biochemistry, host-virus interactions

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Title - In-silico docking platform with serine protease inhibitor (SERPIN) structures identifies host cysteine protease targets with significance for SARS-CoV-2

Authors - Joaquín J Rodriguez Galvan, Maren de Vries, Shiraz Belblidia, Ashley Fisher, Rachel A Prescott, Keaton M Crosse, Walter F. Mangel, Ralf Duerr, Meike Dittmann

Summary

The finding that PAI-1 has cross class serpin functions is of definite interest given the roles of PAI-1 in regulation of physiological processes as well as in driving pathology. PAI-1 is generally considered to be a key regulator of thrombolysis and thus an effect on other pathways and even intracellular pathways is of interest. Examining airway epithelial proteases and serpins is of definite interest in respiratory viral infections. Broadening the targets for serpins is also of very definite interest. This study ranges from an overview of prior published work on bronchoalveolar lavage samples and serpin expression, a tissue culture analysis of lung epithelial cells and expression of proteases and serpins is assessed. In addition selective changes in serpin expression and protease targets are assessed by in silico analysis as well as proof of concept via Western blot and fluorometric analysis. This is an extensive study and of definite interest.

There are some limitations as with any study, albeit the study overall is excellent. The authors do not reference the prior work which has examined cross class serpins, viral and mammalian, - this should be noted as alternative protease targets are known. They do mention the modulation of protease targets by glycosaminoglycans in the discussion. Further, serpins are inhibitors, thus while the RCL provides a target for a protease, but the response may not be fully selective in vivo as the protease has to be present and active to complete the serpin protease interaction. The bronchiolar lavage analysis is excellent but cannot differentiate epithelial cell and associated immune cells and their roles in the response. PAI-1 does not seem to be present in the bronchoalveolar lavage samoles. Further discussion of prior work with cross class serpins and also the limitations of the in silico analyses and the lavage specimens should be provided. Further analysis of the detected proteases that are reported here to bind to PAI-1 would be of great interest. The data from the bronchoalveolar lavage is published.

Overall this is a very simple, although extensive and excellent, study analyzing a wide range of data form patients with bronchoalveolar lavage and epithelial cell samples, human epithelial cell cultures after infection with a range of respiratory viruses as well as the development of a 3D in silico analysis of potential protease and serpin interactions. These correlations between changes in serpin and protease expression with viral infections and potential new interactions for serpins with previously non identified proteases is of clear interest. This shows an excellent correlation but as with big data sets this does not provide a true cause and effect - rather providing new potential directions for analysis of these interactions in viral infections in lung epithelium and this is valuable as a basis for ongoing studies. Prior work evidencing other cross class serpin protease targets as well as limitations related to the analyses as discussed in the critiques above should be noted and the abstract and title could better describe and define the studies as performed.

Critique

Major

1. Cross class serpin interactions are known and have been reported for at least two viral serpins Serp-1 and CrmA - both of which bind cysteine proteases as well as serine proteases as well as the mammalian SCCA serpins
2. The protease targets are reported to vary when interacting with glycosaminoglycans such as heparan sulfate - PAI-1 inhibits thrombin in the presence of heparin - thus while a canonical serpin suicide inhibition is considered specific - it can vary. This is noted in the discussion
3. What is the potential impact of the noted interactions of PAI-1 with other proteases such as cathepsin - PAI-1 is considered to have predominately extracellular functions, but prior work indicates internalization of PAI-1 when bound to the uPA/uPAR complex with alterations in intra cellular activation
4. This is supported by basic in vivo and in vitro serpin and protease interactions that are demonstrated confirming in silico analyses, eg. gel shift analyses or even Mass spectrometry analysis particularly for PAI-1
5. Per the authors "To date, three SERPINs have been studied in the context of innate antiviral defense: PAI- 1 (encoded by SERPINE1) against influenza viruses encoding hemagglutinin H1 and SARS-CoV-2, by impeding the proteolytic maturation of H1 or spike, respectively19,20; alpha-1-antitrypsin (encoded by SERPINA1) and antithrombin (encoded by SERPINC1) against SARS-CoV-2, likely through the inhibition of TMPRSS2, by reducing maturation of spike, although direct inhibition of TMPRSS2 by either SERPIN was not shown". This is partially complete however other serpins such as C1Inh and one virus derived serpin that have been analyzed for efficacy in treating SARS
6. While TMPRSS2 is indeed a serine protease - Beneficial effects of some serpins may be due to modulation of the immune response as opposed to selective anti-viral responses. The immune / cytokine storm and coagulopathies (with clotting and even hemorrhage) seen in the excess inflammatory response that causes respiratory vascular leak and severe viral sepsis. PAI-1 targets tPA and uPA - uPA has marked proinflammatory actions when bound to the uPA receptor (uPAR) and can activate growth factors and MMPs which can enhance immune cell invasion - PAI-1 binds to the uPA / uPAR complex which can thus also alter inflammatory cell responses and cell activation when internalized.
7. The RCL does in general incorporate P4 to P4' but can vary from this specific P4 to P4' sequence
8. How accurately does in silico protease serpin analysis predict real interactions? - this should be discussed as HADDOCK may have some limitations - This is outside my field of expertise
9. The data from a published study examining bronchoalveolar lavage fluid single cell transcriptional analysis from patients with and without COVID - mild and severe - and with comparison to patients without COVID does demonstrate altered protease and serpin activity - but does not indicate specific interactions
10. What is the significance for changes in gene expression in epithelial cells versus macrophage T and B cells looks - This looks like a small change like a small change in the mean values Figure 1b
11. The more common names for the SERPINS as detected in COVID alveolar lavage samples would be helpful in figure 1 - and specifically labelling PAI-1 as this is a focus for this study - together with the known SERPIN nomenclature or under abbreviations - For example SERPINB2 is PAI-2 and SERPING1 is C1INH and SERPINA1 is alpha 1 antitrypsin
12. Of interest - is the brocholaveolar lavage fluid likely to contain both epithelial cells as well as immune response macrophage, T cells and NK cells etc - one assumes single cells were identified and isolated- Is this defined?
13. The known previously reported target proteases for PAI-1 should be noted
14. SERPINE1 is not noted in figure 1 - this is PAI-1 - but is seen in the HAEC infection model data
15. "To overcome this limitation, we developed a computational method to predict 3D interactions between SERPINs and proteases, simulating the binding process depicted in Supplemental Figure 1a. Specifically, we employed High Ambiguity Driven protein- protein Docking (HADDOCK), a tool that predicts complex structures, integrating experimental and computational data35,36." This analysis looks to be extensive however this is a correlation - not a true analysis of cause and effect This does however have the potential to identify significant interactions - In future it might be of interest to assess PAI-1 given to infected cultures to assess viral replication and titers or perhaps examine a knock out cell model?
16. Why does supplemental figure 2 show SERPINB1 and not PAI-1
17. As PAI-1 was identified as having new cathepsin protease binding in addition to TMPRSS2 - the authors did demonstrate inhibition of the new targets on fluorometric analysis and also demonstrated interaction by gel shift - This is excellent
18. The title and the abstract could be better written and more clearly indicate the extent of the analyses performed and the discovery of alternate protease targets for PAI-1
19. Was the SARS CoV2 lung epithelial cell culture analysis performed in BSL3?

Minor critiques

1. Results section heading "SERPINs are differentially expressed individuals with COVID-19 and in response to respiratory virus infection in a model of the human airway epithelium." The word in needs to be inserted between expressed and individuals

Significance

Overall this is a very simple, although extensive and excellent, study analyzing a wide range of data form patients with bronchoalveolar lavage and epithelial cell samples, human epithelial cell cultures after infection with a range of respiratory viruses as well as the development of a 3D in silico analysis of potential protease and serpin interactions. These correlations between changes in serpin and protease expression with viral infections and potential new interactions for serpins with previously non identified proteases is of clear interest. This shows an excellent correlation but as with big data sets this does not provide a true cause and effect - rather providing new potential directions for analysis of these interactions in viral infections in lung epithelium and this is valuable as a basis for ongoing studies. Prior work evidencing other cross class serpin protease targets as well as limitations related to the analyses as discussed in the critiques above should be noted and the abstract and title could better describe and define the studies as performed.

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: Authors performed a metatranscriptomic analysis from publicly-available datasets of whole blood from 3 places in Indonesia. Their goal was to explore which pathogens were present on the blood of those 117 healthy individuals. It was interesting that reads from Flaviviridae and Plasmodium were detected in asymptomatic subjects.

Major comments: 1) How did the authors assess and correct batch-effects between different datasets?

Our response: We have sequencing batch information for the Indonesian dataset and saw no clear clustering based on batches in the first 8 PCs. We recognize that sampling variations may exist between islands, though the taxa matrix we acquired from the unmapped reads are very scarce that such variations did not have a strong enough effect to introduce batch effects in our microbiome analyses, and that the signals were driven by pathogenic reads. For our comparative analyses between datasets, we made sure that all three datasets shared similar processing (collected using Tempus Blood RNA Tubes and went through globin depletion method) and have trimmed both Indonesian and Malian reads to match the length of the UK reads (75BP).

2) Did the RNA-seq capture poly-A mRNAs? If so... these reads that did not map the human genome were captured because of internal priming. Can they find internal poly A sequences in the genome of Flaviviridae and Plasmodium pathogens? I would like to know that to understand the source of the reads and which other pathogens may be missing (due to the lack of internal priming).

__Our response: __No, our dataset did not capture poly-A mRNAs. We performed ribosomal RNA (rRNA) and globin mRNA depletion.

3) Principal coordinates analysis (PCoA) is often utilized in metagenomics analysis. Although they are equivalent, is there a reason for using PCA?

Our response: Since we used CLR transformation, the resulting matrix lies in Euclidian space. PCA is just a form of PCoA in Euclidian space.

Minor comments: 1) "Indonesia is a country with large numbers of endemic and emerging infectious diseases [16], making it a crucially important location to monitor and understand the effects of pathogens on human hosts." Is there any epidemiological data that shows differences in infectious diseases across these 3 places? Can the authors provide a map and better explanation about the importance in comparing these 3 areas?

__Our response: __We have added references to malaria infection being more prevalent in the eastern side of Indonesia in the discussion section.

2) Why is it so hard to try to identify (only for Flaviviridae reads) reads that map to very relevant viruses, such as Zika, Dengue, and Yellow Fever? Why did the authors state that they "were unable to refine this assignment further" if this is one of the most interesting finding?

__Our response: __Our reanalysis showed a small percentage of the Flaviviridae reads to be assigned to the Pegivirus genus. As more diverse microbial genomes are added to reference databases and identical regions become more common between them, it becomes harder for the classifer to further define reads to species level (https://link.springer.com/article/10.1186/s13059-018-1554-6). Flaviviridae has distinct species spread across six different genera (https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=11050). In comparison, despite Plasmodiidae having more species recorded compared to Flaviviridae, an overwhelming majority of the species is part of the Plasmodium genus, hence we were able to refine them down to species-level (https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=1639119).

3) Is the script available at https://gitlab.unimelb.edu.au/igr-lab/Epi_Study ? This reviewer could not access it. __Our response: __We thank Reviewer 1 for pointing this out and have amended the link, now accessible here: https://gitlab.svi.edu.au/muhamad.fachrul/indo_blood_microbiome

Reviewer #1 (Significance (Required)):

Interesting paper that enable to extract additional knowledge from whole blood RNA-seq data. There are already several papers that do this and I think authors could go one step forward (for instance, PCR validation of additional individuals). I don't think this can be used for surveillance if it cannot identify species, it is more expensive than running targeted assays, and that may be many false negative pathogens in the samples.

__Our response: __We thank Reviewer 1 for their comments. We have updated our manuscript to reflect our updated analyses which minimizes false positive taxa and the project’s significance not as a mainline surveillance tool, but a retrospective one.

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

Summary:

Bobowik and colleagues perform a computational analysis of whole blood RNA-seq datasets from healthy individuals of three different regions of Indonesia. Their goal is to identify infecting pathogens and other microbes and correlate their abundances to host gene expression patterns or health characteristics in these populations. They find a broad range of bacterial, viral and microeukaryote taxa. When comparing the three Indonesian populations, they find that the Korowai population is the most diverse and different from the other two, possibly driven by the higher prevalence and abundance of Plasmodium (Apicomplexa) in this population.

Then, the authors conduct a statistical decomposition of human gene expression in these samples in independent factors using ICA, and correlate each of these factors to the abundances of the microbial taxa detected. This analysis allows researchers to associate specific patterns of gene expression, such as immune-related pathways, to the presence of members of the Apicomplexa and Kitrinoviricota phyla.

Lastly, the authors use previously published data from other two cohorts (from Mali and the UK) to contextualize their blood microbiome findings. They find microbial reads in all datasets. The Mali cohort is characterized by a large abundance of archaea, not found in the other two populations, while the UK cohort has the lower diversity. Altogether, the authors propose the use of RNA-seq data from human whole blood as a way to study the blood microbiome and establish potential associations between blood resident microbes and host gene expression

Major comments:

1) The methodology to filter and remove reads from potential contaminants needs to be more stringent to ensure the results do not contain spurious contaminants and that the conclusions are correct. It has been described that genomic databases are heavily contaminated with human sequences (Steinegger and Salzberg, 2020), and in this manuscript, even after a two-pass alignment with STAR, reads mapping to helminths also corresponded to the human genome. Additionally, ad-hoc removal of specific taxa (Metazoa and Viridiplantae) was only performed after suspicion of contamination. However, this ad-hoc removal cannot be performed with microbial (bacterial, viral, etc.) contaminants as there is a risk of removing actual bacteria from the samples. But it has been confirmed that many microbial assemblies also suffer from human contamination. Possible actions to take are the following: a.Perform the human mapping with more lenient parameters to avoid human reads to map to other (likely contaminated) genomes in genome databases. b.Remove common contaminants that have been documented, for instance in blood (Chrisman et al., 2022). c.Run a tool to detect contaminated contigs in the database used to map reads to microbes and remove these problematic contigs from further analysis.

Our response: We thank Reviewer 2 for the suggestions, especially to address contaminants. We have reanalyzed our data which resulted in much fewer taxa yet still retained the main pathogenic findings.

2) In line with the above, removing singletons (as I have understood these are taxa that are represented by a single read), is a way to minimize the risk of contamination. To take advantage of the functional profiling of RNA-seq, a measure to ensure that microbes found in blood are active would be to include in the analysis only taxa for which expression of more than a few genes is detected. This type of filtering has been previously applied in studies where very low microbial loads are expected (Lloréns-Rico et al., 2021). In this study, it has only been applied to the specific case of the archaeal taxon Methanocaldococcaceae. However, I would expect cleaner results if applied consistently to all taxa detected.

__Our response: __We have reanalyzed the data and applied this to all taxa detected.

3) The specificity of Methanocaldococcaceae in the samples from Mali is very striking. I am highly suspicious that this only occurs due to a batch effect, even though the authors were highly selective in their cohorts to avoid these. In fact, I extracted the genes spanning the regions highlighted in Supplementary Figure 9 of the Methanocaldococcus jannaschii genome. A BLAST search of these sequences returned, among Methanocaldococcus hits, hits from the ERCC synthetic spike-in sequences, used as internal controls in many RNA-seq experiments. ERCC synthetic spike-in hits appeared for all 4 regions in the genome of M. jannaschii highlighted in this figure. In the original publications of this dataset, there is no reference to the use of these ERCC controls, but given the observed matches, I suggest the authors to perform an extra step in their filtering pipeline to remove all reads mapping to these ERCC standards in all their three cohorts to prevent these sort of batch effects.

__Our response: __We thank Reviewer 2 for pointing this out. Our reanalysis, which now used proper 2-pass mapping and further downstream classification with both pairs of the reads, no longer detected any archaea.

4) I am puzzled by the inconsistencies shown between forward and reverse reads when mapping paired-end data. I expect these inconsistencies at lower taxonomic ranks (species or genus level) due to incomplete genomes, but not at higher taxonomic ranks. I wonder if, by performing more stringent filtering of contaminants as suggested above, the consistency between forward and reverse reads increases and both mates can be used, making the mapping more reliable.

__Our response: __We have reanalyzed the data using both pairs of the reads for classification, resulting in less detected taxa. We believe the new results are more robust as it no longer includes taxa that are not typically found in humans (such as the archae Methanocaldococcus and other environmental bacteria).

In summary, my main concerns regarding this manuscript involve the possibility that contaminants in the sequencing data may be the cause of some of the results presented, and I tried to propose ways of dealing with these contaminants. While some of the results may not be affected by detection of contaminants (i.e. the association between Apicomplexa and some ICs), others such as the diversity measures or the comparison across cohorts may be severely affected. I will consider these results highly preliminary until a more thorough and stringent approach for contaminant removal is applied.

Our response: We thank Reviewer 2 for the suggestions and have updated our manuscript with results updated analyses that are more stringent towards contaminants, as can be seen from our updated findings.

Minor comments:

1) I would appreciate some of the analyses done at lower taxonomic levels if the sparsity of the data allows it, after removing contaminants. Given that the CLR transformation does not allow for zeros, other alternatives such as GMPR (Chen et al., 2018) or adding a pseudocount would allow these analyses?

__Our response: __After our reanalysis, we ended up with even sparser data and therefore could not perform the analyses at lower taxonomic levels.

2) In the PCA shown in figure 1, does the number of microbial reads detected correlate with any of the first two components?

__Our response: __Yes Plamosdiidae correlates well with PCs 1 and 2 (0.66 & 0.73) and Flaviviridae correlates very strongly with PC1 (0.917). We have added this detail in the results section.

3) In Figure 1C, the x axis is wrongly named PC2.

__Our response: __We thank Reviewer 2 for pointing this out and have amended this detail.

4) There is a typo in the legend of Figure 1A ("showeing")

__Our response: __We thank Reviewer 2 for pointing this out and have amended this detail.

5) In the alpha diversity estimates comparison across the three different cohorts, after subsampling each population to achieve similar sample size in each cohort, it is stated that "after subsampling, each population had similar diversity estimates". However, the numbers shown afterwards corresponding to the mean values of alpha diversity, without confidence intervals or a boxplot/violin plot together with an accompanying statistical test, are not enough to assess similarity. I would appreciate a figure (similar to Figure 3E and F) or a test accompanying these mean values.

__Our response: __We thank Reviewer 2 for pointing this out and have amended this detail.

6) In the volcano plots (Figure 3A, B and others throughout the manuscript) it would help the reader to add lines for the thresholds chosen for the effect size and -log10(p-value) to separate significant results.

__Our response: __We thank Reviewer 2 for pointing this out and have amended this detail.

7) In Figure 3E and F, I would appreciate having bars for the statistically significant comparisons.

__Our response: __We thank Reviewer 2 for pointing this out and have amended this detail.

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:

Bobowik and colleagues perform a computational analysis of whole blood RNA-seq datasets from healthy individuals of three different regions of Indonesia. Their goal is to identify infecting pathogens and other microbes and correlate their abundances to host gene expression patterns or health characteristics in these populations. They find a broad range of bacterial, viral and microeukaryote taxa. When comparing the three Indonesian populations, they find that the Korowai population is the most diverse and different from the other two, possibly driven by the higher prevalence and abundance of Plasmodium (Apicomplexa) in this population.

Then, the authors conduct a statistical decomposition of human gene expression in these samples in independent factors using ICA, and correlate each of these factors to the abundances of the microbial taxa detected. This analysis allows researchers to associate specific patterns of gene expression, such as immune-related pathways, to the presence of members of the Apicomplexa and Kitrinoviricota phyla.

Lastly, the authors use previously published data from other two cohorts (from Mali and the UK) to contextualize their blood microbiome findings. They find microbial reads in all datasets. The Mali cohort is characterized by a large abundance of archaea, not found in the other two populations, while the UK cohort has the lower diversity. Altogether, the authors propose the use of RNA-seq data from human whole blood as a way to study the blood microbiome and establish potential associations between blood resident microbes and host gene expression

Major comments:

1. The methodology to filter and remove reads from potential contaminants needs to be more stringent to ensure the results do not contain spurious contaminants and that the conclusions are correct. It has been described that genomic databases are heavily contaminated with human sequences (Steinegger and Salzberg, 2020), and in this manuscript, even after a two-pass alignment with STAR, reads mapping to helminths also corresponded to the human genome. Additionally, ad-hoc removal of specific taxa (Metazoa and Viridiplantae) was only performed after suspicion of contamination. However, this ad-hoc removal cannot be performed with microbial (bacterial, viral, etc.) contaminants as there is a risk of removing actual bacteria from the samples. But it has been confirmed that many microbial assemblies also suffer from human contamination. Possible actions to take are the following:
   • a.Perform the human mapping with more lenient parameters to avoid human reads to map to other (likely contaminated) genomes in genome databases.
   • b.Remove common contaminants that have been documented, for instance in blood (Chrisman et al., 2022).
   • c.Run a tool to detect contaminated contigs in the database used to map reads to microbes and remove these problematic contigs from further analysis.
2. In line with the above, removing singletons (as I have understood these are taxa that are represented by a single read), is a way to minimize the risk of contamination. To take advantage of the functional profiling of RNA-seq, a measure to ensure that microbes found in blood are active would be to include in the analysis only taxa for which expression of more than a few genes is detected. This type of filtering has been previously applied in studies where very low microbial loads are expected (Lloréns-Rico et al., 2021). In this study, it has only been applied to the specific case of the archaeal taxon Methanocaldococcaceae. However, I would expect cleaner results if applied consistently to all taxa detected.
3. The specificity of Methanocaldococcaceae in the samples from Mali is very striking. I am highly suspicious that this only occurs due to a batch effect, even though the authors were highly selective in their cohorts to avoid these. In fact, I extracted the genes spanning the regions highlighted in Supplementary Figure 9 of the Methanocaldococcus jannaschii genome. A BLAST search of these sequences returned, among Methanocaldococcus hits, hits from the ERCC synthetic spike-in sequences, used as internal controls in many RNA-seq experiments. ERCC synthetic spike-in hits appeared for all 4 regions in the genome of M. jannaschii highlighted in this figure. In the original publications of this dataset, there is no reference to the use of these ERCC controls, but given the observed matches, I suggest the authors to perform an extra step in their filtering pipeline to remove all reads mapping to these ERCC standards in all their three cohorts to prevent these sort of batch effects.
4. I am puzzled by the inconsistencies shown between forward and reverse reads when mapping paired-end data. I expect these inconsistencies at lower taxonomic ranks (species or genus level) due to incomplete genomes, but not at higher taxonomic ranks. I wonder if, by performing more stringent filtering of contaminants as suggested above, the consistency between forward and reverse reads increases and both mates can be used, making the mapping more reliable.

In summary, my main concerns regarding this manuscript involve the possibility that contaminants in the sequencing data may be the cause of some of the results presented, and I tried to propose ways of dealing with these contaminants. While some of the results may not be affected by detection of contaminants (i.e. the association between Apicomplexa and some ICs), others such as the diversity measures or the comparison across cohorts may be severely affected. I will consider these results highly preliminary until a more thorough and stringent approach for contaminant removal is applied.

Minor comments:

1. I would appreciate some of the analyses done at lower taxonomic levels if the sparsity of the data allows it, after removing contaminants. Given that the CLR transformation does not allow for zeros, other alternatives such as GMPR (Chen et al., 2018) or adding a pseudocount would allow these analyses?
2. In the PCA shown in figure 1, does the number of microbial reads detected correlate with any of the first two components?
3. In Figure 1C, the x axis is wrongly named PC2.
4. There is a typo in the legend of Figure 1A ("showeing")
5. In the alpha diversity estimates comparison across the three different cohorts, after subsampling each population to achieve similar sample size in each cohort, it is stated that "after subsampling, each population had similar diversity estimates". However, the numbers shown afterwards corresponding to the mean values of alpha diversity, without confidence intervals or a boxplot/violin plot together with an accompanying statistical test, are not enough to assess similarity. I would appreciate a figure (similar to Figure 3E and F) or a test accompanying these mean values.
6. In the volcano plots (Figure 3A, B and others throughout the manuscript) it would help the reader to add lines for the thresholds chosen for the effect size and -log10(p-value) to separate significant results.
7. In Figure 3E and F, I would appreciate having bars for the statistically significant comparisons.

References:

Chen, L., Reeve, J., Zhang, L., Huang, S., Wang, X., and Chen, J. (2018). GMPR: A robust normalization method for zero-inflated count data with application to microbiome sequencing data. PeerJ 6, e4600. https://doi.org/10.7717/peerj.4600.

Chrisman, B., He, C., Jung, J.-Y., Stockham, N., Paskov, K., Washington, P., and Wall, D.P. (2022). The human "contaminome": bacterial, viral, and computational contamination in whole genome sequences from 1000 families. Sci Rep 12, 9863. https://doi.org/10.1038/s41598-022-13269-z.

Lloréns-Rico, V., Gregory, A.C., Van Weyenbergh, J., Jansen, S., Van Buyten, T., Qian, J., Braz, M., Menezes, S.M., Van Mol, P., Vanderbeke, L., et al. (2021). Clinical practices underlie COVID-19 patient respiratory microbiome composition and its interactions with the host. Nat Commun 12, 6243. https://doi.org/10.1038/s41467-021-26500-8.

Steinegger, M., and Salzberg, S.L. (2020). Terminating contamination: large-scale search identifies more than 2,000,000 contaminated entries in GenBank. Genome Biol 21, 115. https://doi.org/10.1186/s13059-020-02023-1.

Significance

The research reported in this manuscript may have both technical and clinical significance, once the concerns raised above are adequately addressed. At the technical level, once contamination can be ruled out or securely minimized, this work can provide guidelines for microbial identification from whole blood RNA-seq data, applicable to both prospective studies as well as to retrospective studies using previously generated datasets. From this perspective, this work would add to the existing body of bioinformatics pipelines aimed at detecting microbes from host RNA-seq data (Simon et al., 2018). From a clinical perspective, it can provide an additional means of pathogen and disease surveillance without the need of microbial culturing or pathogen-specific tests. However, the requirement of blood samples may still hamper use in rural or underdeveloped areas. Lastly, another advantage is the possibility to directly link microbial abundances to gene expression patterns in the host.

Field of expertise: bacterial transcriptomics, metatranscriptomics, low-biomass microbiome analyses.

Limitations in my expertise: I cannot evaluate the clinical implications of the associations between host gene expression patterns and microbial abundances. Also, I am not familiar with the ICA methodology.

Reference:

Simon, L.M., Karg, S., Westermann, A.J., Engel, M., Elbehery, A.H.A., Hense, B., Heinig, M., Deng, L., and Theis, F.J. (2018). MetaMap: an atlas of metatranscriptomic reads in human disease-related RNA-seq data. GigaScience 7, giy070. https://doi.org/10.1093/gigascience/giy070.

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:

Authors performed a metatranscriptomic analysis from publicly-available datasets of whole blood from 3 places in Indonesia. Their goal was to explore which pathogens were present on the blood of those 117 healthy individuals. It was interesting that reads from Flaviviridae and Plasmodium were detected in asymptomatic subjects.

Major comments:

1. How did the authors assess and correct batch-effects between different datasets?
2. Did the RNA-seq capture poly-A mRNAs? If so... these reads that did not map the human genome were captured because of internal priming. Can they find internal poly A sequences in the genome of Flaviviridae and Plasmodium pathogens? I would like to know that to understand the source of the reads and which other pathogens may be missing (due to the lack of internal priming).
3. Principal coordinates analysis (PCoA) is often utilized in metagenomics analysis. Although they are equivalent, is there a reason for using PCA?

Minor comments:

1. "Indonesia is a country with large numbers of endemic and emerging infectious diseases [16], making it a crucially important location to monitor and understand the effects of pathogens on human hosts." Is there any epidemiological data that shows differences in infectious diseases across these 3 places? Can the authors provide a map and better explanation about the importance in comparing these 3 areas?
2. Why is it so hard to try to identify (only for Flaviviridae reads) reads that map to very relevant viruses, such as Zika, Dengue, and Yellow Fever? Why did the authors state that they "were unable to refine this assignment further" if this is one of the most interesting finding?
3. Is the script available at https://gitlab.unimelb.edu.au/igr-lab/Epi_Study ? This reviewer could not access it.

Significance

Interesting paper that enable to extract additional knowledge from whole blood RNA-seq data. There are already several papers that do this and I think authors could go one step forward (for instance, PCR validation of additional individuals). I don't think this can be used for surveillance if it cannot identify species, it is more expensive than running targeted assays, and that may be many false negative pathogens in the samples.

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

Learn more at Review Commons

---

Reply to the reviewers

Response to Reviewer 1

---

__Glycosaminoglycan (GAG)-binding proteins regulating essential processes such as cell growth and migration are essential for cell homeostasis. It is reported that the GAG has the ability to bind to Herpin sulfate. As both GAGs and the LPS lipid A disaccharide core of gram-negative bacteria contain negatively charged disaccharide units, the researchers proposed that heparin-binding peptides might have cryptic antimicrobial peptide motifs. To prove the hypothesis, they have synthesized five candidates [HBP1-5], which showed a binding affinity towards heparin and LPS binding. By using various methods, they showed that these molecules have antimicrobial activity. The key finding in this study is the finding of the CPC domain, where C is a cationic amino acid and P is a polar amino acid. __

Major comments

1. __ Even though the Authors propose here that CPC' clip motif is needed for antimicrobial activity. However, various studies have demonstrated that the mere presence of cationic amino or hydrophobic amino acids does not give the activity, the location of these amino acids at the strategic position is critically needed. The major issue in this work, the authors have not presented, whether there was a single CPC motif or multiple in the 5 peptides they have synthesized. Further, they need to demonstrate how are the charged and hydrophobic amino acids distributed in the peptides. these things will clearly explain the difference in the activity as well spectrum of the peptides. The authors should make an extra figure or add information highlighting this unique characteristic for better understanding to the reader.__

We thank the reviewer for his/her comments and suggestions. We concur that the distribution of amino acids is crucial for the antimicrobial activity of the peptides and their ability to bind heparin. We also agree with the suggestion of illustrating the location of the CPC' motifs of HBPs in the context of the parental proteins and have accordingly done so in the new Supplementary Figure 1. In all cases, only one CPC' motif was identified in the antimicrobial region, as highlighted in the figure, and the inter-residue distances measured are consistent with the CPC' motif definition. Thus, we demonstrate that a CPC' motif exists in all five HBPs, which explains how they recognize and bind heparin.

To illustrate the distribution of charged and hydrophobic amino acids in HBPs, we have also prepared new Supplementary Figure 2, displaying electrostatic potentials in the predicted HBP structures, and showing how the distribution of charged residues creates hydrophobic and cationic patches on the surface of the peptides. Our analysis reveals cationic patches to be surrounded by hydrophobic residues, which may explain the ability of the peptides to disrupt membranes and exert antimicrobial activity.

__ It is strange to observe that there are quite a number of reports showing that the peptides derived from the Herprin binding proteins have antimicrobial activity, but no one has reported their efficacy in the in vivo mouse model. if possible, the authors could add their observations if in vivo studies were done. or as a future line of study.__

We thank the reviewer for his/her comment on the observation of antimicrobial activity in peptides derived from heparin-binding proteins. Indeed, a few such studies have appeared in the literature, some with moderate success [1]. It is possible that a lack of understanding on how to identify heparin-binding regions in proteins and AMPs underlies their relative paucity. In this context, we believe our results will spur further efforts, specifically by providing a rationale on how to identify CPC' motifs hence heparin-binding regions in protein sequences.

Regarding the suggestion of assessing the in vivo efficacy of HBPs, we would agree that it would be helpful for better understanding their potential therapeutic applications. However, we feel that such experiments are beyond the scope of our manuscript, which offers ample, compelling in vitro and in silico evidence of how heparin-binding proteins can be a source of AMPs. We have done this by showing that CPC' motifs embedded in such proteins can be unveiled, accurately defined in structural terms, and experimentally shown to possess antimicrobial activity. Furthermore, we have shown that heparin binding correlates with LPS binding, allowing us to propose a mechanistic explanation for how heparin binding can be related to antimicrobial activity.

Translating these results to animal models is possibly premature at this stage as, from a classical medicinal chemistry perspective, it would require previous structural elaboration in terms of, e.g., optimized serum half-life or serum protein binding, both of which can modulate activity in in vivo studies regardless of heparin affinity or bactericidal activity per se. Ongoing work in our laboratories is focused in these directions and will be reported in due time.

*Referees cross-commenting**

Minor comments

1. __ The presence of Cryptic antimicrobial domain in various heparin-binding proteins like laminin isoforms, von Willebrand factor, vitronectin, protein C inhibitor, matrix glycoproteins thrombospondin, proline arginine-rich end leucine-rich repeat protein and fibronectin, have been reported previous. It is not clear why the authors did not refer to that work. The authors should refer to the works. (same as reviewer 3)__

We were aware of other prior studies on heparin-binding proteins and did indeed cite some of them, though not exhaustively for conciseness' sake. However, as encouraged by reviewers 1 and 3 we have cited the following studies:

Malmström E, Mörgelin M, Malmsten M, Johansson L, Norrby-Teglund A, Shannon O, Schmidtchen A, Meijers JC, Herwald H. Protein C inhibitor--a novel antimicrobial agent. PLoS Pathog. 2009 Dec;5(12):e1000698. doi: 10.1371/journal.ppat.1000698. Epub 2009 Dec 18. PMID: 20019810; PMCID: PMC2788422.

Ishihara, J., Ishihara, A., Fukunaga, K. et al. Laminin heparin-binding peptides bind to several growth factors and enhance diabetic wound healing. Nat Commun 9, 2163 (2018). https://doi.org/10.1038/s41467-018-04525-w

Chillakuri Chandramouli R, Jones Céline and Mardon Helen J(2010), Heparin binding domain in vitronectin is required for oligomerization and thus enhances integrin mediated cell adhesion and spreading, FEBS Letters, 584, doi: 10.1016/j.febslet.2010.06.023

Papareddy P, Kasetty G, Kalle M, Bhongir RK, Mörgelin M, Schmidtchen A, Malmsten M. NLF20: an antimicrobial peptide with therapeutic potential against invasive Pseudomonas aeruginosa infection. J Antimicrob Chemother. 2016 Jan;71(1):170-80. doi: 10.1093/jac/dkv322. Epub 2015 Oct 26. PMID: 26503666.

All the earlier studies related to the antimicrobial activity of the peptides derived from the Heparin-binding protein reported a consensus Cardin and Weintraub motifs i.e, XBBBXXBX or XBBXBX, where X represents hydrophobic or uncharged amino acids, and B represents basic amino acids. However, in this work, the researchers report about the presence of the new CPC motif. So, this is unique and a novelty in the study.

We thank the reviewers for these observations. Indeed, our quest to unveil CPC' motifs in antimicrobial regions of heparin-binding proteins is the key point of our investigation, and what distinguishes it from previous studies on consensus motifs such as XBBBXXBX or XBBXBX. We believe our definition of CPC' motifs in simple, structure-based, and experimentally verifiable terms is not only a significant departure but also a step forward from earlier views, highlighting the importance of a structural perspective in defining heparin-binding regions. In point of fact, we show that our peptides, even without consensus Cardin-Weintraub motifs, bind heparin with high affinity. The presence of the CPC' motif is crucial for such binding, as well as for LPS binding, and the new experiments performed at editor/reviewer's request, where the CPC motif in HBP5 is abolished, with predictable impact, fully support our view, see new section "Insights into the CPC' motif of HBP-5 and its implication on the antibacterial mechanism" and new Table 3 in the revised manuscript.

__ Even though the researchers report on the role of the CPC motif in the antimicrobial activity and binding to the heprin, the authors did not show any data or draw the conclusions related to the CPC domain when it comes to differences in the activity. this is the weakness of the manuscript. (same as reviewer 2)__

We welcome the reviewer's observation. To address it, we made and tested three HBP-5 mutants aimed at showing how alterations in the CPC' motif might influence interaction with heparin and LPS, as well as antimicrobial properties. The first two mutants involved replacing positively charged R10 and R14 residues with glutamine, similar in size and polarity but uncharged. As shown in the new section "Insights into the CPC' motif of HBP-5 and its implication on the antibacterial mechanism" and on the new Table 3 of the revised manuscript, the changes reduced heparin binding, i.e., shorter retention times on affinity chromatography, as well as LPS binding, i.e., a decrease in EC50 in the cadaverine assay (Table 3). The modifications had a lesser impact on antimicrobial activity, most likely due to the low resolution of MIC assays.

In a further step to assess the effect of the CPC' motif on antimicrobial activity, we deleted it in full by replacing residues H9, R10 and R14 of HBP-5 by alanine. As expected, this DCPC' peptide showed a sharp reduction in both heparin and LPS binding (Table 3) and, most importantly, a significant and asymmetric change in antimicrobial activity, with substantial impact on Gram-negatives yet practically no effect on Gram-positives, suggesting that LPS plays a key role in this selective response. Altogether, these observations align with our hypothesis that heparin-binding proteins might exploit their intrinsic affinity for heparin as an opportunity to developing antimicrobial properties by leveraging structural similarities between glycosaminoglycans and LPS.

__ It is strange to observe that there are quite a number of reports showing that the peptides derived from the Herprin (sic) binding proteins have antimicrobial activity, but no one has reported their efficacy in the in vivo mouse model. if possible, the authors could add their observations if in vivo studies were done. or as a future line of study. (Same as reviewer 2)__

We would kindly direct attention to #2 in the response to reviewer 1 above.

__ There are more than 20 different AMP databases or prediction software. however, not all of them are 100 % current, their success rate varies from 30-50% only. It needs to be investigated if adding this search in the hit peptides might increase the success rate of the extra in silico-based AMPs prediction software.__

If we understand the question correctly, the reviewer wonders whether including a CPC' motif predictor would increase the accuracy of AMP search algorithms. In our view, this strategy has two main limitations to be considered: (i) locating a CPC' motif in a peptide sequence typically requires a known 3D structure. Unfortunately, this is not always the case, and for proteins lacking reliable 3D data it can be a challenging and resource-intensive process; (ii) while CPC' motifs may predispose proteins to evolve antimicrobial properties, it is unclear if this is a required feature for all AMPs. Imposing the presence of a CPC' motif may not be applicable to all AMPs, although it might help identifying peptides with specific activity against gram-negative strains.

In summary, while the query of including a CPC' motif search tool in AMP predictors is intriguing and worthy of exploration for its potential bearing on antimicrobial research, it is technically complicated and beyond the scope of our manuscript.

__Reviewer #1 (Significance (Required)): __

__All the earlier studies related to the antimicrobial activity of the peptides derived from the Heparin-binding protein reported a consensus Cardin and Weintraub motifs i.e, XBBBXXBX or XBBXBX, where X represents hydrophobic or uncharged amino acids, and B represents basic amino acids. However, in this work, the researchers report about the presence of the new CPC motif. So this is unique and a novelty in the study. __

Even though the researchers report on the role of the CPC motif in the antimicrobial activity and binding to the heparin, the authors did not show any data or draw conclusions related to the CPC domain when it comes to differences in the activity. This is the weakness of the manuscript.

We would direct reviewer's attention to #1 in the Referee's cross-commenting section above.

---

Response to Reviewer 2

---

This is a very nice paper by the Andreu and Torrent groups that report the antimicrobial and heparin-binding of several encrypted peptides. Overall, this study presents an intriguing exploration into the potential dual functionality of glycosaminoglycan (GAG)-binding proteins, specifically heparin-binding proteins (HBPs), in recognizing lipopolysaccharide (LPS) and exhibiting antimicrobial properties. The findings, particularly the identification and characterization of novel encrypted peptides, such as HBP-5, are promising and contribute to our understanding of the intricate interplay between GAG-binding proteins and immunity. The data provided and methodology are thorough and well described. In sum, this is a very nice work. Please see below my minor comments.

---

Minor comments:

1. __ Fig. 1 legend does not show antimicrobial activity. Please remove from the figure legend title.__

As pointed out by the reviewer, the legend was incorrect and has been corrected accordingly and now reads "Figure 1. Structural and bioinformatics analysis of HBPs".

__ Discussion section: the authors should expand this section a bit to discuss recent work in the encrypted/cryptic peptide area. There are some recent relevant papers published in the past 3 years that should be discussed.__

We agree with the reviewer's suggestion to expand the discussion section to address recent work in the field of encrypted/cryptic peptides. We have carefully reviewed the recent literature and added several references in this topic:

Torres MDT, Melo MCR, Flowers L, Crescenzi O, Notomista E, de la Fuente-Nunez C. Mining for encrypted peptide antibiotics in the human proteome. Nat Biomed Eng. 2022 Jan;6(1):67-75. doi: 10.1038/s41551-021-00801-1. Epub 2021 Nov 4. Erratum in: Nat Biomed Eng. 2022 Dec;6(12):1451. PMID: 34737399.

• *

Santos MFDS, Freitas CS, Verissimo da Costa GC, Pereira PR, Paschoalin VMF. Identification of Antibacterial Peptide Candidates Encrypted in Stress-Related and Metabolic Saccharomyces cerevisiae Proteins. Pharmaceuticals (Basel). 2022 Jan 28;15(2):163. doi: 10.3390/ph15020163. PMID: 35215278; PMCID: PMC8877035.

• *

Boaro A, Ageitos L, Torres MT, Blasco EB, Oztekin S, de la Fuente-Nunez C. Structure-function-guided design of synthetic peptides with anti-infective activity derived from wasp venom. Cell Rep Phys Sci. 2023 Jul 19;4(7):101459. doi: 10.1016/j.xcrp.2023.101459. PMID: 38239869; PMCID: PMC10795512.

__ References provided are a bit outdated and do not accurately reflect the latest in the field (see comment above).__

We thank the reviewer for this comment. Older references were updated as suggested.

__ Gram should be capitalized throughout the text.__

Gram has been capitalized as suggested by the reviewer.

__ Can the authors comment on the potential translatability of HBP-5? Please also comment on the potential advantages of having peptides that 1) bind to heparin; and 2) kill bacteria.__

We appreciate the reviewer's interest in the potential of HBP-5. Indeed, we believe it has promise for clinical applications due to its unique attributes, but further studies, including in vivo experiments and pharmacokinetic assessments, are needed to fully evaluate its potential. The advantages of peptides that bind to heparin and kill bacteria include targeted delivery or localization of therapeutic agents, enhanced efficacy, and minimized off-target effects. HBP-5's ability to perturb outer membrane LPS, a crucial aspect of its antibacterial activity, makes it a promising approach to combat Gram-negative bacterial infections, which are often challenging to treat. By disrupting the outer membrane integrity, HBP-5 may also enhance the susceptibility of Gram-negative bacteria to other antimicrobial agents or host immune responses, underscoring its translational potential for treating bacterial infections.

__ More details on the computational tools and methods used to mine the peptides are needed.__

We have updated the Methods section to provide more details on the computational tools used for defining AMPs. Briefly, from the library of heparin-binding proteins obtained from previous studies [2] and AMP scanning for all these proteins was performed using the AMPA tool. The predicted antibacterial segments were located in the 3D structure of their respective proteins. Then, the CPC' motifs were searched in each segment following the criteria previously reported in [3, 4]. The motif involves two cationic residues (Arg or Lys) and a polar residue (preferentially Asn, Gln, Thr, Tyr or Ser), with fairly conserved distances between the carbons and the side chain center of gravity, defining a clip-like structure where heparin would be lodged. This structural motif is highly conserved and can be found in many proteins with reported heparin binding capacity. Finally, for all these regions, docking with a heparin disaccharide was performed using AutoDock Vina to evaluate the potential binding energy.

---

---

Response to Reviewer 3

---

__Summary: This manuscript has identified and investigated antimicrobial peptides from GAG binding proteins. Authors hypothesized that due to physiochemical similarity between GAG and LPS, fragments of GAG binding proteins might exert antimicrobial activity particularly against G- bacteria. Authors have identified few such AMPs that demonstrate LPS binding and displayed antibacterial activity. They have also solved NMR structure of the potent peptide and mode of action. __

Major comments: AMPs are promising molecules that can serve as lead for the development of therapeutics against MDR bacteria. In particular, currently therapeutic options to treat MDR Gram negative pathogens are limited. The current study is interesting and provides new non-toxic AMPs. Conclusions drawn from the works are largely valid. However, authors should address following comments:

1. __ The design and characterization of the peptide YI12WF is not described. Previous studies had shown design of β-boomerang peptides (Bhattacharjya and coworkers) that target LPS.__

---

We thank the reviewer for this comment. YI12WF (YVLWKRKRFIFI-amide) has been previously reported [4, 5] and shown to bind LPS with high affinity. YI12WF also contains a CPC' motif that, if deleted, reduces heparin binding [4]. References have been added in the text.

__ Mutations or substitution of the key residues peptide 5 might improve the novelty of the work.__

We thank the reviewer for this comment and agree that targeted substitutions in HBP-5 might shed light on the importance of the CPC' motif. As this point was also raised by reviewer 1, we would direct the reviewer's attention to #2 in the *Referees cross-commenting** section above.

__ How these peptides disrupt LPS permeability is not investigated. As LPS is the major target.__

We thank the reviewer for this suggestion and have accordingly evaluated the outer membrane (OM) permeability of the peptides by the 1-N-phenyl-naphthylamine (NPN) assay, a widely used method to assess OM integrity in Gram-negative bacteria. NPN is typically unable to cross the intact outer membrane; however, when the membrane is damaged or disrupted, it can penetrate and interact with lipids and proteins inside the cell, leading to an increase in fluorescence which is directly correlated with the degree of OM permeability and serves as an indicator of membrane damage.

Our results, illustrated in the new Figure 2D, show that all peptides are able to disrupt the OM of Gram-negative bacteria comparably to the LL-37 positive control, except for HBP2. Notably, HBP-5 exhibits the highest activity against OM, consistent with findings elsewhere in the manuscript and altogether confirming the ability of HBPs to bind to and disrupt the LPS structure.

__ Are the D-enantiomers of the peptides active against bacteria?__

We tested the antibacterial activity of the D-enantiomer of HBP5 (dHBP-and 5) and found it to be even higher than that of all-L HBP-5 against both Gram-negative and -positive bacteria, probably due to increased proteolytic stability as found in many AMP studies [6, 7]. As for LPS and heparin affinity, L- and D-HBP-5 behaved similarly (Table R1). As expected, the CD signatures of L- and D-HBP-5 were mirror images (Figure R1). These results suggest that the conformation of the CPC' motif is preserved in dHBP5, in tune with all previous results.

Antibacterial Activity

ID

E. Coli

P. Aeruginosa

A. Baumannii

S. Aureus

E. Faecium

L. monocytognes

HPB-5

0.4

0.8

0.2

6.3

25

1.6

dHBP-5

0.1

0.2

0.2

1.6

0.4

0.2

---

---

Binding Affinity

---

LPS (EC50, µM)

Heparin (% Elution buffer)

HPB-5

0.9 {plus minus} 0.7

98.0

dHBP-5

1.1 {plus minus} 0.8

97.2

Table R1. Antimicrobial activity of HBP-5 and dHBP-5

---

---

---

---

---

---

---

---

Figure R1. CD spectra of HBP-5 (red line) and dHBP-5 (green line) in LPS (left panel) and heparin (right panel).

---

__ 3D structure of peptide 5 is solved in DPC micelle which is a mimic for eukaryotic cells. Authors should attempt to determine structure in LPS as shown in several recent studies with potent AMPs thanatin, MSI etc.__

We appreciate the suggestion and have indeed attempted to obtain NMR spectra of HBP-5 in LPS micelles. However, we've been hindered by peptide precipitation and, despite considerable efforts, have not been able to obtain satisfactory results thus far. In contrast, we have succeeded in obtaining CD spectra of HBP5 in LPS micelles, showing an a-helix conformation similar to the one in SDS micelles, hence suggesting similar conformation in both environments.

Minor comments: There are examples of AMPs derived from human proteins. Authors should highlight such works.

Other studies have been cited according to the reviewers' comments:

Malmström E, Mörgelin M, Malmsten M, Johansson L, Norrby-Teglund A, Shannon O, Schmidtchen A, Meijers JC, Herwald H. Protein C inhibitor--a novel antimicrobial agent. PLoS Pathog. 2009 Dec;5(12):e1000698. doi: 10.1371/journal.ppat.1000698. Epub 2009 Dec 18. PMID: 20019810; PMCID: PMC2788422.

Ishihara, J., Ishihara, A., Fukunaga, K. et al. Laminin heparin-binding peptides bind to several growth factors and enhance diabetic wound healing. Nat Commun 9, 2163 (2018). https://doi.org/10.1038/s41467-018-04525-w

Chillakuri Chandramouli R, Jones Céline and Mardon Helen J(2010), Heparin binding domain in vitronectin is required for oligomerization and thus enhances integrin mediated cell adhesion and spreading, FEBS Letters, 584, doi: 10.1016/j.febslet.2010.06.023

Papareddy P, Kasetty G, Kalle M, Bhongir RK, Mörgelin M, Schmidtchen A, Malmsten M. NLF20: an antimicrobial peptide with therapeutic potential against invasive Pseudomonas aeruginosa infection. J Antimicrob Chemother. 2016 Jan;71(1):170-80. doi: 10.1093/jac/dkv322. Epub 2015 Oct 26. PMID: 26503666.

---

---

References

1. Papareddy, P., et al., An antimicrobial helix A-derived peptide of heparin cofactor II blocks endotoxin responses in vivo. Biochimica et Biophysica Acta (BBA) - Biomembranes, 2014. 1838(5): p. 1225-1234.
2. Ori, A., M.C. Wilkinson, and D.G. Fernig, A systems biology approach for the investigation of the heparin/heparan sulfate interactome. J Biol Chem, 2011. 286(22): p. 19892-904.
3. Torrent, M., et al., The "CPC Clip Motif": A Conserved Structural Signature for Heparin-Binding Proteins.PLOS ONE, 2012. 7(8): p. e42692.
4. Pulido, D., et al., Structural similarities in the CPC clip motif explain peptide-binding promiscuity between glycosaminoglycans and lipopolysaccharides. J R Soc Interface, 2017. 14(136).
5. Bhunia, A., et al., Designed beta-boomerang antiendotoxic and antimicrobial peptides: structures and activities in lipopolysaccharide. J Biol Chem, 2009. 284(33): p. 21991-22004.
6. Varponi, I., et al., Fighting Pseudomonas aeruginosa Infections: Antibacterial and Antibiofilm Activity of D-Q53 CecB, a Synthetic Analog of a Silkworm Natural Cecropin B Variant. Int J Mol Sci, 2023. 24(15).
7. Chen, Y., et al., Comparison of Biophysical and Biologic Properties of α-Helical Enantiomeric Antimicrobial Peptides. Chemical Biology & Drug Design, 2006. 67(2): p. 162-173.

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 has identified and investigated antimicrobial peptides from GAG binding proteins. Authors hypothesized that due to physiochemical similarity between GAG and LPS, fragments of GAG binding proteins might exert antimicrobial activity particularly against G- bacteria. Authors have identified few such AMPs that demonstrate LPS binding and displayed antibacterial activity. They have also solved NMR structure of the potent peptide and mode of action.

Major comments: AMPs are promising molecules that can serve as lead for the development of therapeutics against MDR bacteria. In particular, currently therapeutic options to treat MDR Gram negative pathogens are limited. The current study is interesting and provides new non-toxic AMPs. Conclusions drawn from the works are largely valid. However, authors should address following comments

1. The design and characterization of the peptide YI12WF is not described. Previous studies had shown design of b-boomerang peptides (Bhattacharjya and coworkers) that target LPS.
2. Mutations or substitution of the key residues peptide 5 might improve the novelty of the work.
3. How these peptides disrupt LPS permeability is not investigated. As LPS is the major target.
4. Are the D-enantiomers of the peptides active against bacteria?
5. 3-D structure of peptide 5 is solved in DPC micelle which is a mimic for eukaryotic cells. Authors should attempt to determine structure in LPS as shown in several recent studies with potent AMPs thanatin, MSI etc,

Minor comments: There are examples of AMPs derived from human proteins. Authors should highlight such works.

Significance

The work described in the manuscript is novel and hold promises to develop antimicrobials in future.

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 very nice paper by the Andreu and Torrent groups that report the antimicrobial and heparin-binding of several encrypted peptides. Overall, this study presents an intriguing exploration into the potential dual functionality of glycosaminoglycan (GAG)-binding proteins, specifically heparin-binding proteins (HBPs), in recognizing lipopolysaccharide (LPS) and exhibiting antimicrobial properties. The findings, particularly the identification and characterization of novel encrypted peptides, such as HBP-5, are promising and contribute to our understanding of the intricate interplay between GAG-binding proteins and immunity. The data provided and methodology are thorough and well described. In sum, this is a very nice work. Please see below my minor comments.

Minor comments:

• Fig. 1 legend does not show antimicrobial activity. Please remove from the figure legend title.
• Discussion section: the authors should expand this section a bit to discuss recent work in the encrypted/cryptic peptide area. There are some recent relevant papers published in the past 3 years that should be discussed.
• References provided are a bit outdated and do not accurately reflect the latest in the field (see comment above).
• Gram should be capitalized throughout the text.
• Can the authors comment on the potential translatability of HBP-5? Please also comment on the potential advantages of having peptides that 1) bind to heparin; and 2) kill bacteria.
• More details on the computational tools and methods used to mine the peptides are needed.

Significance

The data provided and methodology are thorough and well described. In sum, this is a very nice 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 #1

Evidence, reproducibility and clarity

Glycosaminoglycan (GAG)-binding proteins regulating essential processes such as cell growth and migration are essential for cell homeostasis. It is reported that the GAG has the ability to bind to Herpin sulfate. As both GAGs and the LPS lipid A disaccharide core of gram-negative bacteria contain negatively charged disaccharide units, the researchers proposed that heparin-binding peptides might have cryptic antimicrobial peptide motifs. To prove the hypothesis, they have synthesized five candidates [HBP1-5], which showed a binding affinity towards heparin and LPS binding. By using various methods, they showed that these molecules have antimicrobial activity. The key finding in this study is the finding of the CPC domain, where C is a cationic amino acid and P is a polar amino acid.

Major comments

1. Even though the Authors propose here that CPC' clip motif is needed for antimicrobial activity. However, various studies have demonstrated that the mere presence of cationic amino or hydrophobic amino acids does not give the activity, the location of these amino acids at the strategic position is critically needed. The major issue in this work, the authors have not presented, whether there was a single CPC motif or multiple in the 5 peptides they have synthesised. Further, they need to demonstrate how are the charged and hydrophobic amino acids distributed in the peptides. these things will clearly explain the difference in the activity as well spectrum of the peptides. The authors should make an extra figure or add information highlighting this unique characteristic for better understanding to the reader.
2. It is strange to observe that there are quite a number of reports showing that the peptides derived from the Herprin binding proteins have antimicrobial activity, but no one has reported their efficacy in the in vivo mouse model. if possible, the authors could add their observations if in vivo studies were done. or as a future line of study.

Minor comments

1. The presence of Cryptic antimicrobial domain in various heparin-binding proteins like laminin isoforms, von Willebrand factor, vitronectin, pro-tein C inhibitor, matrix glycoproteins thrombospondin, proline arginine-rich end leucine-rich repeat protein and fibronectin, have been reported previous. It is not clear why the authors did not refer to that work. the authors should refer to the works.

Referees cross-commenting

Minor comments

1. The presence of Cryptic antimicrobial domain in various heparin-binding proteins like laminin isoforms, von Willebrand factor, vitronectin, pro-tein C inhibitor, matrix glycoproteins thrombospondin, proline arginine-rich end leucine-rich repeat protein and fibronectin, have been reported previous. It is not clear why the authors did not refer to that work. the authors should refer to the works. (same as reviewer 3)

All the earlier studies related to the antimicrobial activity of the peptides derived from the Heparin-binding protein reported a consensus Cardin and Weintraub motifs i.e, XBBBXXBX or XBBXBX, where X represents hydrophobic or uncharged amino acids, and B represents basic amino acids. However, in this work, the researchers report about the presence of the new CPC motif. So this is unique and a novelty in the study.

Even though the researchers report on the role of the CPC motif in the antimicrobial activity and binding to the heprin, the authors did not show any data or draw the conclusions related to the CPC domain when it comes to differences in the activity. this is the weakness of the manuscript. (same as reviwer 2) 2. It is strange to observe that there are quite a number of reports showing that the peptides derived from the Herprin binding proteins have antimicrobial activity, but no one has reported their efficacy in the in vivo mouse model. if possible, the authors could add their observations if in vivo studies were done. or as a future line of study.(Same as reviewer 2)

Significance

All the earlier studies related to the antimicrobial activity of the peptides derived from the Heparin-binding protein reported a consensus Cardin and Weintraub motifs i.e, XBBBXXBX or XBBXBX, where X represents hydrophobic or uncharged amino acids, and B represents basic amino acids. However, in this work, the researchers report about the presence of the new CPC motif. So this is unique and a novelty in the study.

Even though the researchers report on the role of the CPC motif in the antimicrobial activity and binding to the heprin, the authors did not show any data or draw the conclusions related to the CPC domain when it comes to differences in the activity. this is the weakness of the manuscript.

There are more than 20 different AMP databases or prediction software. however, not all of them are 100 % current, their success rate varies from 30-50% only. It needs to be investigated if adding this search in the hit peptides might increase the success rate of the extra in silico-based AMPs prediction software

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:

Although HEK cells are effective for studying molecular mechanisms and post-translational modifications through siRNA and variant overexpression manipulations, they lack functional relevance in a neuronal context. Consequently, the connection between molecular findings and observed phenotypes in mice is tenuous. It is suggested that the authors attempt to replicate these results (Figures 4 and 5) using a neuronal differentiation model employing ESCs or iPSCs.

We have previously attempted to generate DDX3XSer584Ala knock-in ESCs via CRISPR-Cas9 because, as the reviewer points out, this would facilitate investigating the role of DDX3X O-GlcNAcylation in a neuronal differentiation model. However, clones derived after puromycin selection stop proliferating and perish during clonal outgrowth - we will include a statement to this effect in the revised manuscript. A similar phenomenon has been reported previously in Neuro2a cells by Lennox et al. (2020), who reported that installation of DDX3X patient variants is potentially toxic in certain cell lines. Therefore, the HEK293 KD/overexpression approach, also used for the study of clinically relevant DDX3X variants in other studies, while sub-optimal, is the best possible currently accessible model.

While employing variant overexpression following siRNA-mediated reduction of the endogenous protein is a direct method to illustrate the effects of mutated DDX3X variants, the authors stress a connection between this regulatory mechanism and neurodevelopmental defects. Therefore, it would be justifiable for the authors to create cell lines by editing the endogenous DDX3X gene and demonstrate the effects of O-GlcNAc, disruption of DDX3X target levels, and cell cycle regulation. Combining these approaches (from points 1 and 2), the authors could generate iPSC/ESC lines containing the DDX3X mutations and examine their effects within a neuronal differentiation context. Such an approach would significantly enhance the impact of this study.

As mentioned above for point 1, we have previously tried to edit the endogenous DDX3X gene with a Ser584Ala point mutation for this purpose. However, after trying this approach in both ESCs and HEK293T cells, we consistently observed stalled proliferation and cell death during clonal outgrowth. Therefore, as desirable as this experiment is, we are limited to siRNA-mediated reduction of DDX3X and rescue via over-expression, as also extensively used in other studies.

Given the points raised by reviewers 1 and 3 about the appropriateness of model system and the links drawn between DDX3X O-GlcNAcylation and neurodevelopmental defects, we will revise the manuscript to highlight the correlative nature of this link. In addition we will add further data from OGT-CDG mouse models that strengthens this possible link (see response to point 4 by reviewer 1).

The finding of diminished DDX3X levels in OGT mutant mice and the consequent reduction in O-GlcNAc represents a pivotal connection to the observed neurodevelopmental defects in OGT-CDG. However, this aspect of the research remains somewhat unclear, as it has not been definitively demonstrated that O-GlcNAc levels of DDX3X in OGT mutant mice are indeed decreased. Without this confirmation, the causal relationship between OGT malfunction, O-GlcNAc, and reduced DDX3X levels cannot be firmly established. There is a possibility of indirect effects, and merely observing correlation does not suffice to draw the robust conclusions presented in this paper. To address the uncertainty surrounding Figure 6D, attributed to the antibody's declared lack of specificity, the authors should conduct additional experiments.

The aim of this study is not to confirm whether there is a causal link between OGT catalytic deficiency and DDX3X, but rather to report the function of DDX3X O-GlcNAcylation and propose a possible link between OGT catalytic deficiency, DDX3X loss of activity, and neurodevelopmental defects. However, we do agree that further investigation is required to determine whether there is a correlative link between OGT catalytic deficiency and DDX3X levels (and O-GlcNAcylation) in the mouse brain. Towards this end, we will repeat the immunoprecipitation of DDX3X from mouse brain lysate of wild type and OGT-CDG mice and blot for O-GlcNAc using different pan-specific O-GlcNAc antibodies/ far western techniques (CTD110.6, GST-CpOGAD298N). We will also incorporate an internal negative control (competition with free GlcNAc) to verify the specificity of the O-GlcNAc signals.

The study places significant emphasis on this phenotype and seeks to elucidate it, at least partially, through the O-GlcNAcylation of DDX3X. However, a precise description or depiction of this phenotype is absent. Understanding the phenotype of the OGT-CDG mice necessitates consulting existing literature. The authors ought to contemplate providing brain sections with relevant staining to (i) showcase the microcephaly phenotype and (ii) bolster their assertion regarding the dysregulated cell cycle by utilising appropriate marker stainings for the progenitor cells during embryonic development.

We have recently published a manuscript reporting a mouse model of OGT-CDG (OGTC921Y). OGTC921Y mice display microcephaly as determined by brain weight and skull length. An additional mouse model of the N648Y OGT-CDG variant also displays microcephaly (Authier et al., 2024 reports the C921Y mouse; the N648Y mice line is on BioRxiv: https://doi.org/10.1101/2023.08.23.554427 and subject to peer review elsewhere). As part of the revisions for this manuscript, we will report a micro CT-based analysis of reduced skull/brain volume that supports the microcephaly phenotype. O-GlcNAc, OGT, OGA, DDX3X and cyclin E1 western blots for three brain regions from these mouse models will also be provided. Furthermore, we will include NeuN staining of cortical brain sections to establish whether cortical density is reduced in OGTN648Y mouse brains. The proposed staining of progenitor cells during embryonic corticogenesis is a very good suggestion for future investigation, but is a time-consuming experiment (est. 1 year) that falls beyond the scope of this study and would delay sharing of our current findings.

The proposed staining of progenitor cells during embryonic corticogenesis is a time consuming experiment (est. 1 year) that falls beyond the scope of this study.

Reviewer 2:

Figure 6 D - text last paragraph on page 16, and in supplement where you use RL2 - You need to do the control to show that RL2 staining goes away in the presence of free GlcNAc or when you galactosylate the protein. This control would indicate that you are detecting the sugar, not the protein backbone.

We agree with the reviewer that an internal negative control will help validate the specificity of the RL2 signal. We will repeat the immunoprecipitation/ O-GlcNAc western blot with a free GlcNAc control.

• Reviewer 3:*

Lack of Appropriate Model System: Although the authors initially address the role of O-GlcNAcylation (OGT/OGA) and DDX3X in cerebral development, given that mutations in these enzymes are causative for neurological malformations and disabilities, the experiments addressing the consequences of DDX3X or OGT silencing are predominantly performed in HEK293T cells rather than a nervous system model. This limits the relevance of the findings.

As discussed above (see Reviewer 1, points 1 and 2), we have previously tried to generate ESC knock-ins of the DDX3XSer584Alain ESCs to establish a neuronal differentiation model. However, clones perish during outgrowth, and thus these experiments are not possible. We have however performed preliminary biochemical analysis of DDX3X levels in the brains of OGT-CDG mice. It is important to emphasise that whether there is a causal link between OGT catalytic deficiency, DDX3X and cerebral development, lies beyond the remit of this manuscript. This manuscript aims to highlight DDX3X loss of activity as a candidate conveyor of neurodevelopmental defects in the mouse brain.

Inconclusive Cell Cycle Analysis: The authors' analysis regarding cell cycle characterization is not sufficiently conclusive. First, they need to accurately define the link between DDX3X and cyclin E1. The study they refer to (Lai et al., 2010) is rather superficial, and requires a more in-depth analysis, in order to appreciate the existing link between the two given molecules. Indeed, the current experiments do not clarify whether cells are stuck in G1 due to cyclin E1 downregulation or if cyclin E1 is downregulated because cells are blocked in G1. A suggested approach would be to perform rescue experiments with cyclin E1 overexpression, by using Quantitative Image-Based Cytometry (QIBC) or flow cytometry (EdU incorporation + Hoecsht staining) to monitor cell cycle changes and define the interplay between these molecules mechanistically. However, this alone cannot exclude the presence of alternative substrates of DDX3X regulation influencing cell cycle phase transition. A more holistic approach, such as an interactome analysis through mass spectrometry, may be helpful. Additionally, mild mitotic stress often results in cell cycle arrest in the subsequent G0/G1 phase, which can resemble the G1/S transition impairment described by the authors. Consequently, the statement "Here we identify Ser584 O-GlcNAcylation of DDX3X (...) as a key regulator of G1/S transition" is not well-supported (to do so, the authors should define the temporal pattern of such post-translational modification). It would also be interesting to determine whether the reduction in cell viability is due to a simple slowdown of the cell cycle or apoptotic induction.

We agree that performing cyclin E1 over-expression could provide mechanistic insights into the link between DDX3X, cyclin E1, and the cell cycle. We will therefore repeat the cell cycle analysis by flow cytometry shown in Fig. 5B with the addition of cyclin E1 over-expression in cells co-transfected with siRNA against DDX3X and siRNA-resistant DDX3XSer584Ala, to investigate whether cyclin E1 rescues the observed accumulation of cells in G1.

We acknowledge the reviewer's point that quiescence (G0 entry) and G1/S stalling can provide similar cell cycle profiles to that observed in Fig. 5B. We will therefore re-write the necessary sections of the manuscript to emphasise that we cannot be certain whether the observed cell cycle defects resulting from loss of DDX3X Ser584 O-GlcNAcylation stem from G1/S phase stalling or mitotic stress followed by quiescence.

Regarding the possibility that our observed reductions in the number of viable cells stems from stalled cell cycle progress or apoptosis, we will knock-down DDX3X and blot for cleaved caspase-3 as a marker for apoptosis to investigate the latter.

The proposed interactome analysis of DDX3X is a tangential experiment. Given that DDX3X regulates cell cycle progression through its RNA helicase and transcriptional co-regulator functions, interactome analysis would not provide direct (or indeed, useful) readouts of how loss of DDX3X Ser584 O-GlcNAcylation affects cell cycle progression (i.e. it is possible there are significant effects on DDX3X activity without affecting the interactome).

Weak Correlative Link: While the link between OGT/OGA and the cell cycle is well-established (seereviewSaunders et al., 2023) due to the multitude of targets subjected to this post-translational regulation, the correlative link between DDX3X mutations and cell cycle effects is further weakened by the fact that cyclin E1 knockout mice are indistinguishable from their wild-type littermates.

DDX3X controls the transcription and translation of hundreds of genes, not just cyclin E1. For example, DDX3X controls Klf4 transcription which, in certain cell lines, regulates S-phase entry (Canizarro et al., FEBS Lett. 2018). Thus, it is possible for one candidate conveyor of the observed cell cycle defects to not produce significant phenotypes in a KO mouse. In the interests of emphasising this point, we will add a discussion point regarding the multiple pathways through which loss of DDX3X O-GlcNAcylation may affect the cell cycle in the discussion.

Throughout the text, the authors refer to figures out of alphanumerical order, making the reading experience extremely difficult. To enhance readability, it is essential to present figures in a logical, sequential manner. Additionally, for further comments I would suggest to implement the text with line numbers.

We will correct the text to ensure all figures are referenced and presented in a sequential manner, and line numbers will be added.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary:

In the manuscript, Mitchell et al. explore the interplay between O-GlcNAcylation and the cell cycle, with a particular focus on how DDX3X, an RNA helicase of the DEAD box family, influences the G1/S phase transition. The authors begin with a bioinformatics strategy to identify patterns of correlation between cell cycle regulators and the enzymes OGT and OGA in a dataset of temporal transcriptomics of the human prefrontal cortex. They then narrow their analysis to DDX3X, due to its strong potential correlation with the cell cycle and its candidacy as a substrate for O-GlcNAcylation. Subsequently, the authors biochemically investigate how O-GlcNAcylation regulates DDX3X. Finally, given that DDX3X has been previously shown to regulate cyclin E1, the authors assess the effect of DDX3X depletion on the cell cycle in vitro.

Major comments:

Despite the authors' interesting identification of a novel substrate for O-GlcNAcylation, most conclusions drawn from the study are correlative. The manuscript suffers from three major drawbacks:

1. Lack of Appropriate Model System: Although the authors initially address the role of O-GlcNAcylation (OGT/OGA) and DDX3X in cerebral development, given that mutations in these enzymes are causative for neurological malformations and disabilities, the experiments addressing the consequences of DDX3X or OGT silencing are predominantly performed in HEK293T cells rather than a nervous system model. This limits the relevance of the findings.
2. Inconclusive Cell Cycle Analysis: The authors' analysis regarding cell cycle characterization is not sufficiently conclusive. First, they need to accurately define the link between DDX3X and cyclin E1. The study they refer to (Lai et al., 2010) is rather superficial, and requires a more in-depth analysis, in order to appreciate the existing link between the two given molecules. Indeed, the current experiments do not clarify whether cells are stuck in G1 due to cyclin E1 downregulation or if cyclin E1 is downregulated because cells are blocked in G1. A suggested approach would be to perform rescue experiments with cyclin E1 overexpression, by using Quantitative Image-Based Cytometry (QIBC) or flow cytometry (EdU incorporation + Hoecsht staining) to monitor cell cycle changes and define the interplay between these molecules mechanistically. However, this alone cannot exclude the presence of alternative substrates of DDX3X regulation influencing cell cycle phase transition. A more holistic approach, such as an interactome analysis through mass spectrometry, may be helpful. Additionally, mild mitotic stress often results in cell cycle arrest in the subsequent G0/G1 phase, which can resemble the G1/S transition impairment described by the authors. Consequently, the statement "Here we identify Ser584 O-GlcNAcylation of DDX3X (...) as a key regulator of G1/S transition" is not well-supported (to do so, the authors should define the temporal pattern of such post-translational modification). It would also be interesting to determine whether the reduction in cell viability is due to a simple slowdown of the cell cycle or apoptotic induction.
3. Weak Correlative Link: While the link between OGT/OGA and the cell cycle is well-established (see review Saunders et al., 2023) due to the multitude of targets subjected to this post-translational regulation, the correlative link between DDX3X mutations and cell cycle effects is further weakened by the fact that cyclin E1 knockout mice are indistinguishable from their wild-type littermates.

Minor comments:

Throughout the text, the authors refer to figures out of alphanumerical order, making the reading experience extremely difficult. To enhance readability, it is essential to present figures in a logical, sequential manner. Additionally, for further comments I would suggest to implement the text with line numbers.

Significance

Overall, the major novelty of the manuscript is the interesting link between DDX3X, its O-GlcNAcylation, and cell cycle regulation. However, this section requires the most intense revision to ensure robustness and clarity of the findings.

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

Missense mutations in OGT are causative in X-linked intellectual disability disorders, termed OGT-CDG. This paper identifies O-GlcNAcylation (OGN) at Ser584 of DDX3X, a known intellectual disability and microcephaly associated protein, as a regulator of G1/S transition, inhibiting the proteasome degradation of DDX3X. This OGN site controls the degradation of DDX3X. Lack of OGN at Ser 584 results in more rapid degradation of the protein, which affects the cell cycle. The study shows that dysregulation of DDX3X-dependent translation and concomitant impairments in cortical neurogenesis as a pathway disrupted in OGT-CDG. Overall, a well written paper. The data support the author's conclusions.

Figure 6 D - text last paragraph on page 16, and in supplement where you use RL2 - You need to do the control to show that RL2 staining goes away in the presence of free GlcNAc or when you galactosylate the protein. This control would indicate that you are detecting the sugar, not the protein backbone.

Significance

This well prepared paper is highly significant. 1) It provides mechanistic insights into a cause of human intellectual disability; 2) It helps elucidate the role of O-GlcNAcylation in nutrient regulation of the cell cycle. 3) It provides more data on the roles of DDX3X.

The paper is of significant general interest.

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

O-GlcNAcylation is a post-translational modification and plays a crucial role in neurodevelopment. Variants of O-GlcNAc transferase (OGT) are linked to OGT Congenital Disorder of Glycosylation (OGT-CDG), a syndrome causing intellectual disability. Microcephaly in OGT-CDG patients suggests the involvement of cell cycle dysregulation and abnormal neurogenesis. Mitchell et al. identify Ser584 O-GlcNAcylation of DDX3X, a protein associated with intellectual disability and microcephaly, as a regulator of G1/S-phase transition. They show that this PTM of DDX3X stabilises and prevents the targeting of DDX3X to the proteasome and therefore its degradation. Reduced DDX3X levels in an OGT-CDG mouse model and decreased expression of DDX3X-target gene cyclin E1 suggest impaired cell cycle control and cortical neurogenesis as pathways affected in OGT-CDG.

This study introduces DDX3X as a novel target for O-GlcNAcylation, which enhances its stability and ensures proper regulation of the cell cycle, particularly through its target cyclin E1. While the findings regarding DDX3X O-GlcNAcylation and the identification of the modified residue are compelling, addressing several issues regarding the link between O-GlcNAc of DDX3X-cell cycle regulation and neurodevelopmental impact would enhance the study's robustness.

Major points requiring attention:

1. Although HEK cells are effective for studying molecular mechanisms and post-translational modifications through siRNA and variant overexpression manipulations, they lack functional relevance in a neuronal context. Consequently, the connection between molecular findings and observed phenotypes in mice is tenuous. It is suggested that the authors attempt to replicate these results (Figures 4 and 5) using a neuronal differentiation model employing ESCs or iPSCs.
2. While employing variant overexpression following siRNA-mediated reduction of the endogenous protein is a direct method to illustrate the effects of mutated DDX3X variants, the authors stress a connection between this regulatory mechanism and neurodevelopmental defects. Therefore, it would be justifiable for the authors to create cell lines by editing the endogenous DDX3X gene and demonstrate the effects of O-GlcNAc, disruption of DDX3X target levels, and cell cycle regulation. Combining these approaches (from points 1 and 2), the authors could generate iPSC/ESC lines containing the DDX3X mutations and examine their effects within a neuronal differentiation context. Such an approach would significantly enhance the impact of this study.
3. The finding of diminished DDX3X levels in OGT mutant mice and the consequent reduction in O-GlcNAc represents a pivotal connection to the observed neurodevelopmental defects in OGT-CDG. However, this aspect of the research remains somewhat unclear, as it has not been definitively demonstrated that O-GlcNAc levels of DDX3X in OGT mutant mice are indeed decreased. Without this confirmation, the causal relationship between OGT malfunction, O-GlcNAc, and reduced DDX3X levels cannot be firmly established. There is a possibility of indirect effects, and merely observing correlation does not suffice to draw the robust conclusions presented in this paper. To address the uncertainty surrounding Figure 6D, attributed to the antibody's declared lack of specificity, the authors should conduct additional experiments.
4. The study places significant emphasis on this phenotype and seeks to elucidate it, at least partially, through the O-GlcNAcylation of DDX3X. However, a precise description or depiction of this phenotype is absent. Understanding the phenotype of the OGT-CDG mice necessitates consulting existing literature. The authors ought to contemplate providing brain sections with relevant staining to (i) showcase the microcephaly phenotype and (ii) bolster their assertion regarding the dysregulated cell cycle by utilising appropriate marker stainings for the progenitor cells during embryonic development.

Minor issue:

Consider replacing in cellulo with in vitro

Significance

The study provides an in-depth biochemical analysis of the O-GlcNAcylation of DDX3X, identifying it as a novel target. However, limitations stem from the absence of a robust causal connection between OGT-DDX3X and neurodevelopmental effects, as well as the utilization of HEK cells rather than a neuronal model. Moreover, the study would gain from exploring endogenous proteins instead of solely relying on siRNA and OE methods to investigate the cellular and functional impacts of DDX3X O-GlcNAcylation. Overall, the study provides a mechanistic advance regarding O-GlcNAc PTM and the targets of OGT. This work could be of interest to audiences interested in neurodevelopment, as well as PTM of non-histone proteins.

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)):

This manuscript uses C. elegans as a model to interrogate the effects of autism-associated variants of previously unknown function in the RNA-binding protein RBM-26/RBM27.

Despite its potential impact, there are several concerns related to the technical rigor and specificity of the observed effects.

Major concerns: 1. The effects on PLM are interesting, but why was this neuron selected for study? Was this a lucky guess or are other axons also affected? It is important to clarify whether the effects of RBM-26 are specific to this neuron or act pleiotropically across many or all neurons. According to CeNGEN, rbm-26 is strongly expressed in the well-characterized neurons ASE, PVD, and HSN. Are there morphological defects in these neurons, or others? As a note, there are also functional assays for these neurons (salt sensing, touch response, and egg laying, respectively).

We have added new data to the supplemental materials showing that loss of rbm-26 function also causes the beading phenotype in the axons and dendrites of the PVD neuron (Figure S4 and lines 196-199). We have focused on the PLM neuron because our preliminary studies indicated that it had a higher penetrance of axon defects relative to the PVD neuron. Moreover, we observed expression of endogenously tagged RBM-26 in the PLM neuron (Figure 3A-C and lines 210-215).

Similarly, the choice of the MALSU homolog seemed like a shot in the dark. It is ranked 46th (out of 63 genes) for fold-enrichment following RBM-26 pull-down, and 9th for p-value. Were any of the mRNAs with greater fold-enrichment or smaller p-values examined further? It is important to determine whether many or all of these interacting genes are overexpressed in the absence of RBM-26 and whether they are also required for the phenotypic effects of RBM-26 mutants, or if the MALSU homolog is special.

We have clarified our reasoning for selecting the MALS-1 ortholog of MALSU1 for further study (see lines 283-284 and Table S2). Amongst binding partners with human orthologs, MALS-1 was by far the top ranked candidate. The adjusted p-value for MALS-1 was 0.0008. The next smallest adjusted p-value was two orders of magnitude larger (0.028 for dpy-4). Moreover, the log2fold fold enrichment for MALS-1 was 1.98, about the same as the largest (ACADS with 2.13). Nonetheless, we agree that some of the other interactors may also be of interest and have thus included them in the supplemental table S2. Although these other potential binding partners are outside the scope of this study, we expect that future studies by ourselves or others may focus on the roles of these other binding partners.

In addition to the specificity controls mentioned above, positive and negative controls are needed throughout the results. While each of these may be relatively minor by itself, as a group they raise questions about the technical rigor of the study. Briefly these include: Fig 1C. Missing loading controls and negative control (rbm-26 null allele). Additional exposures should be included to show whether RBM-26(P80L) protein or the lower band for RBM-26(L13V) are present at all, relative to the null allele.

We have added no-stain loading controls to figure 1C. We have also switched to using ECL detection, which is much more sensitive and reveals faint bands for RBM-26(P80L) and additional faint bands for RBM-26(L13V). In addition, we have included a longer exposure for the blot (Figure S1). We are unable to test the null, as we can only produce a limited number of small maternally rescued progeny, thereby precluding western blot analysis.

Fig 2. Controls to distinguish overextension of PLM axon from posterior mispositioning of ALM cell body are needed. Quantification of PLM axon lengths in microns (or normalized to body size) with standard deviation, not error of proportion, should be shown. Measurement of "beading phenotype" should be more rigorous, see for example the approach in Rawson et al. Curr. Biol. 2017 https://doi.org/10.1016/j.cub.2014.02.025 . The developmental stage examined, and the reason for choosing that stage, should be described for this and all figures.

We have added new data that shows PLM axon length relative to body length for each of the RBM-26 mutants (Figure S2 and lines 183-185). These results indicate that the PLM axon has a larger axon length to body length ration, suggesting that the PLM/ALM overlap phenotype is a result of PLM axon overextension. For most experiments, we retain penetrance, as this has been standard practice in the field and allows for a much larger sample size (see examples listed below). We have also added examples of how the beading phenotype was measured (Figure S3). Moreover, we have now analyzed this phenotype and others at multiple developmental stages (Figures 2D-H and Table S1). In general, we have conducted experiments at the L3 stage because the rbm-26(null) mutants don't survive past this stage. However, for many of our experiments we have also included additional stages as well. We have added this explanation to the methods section of phenotype analysis and also at various locations throughout the text. We have also labeled all graphs to clearly indicate the developmental stages and included.

10.1038/s41467-019-12804-3 Article by laboratory of Brock Grill

10.1371/journal.pgen.1002513 Article by laboratory of Ian Chin-Sang

doi.org/10.1073/pnas.1410263111 Article by laboratory of Chun-Liang Pan

10.1016/j.neuron.2007.07.009 Article by laboratory of Yishi Jin

doi.org/10.1523/JNEUROSCI.5536-07.2008 Article by laboratory of William Wadsworth

Fig 3. Controls without auxin and with neuronal TIR1 expression alone should be included. Controls demonstrating successful RBM-26 depletion, in larvae as well as in embryos at the time of PLM extension, should be included (weak embryonic depletion might explain why the overextension phenotype is only 14% instead of 40% as in the null). According to CeNGEN, rbm-26 expression in PLM is barely detected, thus depletion with a PLM-specific TIR1 should also be tested. To confirm the authors' identification of the cell marked "N" as the PLM cell body, co-expression of rbm-26 and a PLM-specific marker should be added. Rescue of the rbm-26 mutants with neuronal (and PLM-only) expression should be included to test sufficiency in PLM, and as a further control for potential artifacts of the AID system.

We have added new data showing that an endogenously tagged RBM-26::Scarlet protein is expressed in the PLM neuron (Figure 3A-C). Moreover, we have added rescue experiments, showing that a Pmec-7::rbm-26::scarlet transgene can rescue the beading phenotype and the PLM/ALM overlap phenotype (Figure 3 F-G). We have also added controls without auxin (Figure S7) __and without the rbm-26::scarlet::aid gene (Figure S8). We have added a new figure showing auxin-mediated depletion of RBM-26::Scarlet::AID in the PLM neuron (Figure S10)__. We examined auxin-mediated depletion at the L3 stage for consistency with our auxin-mediated phenotypic experiments. Moreover, these were done at the L3 stage for consistency with other experiments that included the rbm-26(null) mutants, which don't survive past this stage.

In general, auxin-mediated knockdown tends to be hypomorphic in neurons. This is likely due to the fact that the neuronal TIR1 driver is expressed at much lower levels relative to the other drivers. In addition, the lower penetrance observed in auxin-mediated PLM/ALM overlap phenotype could reflect the fact that this phenotype resolves by the L4 stage in the hypomorphic mutants. For example, in P80L mutants at the L3 stage we see only about a 20% penetrance of the PLM/ALM overlap phenotype (relative to about 15% in auxin-mediated knockdown).

Fig 4. More rigorous quantification of the distribution of mitochondria along the axon should be included, not only total number, and it should be clarified what region of the axon the images are taken from. Including the AID-depletion strain with and without auxin would further add to the sense of rigor. For the mitoTimer experiments, why is RBM-26(L13V) not included and why do wild-type values differ ~5-fold between experiments (despite error bars being almost non-existent)? A more rigorous approach to standardizing imaging conditions may be needed. Positive controls using compounds that affect oxidation should be included. Measurements of individual mitochondria with standard deviations should be shown, rather than aggregate averages with error of proportion.

We have changed our methodology for measuring mitochondria, so that we now report the density of mitochondria in the axon (number per 100µm), (Figure 4E-F). We agree that this method is much better than counting the total number of mitochondria per axon, as it corrects for differences in body length and axon length). We also now include data for the whole axon (Figure 4E), proximal axon (Figure 4G), and distal axon (Figure 4H). These data suggest that the mitochondrial density defects occur in the proximal axon but not in the distal axon. Using the null allele, we have also examined the timing of mitochondria defects in the axon and report that the defects begin in the L1 stage and continue throughout larval development (Figure 4F). Individual datapoints have been added for all graphs in Figure 4.

For the mitoTimer experiments (Figure 5), we have added data for L13V and have added the individual datapoints to the graph. In the prior version, the values did not differ 5-fold between experiments with the same stage, rather the different graphs were from different stages (as noted in the figure legends/main text) and the L4 stage has much more oxidation than the L2 stage. To clear this up, we have added labels to the graphs to indicate the stages for each experiment. We have also added new data, so that we now show results for the L2, L3, and L4 stages for all three rbm-26 mutants (see Figure 5C-E). We didn't test the L1 stage because the signal was not sufficient for accurate quantitation.

Fig 5. Additional positive and negative controls should be added, including additional rbm-26 alleles, the AID-tagged strain with and without auxin, and a rescued mutant.

The old Figure 5 has become Figure 6 in the new version. We have added the rbm-26(L13V) allele to each experiment, (Figure 6B-D). We have also added the loading controls for the western blot along with quantification for 3 biological replicates of the western blot analysis (Figure 6D). We agree that these additions significantly strengthen the data because they show that two independent alleles of rbm-26 cause very substantial increase in the expression of mals-1 at both the mRNA and protein levels. We did not do these experiments with the rescuing transgene or with the AID-tagged strain because these experiments are done on whole worm lysates, whereas the AID-tagged and rescuing transgene are neuron-specific.

Fig 6. Controls showing whether the Scarlet-tagged protein is functional are needed, to rule out dominant negative or toxicity-related effects.

This is Figure 7 in the new version. For this experiment, we are showing that overexpression of MALS-1 does cause defects. The idea is that excessive amounts of MALS-1 causes deleterious effects to the mitochondria. In fact, these defects could be considered as dominant negative or toxic. We considered the possibility of crossing the Pmec-7::mals-1::scarlet transgene with rbm-26; mals-1 double mutants. However, this does not seem workable, because the single copy Pmec-7::mals-1::scarlet transgene produces the phenotypes at penetrances that are similar to what we observe in rbm-26; mals-1 double mutants. We concede that the results of the overexpression experiments in Figure 7 are limited when considered in isolation. However, we think that they are meaningful when considered in combination with the results on the mals-1;rbm-26 double mutants in Figure 8.

Fig 8. Controls for other mitochondrial components need to be included. It is important to determine if the decrease in ribosomes is specific or reflects a general decrease in mitochondria. If there are fewer mitochondria as suggested in Fig. 4, then of course mitochondrial ribosomal protein levels are also reduced. Additional rbm-26 alleles should be included here as well. Is this effect dependent on the MALSU homolog?

This is Figure 8D-E in the new version. We have added new data showing that the decrease in MRPL-58 expression that is caused by the rbm-26(P80L) mutation is dependent on MALS-1. We concede that these experiments cannot be used to determine anything about the mitoribosomes per se, but rather serve as an alternative way of testing the effect of rbm-26 on mitochondria. We have revised the text accordingly (lines 355-357). Given these limitations we have elected not to try additional mitochondrial markers and have also not included additional rbm-26 alleles for this experiment.

Finally the authors should address concerns about image manipulation, which amplify the concerns about technical rigor outlined above. The image in Fig. 2A appears to have a black box placed over the lower-right portion of the field to hide some features. Black boxes also appear to have been placed over the tops of images in Fig. 4B and 4D and at the left of Fig. 6A, 6B, and 6C. While these manipulations probably do not affect the conclusions, they further undermine confidence in data integrity and experimental rigor.

We have corrected all of these image processing errors. The box in 2A was for the purpose of squaring off a corner that was clipped during image rotation. The boxes in Figures 4 and 6 (of the prior version) were added to give space for labels (without obscuring image features). We have now used alternative methods to accomplish the same goals. For example, in Figures 4-D we have placed the labels outside of the images.

Minor points. 1. C. elegans nomenclature conventions should be followed: - C. elegans gene names have three or four letters, thus the MALSU homolog cannot be named "malsu-1". Please have new gene names approved by WormBase BEFORE submitting for publication http://tazendra.caltech.edu/~azurebrd/cgi-bin/forms/gene_name.cgi

We have changed malsu-1 to mals-1. In addition, both mals-1 and mrpl-58 have now been approved by wormbase and will be listed on the website upon its next update.

• If two sequential CRISPR edits are made on the same gene then they should be listed as a compound allele, such as rbm-26(cue22cue25)

We have updated our gene names to reflect this convention.

• Genes on the same chromosome should not be separated with a semicolon, for example rbm-26(cue40) K12H4.2(syb6330)

We have updated our gene names to reflect this convention.

Describing the defects as "neurodevelopmental" is misleading in the case of axon beading or degeneration. Similarly, there is no evidence for an "axon targeting" defect as stated in the abstract.

We have revised such that instead of referring to degeneration phenotypes as neurodevelopmental, we now refer to axon degeneration phenotypes that occur during development. For example, in the abstract we now say, "These observations reveal a mechanism that regulates expression of a mitoribosomal assembly factor to protect against axon degeneration during neurodevelopment.

Regarding targeting defects, this was meant to refer to the misplacement of the PLM axon tip (which contains electrical synapses). However, our subsequent analysis has revealed that these defects are transient in P80L and L13V mutants, as they resolve by the L4 stage. The rbm-26 null axon development defects do not resolve, though these mutant die prior to the L4 stage. Given these findings, we have decided not to use the term of targeting defects. Instead, we now refer to this as an axon tiling defect or PLM/ALM overlap phenotype.

In Fig. 5A, the symbol that appears to correspond to F59C6.15 (lowest p-value) is a different size than the others and is colored as ncRNA, whereas WormBase annotates this gene as snoRNA.

This error has been corrected.

In the Introduction, the last sentences of the first two paragraphs should be varied ("However, little is known about the [...] mechanisms that protect [...] during neurodevelopment.")

This has been done.

Why is RBM-26 protein running as a doublet at both sizes?

We have improved our western blotting methodology by using 12% gel, allowing for better resolution. We have also switched from colorimetric detection to ECL detection, allowing for greater sensitivity. In our new blots, we identify 6 different RBM-26 protein bands. We don't know the reason for these bands, but speculate that they are the result of post-translational processing (148-150).

When showing the RBM-26 expression pattern (Fig. 3) please include a lower-magnification image of the entire animal.

This has been done (Figure S6)

It is confusing to refer to the RNA IP experiments as an "unbiased screen", which in C. elegans typically refers to a genetic screen.

We now refer to this as a "biochemical screen".

The relationship between axon overextension, beading, and mitochondrial localization is not clear. What causal connection between these is being proposed? The causal connections between these phenotypes, if any, should be clarified experimentally. For example, if the axon extension defects develop before mitochondrial localization defects, then it is unlikely that mitochondrial defects cause axon overextension.

We have added new data showing that the reduction in mitochondrial density within the axon begins during the L1 stage and increases throughout larval development (Figure 4F). We have also added additional data showing that the increase in mitochondrial oxidation is weak in the L2 stage and surges in the L3 stage (Figure 5C-E), coincident with the beginning of the axon degeneration phenotypes. We propose (lines 383-391) that a low level of mitochondrial defects is present in L1 larvae, giving rise to the axon tiling defects. In the L3 stage there is a surge in excessive mitochondrial oxidation, giving rise to the axon degeneration phenotypes. We have added a new section to the discussion that addresses the relationship between defects in axon development and axon degeneration (lines 375-405).

Please explain how to interpret the difference in axon beading in the two deletion alleles of the MALSU homolog (axon beading defects in tm12122 but not in syb6330). Is syb6330 not a null allele? Or are the defects in tm12122 due to other mutations in this strain background?

One likely reason for this difference is that tm12122 is predicted to cause a partial deletion of the mals-1 coding sequence, whereas the syb6330 is a full deletion. Thus, the tm12122 could be acting as a dominant negative. In fact, prior work on the MALSU1 ortholog has indicated that this protein is subject to interference by a dominant negative construct (see Rorbach et al, Nucleic Acids Res 2012). Nonetheless, we cannot rule out the possibility of a linked second mutation in tm12122. However, since we have found similar phenotypes and genetic interactions with both alleles, we can conclude that these phenotypes and interactions are due to loss of MALS-1, rather than a second mutation.

Are mitochondria reduced in number or mislocalized? If they are reduced in number, is this due to altered balance of fission/fusion?

We have adjusted our methods for quantifying mitochondria and have also analyzed the proximal vs distal axon (Figure 4). We find that the density of mitochondria is decreased in the proximal axon, but not in the distal axon. We speculate that this might reflect a higher demand on mitochondria in the proximal axon, due to a higher amount of trafficking activity in the proximal axon (lines 255-257). We propose that the loss of RBM-26 causes dysfunction in mitochondria. Since fission and fusion are mechanisms that can help to repair damaged mitochondria, it is likely that they would be involved in the phenotypes that we observe.

In Fig. 3A-D, please keep the labels in the same position in all panels and do not alter brightness settings between single-color and merged panels.

These images have been moved to the supplemental data section (Figure S5). We have adjusted the labels as suggested. We have not changed the brightness settings, as they were already the same in all panels. However, the blue signal in the merged panel does obscure some of the red signal, giving an appearance of an alteration in color balance.

The claim that rbm-26 acts cell-autonomously requires PLM-specific depletion and rescue experiments.

We have added new data indicating that a Pmec-7::rbm-26::scarlet transgene can rescue the beading phenotype (Figure 3F-G).

**Referees cross-commenting** I appreciate the use of the consultation session to resolve differences between reviewers, but in this case I fully agree with the content and tone of all the comments from the other reviewer -- I think our remarks are very well aligned!

Reviewer #1 (Significance (Required)):

The study engineers autism-associated variants in conserved residues of RBM27 into the C. elegans homolog RBM-26 and identifies neuronal phenotypes potentially relevant to autism and a potential molecular mechanism involving regulation of mitochondrial ribosome assembly.

The key claims of the study are 1} that autism-associated variants in RBM-26 decrease its protein expression; 2} that impaired RBM-26 function leads to a variety of defects in development and maintenance of a single neuron called PLM, including altered axonal localization of mitochondria; 3} that RBM-26 normally binds the mRNA for the C. elegans homolog of MALSU, a mitochondrial ribosomal assembly factor; 4} that loss of RBM-26 leads to overexpression of the MALSU homolog; and 5} that MALSU is required for some of the deleterious effects on the PLM neuron seen in RBM-26 mutants.

This study will be of interest to the autism research community because it bolsters the idea that variants in RBM27 are likely to disrupt gene function and to affect neuronal health. It will also be of interest to the broader cell biology community because it suggests an interesting potential nucleus-to-mitochondria signaling mechanism, in which a nuclear RNA-binding protein might regulate assembly of mitochondrial ribosomes.

My field of expertise is developmental biology in C. elegans.

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

Summary In this manuscript, the authors studied an ASD-associated gene, rbm-26 in neuronal morphology using the touch receptor neuron PLM in C. elegans, and found that loss-of-function rbp-27 causes overextension and the formation of bulb-like structures in the axon. Using UV-crosslinking RNA immunoprecipitation and RNA-Seq, they identify malsu-1 as a target of rbm-26. Genetic analyses suggest malsu-1 likely functions downstream of rbm-26 in controlling the PLM morphology. Major comments:

• The authors describe RBM27 is associated with ASD and ID while they only cite SFARI paper that describes a weak association of RBM27 to ASD. The appropriate referenced that show link between RBM27 and ID should be provided. The link with ID was an error. We had meant to say "ASD or other neurodevelopmental disorders." This has been corrected.

• SFARI database only has three (P79L, R190Q, G348D) mutations listed as ASD-associated. Where are other mutations L13V and R455H, particularly L13V that the authors used to generate the C. elegans mutant come from? Are they associated with intellectual disabilities? The others came from the devovo-DB. We have added a reference for this database and have also added the primary source references for each of the five de novo variants (see line 121).

• The authors should be very careful when describing 'gene X causes Y diseases'. Many (if not all) of the examples described in this manuscript are disease-associated genes without validation to be causal genes. We have revised accordingly. For example on lines 433-435, we now say," For example, mutations in the EXOSC3, EXOSC8 and EXOSC9 are thought to cause syndromes that include defects in brain development such as hypoplasia of the cerebellum and the corpus callosum". We have decided to use the phrase "thought to cause" because three of the five referenced articles on these genes use titles that indicate causation.

• The authors refer PLM axon beading and overextension phenotypes to 'axon degeneration and targeting defects'. The authors must provide additional evidence of axon degeneration (see below). Also the term 'targeting defects' is misleading as the authors did not examine if overextension of the PLM axon causes targeting defects. At least they should examine some synaptic markers. To provide more evidence of degeneration we have analyzed several additional phenotypes at multiple developmental stages (Figure 2 and Table S1). Regarding targeting defects, this was meant to refer to the misplacement of the PLM axon tip (which contains electrical synapses). However, our subsequent analysis has revealed that these defects are transient in P80L and L13V mutants, as they resolve by the L4 stage. The rbm-26 null axon development defects do not resolve, though these mutant die prior to the L4 stage. Given these findings, we have decided not to use the term of targeting defects. Instead, we now refer to this as an axon tiling defect or PLM/ALM overlap phenotype.

• Neuronal phenotypes (axon overextension and beading) should be examined at different developmental timepoints (larval, young adult, and aged animals) to test if these phenotypes are indeed degenerative instead of developmental defects. We have included new data to observe all of these phenotypes at multiple developmental time points (Figure 2 and Table S1).

• The authors use the blebbing (beading) phenotype in the axon as the sole evidence of neurodegenerative properties of the PLM neuron. A more thorough analysis of this phenotype as done by others (Pan PNAS 2006) must be provided to support the authors' claim that this phenotype represents neurodegeneration. We have included new data on multiple degenerative phenotypes in axons including: blebbing, beading, waviness and breaks (Table S1).

• The number of beads per axon should be quantified to better represent the severity of rbm-26 mutant. Individual samples should be plotted in the quantification instead of showing the percentage of animals. We have added data on the density of beads in rbm-26(null), rbm-26(P80L), and rbm-26(L13V) mutants (Figure S3). For most experiments we have decided to use penetrance to measure axon degeneration because this is a standard in the field and allows for a larger sample size. For examples please see:

10.1523/JNEUROSCI.1494-11.2012 (Toth et al, 2012)

https://doi.org/10.1016/j.cub.2014.02.025 (Rawson et al, 2014)

10.1073/pnas.1011711108 (Pan et al, 2012)

https://doi.org/10.7554/eLife.80856 (Czech et al, 2023)

https://doi.org/10.1016/j.celrep.2016.01.050 (Nichols et al, 2016)

• Based on the single gel image in Fig. 1C with no loading control, the P80L mutant appears to have no protein expression. How is the P80L viable while the null mutant is lethal? The authors should quantify the protein expression levels from multiple blots with proper loading controls. If P80L mutation is introduced into RBM-26::mScarlet strain can it cause depletion of the signal in vivo? We have added new data showing that the RBM-26::Scarlet signal is diminished by the P80L mutation in vivo (Figure 1E-F). We have also added quantification from 3 biological replicate blots (Figure 1D). Finally, we have improved the sensitivity of our blots by using ECL detection and also show various exposures to highlight the fainter bands (Figures 1C and S1). Therefore, we are now able to detect low level expression of RBM-26(P80L) mutant protein. It is likely that the low level of RBM-26(P80L) and RBM-26(L13V) seen on western blots is sufficient to prevent the lethal phenotype.

• 'Moreover, loss of either the SPTBN1 or ADD1 genes causes a neurodevelopmental syndrome that includes autism and ADHD' References are missing, and as described above, be extra careful when indicating causality. Very few genes are known to cause ASD and ADHD. We have added the citations for this work (line 81). We also note that the titles for both of the cited articles indicate causation. To be on the safe side we have revised this line to say, "Moreover, loss of either the SPTBN1 or ADD1 genes are thought to cause a neurodevelopmental syndrome that includes autism and ADHD"

• Fig. 3E F, the authors should use the strains that express TIR1 specifically in the touch receptor neurons to argue cell autonomous function of RBM-26. Alternatively, the authors may conduct PLM neuron-specific rescue experiments to test the sufficiency. We have added new data indicating that a Pmec-7::rbm-26::scarlet transgene can rescue the beading phenotype and the PLM/ALM overlap phenotype (see Figure 3F-G).

• 'Loss of RBM-26 causes mitochondria dysfunction in axons.' The authors did not examine mitochondria function in axons. They only examined the number of mitochondria, and ROS production in the soma. The authors should provide additional evidence to support the idea that elevated ROS production in the soma is due to mitochondrial dysfunction in axons. Also, the authors should use both P80L and L13V for this experiment, and indicate individual datapoint as dots. Here, they quantified at the L4 stage, which the authors should justify. We have added the L13V data to this experiment and now show the individual data points. In addition, we have now conducted this analysis at the L2, L3 and L4 stages (Figure 5C-E). We have also revised the text to indicate that loss of rbm-26 function causes mitochondrial dysfunction in the cell body which could potentially cause a reduction of mitochondria in the axon (see lines 100-101 and 268-270). We speculate that mitochondria in the axon are also dysfunctional. However, the mitoTimer signal is not bright enough in axons to allow for quantification.

• Figure 5B and C: the authors should also use L13V to quantify malsu-1 mRNA and protein level, and include quantifications in panel C (from multiple blots). This is Figure 6 in the new version. We have added new data for expression of mals-1 mRNA and protein in rbm-26(L13V) mutants (Figure 6B-D). We have also included quantifications from 3 biological replicates (Figure 6D).

• In the rbm-26 mutant, the number of mitochondria is reduced, while the amount of MALSU-1 protein is increased. If MALSU-1 is specifically localized at mitochondria in wild type, where does the excessive MALSU-1 go in the rbm-26 mutants? Quantification of MALSU-1 signal intensity should be provided. Our Pmec-7::mals-1::scarlet transgene uses the tbb-2 3'UTR and causes an overexpression phenotype. To address the question posed by the reviewer, we would need to express MALS-1 at endogenous levels. Given that endogenous levels of MALS-1 are very low, it is unlikely that we would be able to visualize its expression. Nonetheless, as a way to address this question we have attempted to create a single copy Pmec-7::mals-1::scarlet transgene that utilizes the mals-1 endogenous 3'UTR. We have tried multiple approaches for generating this construct, but all have failed, likely due to sequence complexities within the mals-1 3'UTR. While we cannot say where the extra MALS-1 protein goes, we think that it is likely overloaded into the remaining mitochondria and could also be in the cytosol as well.

• Figure 7C: malsu-1 knockout mutants exhibit PLM overextension phenotype, which is not consistent with their model. The authors should discuss this in detail. We have added a paragraph to the discussion explaining that mitochondria function could be disrupted by either MALS-1 overexpression or by MALS-1 loss of function (lines 471-480).

• 'To validate these findings, we also repeated these experiments with an independent allele of malsu-1, malsu-1(tm12122) and found similar results (Fig. 7A-C).' The malsu-1(tm12122) exhibits beading phenotype and more severe overextension phenotype which the authors must describe and discuss more carefully. One likely reason for this difference is that tm12122 is predicted to cause a partial deletion of the mals-1 coding sequence, whereas the syb6330 is a full deletion. Thus, the tm12122 could be acting as a dominant negative. In fact, prior work on the MALSU1 ortholog has indicated that this protein is subject to interference by a dominant negative construct (see Rorbach et al, Nucleic Acids Res 2012). Nonetheless, we cannot rule out the possibility of a linked second mutation in tm12122. However, since we have found similar phenotypes and genetic interactions with both alleles, we can conclude that these phenotypes and interactions are due to loss of MALS-1, rather than a second mutation (albeit at a slightly different penetrance). We have added these considerations to the results section (lines 342-244).

• Figure 8: The authors should include data from L13V, malsu-1 and rbm-26; malsu-1 mutants. Quantification from multiple blots should be provided. This is Figure 8D in the new version. We have added the malsu-1 and rbm-26;malsu-1 double mutants to this experiment. We have also added quantification from multiple biological replicate blots. As pointed out by the other reviewer, we think that this experiment does not give specific information about mitoribosomes, but is an alternative approach to looking at the reduction in mitochondria. Given this limitation and considering that we have added L13V data to the mitochondria experiment in Figure 8B, we have elected not to add additional data on L13V to the western blot experiment in Figure 8D

Minor comments: • 'Consistent with a role for mitochondria in neurodevelopmental disorders, some of these disorders include a neurodegenerative phenotype.' Why is it consistent to have neurodegenerative phenotypes if mitochondria is associated with neurodevelopmental disorders? A better explanation would help.

We have changed this sentence to, "Some neurodevelopmental syndromes feature neurodegenerative phenotypes that occur during neuronal development."

• L13V is generally more severe in axon overextension phenotype than P80L while protein level is more abundant. The authors should discuss about this. We have also added a time course for the PLM/ALM overlap phenotype mutants (Figure 2D). This new data shows that the PLM/ALM overlap is quite similar overall between the P80L and L13V mutants. Both of these mutations cause an increase in PLM/ALM overlap in early larval development that is resolved by the L4 stage. The P80L phenotype resolves slightly sooner for reasons that are unknown. This could reflect differences in expression within the PLM that are not reflected in the whole worm lysate. This could also be due to a slight difference in the genetic background or other stochastic factors. The key point is that these two independent alleles cause similar phenotype overall, indicating that this phenotype is the result of loss in RBM-26 function.

• Fig. 2E, F: 'Beading refers to focal enlargement or bubble-like lesions which were at least twice the diameter of the axon in size.' How are the diameters of axons measured? A more detailed quantification method, and examples of measurement should be provided. We have added example measurements to the supplemental section (Figure S3). Additional detail on the measurements are in the Methods section (lines 517-518).

• Figure 3: The authors should also include low-magnification images to show where RBM-26 is expressed. The current image does now allow identifying cells. The transgene that labels the nuclei of hypodermis should be indicated in the manuscript. Specifically, the expression of the RBM-26 in the PLM should be shown. We have added a low magnification image (Figure S6) and have also added images of endogenously tagged RBM-26:Scarlet in the PLM (Figure 3A-C). The transgenic label for the hypodermis has been added to the legend of Figure S5.

• Figure 3: 'Tissue specific degradation of RBM-26::SCARLET::AID was achieved due to cell-type specific TIR-1 driver lines (see methods for details).' This information is not provided in the method section. This information has been added to methods section, "Auxin proteindegredation"

• Fig. 4 E. Values from individual samples should be indicated as dots. Representative images of P80L and L13V should be included. Conduct quantifications at adult stage as the authors use in other quantifications, or justify use of specific developmental stage (L3) they used. Figure 4 has become Figures 4 and 5 in the revised version. We have updated the graphs to include dots for individual data points. We have added quantifications of the mitoTImer experiments for the L2, L3 and L4 stages (Figure 5C-E). We note that our other experiments were done at the L1, L2, L3 and L4 and adult stages. The mitoTimer signal is not sufficient at the L1 stage for quantification. At the adult stage, the red signal becomes saturated. We have added representative images for mitoTimer in P80L and L13V mutants (Figure S9).

• The genes malsu-1 and mrpl-58 are not listed on wormbase. If the authors would like to designate names to these gene, they should clearly indicate that along with the sequence name. We have changed malsu-1 to mals-1. In addition, both mals-1 and mrpl-58 have now been approved by wormbase and will be listed on the website upon its next update.

• The authors found that MRPL-58 amount is reduced in rbm-26 mutants (which require additional verifications). This can be explained by the fact that axonal mitochondria number is reduced in the rbm-26 mutants. How did the authors confirm that the reduction in MRPL-58 level is due to the disruption of mitoribosome assembly? This is Figure 8D-E in the new version. We have added new data showing that the decrease in MRPL-58 expression that is caused by the rbm-26(P80L) mutation is dependent on MALS-1. We concede that these experiments cannot be used to determine anything about the mitoribosomes per se, but rather serve as an alternative way of testing the effect of rbm-26 on mitochondria. We have revised the text accordingly (lines 355-357).

• 'MALSU-1 is a mitoribosomal assembly factor that functions as part of the MALSU1:LOR8F8:mtACP anti-association module [37-39].' I don't think these are known for C. elegans MALSU-1. We have revised to, "MALS-1 is an ortholog of the MALSU1 mitoribosomal assembly factor that functions as part of the MALSU1:LOR8F8:mtACP anti-association module"

• 'Moreover, our results also suggest that disruption of this process can give rise to neurodevelopmental disorders.' I feel this is a quite a bit of stretch.

This has been replaced with, "Therefore, we speculate that human RBM26/27 could function with the RNA exosome complex to protect against neurodevelopmental defects and axon degeneration in infants." (lines 371-373)

**Referees cross-commenting** Yes, many of our comments overlap, and I fully agree with all comments from the other reviewer too. Reviewer #2 (Significance (Required)):

I found the manuscript interesting particularly the use of innovative techniques in identifying the target of RBM-26, The genetic analyses of rbm-26 and malsu-1 generally support the authors main conclusions that rbm-26 inhibits malsu-1 and be of potential interest to basic neuroscientists and cell biologists. However, the current manuscript looked premature which made my reading experience less pleasant. The phenotypic analyses is superficial compared to works similar to this work, which are insufficient to support the authors' claim of 'axon degeneration and targeting defects'. A number of issues listed above should be addressed before this manuscript is published. The reviewer's expertise: neurodevelopment in model 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

Summary

In this manuscript, the authors studied an ASD-associated gene, rbm-26 in neuronal morphology using the touch receptor neuron PLM in C. elegans, and found that loss-of-function rbp-27 causes overextension and the formation of bulb-like structures in the axon. Using UV-crosslinking RNA immunoprecipitation and RNA-Seq, they identify malsu-1 as a target of rbm-26. Genetic analyses suggest malsu-1 likely functions downstream of rbm-26 in controlling the PLM morphology.

Major comments:

• The authors describe RBM27 is associated with ASD and ID while they only cite SFARI paper that describes a weak association of RBM27 to ASD. The appropriate referenced that show link between RBM27 and ID should be provided.
• SFARI database only has three (P79L, R190Q, G348D) mutations listed as ASD-associated. Where are other mutations L13V and R455H, particularly L13V that the authors used to generate the C. elegans mutant come from? Are they associated with intellectual disabilities?
• The authors should be very careful when describing 'gene X causes Y diseases'. Many (if not all) of the examples described in this manuscript are disease-associated genes without validation to be causal genes.
• The authors refer PLM axon beading and overextension phenotypes to 'axon degeneration and targeting defects'. The authors must provide additional evidence of axon degeneration (see below). Also the term 'targeting defects' is misleading as the authors did not examine if overextension of the PLM axon causes targeting defects. At least they should examine some synaptic markers.
• Neuronal phenotypes (axon overextension and beading) should be examined at different developmental timepoints (larval, young adult, and aged animals) to test if these phenotypes are indeed degenerative instead of developmental defects.
• The authors use the blebbing (beading) phenotype in the axon as the sole evidence of neurodegenerative properties of the PLM neuron. A more thorough analysis of this phenotype as done by others (Pan PNAS 2006) must be provided to support the authors' claim that this phenotype represents neurodegeneration.
• The number of beads per axon should be quantified to better represent the severity of rbm-26 mutant. Individual samples should be plotted in the quantification instead of showing the percentage of animals.
• Based on the single gel image in Fig. 1C with no loading control, the P80L mutant appears to have no protein expression. How is the P80L viable while the null mutant is lethal? The authors should quantify the protein expression levels from multiple blots with proper loading controls. If P80L mutation is introduced into RBM-26::mScarlet strain can it cause depletion of the signal in vivo?
• 'Moreover, loss of either the SPTBN1 or ADD1 genes causes a neurodevelopmental syndrome that includes autism and ADHD' References are missing, and as described above, be extra careful when indicating causality. Very few genes are known to cause ASD and ADHD.
• Fig. 3E F, the authors should use the strains that express TIR1 specifically in the touch receptor neurons to argue cell autonomous function of RBM-26. Alternatively, the authors may conduct PLM neuron-specific rescue experiments to test the sufficiency.
• 'Loss of RBM-26 causes mitochondria dysfunction in axons.' The authors did not examine mitochondria function in axons. They only examined the number of mitochondria, and ROS production in the soma. The authors should provide additional evidence to support the idea that elevated ROS production in the soma is due to mitochondrial dysfunction in axons. Also, the authors should use both P80L and L13V for this experiment, and indicate individual datapoint as dots. Here, they quantified at the L4 stage, which the authors should justify.
• Figure 5B and C: the authors should also use L13V to quantify malsu-1 mRNA and protein level, and include quantifications in panel C (from multiple blots).
• In the rbm-26 mutant, the number of mitochondria is reduced, while the amount of MALSU-1 protein is increased. If MALSU-1 is specifically localized at mitochondria in wild type, where does the excessive MALSU-1 go in the rbm-26 mutants? Quantification of MALSU-1 signal intensity should be provided.
• Figure 7C: malsu-1 knockout mutants exhibit PLM overextension phenotype, which is not consistent with their model. The authors should discuss this in detail.
• 'To validate these findings, we also repeated these experiments with an independent allele of malsu-1, malsu-1(tm12122) and found similar results (Fig. 7A-C).' The malsu-1(tm12122) exhibits beading phenotype and more severe overextension phenotype which the authors must describe and discuss more carefully.
• Figure 8: The authors should include data from L13V, malsu-1 and rbm-26; malsu-1 mutants. Quantification from multiple blots should be provided.

Minor comments:

• 'Consistent with a role for mitochondria in neurodevelopmental disorders, some of these disorders include a neurodegenerative phenotype.' Why is it consistent to have neurodegenerative phenotypes if mitochondria is associated with neurodevelopmental disorders? A better explanation would help.
• L13V is generally more severe in axon overextension phenotype than P80L while protein level is more abundant. The authors should discuss about this.
• Fig. 2E, F: 'Beading refers to focal enlargement or bubble-like lesions which were at least twice the diameter of the axon in size.' How are the diameters of axons measured? A more detailed quantification method, and examples of measurement should be provided.
• Figure 3: The authors should also include low-magnification images to show where RBM-26 is expressed. The current image does now allow identifying cells. The transgene that labels the nuclei of hypodermis should be indicated in the manuscript. Specifically, the expression of the RBM-26 in the PLM should be shown.
• Figure 3: 'Tissue specific degradation of RBM-26::SCARLET::AID was achieved due to cell-type specific TIR-1 driver lines (see methods for details).' This information is not provided in the method section.
• Fig. 4 E. Values from individual samples should be indicated as dots. Representative images of P80L and L13V should be included. Conduct quantifications at adult stage as the authors use in other quantifications, or justify use of specific developmental stage (L3) they used.
• The genes malsu-1 and mrpl-58 are not listed on wormbase. If the authors would like to designate names to these gene, they should clearly indicate that along with the sequence name.
• The authors found that MRPL-58 amount is reduced in rbm-26 mutants (which require additional verifications). This can be explained by the fact that axonal mitochondria number is reduced in the rbm-26 mutants. How did the authors confirm that the reduction in MRPL-58 level is due to the disruption of mitoribosome assembly?
• 'MALSU-1 is a mitoribosomal assembly factor that functions as part of the MALSU1:LOR8F8:mtACP anti-association module [37-39].' I don't think these are known for C. elegans MALSU-1.
• 'Moreover, our results also suggest that disruption of this process can give rise to neurodevelopmental disorders.' I feel this is a quite a bit of stretch.

Referees cross-commenting Yes, many of our comments overlap, and I fully agree with all comments from the other reviewer too.

Significance

I found the manuscript interesting particularly the use of innovative techniques in identifying the target of RBM-26, The genetic analyses of rbm-26 and malsu-1 generally support the authors main conclusions that rbm-26 inhibits malsu-1 and be of potential interest to basic neuroscientists and cell biologists. However, the current manuscript looked premature which made my reading experience less pleasant. The phenotypic analyses is superficial compared to works similar to this work, which are insufficient to support the authors' claim of 'axon degeneration and targeting defects'. A number of issues listed above should be addressed before this manuscript is published.

The reviewer's expertise: neurodevelopment in model organisms.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

This manuscript uses C. elegans as a model to interrogate the effects of autism-associated variants of previously unknown function in the RNA-binding protein RBM-26/RBM27.

Despite its potential impact, there are several concerns related to the technical rigor and specificity of the observed effects.

Major concerns:

1. The effects on PLM are interesting, but why was this neuron selected for study? Was this a lucky guess or are other axons also affected? It is important to clarify whether the effects of RBM-26 are specific to this neuron or act pleiotropically across many or all neurons. According to CeNGEN, rbm-26 is strongly expressed in the well-characterized neurons ASE, PVD, and HSN. Are there morphological defects in these neurons, or others? As a note, there are also functional assays for these neurons (salt sensing, touch response, and egg laying, respectively).
2. Similarly, the choice of the MALSU homolog seemed like a shot in the dark. It is ranked 46th (out of 63 genes) for fold-enrichment following RBM-26 pull-down, and 9th for p-value. Were any of the mRNAs with greater fold-enrichment or smaller p-values examined further? It is important to determine whether many or all of these interacting genes are overexpressed in the absence of RBM-26 and whether they are also required for the phenotypic effects of RBM-26 mutants, or if the MALSU homolog is special.
3. In addition to the specificity controls mentioned above, positive and negative controls are needed throughout the results. While each of these may be relatively minor by itself, as a group they raise questions about the technical rigor of the study. Briefly these include:

Fig 1C. Missing loading controls and negative control (rbm-26 null allele). Additional exposures should be included to show whether RBM-26(P80L) protein or the lower band for RBM-26(L13V) are present at all, relative to the null allele.

Fig 2. Controls to distinguish overextension of PLM axon from posterior mispositioning of ALM cell body are needed. Quantification of PLM axon lengths in microns (or normalized to body size) with standard deviation, not error of proportion, should be shown. Measurement of "beading phenotype" should be more rigorous, see for example the approach in Rawson et al. Curr. Biol. 2017 https://doi.org/10.1016/j.cub.2014.02.025 . The developmental stage examined, and the reason for choosing that stage, should be described for this and all figures.

Fig 3. Controls without auxin and with neuronal TIR1 expression alone should be included. Controls demonstrating successful RBM-26 depletion, in larvae as well as in embryos at the time of PLM extension, should be included (weak embryonic depletion might explain why the overextension phenotype is only 14% instead of 40% as in the null). According to CeNGEN, rbm-26 expression in PLM is barely detected, thus depletion with a PLM-specific TIR1 should also be tested. To confirm the authors' identification of the cell marked "N" as the PLM cell body, co-expression of rbm-26 and a PLM-specific marker should be added. Rescue of the rbm-26 mutants with neuronal (and PLM-only) expression should be included to test sufficiency in PLM, and as a further control for potential artifacts of the AID system.

Fig 4. More rigorous quantification of the distribution of mitochondria along the axon should be included, not only total number, and it should be clarified what region of the axon the images are taken from. Including the AID-depletion strain with and without auxin would further add to the sense of rigor. For the mitoTimer experiments, why is RBM-26(L13V) not included and why do wild-type values differ ~5-fold between experiments (despite error bars being almost non-existent)? A more rigorous approach to standardizing imaging conditions may be needed. Positive controls using compounds that affect oxidation should be included. Measurements of individual mitochondria with standard deviations should be shown, rather than aggregate averages with error of proportion.

Fig 5. Additional positive and negative controls should be added, including additional rbm-26 alleles, the AID-tagged strain with and without auxin, and a rescued mutant.

Fig 6. Controls showing whether the Scarlet-tagged protein is functional are needed, to rule out dominant negative or toxicity-related effects.

Fig 8. Controls for other mitochondrial components need to be included. It is important to determine if the decrease in ribosomes is specific or reflects a general decrease in mitochondria. If there are fewer mitochondria as suggested in Fig. 4, then of course mitochondrial ribosomal protein levels are also reduced. Additional rbm-26 alleles should be included here as well. Is this effect dependent on the MALSU homolog? 4. Finally the authors should address concerns about image manipulation, which amplify the concerns about technical rigor outlined above. The image in Fig. 2A appears to have a black box placed over the lower-right portion of the field to hide some features. Black boxes also appear to have been placed over the tops of images in Fig. 4B and 4D and at the left of Fig. 6A, 6B, and 6C. While these manipulations probably do not affect the conclusions, they further undermine confidence in data integrity and experimental rigor.

Minor points.

1. C. elegans nomenclature conventions should be followed:
   • C. elegans gene names have three or four letters, thus the MALSU homolog cannot be named "malsu-1". Please have new gene names approved by WormBase BEFORE submitting for publication http://tazendra.caltech.edu/~azurebrd/cgi-bin/forms/gene_name.cgi
   • If two sequential CRISPR edits are made on the same gene then they should be listed as a compound allele, such as rbm-26(cue22cue25)
   • Genes on the same chromosome should not be separated with a semicolon, for example rbm-26(cue40) K12H4.2(syb6330)
2. Describing the defects as "neurodevelopmental" is misleading in the case of axon beading or degeneration. Similarly, there is no evidence for an "axon targeting" defect as stated in the abstract.
3. In Fig. 5A, the symbol that appears to correspond to F59C6.15 (lowest p-value) is a different size than the others and is colored as ncRNA, whereas WormBase annotates this gene as snoRNA.
4. In the Introduction, the last sentences of the first two paragraphs should be varied ("However, little is known about the [...] mechanisms that protect [...] during neurodevelopment.")
5. Why is RBM-26 protein running as a doublet at both sizes?
6. When showing the RBM-26 expression pattern (Fig. 3) please include a lower-magnification image of the entire animal.
7. It is confusing to refer to the RNA IP experiments as an "unbiased screen", which in C. elegans typically refers to a genetic screen.
8. The relationship between axon overextension, beading, and mitochondrial localization is not clear. What causal connection between these is being proposed? The causal connections between these phenotypes, if any, should be clarified experimentally. For example, if the axon extension defects develop before mitochondrial localization defects, then it is unlikely that mitochondrial defects cause axon overextension.
9. Please explain how to interpret the difference in axon beading in the two deletion alleles of the MALSU homolog (axon beading defects in tm12122 but not in syb6330). Is syb6330 not a null allele? Or are the defects in tm12122 due to other mutations in this strain background?
10. Are mitochondria reduced in number or mislocalized? If they are reduced in number, is this due to altered balance of fission/fusion?
11. In Fig. 3A-D, please keep the labels in the same position in all panels and do not alter brightness settings between single-color and merged panels.
12. The claim that rbm-26 acts cell-autonomously requires PLM-specific depletion and rescue experiments.

Referees cross-commenting I appreciate the use of the consultation session to resolve differences between reviewers, but in this case I fully agree with the content and tone of all the comments from the other reviewer -- I think our remarks are very well aligned!

Significance

The study engineers autism-associated variants in conserved residues of RBM27 into the C. elegans homolog RBM-26 and identifies neuronal phenotypes potentially relevant to autism and a potential molecular mechanism involving regulation of mitochondrial ribosome assembly.

The key claims of the study are 1} that autism-associated variants in RBM-26 decrease its protein expression; 2} that impaired RBM-26 function leads to a variety of defects in development and maintenance of a single neuron called PLM, including altered axonal localization of mitochondria; 3} that RBM-26 normally binds the mRNA for the C. elegans homolog of MALSU, a mitochondrial ribosomal assembly factor; 4} that loss of RBM-26 leads to overexpression of the MALSU homolog; and 5} that MALSU is required for some of the deleterious effects on the PLM neuron seen in RBM-26 mutants.

This study will be of interest to the autism research community because it bolsters the idea that variants in RBM27 are likely to disrupt gene function and to affect neuronal health. It will also be of interest to the broader cell biology community because it suggests an interesting potential nucleus-to-mitochondria signaling mechanism, in which a nuclear RNA-binding protein might regulate assembly of mitochondrial ribosomes.

My field of expertise is developmental biology in C. elegans.

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

The paper is overall convincing. However, a little more attention to data presentation and possibly the addition of at least another technique (see below) would greatly strengthen the findings.

As we hope to demonstrate below, we have taken steps to improve our manuscript on both fronts (data presentation and experimental evidence).

The absence of statistics catches immediately the eye. I am sure that the shown differences are statistically significant (thanks to the number of analyzed cells), but reporting the result of some statistical test would help the reader in identify the relevant data in a plot. This is somehow necessary considering that sometimes in the text something is deemed to be "significant" or "not significant", and I felt that I really needed that when looking at the plot in Fig. 3D.

To facilitate the interpretation of figures that contain data from multiple strains (such as the one mentioned by the reviewer), we have carried out a nonparametric single-step multiple comparison test (Games-Howell) to identify mutants whose means differ significantly from each other. To avoid overcrowding the figures, we have graphically summarized the p-values of all pairwise comparisons in a small matrix within the corresponding panel, and provided 99% confidence intervals and p-values of all differences in the Supplement.

Related to the previous point: for every N/C distribution analysis, a number of analyzed cells is reported. By the way it is written, it seems that the replication relies solely by the cells in that specific population, i.e.: each cell is treated as a replicate. At least I could not find if that is not the case in the legends or in the methods. I wonder what the results would be (and their significance) if each replicate would be a new assay on another population.

Cell populations exhibit significant variability in their phenotypic characteristics. Consequently, the quantification of a specific feature (e.g., the Sfp1 nuclear/cytoplasmic ratio) across a sample of cells from a given population results in a distribution rather than a single fixed value. For each quantification, we report the number of cells that were used to construct the corresponding distribution, i.e. the sample size. To compare samples from different populations (e.g., different Sfp1 mutant strains), we run them in parallel during microscopy experiments and compare their means, as described above. Throughout our study, we have tried to ensure that we quantify a sufficiently large number of cells to overcome cell-to-cell variability and enhance the reliability of our results.

In this context, the question of the reviewer is not entirely clear to us, as individual measurements of a sample are not replicates. However, one can replicate the entire experiment on a different day by re-growing the different strains, running microscopy, quantifying the new movies etc. In this sense, the experiments shown in the manuscript consist of single replicates, i.e. experiments that were carried out on the same day, with all the relevant mutants and controls quantified together. However, we have monitored many of our mutants multiple times over the course of our work. For example, Fig. 1 below shows replicates of the Sfp1 N/C ratio distributions at steady-state in the analog-sensitive (A) and wild-type (B) background, which were quantified several times across various experiments. While day-to-day variability in the empirical distributions of the same mutant exists to a small extent, it is quite small.

The scale of x axes in N/C ratio plots. Besides not being consistent throughout the figures, it originates from 1, visually enhancing the differences.

We believe the reviewer was referring to the y-axes, as the x-axes represent time. Summarizing the N/C ratio dynamics of different Sfp1 mutants has been challenging. First, the average N/C ratios at steady-state vary considerably across different mutants, as shown in the panels that summarize steady-state N/C ratios. To compare the magnitude and features of their responses, normalization is necessary. We chose to normalize the time series of each mutant to have a mean of 1 prior to the onset of a perturbation. This allows the normalized time series to represent the percentage-wise changes in the Sfp1 N/C ratio upon perturbation.

Using a common y-axis scale for all plots of N/C ratio dynamics not ideal, as some responses are subtler than others. Additionally, we do not believe that N/C dynamics across different figures need to (or should) be compared to each other. However, within a figure, panels that require comparison are placed in the same row and share the same y-axis scale. We believe that this approach optimizes data visualization and facilitates important visual comparisons.

Related to the previous point: it is evident from the plots that the N/C ratio is always positive, even in the most deficient of the analyzed mutants. This implies that a relevant fraction of Sfp1 is still nuclear. I thus wonder what the impact of these mutations would be on the actual function of Sfp1. For this reason, I feel that qPCR evaluation of transcripts of Sfp1 target genes is particularly needed. Since lack of Sfp1 is known to yield some of the smallest cells possible, it would also be cool to have an estimate of the size of mutants where Sfp1 is less nuclear. These analyses could confer phenotypical relevance to the data, but would also help in assessing a currently unexplored possibility, that phosphorylation events by PKA influence Sfp1 function besides its localization, i.e.: the still somehow nuclear fraction is not as functional as wt Sfp1 in promoting transcription.

It is indeed the case that the recorded N/C ratios are larger than 1 in all strains that we have monitored. We have never observed an N/C ratio smaller than 1 using widefield microscopy for two main reasons: first, out-of-focus light from the cytosol above and below the nucleus is added to the nuclear signal, causing the nuclear signal to always be non-zero, even for predominantly cytosolic proteins. Second, both in- and out of focus vacuoles are devoid of the fluorescent protein fusions that we quantify, which reduces the average brightness of the cytosol. For these reasons, even when a protein is largely cytosolic, the average N/C ratio over a cell population is no lower than around 1.5. Keeping these points in mind, one can observe that our most delocalized Sfp1 mutants have an N/C ratio that is around 1.6-1.7, which is very close to the lower limit. This means that these Sfp1 mutants are largely cytosolic, and the nuclear fraction (if non-zero) is quite small.

We agree that assessing the phenotypic relevance of Sfp1 mutations is of interest. However, this was impossible with our original strains, as we introduced each Sfp1 mutant as an extra copy in the HO locus while leaving the endogenous Sfp1 locus intact. This was done in order to avoid any phenotypic changes that might result from changes in Sfp1 activity.

To address the suggestion of the reviewer, we therefore deleted the endogenous Sfp1 copy in strains carrying sfp1PKA2A, sfp1PKA2D and sfp113A, leaving only the mutated Sfp1 copy at the HO locus. Surprisingly, the growth rate and drug sensitivity (determined by halo assays) of these single-copy mutants did not differ much in comparison to the mutants carrying the functional Sfp1 copy and from the wild-type (Supp. Figs. 4J and 7). This observation aligns with findings for the single-copy sfp1-1 mutant in [Lempiäinen et al. 2009], which corresponds to sfp1TOR7A in our work. [Lempiäinen et al. 2009] had suggested that Sch9 compensates for the loss of Sfp1 activity via a feedback mechanism, which could explain our results as well. If this is the case, acute depletion of wild-type Sfp1 could unveil transient changes in cell growth, before the compensatory effect of Sch9 was established. Unfortunately, we were unable to efficiently degrade wild-type Sfp1 carrying a C-terminal auxin-inducible degron. Instead, we followed the same approach with [Lempiäinen et al. 2009] and deleted SCH9.

As we describe in the last section of Results, the difference was dramatic for sfp113A __mutants, which were extremely slow-growing in the absence of Sch9 (doubling time was around 4 hours, but it was hard to estimate because we could not grow the cells consistently). Interestingly, SCH9 deletion had a negative impact on sfp1__PKA2D __but not sfp1__PKA2A __cells (__Supp. Fig. 7). Overall, these results demonstrate that Sch9 can compensate for loss of Sfp1 activity, which makes it challenging to study the impact of Sfp1 mutations on cellular phenotypes.

To further understand to what extent Sch9 compensates for loss of Sfp1 phosphorylation, we carried out RNA-seq on WT and cells carrying a single copy of sfp113A (with the endogenous SFP1 copy removed). Despite the fact that sfp113A __grow as well as WT, RNA-seq picked up several differentially expressed genes related to amino acid biosynthesis. This surprising finding is presented in the last section of Results, and in __Supplementary Figures 8, 9 and 10. We explore the relevance of these results and their connection with past literature on Sfp1 and Sch9 in the Discussion section.

I found some typos here and there, and it would greatly help to report them if in the manuscript line numbers were included.

We apologize for the typos. We have tried to eliminate them, and we have also added line numbers to the manuscript.

Reviewer 2

There is no biochemical evidence presented that the putative PKA sites (S105 and S136) are genuinely phosphorylated by PKA. The fact that they match the PKA consensus motif, alone, does not guarantee this. In order to claim that they are looking at the effect of PKA by mutagenizing these residues, the authors have to demonstrate the PKA-dependency of S105 and S136 phosphorylation by, for example, mass spec experiments or western blotting with phospho-specific antibodies (Cell Signaling Technology #9624 for example). Also, does the band-shift caused by PKA inhibition (Fig 3C) is canceled by the S105A/S136A mutation?

We took several actions to demonstrate that the putative PKA sites are indeed phosphorylated by PKA. We first tried to detect Sfp1 phosphorylation using the antibody mentioned by the reviewer, but failed as the sensitivity of this antibody appears to be quite low. On the other hand, mass spectrometry did not produce the right fragments to detect the sites of interest. We therefore resorted to an in vitro kinase assay using [γ-32P]ATP together with purified PKA and Sfp1. Unfortunately, bacterial overexpression of MBP-tagged Tpk1, Tpk2 and Tpk3 (the catalytic subunits of PKA) was quite challenging and we were unable to produce soluble protein. We therefore resorted to commercially available bovine PKA (bPKA, PKA catalytic subunit, Sigma-Aldrich 539576), which shows high homology to the yeast Tpk kinases [Toda et al. 1987]. Moreover 87% of bPKA substrates have been shown to also be Tpk1 substrates [Ptacek et al. 2005], and bPKA has been used to identify new Tpk substrates in budding yeast [Budovskaya et al. 2005__]. As we show in the revised manuscript, bovine PKA does phosphorylate Sfp1. Moreover, phosphorylation is reduced by 50% in the double S105A, S136A mutant (Fig.1F), and becomes undetectable in the 13A mutant__ (Supp Fig. 6). Together with the rapid response of Sfp1 localization to acute PKA inhibition which we had already reported, we believe that these results provide strong evidence that Sfp1 is a direct PKA substrate, and that the two phosphosites that we identified are functional.

As the above in vivo experiments do not exclude S105/S136 phosphorylation by other kinases downstream of PKA, in order to claim the direct phosphorylation, the authors need in vitro PKA kinase assay. These biochemical experiments are not trivial, but I think absolutely necessary for this story.

One cannot exclude that S105/S136 are also phosphorylated by other kinases of the AGC family (note that [Lempiäinen et al. 2009] has already excluded Sch9). However, as we hope to have shown, PKA indeed phosphorylates Sfp1. Examining if other kinases besides PKA and TORC1 target Sfp1 is a very interesting question that should be addressed in future work.

The authors only look at the localization of Sfp1. To assess its functionality and so physiological impact, it would be informative to measure the mRNA level of target ribosomal genes in various Sfp1 mutants they created.

As we described in our response to Reviewer 1 above, we did perform RNA-seq on WT and cells carrying a single copy of sfp113A. We observed a notable absence of differentially expressed ribosomal genes and ribosome-related categories in the GO analysis (Supp. Figs. 8, 9 and 10). Together with our observations on SCH9 deletion (Supp. Fig. 7), these results suggest that Sch9 can largely compensate for the loss of Sfp1 activity. On the other hand, the emergence of differentially expressed amino acid biosynthesis genes is a finding that merits further investigation, as it connects with previous observations made with Sch9 deletion mutants and the [ISP+] prion form of Sfp1 (cf. Discussion).

In the experiments using analog-sensitive PKA (Fig 1D and E for example), they directly compare wildtype-PKA versus analog sensitive-PKA, or with 1-NM-PP1 versus without 1-NM-PP1. This makes interpretation difficult, particularly because 1-NM-PP1 itself has a significant impact even in the wild PKA strain. The real question is the difference between wild-type Sfp1 versus mutant Sfp1. In the current form, they compare Fig 1D versus 1E, these two do not look like a single, side-by-side experiment. They should compare wild-type Sfp1 versus mutant Sfp1 side-by-side.

Figure 1D shows that 1-NM-PP1 has a transient off-target effect on Sfp1 localization in WT cells, which could also affect Sfp1 mutants. This observation prompted us to use wild-type PKA as a control when testing the effect of 1-NM-PP1 on sfp1PKA2D in cells carrying PKAas (Figure 1E). As Fig. 1E shows, the effect of 1-NM-PP1 on sfp1PKA2D localization in PKAas cells is quite similar to the off-target effect in cells carrying sfp1__PKA2D __and wild-type PKA. This behavior of sfp1__PKA2D __is clearly different from the response of wild-type Sfp1 to PKAas inhibition, which results in sustained delocalization. We have made the latter observation repeatedly, both in this study and our previously published work [Guerra et al. 2021].

In Figure 3, the argument around the additive effects of PKA and TORC1 is confusing. The authors say they are additive referring Figure 3E, but say they are not additive referring Figure 3B. Which is true? In fact, Figure 3B appears to show an additive effect as well.

We did not use the word "additive" in the text, because we find it difficult to interpret. Instead, we state that PKA and TORC1 appear to control Sfp1 phosphorylation independently of each other. PKA and TORC1 phosphorylation converges to the same response, affecting Sfp1 localization. It appears that loss of either kinase delocalizes Sfp1, while loss of both kinases may only have a small additional effect.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

The authors investigated how Sfp1, a transcription factor for ribosomal genes, integrates signals from TORC1 and PKA pathways. They did so by analyzing the nuclear localization of the GFP-tagged Sfp1 variants harboring unphosphorylatable or phosphomimetic mutations on either TORC1 target sites, putative PKA target sites, or a combination of both. This approach was complemented by examining the effect of pharmacological inhibition of either pathway on Sfp1 localization. The obtained results support that TORC1 and PKA independently promote nuclear localization of Snf1, provided that the putative PKA sites are genuinely PKA sites (see Major point). In course of their investigation, the authors made two novel findings about the regulatory mechanism of Sfp1 localization. First, they identified the 98-106aa region as a nuclear export signal (NES). Because this region overlaps with a putative PKA site, it is conceivable that PKA regulates Sfp1 localization via altering the functionality of NES. In addition, they found that the nuclear localization of Snf1 requires its C-terminal zinc fingers, although this domain appears to work independently from TORC1- and PKA-dependent regulations.

Major points:

1. There is no biochemical evidence presented that the putative PKA sites (S105 and S136) are genuinely phosphorylated by PKA. The fact that they match the PKA consensus motif, alone, does not guarantee this. In order to claim that they are looking at the effect of PKA by mutagenizing these residues, the authors have to demonstrate the PKA-dependency of S105 and S136 phosphorylation by, for example, mass spec experiments or western blotting with phospho-specific antibodies (Cell Signaling Technology #9624 for example). Also, does the band-shift caused by PKA inhibition (Fig 3C) is canceled by the S105A/S136A mutation?
2. As the above in vivo experiments do not exclude S105/S136 phosphorylation by other kinases downstream of PKA, in order to claim the direct phosphorylation, the authors need in vitro PKA kinase assay. These biochemical experiments are not trivial, but I think absolutely necessary for this story.

Minor points:

1. The authors only look at the localization of Sfp1. To assess its functionality and so physiological impact, it would be informative to measure the mRNA level of target ribosomal genes in various Sfp1 mutants they created.
2. In the experiments using analog-sensitive PKA (Fig 1D and E for example), they directly compare wildtype-PKA versus analog sensitive-PKA, or with 1-NM-PP1 versus without 1-NM-PP1. This makes interpretation difficult, particularly because 1-NM-PP1 itself has a significant impact even in the wild PKA strain. The real question is the difference between wild-type Sfp1 versus mutant Sfp1. In the current form, they compare Fig 1D versus 1E, these two do not look like a single, side-by-side experiment. They should compare wild-type Sfp1 versus mutant Sfp1 side-by-side.
3. In Figure 3, the argument around the additive effects of PKA and TORC1 is confusing. The authors say they are additive referring Figure 3E, but say they are not additive referring Figure 3B. Which is true? In fact, Figure 3B appears to show an additive effect as well.

Significance

TORC1 and PKA are major pro-growth signaling pathways widely conserved in eukaryotes, that often converge on the same target proteins. How the information from the two pathways is integrated is an interesting question, which the authors directly and meticulously address here with yeast Sfp1 as an example. Provided that they can demonstrate that the putative PKA sites are the real ones (this is really important- TORC1 sites are already known, what is new here is PKA sites), their data and conclusion should be of interest to the signal transduction field.

Their additional discovery of NES and the role of zinc fingers in Sfp1 localization should be of interest to those working on Sfp1, or transcriptional regulation of ribosomal genes in general.

My area of expertise: yeast TOR

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary:

In this paper, Vuillemenot and Milias-Argeitis investigate in budding yeast the role of Protein Kinase A (PKA) in regulating through phosphorylation the subcellular localization of the transcription factor Sfp1, known for controlling transcription of RP genes. Sfp1 is very well known for being regulated by another signaling pathway, centered on the kinase TORC1. Thus, regulation of Sfp1 by PKA raises the intriguing possibility of a downstream crosstalk between the two pathways. Indeed, the authors find that Sfp1 is regulated by PKA independently from TORC1. In the study, the authors employ mainly single-cell microscopy to monitor the nucleo/cytosolic localization of Sfp1 mutants, an experimental set-up they established in a previous paper, with some contribution by PhosTag bandshift assays.

Major comments:

The paper is overall convincing. However, a little more attention to data presentation and possibly the addition of at least another technique (see below) would greatly strengthen the findings. Summarizing my major concerns: - The absence of statistics catches immediately the eye. I am sure that the shown differences are statistically significant (thanks to the number of analyzed cells), but reporting the result of some statistical test would help the reader in identify the relevant data in a plot. This is somehow necessary considering that sometimes in the text something is deemed to be "significant" or "not significant", and I felt that I really needed that when looking at the blot in Fig. 3D. - Related to the previous point: for every N/C distribution analysis, a number of analyzed cells is reported. By the way it is written, it seems that the replication relies solely by the cells in that specific population, i.e.: each cell is treated as a replicate. At least I could not find if that is not the case in the legends or in the methods. I wonder what the results would be (and their significance) if each replicate would be a new assay on another population. - The scale of x axes in N/C ratio plots. Besides not being consistent throughout the figures, it originates from 1, visually enhancing the differences. - Related to the previous point: it is evident from the plots that the N/C ratio is always positive, even in the most deficient of the analyzed mutants. This implies that a relevant fraction of Sfp1 is still nuclear. I thus wonder what the impact of these mutations would be on the actual function of Sfp1. For this reason, I feel that qPCR evaluation of transcripts of Sfp1 target genes is particularly needed. Since lack of Sfp1 is known to yield some of the smallest cells possible, it would also be cool to have an estimate of the size of mutants where Sfp1 is less nuclear. These analyses could confer phenotypical relevance to the data, but would also help in assessing a currently unexplored possibility, that phosphorylation events by PKA influence Sfp1 function besides its localization, i.e.: the still somehow nuclear fraction is not as functional as wt Sfp1 in promoting transcription.

Minor comments:

Experimental issues and suggestions on data presentation are reported in the major comments section, since I felt those were major issues.

Just a side remark: I found some typos here and there, and it would greatly help to report them if in the manuscript line numbers were included.

Significance

The finding that both PKA and TORC1 impinge on Sfp1, and therefore presumably on protein synthesis, is a valuable conceptual addition to the field of cell signaling. The audience potentially interested by the findings of the study include not only yeast cell biologists, but also computational biologists interested in modeling crucial cellular processes. One example is the regulation of cell size, where TORC1, PKA and Sfp1 are already know to play a role, but were potentially missing a crosstalk link.

As requested by Review Commons, I specify that my expertise is on TORC1/AMPK/PKA pathways, on their crosstalk and their regulation by metabolic intermediates.

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

The nucleus is recognised as a core component of mechanotransduction with many mechano-sensitive proteins shuttling between the nucleus and cytoplasm in response to mechanical stimuli. In this work, Granero-Moya et al characterise a live florescent marker of nucleocytoplasmic transport (NCT) and how it responds to a variety of cues. This work follows on from the authors previous study (Andreu 2022) where they examined the response of passive and active NCT to mechanical signalling using a series of artificial constructs. One of these constructs (here named Sencyt) showed a differential localisation depending on substrate stiffness, accumulating in the nucleus on stiffer substrates (which the authors previously showed was due to differences in mechano-sensitivity of passive versus facilitated NCT). Here the authors use Sencyt as a tool to probe how different cues affect NCT and thus nuclear force-sensing in two different cell lines (one epithelial, one mesenchymal). *

They have established a 3D image segmentation pipeline to measure both the nuclear/cytoplasmic ratio of Sencyt and 3D nuclear shape parameters. As a proof-of -principle, they show that hypoosmotic shock (which inflates the nucleus and would be expected to increase nuclear tension) and hyper-osmotic shock (which shrinks and deforms the nucleus) alter Sencyt nuclear-cytoplasmic ration as expected. They then show that inhibiting acto-myosin, which would be expected to block force transduction to the nucleus, reduces NCT, although interestingly this is without any changes to nuclear morphology. They then examine how cell density affects NCT and show that Sencyt localisation correlates only weakly with density but much more strongly with nuclear deformation (especially as measured by solidity). This is surprising considering that mechano-sensitive transcription factors such as YAP have been shown to exit the nucleus at high cell densities. Therefore, the authors directly compare Sencyt and Yap nucleo/cytoplasmic localisation and show that Sencyt behaves differently to YAP with YAP localisation correlating strongly with cell density. This reveals an added layer of complexity in YAP regulation beyond pure changes to NCT.* Major points *

The data presented throughout this work are high quality and rigorous. The controls used are appropriate (including the use of a freely diffusing mCherry to illustrate the specificity of the Sencyt probe in osmotic shock experiments - figure S2). Experiments are properly replicated and the statistical analysis is appropriate. The data are beautifully presented in figures and the manuscript is well written and very clear. Overall this is a high quality work.

We thank the reviewer for the positive assessment of the manuscript.

* The discussion is careful and the conclusions are supported by the data. My only small concern is that the authors place too much emphasis on how this work is in 'multicellular systems' as opposed to their previous work in single cells (for example "Here, we demonstrate that mechanics also plays a role in multicellular systems, in response to both hypo and hyper-osmotic shocks, and to cell contractility. L212). Cell density is only controlled in figures 3 and 4 and in some of the earlier experiments, cells look quite sparse (eg Figure 2). It's also debatable how far a monolayer of cancer cells, which lack contact inhibition of growth, is a multicellular system. Furthermore, the authors don't specifically look at cell/cell adhesion or observe major differences between the epithelial or mesenchymal lines. For this reason, the authors should tone down this discussion before publication. *

• *

We agree with the reviewer that properly assessing cell-cell adhesion is important in the context of the work. To this end, we have stained for E-cadherin in both cell lines. As expected and as described previously, the results confirm that MCF7 cells do have clear cadherin-mediated cell-cell adhesions, with a cadherin staining localized specifically in cell-cell junctions. Also as expected, C26 cells show much lower cadherin expression, without a clear pattern. Further confirming this difference, MCF7 cells show clearly distinct actin organizations in their apical and basal sides, whereas C26 cells do not. Thus, we believe that the two cell models do represent a reasonable assessment of epithelial versus mesenchymal phenotypes, in a multicellular context. The data are presented in new supplementary fig. 1, and discussed in page 3 of the manuscript (first paragraph). We have also included a paragraph in the discussion to comment on the differences between cell types (page 7, 2nd paragraph).

* Optional experimental suggestions: For me, the most compelling finding is that nuclear deformation has a greater correlation with NCT than cell density and that this is different from the behaviour of YAP. To cement the importance of nuclear deformation, the authors could induce deformation in single cells, for example by culture on very thin micropatterned lines and assess the localisation of Sencyt and YAP. It would also be interesting to assess the role of force transduction in this context or in different densities by removing actin, which affects NCT without inducing nuclear shape changes. These functional experiments would allow the authors to draw stronger conclusions about the role of nuclear shape and deformation but they aren't necessary for publication. *

• *

This is a very interesting suggestion. Following the reviewer's advice, we have now carried out experiments in which we have seeded cells on micropatterns of different sizes, and measured both sencyt and YAP ratios. In C26 cells, we have found as expected that increasing spreading leads to progressive nuclear deformation (as measured through nuclear solidity) and progressive increase in both sencyt and YAP ratios. Interestingly, cell spreading in MCF7 did not affect nuclear solidity, sencyt ratios, or YAP ratios. This further confirms the relationship between nuclear deformation and nucleocytoplasmic transport, and shows as well that different cell lines have different sensitivities. The lack of response of MCF7 cells is consistent with the lower sencyt response, and lower sencyt/nuclear shape correlation measured in fig. 4. It suggests that MCF7 cells may have mechanisms to shield the nucleus from deformation, something which we have reported in a different context (Kechagia et al., Nat. Mater. 2023). The new results are reported in new fig. 3, and supplementary fig. 8, and discussed in pages 5 (1st paragraph) and 6 (1st paragraph) of the manuscript results.

• *

Minor points

* - I'd like to see better examples of 3D reconstructions of nuclei (ie fig 1C but bigger) in different conditions. This is especially important in figure 3 where it would be helpful to see examples of nuclei with high or low solidity. The differences in oblateness are clear to see from the images in 3a and 3f but solidity could be better illustrated. *

• *

We have now added 3D reconstructions as requested, which illustrate the nuclear shape changes that take place. This is shown in figs. 1, 4 (which corresponds to figure 3 in the previous version of the manuscript), s3, and s7.

*

• Where Sencyt index is plotted, it would be clearer to add labels to at least figure 1 which indicate whether it is more cytoplasmic or nuclear. *

• *

We have done this as requested in figure 1.

* Reviewer #1 (Significance (Required)): *

* In this work, Granero-Moya et al characterise a new tool for measuring NCT and show that it is mechanically regulated. Given the importance of NCT in mechano-transduction, this tool will be a great asset to the mechano-biology community and will likely be adopted by multiple groups in the future. The findings about the effects of cell density on NCT and differences from YAP are interesting but could be further fleshed out. This work is likely to be of greatest interest to a specialised audience working in the fields of mechano-biology and nuclear transport. *

• *

We thank the reviewer for the positive assessment.

* *

• *

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

* The study conducted by Granero Moya and colleagues describes the application of a synthetic protein which is observed to enter the nucleus in response to mechanical strains, rather than being influenced by cell density. However, the novelty of this work is minimal since the conceptual framework and the utilization of this identical or similar tool have been previously reported by the same team in earlier publications. *

• *

We respectfully disagree with the assessment of the reviewer. Please see below for a detailed response regarding novelty.

• *

*In their experiments, they employ this GFP-based sensor, referred to as Sencyt, in cells subjected to osmotic shocks. These shocks are highly stressful and impact a range of cellular processes, including stress response pathways MAPK and others; Osmoregulatory pathways; cell cycle regulations, autophagy and death pathway; ion channel regulations and others. The second findings are on cells treated with a combo of drugs affecting the actin cytoskeleton. The justification for using a combination of two specific drugs remains unclear, as the study does not adequately explain the rationale behind this choice. Additionally, there is a lack of information regarding the full range of targets these drugs affect. This raises questions about the comprehensiveness and applicability of the findings, as understanding the complete scope of the drugs' targets is crucial for interpreting the results within a minimal frame of physiological context. *

• *

The two drugs used are paranitroblebbistatin (a photostable version of blebbistatin) and Ck666. We apologize for not explaining in more detail the action of these drugs, both of which have been characterized and used extensively in the literature. Paranitroblebbistatin binds to myosin, preventing its ATPase activity and therefore impairing actomyosin contractility (https://doi.org/10.1002/anie.201403540). It acts on different myosin isoforms, including non-muscle myosin II, the main type of myosin responsible for actomyosin contractility in non-muscle cells. CK666 binds to and inhibits arp2/3, a protein responsible for nucleating branched actin (https://doi.org/10.1016/j.chembiol.2013.03.019). This impairs lamellipodial formation and therefore cell spreading (see for instance https://doi.org/10.1371/journal.pone.0100943).

The rationale for using both drugs in combination was explained in page 4 of the manuscript. In our previous work, we determined that myosin inhibition with blebbistatin is not sufficient to inhibit nuclear mechanotransduction. Indeed, in an epithelial context, we observed that due to reduced contractility, blebbistatin-treated epithelial cells in fact spread more on their substrate. This leads to more deformed (flattened) nuclei, leading to the counterintuitive result that YAP nuclear localization increases rather than decreases. If cell spreading is impaired by interfering with branched actin nucleation, then this spreading is prevented, and the combination of drugs leads to reduced nuclear deformation, and reduced YAP nuclear localization (see supplementary fig. 7 in Kechagia et al, Nat. Mater. 2023, https://doi.org/10.1038/s41563-023-01657-3). Similar results had been published previously by the group of Clare Waterman (https://doi.org/10.1074/jbc.M115.708313).

Thus, the combination of drugs was designed to ensure that we were impairing nuclear mechanotransduction. Of course, we agree with the reviewer that all perturbations have potential side effects. Osmotic shocks will affect a range of cellular processes (as mentioned in the discussion of the manuscript), and any drug treatment can potentially have off-target effects. However, the fact that two orthogonal perturbations with different potential side effects (osmotic shocks versus actomyosin-targeting drugs) lead to the same effects in sencyt strongly suggests that the effect is mediated by mechanics, and not other factors. To reinforce this, we have now added an additional mechanical manipulation: seeding cells on micropatterned islands of different sizes. As spreading increases, cells are known to increase actomyosin contractility, and nuclear deformation (https://doi.org/10.1529/biophysj.107.116863, https://doi.org/10.1073/pnas.0235407100, https://www.nature.com/articles/ncomms1668, https://doi.org/10.1073/pnas.1902035116). As expected, nuclear solidity, sencyt ratios, and Yap ratios all increased with cell spreading. Interestingly, this occurred only for C26 and not MCF7 cells, where no changes were measured in solidity, sencyt, or YAP. The lack of response of MCF7 cells is consistent with the lower sencyt response, and lower sencyt/nuclear shape correlation measured in fig. 4. It suggests that MCF7 cells may have mechanisms to shield the nucleus from deformation, something which we have reported in a different context (Kechagia et al., Nat. Mater. 2023).

The new results are shown in figs. 3 and s8. We have also expanded the explanation of drug treatments in page 4 (3rd paragraph).

* The novelty is on the specificity of this synthetic fusion protein for these manipulations and not on cell density. Yet, the reasons behind this selective response remain unexplained, potentially attributable to the unique characteristics or sensitivity thresholds of their synthetic probe. As comparison, YAP localization and this is sensitive to both inputs, but this is also already published (fig4). The focus is anyway on Sencyt for which they offer simple observations and quantifications. *

• *

The main novelty of the work lies in the characterization of the role of nucleocytoplasmic transport in mechanotransduction, in the context of multicellular systems. We and others had shown that nucleocytoplasmic transport responds to mechanical force in the context of single cells (see for instance Andreu et al. 2022 from our group, but also https://doi.org/10.1126/science.abd9776 from the Martin Beck group). However, to what extent this applies to multicellular systems was unknown. It is true that in multicellular systems, the response of YAP and other mechanosensitive transcription factors has been characterized (such as in our Elosegui-Artola 2017 paper, mostly done at the single cell level but including one figure panel on epithelial cell monolayers). The reviewer argues here and in the consultation comments with other reviewers (see below) that this demonstrated the role of nucleocytoplasmic transport in multicellular systems. However, we respectfully disagree. As also noted by reviewer 3 in the consultation, the response of YAP, and of any transcription factor, may include effects on nucleocytoplasmic transport, but will also likely include effects caused by the complex biochemical signalling pathways that regulate them. Disentangling such effects requires a sensor that only responds to nucleocytoplasmic transport, and this is precisely what Sencyt provides.

The reviewer also states that our manuscript does not explain why sencyt responds to mechanics and not cell density. We disagree: sencyt responds to mechanics for the reasons explained in our previous work (Andreu et al., Nat. Cell Biol. 2022), and there is no reason to expect a specific response to cell density. In this regard, we don't think there are any sensitivity thresholds to detect cell density, as the probe is not designed to sense this parameter in the first place. The fact that YAP responds to both mechanics and cell density shows that the response to density cannot be merely explained by mechanics, and is rather due to signalling through other means. Of course, we agree that we do not explain the mechanism by which YAP senses cell density, but we think this lies clearly out of the scope of our manuscript.

In terms of novelty, our work also characterizes a tool to assess nucleocytoplasmic transport live in cells. We agree with the reviewer that the specific construct had been reported in our previous paper, but it had not been characterized in detail. This is done here, enabling its use by the community as a tool to measure nucleocytoplasmic transport in any context, be it related to mechanics or not.

• *

When reviewing the figures presented, I find it challenging to detected marked differences, despite their quantitative data suggesting otherwise.

• *

We assume here that the reviewer refers to differences in sencyt nuclear localization, that is, the sencyt index. We have now checked the example images showing changes in sencyt index, in figures 1 and 2. In figure 1, the example cells under hypo-osmotic shocks increase their sencyt index from 1.2 to 1.45 (C26). In figure 1, the example cells under hyper-osmotic shocks decrease their sencyt index from 0.9 to 0.3 (MCF7) and from 1.4 to 0.5 (C26). In figure 2, the example cells increase their sencyt index upon drug washout from 0.2 to 1.4 (MCF7) and from 0 to 0.9 (C26). Of course, these individual values don't reflect exactly average values, but they do reflect the reported average trends and their magnitudes faithfully. Here we note that even though sencyt changes with the different treatments, it is always more nuclear than cytosolic (sencyt index >0, as it has an NLS). Thus, to the naked eye, sencyt always seems to show a "bright" nucleus, and it is hard to intuitively see changes in its localization. Further, we also note that osmotic shocks lead to overall changes in fluorescence levels due to volume changes (as GFP molecules get diluted or concentrated in hypo or hyper osmotic shocks, respectively). This does not affect ratiometric quantifications as assessed with our mcherry control, but means that changes in ratios are hard to see by eye. To help in this visualization, we have now changed the images from green to grayscale, which is better perceived by the human eye. We have also specified the issue of fluorescence intensity changes in the legend of the figure.

In addition to this, we have seen that there is indeed a case in which examples were not following average trends. In the case of hypo-osmotic shocks in figure 1, example MCF7 cells were barely changing their sencyt index with treatment. We apologize for choosing this non-representative image for the figure, we have now changed the figure to show more representative cells.

• Furthermore, the study attempts to correlate the behavior of Sencyt with the nuclear geometric parameter of solidity, a connection that seems to lack a clear basis in cell biology and could potentially lead to misconceptions. *

• *

Mechanical effects on nucleocytoplasmic transport are mediated by mechanical tension application to nuclear pores, which are embedded in the nuclear membrane (nuclear envelope). Whereas nuclear envelope tension is very challenging to measure directly, it can be indirectly related to nuclear shape. Indeed, a tense membrane will tend to even out membrane irregularities and appear rounded, whereas a membrane under low tension will tend to show wrinkles. Nuclear solidity is a geometric parameter that compares actual nuclear volume to the volume of the convex hull (intuitively, the volume of the smallest wrinkle-free object containing all of the nucleus). Thus, it is the geometric parameter that best reflects the presence of wrinkles, folds or irregularities, and as such the one that should best correlate to membrane tension. Of course, this correlation is not perfect, and there could be many situations in which changes in membrane tension may not directly affect nuclear solidity. But we do believe that solidity is the geometrical parameter that should best reflect membrane tension, and this is why we focus on it. Consistent with our hypothesis, solidity is the geometrical parameter that best correlates with sencyt. To further clarify this, we now explain this rationale in detail in page 4 of the manuscript (1st paragraph).

* Reviewer #2 (Significance (Required)): *

* In sum, I think the MS is of interest for a very specialistic audience. There are no clear interpretations. The work is done in one or two cellular model systems in vitro; and the general significance of these observations is of very limited impact and no novelty. *

We strongly disagree. The study is done on two cellular models, one with epithelial and the other with mesenchymal phenotype, and thus highly relevant for multicellular systems. Following suggestions by reviewers 1 and 2, we have now characterized the epithelial/mesenchymal behaviour of the cell types in detail (see supp. fig. 1). The results are novel in that they demonstrate the role of nucleocytoplasmic transport in multicellular systems, something which as argued above had not been done before. The difference with YAP, and the disentanglement between transport and signalling, is also novel. Finally, we believe the manuscript will be impactful because of this novelty, but also because of the availability of sencyt as a tool for the community. In fact, since placing this manuscript in biorxiv, we have received many requests (directly and through addgene) to share sencyt, which is currently being used in several labs across the world.

• *

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

• *

In this very well-written manuscript, Pere Roca-Cusachs and colleagues investigated the response of nucleocytoplasmic transport (NCT) to mechanical stress and tested whether this response is similar in epithelial and mesenchymal cells using a combination of quantitative approaches. This study builds upon their earlier findings, which elegantly demonstrated that NCT is sensitive to mechanical forces transmitted to the nuclear membrane. Using a similar approach to their recent work, they quantitatively analyzed NCT and compared the two cell types using various treatments that impact nuclear membrane tension. The study is straightforward and experimentally sound, with an adequate number of replicates and independent experiments. While one might consider the limitations given their previous work, none have demonstrated that NCT is mechanosensitive in epithelial cells. Additionally, they provide a simple approach to measure NCT, which should be of interest in the field. However, it is unclear how the authors defined the epithelial phenotype in this work and whether they solely based this characterization on the tissue/cell's origin. Epithelia can be defined ultrastructurally with reference to their apico-basal polarity and specific cell-cell junctions (Alberts et al., 1994; Davies and Garrods, 1997). Changing cell density should affect cell/cell adhesion, but the authors provide no evidence that the cells tested in the study are attached to their neighbors on all sides and form an epithelium. While I recognize that the objective of this study is not to mimic the in vivo behavior of epithelial tissue, the authors should at least ensure that cells form a monolayer by quantitatively assessing cell-cell junctions (or they should adjust their conclusions adequately). This control is specifically important for Figure 3 and 4, whose objective is to test the impact of cell/cell contacts. But it would also be important to provide this essential control for Figure 1 and 2, as it is unclear from the images provided if MCF7 cells are forming an epithelium (and form cell/cell junctions).

• *

We thank the reviewer for the positive assessment of our work. We fully agree with the reviewer that properly assessing cell-cell adhesion is important in the context of the work. To this end, we have stained for E-cadherin in both cell lines. As expected and as described previously, the results confirm that MCF7 cells do have clear cadherin-mediated cell-cell adhesions, with a cadherin staining localized specifically in cell-cell junctions. Also as expected, C26 cells show much lower cadherin expression, without a clear pattern. Further confirming this difference, MCF7 cells (but not C26 cells) show a clear apico-basal polarization, with distinct actin organizations in their apical and basal sides. Thus, we believe that the two cell models do represent a reasonable assessment of epithelial versus mesenchymal phenotypes, in a multicellular context. The data are presented in new supplementary fig. 1. We have also included a paragraph in the discussion to comment on the differences between cell types (page 7, 2nd paragraph).

• Reviewer #3 (Significance (Required)): *

• *

The mechanosensitivity of NCT is an important question central to many aspects of cell biology. One might consider the impact of the proposed work limited, given their previous research. However, none have demonstrated that NCT is mechanosensitive in epithelial cells, making it a crucial question that needs to be addressed. Additionally, they provide a simple approach to measure NCT, which should be of interest to a broad audience.

We thank again the reviewer for this positive assessment.

• *

*Referees cross-commenting *

* Here comments from all 3 reviewers are reported *

* Reviewer 1: *

* I disagree with R2's comment that there is 'no novelty' here. Although this work is going to be of greater interest to a specialised rather than general audience, it characterises in depth a simple tool to measure NCT which will be useful for mechanobiology field. Also, using 'two cellular model systems in vitro' is very standard in the field when assessing subcellular processes like NCT. Using this approach in vivo would be very interesting but challenging and would be an entirely different study . *

• *

*I agree with R2's comments that the authors should better justify their combination of two actin inhibitors and R3s point on better assessing cell/cell junctions. *

• *

We thank the reviewer for these comments. Both issues have been addressed, as described in the response to reviewers above.

* Reviewer 2 *

* About Reviewer 3's comments, I believe it's a stretch to highlight the strength and novelty based on "NCT's mechanosensitivity in epithelial cells has not been demonstrated,". There are thousands of papers on the Hippo pathway, that is known to be mechanosensitive, on the regulation of YAP, that enters in the nucleus in Hippo inhibited conditions and exits to the cytoplasm in Hippo induced cells, including downstream of mechanical signals. The phenomenon of nuclear-cytoplasmic shuttling being a common event from neurons to endothelial and multiple types of epithelial, immune, and fibroblast cells is already established through NCT of this and other endogenous proteins. This is simply an accepted fact. Then, The Nature cell Biology 2022 was offering a very general claim. No warning that conclusions could have been cell type specific. In the Artola 2017 Cell paper they also showed NCT in mammary epithelial cells. We should definitively conclude that NCT's mechanosensitivity in epithelial cells has been well demonstrated. *

• *

We disagree with this assessment, for the same reasons also exposed by reviewer 3 below. Previous work on YAP and other transcription factors cannot be seen as a demonstration of the role of nucleocytoplasmic transport per se. The localization of any transcription factor is highly regulated by complex signalling pathways, and can be affected by many factors. One of them is nucleocytoplasmic transport, but signalling events (for instance through phosphorylation) could change localization by promoting binding to cytosolic or nuclear binding partners, by promoting protein degradation, by masking nuclear localization signals, and others. To isolate the role of nucleocytoplasmic transport, a probe sensitive only to this factor should be designed. This is exactly what sencyt provides. In fact, this has allowed us to answer an important open question: is the sensitivity of YAP to cell density mediated by mechanics and nucleocytoplasmic transport, or is it mediated by some other factor? Our results suggest that some other factor, likely mediated by the Hippo pathway and not necessarily mechanotransduction, explains this sensing of cell density. This is a novel finding, which was not provided in either our Elosegui-Artola 2017 paper or our Andreu 2022 paper.

* About Reviewer 1: I find it challenging to grasp the point made in the comment. On novelty, in their previous study in NBC 2022 Syncet was already shown to undergo NCT. The reviewer states that the study presents "a simple tool to measure nuclear-cytoplasmic transport (NCT) beneficial for the mechanobiology field, and evidence that this demonstrates a novel layer of regulation in hippo signaling (also because this is observational and not a mechanistic study). The tool in question is far from simple. Its application requires transfection into cell cultures, conducting live imaging, etc. If one aims to measure NCT of endogenous proteins, straightforward immunofluorescence or live imaging of endogenous proteins (like GFP-tagged YAP, Twist, Smads, etc.) using the same experimental setup should suffice to demonstrate relevance, without necessitating any additional experiments. What then, is the unique benefit of this proposed tool? Given it's an artificial construct combining NLS-GFP with a bacterial protein, questions arise about the effects of the forced nuclear localization signal (NLS) or the bacterial component. It is an empirical artificial construct and there is no mechanism to explain its behavior.The comparison of Syncet with YAP seems to me questionable and of limited utility. *

As also noted by reviewer 3 below, the use of genetically encoded fluorescent sensors that require transfection is by now absolutely standard in biology, and cannot be considered to be "far from simple". And as stated above, imaging of endogenous transcription factors (which also requires transfection if it is done live) does not isolate the role of nucleocytoplasmic transport. We also disagree that "there is no mechanism to explain its behaviour". Sencyt was developed in our previous andreu et al 2022 paper, where the mechanism is explained in detail.

• *

*It's unsurprising that an artificial construct only mirrors some aspects of what is considered a genuine mechanosensitive protein. The utility of a synthetic tool lies in its ability to replicate actual phenomena, not in what it fails to do. In comparison to their NBC 2022 study, this manuscript focuses on what their reporter fails to detect. *

We disagree that a synthetic tool is only useful if it replicates the behaviour of endogenous proteins. A synthetic tool, precisely due to its engineered, artificial nature, can be made to respond only to specific factors (in this case, nucleocytoplasmic transport). This can then be used to disentangle the role of such specific factors, as done here.

The osmotic shock was the assay in their 2017 Cell paper. Here they demonstrate that a combination of Blebbistatin+CK (an unclear choice of drugs) is ineffective, as is cell density. Are there other specific peculiarities associated with this construct?

Here, we note that our osmotic shock experiments in our 2017 paper were done for YAP (not nucleocytoplasmic transport in general). Regarding the choice of drugs, please refer to our answer to the reviewer comments above for a full explanation. Also, we want to clarify that this combination is not ineffective, as it leads to clear changes in sencyt. * *

* My other concern is on the minor quantitative changes reported, which seem inconsistent with the provided representative images, where significant differences are difficult to appreciate. For instance, the claim that the transfected sensor differs from an endogenous NCT protein, YAP, after cell density treatment, is hard to detect in their images. In Figure 4, comparing YAP and Syncet in C26 cells, YAP appears uniformly nuclear at high cell density, potentially more nuclear than the synthetic sensor, which is not coherent with their claim.*

• *

Regarding the concern of the minor changes seen in images, please refer to our full response to the reviewer comments above. Regarding the comparison between sencyt and YAP, we want to clarify that in our manuscript we do not compare the absolute values of nuclear localization between YAP and sencyt. As the reviewer notes, these are two different proteins, so which one is more nuclear does not really provide useful information. So whether YAP is more or less nuclear than sencyt is unrelated to (not incoherent with) our claim. What we state in figure 4 is that YAP responds to cell density, whereas sencyt does not. This is clear from the quantifications and also from the images.

• *

• From the Hippo perspective, there is really an unusual amount of nuclear YAP left in their cells. This should be almost completely cytoplasmic from prior contact inhibition studies in the Hippo field. Syncet could be simply less sensitive than YAP in these borderline conditions. Although there's a more noticeable cytoplasmic noise in dense cells with YAP compared to Syncet, this could be attributed to several factors, including differences in protein degradation rates, which I suspect to be quicker for a synthetic protein. From a technical perspective it is complex to get strong conclusions after comparing something so unrelated with each other. One is a live GFP detection and the other is a staining by immunofluorescence. the nature of the background is also different and so conclusions from comparisons between unrelated systems is not justified. *

• *

In conditions of high density, average YAP ratios are close to one (zero in logarithmic scale, as reported in the figures) for MCF10A cells, so there is no nuclear localization. This is similar to what we and others have previously reported in similar conditions (Elosegui Artola et al 2017, Kechagia et al. 2023, for example). In C26 cells, YAP levels at high density are a bit higher. This is likely due to their mesenchymal nature, and therefore diminished cell-cell contact inhibition (as assessed in detail in this revision). This in fact further suggests that the response of YAP to cell-cell contacts is different from a mere mechanical factor, supporting our hypothesis. Regarding the issue of noise, background noise is removed from quantifications, and potential noise coming from non-specificities or autofluorescence is also cancelled by the fact that we compute fluorescence ratios between nucleus and cytoplasm (and not absolute values). Thus, we don't think noise is an issue. Further, we note again that we do not directly compare values between sencyt and yap.

* This suggests caution on what is heralded as the main claim here put forward. *

* Reviewer 1: *

*I do have some sympathy with R2s comments in the consultation. I agree that showing that NCT is mechanosensitive in an epithelium is not new. I also agree that sometimes it is difficult to see the quantitative differences by eye. This second point could be addressed by including more details of the segmentation and analysis in the supplemental material (along with some example images). *

• *

We thank the reviewer for the suggestions. Regarding the novelty, please see above for a detailed discussion, and also the comments of reviewer 3 below (previous work studied not NCT but transcription factors, affected by many parameters). Regarding quantitative differences, we have now addressed this issue by showing images in grayscale rather than green, and also by replacing one example cell in figure 1 which indeed did not reflect the average measured trends. We now also show examples of 3D rendered images of the nuclei in different conditions. We have also gone through the methods and clarified in detail how ratios are calculated, the segmentation procedure is also explained in detail.

* Regarding novelty, I would be interested to know if R2 thinks that there are experiments that the authors could do to improve the work. Or do they need to simply tone down their claims? It's perfectly acceptable to publish a well characterised tool with a series of observations and it's beneficial to the community to do so.*

• Reviewer 3 *

* Thanks to Reviewers #1 and #2 for using this consultation option; I truly appreciate their feedback on my comments and find it extremely valuable. I agree with Reviewer #1 that the method proposed here is relatively simple. Transfecting cells and conducting live fluorescent imaging can hardly be considered difficult. I believe the construct used/designed by the authors is the main advantage as it provides a specific way to quantitatively assess NCT and not limit the analysis to a single nuclear protein (such as YAP). Reviewer #2 suggests using immunofluorescence staining of YAP or live imaging of fusion fluorescent protein (following transfection) to analyze NCT, but this approach would yield a readout not only based on NCT but also on the many other interacting partners/mechanisms that regulate the candidate localization, resulting in an unspecific readout (and similar transfection/live imaging set-up). *

• *

We thank the reviewer for this comment, we fully agree and have elaborated on this in our responses above.

* Regarding the impact of the study, I agree that it is certainly not as impactful as previous publications on this topic. Although I find reviewer#2 argument on Yap irrelevant, as YAP is not the main focus of this paper. Some experiments have been done with cells of epithelial origin, but NCT mechanosensitivity has not been clearly tested in epithelial monolayer, which is the main claim of the proposed study here. The 2017 Cell paper focused on YAP transport into the nucleus (and not NCT in general) and they showed a correlation between YAP nuclear localization and traction force in MCF10A. I am not sure if one would say that "NCT mechanosensitivity has been well demonstrated in epithelial cells" based on this single panel. The impact of the proposed study is certainly not outstanding but offering a thorough analysis in epithelial cells (as monolayers and not as individual cells) and presenting a well-defined experimental approach should be of interest in the field. I agree with comments from reviewer#2 that some reported effects in graph are unclear on main images. More experimental details should hopefully clarify this aspect.*

• *

We fully agree with the reviewer. Regarding quantitative differences, we have now addressed this issue by showing images in grayscale rather than green, and also by replacing one example cell in figure 1 which indeed did not reflect the average measured trends.

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 very well-written manuscript, Pere Rochas-Cusachs and colleagues investigated the response of nucleocytoplasmic transport (NCT) to mechanical stress and tested whether this response is similar in epithelial and mesenchymal cells using a combination of quantitative approaches. This study builds upon their earlier findings, which elegantly demonstrated that NCT is sensitive to mechanical forces transmitted to the nuclear membrane. Using a similar approach to their recent work, they quantitatively analyzed NCT and compared the two cell types using various treatments that impact nuclear membrane tension. The study is straightforward and experimentally sound, with an adequate number of replicates and independent experiments. While one might consider the limitations given their previous work, none have demonstrated that NCT is mechanosensitive in epithelial cells. Additionally, they provide a simple approach to measure NCT, which should be of interest in the field. However, it is unclear how the authors defined the epithelial phenotype in this work and whether they solely based this characterization on the tissue/cell's origin. Epithelia can be defined ultrastructurally with reference to their apico-basal polarity and specific cell-cell junctions (Alberts et al., 1994; Davies and Garrods, 1997). Changing cell density should affect cell/cell adhesion, but the authors provide no evidence that the cells tested in the study are attached to their neighbors on all sides and form an epithelium. While I recognize that the objective of this study is not to mimic the in vivo behavior of epithelial tissue, the authors should at least ensure that cells form a monolayer by quantitatively assessing cell-cell junctions (or they should adjust their conclusions adequately). This control is specifically important for Figure 3 and 4, whose objective is to test the impact of cell/cell contacts. But it would also be important to provide this essential control for Figure 1 and 2, as it is unclear from the images provided if MCF7 cells are forming an epithelium (and form cell/cell junctions).

Significance

The mechanosensitivity of NCT is an important question central to many aspects of cell biology. One might consider the impact of the proposed work limited, given their previous research. However, none have demonstrated that NCT is mechanosensitive in epithelial cells, making it a crucial question that needs to be addressed. Additionally, they provide a simple approach to measure NCT, which should be of interest to a broad audience.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

The study conducted by Granero Moya and colleagues describes the application of a synthetic protein which is observed to enter the nucleus in response to mechanical strains, rather than being influenced by cell density. However, the novelty of this work is minimal since the conceptual framework and the utilization of this identical or similar tool have been previously reported by the same team in earlier publications. In their experiments, they employ this GFP-based sensor, referred to as Sencyt, in cells subjected to osmotic shocks. These shocks are highly stressful and impact a range of cellular processes, including stress response pathways MAPK and others; Osmoregulatory pathways; cell cycle regulations, autophagy and death pathway; ion channel regulations and others. The second findings are on cells treated with a combo of drugs affecting the actin cytoskeleton. The justification for using a combination of two specific drugs remains unclear, as the study does not adequately explain the rationale behind this choice. Additionally, there is a lack of information regarding the full range of targets these drugs affect. This raises questions about the comprehensiveness and applicability of the findings, as understanding the complete scope of the drugs' targets is crucial for interpreting the results within a minimal frame of physiological context.

The novelty is on the specificity of this synthetic fusion protein for these manipulations and not on cell density. Yet, the reasons behind this selective response remain unexplained, potentially attributable to the unique characteristics or sensitivity thresholds of their synthetic probe. As comparison, YAP localization and this is sensitive to both inputs, but this is also already published (fig4). The focus is anyway on Sencyt for which they offer simple observations and quantifications. When reviewing the figures presented, I find it challenging to detected marked differences, despite their quantitative data suggesting otherwise. Furthermore, the study attempts to correlate the behavior of Sencyt with the nuclear geometric parameter of solidity, a connection that seems to lack a clear basis in cell biology and could potentially lead to misconceptions.

Significance

In sum, I think the MS is of interest for a very specialistic audience. There are no clear interpretations. The work is done in one or two cellular model systems in vitro; and the general significance of these observations is of very limited impact and no novelty.

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 nucleus is recognised as a core component of mechanotransduction with many mechano-sensitive proteins shuttling between the nucleus and cytoplasm in response to mechanical stimuli. In this work, Granero-Moya et al characterise a live florescent marker of nucleocytoplasmic transport (NCT) and how it responds to a variety of cues. This work follows on from the authors previous study (Andreu 2022) where they examined the response of passive and active NCT to mechanical signalling using a series of artificial constructs. One of these constructs (here named Sencyt) showed a differential localisation depending on substrate stiffness, accumulating in the nucleus on stiffer substrates (which the authors previously showed was due to differences in mechano-sensitivity of passive versus facilitated NCT). Here the authors use Sencyt as a tool to probe how different cues affect NCT and thus nuclear force-sensing in two different cell lines (one epithelial, one mesenchymal).

They have established a 3D image segmentation pipeline to measure both the nuclear/cytoplasmic ration of Sencyt and 3D nuclear shape parameters. As a proof-of -principle, they show that hypoosmotic shock (which inflates the nucleus and would be expected to increase nuclear tension) and hyper-osmotic shock (which shrinks and deforms the nucleus) alter Sencyt nuclear-cytoplasmic ration as expected. They then show that inhibiting acto-myosin, which would be expected to block force transduction to the nucleus, reduces NCT, although interestingly this is without any changes to nuclear morphology. They then examine how cell density affects NCT and show that Sencyt localisation correlates only weakly with density but much more strongly with nuclear deformation (especially as measured by solidity). This is surprising considering that mechano-sensitive transcription factors such as YAP have been shown to exit the nucleus at high cell densities. Therefore, the authors directly compare Sencyt and Yap nucleo/cytoplasmic localisation and show that Sencyt behaves differently to YAP with YAP localisation correlating strongly with cell density. This reveals an added layer of complexity in YAP regulation beyond pure changes to NCT.

Major points

The data presented throughout this work are high quality and rigorous. The controls used are appropriate (including the use of a freely diffusing mCherry to illustrate the specificity of the Sencyt probe in osmotic shock experiments - figure S2). Experiments are properly replicated and the statistical analysis is appropriate. The data are beautifully presented in figures and the manuscript is well written and very clear. Overall this is a high quality work.

The discussion is careful and the conclusions are supported by the data. My only small concern is that the authors place too much emphasis on how this work is in 'multicellular systems' as opposed to their previous work in single cells (for example "Here, we demonstrate that mechanics also plays a role in multicellular systems, in response to both hypo and hyper-osmotic shocks, and to cell contractility. L212). Cell density is only controlled in figures 3 and 4 and in some of the earlier experiments, cells look quite sparse (eg Figure 2). It's also debatable how far a monolayer of cancer cells, which lack contact inhibition of growth, is a multicellular system. Furthermore, the authors don't specifically look at cell/cell adhesion or observe major differences between the epithelial or mesenchymal lines. For this reason, the authors should tone down this discussion before publication.

Optional experimental suggestions: For me, the most compelling finding is that nuclear deformation has a greater correlation with NCT than cell density and that this is different from the behaviour of YAP. To cement the importance of nuclear deformation, the authors could induce deformation in single cells, for example by culture on very thin micropatterned lines and assess the localisation of Sencyt and YAP. It would also be interesting to assess the role of force transduction in this context or in different densities by removing actin, which affects NCT without inducing nuclear shape changes. These functional experiments would allow the authors to draw stronger conclusions about the role of nuclear shape and deformation but they aren't necessary for publication.

Minor points

• I'd like to see better examples of 3D reconstructions of nuclei (ie fig 1C but bigger) in different conditions. This is especially important in figure 3 where it would be helpful to see examples of nuclei with high or low solidity. The differences in oblateness are clear to see from the images in 3a and 3f but solidity could be better illustrated.
• Where Sencyt index is plotted, it would be clearer to add labels to at least figure 1 indicting which indicate whether it is more cytoplasmic or nuclear.

Referees cross-commenting

Here comments from all 3 reviewers are reported

Reviewer 1:

I disagree with R2's comment that there is 'no novelty' here. Although this work is going to be of greater interest to a specialised rather than general audience, it characterises in depth a simple tool to measure NCT which will be useful for mechanobiology field. Also, using 'two cellular model systems in vitro' is very standard in the field when assessing subcellular processes like NCT. Using this approach in vivo would be very interesting but challenging and would be an entirely different study .

I agree with R2's comments that the authors should better justify their combination of two actin inhibitors and R3s point on better assessing cell/cell junctions.

Reviewer 2

About Reviewer 3's comments, I believe it's a stretch to highlight the strength and novelty based on "NCT's mechanosensitivity in epithelial cells has not been demonstrated,". There are thousands of papers on the Hippo pathway, that is known to be mechanosensitive, on the regulation of YAP, that enters in the nucleus in Hippo inhibited conditions and exits to the cytoplasm in Hippo induced cells, including downstream of mechanical signals. The phenomenon of nuclear-cytoplasmic shuttling being a common event from neurons to endothelial and multiple types of epithelial, immune, and fibroblast cells is already established through NCT of this and other endogenous proteins. This is simply an accepted fact. Then, The Nature cell Biology 2022 was offering a very general claim. No warning that conclusions could have been cell type specific. In the Artola 2017 Cell paper they also showed NCT in mammary epithelial cells. We should definitively conclude that NCT's mechanosensitivity in epithelial cells has been well demonstrated.

About Reviewer 1: I find it challenging to grasp the point made in the comment. On novelty, in their previous study in NBC 2022 Syncet was already shown to undergo NCT. The reviewer states that the study presents "a simple tool to measure nuclear-cytoplasmic transport (NCT) beneficial for the mechanobiology field, and evidence that this demonstrates a novel layer of regulation in hippo signaling (also because this is observational and not a mechanistic study). The tool in question is far from simple. Its application requires transfection into cell cultures, conducting live imaging, etc. If one aims to measure NCT of endogenous proteins, straightforward immunofluorescence or live imaging of endogenous proteins (like GFP-tagged YAP, Twist, Smads, etc.) using the same experimental setup should suffice to demonstrate relevance, without necessitating any additional experiments. What then, is the unique benefit of this proposed tool? Given it's an artificial construct combining NLS-GFP with a bacterial protein, questions arise about the effects of the forced nuclear localization signal (NLS) or the bacterial component. It is an empirical artificial construct and there is no mechanism to explain its behavior. The comparison of Syncet with YAP seems to me questionable and of limited utility. It's unsurprising that an artificial construct only mirrors some aspects of what is considered a genuine mechanosensitive protein. The utility of a synthetic tool lies in its ability to replicate actual phenomena, not in what it fails to do. In comparison to their NBC 2022 study, this manuscript focuses on what their reporter fails to detect. The osmotic shock was the assay in their 2017 Cell paper. Here they demonstrate that a combination of Blebbistatin+CK (an unclear choice of drugs) is ineffective, as is cell density. Are there other specific peculiarities associated with this construct?

My other concern is on the minor quantitative changes reported, which seem inconsistent with the provided representative images, where significant differences are difficult to appreciate. For instance, the claim that the transfected sensor differs from an endogenous NCT protein, YAP, after cell density treatment, is hard to detect in their images. In Figure 4, comparing YAP and Syncet in C26 cells, YAP appears uniformly nuclear at high cell density, potentially more nuclear than the synthetic sensor, which is not coherent with their claim. From the Hippo perspective, there is really an unusual amount of nuclear YAP left in their cells. This should be almost completely cytoplasmic from prior contact inhibition studies in the Hippo field. Syncet could be simply less sensitive than YAP in these borderline conditions. Although there's a more noticeable cytoplasmic noise in dense cells with YAP compared to Syncet, this could be attributed to several factors, including differences in protein degradation rates, which I suspect to be quicker for a synthetic protein. From a technical perspective it is complex to get strong conclusions after comparing something so unrelated with each other. One is a live GFP detection and the other is a staining by immunofluorescence. the nature of the background is also different and so conclusions from comparisons between unrelated systems is not justified. This suggests caution on what is heralded as the main claim here put forward.

Reviewer 1: I do have some sympathy with R2s comments in the consultation. I agree that showing that NCT is mechanosensitive in an epithelium is not new. I also agree that sometimes it is difficult to see the quantitative differences by eye. This second point could be addressed by including more details of the segmentation and analysis in the supplemental material (along with some example images).

Regarding novelty, I would be interested to know if R2 thinks that there are experiments that the authors could do to improve the work. Or do they need to simply tone down their claims? It's perfectly acceptable to publish a well characterised tool with a series of observations and it's beneficial to the community to do so.

Reviewer 3

Thanks to Reviewers #1 and #2 for using this consultation option; I truly appreciate their feedback on my comments and find it extremely valuable. I agree with Reviewer #1 that the method proposed here is relatively simple. Transfecting cells and conducting live fluorescent imaging can hardly be considered difficult. I believe the construct used/designed by the authors is the main advantage as it provides a specific way to quantitatively assess NCT and not limit the analysis to a single nuclear protein (such as YAP). Reviewer #2 suggests using immunofluorescence staining of YAP or live imaging of fusion fluorescent protein (following transfection) to analyze NCT, but this approach would yield a readout not only based on NCT but also on the many other interacting partners/mechanisms that regulate the candidate localization, resulting in an unspecific readout (and similar transfection/live imaging set-up). Regarding the impact of the study, I agree that it is certainly not as impactful as previous publications on this topic. Although I find reviewer#2 argument on Yap irrelevant, as YAP is not the main focus of this paper. Some experiments have been done with cells of epithelial origin, but NCT mechanosensitivity has not been clearly tested in epithelial monolayer, which is the main claim of the proposed study here. The 2017 Cell paper focused on YAP transport into the nucleus (and not NCT in general) and they showed a correlation between YAP nuclear localization and traction force in MCF10A. I am not sure if one would say that "NCT mechanosensitivity has been well demonstrated in epithelial cells" based on this single panel. The impact of the proposed study is certainly not outstanding but offering a thorough analysis in epithelial cells (as monolayers and not as individual cells) and presenting a well-defined experimental approach should be of interest in the field. I agree with comments from reviewer#2 that some reported effects in graph are unclear on main images. More experimental details should hopefully clarify this aspect.

Significance

In this work, Granero-Moya et al characterise a new tool for measuring NCT and show that it is mechanically regulated. Given the importance of NCT in mechano-transduction, this tool will be a great asset to the mechano-biology community and will likely be adopted by multiple groups in the future. The findings about the effects of cell density on NCT and differences from YAP are interesting but could be further fleshed out. This work is likely to be of greatest interest to a specialised audience working in the fields of mechano-biology and nuclear transport.

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

First, we would like to extend our gratitude to all reviewers for their supportive and enthusiastic feedback, which acknowledges our study as an interesting, well-executed, and well-documented contribution to the field. We are also pleased that the novelty and significance of our work have been recognized and appreciated.

As highlighted by reviewers 2, 3, and 4, our research represents a substantial advancement in understanding the mechanisms that coordinate the development of different cell types. Our findings have broader implications for developmental biology. We also thank the reviewers for their valuable insights, which have significantly improved the overall readability of our manuscript. We have carefully considered all minor corrections and text modifications they had suggested and made amendments accordingly.

The reviewers proposed several complementary experiments to enhance and clarify our points. We have conducted most of these experiments, with one exception (detailed below), and incorporated the corresponding results into this revised version of the manuscript.

Additionally, reviewers agreed that higher resolution images depicting the interactions between tendon and myoblast membranes would strengthen our manuscript. In response, we are pleased to present new high-resolution images of Ama::EGFP localization with respect to tendon and muscle cells, obtained using Zeiss Airyscan technology. We also provide new images using newly generated flies that allow simultaneous observation of both myoblast and tendon membranes.

We believe these modifications substantially enhance the quality and interest of our results, as already highlighted by the reviewers.

__Referees cross-commenting: __

All reviewers agreed* that "this is an interesting study that is well done and well documented. I agree with reviewer 1 that the study would further benefit from better imaging of the cellular extensions of tendons and myoblasts to see how both cell types interact." *

Reply: We agree with this point. To address it, we analyzed leg discs from Sr-Gal4>UAS-myrGFP line (labelling tendon membranes) crossed with a newly generated line R32D05::CD4TdTomato (myoblast specific expression of membrane tagged Tomato protein). Using confocal Zeiss Airy Scan technology, we generated high resolution images for which both tendon cell extensions and myoblast membranes are simultaneously visualized. These images are included in the new Figure 2 (O, P and O', P'). To be noticed: as we also provide new high resolution images of Ama::EGFP and Nrt localizations (Fig. 2 M, N and M', N'), we removed images of zoom in of 5h APF leg disc.

REVIEWER 1

Moucaud et al carried out single cell sequencing on myoblasts from the developing drosophila leg muscles, focusing on gene expressions overlapping with tendon and muscle cells. This study proposes that neuronal cell adhesion molecules Ama and Nrt interact in myoblast and tendon adhesion to support tendon and in proliferation of muscle progenitors. This study traces Ama and Nrt expression with various drosophila mutant strains and provides evidence to support its claims using single cell sequencing, immuno-fluoresence and in situ hybridisation. *The authors report novel markers to study the interactions between muscle and tendon progenitors in the Drosophila leg provide convincing evidence of their functions in muscle and muscle and tendon formation. *

• The authors report novel markers to study the interactions between muscle and tendon progenitors in the Drosophila leg provide convincing evidence of their functions in muscle and muscle and tendon formation. *

__ Reply:__ We are grateful to the Reviewer 1 for his/her supportive comments on the quality of our work.

Page 2 "cell types*..." might be worth including other cell types such as vascular/endothelial if listing all cell types in the limb, as the sentence is suggesting. *

__ Reply__: "blood vessels" have been added as components of the limb musculoskeletal system.

Reviewer's comment: The authors discuss the interactions between the myoblasts and tendon cells but do not show any cellular resolution of the interaction between the cells and the secreted adhesion proteins. It would enhance the manuscript if the authors could show high resolution images of these cellular interactions with the secreted protein in vivo.

__ Reply__: see reply to referees cross-commenting about newly generated high-resolution images shown in Fig. 2.

Lots of examples of definite article (the) missing throughout the text.

Reply: The text has been edited and missing articles added

Second line of Abstract does not flow: Ama encodes secreted proteins to "Ama encodes a secreted protein"

__ Reply__ the correction has been made accordingly

2nd para Intro- this para is essentially discussing vertebrate limb muscle/tendon precursors, although includes a non-vertebrate citation. It could be helpful to (briefly) compare/contrast the non-vertebrate vs vertebrate literature on this topic.

__ Reply__: Indeed, this paragraph is primarily focused on the development of the musculoskeletal system in vertebrates. The comparison (from a molecular standpoint) with the muscle system of the Drosophila leg appears in the following paragraph. For clarity, we have included a brief, more general description (end of second paragraph) about the muscle/tendon system in Drosophila to highlight certain divergences between vertebrate and invertebrate systems and to introduce the subsequent paragraph.

"in limb of chick embryo add "the limb"

__ Reply:__ the correction has been made accordingly

p6 because these two antibodies were raised in rabbit, as the Twist antibody, needs some additional explanatory text.

__ Reply__: We have modified the text to give a more accurate explanation: "Because these two antibodies were raised in rabbit, as was the Twist antibody, we could not use this latter to visualize the myoblasts"

P9 discussion creeping into results section-with some speculation on Ama forming homophilic adhesions which has not been experimentally tested.

__ Reply:__ Because we chose to submit this work as a short format paper, Results and Discussion sections are indeed combined. However, we agree that homophilic adhesion properties of Ama have been shown only in cell culture and not tested in physiological context. To clarify this point, we have modified the corresponding part of the text and only suggest that Ama could directly bind to membranes through its putative GPI modification as proposed by Seeger et al.

Ama depletion affects both viability and the proliferation rate of leg disc myoblasts (in a Nrt-independent way) Does it have similar role in tendon precursors? Could the authors provide any evidence of apoptosis given proposed role of Ama in glial cells?

__ Reply: __As asked by Reviewer 1, we have tested these two points and included the results in suppl figure 3 (A-C). As expected, the proliferation rate of tendon cells is not affected as we have previously showed that tendon cells are post-mitotic cells (Laurichesse et al. 2021). Moreover, we now show that Ama depletion does not lead to apoptosis of tendon cells. See Supp. Figure 3 (A-C), the main text has also been modified accordingly.

REVIEWER 2:

In this well-written, comprehensive, and interesting manuscript, the authors study the molecular circuitry that supports the coordinated activity of tendon cells and myoblasts during development. As the authors themselves point out in the introduction, the assembly of tissues within the musculoskeletal system provides a particularly attractive system in which to study how different cell types coordinate their behaviours to form higher-order structures. Using single-cell transcriptomics, the authors first identify the cell adhesion molecule Ama and transmembrane protein Nrt as enriched in Drosophila myoblasts and tendon cells. Their transcriptomic data suggest that Nrt is specifically expressed in the tendon cells while Ama is expressed in both. They support these data with a variety of in situ, antibody, and endogenous stainings. Using a series of genetic manipulations, they then convincingly show that Ama controls the total number of myoblasts during the larval stages: Ama knockdown is associated with both decreased proliferation and increased apoptosis of myoblasts. Ama's role in regulating myoblast number is shown to be independent of Nrt and likely under the control of the FGF/RTK pathway. Finally, the authors show that the loss of either Nrt or Ama activity is associated with a loss of adhesion between myoblasts and tendon cells and with the stunted growth of the long tendon. Thus, their data point to Ama playing dual roles in muscle development by regulating both myoblast number and cell adhesion.

* I very much enjoyed reading the paper, which I think makes an important contribution to our understanding of both the developing musculature and inter-cell-type coordination during development more broadly. I have only a handful of grammatical errors to point out.*

__Reply: __We appreciate these enthusiastic and supportive comments, and we would like to thank the Reviewer for highlighting the broader contribution of our work to the understanding of the mechanisms of coordination between different cell types.

*- Grammar: 'This prompted us to use Drosophila model to search' should read 'This prompted us to use the Drosophila model to search' - Grammar: '...identify Neurotactin (Nrt) and its binding partner, Amalgam (Ama) as candidates...' should read '...identify Neurotactin (Nrt) and its binding partner, Amalgam (Ama), as candidates...' - Grammar: 'As tendon precursors in leg disc' should read 'As tendon precursors in the leg disc'. - Grammar : '...we performed a series of in situ hybridization...' should read '...we performed a series of in situ hybridizations...' - Grammar: 'Because these two antibodies were raised in rabbit, as the Twist antibody' should read 'Because these two antibodies were raised in rabbit, as was the Twist antibody' - Grammar: 'Statistical analysis reveals an increase myoblast total number when overexpressing an activated ERK' should read 'Statistical analysis reveals an increase in the total number of myoblasts when overexpressing an activated ERK' - Grammar: The following section header needs rephrasing: 'Ama, potential downstream effector of FGF pathway in the regulation of myoblast number'. Maybe 'Ama is a potential downstream effector of the FGF pathway in the regulation of myoblast number' or Ama: a potential downstream effector of the FGF pathway in the regulation of myoblast number. - Grammar: 'whereas its reduction (UAS-StyRNAi) lead to more myoblasts' should read 'whereas its reduction (UAS-StyRNAi) leads to more myoblasts' - For clarity 'The expression of the constitutively active form of ERK could rescue the phenotype of Ama depletion in glial cells (Ariss et al. 2020)' might read better as 'Previous work has shown that the expression of the constitutively active form of ERK can rescue the phenotype of Ama depletion in glial cells (Ariss et al. 2020). Therefore, we tried...' - Grammar: 'showed a significant higher number of myoblasts compared' should read 'showed a significantly higher number of myoblasts compared' - Grammar: 'Another, not exclusive, possibility' should read Another, non-mutually exclusive, possibility'. - Grammar: 'We measured the length of the tilt relatively to the length' should read '. We measured the length of the tilt relative to the length' *

Reply: All the modifications suggested above by Reviewer 2 are now included in the text.

REVIEWER 3:

*Myoblast and tendon precursors stem from different developmental origins. Hence, they need to find each other to build a functional muscle-skeleton. How they do so is an exciting biological problem, not only for this reviewer who is working on Drosophila muscle development, too. As we currently understand little about how myoblasts communicate with tendons during development, I find this manuscript a generally interesting contribution unravelling a new mechanism of cell-cell communication between these two cell types. It proposes a role for 2 interesting proteins that are little studied. Furthermore, Drosophila leg muscle-tendon development is complex and results in an intricate final architecture. Thus, a better understanding of its molecular mechanisms of development is exciting to this reviewer and to the muscle and tendon fields. *

Reply: We express our gratitude to Reviewer 3 for his/her keen interest in our work and for emphasizing its significance within the field of developmental biology.

*While some Ama mRNA expression in myoblasts was confirmed with in situ hybridisation, it was also shown that Ama mRNA is expressed in other sources including tendon precursors. As the interesting AmaGFP protein overlapping with the developing tendon cells is found at some distance from the myoblasts, the source for this Ama protein population is not entirely clear. To identify if it is secreted from myoblasts I suggest to stain for Ama-GFP in the muscle-specific Ama knock-down discs at 5h APF. This could use the late knock-down condition. *

__ Reply:__ In Suppl Fig.2 in a close-up view of the femur region, we show that at 5h APF Ama is transcribed in addition to myoblasts also in the developing tilt tendon. This tendon associated Ama expression is specific as it is detected after myoblast specific Ama knockdown. Thus, at 5h APF, the Ama-GFP signal detected at the interface of muscle and tendon precursors could in part correspond to Ama secreted by the tilt tendon cells. However, we also observed clear Ama-GFP signal at the interface of myoblasts and tendon precursors at 0h APF when Ama is not yet transcriptionally activated in tilt tendon precursors (not shown). Thus, we are confident that the myoblasts are the main source of secreted Ama protein that ensure close proximity of myoblast and tendon precursor cells. A view supported by the loss of myoblast-tendon cell proximity in myoblast-specific Ama knockdown. However, to clarify this point, we immunostained myoblast-specific Ama knock-down discs for the Ama protein in at 5h APF as suggested. As stated in the text, we were concerned that GFP tag could influence the life-time of the Ama protein, as GFP itself is pretty stable. This is why we used anti-Ama antibody kindly provided by Dr. Silman to determine whether myoblast-specific Ama knockdown (using R32D05-Gal4 driver) would completely abolish Ama protein at 5h APF. We indeed observed a strong reduction of Ama protein at this stage indicating that the contribution of Ama protein from tendon cells is minimal (but cannot be completely excluded), with myoblasts remaining the major source at this stage. This new result is now presented in Suppl Fig. 2M-P. This result is also in accordance with our new result showing that tendon-specific AmaKD has no effect on tendon growth (see reply to the comment below regarding tendon length). In light of this new result, we have modified the text accordingly in the corresponding paragraph (p5-6).

Generally, it might be useful to move the part of Figure 2 that shows the Ama-GFP Nrt co-staining to the later part in the text that addresses the interaction of both cell types and keep the autonomous Ama role in muscle for the start of paper only.

__ Reply: __We have indeed debated extensively about this possibility before submitting this work. While such a presentation would have some logical coherence, it also has the disadvantage of having to resume, at least partially, the expression of Ama, leading to certain redundancies. Additionally, we chose to begin with a comparison of the new myoblast transcriptomic data with pre-established tendon data to highlight the presence of ligand-receptor pairs. In this context, it seemed to us more pertinent to present the expression patterns of Ama and Nrt together in the initial figures.

To quantify the interaction of the myoblast cell membranes and the tendon cells better it would be useful to combine sr>CAAXmCherry with a myoblast membrane maker (possibly Him-CD8-GFP or use R15B03-Gal4 with R79D08-lexA). This could also improve the "mean distance" measurements. As currently presented, it is not so clear how the mean distance was measured. It could be helpful to indicate some examples in zoom-in vies on Figure 5. Does a distance of 4 µm in wild type mean that the myoblast is not touching the tendon precursors, or is only the myoblast nucleus that is Twi positive at this distance?

__ Reply: We are grateful to this reviewer for its relevant suggestion. Thus, as stated above (referees cross-commenting), we provide new high-resolution images with labelled membranes of both tendon cells and myoblasts (fig 2 O-P). As shown here, myoblast membranes are very closed to each other, and nuclei occupy an important part of myoblast volumes. So, we found more accurate to use the myoblast nucleus (stained with Twist antibody) to detect individual myoblasts using Imaris Spot function rather than myoblast membranes. We also believe that the distances between the center of myoblast nuclei and the tendon surface are representative of the distance between these two cell types as nucleus myoblast occupies most of the cell volume. We addressed this point in the new Suppl. Fig.5. __Regarding the distance of 4____ µm: As mentioned in the original text, the 4 µm distance represents the average distance between myoblasts and the tendon surface in wild type discs. We do not perceive this distance as indicative of a threshold distinguishing myoblasts that interact physically with tendons from those that do not. We rather use this mean distance to quantify the distribution of myoblasts around the tendon and their dispersion/mis-distribution in Ama and Nrt knockdown leg discs. To clarify this point, we have modified the corresponding paragraph: "This result indicates that AmaKD leads to myoblasts mis-distribution around the tilt, suggesting that the reduction of Ama level could affect myoblast-tendon adhesion".

For a better understanding of how the mean distance was measured, we added a new Supplementary Figure 5 (rather than a zoom-in in the main figure as suggested by this reviewer), with a corresponding description of how this distance was measured in addition to the explanations in the material and method section.

Is the tendon elongation phenotype seen after Ama RNAi in muscle and in Nrt mutants due to the fact that myoblasts are further away from tendons or is it an Ama/Nrt role that is autonomous to tendons? This could be tested by assaying tendon elongation after tendon-specific Ama knock-down as shown in Figure S2.

__ Reply: __As asked by this reviewer, we have performed this experiment using Sr-Gal4 driver to induce tendon-specific Ama knockdown and assessed tendon elongation using R79D08-lexA>lexAop-GFP marker. Overall statistical analysis is now included in Fig 5F and G graphs. This analysis shows that tendon-specific Ama knockdown does not affect tendon elongation. This is in concordance with the fact that Ama knockdown in myoblasts leads to tendon defects similar to that of Nrt loss of function in tendon clearly indicating that the observed phenotypes are due to Ama's role in myoblasts. This does not exclude an additional subsequent Ama function in growing tendon precursors in later development.

Minor : Is Figure 2Q a zoom-in from Figure 2P? If yes, it would be helpful to indicate the rough position of it in the lower magnification image.

Reply: Figure 2Q was not a zoom-in from Figure 2P in the previous version of the paper. As stated above Fig. 2 has now been modified.

Minor page 6 - w[1118] is with small "w".

Reply: modifications have been made accordingly in main and figure texts.

REVIEWER 4:

*The manuscript is well-organized, with clear descriptions of methods and results. The use of transcriptomic datasets and gene expression analyses provides insights into the molecular mechanisms underlying the interaction between muscle and tendon precursors. *

*The immunostaining and in situ hybridization experiments well illustrate the expression patterns of Ama and Nrt in muscle and tendon cells during leg disc development in Drosophila. *

*The functional analyses, including knockdown experiments, support the conclusion that Ama plays a crucial role in maintaining the pool of leg muscle precursor cells and coordinating tendon and muscle precursor growth. *

The manuscript significantly enhances our understanding of cell-cell interactions in the musculoskeletal system of Drosophila. The findings have broader implications for the field of developmental biology. In general, this manuscript provides valuable insights into the molecular processes governing leg muscle and tendon development.

Reply: We are indebted to Reviewer 4 for highlighting that our manuscript is well-organized and well-illustrated. We are also grateful to Reviewer 4 for highlighting the valuable insights of our work.

some aspects in the manuscript, for example how Ama regulates myoblast number and its interaction with the FGFR pathway, could be further explored or clarified.

__ Reply:__ Regarding Ama's contribution for maintaining the myoblast pool through its interaction with the FGF pathway, we demonstrate here that, contrary to what has been proposed for glial cells, Ama acts downstream of this pathway, although we emphasize that there is a synergistic effect with the MAPK pathway inhibitor, Sprouty. These findings thus reveal complex and variable regulatory mechanisms between Ama and the FGF pathway that would require specific investigation, the entirety of which appears challenging to integrate into this same publication.

the organization of the abstract could be improved to provide a clearer and more comprehensive overview of the study. The abstract currently lacks a structured presentation of essential components such as methods, results, and conclusions. It would greatly benefit from a more systematic arrangement.

__ Reply:__ We have made modifications to propose a more structured abstract.

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

The manuscript by Moucaud et al. titled "Amalgam plays a dual role in controlling the number of leg muscle progenitors and regulating their interactions with developing tendon" presents a comprehensive investigation of cell-cell interaction between skeletal muscle and tendon precursor cells. The authors use the Drosophila leg development model to findr candidates involved in early interactions between muscle and tendon precursors. The manuscript focuses on the role of Amalgam (Ama) and Neurotactin (Nrt) in these two cell populations communication during musculoskeletal system development .

The results demonstrate that Ama and Nrt are selectively expressed in skeletal muscle and tendon precursors, respectively. Moreover, Ama and Nrt are required to keep myogenic and tendom precursors closely associated, thus is essential for leg muscles development. Furthermore, Ama plays also important role in maintaining the pool of leg muscle precursor cells. Additionally, the expression patterns of Ama and Nrt suggest a potential dual role for Ama, not only in interacting with Nrt but also in a Nrt-independent manner. Summarizing, the study shows the importance of specific bi-directional communication between different cell populations in the formation of functional organs in suggests that Ama and Nrt plays key role during musculoskeletal system development.

The manuscript is well-organized, with clear descriptions of methods and results. The use of transcriptomic datasets and gene expression analyses provides insights into the molecular mechanisms underlying the interaction between muscle and tendon precursors. The immunostaining and in situ hybridization experiments well illustrate the expression patterns of Ama and Nrt in muscle and tendon cells during leg disc development in Drosophila. The functional analyses, including knockdown experiments, support the conclusion that Ama plays a crucial role in maintaining the pool of leg muscle precursor cells and coordinating tendon and muscle precursor growth. Additionally, the authors explore the potential link between Ama and the FGFR pathway, suggesting that Ama may act downstream of the pathway.

However, some aspects in the manuscript, for example how Ama regulates myoblast number and its interaction with the FGFR pathway, could be further explored or clarified. Moreover, the organization of the abstract could be improved to provide a clearer and more comprehensive overview of the study. The abstract currently lacks a structured presentation of essential components such as methods, results, and conclusions. It would greatly benefit from a more systematic arrangement.

Significance

The manuscript significantly enhances our understanding of cell-cell interactions in the musculoskeletal system of Drosophila. The findings have broader implications for the field of developmental biology. In general, this manuscript provides valuable insights into the molecular processes governing leg muscle and tendon development.

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 is investigating mechanisms how muscle and tendon development is properly coordinated. By using differential expression analysis of developing Drosophila legs, the authors find that Amalgam (Ama) is expressed in the developing leg myoblasts and Neurotactin (Nrt) is expressed in one class of tendon precursors, the tilt in the dorsal femur. Using a new Ama-GFP knock-in line, Ama protein was found in the proximity to the developing tendons at 5h APF, when myoblasts and tendons likely interact.

Using myoblast specific knock-down of Ama, the authors find that Ama acts autonomously in the leg myoblasts to promote their proliferation and survival. A similar function for Ama had been described for flight muscle myoblasts in larval wing discs (Zappa et al. 2020). In leg myoblasts ama likely acts downstream of FGF signalling in the myoblasts to promote their proliferation.

As Ama and Nrt were proposed as a ligand-receptor pair promoting adhesion, the authors knocked-down Ama later during myoblast development, after they have proliferated, and tested where these myoblasts are positioned. They find that Ama knock-down myoblasts stay at a larger distance from tendons compared to wild type; the same was found in Nrt mutants. Thus, the authors propose that Ama protein secreted from myoblasts acts via the membrane-bound Nrt protein on the tendon precursors to coordinate muscle-tendon adhesion and proper positioning.

1. While some Ama mRNA expression in myoblasts was confirmed with in situ hybridisation, it was also shown that Ama mRNA is expressed in other sources including tendon precursors. As the interesting AmaGFP protein overlapping with the developing tendon cells is found at some distance from the myoblasts, the source for this Ama protein population is not entirely clear. To identify if it is secreted from myoblasts I suggest to stain for Ama-GFP in the muscle-specific Ama knock-down discs at 5h APF. This could use the late knock-down condition. Generally, it might be useful to move the part of Figure 2 that shows the Ama-GFP Nrt co-staining to the later part in the text that addresses the interaction of both cell types and keep the autonomous Ama role in muscle for the start of paper only.
2. To quantify the interaction of the myoblast cell membranes and the tendon cells better it would be useful to combine sr>CAAXmCherry with a myoblast membrane maker (possibly Him-CD8-GFP or use R15B03-Gal4 with R79D08-lexA). This could also improve the "mean distance" measurements. As currently presented, it is not so clear how the mean distance was measured. It could be helpful to indicate some examples in zoom-in vies on Figure 5. Does a distance of 4 µm in wild type mean that the myoblast is not touching the tendon precursors, or is only the myoblast nucleus that is Twi positive at this distance?
3. Is the tendon elongation phenotype seen after Ama RNAi in muscle and in Nrt mutants due to the fact that myoblasts are further away from tendons or is it an Ama/Nrt role that is autonomous to tendons? This could be tested by assaying tendon elongation after tendon-specific Ama knock-down as shown in Figure S2.

Minor:

1. Is Figure 2Q a zoom-in from Figure 2P? If yes, it would be helpful to indicate the rough position of it in the lower magnification image.
2. page 6 - w[1118] is with small "w".

Referees cross-commenting

All reviewer agree that this is an interesting study that is well done and well documented. I agree with reviewer 1 that the study would further benefit from better imaging of the cellular extensions of tendons and myoblasts to see how both cell types interact.

Significance

Myoblast and tendon precursors stem from different developmental origins. Hence, they need to find each other to build a functional muscle-skeleton. How they do so is an exciting biological problem, not only for this reviewer who is working on Drosophila muscle development, too.

As we currently understand little about how myoblasts communicate with tendons during development, I find this manuscript a generally interesting contribution unravelling a new mechanism of cell-cell communication between these two cell types. It proposes a role for 2 interesting proteins that are little studied. Furthermore, Drosophila leg muscle-tendon development is complex and results in an intricate final architecture. Thus, a better understanding of its molecular mechanisms of development is exciting to this reviewer and to the muscle and tendon fields.

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 well-written, comprehensive, and interesting manuscript, the authors study the molecular circuitry that supports the coordinated activity of tendon cells and myoblasts during development. As the authors themselves point out in the introduction, the assembly of tissues within the musculoskeletal system provides a particularly attractive system in which to study how different cell types coordinate their behaviours to form higher-order structures. Using single-cell transcriptomics, the authors first identify the cell adhesion molecule Ama and transmembrane protein Nrt as enriched in Drosophila myoblasts and tendon cells. Their transcriptomic data suggest that Nrt is specifically expressed in the tendon cells while Ama is expressed in both. They support these data with a variety of in situ, antibody, and endogenous stainings. Using a series of genetic manipulations, they then convincingly show that Ama controls the total number of myoblasts during the larval stages: Ama knockdown is associated with both decreased proliferation and increased apoptosis of myoblasts. Ama's role in regulating myoblast number is shown to be independent of Nrt and likely under the control of the FGF/RTK pathway. Finally, the authors show that the loss of either Nrt or Ama activity is associated with a loss of adhesion between myoblasts and tendon cells and with the stunted growth of the long tendon. Thus, their data point to Ama playing dual roles in muscle development by regulating both myoblast number and cell adhesion.

I very much enjoyed reading the paper, which I think makes an important contribution to our understanding of both the developing musculature and inter-cell-type coordination during development more broadly. I have only a handful of grammatical errors to point out.

Grammar: 'This prompted us to use Drosophila model to search' should read 'This prompted us to use the Drosophila model to search'

Grammar: '...identify Neurotactin (Nrt) and its binding partner, Amalgam (Ama) as candidates...' should read '...identify Neurotactin (Nrt) and its binding partner, Amalgam (Ama), as candidates...'

Grammar: 'As tendon precursors in leg disc' should read 'As tendon precursors in the leg disc'.

Grammar : '...we performed a series of in situ hybridization...' should read '...we performed a series of in situ hybridizations...'

Grammar: 'Because these two antibodies were raised in rabbit, as the Twist antibody' should read 'Because these two antibodies were raised in rabbit, as was the Twist antibody'

Grammar: 'Statistical analysis reveals an increase myoblast total number when overexpressing an activated ERK' should read 'Statistical analysis reveals an increase in the total number of myoblasts when overexpressing an activated ERK'

Grammar: The following section header needs rephrasing: 'Ama, potential downstream effector of FGF pathway in the regulation of myoblast number'. Maybe 'Ama is a potential downstream effector of the FGF pathway in the regulation of myoblast number' or Ama: a potential downstream effector of the FGF pathway in the regulation of myoblast number.

Grammar: 'whereas its reduction (UAS-StyRNAi) lead to more myoblasts' should read 'whereas its reduction (UAS-StyRNAi) leads to more myoblasts'

For clarity 'The expression of the constitutively active form of ERK could rescue the phenotype of Ama depletion in glial cells (Ariss et al. 2020)' might read better as 'Previous work has shown that the expression of the constitutively active form of ERK can rescue the phenotype of Ama depletion in glial cells (Ariss et al. 2020). Therefore, we tried...'

Grammar: 'showed a significant higher number of myoblasts compared' should read 'showed a significantly higher number of myoblasts compared'

Grammar: 'Another, not exclusive, possibility' should read Another, non-mutually exclusive, possibility'.

Grammar: 'We measured the length of the tilt relatively to the length' should read '. We measured the length of the tilt relative to the length'

Significance

I very much enjoyed reading the paper, which I think makes an important contribution to our understanding of both the developing musculature and inter-cell-type coordination during development more broadly

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

Moucaud et al carried out single cell sequencing on myoblasts from the developing drosophila leg muscles, focusing on gene expressions overlapping with tendon and muscle cells. This study proposes that neuronal cell adhesion molecules Ama and Nrt interact in myoblast and tendon adhesion to support tendon and in proliferation of muscle progenitors. This study traces Ama and Nrt expression with various drosophila mutant strains and provides evidence to support its claims using single cell sequencing, immuno-fluoresence and in situ hybridisation.

Minor comments:

Page 2 "cell types..." might be worth including other cell types such as vascular/endothelial if listing all cell types in the limb, as the sentence is suggesting.

The authors discuss the interactions between the myoblasts and tendon cells but do not show any cellular resolution of the interaction between the cells and the secreted adhesion proteins. It would enhance the manuscript if the authors could show high resolution images of these cellular interactions with the secreted protein in vivo.

Minor typographical

Lots of examples of definite article (the) missing throughout the text

Second line fo Abstract does not flow Ama encodes secreted proteins to "Ama encodes a secreted protein"

2nd para Intro- this para is essentially discussing vertebrate limb muscle/tendon precursors, although includes a non-vertebrate citation. It could be helpful to (briefly) compare/contrast the non-vertebrate vs vertebrate literature on this topic.

in limb of chick embryo add "the limb"

p6 because these two antibodies were raised in rabbit, as the Twist antibody, needs some additional explanatory text

p8 or FGF receptor add "the"

P9 discussion creeping into results section-with some speculation on Ama forming homophilic adhesions which has not been experimentally tested.

Significance

Thge authors report novel markers to study the interactions between muscle and tendon progenitors in the Drosophila leg provide convincing evidence of theri functions in muscle and muscle and tendon formation.

Ama depletion affects both viability and the proliferation rate of leg disc myoblasts (in a Nrt-independent way)

Does it have similar role in tendon precursors?

Could the authors provide any evidence of apoptosis given proposed role of Ama in glial cells?

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 referees for their insightful comments and constructive feedback, which have undoubtedly strengthened our manuscript.

Reviewer #1 Comment:

Absence of evidence demonstrating the presence of prions or prion-seeding activity in extraneural tissues.

Response: Our study focuses on all pathophysiological alterations brought about by prion infections in extraneural organs, be they caused directly by prion infections of said organs or through indirect mechanisms. Nevertheless, we share the reviewer's curiosity about a possible correlation between these changes and the presence of prion-seeding activity. We propose to include additional data demonstrating prion presence and seeding activity in skeletal muscle at various timepoints. This will be achieved by employing proteinase K digestion followed by Western blot (PK-WB) and real-time quaking-induced conversion (RT-QuIC) assays to provide a robust correlation with our transcriptomic analyses and Glul upregulation.

Revision: We have conducted preliminary experiments using PK-WB and RT-QuIC assays. These experiments were performed on terminal-stage prion-infected samples and related controls. If prion presence is detected at this stage, we plan to extend the analysis to earlier stages, specifically at 16 weeks post-inoculation (wpi) and 8 wpi, to track the progression of prion-seeding activity over time.

Western blot analysis was performed on skeletal muscle homogenates at the terminal stage of prion disease in mice infected with three prion strains (RML6, ME7, and 22L) and related NBH control. Control samples include CNS brain homogenates, showing prion presence (PrPSc) in RML6 but not in NBH. Neither skeletal muscles from NBH nor from prion-infected samples show detectable prions, indicating that PK-WB may lack the sensitivity to detect prions in skeletal muscle or that prion levels are below detection thresholds (Revision Figure 1A) for this specific technique.

Next, we performed RT-QuIC assays on muscle homogenates using different protocols, including sodium phosphotungstic acid (NaPTA) enrichment (Revision Figure 1B). NaPTA binds and precipitates PrPSc in the presence of MgCl2, removing contaminants and concentrating PrPSc. For this protocol, 100 µg of muscle homogenates were treated with benzonase to degrade DNA contaminants. Then, 4% NaPTA and 170 mM MgCl2 were added, resulting in a final NaPTA concentration of 0.3%. The samples were incubated at 37{degree sign}C while shaking at 1500 rpm for 2 hours, followed by centrifugation at 15,000 g for 30 minutes to precipitate PrPSc. The resulting pellets were used to seed RT-QuIC reactions. Each biological replicate was run in quadruplicate, and replicates were considered positive only if at least 3 out of 4 technical replicates showed detection. Results indicated no amplification for NBH samples. For RML6 skeletal muscles, 2 out of 3 biological replicates were positive. For ME7, 0 out of 4 biological replicates were positive. For 22L, 1 out of 3 biological replicates was positive.

Using the same protocol without NaPTA (Revision Figure 1C), two positive samples were observed in the NBH condition, suggesting that NaPTA is useful for specific prion enrichment.

Similarly, we combined NaPTA and sarkosyl in our further trial. 10% weight/volume muscle homogenates in 1x PBS containing 2% sarkosyl were centrifuged at 80 g for 1 minute, and the supernatant (500 µl) was collected. If still dirty, further centrifugation at 2700 g for 5 minutes was performed. Then, 500 µl PBS containing 2% sarkosyl was added and incubated for 10 minutes at 37{degree sign}C. Benzonase and MgCl2 (final concentrations 50 U/ml and 1 mM, respectively) were added and incubated for 30 minutes at 37{degree sign}C with shaking at 1500 rpm. NaPTA was added to a final concentration of 0.3% and incubated for 30 minutes at 37{degree sign}C with shaking. Samples were centrifuged at 15,000 g for 30 minutes and resuspended in 20 µl of 0.1% sarkosyl-containing PBS condition (Revision Figure 1D). Using this protocol, we also detected signals in the NBH control.

These results indicate ongoing challenges in optimizing prion extraction from skeletal muscle. Unlike brain tissue, where prion levels are significantly higher, skeletal muscle presents difficulties due to lower prion concentrations. In brain samples, dilutions still result in positive signals only from prion-enriched conditions. However, for skeletal muscle, prion extraction is not as straightforward, highlighting the need for further refinement of the protocol to achieve reliable detection and differentiation between prion-infected and control samples.

Inclusion of prion strain with limited extraneural replication. Reliance on three prion strains limits the relevance. Inclusion of a strain with limited extraneural replication is suggested.

Response: To address this limitation, we propose a comprehensive discussion on the systemic nature of prion diseases, emphasizing the need for future research to explore potential strains with restricted replication patterns.

Revision: There is a significant interest in prion deposits in skeletal muscles as potential sources of prion spreading. The consumption of beef products from cattle infected with bovine spongiform encephalopathy (BSE) prions caused new variant Creutzfeldt-Jakob disease, raising early concerns about the transmission of prions from cervids to humans (1-3). This remains a hot topic in the field (4-7), underscoring the importance of our longitudinal transcriptomic analysis in different extraneural organs. However, prion strains with restricted extraneural replication which we could use as control have not been described in mice used as prion animal models. According to our knowledge and the existing literature, there is no documentation of any mouse-adapted prion strains that are unable to propagate prions outside the central nervous system (CNS). Although this does not apply to our study, it is important to note that hamster-derived prion strains such as HY and DY exhibit different replication patterns. Hamsters infected with HY TME prions show detectable infectivity and/or PrPSc in the CNS, lymphoreticular system, skeletal muscle, nasal secretions, and blood (8-11). Conversely, prion infectivity and/or PrPSc in DY TME-infected hamsters is restricted to the CNS (12-14).

In mice, the situation is quite different, and there are no prion strains with restricted extraneural replication. Instead, studies have focused on models where prion protein (PrPC) is absent in all tissues except skeletal muscle, which is essentially the opposite of the condition requested by the reviewer. For instance, research has demonstrated that prion levels in skeletal muscles are 5-10% of those observed in the brain (1). Here, transgenic mice (Tg(α-actin-MoPrP)6906/Prnp0/0) expressing PrPC only in skeletal muscles (and barely detectable in the CNS) were created. After intramuscular prion injection, these mice showed that skeletal muscles could propagate prions (1). Additionally, another study found that prions were not detectable in skeletal muscle at early stages (32 and 60 days post intracerebral prion inoculation) unless experimental autoimmune myositis was induced, which increased prion spread to skeletal muscle (15).

This comprehensive discussion underscores the absence of a mouse model prion strain with limited extraneural replication, highlighting a gap in current research that our study aims to address indirectly through our systemic approach.

Clarification on statistical methods. Lack of details on statistical tests used for comparing GLUL levels in Figures 3 and 4.

Response: We clarified the statistical tests used, specifying whether they are parametric or non-parametric, and provide a rationale for the chosen methods. It is important to note that GLUL upregulation is significant, as evident from Figure 4. At 8 wpi, the fold change for RML6 is above 3, for ME7 is above 1.5, and for 22L is above 2. The fold change in later timepoints is increasingly larger.

Revision: For Figure 3E, normalized raw counts for the GLUL gene in control and sCJD patients were analyzed using the DESeq2 package, with related false discovery rate (FDR) calculations. DESeq2 is appropriate for RNA-seq data as it models count data using a negative binomial distribution, suitable for overdispersed count data commonly found in RNA-seq experiments. The normalization and FDR calculation ensure that the comparisons between control and sCJD patients are statistically robust.

In Figure 3G, Western blot densitometry data were analyzed using the Mann-Whitney U test, resulting in a p-value of 0.01072. The Mann-Whitney U test is a non-parametric test that does not assume normal distribution, making it suitable for small sample sizes and non-normally distributed data, which is often the case in Western blot densitometry. For statistical analyses in Figure 4A, Mann-Whitney U test was used. In Figure 4C, we applied the t-test (as the standard deviations were consistent) with Bonferroni correction to account for multiple comparisons. The t-test is a parametric test suitable for normally distributed data, and the Bonferroni correction adjusts for the increased risk of Type I errors when multiple comparisons are made, ensuring the results are not due to chance. Additionally, we used one-way ANOVA corrected with Kruskal-Wallis, a non-parametric method, to confirm our findings (Revision Figure 2). The results from both statistical tests were in strong agreement, validating our analysis.

In Supplementary Figure 6, the Mann-Whitney U test was used due to its non-parametric nature, which is suitable for data that do not assume a normal distribution. This test was chosen to provide a robust analysis of the data, which did not fit the assumptions required for parametric tests.

These methods were selected based on the data distribution and the need for accurate statistical analysis to validate the significance of our findings. The choice of DESeq2 for RNA-seq data, Mann-Whitney U for non-normally distributed data, and t-tests with Bonferroni correction for normally distributed data ensures that our analyses are appropriately tailored to the characteristics of the data, providing reliable and valid results.

Details on CJD cases analyzed. Information regarding the types of CJD analyzed is missing.

Response: We ensured that details on the types of CJD cases analyzed, including genotype, strain type, and age, are clearly presented in the main text and supplementary materials.

Revision: A comprehensive description of the sCJD cases, including genotype, strain type, and age, was accessible in Supplementary Table 5 (and here, in Revision Figure 3). This table also provides the reasons why some biosamples were excluded from the final bulk RNA sequencing and downstream analysis.

Recent publications on prion seeding activity. Mention recent publications showing prion seeding activity in extraneural tissues.

Response: We will update the Introduction section to include references to recent studies demonstrating prion seeding activity in extraneural tissues of sCJD and vCJD patients using RT-QuIC or PMCA assays.

Revision: We will discuss and cite additional papers on this topic, highlighting the growing body of evidence for prion seeding activity in extraneural tissues. These references will provide a comprehensive background on the detection and significance of prion seeding in peripheral tissues, thereby strengthening the context and relevance of our study.

Reviewer #2 Comment:

The RNA sequencing of human skeletal muscle samples identified only one common gene between human and mouse conditions. There is concern that this gene may be a bystander result of terminal disease stage pathophysiology in both animals and human.

Response: We strengthened the evidence supporting GLUL as early and altered gene by including additional timepoint analyses and showing its presence at earlier disease stages.

Revision: The upregulation of GLUL cannot be attributed to a bystander effect result of terminal disease stage pathophysiology, as it is consistently upregulated across all analyzed timepoints. We performed weighted gene co-expression network analysis (WGCNA) grouped by disease stages and GLUL belongs to the orange module (upregulated genes throughout all timestages in both main (Figure 2A - original manuscript) and validation (Figure 3A - original manuscript) cohort). We also included a comparison of GLUL expression between RML6 and NBH. As shown in the figure (Revision Figure 4), GLUL is upregulated at all individual timepoints. This finding is corroborated at both the transcriptional and protein levels, including other prion strains (Figure 4 - original manuscript) from a further animal cohort for RML6 condition. This consistent upregulation across various stages supports GLUL as a robust altered genes and possible biomarker for prion disease progression.

Cautious interpretation of GLUL dysregulation specificity. The claim that GLUL dysregulation is specific to prion diseases should be mentioned more cautiously due to the small sample number of other neurodegenerative diseases (NDs). The finding would be stronger if a meta-analysis of possible available data from human ND cohorts could be examined.

Response: We will rephrase our conclusion to acknowledge the sample size limitation and suggest further studies for confirmation. However, due to the poor sample availability of skeletal muscles biopsis from other NDs, related metadata are complicated to be found.

Revision: Modify the discussion section to reflect a more cautious interpretation, emphasizing the need for larger cohort studies to confirm GLUL specificity.

Post-transcriptional modifications of GLUL. Explore the possibility of GLUL being modified through RNA editing affecting its expression.

Response: We investigated the potential post-transcriptional modifications of GLUL, such as RNA editing, and their impact on its expression and function.

Revision: We will add a paragraph named "Lack of post-transcriptional changes in extra-neural organs of prion-inoculated mice".

Results: We calculated the genome-wide adenosine-to-inosine editing index (AEI) to measure global RNA editing levels (16), the preferential site of RNA editing in mammals. Blood global editing levels rose steadily during aging but were independent of prion inoculation (Revision Figure 5A). No AEI differences were seen in muscle or spleen (Revision Figure 5B and C). To determine recoding of individual transcripts, we aligned our sequencing results to previously published high-confidence AEI recoding sites (17). We found Flnb and Copa in the spleen and Cog3 in blood to be significantly recoded (Revision Figure 5D). However, Glul did not show significant recoding.

Alternative splicing can give rise to disease-associated differentially used transcripts (18). In contrast to our previous results in the brain (19), the present alternative splicing analyses in extraneural organs showed only minor alterations (Revision Figure 5E). Necap2, Myl6 and Srsf5 transcripts were alternatively spliced across multiple organs and prion incubation times (Revision Table 1). Only in two out of a total of 21 splice variants differential transcript usage was accompanied by differential gene expression: upregulation of Myl6 in blood at 4 wpi and downregulation of Ms4a6c in blood at 14 wpi.

Discussion: Except for Flnb, Copa and Cog3, we were unable to find evidence for broad dysregulation of posttranscriptional RNA editing, in contrast a recent report (20) but in line with our previous findings (19). Furthermore, splicing analysis suggests that alternative splicing was largely unlinked from gene expression changes.

Method: Adenosine-to-inosine editing index (AEI) was calculated as previously published (16). Herein, raw fastq reads were uniquely aligned to a murine mm10 reference genome using STAR v2.7.3 with the filter outFilterMultimapNmax=1. RNAEditingIndexer (https://github.com/a2iEditing/RNAEditingIndexer) was used to calculate per-sample AEI.

We identified gene-specific RNA editing based on a recently published list of high-confidence targets of Adar (17)as follows. RediToolsKnown.py from REDItools (21) was applied on uniquely aligned samples as mentioned above. This yielded per-site lists of A-to-I editing on which we applied the following thresholds: (a) a minimum of 3 alternative reads per site per sample (b) a minimal editing frequency of 1 % per site (c) criteria a) and b) are fulfilled in at least floor(2/3 * n) biological replicates, n is total number of biological replicates per group (d) transcripts of site present in at least 2 biological control replicates. Multiple testing of sites passing above-mentioned thresholds was performed using REDIT (https://github.com/gxiaolab/REDITs) and adjusted for false discovery rate (FDR) according to Benjamini-Hochberg, we considered sites with an FDR For alternative splicing, SGSeq R package (22) was employed to find splicing events characterized by two or more splice variants. Exons and splice junction predictions were obtained from BAM filesPrediction of exons and splice junctions was first made for each sample individually. Then the predictions for all samples were merged and we obtained a common set of transcript features. Overlapping exons were disjoint into non-overlapping exon bins and a genome-wide splice graph was compiled based on splice junctions and exon bins. A single value for each variant was produced by adding up the 5' and 3' counts, or, if these represented the same transcript features, by considering the unique value. These counts were then fed to DEXSeq (23). We analyzed differential usage of variants across a single event, in-stead of quantifying differential usage of exons across a single gene. We retained only variants with at least five counts in at least three samples (of any condition). After filtering, the events associated with a single variant were discarded. Differential analysis was then performed implementing a sam-ple+exon+condition:exon model in DEXSeq. Differentially expressed isoforms were defined as isoforms changing with FDR < 0.05. In the case of differentially used splice variants in muscle on 12 wpi, this dataset was considered as an outlier and hence excluded due to excessively reported splice variants (1,788 events compared to 5 or less on all other time-points and extraneural organs).

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary: The authors present a comprehensive transcriptomic analysis of different organ tissue samples (blood, spleen and muscle) from established prion models RML6, or 22L, or ME7 strain mouse-adapted scrapie, at a significant number of timepoints recapitulating early, asymptomatic preclinical, clinical and terminal stages of disease. The transcriptome profile of each tissue at each progression stage is a very useful information, on disease relevant transcriptional changes of variable significance. Eigengene networks were constructed to study the relationships between 39 modules only 2 were significant (orange and dark green) in skeletal muscle from all 3 tissues examined. The authors support they reveal a higher order organization of the late-stage disease transcriptome and provide insights into the modular architecture of gene expression during disease progression (early, presymptomatic and symptomatic) followed by 20 gene hub identification. In the context of network analysis, a high preservation with a Z-summary statistic > 1.96 suggests that the module's connectivity pattern is significantly preserved across different networks of the module's structure as RML6 disease progresses and it is not random compared to control.

The RNA sequencing of human skeletal muscle samples further validated and only 1 common gene between human and mouse condition confirmed by immunoblotting. The GLUL gene was further investigated in terms of gene and protein expression in various mouse disease models along with human skeletal muscle CJD autopsy material. Their findings did not correlate between human and all mice models completely. The increase in GLUL expression is accompanied by changes in glutamate/glutamine metabolism and reduced glutamate levels in CJD skeletal muscle. These alterations were only specific to prion diseases, as they were not confirmed in other neurodegenerative conditions such as amyotrophic lateral sclerosis, Alzheimer's disease, or dementia with Lewy bodies. The authors propose GLUL dysregulation as a potential novel biomarker for prion disease progression, during the preclinical stages with potentially useful efficacy for monitoring of therapies.

Comments

• The RNA sequencing of human skeletal muscle samples identified only 1 common gene between human and mouse condition confirmed by immunoblotting at the terminal stage of the disease. How can they conclude that this specific gene is not a bystander result of the known pathophysiology at the terminal disease stage? The mouse data are not solidly consistent with the biomarker expression.
• The claim that GLUL dysregulation with a result in glutamate/glutamine metabolism is only specific to prion progression only eventhough interesting should be mentioned with a more cautious way as the sample number of other NDs is small. The finding would be significantly stronger if metanalysis of possible available data of human ND cohorts could be examined.
• Post-transcriptional modifications have been described as potential contributors to prion pathogenesis therefore, the authors should also explore the possibility of GLUL being modified through RNA editing affecting its expression.
• The experiments are well designed and executed and the analysis methods are explained in detail. The figures are elegantly presented with adequate information.

Significance

The authors present a very well designed and comprehensive transcriptomic analysis of several tissue organs related to prion disease progression and validation data from multiple mouse models as well as human CJD skleletal muscle tissue.

They provide a strong and logic experimental strategy with adequate validation comparing multiple strains with disease onset differences and human tissue. The result was to identify GLUL as a potential biomarker of prion specific disease progression without any overlap with other NDs sharing pathophysiology.

The finding is interesting to the field but most importantly the established transcriptome profiles available will by of great use for future use from relevant basic research and also translational studies.

The presented study is an important addition to the field without any other comparable datasets regarding skeletal muscle analysis in CJD.

Any novel biomarker for progression of prion disease is extremely important and its potential link with pathogenesis would be of paramount importance.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

The manuscript by Caredio et al is a follow-up of their previous work (Sorce et al,PloS Pathogens 2020), wherein they conducted genome-wide transcriptomic analyses on the brains of prion-infected mice throughout the course of the experimental disease. In the present study, the authors extended their analysis to extraneural tissues of prion-infected mice, including hindlimb skeletal muscles, spleen and blood. Their key findings indicate upregulation of the glutamate-ammonia ligase (GLUL) gene in the skeletal muscle, a pattern also observed across different mouse prion strains and notably in human cases of sporadic CJD, albeit in a relatively small cohort.

A major limitation of the manuscript is the absence of evidence demonstrating the presence of prions or prion-seeding activity, and the lack of correlation with transcriptomic analyses, in any of the extraneural tissues and different timepoints. This omission is surprising given its inclusion in their initial publication in PloS Pathogens. Particularly concerning are the mouse experiments, where intracerebral inoculations were performed, suggesting potential presence of prions in terminal nerve endings and then muscles only at late stage of the disease, due to anterograde axonal transport. The reliance on three prion strains (22L, ME7, and RML) that replicate extraneurally limits the relevance of the study. Including at least one strain with limited or no capacity for extraneural replication could help distinguish whether observed transcriptomic alterations are directly linked to prion replication or are indirect consequences, particularly within the brain. This would also have prevented a significant misinterpretation in the discussion section. The authors delve into the alterations observed in the spleens of affected mice at the terminal disease stage. They attribute these alterations to the consequence of intracerebral inoculation, suggesting a delayed accumulation of prions in lymphatic tissues compared to oral or intraperitoneal inoculation routes. However, considering the high volume and dose inoculated (30 µL 10%), there is likely spillover from the brain, resulting in an intravenous-like inoculation concurrent with the intracerebral infection. Consequently, prion replication occurs rapidly in the lymphoid tissue. Given this, the tardy transcriptomic alterations observed in the spleens become even more surprising.

Information regarding the statistical tests used to compare GLUL levels in Figures 3 and 4, and whether these tests are parametric or not, is missing. Given the relatively low differences observed and the substantial SEM, clarification on statistical methods is imperative for interpreting the results accurately.

Providing details on the types of CJD analyzed, such as genotype, strain type, and age, would enhance the manuscript's comprehensiveness. While this information may be available in supplementary Table 8, we had no access to it.

Minor point: in the introduction, it may be worth mentioning recent publications showing presence of prion seeding activity in many extraneural tissues from humans infected with sporadic CJD and or vCJD, using RT-QuIC or PMCA assays.

Significance

Given the aforementioned concerns, the correlates between prion replication and the GLUL/glutamate-glutamine metabolism alterations are thus highly uncertain. This limits the general significance of the study.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

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

Brunet and colleagues utilized expansion microscopy to identify the distinct localizations of Alms1a and Alms1b at the centrosome in transgenic flies expressing Alms1a-Tomato or Alms1b-GFP-6Myc. Their imaging results are of exceptional quality and include insightful observations and discussion. However, these experiments were exclusively conducted in the transgenic flies, and no evidence has been provided for the expression of endogenous Alms1b proteins. Moreover, the defect in centriole duplication was observed only when Alms1 was depleted via RNA interference, not through gene knockout. Although some discussion is provided to address this discrepancy, the evidence remains unconvincing. More robust evidence is necessary to support their findings.

Major comments

1. One of the primary observations is the defect in centriole duplication in the absence of Alms1. This finding was only observed with RNAi and not with gene knockout. The authors performed the RNAi experiments in Alms1-deleted cells (Alms1del3) and did not observe the defects in centriole duplication. Therefore, authors concluded that acute loss of Alms1 is responsible for centriole duplication in germline stem cells. However, it is still difficult for me to reach that conclusion with only this evidence.

A. Could the authors provide the western blot and staining results of Alms1a and Alms1b proteins after RNAi and gene knockout?

B. The most reliable and widely accepted method to determine the specificity of RNAi (whether it is an off-target effect or not) is through genetic rescue experiments. Could the authors provide rescue data? It would be great if the authors could show the rescue of centriolar signals of Plk4, Sas-6, and Ana2, in addition to centriole duplication.

C. The authors failed to discriminate which gene is crucial for centriole duplication in GSGs due to the sequence similarity. Linked to the above point, could the authors show the rescue results with either or both Alms1a and/or Alms1b genes? If only one gene can rescue the centriole duplication defects, it would clarify the specific functions of Alms1a and Alms1b in this process.

D. (Optional) Considering the results in the manuscript, the authors appear to have good techniques in genetic manipulation in flies. If so, generating transgenic flies of Alms1a and Alms1b tagged with a degron (e.g., dTAG or destabilization domain [DD]) to observe centriole duplication defects upon rapid protein degradation might be feasible to support their observations. 2. There were no results with endogenous Alms1a and Alms1b in the manuscript. Could the authors provide immunostaining results for endogenous Alms1a and Alms1b? If expansion microscopy is not available due to antibody specificity issues, standard confocal imaging would be sufficient. 3. In figure 3C and D, the authors noted measurements from 7 to 8 testes per group. It would be helpful to also know how many centrosomes were measured, as done in Figure 4C.

Minor comments

1. In figure S1, the orientation of the gene is not consistent between endogenous and transgenic Alms1. Could the authors make it consistent for readers to intuitively recognize it? Additionally, the promoter used for transgenic gene expression is not specified. Please provide this information as well.
2. On page 5, the statement '~, whereas Alms1b is associated with post-duplication maturation of the centriole." needs revision. The authors observed strong Alms1b-GFP signals in the centrioles at the end of the SC stage or during meiotic division but did no observe any defects in centriole maturation at the current stages. Furthermore, Alms1b was not detected from rapidly duplicating centrioles in syncytial embryos and dividing neuroblasts (Page 6), making it making it hard to understand that that Alms1b has a role in post-duplication centriole maturation.
3. In figure S2, the centrosome drawings are unclear (initially thought to be 'v' marks). Could a centrosome drawing be added to the legend for clarity?

Significance

ALMS1 is a well-known protein which is important for cilia formation in human cells. Recent research by Chen and Yamashita (eLife, 2020) highlighted its significance in centrosome duplication in Drosophila's germ line stem cells (GSGs). Brunet and colleagues' findings align with these previous studies but go further by elucidating the localization of Alms1a as a centriolar and pericentriolar material protein, and Alms1b solely as a centriolar protein, using expansion microscopy. Moreover, their research suggests that only acute, not chronic, loss of Alms1 leads to defects in centriole duplication, proposing that cells may develop compensatory mechanisms for Alms1 loss in GSGs.

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 study takes a look at the role of Alms1a and b in centriole formation in Drosophila tissues using genetics and microscopy approaches. Alsm1a RNAi was previously shown to lead to loss of centrioles on the male germ line, but the molecular mechanism of how Alsm1a provides this function is not known. This is what this manuscript addresses.

The general proposal of this manuscript is that Alsm1a is a key regulator of centriole formation in Drosophila tissues (embryos, male germ line and neuroblasts analysed). Alms1a is an inner PCM protein and functions downstream of PLK4 to drive procentriole formation.

This is an interesting advance in understanding of centriole formation in flies. The authors managed to pull of beautiful expansion microscopy and produced images of centrioles both by immunofluorescence and electron microscopy. The data quality is very high, the quantifications lack a little behind in places. The manuscript would benefit from thinking about structure and presentation. The proposal that chronic loss of Alms1a (mutants) is well tolerated while acute (RNAi) is not is a bit puzzling. It is not impossible, but RNAi is not that acute either. It would be important to clarify this. To make sure there is no glitch in the tools, the newly generated alms1 deletions should be better characterised to clarify whether a compensatory mechanism exists, that upon chronic depletion rescues all phenotypes described by RNAi. There are also ways to test whether the Alsm1a fluorescent transgene introduce confounding effects, which if tested would be an improvement (see below). The genetic and immunofluorescence analysis got the authors quite far, the manuscript lacks biochemical analysis to strengthen the proposed molecular mechanism and clarify whether key and easy to predict interactions of Alms1 actually do occur. This would be a big plus but is not limiting. There are also inconsistencies that need clarification, for instance the title of figure 6 is that Alms1 proteins act downstream of Plk4, yet the model in the discussion proposes that Alsm1 stabilises Plk4 which does not fit and hinges on quantification of Plk4 levels that is perhaps currently not robust enough. The statement that Alms1 is required for PLK4 activity is too strong based on the data provided or provide data on PLK4 activity. The attempt to check what is going on in other systems is appreciated, in RPE cells Alms1 comes in only after 120nm elongation. Some of the quantifications could be done more robustly, using ratiometric analsysis rather than directly comparing intensity levels. Referencing in the manuscript and discussion should be improved.

Specific points

1. Provide a thorough characterisation of the alsm1 deletions generated using qPCR to measure RNA levels in the flies. Provide whether the alleles are viable and fertile and clarify all genotypes in the manuscript. Can the mutants suppress the Plk4-ND overexpression effect?
2. BamGal4 alms1a RNAi, SCs loose centrioles, those that keep centrioles, have Alms1a, but fail to initiate procentriole formation. To strengthen this view please provide nos-Gal4 alms1a RNAi, Alms1a-Tomatoe data showing that now Alms1a-Tomato is not present accept on the mother centrosomes in GCS, ruling out anything unpredicted happens by introducing the Alsm1a-Tomatoe.
3. Quantifications, measure fluorescence ratios (e.g. Figure 3 C,D quantify the ratio of Asl/PlP to Bld10, similarly for PLK4-ND in the relevant figure).
4. A model for Alms1a in the style of Figur5A would be great.
5. What happens to Alsm1a upon Plk4 inhibition? Experiments like this could strengthen the validity of the hierarchy of events proposed.

Other comments

The organisation and presentation should be improved. The figures reorganised to group what belongs together, I would suggest moving human RPE cell analysis to supplementary data and bring the beautiful EM of BamGal4-Alms1 RNAi, Alsm1a-Tomatoe into the main manuscript.

In the deletions asterless levels are reduced in SC but not in the testis tip, why is that?

Conclusion: "while Alms1b is only detected on mature centrioles as it is absent from rapidly duplicating centrioles of syncytial embryos or from duplicating centrioles of dividing neuroblasts." Perhaps add, Alms1b when expressed.

BamGal4 alms1a RNAi, SCs loose centrioles, those that keep centrioles, have Alms1a, but fail to initiate procentriole formation. To strengthen this view please provide nos-Gal4 alms1a RNAi, Alms1a-Tomatoe data showing that now Alms1a-Tomato is not present accept on the mother centrosomes in GCS.

The idea to compare nos-gal4 and bam-gal4 driving Alms1a RNAi is good. Providing a scheme of when these are expressed in the male germ line will help illustrate the experimental strategy. BamGal4 Alsm1a RNAi leads to loss of centrioles in SCs

Alms1b (CG12184) is according to published RNAseq data not expressed in neuroblasts, so it makes sense they do not observe (Knoblich lab data), which could be cited here: "in agreement with RNA-seq data showing low expression of alms1b in neurons and glial cells (Li et al., 2022)."

Fig S5 what cells were analysed by TEM?

Fig S6 monochrome images confirm what is what.

Figure 4 A for clarity it would be helpful to provide landmarks, marker for stem cells (Vasa) or the Hub to be able to understand what we are looking at. Same for Fig 4E, Mira or Dpn? The DAPI staining does not allow in the images provided to identify NBs.

Figure 5 For the fluorescence profile plots it would be nice to see the average plus the standard deviation of the signal in the quantifications.

Figure 6 D is not convincing, how were the dots visible chosen for the quantification?

Figure 1C. Quantification of protein levels is not provided (is this because the expanded ultrastructural approach is not linear precluding quantification?)

Figure 2 B radial dimensions, it would be great to show the sample average and standard deviation.

I don't find figure S2 helpful

Significance

Understanding the process of centriole duplication is an important topic that should be relevant to a broader cell biological community especially since the proteins of interest have disease relevance.. This study provides new insights into how Plk4 dependent centriole duplication takes place in Drosophila.

The manuscript is strong on the microscopic images both immune fluorescence and TEM. The function of Alms proteins is dissected genetically no biochemical analysis is provided, which is a limitation. Another limitation currently is the uncertainty about the discrepancy between the RNAi and the mutant results. If the mutants are confirmed and technical issues can be ruled out the finding that a compensatory mechanism exists that suppresses chronic loss of Alms1 proteins could bar very interesting.

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

Much is known about the centriole duplication machinery and the centriole duplication cycle thanks to key screens performed in C.elegans in the past 20-25 years and subsequent work in Drosophila or human cells. It has been proposed that the core duplication machinery consists of 5 proteins implicating Cep192-PLK4-SAS6-SAS5 and SAS4, with specific variations according to the model system or cell line studied. Recently, a role for ALM proteins in the duplication of centrioles in the Drosophila male germ line has been reported, but in this study the mechanism involved has not been identified or described.

In the Brunet study, the authors show a role for Alm proteins in centriole duplication in different tissues in flies- showing that these proteins have more a ubiquitous role in centriole duplication rather than a restricted role in the male germ line as initially described. This is already accounting for the novelty of this paper. However, this study goes well beyond previous studies as it proposes two very novel concepts. The first one is that Alm proteins can stabilize PLK4, the sole kinase implicated in centriole duplication, and so be responsible for maintaining kinase activity during a time window where daughter centrioles are generated. This is extremely interesting and makes a lot of sense and the data is very convincing. The second novel concept is that loss of Alm can have different consequences according to the- I do not even know how to call it- loss condition- leading to Alm deficiency. While I think this is quite novel, interesting and maybe even real, I think the authors conclude very strongly on the differences between a gene knock out or gene knock down conditions. I think that they may want to tune down their conclusions related to this part, as many more data would be required to conclude in this way.

In other words, I think the paper is of high interest to a broad field of cell biology with interests in centrosome, cilia and the regulation of centriole duplication.

Major points:

1. I did not find a rescue experiment of the Alm deletions with the Alm transgenes.
2. If I understood the authors correctly, Alm def compensate for centrosome duplication by a yet unknown process, while Gal4-induced depletions do not. First, calling Gal4- induce RNAi- acute depletion is not correct. This is certainly not acute and so another designation has to be found. Acute is something like a degron such as Auxin or dTag where the protein is degraded in a very acute manner. RNAi targets the mRNA, so if the protein is already made, it will not suffer from the RNAi treatment. Second, are the authors sure that either their crispr strategy did not generate any other knock out- hence the essentiality of rescuing these mutations. Or alternatively, are they sure about their RNAi conditions? So can they add the RNAi conditions on the background of their deficiency and see that nothing changes? Third if depletion through RNAi indeed leads to a more evident role of Alm proteins, one is expected to see this over time. Can they do a clone-using an FRT site recombined with their Alm mutation, so that the initial cell divisions does not contain Alm, and so it is expected to fail duplication, which may be overcome with time? So a large clone might have progeny with cells with centrosomes (the young ones) and without (the older ones). Can they show that RNAi depletion in cells that will generate sensory cilia- these are not assembled? Because I am assuming that the mutants are not uncoordinated...
3. If they think that Almdef flies compensate for Alm loss - can they analyse levels of PLK4, SAS6, Ana2 and SAS4 in the mother centrioles?

Minor points

1. Some figures (the majority) lack scale bars.
2. I do not think that one can consider centrioles that are not in rosettes to be made de novo. They might just have disengaged. The "novo" centriole should be removed. Actually, PLK4 ND generates extra centrosomes, this is sufficient.

Significance

The article by Brunet and colleagues investigates the role of ALMs proteins in centriole duplication. Centrioles are the core constituents of centrosomes and as such contribute to microtubule nucleation. Centrioles can behave as basal bodies, providing essential function sin cilia assembly and function.

Here the authors have use Drosophila to characterize the role of Alm proteins. They show that 2 isoforms with distinct behaviour are expressed with different localizations. Further, through the assessment of different tissues and loss-of-function conditions, they propose a role for Alm proteins in centriole duplication.

Overall the paper is very well written and easy to follow.

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

In this manuscript, the authors performed extensive genetic analyses in yeast on the functions of Med15 regions under multiple stress conditions, linking cell growth phenotypes, gene expression and protein-protein interaction. Med15 is an activator-contacting subunit of the Mediator complex and the functions of yeast Med15 have been extensively studied. In particular, the authors attempted to understand the roles of a poly-Q region (Q1), its length and composition in stress response phenotypes, Med15-mediated gene expression, and interaction with transcription factor Msn2. Results from this work are consistent with several previous studies and revealed some new insights. The authors concluded that robust Med15 activities required the Q1 tract and the length of Q1 tract modulates activity in a context-dependent manner. While the study is well executed and the conclusions are generally sound, several concerns listed below should be addressed and some clarifications should be made.

Major comments:

1. Abstract, "We also observed that distinct glutamine tracts and Med15 phosphorylation affected the activities of the KIX domain". Fig. 1 shows the effects of KIXQ2Q3 deletion and p7 phosphor-dead mutant under Acetic acid and Ketoconazole treatment, but does not demonstrate that these domains or phosphorylation affects KIX domain activities.
2. Is it known that Med15 is dephosphorylated under stress conditions other than osmotic challenge? Another explanation for D7P/D30P mutant results (Fig. 1B) might be that Med15 phosphorylation in unstressed cells is important for certain types of stress response (acetic acid and Keto). In contrast, the observation that D7P has no effects on osmotic stress (Fig. 1B) might suggest that phosphor-Med15 is dispensable for function. Some explanations on how to ascertain the roles of Med15 phosphorylation would be needed.
3. Fig. 4, what is the rationale of analyzing basal expression rather than activated expression of Gcn4 and Msn2 dependent genes? Gal4 and Hap5 dependent genes could be measured as well, in order to complete the gene expression-phenotype correlation that the authors strive to make in this paper.
4. Fig. 4, Error bars should be provided on gene expression analysis. Gcn4 and Msn2 target genes should be highlighted separately to facilitate comparisons.
5. Results from Fig. 4 and 5 indicate that Spacer-Q1 and 12PQ-Q1, being the strongest interactors to Msn2, actually reduced HSP12 expression (a known Msn2 target gene). Some explanations would be needed. Page 12 Discussion paragraph 2 "Q1 substitutions that interfered with coiled-coil propensity had no effect on TF activity" would need to be revised and to include some discussions on this result.
6. Fig. 5C, additional explanation is needed on how interaction rank is determined and how error bars are obtained.
7. The idea that Q1 provides a molecular hinge to facilitate intramolecular interactions is interesting, but sounds like a possible scenario without further evidence. Are there any published structural studies on Med15/Mediator complex that might support this idea?

Minor comments:

1. SC-HLUM and SC-HLMU is used interchangeably in the legend and text. Please keep consistent. Explanations for these acronyms are not found in the Methods.
2. Fig. 2, 3. AcOH should be Acetic acid, to be consistent with Fig. 1.
3. Fig. 3B, error bars should be provided for growth measurements.

Significance

This work provided a detailed analysis on the roles of a specific poly-glutamine region in yeast Med15 functions and regulation. One conceptual advance of this work is that the structural flexibility rather than the sequence itself of Q1 tract proves to be critical for Med15 function. The ability to correlate Med15-Msn2 interaction with gene expression analysis demonstrated some technical novelty, given the power of genetic manipulation in yeast.

Med15 is a key Mediator subunit contacting several sequence-specific transcription activators. Its interaction with a number of transcription activators in yeast such as Gcn4 and Gal4, was previously studied as referenced in this manuscript. This manuscript first provided a quite comprehensive genetic mutational analysis and confirmed several findings in previous studies. The identification of critical Med15 regions for acetic acid response (and Hap5-dependent gene activation) and the analysis of Msn2-Med15 interactions appear to be novel. Researchers interested in eukaryotic transcriptional regulation would benefit from reading this study.

Field of expertise of this reviewer: mechanisms of transcriptional regulation, genetics, nuclear organization and function

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

The Mediator complex, a multicomponent complex, regulates the interaction between transcription factors and RNA polymerase II using protein interactions. In particular, Intrinsically Disordered Regions (IDRs) with in Med15. Copper and Frassler made extensive mutations to the three IDRs that are characterized with high glutamate content (poly-Q), the KIX domain which interacts with transcription factors and the MAD which interacts with the rest of the Mediator complex. The three poly-Q repeats are adjacent to Activator Binding Domains (ABD). The impact of mutant Med15 was measured with growth assays, co-IPs with a transcription factor, and transcriptional activation of a reporter during different stresses. Med15 is particular important in stress responsive transcription rather than basal transcription because in yeast it is nonessential. It can be repress and activation translation of genes. Using a series of internal deletions and substitutions impact of the mutations was tested on by measuring growth of strains, expression of Med15 regulated genes, and interaction with Msn2, a transcription factor that regulates to various stresses. This work adds to the body of research confirming that multiple weak/ transient interaction domains regulate Med15 function.

Major comments:

1. Without knowing the protein levels of the different mutants, it is difficult to contribute deletions of different regions with phenotypes measured. Various internal deletions decrease Med15 protein levels (Jedidi et al. 2010) while other affect the integrity of the Mediator complex. This study did not measure mRNA or protein levels of their mutants. However, (Jedidi et al. 2010) used Myc-tagged Med15 which affects regulation of Med15 via SNF1 (Gallagher et al. 2020). In another study, Med15 was N-terminally tagged and protein levels of some deletions increased (Herbig et al. 2010). It's unknown if other tags such as HA, TAP or FLAG affect Med15 regulation via SNF1. This study used untagged Med15 expressed from the native promoter which avoids these complications. It's also unknown if the differences in Med15 deletions are from reduced transcription, translation, or protein stability. There are commercial antibody that may work (https://www.genetex.com/Product/Detail/Gal-11-S-cerevisiae-antibody/GTX64110). There are several commercial antibodies to human Med15 but the cross reactivity has not been tested.
2. Quantification with ImageJ on spot assays is difficult because once growth has maxed out on the dense spots there is no resolution. Using more dilute spots is challenging because colony size is affected by the nearby colonies. The error bars are the mutants are large. Can quantitative growth curves be carried out in flasks or an automatic plate reader for better quantification?

Minor comments:

1. Why were the stress conditions chosen in figure 1B? These are only a subset of conditions that the med15 deletion is sensitive to. Aside from acetic acid the phenotypic profile of each deletion is similar. The bigger the deletion, the more severe the growth defect. The keto plate appears under loaded by comparing the number of colonies in the third dilution on the keto plate and the fourth spot on the YPD plate in the BY4742 (MED15 wild-type strain). Does keto lyse cells? Or was there that much variation between mutants? Perhaps the dose of keto needs to be calibrated. In figure 2B it looks like wildtype growth in keto was 70% of untreated growth.
2. Figures 1C and 1D are not discussed in the results. The authors should remind what the hGR assay measures when discussing the results. How is it different from the GAL4 transcriptional reporter?
3. In the D to A mutants some appear to be required for acetic acid tolerance. What was the pH of the media?
4. The labels between Figure 1 and 2 are inconsistent. 90 mM acetic versus 80mM AcOH, YPGal versus YPGalactose, SD+LKHU versus SD+KLUH. The mutants on 0.97M NaCl at 37oC from figure 2A grew more than 0.9M NaCl at 38oC. Also, in the text it says 37 oC.
5. Is the MED15 strain BY4742? In Figure 1 was it also transformed the pRS315? How was the plasmid maintained on the plates, specifically YPD?
6. Genes such as GAL and URA3 should in italics.
7. In the split Ub assays, was wildtype Msn2 and Med15 also present?
8. There is inconsistently naming of media in Media and Phenotype Testing section. The media is called synthetic complete media is labeled SC-URA or SC-LEU and at times in Results its called SD+K or SC-HULM. Is SC with all amino acids and SD without any amino acids?
9. The plasmid names in the supplemental table don't match the ones labeled in the figures.
10. Why was ALG9 used for normalization of qRT PCR?
11. The background of the strains is confusing. There appears to be two different med15 knockouts OY320 and JF1368. Which ones were used in which experiments? Some of the trains have a trp1 auxotrophic which affects stress response on it's own (González et al. 2008; Schroeder and Ikui 2019).

Significance

• General assessment: provide a summary of the strengths and limitations of the study. What are the strongest and most important aspects? What aspects of the study should be improved or could be developed? Extensive mutational analysis of Med15.
• Advance: compare the study to the closest related results in the literature or highlight results reported for the first time to your knowledge; does the study extend the knowledge in the field and in which way? Describe the nature of the advance and the resulting insights (for example: conceptual, technical, clinical, mechanistic, functional,...). This work confirms numerous other studies on the contribution of various Med15 domains on function.
• Audience: describe the type of audience ("specialized", "broad", "basic research", "translational/clinical", etc...) that will be interested or influenced by this research; how will this research be used by others; will it be of interest beyond the specific field? Incorporation of human domain substitutions could influence how people outside the field would interpret how Med15 interacts with transcription factors.
• Please 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. Expert in using yeast genetics and natural genetic variation to address underlying mechanisms of stress response to environmental toxins with a particular focus on transcription factors and TORC1.

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

Major Comments

The authors show that the amino acid content and length of Q1 affects transcription activity in a media-dependent way in a construct that includes Q1-ABD1 and a tailing Q/N rich region (Q1R). Briefly, different media conditions used as proxies for specific target TF activities varied in their sensitivity to the Q1 sequence content. However, the reason for this variation between target TF activities is not addressed, so the observations seem more anecdotal than insightful. One test performed suggests some of the Q1 sequence dependence may be due to changes in AD-ABD interactions, but this interesting possibility is not investigated further.

A split-ubiquitin two-hybrid assay, meant to detect interactions between Msn2 TAD and Med15-Q1R, showed clear Q1 sequence/composition-dependence when changed from polyQ tracts. In particular, replacement with leucine-rich tracts (12L and RvHs) significantly reduced interactions (as inferred from growth requirements in Fig 5B). Q1 consisting of just 10 spacer residues, 0 to 24 Q residues, or PQ repeats all had quite similar results suggesting retention of some Msn2 and Med15 interactions. Replacement with a helix-forming sequence from NAB3 gave intermediate results. Again, no explanation was offered for the observation but it seems probable that the NAB3 Q1 system is no longer reporting on Msn2 Med15 interactions.

The manuscript presents extensive assays, but a lack of consistency in conditions and constructs tested makes comparing different assays difficult. In particular, it would be valuable to have NAB3-Q1, FrHs-Q1, and RvHS-Q1 tested under conditions of high salt as that is indicated to be the Msn2 target condition (e.g. an additional result that would be presented in Fig 3B); this would be valuable to compare to the two-hybrid results. The relationship between Q1 polyQ length and Msn2 TAD-Med15 ABD1 binding is not clear from this assay as all had similar growth on the plates. A possible explanation for the inferred reduction in TAD-ABD1 binding in the leucine rich Q1 constructs is that this highly hydrophobic linker itself binds to ABD1 and is therefore self-inhibitory. There is also the unexplored/not discussed possibility that NAB3, 12L, and RvHs have off-target interactions that disrupt the TAD-ABD1 interactions.

The framing of the study and the title of the manuscript strongly suggest that there might be a relationship between coiled-coil formation and transcription activity. This is the basis for selection of many of the Q1 sequences tested, with the premise of either increasing or disrupting coiled-coil structure. These 'propensities' are quantified in Supp Fig 1; however, a significant limitation of this interpretation is that these propensities are bulk properties that presume formation of homo-dimers or homo-trimers, a situation that is not shown to be relevant for Med15 at a promoter. This means that Q1 is potentially only one of the multiple partners required for coiled-coil formation. So even if a tested sequence has high coiled-coil propensity, that may not be the case in the actual biological systems at play here. Another consideration to be entertained is how different solvent conditions (different media) may affect coiled-coil propensity. An unanswered question is whether Q1 may form coiled-coil structure either with other regions of Med15 and/or with other Mediator subunits or even other co-factors entirely. This is a question implied by the title of this paper, but the data presented address neither intra- nor inter-molecular interactions of the polyQ regions (the two-hybrid study is designed to probe the ABD-AD interactions).

A final proposed hypothesis was that Q1 acts as a hinge in a way analogous to what was reported for the huntingtin protein (ref 7). This is an attractive model but remains untested in this work. In particular, the Med15-Q1R construct used does not have multiple ABDs that would potentially be brought in close contact, so the results here cannot be interpreted as analogous to the huntingtin hinge model. Minor Comments:

Please explain the choice of the 10-residue spacer instead of a 12-residue spacer.

Page 14: "We observed that Q1 substitutions with increased coiled-coil propensity (Supplementary Figure 1) diminished TF activity while Q1 substitutions that interfered with coiled-coil propensity had no effect on TF activity (Fig. 3, 4), suggesting that the flexibility of the sequence is an important feature." There was no demonstration that those sequences in this context form CCs. There's no evidence of what is actually being modulated whether it's length, flexibility, or ability to interact with other regions of Med15 or even with other co-factors.

Page 15: "We confirmed that Msn2-dependent activities of Med15 are encoded by the region containing the Q1 tract and ABD1 (aa 116-277) and found that the KIX domain alone could also mediate an interaction with Msn2 (Fig. 5). This contrasts with the Gcn4- or Gal4-dependent growth or stress responses which are the result of additive interactions with Med15 that are characterized by weak, highly dispersed, multivalent interfaces. While it is not yet entirely clear if the interaction with Msn2 is similarly multivalent, we have shown that either the KIX domain alone or the Q1R region alone of Med15 was sufficient with no evidence of additivity." These statements are unsupported. While Gcn4 and Gal4 transcription activity has been shown to depend on multiple AD-ABD interactions, none of the data reported here shows that Msn2 does not (as is stated here, which undermines the "contrasts" argument. Further, based on the assays presented in Figure 1B, Msn2, Gal4, and Gcn4 behave similarly for the various Med15 constructs.

Page 16: "In all instances TF activity was reduced in the absence of the Med15 Q1 tract." This seems false based on the data presented. Met10 activity appears to have increased in Figure 4A.

Page 16: Reference to Figure 2C and Figure 2B are mislabeled. Should be Figure 2D and 2C, respectively.

Page 17: "The fact that residues at Q1 were not functionally constrained to be glutamine residues suggests the Q1 tract is not an interaction motif participating directly in protein-protein interactions." This is completely unsupported. There are no data presented that address interactions between Q1 and anything else.

Figure 2: Not clear which assays were at 30{degree sign}C vs 22{degree sign}C as they are not labeled in the figure. In Figure 2A, the label med15 should be med15Δ.

Figure 4: Interpretation of these results seems limited by only reporting YPD media conditions. May be helpful to include the conditions reported in Figure 1.

Figure 6: It is not clear what some elements of this figure are meant to represent. Is saw tooth always polyQ? or Is ABD1 always blue and ABD2 is always red. What then are the loops? The general premise of this figure does not seem to be supported by the actual experiments performed.

Supp Figure 3: "K is the Med15 fragment encompassing the KIX domain, aa 1-277." This aa range is KQ in the main text. Either the residue range is wrong, or the label is wrong.

Significance

This manuscript addresses an interesting topic. There appears to be a disconnect between the stated motivation and what was actually done. The large array of assays and conditions are difficult to compare, leaving the reader with a feeling that the authors have catalogued a lot of possibilities but that no generalizable or unifying insights are at hand. The attempt to present a model (Figure 6) is difficult to parse and is not directly supported by the data presented. Addressing the issues raised here could result in a work that is useful to the specific field of Med15 structure and function but of limited use at the moment to a wider audience.

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 #2

Evidence, reproducibility and clarity

In the manuscript by Nabeel-Shah et al. the authors identify ZBTB48 as a novel interactor of FTO. They show that ZBTB48 helps recruiting FTO to mRNA as well as the noncoding RNA Terra. Their results further suggest that this recruitment is required for FTO to demethylated m6A and m6Am RNA modification in target RNAs. This affects cellular rates of RNA turnover. Furthermore, the mechanism is involved in repressing colorectal cancer cell growth.

Overall, the authors present a new role of ZBTB48 in m6A and m6Am mediated RNA metabolism. They also suggest a nice model of ZBTB48 action, via the recruitment of the demethylase FTO. These findings will be of interest for the general RNA community and might be also relevant for cancer treatment. However, I have some concerns about the quality and analysis of the obtained data, especially the iCLIP and miCLIP experiments. The concerns are detailed below and should be addressed before publication.

Major concerns:

1. Figure 1B) I think that the results shown in the autoradiograph are not very convincing and suggest that the purification of ZBTB48 is not very clean. The radioactive signal covers the lane from 50 kDa to 200 kDa. The ZBTB48 alone is running around 80 kDa. If the purification is specific most of the signal should be above 80 kDa. Maybe it helps to also use higher RNAse concentrations: Having very short RNA pieces will allow to evaluate specificity of the purification, since protein-RNA complexes will just a bit above the size of the protein. This also applies to Figures S1E,F.

This concern could also apply to the generated iCLIP libraries and indicates that at least part of the obtained reads does not originate from ZBTB48 crosslinked RNA. The validation of protein-RNA interactions with RIP shown in Figure S2E supports the quality of the iCLIP data. Here some control RNAs should be analyzed to show that the RIP is not unspecifically enriching any RNA. 2. Regarding the co-occurrence of m6A sites as well as FTO and ZBTB48 binding sites shown in Figures 2B, C, F and G. The CLIP signal is a lot affected by crosslinking bias and read mappability. Therefore, to make these results more convincing it would be important to include additional iCLIP datasets (published other RBPs) for comparison. 3. Regarding Figure 2D and related analyses: I very much like this experiment and the results obtained here! I just wondered where the remaining reads go in the ZBTB48 knockdown. The introns? Maybe this becomes clearer in the meta profile representation used in Figure 1F.

For me the changes in the 5' UTR look most dramatic. Maybe this means that ZBTB48 is most important for recruitment of FTO to m6Am sites. Therefore, I think it would be good to differentiate the analyses in the remainder of the manuscript for m6A and m6Am sites. I first step would be to treat sites in the 5'UTR separate from the CDS and 3'UTR sites in the following analysis. 4. I have some concerns about the analysis of the miCLIP data. In Figure 3A antibody crosslinking in all conditions appears similar. Yet in panel B there seems less signal for FTO and ZBTB48 overexpression in all areas. Have there been more reads generated for the GFP control? Where is the rest of the reads going? I think it would be required here to identify peaks that significantly change between the conditions. Then ask do those peaks coincide with DRACH and were are they located. Also, in the Genome browser pictures the signal is going down at all locations. Why is that? Usually, miCLIP generates a lot of background peaks. These should be unchanged. 5. I am a bit confused by the author's interpretation of the results shown in Figure 5E. For me this plot shows that target transcripts are less downregulated and less upregulated than non-targets? Basically they are less regulated overall. In this context I also think that the representation in of the mean in figure 5E and F is misleading. Data distribution should be visible as violin or boxplot.

Minor points:

Figure 1D) I think the legend in the panel is confusing and does not add information. Especially since its in a different order than the categories on the y-Axis.

Figure S2D) I wondered if the authors controlled for mappability of reads when picking random sites. How do the authors account for that in iCLIP there will be less reads in introns compared to exons?

5E,F boxplots would be more suitable than barplots.xx

Figure 2I: Please do not use "+ve"

Figures with Microscopy pictures of cells have no scale bars or way too small.

The model that is shown in figure 7 is somehow misleading as it shows FTO binding only to ZBTB48 and not to the RNA.

Significance

Overall, the authors present a new role of ZBTB48 in m6A and m6Am mediated RNA metabolism. They also suggest a nice model of ZBTB48 action, via the recruitment of the demethylase FTO. These findings will be of interest for the general RNA community and might be also relevant for cancer treatment. However, I have some concerns about the quality and analysis of the obtained data, especially the iCLIP and miCLIP experiments. The concerns are detailed below and should be addressed before publication.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In this manuscript, Nabeel-Shah et al. identified that the telomeric zinc finger protein ZBTB48 help recruit FTO onto target RNA, including mRNAs and the telomere-associated regulatory RNA TERRA, to achieve cellular RNA m6A/m6Am demethylation, thereby regulating cell biology such as tumour growth studied in this study. The biochemistry and molecular biology experiments were done in HEK293 cells, and the cell models for tumour growth was done in HCT116 cells. However, I have identified both major and minor concerns that, when addressed, could further strengthen the findings and enhance the impact on the field.

Major concerns

1. Regarding the ZBTB48/FTO targets and m6A/m6Am level that are regulated by ZBTB48/FTO axis. Wei et al Molecular Cell 2018 (PMID: 30197295) (Figure 1B) quantified the m6A/m6Am levels upon knockdown of FTO in HeLa, HEK293, and 3T3-L1 cells, and found that the m6A/m6Am levels are generally mildly (10-20%) yet significantly upregulated on poly(A)+ RNAs. Here shown in the Figure 4A in this manuscript, the authors show that (1) siFTO and siZBTB48 led to ~2-3 fold upregulation of m6A and m6Am levels using total RNAs (dominant by rRNA in the population), and (2) that the m6A and m6Am levels are similar between siFTO and siZBTB48. Regarding (1), can the authors explain the discrepancy? This point is also relevant to the m6AIP-qRT-PCR results in this manuscript. Regarding (2), does this result suggest that ZBTB48 helps FTO to demethylate nearly all its targets, rather than a subset?
2. Considering the significance of how FTO achieves target specificity, can the author anticipate the extent of applicability of the proposed model? Does the interaction between ZBTB48 and FTO also exist in various human and mouse cell lines? If confirmed, this discovery would hold substantial value for the field. This interaction at least needs to be confirmed in HCT116 cells used for tumour growth model in this study.

Minor concerns.

1. The authors realised that knockdown of ZBTB48 does not change FTO levels, whereas overexpression of ZBTB48 leads to elevated FTO. It is unclear about the rationale behind overexpression studies?
2. Considering the multifunctional nature of ZBTB48, it's important to disentangle transcriptional and post-transcriptional roles of ZBTB48 to draw conclusions. I appreciate the analysis conducted in this manuscript. Is it possible to overexpress an DNA-binding mutant of ZBTB48 or RNA-binding mutant of ZBTB48 proteins?
3. In Fig 2B, the innet panel does not match the metagene profile regarding the difference.
4. In Fig 5H, there is a lack of experiments for comparison, i.e. siFTO and o/e FTO.
5. Page 3, "Genome-wide studies" should be "Transcriptome-wide studies".

Significance

Strengths: The biochemistry and molecular biology experiments are comprehensive and well designed. The analysis is robust, and the conclusions generally align with the presented data.

Limitations: Cell line-specific and/or species-specific interaction?

Advance: filling a gap in our knowledge of how FTO achieves target specificity.

Audience: Basica research in the field of RNA modifications and RNA Biology.

My expertise: Bioinformatics, gene regulation by RNA m6A modification

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

'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

The authors expand upon a previous method (PROMIS) that was developed to detect metabolite-protein interactions from cell extracts through co-elution of proteins and metabolites in size exclusion chromatography. Here, the authors use ion exchange chromatography. By comparing PROMIS and IEX datasets, the authors increase the confidence of detected interactions being true hits. The addition of IEX data significantly improves the power of the method to identify previously known/predicted protein-metabolite interactions. The authors also purify two proteins identified by the method and experimentally validates their novel interactions with dipeptide metabolites.

In general the paper is well written and provides a clear assessment of the two methods. The validation of two interactions is welcomed.

There are some issues that could be addressed with better clarification:

• While the two datasets for omniTICC appear to show good overlap, the two PROMIS replicates appear to have very little. Due to this low overlap, one of the metabolite interactions chosen for validation (dipeptide) was chosen by manually lowering the correlation cutoff of PROMIS as well as relying on two previous PROMIS datasets. Authors could comment on possible reasons for this low overlap.
• Experimentally validating the effect of Val-Leu on the enzyme activity of FabF would be a very good addition, but is not crucial to the paper.

Minor:

Line 173: In STITCH the authors find 1012 known or predicted interactions, while their method finds 92 of these interactions. Authors could comment on limitation of the unbiased method which may lead to missing these interactions (for example, sensitivity of MS detection of low abundance metabolites).

Minor: - Authors could comment on what are some drawbacks to using docking to estimate binding sites. Computational screening could be a powerful way to prioritize hits for validation, so could be worth a more detailed discussion here. - In figure 2 the color labelling of the experiments could be presented more clearly. - Authors could add in the text or methods how they calculate the Pearson correlation coefficient they use for determining significance. Overall the methods are well presented and the technical approach is very well described and impressive. In general the authors present the statistics such as likelihood over chance detection, in a way that helps to evaluate the accuracy of the separate and combined CF approaches.

Some comments:

• The PROMIS method and the omniTICC method are performed only in duplicate, where the two PROMIS experiments also differ in proteomics workflow. Could that be why the overlap between PROMIS experiments is so poor? Why not perform the experiments in biological triplicates?
• Authors correct for poor replicate overlap by lowering the requirements of one subset of metabolites. What's the argument for not doing this with all metabolite subsets? Would this interaction have been discovered just by using PROMIS in three different experiments (i.e. more biological replicates)? The benefit of adding the IEX analysis is clouded by the poor overlap of PROMIS data
• Line 353 - It would also be good to elaborate as to why other methods would be likely to miss this interaction. Is it because the authors method does not require the users to choose specific metabolites or proteins to focus on beforehand?
• Authors could comment on whether any of the high confidence interactions were also been observed in different IP experiments with E. coli (such the Lip-SMAP method reported by Piazza et al., 2018).
• Authors could comment on other ways to identify the potential binding site. For example, limited proteolysis of the protein in presence of metabolite has been used previously
• There is massive interconnectivity between NMPs and ribosomal proteins. It would be good to comment on this. Do the authors believe these are true interactions? Could there be another reason for this cluster? Do the STITCH interactions also show that these proteins interact with so many NMPs?
• Line 157 - The authors state that "isopropylmalate co-eluate with tens of proteins". This feels like a bit of an understatement if the actual numbers are: 92, 119, 303, 287

Minor:

• How did the number 1479 proteins come about? Are those detected in all 4 datasets? What about those detected in only some datasets? This information could be more clear

Significance

General assessment

The paper is a welcome comparison of using two co-elution MS methods for identifying protein-metabolite interactions. It is clear that these types of interactions are important for modulating protein activity but are not well studied. The paper provides a clear workflow for both the experimental and data analysis portions of interaction proteomics. The relatively low number of validated hits could limit its significance outside the specialised field.

Advance.

The paper makes a conceptual advance in interaction proteomics. It is perhaps not unexpected that combining two different interaction proteomics methods gives more accurate target identification than a single method alone. The paper also strengthens the evidence that dipeptides play a role in regulation that may be conserved in bacteria and plants.

Audience.

The audience for this paper is likely researchers that are actively involved in the field of interaction proteomics. The link between dipeptides and feedback regulation has been developed by the group over a few papers, and this report provides further evidence. The paper also shows that a weak kinetic effect in vitro can actually lead to a significant effect in vivo, which is a very interesting finding that is a good point of reference for future in vitro validation efforts. Often the effect of metabolites in vitro is weak, which could lead to the (possibly false) conclusion that it is not relevant in vivo (see e.g. weak effects of metabolites on enzyme activity in vitro in a recent LiP-SMap paper (https://doi.org/10.1038/s42003-023-05318-8)

Reviewer background:

Reviewers are active in the field of interaction proteomics and have previously used the Lip-SMAP method as well as thermal proteome shift method.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

In this work, Wagner et al aimed to comprehensively map protein-metabolite interaction in E. coli. They do so by extending a previously developed SEC-based co-elution method with an IEX-based co-elution method. The results is a more extensive and robust protein-metabolite network. As a validation, selected predicted protein-metabolite interactions are confirmed at the level of binding, as well as enzyme kinetics and even phenotype.

The key conclusions of the paper are mostly convincing, even more so because the authors provide a critical evaluation of the method. My only concern is that the results from the PROMIS replicates seem to be quite different (e.g. Fig 1h and Suppl Fig 1a). One wonders whether, given this apparent variation, two replicates are sufficient to define the protein-metabolite interactions with great confidence.

The author took care to make all data (proteomics & metabolomics) publicly available and the methods are clearly described.

The manuscript is well-written and the figures are clear, although Fig 1 would benefit from and explanation of the color coding in the legend. Personally, I feel that less emphasis on lumichrome would help focus the conclusion section.

Significance

This work is of high significance because it describes a method to comprehensively map protein-metabolite interaction that could relatively easily be applied to any organism. The work is of high quality and similar to work by for example Link et al (https://doi.org/10.1038/nbt.2489) or Piazza et al (https://doi.org/10.1016/j.cell.2017.12.006), but technically less challenging which increases its potential impact.

Beyond these technical advances, global studies like these are a great resource for anyone working on the functional characterization of proteins. Often, a (predicted) protein-metabolite interaction can be a crucial lead to find the function of a protein.

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, the authors demonstrate a strategy for uncovering metabolite-protein interactions, rooted in the concept of identifying co-fractionating metabolites and proteins through independent analyses of each fraction using untargeted proteomics and metabolomics. This approach is applied to E. coli protein lysates.

The same research group has previously advocated for utilizing the simultaneous elution of metabolites and proteins as an indication of interactions, as seen in Luzarowski et al. (2019) and Veyel et al. (2017), where metabolite-protein interactions in yeast and plants were investigated, respectively. Here, a variation of the original proposed method (PROMIS) is introduced, involving an additional chromatography separation (ion exchange) alongside the original size exclusion separation proposed in PROMIS. By incorporating an extra dimension to reduce sample complexity, both experimental and data analysis efforts are effectively doubled, entailing the processing of 48 additional fractions for proteomic and metabolomic analyses, in addition to the 40 already included in the original PROMIS protocol.

While previous studies have demonstrated the isolation of small molecule-protein complexes following ion-exchange based separation (Chan et al., 2012), the method described here does not introduce any fundamentally new concepts. However, it remains to be determined whether the addition of an extra separation dimension genuinely aids in accurately classifying new interactions. Although coincidental co-elution effects may be mitigated, there is a risk of disrupting genuine complexes through excessive handling of lysates.

In my view, there are several points that require clarification:

1. Line 138: There's an assumption that the metabolite profile will alter in IEX upon binding to proteins. While this might occur, it's not definite. Additionally, it's unclear why peaks associated with protein binding could also increase in intensity, and how this signifies a binding event.
2. Lines 147-154: The most crucial dataset in this paper isn't adequately delineated here. We learn that by combining SEC and IEX, 1479 proteins and 58 metabolites are identified, but what about when only one separation method is employed? What advantages does using IEX provide?
3. Line 170: Similar to point 2 regarding PMI.
4. Line 174: To gauge whether the proposed strategy aids in accurately classifying new interactions, the authors examined if their predicted interactions also appear in the STITCH database. Out of the 994 PMI in the network, 92 were found in STITCH. I'm uncertain if STITCH is the most suitable metric for this assessment, given it likely hasn't been updated since 2016. How does this PROMIS-IEX protocol mirror known interactions in E. coli's central carbon metabolism, for instance, such as those detailed in this publication: doi: 10.15252/msb.20199008?
5. Certain details of Figure 1 are challenging to grasp and inadequately explained in the figure legend. What do the colours of the heatmap in 1d represent?
6. In Figure 1h, the proteins co-eluting with 2-isopropylmalic acid in the four separations decrease to 5 from tens of proteins, with only one known interactor of 2-isopropylmalic acid (LeuA) among them. Is this outcome favourable or unfavourable? Are the other four proteins false positives?
7. Figure 2b: Why is this expected to work particularly well for NMPs? Is there a specific biological rationale behind it?

Furthermore, among the classified interactions, two were corroborated through microscale thermophoresis and protein docking. One involved the enzyme FabF and the dipeptide Val-Leu. The decision to delve deeper into this pair likely stemmed from the prior focus on dipeptide-protein interactions, extensively discussed in Luzarowski et al.'s 2019 manuscript. The second interaction pertained to lumicrome and PyrE.

Significance

To summarise, this paper advocates for heightening the complexity of experimental and computational analyses in studying metabolite-protein interactions through co-fractionation techniques. While it's anticipated that increased separation would enhance results, I remain unconvinced that the data presented conclusively demonstrates this. Overall, I believe the proposed method possesses only a modest level of originality and novelty, as outlined at the beginning of my review. Nonetheless, the substantial experimental effort and data generation warrant publication following additional meticulous quality control evaluation.

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

In this manuscript, the authors performed extensive genetic analyses in yeast on the functions of Med15 regions under multiple stress conditions, linking cell growth phenotypes, gene expression and protein-protein interaction. Med15 is an activator-contacting subunit of the Mediator complex and the functions of yeast Med15 have been extensively studied. In particular, the authors attempted to understand the roles of a poly-Q region (Q1), its length and composition in stress response phenotypes, Med15-mediated gene expression, and interaction with transcription factor Msn2. Results from this work are consistent with several previous studies and revealed some new insights. The authors concluded that robust Med15 activities required the Q1 tract and the length of Q1 tract modulates activity in a context-dependent manner. While the study is well executed and the conclusions are generally sound, several concerns listed below should be addressed and some clarifications should be made.

Major comments:

1. Abstract, "We also observed that distinct glutamine tracts and Med15 phosphorylation affected the activities of the KIX domain". Fig. 1 shows the effects of KIXQ2Q3 deletion and p7 phosphor-dead mutant under Acetic acid and Ketoconazole treatment, but does not demonstrate that these domains or phosphorylation affects KIX domain activities.
2. Is it known that Med15 is dephosphorylated under stress conditions other than osmotic challenge? Another explanation for D7P/D30P mutant results (Fig. 1B) might be that Med15 phosphorylation in unstressed cells is important for certain types of stress response (acetic acid and Keto). In contrast, the observation that D7P has no effects on osmotic stress (Fig. 1B) might suggest that phosphor-Med15 is dispensable for function. Some explanations on how to ascertain the roles of Med15 phosphorylation would be needed.
3. Fig. 4, what is the rationale of analyzing basal expression rather than activated expression of Gcn4 and Msn2 dependent genes? Gal4 and Hap5 dependent genes could be measured as well, in order to complete the gene expression-phenotype correlation that the authors strive to make in this paper.
4. Fig. 4, Error bars should be provided on gene expression analysis. Gcn4 and Msn2 target genes should be highlighted separately to facilitate comparisons.
5. Results from Fig. 4 and 5 indicate that Spacer-Q1 and 12PQ-Q1, being the strongest interactors to Msn2, actually reduced HSP12 expression (a known Msn2 target gene). Some explanations would be needed. Page 12 Discussion paragraph 2 "Q1 substitutions that interfered with coiled-coil propensity had no effect on TF activity" would need to be revised and to include some discussions on this result.
6. Fig. 5C, additional explanation is needed on how interaction rank is determined and how error bars are obtained.
7. The idea that Q1 provides a molecular hinge to facilitate intramolecular interactions is interesting, but sounds like a possible scenario without further evidence. Are there any published structural studies on Med15/Mediator complex that might support this idea?

Minor comments:

1. SC-HLUM and SC-HLMU is used interchangeably in the legend and text. Please keep consistent. Explanations for these acronyms are not found in the Methods.
2. Fig. 2, 3. AcOH should be Acetic acid, to be consistent with Fig. 1.
3. Fig. 3B, error bars should be provided for growth measurements.

Significance

This work provided a detailed analysis on the roles of a specific poly-glutamine region in yeast Med15 functions and regulation. One conceptual advance of this work is that the structural flexibility rather than the sequence itself of Q1 tract proves to be critical for Med15 function. The ability to correlate Med15-Msn2 interaction with gene expression analysis demonstrated some technical novelty, given the power of genetic manipulation in yeast.

Med15 is a key Mediator subunit contacting several sequence-specific transcription activators. Its interaction with a number of transcription activators in yeast such as Gcn4 and Gal4, was previously studied as referenced in this manuscript. This manuscript first provided a quite comprehensive genetic mutational analysis and confirmed several findings in previous studies. The identification of critical Med15 regions for acetic acid response (and Hap5-dependent gene activation) and the analysis of Msn2-Med15 interactions appear to be novel. Researchers interested in eukaryotic transcriptional regulation would benefit from reading this study.

Field of expertise of this reviewer: mechanisms of transcriptional regulation, genetics, nuclear organization and function

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

The Mediator complex, a multicomponent complex, regulates the interaction between transcription factors and RNA polymerase II using protein interactions. In particular, Intrinsically Disordered Regions (IDRs) with in Med15. Copper and Frassler made extensive mutations to the three IDRs that are characterized with high glutamate content (poly-Q), the KIX domain which interacts with transcription factors and the MAD which interacts with the rest of the Mediator complex. The three poly-Q repeats are adjacent to Activator Binding Domains (ABD). The impact of mutant Med15 was measured with growth assays, co-IPs with a transcription factor, and transcriptional activation of a reporter during different stresses. Med15 is particular important in stress responsive transcription rather than basal transcription because in yeast it is nonessential. It can be repress and activation translation of genes. Using a series of internal deletions and substitutions impact of the mutations was tested on by measuring growth of strains, expression of Med15 regulated genes, and interaction with Msn2, a transcription factor that regulates to various stresses. This work adds to the body of research confirming that multiple weak/ transient interaction domains regulate Med15 function.

Major comments:

1. Without knowing the protein levels of the different mutants, it is difficult to contribute deletions of different regions with phenotypes measured. Various internal deletions decrease Med15 protein levels (Jedidi et al. 2010) while other affect the integrity of the Mediator complex. This study did not measure mRNA or protein levels of their mutants. However, (Jedidi et al. 2010) used Myc-tagged Med15 which affects regulation of Med15 via SNF1 (Gallagher et al. 2020). In another study, Med15 was N-terminally tagged and protein levels of some deletions increased (Herbig et al. 2010). It's unknown if other tags such as HA, TAP or FLAG affect Med15 regulation via SNF1. This study used untagged Med15 expressed from the native promoter which avoids these complications. It's also unknown if the differences in Med15 deletions are from reduced transcription, translation, or protein stability. There are commercial antibody that may work (https://www.genetex.com/Product/Detail/Gal-11-S-cerevisiae-antibody/GTX64110). There are several commercial antibodies to human Med15 but the cross reactivity has not been tested.
2. Quantification with ImageJ on spot assays is difficult because once growth has maxed out on the dense spots there is no resolution. Using more dilute spots is challenging because colony size is affected by the nearby colonies. The error bars are the mutants are large. Can quantitative growth curves be carried out in flasks or an automatic plate reader for better quantification?

Minor comments:

1. Why were the stress conditions chosen in figure 1B? These are only a subset of conditions that the med15 deletion is sensitive to. Aside from acetic acid the phenotypic profile of each deletion is similar. The bigger the deletion, the more severe the growth defect. The keto plate appears under loaded by comparing the number of colonies in the third dilution on the keto plate and the fourth spot on the YPD plate in the BY4742 (MED15 wild-type strain). Does keto lyse cells? Or was there that much variation between mutants? Perhaps the dose of keto needs to be calibrated. In figure 2B it looks like wildtype growth in keto was 70% of untreated growth.
2. Figures 1C and 1D are not discussed in the results. The authors should remind what the hGR assay measures when discussing the results. How is it different from the GAL4 transcriptional reporter?
3. In the D to A mutants some appear to be required for acetic acid tolerance. What was the pH of the media?
4. The labels between Figure 1 and 2 are inconsistent. 90 mM acetic versus 80mM AcOH, YPGal versus YPGalactose, SD+LKHU versus SD+KLUH. The mutants on 0.97M NaCl at 37oC from figure 2A grew more than 0.9M NaCl at 38oC. Also, in the text it says 37 oC.
5. Is the MED15 strain BY4742? In Figure 1 was it also transformed the pRS315? How was the plasmid maintained on the plates, specifically YPD?
6. Genes such as GAL and URA3 should in italics.
7. In the split Ub assays, was wildtype Msn2 and Med15 also present?
8. There is inconsistently naming of media in Media and Phenotype Testing section. The media is called synthetic complete media is labeled SC-URA or SC-LEU and at times in Results its called SD+K or SC-HULM. Is SC with all amino acids and SD without any amino acids?
9. The plasmid names in the supplemental table don't match the ones labeled in the figures.
10. Why was ALG9 used for normalization of qRT PCR?
11. The background of the strains is confusing. There appears to be two different med15 knockouts OY320 and JF1368. Which ones were used in which experiments? Some of the trains have a trp1 auxotrophic which affects stress response on it's own (González et al. 2008; Schroeder and Ikui 2019).

Significance

• General assessment: provide a summary of the strengths and limitations of the study. What are the strongest and most important aspects? What aspects of the study should be improved or could be developed? Extensive mutational analysis of Med15.
• Advance: compare the study to the closest related results in the literature or highlight results reported for the first time to your knowledge; does the study extend the knowledge in the field and in which way? Describe the nature of the advance and the resulting insights (for example: conceptual, technical, clinical, mechanistic, functional,...). This work confirms numerous other studies on the contribution of various Med15 domains on function.
• Audience: describe the type of audience ("specialized", "broad", "basic research", "translational/clinical", etc...) that will be interested or influenced by this research; how will this research be used by others; will it be of interest beyond the specific field? Incorporation of human domain substitutions could influence how people outside the field would interpret how Med15 interacts with transcription factors.
• Please 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. Expert in using yeast genetics and natural genetic variation to address underlying mechanisms of stress response to environmental toxins with a particular focus on transcription factors and TORC1.

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

Major Comments

The authors show that the amino acid content and length of Q1 affects transcription activity in a media-dependent way in a construct that includes Q1-ABD1 and a tailing Q/N rich region (Q1R). Briefly, different media conditions used as proxies for specific target TF activities varied in their sensitivity to the Q1 sequence content. However, the reason for this variation between target TF activities is not addressed, so the observations seem more anecdotal than insightful. One test performed suggests some of the Q1 sequence dependence may be due to changes in AD-ABD interactions, but this interesting possibility is not investigated further.

A split-ubiquitin two-hybrid assay, meant to detect interactions between Msn2 TAD and Med15-Q1R, showed clear Q1 sequence/composition-dependence when changed from polyQ tracts. In particular, replacement with leucine-rich tracts (12L and RvHs) significantly reduced interactions (as inferred from growth requirements in Fig 5B). Q1 consisting of just 10 spacer residues, 0 to 24 Q residues, or PQ repeats all had quite similar results suggesting retention of some Msn2 and Med15 interactions. Replacement with a helix-forming sequence from NAB3 gave intermediate results. Again, no explanation was offered for the observation but it seems probable that the NAB3 Q1 system is no longer reporting on Msn2 Med15 interactions.

The manuscript presents extensive assays, but a lack of consistency in conditions and constructs tested makes comparing different assays difficult. In particular, it would be valuable to have NAB3-Q1, FrHs-Q1, and RvHS-Q1 tested under conditions of high salt as that is indicated to be the Msn2 target condition (e.g. an additional result that would be presented in Fig 3B); this would be valuable to compare to the two-hybrid results. The relationship between Q1 polyQ length and Msn2 TAD-Med15 ABD1 binding is not clear from this assay as all had similar growth on the plates. A possible explanation for the inferred reduction in TAD-ABD1 binding in the leucine rich Q1 constructs is that this highly hydrophobic linker itself binds to ABD1 and is therefore self-inhibitory. There is also the unexplored/not discussed possibility that NAB3, 12L, and RvHs have off-target interactions that disrupt the TAD-ABD1 interactions.

The framing of the study and the title of the manuscript strongly suggest that there might be a relationship between coiled-coil formation and transcription activity. This is the basis for selection of many of the Q1 sequences tested, with the premise of either increasing or disrupting coiled-coil structure. These 'propensities' are quantified in Supp Fig 1; however, a significant limitation of this interpretation is that these propensities are bulk properties that presume formation of homo-dimers or homo-trimers, a situation that is not shown to be relevant for Med15 at a promoter. This means that Q1 is potentially only one of the multiple partners required for coiled-coil formation. So even if a tested sequence has high coiled-coil propensity, that may not be the case in the actual biological systems at play here. Another consideration to be entertained is how different solvent conditions (different media) may affect coiled-coil propensity. An unanswered question is whether Q1 may form coiled-coil structure either with other regions of Med15 and/or with other Mediator subunits or even other co-factors entirely. This is a question implied by the title of this paper, but the data presented address neither intra- nor inter-molecular interactions of the polyQ regions (the two-hybrid study is designed to probe the ABD-AD interactions).

A final proposed hypothesis was that Q1 acts as a hinge in a way analogous to what was reported for the huntingtin protein (ref 7). This is an attractive model but remains untested in this work. In particular, the Med15-Q1R construct used does not have multiple ABDs that would potentially be brought in close contact, so the results here cannot be interpreted as analogous to the huntingtin hinge model. Minor Comments:

Please explain the choice of the 10-residue spacer instead of a 12-residue spacer.

Page 14: "We observed that Q1 substitutions with increased coiled-coil propensity (Supplementary Figure 1) diminished TF activity while Q1 substitutions that interfered with coiled-coil propensity had no effect on TF activity (Fig. 3, 4), suggesting that the flexibility of the sequence is an important feature." There was no demonstration that those sequences in this context form CCs. There's no evidence of what is actually being modulated whether it's length, flexibility, or ability to interact with other regions of Med15 or even with other co-factors.

Page 15: "We confirmed that Msn2-dependent activities of Med15 are encoded by the region containing the Q1 tract and ABD1 (aa 116-277) and found that the KIX domain alone could also mediate an interaction with Msn2 (Fig. 5). This contrasts with the Gcn4- or Gal4-dependent growth or stress responses which are the result of additive interactions with Med15 that are characterized by weak, highly dispersed, multivalent interfaces. While it is not yet entirely clear if the interaction with Msn2 is similarly multivalent, we have shown that either the KIX domain alone or the Q1R region alone of Med15 was sufficient with no evidence of additivity." These statements are unsupported. While Gcn4 and Gal4 transcription activity has been shown to depend on multiple AD-ABD interactions, none of the data reported here shows that Msn2 does not (as is stated here, which undermines the "contrasts" argument. Further, based on the assays presented in Figure 1B, Msn2, Gal4, and Gcn4 behave similarly for the various Med15 constructs.

Page 16: "In all instances TF activity was reduced in the absence of the Med15 Q1 tract." This seems false based on the data presented. Met10 activity appears to have increased in Figure 4A.

Page 16: Reference to Figure 2C and Figure 2B are mislabeled. Should be Figure 2D and 2C, respectively.

Page 17: "The fact that residues at Q1 were not functionally constrained to be glutamine residues suggests the Q1 tract is not an interaction motif participating directly in protein-protein interactions." This is completely unsupported. There are no data presented that address interactions between Q1 and anything else.

Figure 2: Not clear which assays were at 30{degree sign}C vs 22{degree sign}C as they are not labeled in the figure. In Figure 2A, the label med15 should be med15Δ.

Figure 4: Interpretation of these results seems limited by only reporting YPD media conditions. May be helpful to include the conditions reported in Figure 1.

Figure 6: It is not clear what some elements of this figure are meant to represent. Is saw tooth always polyQ? or Is ABD1 always blue and ABD2 is always red. What then are the loops? The general premise of this figure does not seem to be supported by the actual experiments performed.

Supp Figure 3: "K is the Med15 fragment encompassing the KIX domain, aa 1-277." This aa range is KQ in the main text. Either the residue range is wrong, or the label is wrong.

Significance

This manuscript addresses an interesting topic. There appears to be a disconnect between the stated motivation and what was actually done. The large array of assays and conditions are difficult to compare, leaving the reader with a feeling that the authors have catalogued a lot of possibilities but that no generalizable or unifying insights are at hand. The attempt to present a model (Figure 6) is difficult to parse and is not directly supported by the data presented. Addressing the issues raised here could result in a work that is useful to the specific field of Med15 structure and function but of limited use at the moment to a wider audience.

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)):

This manuscript uses C. elegans as a model to interrogate the effects of autism-associated variants of previously unknown function in the RNA-binding protein RBM-26/RBM27.

Despite its potential impact, there are several concerns related to the technical rigor and specificity of the observed effects.

Major concerns: 1. The effects on PLM are interesting, but why was this neuron selected for study? Was this a lucky guess or are other axons also affected? It is important to clarify whether the effects of RBM-26 are specific to this neuron or act pleiotropically across many or all neurons. According to CeNGEN, rbm-26 is strongly expressed in the well-characterized neurons ASE, PVD, and HSN. Are there morphological defects in these neurons, or others? As a note, there are also functional assays for these neurons (salt sensing, touch response, and egg laying, respectively).

We have added new data to the supplemental materials showing that loss of rbm-26 function also causes the beading phenotype in the axons and dendrites of the PVD neuron (Figure S4 and lines 196-199). We have focused on the PLM neuron because our preliminary studies indicated that it had a higher penetrance of axon defects relative to the PVD neuron. Moreover, we observed expression of endogenously tagged RBM-26 in the PLM neuron (Figure 3A-C and lines 210-215).

Similarly, the choice of the MALSU homolog seemed like a shot in the dark. It is ranked 46th (out of 63 genes) for fold-enrichment following RBM-26 pull-down, and 9th for p-value. Were any of the mRNAs with greater fold-enrichment or smaller p-values examined further? It is important to determine whether many or all of these interacting genes are overexpressed in the absence of RBM-26 and whether they are also required for the phenotypic effects of RBM-26 mutants, or if the MALSU homolog is special.

We have clarified our reasoning for selecting the MALS-1 ortholog of MALSU1 for further study (see lines 283-284 and Table S2). Amongst binding partners with human orthologs, MALS-1 was by far the top ranked candidate. The adjusted p-value for MALS-1 was 0.0008. The next smallest adjusted p-value was two orders of magnitude larger (0.028 for dpy-4). Moreover, the log2fold fold enrichment for MALS-1 was 1.98, about the same as the largest (ACADS with 2.13). Nonetheless, we agree that some of the other interactors may also be of interest and have thus included them in the supplemental table S2. Although these other potential binding partners are outside the scope of this study, we expect that future studies by ourselves or others may focus on the roles of these other binding partners.

In addition to the specificity controls mentioned above, positive and negative controls are needed throughout the results. While each of these may be relatively minor by itself, as a group they raise questions about the technical rigor of the study. Briefly these include: Fig 1C. Missing loading controls and negative control (rbm-26 null allele). Additional exposures should be included to show whether RBM-26(P80L) protein or the lower band for RBM-26(L13V) are present at all, relative to the null allele.

We have added no-stain loading controls to figure 1C. We have also switched to using ECL detection, which is much more sensitive and reveals faint bands for RBM-26(P80L) and additional faint bands for RBM-26(L13V). In addition, we have included a longer exposure for the blot (Figure S1). We are unable to test the null, as we can only produce a limited number of small maternally rescued progeny, thereby precluding western blot analysis.

Fig 2. Controls to distinguish overextension of PLM axon from posterior mispositioning of ALM cell body are needed. Quantification of PLM axon lengths in microns (or normalized to body size) with standard deviation, not error of proportion, should be shown. Measurement of “beading phenotype” should be more rigorous, see for example the approach in Rawson et al. Curr. Biol. 2017 https://doi.org/10.1016/j.cub.2014.02.025 . The developmental stage examined, and the reason for choosing that stage, should be described for this and all figures.

We have added new data that shows PLM axon length relative to body length for each of the RBM-26 mutants (Figure S2 and lines 183-185). These results indicate that the PLM axon has a larger axon length to body length ration, suggesting that the PLM/ALM overlap phenotype is a result of PLM axon overextension. For most experiments, we retain penetrance, as this has been standard practice in the field and allows for a much larger sample size (see examples listed below). We have also added examples of how the beading phenotype was measured (Figure S3). Moreover, we have now analyzed this phenotype and others at multiple developmental stages (Figures 2D-H and Table S1). In general, we have conducted experiments at the L3 stage because the rbm-26(null) mutants don’t survive past this stage. However, for many of our experiments we have also included additional stages as well. We have added this explanation to the methods section of phenotype analysis and also at various locations throughout the text. We have also labeled all graphs to clearly indicate the developmental stages and included.

10.1038/s41467-019-12804-3 Article by laboratory of Brock Grill

10.1371/journal.pgen.1002513 Article by laboratory of Ian Chin-Sang

doi.org/10.1073/pnas.1410263111 Article by laboratory of Chun-Liang Pan

10.1016/j.neuron.2007.07.009 Article by laboratory of Yishi Jin

doi.org/10.1523/JNEUROSCI.5536-07.2008 Article by laboratory of William Wadsworth

Fig 3. Controls without auxin and with neuronal TIR1 expression alone should be included. Controls demonstrating successful RBM-26 depletion, in larvae as well as in embryos at the time of PLM extension, should be included (weak embryonic depletion might explain why the overextension phenotype is only 14% instead of 40% as in the null). According to CeNGEN, rbm-26 expression in PLM is barely detected, thus depletion with a PLM-specific TIR1 should also be tested. To confirm the authors' identification of the cell marked "N" as the PLM cell body, co-expression of rbm-26 and a PLM-specific marker should be added. Rescue of the rbm-26 mutants with neuronal (and PLM-only) expression should be included to test sufficiency in PLM, and as a further control for potential artifacts of the AID system.

We have added new data showing that an endogenously tagged RBM-26::Scarlet protein is expressed in the PLM neuron (Figure 3A-C). Moreover, we have added rescue experiments, showing that a Pmec-7::rbm-26::scarlet transgene can rescue the beading phenotype and the PLM/ALM overlap phenotype (Figure 3 F-G). We have also added controls without auxin (Figure S7) __and without the rbm-26::scarlet::aid gene (Figure S8). We have added a new figure showing auxin-mediated depletion of RBM-26::Scarlet::AID in the PLM neuron (Figure S10)__. We examined auxin-mediated depletion at the L3 stage for consistency with our auxin-mediated phenotypic experiments. Moreover, these were done at the L3 stage for consistency with other experiments that included the rbm-26(null) mutants, which don’t survive past this stage.

In general, auxin-mediated knockdown tends to be hypomorphic in neurons. This is likely due to the fact that the neuronal TIR1 driver is expressed at much lower levels relative to the other drivers. In addition, the lower penetrance observed in auxin-mediated PLM/ALM overlap phenotype could reflect the fact that this phenotype resolves by the L4 stage in the hypomorphic mutants. For example, in P80L mutants at the L3 stage we see only about a 20% penetrance of the PLM/ALM overlap phenotype (relative to about 15% in auxin-mediated knockdown).

Fig 4. More rigorous quantification of the distribution of mitochondria along the axon should be included, not only total number, and it should be clarified what region of the axon the images are taken from. Including the AID-depletion strain with and without auxin would further add to the sense of rigor. For the mitoTimer experiments, why is RBM-26(L13V) not included and why do wild-type values differ ~5-fold between experiments (despite error bars being almost non-existent)? A more rigorous approach to standardizing imaging conditions may be needed. Positive controls using compounds that affect oxidation should be included. Measurements of individual mitochondria with standard deviations should be shown, rather than aggregate averages with error of proportion.

We have changed our methodology for measuring mitochondria, so that we now report the density of mitochondria in the axon (number per 100µm), (Figure 4E-F). We agree that this method is much better than counting the total number of mitochondria per axon, as it corrects for differences in body length and axon length). We also now include data for the whole axon (Figure 4E), proximal axon (Figure 4G), and distal axon (Figure 4H). These data suggest that the mitochondrial density defects occur in the proximal axon but not in the distal axon. Using the null allele, we have also examined the timing of mitochondria defects in the axon and report that the defects begin in the L1 stage and continue throughout larval development (Figure 4F). Individual datapoints have been added for all graphs in Figure 4.

For the mitoTimer experiments (Figure 5), we have added data for L13V and have added the individual datapoints to the graph. In the prior version, the values did not differ 5-fold between experiments with the same stage, rather the different graphs were from different stages (as noted in the figure legends/main text) and the L4 stage has much more oxidation than the L2 stage. To clear this up, we have added labels to the graphs to indicate the stages for each experiment. We have also added new data, so that we now show results for the L2, L3, and L4 stages for all three rbm-26 mutants (see Figure 5C-E). We didn’t test the L1 stage because the signal was not sufficient for accurate quantitation.

Fig 5. Additional positive and negative controls should be added, including additional rbm-26 alleles, the AID-tagged strain with and without auxin, and a rescued mutant.

The old Figure 5 has become Figure 6 in the new version. We have added the rbm-26(L13V) allele to each experiment, (Figure 6B-D). We have also added the loading controls for the western blot along with quantification for 3 biological replicates of the western blot analysis (Figure 6D). We agree that these additions significantly strengthen the data because they show that two independent alleles of rbm-26 cause very substantial increase in the expression of mals-1 at both the mRNA and protein levels. We did not do these experiments with the rescuing transgene or with the AID-tagged strain because these experiments are done on whole worm lysates, whereas the AID-tagged and rescuing transgene are neuron-specific.

Fig 6. Controls showing whether the Scarlet-tagged protein is functional are needed, to rule out dominant negative or toxicity-related effects.

This is Figure 7 in the new version. For this experiment, we are showing that overexpression of MALS-1 does cause defects. The idea is that excessive amounts of MALS-1 causes deleterious effects to the mitochondria. In fact, these defects could be considered as dominant negative or toxic. We considered the possibility of crossing the Pmec-7::mals-1::scarlet transgene with rbm-26; mals-1 double mutants. However, this does not seem workable, because the single copy Pmec-7::mals-1::scarlet transgene produces the phenotypes at penetrances that are similar to what we observe in rbm-26; mals-1 double mutants. We concede that the results of the overexpression experiments in Figure 7 are limited when considered in isolation. However, we think that they are meaningful when considered in combination with the results on the mals-1;rbm-26 double mutants in Figure 8.

Fig 8. Controls for other mitochondrial components need to be included. It is important to determine if the decrease in ribosomes is specific or reflects a general decrease in mitochondria. If there are fewer mitochondria as suggested in Fig. 4, then of course mitochondrial ribosomal protein levels are also reduced. Additional rbm-26 alleles should be included here as well. Is this effect dependent on the MALSU homolog?

This is Figure 8D-E in the new version. We have added new data showing that the decrease in MRPL-58 expression that is caused by the rbm-26(P80L) mutation is dependent on MALS-1. We concede that these experiments cannot be used to determine anything about the mitoribosomes per se, but rather serve as an alternative way of testing the effect of rbm-26 on mitochondria. We have revised the text accordingly (lines 355-357). Given these limitations we have elected not to try additional mitochondrial markers and have also not included additional rbm-26 alleles for this experiment.

Finally the authors should address concerns about image manipulation, which amplify the concerns about technical rigor outlined above. The image in Fig. 2A appears to have a black box placed over the lower-right portion of the field to hide some features. Black boxes also appear to have been placed over the tops of images in Fig. 4B and 4D and at the left of Fig. 6A, 6B, and 6C. While these manipulations probably do not affect the conclusions, they further undermine confidence in data integrity and experimental rigor.

We have corrected all of these image processing errors. The box in 2A was for the purpose of squaring off a corner that was clipped during image rotation. The boxes in Figures 4 and 6 (of the prior version) were added to give space for labels (without obscuring image features). We have now used alternative methods to accomplish the same goals. For example, in Figures 4-D we have placed the labels outside of the images.

Minor points. 1. C. elegans nomenclature conventions should be followed: - C. elegans gene names have three or four letters, thus the MALSU homolog cannot be named "malsu-1". Please have new gene names approved by WormBase BEFORE submitting for publication http://tazendra.caltech.edu/~azurebrd/cgi-bin/forms/gene_name.cgi

We have changed malsu-1 to mals-1. In addition, both mals-1 and mrpl-58 have now been approved by wormbase and will be listed on the website upon its next update.

• If two sequential CRISPR edits are made on the same gene then they should be listed as a compound allele, such as rbm-26(cue22cue25)

We have updated our gene names to reflect this convention.

• Genes on the same chromosome should not be separated with a semicolon, for example rbm-26(cue40) K12H4.2(syb6330)

We have updated our gene names to reflect this convention.

Describing the defects as "neurodevelopmental" is misleading in the case of axon beading or degeneration. Similarly, there is no evidence for an "axon targeting" defect as stated in the abstract.

We have revised such that instead of referring to degeneration phenotypes as neurodevelopmental, we now refer to axon degeneration phenotypes that occur during development. For example, in the abstract we now say, “These observations reveal a mechanism that regulates expression of a mitoribosomal assembly factor to protect against axon degeneration during neurodevelopment.

Regarding targeting defects, this was meant to refer to the misplacement of the PLM axon tip (which contains electrical synapses). However, our subsequent analysis has revealed that these defects are transient in P80L and L13V mutants, as they resolve by the L4 stage. The rbm-26 null axon development defects do not resolve, though these mutant die prior to the L4 stage. Given these findings, we have decided not to use the term of targeting defects. Instead, we now refer to this as an axon tiling defect or PLM/ALM overlap phenotype.

In Fig. 5A, the symbol that appears to correspond to F59C6.15 (lowest p-value) is a different size than the others and is colored as ncRNA, whereas WormBase annotates this gene as snoRNA.

This error has been corrected.

In the Introduction, the last sentences of the first two paragraphs should be varied ("However, little is known about the [...] mechanisms that protect [...] during neurodevelopment.")

This has been done.

Why is RBM-26 protein running as a doublet at both sizes?

We have improved our western blotting methodology by using 12% gel, allowing for better resolution. We have also switched from colorimetric detection to ECL detection, allowing for greater sensitivity. In our new blots, we identify 6 different RBM-26 protein bands. We don’t know the reason for these bands, but speculate that they are the result of post-translational processing (148-150).

When showing the RBM-26 expression pattern (Fig. 3) please include a lower-magnification image of the entire animal.

This has been done (Figure S6)

It is confusing to refer to the RNA IP experiments as an "unbiased screen", which in C. elegans typically refers to a genetic screen.

We now refer to this as a “biochemical screen”.

The relationship between axon overextension, beading, and mitochondrial localization is not clear. What causal connection between these is being proposed? The causal connections between these phenotypes, if any, should be clarified experimentally. For example, if the axon extension defects develop before mitochondrial localization defects, then it is unlikely that mitochondrial defects cause axon overextension.

We have added new data showing that the reduction in mitochondrial density within the axon begins during the L1 stage and increases throughout larval development (Figure 4F). We have also added additional data showing that the increase in mitochondrial oxidation is weak in the L2 stage and surges in the L3 stage (Figure 5C-E), coincident with the beginning of the axon degeneration phenotypes. We propose (lines 383-391) that a low level of mitochondrial defects is present in L1 larvae, giving rise to the axon tiling defects. In the L3 stage there is a surge in excessive mitochondrial oxidation, giving rise to the axon degeneration phenotypes. We have added a new section to the discussion that addresses the relationship between defects in axon development and axon degeneration (lines 375-405).

Please explain how to interpret the difference in axon beading in the two deletion alleles of the MALSU homolog (axon beading defects in tm12122 but not in syb6330). Is syb6330 not a null allele? Or are the defects in tm12122 due to other mutations in this strain background?

One likely reason for this difference is that tm12122 is predicted to cause a partial deletion of the mals-1 coding sequence, whereas the syb6330 is a full deletion. Thus, the tm12122 could be acting as a dominant negative. In fact, prior work on the MALSU1 ortholog has indicated that this protein is subject to interference by a dominant negative construct (see Rorbach et al, Nucleic Acids Res 2012). Nonetheless, we cannot rule out the possibility of a linked second mutation in tm12122. However, since we have found similar phenotypes and genetic interactions with both alleles, we can conclude that these phenotypes and interactions are due to loss of MALS-1, rather than a second mutation.

Are mitochondria reduced in number or mislocalized? If they are reduced in number, is this due to altered balance of fission/fusion?

We have adjusted our methods for quantifying mitochondria and have also analyzed the proximal vs distal axon (Figure 4). We find that the density of mitochondria is decreased in the proximal axon, but not in the distal axon. We speculate that this might reflect a higher demand on mitochondria in the proximal axon, due to a higher amount of trafficking activity in the proximal axon (lines 255-257). We propose that the loss of RBM-26 causes dysfunction in mitochondria. Since fission and fusion are mechanisms that can help to repair damaged mitochondria, it is likely that they would be involved in the phenotypes that we observe.

In Fig. 3A-D, please keep the labels in the same position in all panels and do not alter brightness settings between single-color and merged panels.

These images have been moved to the supplemental data section (Figure S5). We have adjusted the labels as suggested. We have not changed the brightness settings, as they were already the same in all panels. However, the blue signal in the merged panel does obscure some of the red signal, giving an appearance of an alteration in color balance.

The claim that rbm-26 acts cell-autonomously requires PLM-specific depletion and rescue experiments.

We have added new data indicating that a Pmec-7::rbm-26::scarlet transgene can rescue the beading phenotype (Figure 3F-G).

**Referees cross-commenting** I appreciate the use of the consultation session to resolve differences between reviewers, but in this case I fully agree with the content and tone of all the comments from the other reviewer -- I think our remarks are very well aligned!

Reviewer #1 (Significance (Required)):

The study engineers autism-associated variants in conserved residues of RBM27 into the C. elegans homolog RBM-26 and identifies neuronal phenotypes potentially relevant to autism and a potential molecular mechanism involving regulation of mitochondrial ribosome assembly.

The key claims of the study are 1} that autism-associated variants in RBM-26 decrease its protein expression; 2} that impaired RBM-26 function leads to a variety of defects in development and maintenance of a single neuron called PLM, including altered axonal localization of mitochondria; 3} that RBM-26 normally binds the mRNA for the C. elegans homolog of MALSU, a mitochondrial ribosomal assembly factor; 4} that loss of RBM-26 leads to overexpression of the MALSU homolog; and 5} that MALSU is required for some of the deleterious effects on the PLM neuron seen in RBM-26 mutants.

This study will be of interest to the autism research community because it bolsters the idea that variants in RBM27 are likely to disrupt gene function and to affect neuronal health. It will also be of interest to the broader cell biology community because it suggests an interesting potential nucleus-to-mitochondria signaling mechanism, in which a nuclear RNA-binding protein might regulate assembly of mitochondrial ribosomes.

My field of expertise is developmental biology in C. elegans.

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

Summary In this manuscript, the authors studied an ASD-associated gene, rbm-26 in neuronal morphology using the touch receptor neuron PLM in C. elegans, and found that loss-of-function rbp-27 causes overextension and the formation of bulb-like structures in the axon. Using UV-crosslinking RNA immunoprecipitation and RNA-Seq, they identify malsu-1 as a target of rbm-26. Genetic analyses suggest malsu-1 likely functions downstream of rbm-26 in controlling the PLM morphology. Major comments:

• The authors describe RBM27 is associated with ASD and ID while they only cite SFARI paper that describes a weak association of RBM27 to ASD. The appropriate referenced that show link between RBM27 and ID should be provided. The link with ID was an error. We had meant to say “ASD or other neurodevelopmental disorders.” This has been corrected.

• SFARI database only has three (P79L, R190Q, G348D) mutations listed as ASD-associated. Where are other mutations L13V and R455H, particularly L13V that the authors used to generate the C. elegans mutant come from? Are they associated with intellectual disabilities? The others came from the devovo-DB. We have added a reference for this database and have also added the primary source references for each of the five de novo variants (see line 121).

• The authors should be very careful when describing 'gene X causes Y diseases'. Many (if not all) of the examples described in this manuscript are disease-associated genes without validation to be causal genes. We have revised accordingly. For example on lines 433-435, we now say,” For example, mutations in the EXOSC3, EXOSC8 and EXOSC9 are thought to cause syndromes that include defects in brain development such as hypoplasia of the cerebellum and the corpus callosum”. We have decided to use the phrase “thought to cause” because three of the five referenced articles on these genes use titles that indicate causation.

• The authors refer PLM axon beading and overextension phenotypes to 'axon degeneration and targeting defects'. The authors must provide additional evidence of axon degeneration (see below). Also the term 'targeting defects' is misleading as the authors did not examine if overextension of the PLM axon causes targeting defects. At least they should examine some synaptic markers. To provide more evidence of degeneration we have analyzed several additional phenotypes at multiple developmental stages (Figure 2 and Table S1). Regarding targeting defects, this was meant to refer to the misplacement of the PLM axon tip (which contains electrical synapses). However, our subsequent analysis has revealed that these defects are transient in P80L and L13V mutants, as they resolve by the L4 stage. The rbm-26 null axon development defects do not resolve, though these mutant die prior to the L4 stage. Given these findings, we have decided not to use the term of targeting defects. Instead, we now refer to this as an axon tiling defect or PLM/ALM overlap phenotype.

• Neuronal phenotypes (axon overextension and beading) should be examined at different developmental timepoints (larval, young adult, and aged animals) to test if these phenotypes are indeed degenerative instead of developmental defects. We have included new data to observe all of these phenotypes at multiple developmental time points (Figure 2 and Table S1).

• The authors use the blebbing (beading) phenotype in the axon as the sole evidence of neurodegenerative properties of the PLM neuron. A more thorough analysis of this phenotype as done by others (Pan PNAS 2006) must be provided to support the authors' claim that this phenotype represents neurodegeneration. We have included new data on multiple degenerative phenotypes in axons including: blebbing, beading, waviness and breaks (Table S1).

• The number of beads per axon should be quantified to better represent the severity of rbm-26 mutant. Individual samples should be plotted in the quantification instead of showing the percentage of animals. We have added data on the density of beads in rbm-26(null), rbm-26(P80L), and rbm-26(L13V) mutants (Figure S3). For most experiments we have decided to use penetrance to measure axon degeneration because this is a standard in the field and allows for a larger sample size. For examples please see:

10.1523/JNEUROSCI.1494-11.2012 (Toth et al, 2012)

https://doi.org/10.1016/j.cub.2014.02.025 (Rawson et al, 2014)

10.1073/pnas.1011711108 (Pan et al, 2012)

https://doi.org/10.7554/eLife.80856 (Czech et al, 2023)

https://doi.org/10.1016/j.celrep.2016.01.050 (Nichols et al, 2016)

• Based on the single gel image in Fig. 1C with no loading control, the P80L mutant appears to have no protein expression. How is the P80L viable while the null mutant is lethal? The authors should quantify the protein expression levels from multiple blots with proper loading controls. If P80L mutation is introduced into RBM-26::mScarlet strain can it cause depletion of the signal in vivo? We have added new data showing that the RBM-26::Scarlet signal is diminished by the P80L mutation in vivo (Figure 1E-F). We have also added quantification from 3 biological replicate blots (Figure 1D). Finally, we have improved the sensitivity of our blots by using ECL detection and also show various exposures to highlight the fainter bands (Figures 1C and S1). Therefore, we are now able to detect low level expression of RBM-26(P80L) mutant protein. It is likely that the low level of RBM-26(P80L) and RBM-26(L13V) seen on western blots is sufficient to prevent the lethal phenotype.

• 'Moreover, loss of either the SPTBN1 or ADD1 genes causes a neurodevelopmental syndrome that includes autism and ADHD' References are missing, and as described above, be extra careful when indicating causality. Very few genes are known to cause ASD and ADHD. We have added the citations for this work (line 81). We also note that the titles for both of the cited articles indicate causation. To be on the safe side we have revised this line to say, “Moreover, loss of either the SPTBN1 or ADD1 genes are thought to cause a neurodevelopmental syndrome that includes autism and ADHD”

• Fig. 3E F, the authors should use the strains that express TIR1 specifically in the touch receptor neurons to argue cell autonomous function of RBM-26. Alternatively, the authors may conduct PLM neuron-specific rescue experiments to test the sufficiency. We have added new data indicating that a Pmec-7::rbm-26::scarlet transgene can rescue the beading phenotype and the PLM/ALM overlap phenotype (see Figure 3F-G).

• 'Loss of RBM-26 causes mitochondria dysfunction in axons.' The authors did not examine mitochondria function in axons. They only examined the number of mitochondria, and ROS production in the soma. The authors should provide additional evidence to support the idea that elevated ROS production in the soma is due to mitochondrial dysfunction in axons. Also, the authors should use both P80L and L13V for this experiment, and indicate individual datapoint as dots. Here, they quantified at the L4 stage, which the authors should justify. We have added the L13V data to this experiment and now show the individual data points. In addition, we have now conducted this analysis at the L2, L3 and L4 stages (Figure 5C-E). We have also revised the text to indicate that loss of rbm-26 function causes mitochondrial dysfunction in the cell body which could potentially cause a reduction of mitochondria in the axon (see lines 100-101 and 268-270). We speculate that mitochondria in the axon are also dysfunctional. However, the mitoTimer signal is not bright enough in axons to allow for quantification.

• Figure 5B and C: the authors should also use L13V to quantify malsu-1 mRNA and protein level, and include quantifications in panel C (from multiple blots). This is Figure 6 in the new version. We have added new data for expression of mals-1 mRNA and protein in rbm-26(L13V) mutants (Figure 6B-D). We have also included quantifications from 3 biological replicates (Figure 6D).

• In the rbm-26 mutant, the number of mitochondria is reduced, while the amount of MALSU-1 protein is increased. If MALSU-1 is specifically localized at mitochondria in wild type, where does the excessive MALSU-1 go in the rbm-26 mutants? Quantification of MALSU-1 signal intensity should be provided. Our Pmec-7::mals-1::scarlet transgene uses the tbb-2 3’UTR and causes an overexpression phenotype. To address the question posed by the reviewer, we would need to express MALS-1 at endogenous levels. Given that endogenous levels of MALS-1 are very low, it is unlikely that we would be able to visualize its expression. Nonetheless, as a way to address this question we have attempted to create a single copy Pmec-7::mals-1::scarlet transgene that utilizes the mals-1 endogenous 3’UTR. We have tried multiple approaches for generating this construct, but all have failed, likely due to sequence complexities within the mals-1 3’UTR. While we cannot say where the extra MALS-1 protein goes, we think that it is likely overloaded into the remaining mitochondria and could also be in the cytosol as well.

• Figure 7C: malsu-1 knockout mutants exhibit PLM overextension phenotype, which is not consistent with their model. The authors should discuss this in detail. We have added a paragraph to the discussion explaining that mitochondria function could be disrupted by either MALS-1 overexpression or by MALS-1 loss of function (lines 471-480).

• 'To validate these findings, we also repeated these experiments with an independent allele of malsu-1, malsu-1(tm12122) and found similar results (Fig. 7A-C).' The malsu-1(tm12122) exhibits beading phenotype and more severe overextension phenotype which the authors must describe and discuss more carefully. One likely reason for this difference is that tm12122 is predicted to cause a partial deletion of the mals-1 coding sequence, whereas the syb6330 is a full deletion. Thus, the tm12122 could be acting as a dominant negative. In fact, prior work on the MALSU1 ortholog has indicated that this protein is subject to interference by a dominant negative construct (see Rorbach et al, Nucleic Acids Res 2012). Nonetheless, we cannot rule out the possibility of a linked second mutation in tm12122. However, since we have found similar phenotypes and genetic interactions with both alleles, we can conclude that these phenotypes and interactions are due to loss of MALS-1, rather than a second mutation (albeit at a slightly different penetrance). We have added these considerations to the results section (lines 342-244).

• Figure 8: The authors should include data from L13V, malsu-1 and rbm-26; malsu-1 mutants. Quantification from multiple blots should be provided. This is Figure 8D in the new version. We have added the malsu-1 and rbm-26;malsu-1 double mutants to this experiment. We have also added quantification from multiple biological replicate blots. As pointed out by the other reviewer, we think that this experiment does not give specific information about mitoribosomes, but is an alternative approach to looking at the reduction in mitochondria. Given this limitation and considering that we have added L13V data to the mitochondria experiment in Figure 8B, we have elected not to add additional data on L13V to the western blot experiment in Figure 8D

Minor comments: • 'Consistent with a role for mitochondria in neurodevelopmental disorders, some of these disorders include a neurodegenerative phenotype.' Why is it consistent to have neurodegenerative phenotypes if mitochondria is associated with neurodevelopmental disorders? A better explanation would help.

We have changed this sentence to, “Some neurodevelopmental syndromes feature neurodegenerative phenotypes that occur during neuronal development.”

• L13V is generally more severe in axon overextension phenotype than P80L while protein level is more abundant. The authors should discuss about this. We have also added a time course for the PLM/ALM overlap phenotype mutants (Figure 2D). This new data shows that the PLM/ALM overlap is quite similar overall between the P80L and L13V mutants. Both of these mutations cause an increase in PLM/ALM overlap in early larval development that is resolved by the L4 stage. The P80L phenotype resolves slightly sooner for reasons that are unknown. This could reflect differences in expression within the PLM that are not reflected in the whole worm lysate. This could also be due to a slight difference in the genetic background or other stochastic factors. The key point is that these two independent alleles cause similar phenotype overall, indicating that this phenotype is the result of loss in RBM-26 function.

• Fig. 2E, F: 'Beading refers to focal enlargement or bubble-like lesions which were at least twice the diameter of the axon in size.' How are the diameters of axons measured? A more detailed quantification method, and examples of measurement should be provided. We have added example measurements to the supplemental section (Figure S3). Additional detail on the measurements are in the Methods section (lines 517-518).

• Figure 3: The authors should also include low-magnification images to show where RBM-26 is expressed. The current image does now allow identifying cells. The transgene that labels the nuclei of hypodermis should be indicated in the manuscript. Specifically, the expression of the RBM-26 in the PLM should be shown. We have added a low magnification image (Figure S6) and have also added images of endogenously tagged RBM-26:Scarlet in the PLM (Figure 3A-C). The transgenic label for the hypodermis has been added to the legend of Figure S5.

• Figure 3: 'Tissue specific degradation of RBM-26::SCARLET::AID was achieved due to cell-type specific TIR-1 driver lines (see methods for details).' This information is not provided in the method section. This information has been added to methods section, ”Auxin proteindegredation”

• Fig. 4 E. Values from individual samples should be indicated as dots. Representative images of P80L and L13V should be included. Conduct quantifications at adult stage as the authors use in other quantifications, or justify use of specific developmental stage (L3) they used. Figure 4 has become Figures 4 and 5 in the revised version. We have updated the graphs to include dots for individual data points. We have added quantifications of the mitoTImer experiments for the L2, L3 and L4 stages (Figure 5C-E). We note that our other experiments were done at the L1, L2, L3 and L4 and adult stages. The mitoTimer signal is not sufficient at the L1 stage for quantification. At the adult stage, the red signal becomes saturated. We have added representative images for mitoTimer in P80L and L13V mutants (Figure S9).

• The genes malsu-1 and mrpl-58 are not listed on wormbase. If the authors would like to designate names to these gene, they should clearly indicate that along with the sequence name. We have changed malsu-1 to mals-1. In addition, both mals-1 and mrpl-58 have now been approved by wormbase and will be listed on the website upon its next update.

• The authors found that MRPL-58 amount is reduced in rbm-26 mutants (which require additional verifications). This can be explained by the fact that axonal mitochondria number is reduced in the rbm-26 mutants. How did the authors confirm that the reduction in MRPL-58 level is due to the disruption of mitoribosome assembly? This is Figure 8D-E in the new version. We have added new data showing that the decrease in MRPL-58 expression that is caused by the rbm-26(P80L) mutation is dependent on MALS-1. We concede that these experiments cannot be used to determine anything about the mitoribosomes per se, but rather serve as an alternative way of testing the effect of rbm-26 on mitochondria. We have revised the text accordingly (lines 355-357).

• 'MALSU-1 is a mitoribosomal assembly factor that functions as part of the MALSU1:LOR8F8:mtACP anti-association module [37-39].' I don't think these are known for C. elegans MALSU-1. We have revised to, “MALS-1 is an ortholog of the MALSU1 mitoribosomal assembly factor that functions as part of the MALSU1:LOR8F8:mtACP anti-association module”

• 'Moreover, our results also suggest that disruption of this process can give rise to neurodevelopmental disorders.' I feel this is a quite a bit of stretch.

This has been replaced with, “Therefore, we speculate that human RBM26/27 could function with the RNA exosome complex to protect against neurodevelopmental defects and axon degeneration in infants.” (lines 371-373)

**Referees cross-commenting** Yes, many of our comments overlap, and I fully agree with all comments from the other reviewer too. Reviewer #2 (Significance (Required)):

I found the manuscript interesting particularly the use of innovative techniques in identifying the target of RBM-26, The genetic analyses of rbm-26 and malsu-1 generally support the authors main conclusions that rbm-26 inhibits malsu-1 and be of potential interest to basic neuroscientists and cell biologists. However, the current manuscript looked premature which made my reading experience less pleasant. The phenotypic analyses is superficial compared to works similar to this work, which are insufficient to support the authors' claim of 'axon degeneration and targeting defects'. A number of issues listed above should be addressed before this manuscript is published. The reviewer's expertise: neurodevelopment in model 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

Summary

In this manuscript, the authors studied an ASD-associated gene, rbm-26 in neuronal morphology using the touch receptor neuron PLM in C. elegans, and found that loss-of-function rbp-27 causes overextension and the formation of bulb-like structures in the axon. Using UV-crosslinking RNA immunoprecipitation and RNA-Seq, they identify malsu-1 as a target of rbm-26. Genetic analyses suggest malsu-1 likely functions downstream of rbm-26 in controlling the PLM morphology.

Major comments:

• The authors describe RBM27 is associated with ASD and ID while they only cite SFARI paper that describes a weak association of RBM27 to ASD. The appropriate referenced that show link between RBM27 and ID should be provided.
• SFARI database only has three (P79L, R190Q, G348D) mutations listed as ASD-associated. Where are other mutations L13V and R455H, particularly L13V that the authors used to generate the C. elegans mutant come from? Are they associated with intellectual disabilities?
• The authors should be very careful when describing 'gene X causes Y diseases'. Many (if not all) of the examples described in this manuscript are disease-associated genes without validation to be causal genes.
• The authors refer PLM axon beading and overextension phenotypes to 'axon degeneration and targeting defects'. The authors must provide additional evidence of axon degeneration (see below). Also the term 'targeting defects' is misleading as the authors did not examine if overextension of the PLM axon causes targeting defects. At least they should examine some synaptic markers.
• Neuronal phenotypes (axon overextension and beading) should be examined at different developmental timepoints (larval, young adult, and aged animals) to test if these phenotypes are indeed degenerative instead of developmental defects.
• The authors use the blebbing (beading) phenotype in the axon as the sole evidence of neurodegenerative properties of the PLM neuron. A more thorough analysis of this phenotype as done by others (Pan PNAS 2006) must be provided to support the authors' claim that this phenotype represents neurodegeneration.
• The number of beads per axon should be quantified to better represent the severity of rbm-26 mutant. Individual samples should be plotted in the quantification instead of showing the percentage of animals.
• Based on the single gel image in Fig. 1C with no loading control, the P80L mutant appears to have no protein expression. How is the P80L viable while the null mutant is lethal? The authors should quantify the protein expression levels from multiple blots with proper loading controls. If P80L mutation is introduced into RBM-26::mScarlet strain can it cause depletion of the signal in vivo?
• 'Moreover, loss of either the SPTBN1 or ADD1 genes causes a neurodevelopmental syndrome that includes autism and ADHD' References are missing, and as described above, be extra careful when indicating causality. Very few genes are known to cause ASD and ADHD.
• Fig. 3E F, the authors should use the strains that express TIR1 specifically in the touch receptor neurons to argue cell autonomous function of RBM-26. Alternatively, the authors may conduct PLM neuron-specific rescue experiments to test the sufficiency.
• 'Loss of RBM-26 causes mitochondria dysfunction in axons.' The authors did not examine mitochondria function in axons. They only examined the number of mitochondria, and ROS production in the soma. The authors should provide additional evidence to support the idea that elevated ROS production in the soma is due to mitochondrial dysfunction in axons. Also, the authors should use both P80L and L13V for this experiment, and indicate individual datapoint as dots. Here, they quantified at the L4 stage, which the authors should justify.
• Figure 5B and C: the authors should also use L13V to quantify malsu-1 mRNA and protein level, and include quantifications in panel C (from multiple blots).
• In the rbm-26 mutant, the number of mitochondria is reduced, while the amount of MALSU-1 protein is increased. If MALSU-1 is specifically localized at mitochondria in wild type, where does the excessive MALSU-1 go in the rbm-26 mutants? Quantification of MALSU-1 signal intensity should be provided.
• Figure 7C: malsu-1 knockout mutants exhibit PLM overextension phenotype, which is not consistent with their model. The authors should discuss this in detail.
• 'To validate these findings, we also repeated these experiments with an independent allele of malsu-1, malsu-1(tm12122) and found similar results (Fig. 7A-C).' The malsu-1(tm12122) exhibits beading phenotype and more severe overextension phenotype which the authors must describe and discuss more carefully.
• Figure 8: The authors should include data from L13V, malsu-1 and rbm-26; malsu-1 mutants. Quantification from multiple blots should be provided.

Minor comments:

• 'Consistent with a role for mitochondria in neurodevelopmental disorders, some of these disorders include a neurodegenerative phenotype.' Why is it consistent to have neurodegenerative phenotypes if mitochondria is associated with neurodevelopmental disorders? A better explanation would help.
• L13V is generally more severe in axon overextension phenotype than P80L while protein level is more abundant. The authors should discuss about this.
• Fig. 2E, F: 'Beading refers to focal enlargement or bubble-like lesions which were at least twice the diameter of the axon in size.' How are the diameters of axons measured? A more detailed quantification method, and examples of measurement should be provided.
• Figure 3: The authors should also include low-magnification images to show where RBM-26 is expressed. The current image does now allow identifying cells. The transgene that labels the nuclei of hypodermis should be indicated in the manuscript. Specifically, the expression of the RBM-26 in the PLM should be shown.
• Figure 3: 'Tissue specific degradation of RBM-26::SCARLET::AID was achieved due to cell-type specific TIR-1 driver lines (see methods for details).' This information is not provided in the method section.
• Fig. 4 E. Values from individual samples should be indicated as dots. Representative images of P80L and L13V should be included. Conduct quantifications at adult stage as the authors use in other quantifications, or justify use of specific developmental stage (L3) they used.
• The genes malsu-1 and mrpl-58 are not listed on wormbase. If the authors would like to designate names to these gene, they should clearly indicate that along with the sequence name.
• The authors found that MRPL-58 amount is reduced in rbm-26 mutants (which require additional verifications). This can be explained by the fact that axonal mitochondria number is reduced in the rbm-26 mutants. How did the authors confirm that the reduction in MRPL-58 level is due to the disruption of mitoribosome assembly?
• 'MALSU-1 is a mitoribosomal assembly factor that functions as part of the MALSU1:LOR8F8:mtACP anti-association module [37-39].' I don't think these are known for C. elegans MALSU-1.
• 'Moreover, our results also suggest that disruption of this process can give rise to neurodevelopmental disorders.' I feel this is a quite a bit of stretch.

Referees cross-commenting Yes, many of our comments overlap, and I fully agree with all comments from the other reviewer too.

Significance

I found the manuscript interesting particularly the use of innovative techniques in identifying the target of RBM-26, The genetic analyses of rbm-26 and malsu-1 generally support the authors main conclusions that rbm-26 inhibits malsu-1 and be of potential interest to basic neuroscientists and cell biologists. However, the current manuscript looked premature which made my reading experience less pleasant. The phenotypic analyses is superficial compared to works similar to this work, which are insufficient to support the authors' claim of 'axon degeneration and targeting defects'. A number of issues listed above should be addressed before this manuscript is published.

The reviewer's expertise: neurodevelopment in model organisms.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

This manuscript uses C. elegans as a model to interrogate the effects of autism-associated variants of previously unknown function in the RNA-binding protein RBM-26/RBM27.

Despite its potential impact, there are several concerns related to the technical rigor and specificity of the observed effects.

Major concerns:

1. The effects on PLM are interesting, but why was this neuron selected for study? Was this a lucky guess or are other axons also affected? It is important to clarify whether the effects of RBM-26 are specific to this neuron or act pleiotropically across many or all neurons. According to CeNGEN, rbm-26 is strongly expressed in the well-characterized neurons ASE, PVD, and HSN. Are there morphological defects in these neurons, or others? As a note, there are also functional assays for these neurons (salt sensing, touch response, and egg laying, respectively).
2. Similarly, the choice of the MALSU homolog seemed like a shot in the dark. It is ranked 46th (out of 63 genes) for fold-enrichment following RBM-26 pull-down, and 9th for p-value. Were any of the mRNAs with greater fold-enrichment or smaller p-values examined further? It is important to determine whether many or all of these interacting genes are overexpressed in the absence of RBM-26 and whether they are also required for the phenotypic effects of RBM-26 mutants, or if the MALSU homolog is special.
3. In addition to the specificity controls mentioned above, positive and negative controls are needed throughout the results. While each of these may be relatively minor by itself, as a group they raise questions about the technical rigor of the study. Briefly these include:

Fig 1C. Missing loading controls and negative control (rbm-26 null allele). Additional exposures should be included to show whether RBM-26(P80L) protein or the lower band for RBM-26(L13V) are present at all, relative to the null allele.

Fig 2. Controls to distinguish overextension of PLM axon from posterior mispositioning of ALM cell body are needed. Quantification of PLM axon lengths in microns (or normalized to body size) with standard deviation, not error of proportion, should be shown. Measurement of "beading phenotype" should be more rigorous, see for example the approach in Rawson et al. Curr. Biol. 2017 https://doi.org/10.1016/j.cub.2014.02.025 . The developmental stage examined, and the reason for choosing that stage, should be described for this and all figures.

Fig 3. Controls without auxin and with neuronal TIR1 expression alone should be included. Controls demonstrating successful RBM-26 depletion, in larvae as well as in embryos at the time of PLM extension, should be included (weak embryonic depletion might explain why the overextension phenotype is only 14% instead of 40% as in the null). According to CeNGEN, rbm-26 expression in PLM is barely detected, thus depletion with a PLM-specific TIR1 should also be tested. To confirm the authors' identification of the cell marked "N" as the PLM cell body, co-expression of rbm-26 and a PLM-specific marker should be added. Rescue of the rbm-26 mutants with neuronal (and PLM-only) expression should be included to test sufficiency in PLM, and as a further control for potential artifacts of the AID system.

Fig 4. More rigorous quantification of the distribution of mitochondria along the axon should be included, not only total number, and it should be clarified what region of the axon the images are taken from. Including the AID-depletion strain with and without auxin would further add to the sense of rigor. For the mitoTimer experiments, why is RBM-26(L13V) not included and why do wild-type values differ ~5-fold between experiments (despite error bars being almost non-existent)? A more rigorous approach to standardizing imaging conditions may be needed. Positive controls using compounds that affect oxidation should be included. Measurements of individual mitochondria with standard deviations should be shown, rather than aggregate averages with error of proportion.

Fig 5. Additional positive and negative controls should be added, including additional rbm-26 alleles, the AID-tagged strain with and without auxin, and a rescued mutant.

Fig 6. Controls showing whether the Scarlet-tagged protein is functional are needed, to rule out dominant negative or toxicity-related effects.

Fig 8. Controls for other mitochondrial components need to be included. It is important to determine if the decrease in ribosomes is specific or reflects a general decrease in mitochondria. If there are fewer mitochondria as suggested in Fig. 4, then of course mitochondrial ribosomal protein levels are also reduced. Additional rbm-26 alleles should be included here as well. Is this effect dependent on the MALSU homolog? 4. Finally the authors should address concerns about image manipulation, which amplify the concerns about technical rigor outlined above. The image in Fig. 2A appears to have a black box placed over the lower-right portion of the field to hide some features. Black boxes also appear to have been placed over the tops of images in Fig. 4B and 4D and at the left of Fig. 6A, 6B, and 6C. While these manipulations probably do not affect the conclusions, they further undermine confidence in data integrity and experimental rigor.

Minor points.

1. C. elegans nomenclature conventions should be followed:
   • C. elegans gene names have three or four letters, thus the MALSU homolog cannot be named "malsu-1". Please have new gene names approved by WormBase BEFORE submitting for publication http://tazendra.caltech.edu/~azurebrd/cgi-bin/forms/gene_name.cgi
   • If two sequential CRISPR edits are made on the same gene then they should be listed as a compound allele, such as rbm-26(cue22cue25)
   • Genes on the same chromosome should not be separated with a semicolon, for example rbm-26(cue40) K12H4.2(syb6330)
2. Describing the defects as "neurodevelopmental" is misleading in the case of axon beading or degeneration. Similarly, there is no evidence for an "axon targeting" defect as stated in the abstract.
3. In Fig. 5A, the symbol that appears to correspond to F59C6.15 (lowest p-value) is a different size than the others and is colored as ncRNA, whereas WormBase annotates this gene as snoRNA.
4. In the Introduction, the last sentences of the first two paragraphs should be varied ("However, little is known about the [...] mechanisms that protect [...] during neurodevelopment.")
5. Why is RBM-26 protein running as a doublet at both sizes?
6. When showing the RBM-26 expression pattern (Fig. 3) please include a lower-magnification image of the entire animal.
7. It is confusing to refer to the RNA IP experiments as an "unbiased screen", which in C. elegans typically refers to a genetic screen.
8. The relationship between axon overextension, beading, and mitochondrial localization is not clear. What causal connection between these is being proposed? The causal connections between these phenotypes, if any, should be clarified experimentally. For example, if the axon extension defects develop before mitochondrial localization defects, then it is unlikely that mitochondrial defects cause axon overextension.
9. Please explain how to interpret the difference in axon beading in the two deletion alleles of the MALSU homolog (axon beading defects in tm12122 but not in syb6330). Is syb6330 not a null allele? Or are the defects in tm12122 due to other mutations in this strain background?
10. Are mitochondria reduced in number or mislocalized? If they are reduced in number, is this due to altered balance of fission/fusion?
11. In Fig. 3A-D, please keep the labels in the same position in all panels and do not alter brightness settings between single-color and merged panels.
12. The claim that rbm-26 acts cell-autonomously requires PLM-specific depletion and rescue experiments.

Referees cross-commenting I appreciate the use of the consultation session to resolve differences between reviewers, but in this case I fully agree with the content and tone of all the comments from the other reviewer -- I think our remarks are very well aligned!

Significance

The study engineers autism-associated variants in conserved residues of RBM27 into the C. elegans homolog RBM-26 and identifies neuronal phenotypes potentially relevant to autism and a potential molecular mechanism involving regulation of mitochondrial ribosome assembly.

The key claims of the study are 1} that autism-associated variants in RBM-26 decrease its protein expression; 2} that impaired RBM-26 function leads to a variety of defects in development and maintenance of a single neuron called PLM, including altered axonal localization of mitochondria; 3} that RBM-26 normally binds the mRNA for the C. elegans homolog of MALSU, a mitochondrial ribosomal assembly factor; 4} that loss of RBM-26 leads to overexpression of the MALSU homolog; and 5} that MALSU is required for some of the deleterious effects on the PLM neuron seen in RBM-26 mutants.

This study will be of interest to the autism research community because it bolsters the idea that variants in RBM27 are likely to disrupt gene function and to affect neuronal health. It will also be of interest to the broader cell biology community because it suggests an interesting potential nucleus-to-mitochondria signaling mechanism, in which a nuclear RNA-binding protein might regulate assembly of mitochondrial ribosomes.

My field of expertise is developmental biology in C. elegans.

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-2024-02413R

Corresponding author(s): Hammond, Gerald

1. General Statements [optional]

We are grateful to the three reviewers for such thorough and thoughtful comments. Data or re-writes that we have on hand that address many of these comments have been incorporated already. We also have a comprehensive experimental plan to address all of the remaining major comments. Reviewer’s comments are in light italics, whilst our responses appear in regular font below. We added reviewer numbering for ease of cross-reference to the original comments, with the format: reviewer X’s comment number N as #X.N

• *

*Overall, we were thrilled that the reviewers agreed that our work is of significance and broad interest: *

• *

“The development of a new, superior PA-sensor is a significant advance in the fields of lipid signaling and specific lipid-protein interactions, that will benefit research on lipid-mediated cellular signaling and intracellular lipid trafficking.” – reviewer 1.

“The lipid biology community would be highly interested in using the new PA-binding tool to study lipid localization in live cells.” – reviewer 2.

“Tracking intracellular phosphatidic acid (PA) in live cells is essential for understanding its cellular functions, leading to the development of genetically encoded lipid biosensors. While several PA biosensors have been developed, they often suffer from limited sensitivity or specificity.” – reviewer 3

2. Description of the planned revisions

#1.2: P 4 and Fig. 2 mention a 'novel domain structure', responsible for binding of PA and PIP2 (in vitro). What exactly is this novel domain structure? Why have the two parts of Nir1-LNS, the AAH and the 263 amino acid domain, not been tested in similar liposome assays as in Fig. 2A? Lipid binding in vivo, as tested in the experiment of Fig. 2D, is confounded by endogenous PA binding proteins.PA is expected to bind the SIDGS motif, as this is conserved from the Lipin catalytic motif (p. 5). However, experimental evidence of this appears to be lacking. Nevertheless, it is several times presented as a fact in the text. Pag. 7: 'This suggests that the SIDGS motif alone is not the sole PA binding pocket as the LNS2 domain requires both that motif and the amphipathic helix for sustained binding to membrane embedded PA.' Pag. 13: '... One reason for this difference could be that the Nir1-LNS2 is not a novel bona fide PA binding pocket. Rather, it requires both the SIDGS motif and an N-terminal amphipathic helix for membrane interactions (Figure 2). These conclusions therefore appear to be not accurate and would need te be rephrased. If the authors could show that PA binding by Nir1-LNS can be eliminated by mutating residues in the SIDGS motif, this would not only substantiate the above claims, but also make for a negative control protein, next to Nir1-LNS2. For future applications of Nir1-LNS2 as PA biosensor in other organisms this would be useful.

In addition to the already included data on cellular binding of the R784E mutant, we do plan to test this variant in the liposome binding assays to show loss of PA binding abilities as the reviewer has suggested. We also plan to evaluate proper folding of the R784 mutant through circular dichroism.

• *

#1.7. Fig. 2A suggests cooperativity in binding of Nir1-LNS2 to PA-containing liposomes. Please mention/comment! Does binding to PIP2-containing liposomes also exhibit cooperativity?

Using a nonlinear fit, we were able to determine Hill coefficients for PA binding. This has now been included in Figure 2B (formerly Figure 2A) and the following text has been added to page 6, left column, third paragraph: “In addition, we found that Nir1-LNS2 bound PA-rich liposomes in a concentration dependent manner (Figure 2B). By fitting the binding curve to this data, we found that the interaction of Nir1-LNS2 with PA provided a Kd value of ~19 mol%. Interestingly, Nir1-LNS2 binds to PA in a highly cooperative manner. The Hill coefficient for the interaction of Nir1-LNS2 with PA was calculated to be approximately 4 (Figure 2B).”

However, due to the liposome binding assay used that utilizes a set total lipid concentration but alters mol% lipids, the Kd that we determined is not a “traditional” Kd. Therefore, we plan to repeat this assay using constant PA concentrations but increasing total concentrations of lipid so that we can make a better fit and get a more accurate Kd value and Hill coefficient. We also plan to do the same assay with PIP2 to determine Kd values and Hill coefficients for that interaction.

• *

#2.2. The authors mention high affinity of Nir1-LNS2, but it lacks in vitro characterization that should demonstrate the higher affinity of Nir1-LNS2 compared to conventional probes such as Spo20. The authors should perform side-by-side comparison in Fig 2 to compare the PA affinity and specificity of Nir1-LNS2 compared to Spo20.

We plan to take the reviewer’s advice and directly compare Spo20-PABDx2 (and/or the single PABD depending on what we can get to purify correctly) and Nir1-LNS2 in the liposome binding assay.

Additionally, we propose to further characterize these sensors in cells as well. To start, we have added a direct comparison of the Spo20-PABDx2 and Nir1-LNS2 response to PA production at the PM (by PMA stimulation) and at mitochondria (by FKBP-DGKa) in Figure 4. The text has been updated to reflect this on p.10, left paragraph, 3rd paragraph: “Importantly, we also observed that Nir1-LNS2 responds to this ectopic PA production quicker and more robustly than NES-PABDx2-Spo20 does, as can be seen when the responses from Figure 4F are plotted together (Figure 4H). When analyzing the responses to PA production at the PM by PMA stimulation in Figure 1D and Figure 1F, we similarly see that the Nir1-LNS2 translocates to the PM more robustly and in a shorter timeframe (Figure 4G). This suggests that the Nir1-LNS2 can serve as a high affinity PA biosensor at various cellular locations.”

Furthermore, as suggested in the “General Assessment” we propose to use the FKBP-DGKa system to produce PA on other organelles such as the Golgi and ER and then we can directly compare the response of Spo20-PABDx2 and Nir1-LNS2 to the increase in PA at these organelles. This data will be added to Figure 4 for a full comparison of the sensors across cellular locations.

• *

#2.4. Fig 3 and Fig 4 need the validation of PIP depletion/production using PIP-binding probes.

We propose to repeat the experiments using TIRF in figure 3 as it will give us increases sensitivity, and also compare selectivity with the currently used spo20-based biosensors.

• *

#2.“General assessment”: The existing PA-binding probe using Spo20 is indeed quite blunt, which takes minutes to see appreciable accumulation of the probe upon PA production. Nir1-LNS2 can be indeed useful if it offers better spatiotemporal precision. However, the advantage of this tool over existing tools is not convincing without head-to-head comparison of either (1) in vitro characterization of PA binding affinity and selectivity between Nir1-LNS2 and Spo20 or (2) response to PA produced on different subcellular localizations other than plasma membrane and mitochondria (e.g., endosomes, golgi, and endoplasmic reticulum).

In order to address the selectivity of Nir1-LNS2 and Spo20, we propose to repeat the experiments in Figure 3 with the PJ enzymes in order to see how the PM PIPs affect Spo20 membrane binding, as described in our response to #2.4. Previously published data, as well as our own unpublished observations suggest that Spo20 interacts with the anionic PIPs to a greater extent than Nir1-LNS2 does (Nakanishi et al., 2004 doi: 10.1091/mbc.e03-11-0798; Horchani et al., 2014 doi: 10.1371/journal.pone.0113484). If we can show that Spo20’s interactions with the PM are significantly influenced by the PIPs, then this will add more evidence to the idea that Nir1-LNS2 is more selective for PA.

As described in response to #2.2, we are also planning a side-by-side comparison of spo20 based protein binding on liposomes alongside Nir1-LNS2.

Also, as discussed above, we agree with the reviewer that looking at the Nir1-LNS2 and Spo20 responses to PA production at other organelles would increase confidence that Nir1-LNS2 has a higher affinity for PA. We propose to add these experiments to Figure 4.

• *

#3.1. The direct measurement of the binding affinity of Nir1-LNS2 with PA, e.g., Kd, is essential; this information will help the field explore the potential usage of Nir1-LNS2.

Using a nonlinear fit, we were able to determine Hill coefficients for PA binding. This has now been included in Figure 2B (formerly Figure 2A) and the following text has been added to p.6, left column, 3rd paragraph: “In addition, we found that Nir1-LNS2 bound PA-rich liposomes in a concentration dependent manner (Figure 2B). By fitting a nonlinear curve to this data, we found that the interaction of Nir1-LNS2 with PA provided a Kd value of ~19 mol%. Interestingly, Nir1-LNS2 binds to PA in a highly cooperative manner. The Hill coefficient for the interaction of Nir1-LNS2 with PA was calculated to be 4.323 (Figure 2B)…. This suggests that the amphipathic helix and the SIDGS-containing domain may both interact with the membrane leading to the cooperative nature of Nir1-LNS2’s binding of PA-rich liposomes (Figure 2B).”

However, due to the liposome binding assay used that utilizes a set total lipid concentration but alters mol% lipids, the Kd that we determined is not a “traditional” Kd. Therefore, we plan to repeat this assay using higher total lipid concentrations with a fixed PA mol% so that we can make a better fit and get a more accurate Kd value and Hill coefficient. Furthermore, we plan to directly compare Spo20-PABDx2 (and/or the single PABD depending on what we can get to purify correctly) and Nir1-LNS2 in the liposome binding assay to directly compare their affinities for PA in vitro, as described in response to #2.2.

• *

#3.2. As mentioned by the authors, there is a confusing inconsistency regarding why Nir1-LNS2 binds to PIP2 in vitro but not in cells. Going beyond what has been discussed in the manuscript, there is a possibility that PIP2 could induce Nir1-LNS2 aggregation, leading to pelleting after centrifugation, among many other possibilities. I recommend the authors perform additional in vitro experiments, including but not limited to the liposome floatation assay to directly examine Nir1-LNS2 binding to the liposomes with varied compositions.

This is an excellent suggestion. We plan to check for aggregation by liposome flotation with Nir1-LNS2 in the presence of high mol% of PA and PIP2. In addition, we will also perform circular dichroism to see if PA or PIP2 liposomes are inducing any unfolding of Nir1-LNS2.

#3.3. In Fig. 2D, it would be beneficial to examine the constructs Nir1-613-630 and Nir1-631-894, comparing them with Nir1-LNS2 using liposome sedimentation and floatation assays to evaluate the contribution of the SIDGS motif and the amphipathic helix in binding PA.

Per our response to #1.2, we looked around the SIDGS motif to find the residue that would mediate the binding of membrane embedded PA, which our data suggests is R784 (Figure 2D). We do plan to test the R784 mutant in the liposome binding assays to show loss of PA binding abilities as the reviewer has suggested. We also plan to evaluate proper folding of the R784 mutant through circular dichroism.

• *

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

#1.1: CCH treatment of HEK293A cells leads to the PM localization of the DAG sensor C1ab-Prkd1 as well as Nir1-LNS2 (Fig. 5), and the kinetics of these changes - Nir1-LNS2 would lag behind C1ab-Prkd1 fluorescence - is taken as evidence for Nir1-LNS2's specific binding of PA rather than DAG: Pag. 10: 'When we look at the first 2-minutes after CCh addition, we see that C1ab-Prkd1 moves to the PM much faster than Nir1-LNS2 does (Figure 5D). The delay in Nir1-LNS2 translocation makes sense given DAG is produced first and then converted into PA, again indicating that Nir1-LNS2 is specific for PA. 'Fig. 5 legend: 'The Nir1-LNS2 response to PLC depends on PA and not DAG.(...) (D) Nir1-LNS2 translocation to the PM (data replicated from Figure 5C) lags behind C1ab-Prkd1 (data replicated from Figure 5B) in response to CCh addition. 'The validity of the conclusion from this experiment seems questionable. The argument relies on the low values of Nir1-LNS2's normalized fluorescence intensity compared to C1ab-Prkd1's, in the first two minutes of stimulation, when DAG is expected to accumulate (Fig. 5D). However, if Nir1-LNS2 would bind DAG, the resulting fluorescence is expected to be low, i.e. compared to the much higher signal resulting from subsequent PA binding. Moreover, in this system, which co-expresses the two sensor proteins, competition in binding may account for the apparent precedence of one sensor over the other. Thus, even if the increase in C1ab-Prkd1 fluorescence would precede Nir1-LNS2 - following the authors' interpretation - this would not exclude binding of DAG by Nir1-LNS2.Fig. 5D shows the confocal images of cells after 30 sec CCH treatment using the two sensors, next to the respective controls, replicated from Fig. 5B/C. However, in Fig. 5D, different colors are used for Nir1-LNS2 than in Fig. 5C, which makes comparison along the time course difficult. In conclusion, the data presented in Fig. 5 do not exclude DAG binding by Nir1-LNS2, and modification of the conclusions from this experiment throughout the ms (including the cited sentences) is recommended. Consider removal of Figure 5D. Despite these considerations, the authors' final conclusion regarding the specificity of Nir1-LNS2 towards PA appears well supported (e.g. by the data presented in Figure 4).

We agree with the reviewer that the interpretation of the kinetic data is ambiguous and does not fully negate the idea that Nir1-LNS2 may bind to DAG. We have modified the interpretation accordingly. However, we have left the kinetic comparison of the DAG vs Nir1-LNS2 biosensors since these reflect the expected dynamics of the two lipids downstream of PLC. The data are now interpreted as follows on p. 10, right column, second paragraph: “The PM accumulation lagged that of DAG, consistent with conversion of DAG to PA by DGKs (Figure. 5D). Alternatively, in cells treated with CCh and then atropine, Nir1-LNS2 localized to the PM after CCh was added but was then observed returning to the cytoplasm over the 15-minute treatment with atropine as PA levels declined (Figure 5C). Overall, this experiment shows that Nir1-LNS2 binding to the PM follows the expected kinetic profile of DGK-produced PA.” Likewise, the legend for figure 5 is now labelled “Nir1-LNS2 detects PLC stimulated PA production.” to remove explicit conclusions about PA vs DAG binding.

#1.2: P 4 and Fig. 2 mention a 'novel domain structure', responsible for binding of PA and PIP2 (in vitro). What exactly is this novel domain structure? Why have the two parts of Nir1-LNS, the AAH and the 263 amino acid domain, not been tested in similar liposome assays as in Fig. 2A? Lipid binding in vivo, as tested in the experiment of Fig. 2D, is confounded by endogenous PA binding proteins.PA is expected to bind the SIDGS motif, as this is conserved from the Lipin catalytic motif (p. 5). However, experimental evidence of this appears to be lacking. Nevertheless, it is several times presented as a fact in the text. Pag. 7: 'This suggests that the SIDGS motif alone is not the sole PA binding pocket as the LNS2 domain requires both that motif and the amphipathic helix for sustained binding to membrane embedded PA.' Pag. 13: '... One reason for this difference could be that the Nir1-LNS2 is not a novel bona fide PA binding pocket. Rather, it requires both the SIDGS motif and an N-terminal amphipathic helix for membrane interactions (Figure 2). These conclusions therefore appear to be not accurate and would need te be rephrased. If the authors could show that PA binding by Nir1-LNS can be eliminated by mutating residues in the SIDGS motif, this would not only substantiate the above claims, but also make for a negative control protein, next to Nir1-LNS2. For future applications of Nir1-LNS2 as PA biosensor in other organisms this would be useful.

We have clarified that the domain architecture of the Nir1-LNS2 is not a novel domain structure generally, but novel for a PA binding protein which are typically just helices such as that seen in Spo20. Figure 2 is now titled “Nir1-LNS2 shows specificity for PA and PIP2 in vitro, based on a novel PA-binding domain.”

We have also clarified that the SIDGS motif is not the actual location of PA binding, but rather is only the motif conserved with the Lipin/Pah active site. R784 appears to be a PA coordinating residue near the SIDGS, as the positive residue can interact with the negative lipid. Furthermore, we agreed with the reviewer that mutating this residue to perturb PA binding was a much more convincing experiment. We have now included this data in Figure 2 and rewritten the following passages.

From Page 6, left column, last paragraph: “The putative Lipin catalytic motif DxDxT is partially conserved in Nir1-LNS2 as a SIDGS motif spanning residues 742-746. We looked for positively charged residues nearby that could bind to the PA in the membrane and coordinate its entrance into the SIDGS site. The active site of the Lipins has a nearby Arg residue which was predicted to perform this role (Khayyo et al., 2020). AlphaFold analysis of Nir1-LNS2 showed that this residue was also conserved in Nir1-LNS2 as R784, and that the side chain of the Arg sticks out toward the membrane interface where it would be able to contact the negatively charged PA (Figure 2C).

The conservation of these features between the Lipins and Nir1-LNS2 suggests that PA binds this positively charged residue near the SIDGS pocket within Nir1-LNS2 (Kim et al., 2013; Khayyo et al., 2020). However, for efficient catalytic activity, the Lipins also require an N-terminal amphipathic helix for membrane interaction. This helix is made up of residues 1-18 in Tetrahymena thermophila Pah2 (Khayyo et al., 2020), and residues 613-630 in the N-terminus of Nir1-LNS2 are predicted to form a similar amphipathic helix (Figure 2C). We therefore tested whether the N-terminal helix of Nir1-LNS2 was necessary for interaction with PA at the PM. We made two truncations of the Nir1-LNS2 construct: Nir1-613-630 is the isolated amphipathic helix, while Nir1-631-894 is the rest of the domain excluding the helix but including the SIDGS motif. Surprisingly, neither truncated construct responded to PMA by binding the PM, and they even showed reduced basal PM localization (Figure 2D).

Although Figure 2D suggests that the SIDGS motif alone is not sufficient for membrane interactions, we probed into the suspected PA binding residue R784 by mutating it into a negatively charged Glu residue, which should disrupt its interaction with the negatively charged lipid. The R784E mutation completely ablated Nir1-LNS2 interactions at the PM after PMA stimulation and showed reduced association with the PM even before PMA stimulation (Figure 2D).

Altogether, our data suggests that the LNS2 domain requires both the larger SIDGS-containing domain and the amphipathic helix for sustained binding to membrane-embedded PA, but that the PA may directly interact with R784 near the SIDGS motif. Therefore, the Nir1-LNS2 provides a novel PA binding domain with a tertiary structure beyond the simple amphipathic helices found in Spo20.”

We have also rewritten this sentence in the discussion, p. 14, right column, second paragraph: “As far as the use of Nir1-LNS2 as a biosensor, the one caveat is the discrepancy in its specificity: in vitro PA and PIP2 were sufficient to recruit Nir1-LNS2 to PC liposomes (Figure 2), but in vivo only PA was sufficient for mitochondrial recruitment (Figure 4). One reason for this difference could be that the Nir1-LNS2 requires R784 near the SIDGS pocket and an N-terminal amphipathic helix for membrane interactions (Figure 2).”

---

#1.3. Pag. 7: '...small caveat to consider when using Nir1-LNS2 to study PA, the data also demonstrates that Nir1-LNS2 is not specifically interacting with any of the PM PIPs in cellular membranes.'This seems not accurate, since the data in Fig. 3 suggest that PI4P could be involved in membrane localization of Nir1-LNS2. It remains however unresolved whether this is a specific interaction with this PIP.

We have rewritten the text on Page 8, right column, first paragraph accordingly: “This data suggests that decreasing the anionic charge of the membrane through depletion of PIPs slightly reduces Nir1-LNS2’s ability to interact with the PM, but it doesn’t fully re-localize the sensor. Therefore, this is a caveat to consider when using Nir1-LNS2 to study PA.”

• *

#1.4. Please note that the presence of PE increases the ionization (and negative charge) of PIP2 (Graber et al., 2012) rather than dilutes the negative charge as stated in the Discussion on p.13. Please revise!

We have updated the text on Page 14, right column, last paragraph: “The presence of other lipids such as PI, the formation of PIP2-rich domains, and even interactions with neighboring proteins can increase hydrogen bonding of PIP2and dilute the negative charge (Graber et al., 2012; Borges-Araújo and Fernandes, 2020). Phosphatidylethanolamine on the other hand, increases PIP2 ionization and its negative charge, though these effects are also thought to be reduced by PIP2 intramolecular hydrogen bonding which competes for the charges on the lipid (Graber et al., 2012).”

#1.5. P 1 The authors may consider adding a 4th criterium for a lipid biosensor: the sensor should not serve as a sink for the lipid by removing/sequestering it from the active pool, thereby interfering with other interactions/conversions.

We agree that biosensors should not sequester a significant fraction of their cognate lipids and affect downstream pathways by competing with endogenous binding partners. We have rewritten the following text regarding Figure 6 to make this distinction more clear:

Page 13, left column, last paragraph: “As Nir1-LNS2 shows high affinity for PA across cell lines, this brings up the concern that use of Nir1-LNS2 will sequester PA and inhibit endogenous signaling pathways that depend on PA…Therefore, we conclude that use of Nir1-LNS2 as a PA biosensor does not sequester significant amounts of PA. It is suggested that cellular homeostasis may compensate for the amount of bound lipid by increasing synthesis of free lipid, as this has been seen with the PIP2 biosensor PH-PLCd1 (Traynor-Kaplan et al., 2017). While PA has a plethora of cellular functions, the fact that Nir1-LNS2 expression does not disrupt MCS formation shows promise that the high affinity of Nir1-LNS2 will not inhibit downstream PA signaling.

#1.6. Nir1 lacks a PITP domain (Fig. 1), yet is referred to as lipid transfer protein: please elaborate/explain.

The following text has been added to Page 2, left column, last paragraph to clarify this point: “This family of proteins, made up of Nir1, Nir2, and Nir3, form ER-PM membrane contact sites (MCS) to exchange PA and phosphatidylinositol (PI) between the compartments (Cockcroft and Raghu, 2016; Kim et al., 2015). While Nir1 lacks a functional PITP domain, it was initially classified as part of the PITP family based on the homology of its other domains with Nir2 and Nir3. Furthermore, Nir1 has a role in lipid transfer by facilitating Nir2 recruitment to the MCS (Quintanilla et al., 2022).”

• *

#1.8. Indicate the concentrations of PC and protein in the legend to Fig. 2 panels A and B. M&M says 2 mM PC, according to the PA-concentrations above panel 2A, this should be 1 mM. Please clarify.

We have corrected the typo in panel 2A (now panel 2B) and have updated the Figure 2 legend as follows, “(A) A representative SDS-PAGE gel is shown for Nir1-LNS2 binding to various PM lipids in POPC liposomes. (B) A representative SDS-PAGE gel is shown for Nir1-LNS2 binding of increasing PA molar concentrations in POPC liposomes. For both A and B, the lipids indicated were mixed with POPC to produce a 2 mM solution, then 50 uL of the resulting liposome mixture was incubated with 50 uL of Nir1-LNS2 at ~1 mg/mL. Supernatant (S) and pellet (P) lanes were quantified using ImageJ to determine percent protein bound. The protein-only control pellet was used as a baseline (input). Nir1-LNS2 appears on the gel at 37 kDa.”

#1.9. In Fig. 2B, PI is missing. Any specific reason?

We have updated the text on Page 6, left column, second paragraph to discuss the low levels of PI at the PM, which is why we did not include this lipid. “Using this same PC background, we tested the efficacy of the PM lipids DAG, PA, PS, PI4P and PIP2 in recruiting Nir1-LNS2 to membranes. While PI serves as a substrate for PI4P and PIP2 synthesis (collectively referred to as the phosphatidylinositol phosphates (PIPs)) at the PM, levels of PI at the PM are very low compared to the PIPs and therefore PI itself was not tested (Zewe et al., 2020; Pemberton et al., 2020).”

#1.10. Move Fig. 2C to the Introduction and extend it to illustrate the shared conserved features of Nir1-LNS2 and lipin.

We would like to keep the diagram of the Nir1-LNS2 in Figure 2 where the features are discussed in more detail than in the introduction. However, we did add this sentence to the introduction __on p.2, right column, second paragraph __that refers the reader to the cartoon in Figure 2. “These features are conserved in the Nir LNS2 domains, except for the catalytic Asp in the DxDxT motif and another Mg2+-coordinating residue (Figure 2C).”

#1.11. P 13. 'While real-time IMPACT does not directly report on PA levels as it does not use the endogenous PLD substrate PC, ...'It is true that this method doesn't directly report on PA levels, but that is because it uses a click chemistry probe as substrate for PLD's transphosphatidylation reaction. Contrary though to what is stated by the authors, this reaction still uses the endogenous substrate of PLD, PC (Liang et al. 2019; www.pnas.org/cgi/doi/10.1073/pnas.1903949116).

We have rewritten the aforementioned sentence in the discussion (Page 15, right column, second paragraph): “While real-time IMPACT does not directly report on PA levels as it creates a unique fluorescent lipid, it offers several advantages such as being able to interrogate lipid trafficking over time.”

#1.12. Fig. 1K: Control must have been treated with PMA plus vehicle (DMSO); if so, please indicate that vehicle was added.

Figure 1K and its legend have been updated. The legend now reads “Stimulating HEK293A cells with 100 nM PMA and 750 nM of the PLD inhibitor FIPI reduces the Nir1-LNS2 response to PMA and cell media (Veh)”

#1.13. Figure 6C: How is DFt/Fpre defined? Add to legend.

We have updated the Figure 6 legend to read “MCS formation was quantified as the change in fluorescence at a given time (Ft) divided by the fluorescence before CCh stimulation (Fpre).”

#1.14. P 4: 'The boundaries of the Nir2-LNS2 (Uniprot: O00562) and Nir3-LNS2 (Uniprot: Q9BZ72) were also defined using AlphaFold predictions of the structure of these domains.' How was the extent of each of these domains determined - they are much larger than the previously published sequences of LNS2 domains (Kim et al. 2013; Embo Rep. 14:891-899. doi:10.1038/embor.2013.113)?

We have clarified the definition of boundaries by updating the following sentence on Page 4, right column, 5thparagraph: “The boundaries of the Nir2-LNS2 (Uniprot: O00562) and Nir3-LNS2 (Uniprot: Q9BZ72) were also defined using AlphaFold predictions of the structure of these domains. Previous definitions of the Nir2-LNS2 domain have considered the domain smaller than we do here (Kim et al., 2013, 2015) . However, according to AlphaFold, the boundaries set previously exclude a large N-terminal beta barrel that is conserved in the Lipin/Pah PAPs, as well as disrupt the domain fold that is homologous to the Lipin active site. Therefore, we are confident that our constructs include the entire LNS2 fold.”

#1.15. In Fig. 3 legend, specify the starting condition of Nir1-LNS2 binding?? Which fluorescence are we looking at?

This figure has now been revised in response to __point #3.5, __which hopefully also clarified this point.

---

#1.16. In legend to Fig. 6 please specify the fluorescent tags used. Have they been shown not to affect protein function?

We have updated the figure legend to specify that GFP-Nir2 was used in conjunction with iRFP-Nir1-LNS2, we also changed the text on Page 13, right column, second paragraph that refers to this experiment. It now reads “We co-expressed a GFP-tagged Nir2 and either iRFP-Nir1-LNS2 or iRFP-TubbyC, a PIP2 biosensor that is not expected to affect MCS formation. It should be noted that although we have used the NG-tagged Nir1-LNS2 the most extensively, the iRFP and mCherry-tagged biosensors have behaved the same as the NG-tagged version in the experiments where we utilized them.”

• *

#2.1. Nir1-LNS2 seems to show variable basal localization across different representative images presented in the manuscript. A part of them were justified by the effect of other anionic species by PIP (such as Fig 3 where they co-expressed PIP-degrading enzymes). For example, cells in Fig 1F and those in Fig 4F show quite different basal localization of Nir1-LNS2. Is it due to difference in expression level, cell conditions, or other factors The significant amount of plasma membrane basal localization seems to indicate that Nir1-LNS2 localization is affected by its binding to PI(4,5)P2.? The significant and potentially variable plasma membrane localization of Nir1-LNS1 can limit the utility of this probe.

We have added Supplemental Figure 1 __to show the range of Nir1-LNS2 basal localization compared to NES-PABDx2-Spo20 and PASS. We believe that this localization is due to variable amounts of basal PA combined with some non-selective anionic interactions at the PM. The following paragraph has been added to __page 7, left column, first paragraph to discuss this point, “Since the R784E mutant showed reduced basal PM localization, we wanted to further characterize the basal localization of the wild-type Nir1-LNS2. The basal localization of wild-type Nir1-LNS2 varies somewhat between cells, but analysis of all of the cells used throughout this study determines that the basal PM/Cyt ratio of the wild-type Nir1-LNS2 is 1.0644 ± 0.0672, which suggests that at resting conditions Nir1-LNS2 is slightly enriched at the PM (Supplemental Figure 1A, 1D). When we did the same analysis for all the cells where we expressed NES-PABDx2-Spo20 or PASS, we obtained a basal PM/Cyt ratio of 1.1318 ± 0.0954 for NES-PABDx2-Spo20 and a ratio of 0.6861 ± 0.0143 for PASS (Supplemental Figure 1B, 1C, 1E). We believe that the basal localization of these sensors reflects variable PA levels in the PM at resting conditions. FRET based imaging of PA has indicated that there are detectable levels of PA under basal conditions, and this approach also showed some variability in basal PA levels as we see with the spread of Nir1-LNS2’s basal localization (Nishioka et al., 2010). Overall, our data suggests that the high affinity of Nir1-LNS2 for PA is reflected in both its basal localization and its response to stimulations such as PMA.”

To address the idea that PIP2 is responsible for the basal localization of Nir1-LNS, we have added the following to the discussion on p.15, left column, second paragraph: “Aside from concerns about specificity, the ability of Nir1-LNS2 to interact with PIP2 in liposomes could suggest that the basal PM localization of Nir1-LNS2 is due to it binding PIP2. However, selective depletion of PI(4,5)P2 did not affect basal Nir1-LNS2 localization to the PM (Figure 3C) and was not able to recruit the probe to mitochondria (Figure 4A-C). We did see FKBP-PJ reduce the association of Nir1-LNS2 with the PM under resting conditions (Figure 3E, 3F), suggesting a possible non-specific ionic interaction with polyanionic inositol lipids. Another mechanism to explain these data would be phosphoinositide-dependence of PA production. Phosphoinositides are well-known to regulate the recruitment of PLD isoforms and type II DGKs to the PM as well as their catalytic activity there (Sciorra et al., 2002; Du et al., 2003; Hodgkin et al., 2000; Liscovitch et al., 1994; Kume et al., 2016). Therefore, we suggest that the effects of FKBP-PJ could be reducing basal PLD and DGK activity and hence lowering resting PA levels. That could explain the loss of both basal Nir1-LNS2 PM association when FKBP-PJ is expressed, and Nir1-LNS2’s PM interactions as FKBP-PJ is recruited to the membrane to further deplete phosphoinositides. While this study cannot fully substantiate this hypothesis, the role of PIP2 in PLD activity and PA production is an interesting hypothesis that warrants further investigation.

• *

#2.3. Fig 1 shows Nir1-LNS2 translocates to plasma membrane upon PMA stimulation in a PLD activity-dependent manner. However, the image in Fig1K is not super convincing since there is already a decent amount of plasma membrane localization of the sensor at t = 0 min, which looks considerably different from the t = 0 min image shown in 1F.

We have updated the images both in Figure 1F and Figure 1K to best represent the mean basal localization as determined in Supplemental Figure 1.

• *

#2.4. Fig 3 and Fig 4 need the validation of PIP depletion/production using PIP-binding probes.

These controls are shown in Figure 4B. We have only included PH-PLCd1 to show PIP2 levels as the large PIP2 production by a PIP5K also indicates the large elevation of the substrate PI4P.

This control data has now been included in Figure 3, and is referenced by the Figure 3 legend and the following text from p.8, left column, 4th paragraph: “As a negative control, we expressed a doubly catalytically dead mutant of PJ. When PJ-Dead was recruited to the PM, we confirmed that PIP2 and PI4P levels remained unaltered by seeing stable association of the PIP2 biosensor Tubby(c) with the PM (Figure 3A). We observed no loss of the PM localization of Nir1-LNS2 with PJ-Dead recruitment (Figure 3A, 3E). When the active PJ was expressed in HEK293A cells, there was a slight loss of Nir1-LNS2 at the PM even before PJ recruitment (Figure 3B, 3E), although this was not significant as compared to pre-stimulated cells expressing PJ-Dead (Figure 3F). However, Nir1-LNS2 did move off the PM into the cytosol after PJ recruitment, to a similar extent that the PIP2 biosensor Tubby(c) moved off the PM (Figure 3B, 3E). AUC analysis of the Nir1-LNS2 response showed there was a significant reduction of Nir1-LNS2 PM localization (Figure 3G).

        Since PJ depletes both PIP2 and PI4P, we examined which of these lipids specifically contribute to Nir1-LNS2 membrane binding. We utilized an FKBP-INPP5E construct that depletes PIP2 but does not deplete PI4P at the PM, as seen by the significant loss of PM-localized Tubby(c) (Figure 3C). Then FKBP-Sac1, an FKBP-PJ construct that has a catalytically dead INPP5E domain, but an active Sac1 domain was used to deplete PI4P without altering PIP2 levels, as seen by removal of the PI4P biosensor P4Mx1 from the PM (Figure 3D). Recruitment of FKBP-INPP5E did not significantly affect Nir1-LNS2 localization (Figure 3C, 3E, 3G). However, recruitment of FKBP-Sac1 slightly, but not significantly affected Nir1-LNS2 localization (Figure 3D, 3E, 3G). This data suggests that decreasing the anionic charge of the membrane through depletion of PIPs slightly reduces Nir1-LNS2’s ability to interact with the PM, but it doesn’t fully re-localize the sensor. Therefore, this is a small caveat to consider when using Nir1-LNS2 to study PA.”

• *

#2.“Advance”: The key significance of the manuscript, which is the superiority of Nir1-LNS2 over existing PA-binding probes, is not clear from the data provided. Other than that part, the study does not seem to include significant finding, since the binding of Nir1-LNS2 to PA itself is already known (EMBO Rep. 2013 Oct; 14(10): 891-899, Mol Biol Cell. 2022 Mar 1;33(3):br2).

While the Kim et al., paper referenced by the reviewer does show that the LNS2 binds to PA, this same group later published data showing that the LNS2 binds to both PA and DAG. (Kim et al., 2015 doi: 10.1016/j.devcel.2015.04.028). Therefore, we believe our data which unequivocally shows that the LNS2 does not bind DAG, is a significant advancement in the field. Aside from the creation of the new biosensor, it progresses our understanding of the mechanism of the Nir family lipid transfer proteins, which are vital to PM lipid homeostasis.

To highlight this point, we have added the following paragraph to the discussion on p.14, right column, 1st paragraph: “The lack of Nir1-LNS2 binding to DAG-rich liposomes (Figure 2), DAG produced at the mitochondria (Figure 4), and DAG analogs (Figure 5) shows that the LNS2 domains only binds to PA rather than to PA and DAG as has been reported previously (Kim et al., 2015). In this study, we redefined the boundaries of the LNS2 domain based on the structure of the Lipin/Pah family domains and the AlphaFold prediction for the Nir1-LNS2. The new boundaries included the entire fold that is conserved between the Lipins/Pahs and the Nirs. Therefore, we suspect that the expansion of the LNS2 domain in our work explains the differences in our data and the published literature regarding DAG binding. Importantly, the data obtained with our amended Nir1-LNS2 suggests that within the context of the lipid transfer cycle and MCS formation, the Nir family of PITPs translocate to the PM solely based on PA. This information will be important as the field continues to determine the exact mechanism of the Nir PITPs in lipid homeostasis.”

• *

#3.4. Due to PA's versatile biological roles, the evidence provided by the MCS experiment is far from enough to conclude that Nir1-LNS2 does not interfere with PA function. Further examination of various endogenous pathways is warranted before making the statement "Therefore, Nir1-LNS2 can be used ...... without concern of affecting downstream events".

We have rewritten the quoted sentence for a more nuanced interpretation on p.13, right column, second paragraph:“While PA has a plethora of cellular functions, the fact that Nir1-LNS2 expression does not disrupt MCS formation shows promise that the high affinity of Nir1-LNS2 will not inhibit downstream PA signaling.”

---

#3.5. In Fig. 3A-D (Left), it is unclear to what extent PIPs are reduced after treatment with FKBP-tagged PIP phosphatases. The treatment depicted in the illustration should be accompanied by data, e.g., % of PIPs being degraded after treatment.

This comment is addressed in our response to #2.4, where we show the addition of control biosensors for PIP2 and PI4P, and also propose new experiments in TIRFM for more sensitive and precise measurements.

• *

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

---

__#1.1a: __Have the authors considered using the DGK inhibitor R59022 to selectively block the conversion of DAG to PA by DGK? Such an experiment could provide additional evidence for the requirement of DGK activity and consequent PA formation for Nir1-LNS2 membrane localization.

We did indeed attempt experiments with R59022, and have made several unexpected findings with the compound that go way beyond the scope of the current manuscript. In short, although R59022 reduces DGK catalytic activity, it also potently drives over-expressed or endogenous DGKalpha to the plasma membrane, and induces large accumulations of PM PA. This complicated interpretation of data obtained with this compound. We are currently preparing a manuscript detailing the novel and unexpected effects.

#3.6. In Fig. 4C, the plasma membrane (PM) localization of Nir1-LNS2 and NES-PABDx2-Spo20, as determined by the "intensity PM/Cyto," should be analyzed following the ectopic production of PI4P and PIP2. Although mitochondria do not apparently recruit Nir1-LNS2 or NES-PABDx2-Spo20 after induced PI4P and PIP2 production, it remains possible that the subsequent trafficking of PI4P and PIP2 from mitochondria might sequester the biosensors away from the PM into the cytoplasm, thereby reducing the "intensity PM/Cyto" of Nir1-LNS2.

We cannot easily determine the PM/cyt ratio in this experiment as we included a mitochondrial marker rather than a PM marker when imaging. However, based on the images, there is no change in the PM intensity of the Nir1-LNS2 and NES-PABDx2-Spo20 biosensors. The images included in Figure 4 are representative of this localization.

#3.7. It would be valuable to determine the half-life (stability) of Nir1-LNS2.

In all of our transient transfections, the Nir1-LNS2 shows good stability where we don’t expect degradation to be a major concern. Furthermore, stability has not usually been factor considered in the creation any of the current widely used lipid biosensors.

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

Weckerly et al. introduced a fluorescently tagged Nir1-LNS2 construct capable of binding to both PA and PIP2 in vitro, yet selectively targeting PA-enriched membranes in cells. Their findings demonstrate that Nir1-LNS2 exhibits heightened responsiveness to PA, allowing the authors to uncover a modest contribution of PLD to PA production downstream of muscarinic receptors, a phenomenon not visualized with previous Spo20-based biosensors. Thus, Nir1-LNS2 is a sensitive biosensor, potentially providing researchers with a powerful new tool for real-time investigation of PA dynamics in live cells. The manuscript is well-written, with major conclusions supported by experimental evidence. The tool developed in this study holds significant importance for the field of lipid biology. However, missing controls and weaknesses from the in vitro analysis reduce the overall impact of this work. The authors are encouraged to address the following comments to further strengthen their conclusions:

Major Points:

1. The direct measurement of the binding affinity of Nir1-LNS2 with PA, e.g., Kd, is essential; this information will help the field explore the potential usage of Nir1-LNS2.
2. As mentioned by the authors, there is a confusing inconsistency regarding why Nir1-LNS2 binds to PIP2 in vitro but not in cells. Going beyond what has been discussed in the manuscript, there is a possibility that PIP2 could induce Nir1-LNS2 aggregation, leading to pelleting after centrifugation, among many other possibilities. I recommend the authors perform additional in vitro experiments, including but not limited to the liposome floatation assay to directly examine Nir1-LNS2 binding to the liposomes with varied compositions.
3. In Fig. 2D, it would be beneficial to examine the constructs Nir1-613-630 and Nir1-631-894, comparing them with Nir1-LNS2 using liposome sedimentation and floatation assays to evaluate the contribution of the SIDGS motif and the amphipathic helix in binding PA.
4. Due to PA's versatile biological roles, the evidence provided by the MCS experiment is far from enough to conclude that Nir1-LNS2 does not interfere with PA function. Further examination of various endogenous pathways is warranted before making the statement "Therefore, Nir1-LNS2 can be used ...... without concern of affecting downstream events".

Minor Points:

1. In Fig. 3A-D (Left), it is unclear to what extent PIPs are reduced after treatment with FKBP-tagged PIP phosphatases. The treatment depicted in the illustration should be accompanied by data, e.g., % of PIPs being degraded after treatment.
2. In Fig. 4C, the plasma membrane (PM) localization of Nir1-LNS2 and NES-PABDx2-Spo20, as determined by the "intensity PM/Cyto," should be analyzed following the ectopic production of PI4P and PIP2. Although mitochondria do not apparently recruit Nir1-LNS2 or NES-PABDx2-Spo20 after induced PI4P and PIP2 production, it remains possible that the subsequent trafficking of PI4P and PIP2 from mitochondria might sequester the biosensors away from the PM into the cytoplasm, thereby reducing the "intensity PM/Cyto" of Nir1-LNS2.
3. It would be valuable to determine the half-life (stability) of Nir1-LNS2.

Significance

Tracking intracellular phosphatidic acid (PA) in live cells is essential for understanding its cellular functions, leading to the development of genetically encoded lipid biosensors. While several PA biosensors have been developed, they often suffer from limited sensitivity or specificity.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

The authors reported a new PA-binding probe Nir1-LNS2, which potentially offers advantages over conventional tools with its higher sensitivity for PA. The authors performed extensive characterization in different cell lines to test the ability of Nir1-LNS2 to selectively bind to PA without disrupting endogenous PA signaling. While the tool is potentially useful as a new PA-binding probe with higher spatiotemporal precision, the data provided in the manuscript are not enough to support their claims and conclusions. Especially, the data do not fully support that the Nir1-LNS2 offers more sensitive and selective binding to PA than conventional PA-binding probes using Spo20.

Major comments:

1. Nir1-LNS2 seems to show variable basal localization across different representative images presented in the manuscript. A part of them were justified by the effect of other anionic species by PIP (such as Fig 3 where they co-expressed PIP-degrading enzymes). For example, cells in Fig 1F and those in Fig 4F show quite different basal localization of Nir1-LNS2. Is it due to difference in expression level, cell conditions, or other factors? The significant amount of plasma membrane basal localization seems to indicate that Nir1-LNS2 localization is affected by its binding to PI(4,5)P2. The significant and potentially variable plasma membrane localization of Nir1-LNS1 can limit the utility of this probe.
2. The authors mention high affinity of Nir1-LNS2, but it lacks in vitro characterization that should demonstrate the higher affinity of Nir1-LNS2 compared to conventional probes such as Spo20. The authors should perform side-by-side comparison in Fig 2 to compare the PA affinity and specificity of Nir1-LNS2 compared to Spo20.

Minor comments:

1. Fig 1 shows Nir1-LNS2 translocates to plasma membrane upon PMA stimulation in a PLD activity-dependent manner. However, the image in Fig1K is not super convincing since there is already a decent amount of plasma membrane localization of the sensor at t = 0 min, which looks considereably different from the t = 0 min image shown in 1F.
2. Fig 3 and Fig 4 need the validation of PIP depletion/production using PIP-binding probes.
3. In Discussion: "while in vivo it solely binds to PA (Fig 4)" - this claim does not seem to be true according to Fig 4, where the overexpression of PIP-degrading enzymes did affect the Nir1-LNS2 basal localization.

Significance

General assessment:

The existing PA-binding probe using Spo20 is indeed quite blunt, which takes minutes to see appreciable accumulation of the probe upon PA production. Nir1-LNS2 can be indeed useful if it offers better spatiotemporal precision. However, the advantage of this tool over existing tools is not convincing without head-to-head comparison of either (1) in vitro characterization of PA binding affinity and selectivity between Nir1-LNS2 and Spo20 or (2) response to PA produced on different subcellular localizations other than plasma membrane and mitochondria (e.g., endosomes, golgi, and endoplasmic reticulum).

Advance:

The key significance of the manuscript, which is the superiority of Nir1-LNS2 over existing PA-binding probes, is not clear from the data provided. Other than that part, the study does not seem to include significant finding, since the binding of Nir1-LNS2 to PA itself is already known (EMBO Rep. 2013 Oct; 14(10): 891-899, Mol Biol Cell. 2022 Mar 1;33(3):br2).

Audience:

The lipid biology community would be highly interested in using the new PA-binding tool to study lipid localization in live cells.

My expertise is PA signaling and deveopment of engineered phospholipase Ds, which can produce PA on demand at various subcellular locations.

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 designed, created, and validated a fluorescently tagged sensor protein that binds with high affinity to the signaling phospholipid PA in cells. The LNS2 PA-binding domain used originates from the lipid transfer protein Nir1, and shares conserved features with lipins. The novel sensor outperforms the commonly used Spo20-based PA probes, although it also suffers from binding to PIP2 in vitro (liposomes) and from PIPs affecting its membrane binding in vivo. Importantly, the authors demonstrate that PA but not DAG or PIP2 is sufficient for membrane binding of Nir1-LNS2 in cells, validating Nir1-LNS2 as a PA-sensor in fluorescence microscopy studies.

Major comments:

1. CCH treatment of HEK293A cells leads to the PM localization of the DAG sensor C1ab-Prkd1 as well as Nir1-LNS2 (Fig. 5), and the kinetics of these changes - Nir1-LNS2 would lag behind C1ab-Prkd1 fluorescence - is taken as evidence for Nir1-LNS2's specific binding of PA rather than DAG: Pag. 10: 'When we look at the first 2-minutes after CCh addition, we see that C1ab-Prkd1 moves to the PM much faster than Nir1-LNS2 does (Figure 5D). The delay in Nir1-LNS2 translocation makes sense given DAG is produced first and then converted into PA, again indicating that Nir1-LNS2 is specific for PA.' Fig. 5 legend: 'The Nir1-LNS2 response to PLC depends on PA and not DAG.(...) (D) Nir1-LNS2 translocation to the PM (data replicated from Figure 5C) lags behind C1ab-Prkd1 (data replicated from Figure 5B) in response to CCh addition.' The validity of the conclusion from this experiment seems questionable. The argument relies on the low values of Nir1-LNS2's normalized fluorescence intensity compared to C1ab-Prkd1's, in the first two minutes of stimulation, when DAG is expected to accumulate (Fig. 5D). However, if Nir1-LNS2 would bind DAG, the resulting fluorescence is expected to be low, i.e. compared to the much higher signal resulting from subsequent PA binding. Moreover, in this system, which co-expresses the two sensor proteins, competition in binding may account for the apparent precedence of one sensor over the other. Thus, even if the increase in C1ab-Prkd1 fluorescence would precede Nir1-LNS2 - following the authors' interpretation - this would not exclude binding of DAG by Nir1-LNS2. Fig. 5D shows the confocal images of cells after 30 sec CCH treatment using the two sensors, next to the respective controls, replicated from Fig. 5B/C. However, in Fig. 5D, different colors are used for Nir1-LNS2 than in Fig. 5C, which makes comparison along the time course difficult. In conclusion, the data presented in Fig. 5 do not exclude DAG binding by Nir1-LNS2, and modification of the conclusions from this experiment throughout the ms (including the cited sentences) is recommended. Consider removal of Figure 5D. Despite these considerations, the authors' final conclusion regarding the specificity of Nir1-LNS2 towards PA appears well supported (e.g. by the data presented in Figure 4).

Have the authors considered using the DGK inhibitor R59022 to selectively block the conversion of DAG to PA by DGK? Such an experiment could provide additional evidence for the requirement of DGK activity and consequent PA formation for Nir1-LNS2 membrane localization. 2. P 4 and Fig. 2 mention a 'novel domain structure', responsible for binding of PA and PIP2 (in vitro). What exactly is this novel domain structure? Why have the two parts of Nir1-LNS, the AAH and the 263 amino acid domain, not been tested in similar liposome assays as in Fig. 2A? Lipid binding in vivo, as tested in the experiment of Fig. 2D, is confounded by endogenous PA binding proteins. PA is expected to bind the SIDGS motif, as this is conserved from the Lipin catalytic motif (p. 5). However, experimental evidence of this appears to be lacking. Nevertheless, it is several times presented as a fact in the text. Pag. 7: 'This suggests that the SIDGS motif alone is not the sole PA binding pocket as the LNS2 domain requires both that motif and the amphipathic helix for sustained binding to membrane embedded PA.' Pag. 13: '... One reason for this difference could be that the Nir1-LNS2 is not a novel bona fide PA binding pocket. Rather, it requires both the SIDGS motif and an N-terminal amphipathic helix for membrane interactions (Figure 2). These conclusions therefore appear to be not accurate and would need te be rephrased.

If the authors could show that PA binding by Nir1-LNS can be eliminated by mutating residues in the SIDGS motif, this would not only substantiate the above claims, but also make for a negative control protein, next to Nir1-LNS2. For future applications of Nir1-LNS2 as PA biosensor in other organisms this would be useful. 3. Pag. 7: '...small caveat to consider when using Nir1-LNS2 to study PA, the data also demonstrates that Nir1-LNS2 is not specifically interacting with any of the PM PIPs in cellular membranes.' This seems not accurate, since the data in Fig. 3 suggest that PI4P could be involved in membrane localization of Nir1-LNS2. It remains however unresolved whether this is a specific interaction with this PIP. 4. Please note that the presence of PE increases the ionization (and negative charge) of PIP2 (Graber et al., 2012) rather than dilutes the negative charge as stated in the Discussion on p.13. Please revise!

Minor comments:

1. P 1 The authors may consider adding a 4th criterium for a lipid biosensor: the sensor should not serve as a sink for the lipid by removing/sequestering it from the active pool, thereby interfering with other interactions/conversions.
2. Nir1 lacks a PITP domain (Fig. 1), yet is referred to as lipid transfer protein: please elaborate/explain.
3. Fig. 2A suggests cooperativity in binding of Nir1-LNS2 to PA-containing liposomes. Please mention/comment! Does binding to PIP2-containing liposomes also exhibit cooperativity?
4. Indicate the concentrations of PC and protein in the legend to Fig. 2 panels A and B. M&M says 2 mM PC, according to the PA-concentrations above panel 2A, this should be 1 mM. Please clarify.
5. In Fig. 2B, PI is missing. Any specific reason?
6. Move Fig. 2C to the Introduction and extend it to illustrate the shared conserved features of Nir1-LNS2 and lipin.
7. P 13. 'While real-time IMPACT does not directly report on PA levels as it does not use the endogenous PLD substrate PC, ...' It is true that this method doesn't directly report on PA levels, but that is because it uses a click chemistry probe as substrate for PLD's transphosphatidylation reaction. Contrary though to what is stated by the authors, this reaction still uses the endogenous substrate of PLD, PC (Liang et al. 2019; www.pnas.org/cgi/doi/10.1073/pnas.1903949116).
8. Fig. 1K: Control must have been treated with PMA plus vehicle (DMSO); if so, please indicate that vehicle was added.
9. Figure 6C: How is Ft/Fpre defined? Add to legend.
10. P 4: 'The boundaries of the Nir2-LNS2 (Uniprot: O00562) and Nir3-LNS2 (Uniprot: Q9BZ72) were also defined using AlphaFold predictions of the structure of these domains.' How was the extent of each of these domains determined - they are much larger than the previously published sequences of LNS2 domains (Kim et al. 2013; Embo Rep. 14:891-899. doi:10.1038/embor.2013.113)?
11. In Fig. 3 legend, specify the starting condition of Nir1-LNS2 binding?? Which fluorescence are we looking at?
12. In legend to Fig. 6 please specify the fluorescent tags used. Have they been shown not to affect protein function?

Significance

The development of a new, superior PA-sensor is a significant advance in the fields of lipid signaling and specific lipid-protein interactions, that will benefit research on lipid-mediated cellular signaling and intracellular lipid trafficking.

This reviewer's expertise encompasses lipid metabolism and lipid-protein interactions, not so much fluorescence microscopy.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Manuscript number: RC-2023-02235

Corresponding author: Adriano, Aguzzi

1. General Statements

We thank the reviewers for providing valuable comments. We are pleased that our study is considered important to advance the knowledge on IL-1-independent inflammatory functions of inflammasomes. We have clarified and revised the manuscript (track changed) as detailed below in the point-by-point response in this letter.

2. Point-by-point description of the revisions

Referee 1

General: In this manuscript, et al., investigates the role of the inflammasome adapter ASC (in AA amyloidosis). This condition involves the aggregation of serum amyloid A (SAA) and is linked to chronic inflammation. Firstly, I can directly say that I do recommend this study for publication. This is a well conducted and well-written study which advances the knowledge on IL-1-independent inflammatory functions of inflammasomes. Furthermore, I find it particularly impressive that despite the inflammasome research community is well aware that amyloidosis is a hallmark of inflammatory diseases, it took a neuroscientist specialized in prion diseases to raise the question whether ASC would be involved in seeding serum AA aggregation. Key findings include: • ASC forms extracellular aggregates that enhance SAA aggregation, as observed through superresolution microscopy. • In a mouse model, the absence of ASC significantly reduced amyloid load, not due to increased phagocytosis but likely due to diminished aggregation. • Treatment with anti-ASC antibodies reduced amyloid load and mitigated weight loss in mice with AA amyloidosis. These findings suggest that ASC plays a crucial role in AA amyloidosis and that targeting ASC could be a potential therapeutic strategy. The study expands our understanding of the involvement of ASC in proteinopathies beyond neural diseases, pointing to its role in systemic conditions like AA amyloidosis.

__Significance: __In conclusion, this manuscript offers valuable insights into the role of ASC in AA amyloidosis, presenting compelling findings that support its potential as a therapeutic target. Addressing the mentioned concerns and making the suggested revisions will further enhance the manuscript's scientific rigor and impact. Overall, this study is a valuable contribution to the field of inflammasome research and its relevance in systemic conditions like AA amyloidosis.

Comment 1: Overall, the experiments are well-conducted and mostly all controls I would expect were included. With few exceptions, the data is convincing. With that said, I have issues with some of the staining employed in Fig 1. In Fig. 1, the authors assess ASC staining in cardiac tissues from a patient with vasculitis and systemic inflammation-related AA amyloidosis, and a control patient who died of a heart attack but had no signs of amyloidosis. However, most of the data shown is related to the AL177 anti-ASC. More importantly, no isotype stainings are included. We have previously demonstrated that the AL177 anti-ASC, used here, reacts quite strongly with ASC−/− cells, and it is one of the less specific anti-ASC commercially available (PMID: 27221487). As this is data from one patient (understandably), I wonder if the authors could counterstain ASC in the same samples using a specific human anti-ASC with a different color (ex: Biolegend HASC), and confirm that the signal overlays with the AL-177.

Response: We conducted additional experiments to address the anti-ASC antibody specificity, as now described in Results, Method, and Fig. S1. We tested a set of anti-ASC antibodies (AL177, MY6745, 1C3D7) for their ASC specificity. We confirmed that both the AL177 and the MY6745 antibodies have high ASC-specificity (Fig. S1A). Moreover, for illustration purposes (and to warn other scientists), we included a third anti-ASC antibody (1C3D7) found to be unspecific as it yielded a strong signal in PYCARD-/- (ASC-/-) THP-1 cells (Fig. S1B). In addition, isotype controls were included in these experiments (Fig. S1A, right panels), as suggested by the reviewer, showing no target protein detection in both, PYCARD+/+ (ASC+/+) and PYCARD-/- cells underscoring the anti-ASC specificity of AL177 and MY6745 antibodies.

• *

Comment 2: Finally, in Figure 1H it seens from the description that another anti-ASC was used: "referred in the legend as ASC (MAB ASC, Yellow)". Is this a monoclonal anti-ASC? Also, the images show large and bright antibody aggregates (middle of the image, top left corner behind the "H", and a massive fluorescence in the bottom right of the image), indicating the presence of staining artifacts. Again, no counterstaining with isotype controls are shown.

Response: We apologize for the confusing jargon in Figure 1H. “MAB ASC” refers to the anti-ASCPYD antibody (MAB/MY6745). We have corrected the antibody terminology in the legend. MAB/MY6745 is a monoclonal antibody generated by Mabylon that is highly reactive to both human and murine ASC. This antibody was generated to 1) perform an immunotherapy in vivo study and to 2) be used as alternative specific antibody in addition to AL177 to show co-localization of SAA and ASC in a human AA patient using STED superresolution microscopy. MAB/MY6745 is a rabbit monoclonal anti-ASC antibody targeting the pyrin domain (PYD) from which the rabbit Fcγ domain was replaced with that of a mouse IgG2a domain to avoid xenogeneic anti-drug responses in recipients and to improve its effector functions in vivo. To examine possible staining artefacts which can occur with Formalin-Fixed Paraffin-Embedded (FFPE) human tissues, we assessed the specificity of a variety of anti-ASC antibodies (Fig. S1). Our data presented in Fig. S1 show that the monoclonal anti-ASC antibody binds specifically. It is conceivable that AL177 and MAB/MY6745 target different epitopes of ASC, resulting in different staining patterns. An isotype control, included in __Fig. S1, __was used to test the specificity of the secondary antibodies, and did not show any nonspecific staining. We have adapted and added this to the text body and figure legend accordingly.

Comment 3: Overall, although I don't dispute the possibility that ASC would co-localize with SAA deposits, I don't think the data presented can safely sustain that claim. I would, therefore, suggest that alternative methods to be employed to substantiate these conclusions: Supposedly, would it be possible to immuno-precipitate (IP) amyloid SAA and assess ASC via western blotting? As well as IP ASC and detect SAA? Or use DSS-crosslinking to find ASC oligomers in tissue areas rich in SAA?

Response: In addition to assessing co-localization by means of STED superresolution microscopy (Fig. 1), we also employed LiP-MS with various forms of ASC (monomeric and ASC specks) and identified a previously unrecognized biophysical interaction of SAA and the ASC PYD domain (Fig. 2C-F). As an orthogonal line of evidence, we provided kinetic data showing that SAA aggregation is enhanced in the presence of ASC specks (Fig. 2A-B). We feel that these results are reasonably convincing, but we agree that co-localization is almost invariably an aspirational finding, and even superresolution microscopy cannot fully exclude the presence artifacts (nor can, in fairness, co-immunoprecipitation, which must often rely on overexpression). A sentence acknowledging this limitation was added to the Discussion.

Comment 4: For example, it would be reasonable to quantify the results in Figure 3G and providing clarification regarding the controls in the figure legend. Though there is significantly less SAA in spleen homogenates from Asc−/−, there also seems to be the case for b-actin in Fig 3G. Moreover, in the figure legend the authors state: "...Spleen homogenate from untreated (-ctrl) and AA+ (+ctrl) C57BL/6 wt mice from an independent experiment served as negative and positive control, respectively." I don't know what the authors mean with that. Is this a montage, or samples from different experiments were run together in one blot? And if so, for what reason? This is confusing and should be clarified.

Response: We reworded the figure legend to provide clarity about the technical assay controls and adjusted the labels in Fig. 3E __accordingly: To ascertain SAA antibody functionality, mouse spleen homogenate from independently obtained and Congo red-confirmed AA+ tissue served as positive, whereas non-induced (AA-) spleen tissue served as negative technical controls. (__Fig 3E). We decided to show the two (positive/AA+ and negative/AA-) technical controls in Fig. 3E.

Comment 5: Furthermore, in the Abstract, a slight rephrasing is suggested to accurately describe ASC specks as molecular aggregates formed inside cells, which are subsequently released into the extracellular space.

Response: We thank the referee for bringing this to our attention. We rephrased the abstract accordingly.

Comment 6: Lastly, enhancing the text size in figures, particularly in Fig 3, is advised to improve legibility and overall clarity.

Response: The legibility and style of main Fig. 3 text sizes has been changed and additional figure formatting has been performed.

Referee 2

General: The manuscript by Losa et al., investigates whether ASC is involved in serum AA amyloidosis. The authors report that ASC colocalizes with SAA in human AA amyloidosis and that purified ASC specks accelerate SAA fibril formation in vitro. In addition, splenic AA amyloid was decreased in Pycard-/- mice compared to Pycard+/+ mice and that treatment with anti-ASC antibodies decreased amyloid loads in Pycard+/+ mice. Lastly, they analyzed serum of 19,334 patients to show that the prevalence of anti-ASC antibodies did not correlate with any specific disease. The authors conclude that ASC to play a role in extraneural proteinopathies of humans and experimental animals and suggest that anti-ASC immunotherapy may contribute to resolving such diseases. The findings in the study are novel and demonstrate a new role for ASC in aggregation proteinopathies. However, there are number of issues that need to be addressed before acceptance for publication.

Significance: __The findings in the study are novel and demonstrate a new role for ASC in aggregation proteinopathies. This study reports a crucial role for ASC in SAA interaction and recruitment, SAA serum level modulation, SAA fibril formation acceleration, and controlling the extent of inflammation associated amyloidosis with respect to AA amyloid deposition __

Comment 1: Figure 3 E depicts Western blots of monomeric SAA in spleen of Pycard+/+ and Pycard-/- mice. The authors should include immunoblots depicting the levels of ASC in these tissues and to demonstrate that the Pycard-/- mice lack ASC.

Response: We did not perform ASC immunoblots for Pycard-/- and Pycard+/+ mice since the absence of the ASC protein in this well-established mouse line has been demonstrated in several key publications, including under inflammation conditions (right side of the figure below, from Mariathasan et al., Nature, 2014). However, we show ASC IHC of Pycard+/+ and Pycard-/- AA+ mice on spleen, confirming the absence of an ASC signal in Pycard-/- mice and its presence in the Pycard+/+ (Fig. 3F). Moreover, our genotyping data confirmed the presence and absence of the Pycard gene in Pycard+/+ and Pycard-/- AA+ mice.

Comment 2: Fig. 3B shows that at 96 hours after injection there was no difference in SAA serum concentration. How do the authors explain this drop in SAA serum concentration? No explanation is provided.

Response: Acute-phase response peaks at 24 hours after injury (i.e., Kushner I, 1982; Gabay et Kushner, 1999; Gitlin et Colten, 1987, Calif.: Academic Press, 1987:123-53). Beyond 24 hours, acute phase proteins decay over time mirroring the process of tissue integrity restoration and the clearance of the insulting stimuli. This is in line with our data, where the inflammatory injury was induced by subcutaneous AgNO3 injection, resulting in a non-statistical serum SAA difference between the Pycard+/+ and Pycard-/- experimental mice at 96 hours post AgNO3 injection. In addition, the majority of SAA in Pycard+/+ mice was incorporated into amyloid deposit. As suggested by the reviewer we have included this explanation/references into the revised manuscript.

Comment 3: Figure 4 shows anti-ASC administration reduces amyloid load. The immunoblot in Figure 4C does not represent the quantification of the blot. In fact, there are only 3 samples per treatment group whereas the quantification shows 5-6 animals per group.

Response: We have performed two independent immunoblots at the same time to perform technical replicates (duplicates). As pointed out by the reviewer, this resulted in 6 samples and data points that were visualized and analyzed in main Fig. 4C. To avoid duplicating data, overloading the main figures with technical replicates, we opted to show only one representative immunoblot in the main Fig. 4C. The other blots are shown in the supplementary figures Fig. S13A and Fig. S13B for full transparency.

Comment 4: Additionally, the authors have not shown that the drug penetrates the target tissue and how much drug is present in spleen to provide a therapeutic effect. What is the half-life of the drug? These parameters are critical to assess the MOA of the anti-ASC used in these studies.

Response: To assess the pharmacokinetics of the anti-ASC antibody, we determined its titers in serum by ELISA at various time points up to 96 hpi after the first injection. The anti-ASC antibody serum levels peaked at 24 hpi and declined to about half maximal serum concentration levels at 96 hpi. This serum half-life, for the injected concentration, is in the range of reported kinetic parameters of engineered monoclonal antibodies (e.g., Unverdorben et al., MAbs, 2016; Foss et al., Nat Comm, 2024) (Fig. 4B). Because of the high permeability of splenic red pulp vasculatures, and because of the absence of any selectively permeable barrier, efficacious imbibement of the splenic extracellular space can be plausibly expected. Theoretically, one could perfuse mice intracardially with PBS and then measure antibody in tissue. Such measurements can work relatively well in the brain, which possesses a highly impermeable barrier. However, here we would find it difficult to convince ourselves that such measurements would not be contaminated by residual blood in splenic capillaries that may be difficult to clean up through perfusion. Therefore, we did not measure the antibody levels in the spleen.

Comment 5: The authors should expand the discussion section to include the work of other groups that have successfully employed anti-ASC antibodies. For example, PMID: 35793783, PMID: 32366256

Response: We thank the referee for pointing out that literature. We extended the discussion section accordingly and added these important references into the discussion.

Comment 6: Methods: The authors provide the number of animals employed in the Supplemental Tables 5 and 7. These numbers should be provided in the methods section or in the Figure legends. Additionally, how many replicates were performed for the data in Figure 2?

Response: __As suggested by the reviewer we now provide the number of animals in the figure legends of main __Fig. 2 and Fig. 3 __in addition to those in Table 5 and Supp Table 7__ to enhance clarity.

Referee 3____

General: The manuscript by Losa et al. explores the co-aggregation of ASC with serum amyloid A (SAA) in vivo and in mouse models, It posits that, similar to Amyloid beta, SAA is cross-seeded by ASC foci both in vitro and in vivo. This review only addresses the co-localization and in vitro cross seeding data (Figs. 1 and 2A, B), not the mouse experiments or mass spectrometry data. The manuscript first shows co-deposition of ASC with SAA amyloid. SAA was stained both with Congo red and ThS, both standard dyes for amyloid staining. Figure S2 shows CR birefringence, the hallmark of amyloid deposits. The authors then move to demonstrate co-localization of SAA and ASC in confocal and STED immuno-fluorescence microscopy.

Significance: The discovery of the role of ASC in Alzheimer's disease generated an exciting new hypothesis to the etiology of sporadic AD, for which the cause is unknown. The current manuscript finds that ASC may also play a role in AA amyloidosis, which is a significant finding.

Comment 1: Confocal images C-E show overlapping staining of markers for both SAA and ASC. Similarly, STED images show co-aggregation of ASC and SAA in amyloidosis patients. However, since confocal images F and G seem to show overlapping staining of the yellow and magenta channels as well, a careful quantitative analysis of the data I needed. Quantify co-localization (Pearson coefficient) in confocal and STED images. STED images from control patients are missing and need to be included.

Response: AA amyloidosis is a relatively rare disease, and tissue samples thereof are even rarer. We only had access to the samples of one patient in both control and SAA groups. This limitation prevented us from conducting quantitative analyses. Rather than looking at the Pearson – or, possibly better, Spearman – correlation coefficient, we opted for an unbiased method of correlation in which we reconstructed the picture using 3D surface rendering with the Imaris software (see Fig. 1). From this reconstruction, we exported the barycenter of each surface on a 3D plot for both SAA and ASC markers (see Fig. S2B-C). Each point represents the center of a surface, while the box plots on the sides represent the distribution of the markers in space, demonstrating the overlap of the markers for ASC and SAA. We also understand the suggestion to conduct STED imaging on control samples to show the absence of co-aggregation. However, we could not be sure of which region to capture and how to decide on the focus, as we did not detect strong signal from confocal images of the control sample. Imaging blindly would almost necessarily lead to irrelevant imaging and aberrant comparison. We do not claim any quantitative data out of these images; however, we report an observation. Quantitative and mechanistic co-aggregation data are presented in Fig. 2 using LiP-MS.

Comment 2: The authors then move on to demonstrate that ASC foci can cross-seed SAA amyloid formation in vitro, by recording SAA aggregation kinetics in the presence and absence of ASC foci. Curves recorded in the presence of ASC foci have accelerated kinetics as shown by a decrease in the time to reach half-maximal fluorescence (t1/2). However, these data (Fig 2A, B) are not very clean. Only three data points out of five curves shown in panel A. are presented in the fitting of the control (yellow) aggregation kinetics in panel B. Why was this done? Panel B shows a significant difference between the control and the kinetics seeded with ASC specks. It looks doubtful that the results are still statistically significant if these data are included, so their exclusion impacts the overall conclusion of the paper. The significance of the cross-seeding results needs to be substantiated experimentally.

__Response: __The in vitro SAA aggregation assay was performed under established conditions (Claus S et al., EMBO Rep 2017) and the resulting data was processed using the AmyloFit software from the Knowles lab in Cambridge, UK (Meisl G et al., Nat Protoc 2016). The AmyloFit technology uses global fitting resulting in high-accuracy kinetics. Given the software algorithm, only curves that show a sigmoidal ThT fluorescence signal over time can be fitted. Therefore, replicates that do not show aggregation (characteristic ThT signal) over time cannot be fitted. As a result, only three out of six curves could be fitted resulting in three t1/2. Conversely, in the presence of ASC specks, all six replicates aggregated in a dose-dependent manner, and could be fitted perfectly, yielding six t1/2 values. Thus, all available data points are plotted and used for statistical analysis. Moreover, the fact that in presence of ASC specks all SAA replicates aggregated/converted successfully in a dose-dependent manner (whereas in the SAA-only condition some replicates do not aggregate) further underscores the pivotal role of ASC specks in SAA seeding, conversion, and aggregation enhancement.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

The manuscript by Losa et al. explores the co-aggregation of ASC with serum amyloid A (SAA) in vivo and in mouse models, It posits that, similar to Amyloid beta, SAA is cross-seeded by ASC foci both in vitro and in vivo. This review only addresses the co-localization and in vitro cross seeding data (Figs. 1 and 2A, B), not the mouse experiments or mass spectrometry data.

The manuscript first shows co-deposition of ASC with SAA amyloid. SAA was stained both with Congo red and ThS, both standard dyes for amyloid staining. Figure S2 shows CR birefringence, the hallmark of amyloid deposits. The authors then move to demonstrate co-localization of SAA and ASC in confocal and STED immuno-fluorescence microscopy.

Confocal images C-E show overlapping staining of markers for both SAA and ASC. Similarly, STED images show co-aggregation of ASC and SAA in amyloidosis patients. However, since confocal images F and G seem to show overlapping staining of the yellow and magenta channels as well, a careful quantitative analysis of the data I needed. Quantify co-localization (Pearson coefficient) in confocal and STED images. STED images from control patients are missing and need to be included. The authors then move on to demonstrate that ASC foci can cross-seed SAA amyloid formation in vitro, by recording SAA aggregation kinetics in the presence and absence of ASC foci. Curves recorded in the presence of ASC foci have accelerated kinetics as shown by a decrease in the time to reach half-maximal fluorescence (t1/2). However, these data (Fig 2A, B) are not very clean. Only three data points out of five curves shown in panel A. are presented in the fitting of the control (yellow) aggregation kinetics in panel B. Why was this done? Panel B shows a significant difference between the control and the kinetics seeded with ASC specks. It looks doubtful that the results are still statistically significant if these data are included, so their exclusion impacts the overall conclusion of the paper. The significance of the cross-seeding results needs to be substantiated experimentally.

Significance

The discovery of the role of ASC in Alzheimer's disease generated an exciting new hypothesis to the etiology of sporadic AD, for which the cause is unknown. The current manuscript finds that ASC may also play a role in AA amyloidosis, which is a significant finding.

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 Losa et al., investigates whether ASC is involved in serum AA amyloidosis. The authors report that ASC colocalizes with SAA in human AA amyloidosis and that purified ASC specks accelerate SAA fibril formation in vitro. In addition, splenic AA amyloid was decreased in Pycard-/- mice compared to Pycard+/+ mice and that treatment with anti-ASC antibodies decreased amyloid loads in Pycard+/+ mice. Lastly, they analyzed serum of 19,334 patients to show that the prevalence of anti-ASC antibodies did not correlate with any specific disease. The authors conclude that ASC to play a role in extraneural proteinopathies of humans and experimental animals and suggest that anti-ASC immunotherapy may contribute to resolving such diseases. The findings in the study are novel and demonstrate a new role for ASC in aggregation proteinopathies. However, there are number of issues that need to be addressed before acceptance for publication.

Major Points:

Figure 3 E depicts Western blots of monomeric SAA in spleen of Pycard+/+ and Pycard-/- mice. The authors should include immunoblots depicting the levels of ASC in these tissues and to demonstrate that the Pycard-/- mice lack ASC. Fig. 3B shows that at 96 hours after injection there was no difference in SAA serum concentration. How do the authors explain this drop in SAA serum concentration? No explanation is provided.

Figure 4 shows anti-ASC administration reduces amyloid load. The immunoblot in Figure 4C does not represent the quantification of the blot. In fact, there are only 3 samples per treatment group whereas the quantification shows 5-6 animals per group. Additionally, the authors have not shown that the drug penetrates the target tissue and how much drug is present in spleen to provide a therapeutic effect. What is the half-life of the drug? These parameters are critical to assess the MOA of the anti-ASC used in these studies.

The authors should expand the discussion section to include the work of other groups that have successfully employed anti-ASC antibodies. For example, PMID: 35793783, PMID: 32366256

Methods: The authors provide the number of animals employed in the Supplemental Tables 5 and 7. These numbers should be provided in the methods section or in the Figure legends. Additionally, how many replicates were performed for the data in Figure 2?

Significance

The findings in the study are novel and demonstrate a new role for ASC in aggregation proteinopathies. This study reports a crucial role for ASC in SAA interaction and recruitment, SAA serum level modulation, SAA fibril formation acceleration, and controlling the extent of inflammation associated amyloidosis with respect to AA amyloid deposition

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, et al., investigates the role of the inflammasome adapter ASC (in AA amyloidosis). This condition involves the aggregation of serum amyloid A (SAA) and is linked to chronic inflammation.

Firstly, I can directly say that I do recommend this study for publication. This is a well conducted and well-written study which advances the knowledge on IL-1-independent inflammatory functions of inflammasomes. Furthermore, I find it particularly impressive that despite the inflammasome research community is well aware that amyloidosis is a hallmark of inflammatory diseases, it took a neuroscientist specialized in prion diseases to raise the question whether ASC would be involved in seeding serum AA aggregation.

Key findings include:

• ASC forms extracellular aggregates that enhance SAA aggregation, as observed through superresolution microscopy.
• In a mouse model, the absence of ASC significantly reduced amyloid load, not due to increased phagocytosis but likely due to diminished aggregation.
• Treatment with anti-ASC antibodies reduced amyloid load and mitigated weight loss in mice with AA amyloidosis.

These findings suggest that ASC plays a crucial role in AA amyloidosis and that targeting ASC could be a potential therapeutic strategy. The study expands our understanding of the involvement of ASC in proteinopathies beyond neural diseases, pointing to its role in systemic conditions like AA amyloidosis. Main Comments: Overall, the experiments are well-conducted and mostly all controls I would expect were included. With few exceptions, the data is convincing. With that said, I have issues with some of the staining employed in Fig 1.

In Fig. 1, the authors assess ASC staining in cardiac tissues from a patient with vasculitis and systemic inflammation-related AA amyloidosis, and a control patient who died of a heart attack but had no signs of amyloidosis. However, most of the data shown is related to the AL177 anti-ASC. More importantly, no isotype stainings are included. We have previously demonstrated that the AL177 anti-ASC, used here, reacts quite strongly with ASC−/− cells, and it is one of the less specific anti-ASC commercially available (PMID: 27221487). As this is data from one patient (understandably), I wonder if the authors could counterstain ASC in the same samples using a specific human anti-ASC with a different color (ex: Biolegend HASC), and confirm that the signal overlays with the AL-177.

Finally, in Figure 1H it seens from the description that another anti-ASC was used: "referred in the legend as ASC (MAB ASC, Yellow)". Is this a monoclonal anti-ASC? Also, the images show large and bright antibody aggregates (middle of the image, top left corner behind the "H", and a massive fluorescence in the bottom right of the image), indicating the presence of staining artifacts. Again, no counterstaining with isotype controls are shown.

Overall, although I don't dispute the possibility that ASC would co-localize with SAA deposits, I don't think the data presented can safely sustain that claim. I would, therefore, suggest that alternative methods to be employed to substantiate these conclusions: Supposedly, would it be possible to immuno-precipitate (IP) amyloid SAA and assess ASC via western blotting? As well as IP ASC and detect SAA? Or use DSS-crosslinking to find ASC oligomers in tissue areas rich in SAA?

Minor comments:

In addition to these main comments, some minor adjustments are recommended:

For example, it would be reasonable to quantify the results in Figure 3G and providing clarification regarding the controls in the figure legend. Though there is significantly less SAA in spleen homogenates from Asc−/−, there also seems to be the case for b-actin in Fig 3G. Moreover, in the figure legend the authors state: "...Spleen homogenate from untreated (-ctrl) and AA+ (+ctrl) C57BL/6 wt mice from an independent experiment served as negative and positive control, respectively." I don't know what the authors mean with that. Is this a montage, or samples from different experiments were run together in one blot? And if so, for what reason? This is confusing and should be clarified.

Furthermore, in the Abstract, a slight rephrasing is suggested to accurately describe ASC specks as molecular aggregates formed inside cells, which are subsequently released into the extracellular space.

Lastly, enhancing the text size in figures, particularly in Fig 3, is advised to improve legibility and overall clarity.

Significance

In conclusion, this manuscript offers valuable insights into the role of ASC in AA amyloidosis, presenting compelling findings that support its potential as a therapeutic target. Addressing the mentioned concerns and making the suggested revisions will further enhance the manuscript's scientific rigor and impact. Overall, this study is a valuable contribution to the field of inflammasome research and its relevance in systemic conditions like AA amyloidosis.

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 would like to thank the reviewers for their attentive reading of our manuscript. We appreciate all the comments and suggestions. We have addressed all the concerns and have included point-by-point responses.

Reviewer #1

Evidence, reproducibility and clarity

• *

Summary:

* Cacioppo et al perform a meta-analysis of public omics data examining AURKA protein and mRNA expression (including mRNA isoforms with alternative cleavage and polyadenylation), and hsa-let-7a miRNA (shown to target AURKA mRNA) in multiple cancer types from The Cancer Genome Atlas. They conclude AURKA mRNA and protein expression may be discordant in cancer in part due to the interplay between alternative polyadenylation and hsa-let-7a miRNA.

Major comments:*

* 1) Unfortunately, there is a major flaw in the TCGA AURKA protein quantification data that underpins much of this study. Following the protein data trail (via https://docs.gdc.cancer.gov/Data/Introduction and its dependents), it appears to rely on the CST anti-AURKA #14475 which is raised to an antigen around Pro70.*

Response: We believe the reviewer refers to work from Bertolin et al. 2018 paper (https://doi.org/10.7554/eLife.38111.001) that describes the appearance of truncated versions of AURKA in mitochondrial fractions of cell extracts and shows they depend upon the presence of PMPCB mitochondrial matrix peptidase. We are not familiar with any other literature describing this phenomenon. In our own hands we find AURKA present in the mitochondrial fraction, but the protein is mostly full-length (Grant et al. 2018, https://doi.org/10.1098/rsob.170272). In both papers the mitochondrial pool is small relative to the total cellular pool of AURKA. In fact, this mitochondrial pool is so difficult to detect in intact cells that it has not been reported by other labs and is not universally acknowledged. Given the small size of the mitochondrial pool, any increased amounts of mitochondrial AURKA in cancers, it would be unlikely to significantly impact the measured total protein levels.

2) Following the flaws identified in the protein foundation data, the study would then benefit from some post-validation of findings with actual biological data derived from their own independent assessment of the cancers being examined.

• *

Response: The literature thoroughly reports empirical evidence on AURKA protein expression levels in the cancers analysed in this study, therefore we don't believe our own post-validation of findings would add any novelty in this sense.

Minor comments:

* 1) All of the Correlation analysis have been tested for statistical significance and these results are available in the supplementary data. However, I think it would be useful if these statistics were also included in the main figures themselves. (Figures 1B, 2B and 2C) A low correlation that is statistically significant is a more powerful statement.*

Response: We agree, and plan to add the results of the statistical analyses in the Figures 1B, 2B and 2C.

2) In the materials and methods, Correlation is separated into distinct degrees: none to very strong, but apart from some lines on the graphs, these degrees of correlation strength are never revisited, so they should be included. Perhaps there is a biological difference between AURKA post transcriptional regulation and protein levels with different R score strength?

Response: We believe that reiterating a discussion on the degrees of correlation strength in the main text would appear repetitive. We do however plan to add a sentence to appropriate points in the main text to redirect the reader to the materials and methods section for information on the distinct degrees of correlation.

3) In Figure 2D a clustering analysis was performed to show the possible relationships between hsa-let-7a and protein levels. The current visualization is hard to understand. A 3D graph with Protein, mRNA and has-let-7a axis's would be easier to follow. I believe it would also be beneficial to do something similar including the APA data as this is the area that the paper lacks depth.

• *

Response: We agree that 3D graphs could aid visualization and plan to provide a link to an interactive 3D view of our analysis.

* 4) Figure 3B and 3C, can you apply a statistical test on the SLR ratios given the magnitude difference between CCND1 and AURKA SLRs?*

• *

Response: Since the values of AURKA and CCND1 SLRs are not always coming from the same dataset and are therefore not matched for patients, we believe it would not be appropriate to make comparisons applying statistical tests.

* 5) Even though the paper does not claim to provide a unifying hypothesis for APA/has-let-7a regulation of AURKA, I think a more in depth look at the data would be useful. The discussion starts off well when describing what was found with the analysis, but as is, is mostly a re-statement of the results without added insight.*

Response: We agree that more in depth analysis of more data would be useful in strengthening conclusions. However, given the variability in interplay between APA and hsa-let-7a we describe, it is well beyond the scope of this study (or the extent of TCGA database) to come up with a unifying hypothesis.

Significance

• *

The study is novel in attempting to show additional layers of AURKA regulation that hadn't been previously investigated. Furthermore, factors controlling AURKA expression are of broad interest. Overall, I would like to say this is an interesting investigation into AURKA mRNA expression in cancers. In our opinion the choice of bioinformatic tools is appropriate and well controlled.*

General Assessment: As noted in the major comments, a major weakness is the reliance on a flawed measure of AURKA protein levels from the foundation dataset. Thus, the study needs to be repeated using an alternative MS derived dataset to accurately quantify total AURKA protein levels. This would greatly improve the study and subsequent claims.

Advance: The study has potential to extend knowledge in the field in a conceptual way, predicting the complex interplay of factors that regulate AURKA mRNA processing and translation.

Audience: Currently the paper is only fully accessible a specialized bioinformatician audience but the topic (factors controlling AURKA expression) has a broad interest in many fields not limited to just cancer but also development and other non-cancer diseases.*

* This review was jointly completed by a mouse model of human disease AURKA biologist with 24 years' experience, and a bioinformatician.*

• *

Reviewer #2

Evidence, reproducibility and clarity

In the manuscript "Post-transcriptional control drives Aurora kinase A expression in human cancers", authors Cacioppo, Lindon and colleagues analyze publicly available data on transcript and protein levels for many cancer types to determine correlations between transcript and protein levels for Aurora A and the microRNA hsa-let-7a. This study builds on a recent publication from their lab where they show that different polyadenylation isoforms of the Aurora A transcript in triple negative breast cancer correlate with patient survival and affect protein abundance. In this study, they aim to extend this analysis to 18 different cancer types to determine if posttranscriptional regulation potentially plays a role in Aurora A protein abundance. The authors find that for certain cancer types, Aurora A protein abundance does not correlate with mRNA abundance, suggesting that posttranscriptional regulation may be responsible for differences in protein expression in these cancer types. Furthermore, they find negative correlations between expression of hsa-let-7a and mRNA and protein abundance in certain cancer types, implicating this microRNA as a potential regulator of Aurora A mRNA stability.*

Major comments:

1. The biggest issue that I have with this analysis relates to the assumption that Aurora A levels will be meaningfully different between individual tumors in all cancer types. For some cancers, the lack of a correlation between mRNA and protein levels for Aurora A could simply be because Aurora A overexpression is not a feature of that cancer type. Looking at the data, the cancer types where they see little-to-no correlation are the cancer types where none of the tumors have high levels of Aurora A mRNA or protein. Therefore, the lack of correlation is likely because differences in protein levels result from noise in the measurements rather than posttranscriptional regulation. Since the lack of correlation between protein and mRNA in these cancer types is the main evidence for the primary conclusion in the paper that "AURKA mRNA and protein expression are often discordant in cancer as a result of dynamic post-transcriptional regulation", I don't think that this conclusion is supported by the data. If anything, the data seems to show that substantial changes in Aurora A protein levels are almost always accompanied by a corresponding change in mRNA levels.

To address this issue, the authors could look at the variability in Aurora A protein levels for each cancer type, and then focus their correlation analyses on cancer types where overexpression of Aurora A is a feature.*

Response: We thank the reviewer for this thoughtful comment. We decided not to consider data on AURKA protein levels between healthy and tumour samples because of the lack of proteomic datasets of matching normal tissues for all cancers (except BRCA) in the TCGA database. For this reason, it cannot be excluded that the tumours where we see little-to-no protein-mRNA correlation have in fact high levels of AURKA protein. Indeed, the literature reports wide empirical evidence that AURKA protein is overexpressed in the cancer tissues where we see little-to-no protein-mRNA correlation (Thyroid cancer: Zhao et al, Cell Biosci, 2022; Jingtai et al, Cell Death Dis, 2023. Prostate cancer: Das et al, Pathol, 2010; Chun Yu Lee et al, Cancer Res, 2006. Kidney cancers: Wen et al, Heliyon, 2024; Li et al, Cell Death Dis, 2022. No evidence available for PCPG). Therefore, we believe that is reasonable to propose that in these cancers, which according to our analysis of TCGA data only show minor or no increase in AURKA mRNA expression compared to the normal tissue, lack of correlation is because of post-transcriptional regulation.

2. The statistical significance of the analyses is often unclear. For the correlations between Aurora A protein levels and hsa-let-7a, authors mention that two cancers have a correlation with "statistical significance", but I cannot find any indication of how that was determined, and it is not shown in the corresponding figure (2C). The only time significance is indicated for a correlation is in Figure 4A. Is this the only correlation in the whole manuscript with a p-value less than .05?

Response: The results of the statistical analyses are included in the corresponding supplementary data (Sup. Fig 1, Sup. Fig. 2A-B). We plan to add them to the Figures 1B, 2B and 2C as requested by another reviewer.

3. The SLR for the Aurora A transcripts is only shown in terms of a ratio between cancer and normal tissue. Without the numbers in the absence of normalization, it is difficult to determine how meaningful this is. Is a two-fold change going from .3 to .6 or .001 to .002?

• *

Response: We plan to add a supplementary table containing the SLR values for matched normal and cancer samples in the absence of normalization.

4. Figure 5B is nearly impossible to interpret due to the extreme differences in overall transcript levels between the cancer types. The differences in scaling of the y-axis between the plots makes this even more challenging. The authors state that "It is evident that each isoform has an individual profile of expression across cancers", but this could only be determined from relative expression levels between the different isoforms instead of absolute levels.

Response: We retrieved this plot from the GEPIA2 platform without possibility of editing the y-axis. We plan to edit the text to "It is likely that each isoform has an individual profile of expression across cancers, however a measure of the relative expression levels between the different isoforms would be required".

Minor comments:*

1. In supplementary figure 3, SLR is plotted on a log scale in A and a linear scale in B.*

Response: We plan to convert the SLR scale in Sup. Fig. 3B to a log scale.

2. Figure 4D is a correlation of correlations. I don't see how to interpret this in a meaningful way.

Response: Figure 4D is not intended for quantitative analysis of correlation of correlations (no quantitative coefficients were in fact calculated), rather to visualize how the link of AURKA SLR with AURKA protein levels and that with hsa-let-7a levels can be differently associated in different cancers.

Significance

Aurora A is overexpressed in a wide variety of cancer types. This overexpression is commonly believed to result primarily from increased mRNA abundance. The identification of additional mechanisms regulating Aurora A protein levels would therefore be of interest to the field, as these regulatory mechanisms could be contributing to Aurora A's role in cancer progression.*

To some degree, the significance of the findings presented here depend on whether they convincingly demonstrate substantial post-transcriptional regulation. My interpretation of the data presented in this manuscript is that it largely supports Aurora A protein levels being extremely well correlated with mRNA levels, which is in line with previous findings.*

• *

• *

• *

Reviewer #3

Evidence, reproducibility and clarity

• *

*Aurora A misregulation at both mRNA and protein levels has been known since the 1990s to be casually associated in vivo, and strongly associated in vitro, with tumourigenesis. The study builds the case that dysregulation of Aurora A mRNA and protein levels (most previously established) are more prevalent in cancer cells than 'normal' cells, using data from TCGA, and extends this to a mechanistic explanation. It evaluates miRNA and the ratio of the two short/long ratio (SLR) isoforms of mRNA across cancer types compared to healthy controls. The work concludes that an interplay between APA (alternative polyadenylation) and hsa-let-7a miRNA (which has known tumor suppressor properties) regulation of AURKA mRNA contributes to alternative splicing, revealing a new factor explaining changes in AURKA expression in many (if not all) cancers. *

• *

*Minor points: *

• *

*1) To strengthen the study, some analysis of AURKB mRNA would be useful in the same datasets, because this is also an M-phase kinase. *

• *

Response: We carried out a specific study of AURKA (and to some extent also of the cell cycle regulator CCND1) using time-limited access to private TCGA datasets. Although we agree that investigation of AURKB would potentially enable us to strengthen some conclusions, this would be a new project that we do not currently have resources for.

*2) What happens to TPX2 or CEP192 mRNA (splicing or levels) in the same samples? For TPX2 in particular, this is described in the literature to help form the oncogenic holoenzyme, as well as dictating AURKA protein stability. *

• *

Response: Again, we like this suggestion but are not in a position to carry out analyses of TPX2 and CEP192 within the scope of this study.

• *

*3) Does an alternative AURKA splicing change G1/S to G2/M-phase roles of AURKA? I understand that mRNA is repressed by hsa- let-7a in G1 and S phases but not in G2, so how does non M-phase AURKA protein get made? This may be beyond the scope of the study at this point. *

• *

Response: Whether alternative AURKA transcripts change non-mitotic roles of AURKA is an open and intriguing question. In acknowledgement of this point raised by the reviewer, we plan to add a discussion on this in the main text: "Although there is no evidence to date that different AURKA transcripts might influence AURKA activity, instances of isoform-dependent protein localization and function are increasingly reported (Mitschka and Mayr, Nat Rev Mol Cell Biol, 2022). In a previous study, we have detected higher nuclear localization of a reporter protein under the regulation of AURKA short 3'UTR (Cacioppo et al., eLife, 2023). Therefore, there is a possibility that AURKA mRNA isoforms are targeted to different subcellular localizations to support localized translation - or that AURKA protein is co-translationally targeted to different compartments - and AURKA may be preferentially localized in the nucleus when coded by the short 3'UTR mRNA".

AURKA protein levels are maintained very low in G1 to S phase compared to G2 and M phases. At the level of translation, this is likely ensured by the absence of factors/mechanisms that activate AURKA translation (e.g., hnRNP Q1) and the presence of factors/mechanisms that repress its translation (e.g., hsa-let-7a), the combination of which results in basal translation of AURKA in G1/S until full translational activation in G2 (where a switch likely occurs whereby activating factors operate while repressing factors are disabled). However, the combination and synergy of these factors/mechanisms are likely cell type- and context-dependent.

• *

Significance

*I think the study is strong overall, and the authors are humble enough to describe the work as an exploratory analysis, which though not directly in my area of expertise (since it relies on data assembly and statistical analysis), has the right team to ask the questions and interrogate the data. It builds on a huge amount of literature and a recent study from this team showing that alternative translation is relevant to activation of AURKA, and which linked let-7a to this process. Overall, the study provides a very useful resource for other researchers, assembling a large amount of data around AURKA mRNA variants, Let-7a miRNA and coming to the conclusions that *

*1) hsa-let-7a potentially negatively controls the rate of degradation or translation of AURKA mRNA in cancer cells. *

*2) Splicing-related architecture of the 5'UTR of AURKA mRNA likely plays a role in determining the context-dependent cancer expression profile of expression. *

Overall, with some extra information around the key regulators of AURKA (TPX2 mRNA?) the work is likely to be cited and spur on future studies.

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

Aurora A misregulation at both mRNA and protein levels has been known since the 1990s to be casually associated in vivo, and strongly associated in vitro, with tumourigenesis. The study builds the case that dysregulation of Aurora A mRNA and protein levels (most previously established) are more prevalent in cancer cells than 'normal' cells, using data from TCGA, and extends this to a mechanistic explanation. It evaluates miRNA and the ratio of the two short/long ratio (SLR) isoforms of mRNA across cancer types compared to healthy controls. The work concludes that an interplay between APA (alternative polyadenylation) and hsa-let-7a miRNA (which has known tumor suppressor properties) regulation of AURKA mRNA contributes to alternative splicing, revealing a new factor explaining changes in AURKA expression in many (if not all) cancers.

Minor points:

1. To strengthen the study, some analysis of AURKB mRNA would be useful in the same datasets, because this is also an M-phase kinase.
2. What happens to TPX2 or CEP192 mRNA (splicing or levels) in the same samples? For TPX2 in particular, this is described in the literature to help form the oncogenic holoenzyme, as well as dictating AURKA protein stability
3. Does an alternative AURKA splicing change G1/S to G2/M-phase roles of AURKA? I undersgtand that mRNA is repressed by hsa- let-7a in G1 and S phases but not in G2, so how does non M-phase AURKA protein get made? This may be beyond the scope of the study at this point.

Significance

I think the study is strong overall, and the authors are humble enough to describe the work as an exploratory analysis, which though not directly in my area of expertise (since it relies on data assembly and statistical analysis), has the right team to ask the questions and interrogate the data. It builds on a huge amount of literature and a recent study from this team showing that alternative translation is relevant to activation of AURKA, and which linked let-7a to this process. Overall, the study provides a very useful resource for other researchers, assembling a large amount of data around AURKA mRNA variants, Let-7a miRNA and coming to the conclusions that

1) hsa-let-7a potentially negatively controls the rate of degradation or translation of AURKA mRNA in cancer cells.

2)Splicing-related architecture of the 5'UTR of AURKA mRNA likely plays a role in determining the context-dependent cancer expression profile of expression.

Overall, with some extra information around the key regulators of AURKA (TPX2 mRNA?) the work is likely to be cited and spur on future studies.