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:

Selective autophagy receptors (SARs) of the Sequestosome-1 like receptor group (SLRs) including SQSTM1(Sequestosome-1)/p62, NBR1, TAX1BP1, NDP52, CALCOCO1 and Optineurin are soluble SARs that engage cargo and ATG8 family proteins as well as components of the core autophagy machinery like FIP200/RBCC1 to bring about the autophagic degradation of the cargo and themselves. In the autophagic degradation of protein aggregates (aggrephagy) the most studied SAR p62 collaborates with the archetypal autophagy receptor NBR1 and also TAX1BP1 to bring about effective turnover of ubiquitinated cargos sequestered into p62 bodies or droplets by liquid-liquid phase separation. How this intricate co-operation of these SARs is orchestrated is incompletely understood. In the paper by North et al entitled "The LC3-interacting region of NBR1 is a protein interaction hub enabling optimal flux" the authors use peptide arrays to map the binding sites for ATG8-family proteins LC3A and GABARAPL1, FIP200 and TAX1BP1 to the autophagy receptor NBR1. The authors find that three short linear interaction motifs (SLiMs), the LIR, FIR and TIR interacting with ATG8 family proteins, FIP200 and TAX1BP1, respectively, partly overlap in a short region of NBR1 that can adopt different conformations to accommodate the different binding partners. In short, the different interactions are mediated by distinct overlapping determinants, rather than a single, convergent, SLiM. While the important binding determinants for ATG8 proteins and FIP200 show more overlap and it was not possible here to find mutations that distinguish LIR and FIR binding, TAX1BP1 bound more to a region downstream of the LIR and a specific mutation in NBR1 and in TAX1BP1 could abolish binding. Checking the role of phosphorylations in augmenting binding using phosphomimetic mutations it was seen that while FIP200 and Atg8-family binding were generally augmented by phosphorylation, TAX1BP1 binding did not respond to these mutations. Very interestingly, the authors found that co-expression of TAX1BP1 with tandem-tagged NBR1 in pentaKO cells (not expressing the SLRs p62, NBR1, NDP52, TAX1BP1 and OPTN) increased significantly the autophagic turnover of NBR1. None of the other SLRs could do this. Instead, this over-expression assay revealed a competition.

Major points:

In Fig 4 the peptide array binding assay is not sufficient as it is only semiquantitative. The data shown should be accompanied by a more direct binding assay allowing the determination of kDs for the binding where the WT peptides are directly compared to the phosphor mimicking mutant peptides. Here the fluorescence anisotropy assay the authors use in Suppl Fig. 1E or ITC, OctetRed96 or another assay suitable for kD determinations should be used.

As this paper is already dominated by the use of peptides it would significantly enhance the quality of the data if the authors had included studied with peptides phosphorylated at the specific positions to allow comparison with the phosphomimetic substitutions to aspartate.

The quality of the 2D peptide array probing of GST-LC3A binding in Fig 3A is poor. Is this a stripped and re-probed membrane? I do not think these data are publication quality and the experiment should be redone unless the authors have very good arguments against my suggestion. It would also be nice to see a 2D peptide array of GABARAPL1 binding too to make the comparative study complete.

For the data shown in Fig 6 it should be noted that although these are very interesting results a clear limitation of the study is that the results on the autophagic turnover is based on overexpressing the SLRs in the pentaKO cells. In a physiological setting with all relevant actors in place and with a different stoichiometry the effects could likely be different.

Minor points:

It would be beneficial for the reader to show a cartoon of the domain organization of both TAX1BP1 and NBR1 in Figure 1. NBR1 is shown in supplemental figure 1, but there is no depiction of the domain organization of TAX1BP1.

The authors say at the bottom of page 4 "Complementary in vivo studies reveal that while SLRs typically compete". But do they actually typically compete? Is this not a result of the experimental strategies employed? There is more a shortage of SLRs based on cargo competition as shown recently by Peter Kim's group that excessive pexophagy may reduce mitophagy etc. (Germain et al. 2023).

In Fig. 3D it should be shown that D, E, A and V are preferred residues at position +1 for LC3A binding.

In such a 2D mutational analysis it is often just as important to determine which residues are not allowed for binding. It would therefore be nice if the authors could summarize/visualize their results in a better way in Fig 3D to also show the residues that lead to loss of binding. These could be shown below the sequence and the use of color to distinguish basic, acidic, hydrophobic and aromatic residues could be attempted.

Line 327: To be clear about the fact that this is an overexpression assay "simultaneous expression" should be corrected to simultaneous overexpression".

There are LIRs and FIRs that overlap and those that do not. To check the degree of overlaps that may occur among known LIRs the authors made a peptide array with 100 established LIR sequences taken from the LIR-Central database (Chatzichristofi et al., 2023). The peptide array was probed with LC3A (29 bound), GABARAPL1 (49 bound), the FIP200 Claw domain (57 bound) and the TAX1BP1 CC2 domain (49 bound). As much as one third (32) of the LIR peptides were not bound by any of the four probes. Do the authors have a good explanation for the fact that so many peptides did not bind? Strangely enough, the NBR1 peptide used in Figure 2A did not bind any of the probes while the NBR1 peptides used in Fig. 1C bound very well. Do the authors have any explanation for this?

Significance

I found this paper very interesting to read with a lot of interesting new detailed and useful information on binding specificity for the proteins and motifs involved. It is a generally well performed study with interesting results. I also very much enjoyed the Discussion section which opens up for several interesting possible scenarios. The study also produced important point mutants that can be used in future studies to selectively abolish TAX1BP1 binding to NBR1. I think this is a "must read" paper for researchers interested in selective autophagy and co-operation between SARs, and more generally for getting some insight into how SLiMs may work. As such, this paper will be of interest for all interested in autophagy research and for a wider audience too as it is in essence about how overlapping SLiMs may be employed to orchestrate multiple protein-protein interactions using distinct overlapping determinants, rather than a single, convergent, SLiM. It is also one of the very few papers I have come across exploiting the power of the peptide array method so extensively with success for mapping protein binding sites. It could perhaps be interesting if the authors discussed their results in relation to another study from the group of Sascha Martens on the role of TAX1BP1 in p62 bodies or condensates (doi: https://doi.org/10.1101/2024.05.17.594671). These two papers should be read together as they are both very interesting and important contributions.

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

Learn more at Review Commons

Reply to the reviewers

We thank the reviewers for their positive and constructive criticism. We answer their points one by one below.

Reviewer #1

1.) In the baf-1 G12T mutants the authors find reduced levels of lamin in hypodermal nuclei. It would be good to also examine the dynamics of lamin in the second tissue that was subjected to DamID (intestinal cells).

We provide a complete analysis of GFP::LMN-1 and EMR-1::mCh in control and baf-1(G12T) day 1 adults in intestine and hypodermis and at 20°C and 25°C. These data demonstrate that GFP::LMN-1 expression is reduced in baf-1(G12T) mutants in both tissue and at both temperatures. In contrast, for EMR-1::mCh a significant reduction was only observed in hypodermal nuclei at 20°C.

The effects on GFP::LMN-1 and EMR-1::mCh in the hypodermis 20°C were reported in Figure 2E-F in the original version of our manuscript. We have moved these data to the new Supplementary Figure S5 and represent instead the data obtained for hypodermis at 25°C in Figure 2E-F for consistency with the data represented in Figure 2A-D. Data on intestine for both markers and both temperatures are also included in the new Supplementary Figure S5.

We have modified the text as follows:

“To test the impact of baf-1(G12T) on LMN-1, EMR-1, and BAF-1 localization in vivo, we quantified these factors at the NE of hypodermal and intestinal cells. We observed a significantly lower median GFP::LMN-1 signal at the NE in baf-1(G12T) mutants in both tissues at 20°C and 25°C (Figure ____2E; Supplementary Figure S____5A-C). In contrast, accumulation of EMR-1 at the NE was unaffected by the baf-1(G12T) mutation in both tissues at 25°C and reduced in the hypodermis at 20°C (Figure ____2F; Supplementary Figure S____5D-F). In human NGPS cells, emerin was observed to be delocalized to the ER (Janssen et al., 2022; Puente et al., 2011), but we detected no increase in cytoplasmic EMR-1::mCh signal in the mutant, indicating that this NGPS phenotype is not present in the C. elegans model. In agreement with these microscopy data, analysis of whole-worm mRNA levels by quantitative RT-PCR also revealed a significant reduction in lmn-1 expression whereas emr-1 was unaffected (Supplementary Figure S4E-F).”

2.) The authors make a statement that EMR-1 expression was reduced in the baf-1 G12T mutant, but do not comment on LMN-1 expression. Can a statement on this be made by RT PCR?

Our gene expression analysis by RAPID determined a significant reduction in emr-1 expression in the intestine of baf-1(G12T) mutants, using a fold change of 2 as threshold. In contrast, expression of emr-1 in hypodermis as well as baf-1 and lmn-1 expression in both tissues were not significantly different between wild type and baf-1(G12T) mutants in our RAPID data.

We performed qRT-PCR on bulk mRNA to compare the expression of baf-1, emr-1 and lmn-1 in control versus baf-1(G12T) mutants. No differences were detected for baf-1 and emr-1 (new Supplementary Figure S4E-F). Considering that the qRT-PCR is on bulk mRNA, the emr-1 result is compatible with the RAPID data that suggest deregulation of emr-1 only in intestine and unaffected expression in the hypodermis. For baf-1 there is agreement between qRT-PCR and RAPID data from both tissues (no difference in the mutant). For lmn-1, the qRT-PCR analysis suggests a modest reduction (23%; not reaching the threshold applied in the RAPID analysis) in baf-1(G12T) mutants, which is concordant with the reduction observed in GFP::LMN-1 intensity in hypodermis and intestine by confocal microscopy (e.g. 14% reduction in median GFP::LMN-1 intensity in hypodermis at 25C; Figure 2E).

The discordance between RAPID and live imaging for emr-1/EMR-1::mCh (a reduction in the intestine or the hypodermis according to RAPID or live imaging, respectively) is not surprising. Although mRNA and protein levels in general correlate well, often, variation in transcription can only explain We have added these two sentences to the manuscript:

“In agreement with these microscopy data, analysis of whole-worm mRNA levels by quantitative RT-PCR also revealed a significant reduction in lmn-1 expression whereas emr-1 was unaffected (Supplementary Figure S4E-F).”

“As described above, the amount of endogenously tagged EMR-1::mCh at the NE of intestinal cells was normal in baf-1(G12T) mutants (Supplementary Figure S5F), suggesting a cellular capacity to buffer the downregulation of emr-1 transcription (Vogel & Marcotte, 2012).”

3.) The authors find few alterations in gene regulation of the loci which have different occupancy WT BAF-1 versus BAF-1 G12T. It was surprising to see the DamID and RNA polymerase DamID experiments be done with worms grown at 20°C, because the more penetrant phenotypes at the organismal level were observed at 25°C. Could this be the reason for the little change of chromatin occupancy of BAF-1 and BAF-1 G12T or few changes in gene expression? Would it make sense to examine the expression of some selected BAF-1 bound loci by single molecule Fish at 25°C and compare expression wt versus baf-1 G12T?

We performed the DamID experiments at 20°C to avoid potential artifacts and/or toxicity by higher expression levels of Dam fusion proteins (Greil, Moorman, & van Steensel, 2006; Schuster et al., 2010). We note that altered UV and tert-butyl hydroperoxide was observed at 20°C, indicating that the baf-1(G12T) allele affects physiology at several temperatures. The original version of our manuscript described the expression of fluorescently tagged LMN-1 and EMR-1 in the hypodermis at 20°C (Figure 2E-F). As described above, in the revised version, we report the expression in the intestine at 20°C and in both tissues at 25°C. For GFP::LMN-1, a similar reduction in the baf-1(G12T) mutant was observed at the two temperatures in both tissues, whereas for EMR-1::mCh a reduction was only seen in the hypodermis at 20°C. Taken together, we conclude that 20°C is a suitable temperature for the DamID experiments.

We appreciate the suggestion to study expression of genes bound by BAF-1 by smFISH. However, we anticipate that because the hypodermis is composed mostly of large syncytia covering the round body of the animal, smFISH would be difficult to quantify. Regarding loci with different occupancy of WT BAF-1 versus BAF-1(G12T), the emr-1 locus was bound in the intestine by Dam::BAF-1 but not by Dam::BAF-1(G12T) (Figure 6B). As mentioned above, we observed that emr-1 expression was reduced in intestine of baf-1(G12T) mutants, suggesting that BAF-1 binding has a positive effect of transcription of this locus.

4.) The finding that BAF-1 nuclear envelope localization remains unchanged in the mutant stems from detection of the inserted GFP epitope. Given that the tag has an influence on the BAF-1 G12 12T mutant viability, this statement should be phrased with more care. The tag could influence the turnover of the protein for example. Maybe Western blots comparing the signal of WT BAF-1 worms and BAF-1 G12T mutant worms would be instructive to compare the levels of the protein (at 20oC and at 25oC, day 1 adults and day 8 adults).

We performed Western blot experiments to address this. As controls, we included strains expressing equal amounts of GFP::BAF-1 and GFP::BAF-1(G12T) strains (Figure 3E and Supplementary Figure 7 in original manuscript reported equal expression of the two proteins). Surprisingly, the polyclonal anti-BAF-1 serum raised against recombinant, full-length wild type BAF-1 (Gorjanacz et al., 2007) has significantly lower affinity for mutant GFP::BAF-1(G12T) than for GFP::BAF-1, which precludes the evaluation of untagged proteins:

Figure 1. [png file provided to reviewers - not possible to include here for technical reasons] Western blot analyses with anti-BAF-1 serum (Gorjanacz et al, 2007). (A) Embryonic extracts. A band of the expected size is observed in wildtype embryos (*), but not in baf-1(G12T) embryos. (B) Extracts from young adults. A faint band of the expected size is observed in wildtype embryos (* in lane 1; longer exposure is shown below), whereas a more prominent band is present corresponding to endogenously tagged GFP::BAF-1 (** in lane 2). The intensity of the potential GFP::BAF-1(G12T) is reduced by >80% (lane 4; >90% reduction was observed in a second experiment).

We point out in the revised manuscript that the conclusion on equal BAF-1 and BAF-1(G12T) expression was based on endogenously tagged proteins: “Quantifying the intensity at the NE or in the nucleoplasm of hypodermal cells did not demonstrate any difference between endogenously GFP-tagged wild-type and mutant BAF-1 (Figure 3E). A small reduction in cytoplasmic signal was observed for BAF-1(G12T), however, no difference was detected in the ratio between nucleoplasmic/cytoplasmic signal (Figure 3E). Quantitative RT-PCR analysis of whole-worm RNA samples also indicated that baf-1 and baf-1(G12T) are expressed at identical levels (Supplementary Figure S4E-F).”

5.) Line 105: typo: remove "s"

Corrected.

6.) Line 154: A conclusion is missing for the fog-2 experiment.

We have modified the text as follows: “To test this possibility, we incubated baf-1(G12T) males with fog-2(q71) feminized worms that only produce oocytes and counted daily offspring. At 25°C, the fog-2(q71) allele prevents spermatogenesis specifically in XX hermaphrodites whereas X0 males are unaffected (Schedl & Kimble, 1988). We observed a reduction in brood size of approximately one third when sperm came from baf-1(G12T) males (Supplementary Figure S2B, C). Thus, we concluded that the baf-1(G12T) mutation has a negative impact on spermatogenesis. The male/female ratio in the progeny was ~1, suggesting that meiotic segregation of chromosomes was normal in baf-1(G12T) males.”

7). Would it make sense to discuss a possible influence of altered lamin binding to the nuclear envelope in the mutant in the context of the gene expression results?

We agree that this point is relevant, and we have added the following text to the Discussion: “At current, we can only speculate about how the NGPS mutation might affect gene expression. Proteomics analyses indicate that BAF interacts with several histones and transcription factors (Montes de Oca, Shoemaker, Gucek, Cole, & Wilson, 2009), and the differences between BAF-1 and BAF-1(G12T)’s chromatin binding profiles reported here might be accompanied by changes in the association of chromatin factors at the deregulated loci. A particularly interesting candidate is GCL-1/germ cell-less 1, a repressive factor involved in spermatogenesis (Holaska, Lee, Kowalski, & Wilson, 2003). Moreover, it is plausible that the diminished recruitment of LMN-1 to the NE in baf-1(G12T) mutants modifies its interaction with the genome and with chromatin factors.”

8). In a nutshell, the authors have established a convincing accessible model system for studying aging, ready for consecutive testing interventions to reduce the pace of premature aging.

We appreciate and share the opinion of the reviewer.

Reviewer #2

1). The value of this work is two-fold: First, it is a very robust characterization of NGPS worms. Second, this will be a very useful model for the study of NGPS. Overall, the study is well-designed, technically strong, and the results are carefully and thoughtfully interpreted, which is nicely exemplified by the discussion of the relatively small number of genes which are differentially bound by BAF1 and are also differentially expressed and the authors do a good job of not overinterpreting the data, but simply state them. The results are convincing and informative.

We thank the reviewer for her/his positive evaluation.

My only minor point that may make this paper marginally better is that it would be nice to have a paragraph in the Discussing elaborating on the potential and the limitations of using the worm model to understand human NGPS, for example, humans have multiple lamin proteins etc.

We agree with the reviewer and have added the following text to the Discussion: “We note that the simplicity of invertebrates also implies certain limitations. For instance, while both human and C. elegans genomes contain a single BAF gene, humans, but not C. elegans, express multiple lamin isoforms in tissue-specific ratios that regulate chromatin organization and nuclear mechanics (Swift et al., 2013). Thus, C. elegans is not suitable to explore potential differences in how wild type and NGPS BAF interacts differently with the various lamin isoforms.”

Reviewer #3

1). Overall, this manuscript strongly supports the major conclusion that this C. elegans line is a powerful model for human NGPS that complements a previously reported Drosophila model. Equally importantly, from the viewpoint of fundamental discovery, this manuscript also reports major advances in understanding how BAF influences gene expression at the molecular level.

We thank the reviewer for her/his positive evaluation.

2). DamID-Baf-1 access to chromatin was unaffected by the G12T mutation (Fig. S7), but they successfully identified subsets of genes 'occupied' by baf-1 in specific cell types, some of which were significantly affected in opposite ways by the NGPS mutation (Fig. 4, Fig. 5). However, these important new results are described too briefly, and discussion is inadequate. E.g., in hypodermal cells, the baf-1 G12T mutation dysregulated genes encoding proteins in five categories (ribosomal, proton transport, cuticle components, cell surface, lysine acetylation), by downregulating genes in three categories (ribosomal, proton transport, histone acetylation) and upregulating three other categories (cuticle components, cell surface, apical region). In intestinal cells, the mutation dysregulated genes in 8 categories (ribosomal, response to X-ray, proton transport, proteasome binding, mitochondrial protein import, endopeptidase activity, carboxy-lyase activity, ATP generation), by downregulating genes in 5 categories (ribosomal, proton transport, peptidase activity, NAD binding, metal cluster binding) and upregulating 3 categories (ribosomal, response to external stimulation, histone acetylation). Opposite results for "ribosomal genes" is confusing. Examples of genes in each affected category are shown in Fig. 6. To fully interpret this data, and address apparently-conflicting results, further analysis is needed to determine if any affected groups of genes have shared regulators. For example, Fig 5E shows "ribosomal protein genes" are both up- and down-regulated by the mutation. The authors should consider: (a) WHICH ribosomal genes are in each category, and (b) does either group of genes have known regulators that might be differentially affected by the baf-1 mutation? Similar consideration of other sets of differentially-affected genes might provide novel insight into specific chromatin-regulatory proteins (e.g., potential baf-1 partners; see next paragraph) affected by the NGPS mutation.

At first it may seem confusing that some ribosomal genes are downregulated while others are upregulated. However, the baf-1(G12T) mutant represents a disease situation and not a process of natural selection where one might expect “meaningful” groups of up- and down-regulated genes. We have looked closer at the individual deregulated ribosomal genes and found genes encoding structural components of large ribosomal subunits that are either upregulated (rpl-10, rpl-29, rpl-36) or downregulated (rpl-1, rpl-3, rpl-30) in the intestine. Although these opposite behaviors might seem confusing, we propose that they reflect deregulation of ribosome biosynthesis, which is in concordance with the observations in NGPS fibroblasts (Breusegem et al., 2022). We agree that it will be important to investigate how the NGPS mutation induces these oppositely directed effects on gene expression. We found a significant higher association of the 13 deregulated ribosomal genes to BAF-1(G12) than to BAF-1 in the intestine, but we believe it goes beyond the scope of this manuscript to focus on the underlying mechanisms.

3). The current manuscript is too strictly focused on establishing C. elegans as a model for NGPS, and neglects the novel discoveries. The authors did not consider or discuss HOW a baf-1 mutation might cause such complex gene expression outcomes- given that baf-1 binds dsDNA nonspecifically. One plausible molecular explanation is that the NGPS mutation might affect baf-1 interactions with: (a) transcription factors (Requiem, RBBP4, DDB1) or chromatin-regulators (PARP1; UV-regulated interactions with DDB2 and CUL4A) identified as BAF-associated in a proteomic study (Montes de Oca et al., 2009), or (b) histone modifiers such as SET/I2PP2A (blocks H3 dephosphorylation) or H3K9 methyltransferase 'G9a' (Montes de Oca et al., 2011), or (c) other regulators that control affected genes identified in this manuscript.

We agree that this point is very relevant, but at this point we do not have experimental support for any of these possibilities. As indicated in the response to Reviewer #2, we have added the following text to the Discussion: “At current, we can only speculate about how the NGPS mutation might affect gene expression. Proteomics analyses indicate that BAF interacts with several histones and transcription factors (Montes de Oca et al., 2009), and the differences between BAF-1 and BAF-1(G12T)’s chromatin binding profiles reported here might be accompanied by changes in the association of chromatin factors at the deregulated loci. A particularly interesting candidate is GCL-1/germ cell-less 1, a repressive factor involved in spermatogenesis (Holaska et al., 2003). Moreover, it is plausible that the diminished recruitment of LMN-1 to the NE in baf-1(G12T) mutants modifies its interaction with the genome and with chromatin factors.”

4). Figures 1, 2, 4, 5: the graphs in Fig 1A,B,D-F and Fig 2B,D and the colorscales in Fig 4F and Fig 5E are uninterpretable when printed in black-and-white. Please fix Figs 1 and 2 using black/light-gray/white/stippled for bar graphs, and black/light-gray/solid/dotted/dashed for line graphs. Fig 2B can be fixed by direct-labeling of class numbers within each bar (instead of 'color-coding' separately).

We thank the reviewer for this suggestion. We have modified the figures to enable better visualization when printed in BW.

5). Revise abstract lines 40-42 ("suggesting a direct relationship between BAF-1 binding [to what?] and gene expression") to reflect the deeper analysis.

We have rephrased this sentence, so it now reads: “Most genes deregulated by the baf-1(G12T) mutation were characterized by a change in BAF-1 association, suggesting a direct relation between association of a gene to BAF-1 and its expression.” However, we prefer to not extend into speculations in the abstract because of lack of experimental evidence.

6). Lines 132-155 (Figure 1): The impact on sperm production suggests the NGPS mutation might affect association with Germ cell-less (GCL), a transcription repressor that competes with BAF for binding to emerin in mammalian cells (Holaska et al., 2003 JBC).

This is indeed an interesting possibility and we have incorporated it into to Discussion (see answer to point 3 above).

7). Lines 151-154: Did not understand the fog-2 'feminized worm' experiments. Please briefly explain for non-worm experts.

Please see our response to Reviewer #1’s point 6.

8). Line 190: Clarify that nuclear shapes were categorized manually by single-blind observer.

We have amended the text: “Nuclei were manually classified by single-blind observer based on their morphology as previously described (Perez-Jimenez, Rodriguez-Palero, Rodenas, Askjaer, & Munoz, 2014), except that we introduced a fourth class to describe the most irregular nuclei (see Materials and Methods).”

9). Line 237-252: Abnormal chromosome segregation and postmitotic nuclear assembly in all gfp::baf-1(G12T) embryos is fully consistent (not 'presumably causative'; line 251) with the embryonic loss-of-function phenotype for baf-1 (Margalit et al., 2005, PNAS) and is consistent with mutational disruption of binding to lamin (Liu J et al., 2000, MBC) and/or LEM-domain proteins (Liu J, Lee KK et al., 2003, PNAS).

We thank the reviewer for pointing this out. We have added the following sentence: “These phenotypes are consistent with the effects of embryonic depletion of BAF-1 or LMN-1 (Liu et al., 2000; Margalit, Segura-Totten, Gruenbaum, & Wilson, 2005).”

10). Lines 530-533: Baf-1 localization (mobility) in intestinal cells is known to change profoundly in response to heat shock, caloric restriction or food deprivation (Bar et al., 2014, MBC). It would be worthwhile testing, in future, whether the NGPS mutation affects baf-1 localization in response to these stresses.

We appreciate this suggestion, and we agree with the reviewer that it would be important to test this in future studies.

Other changes:

Missing column in Table S3 added.

Mistake if column heading in Table S4 corrected.

Breusegem, S. Y., Houghton, J., Romero-Bueno, R., Fragoso-Luna, A., Kentistou, K. A., Ong, K. K., . . . Larrieu, D. (2022). A multiparametric anti-aging CRISPR screen uncovers a role for BAF in protein translation. bioRxiv. doi:10.1101/2022.10.07.509469

Gorjanacz, M., Klerkx, E. P., Galy, V., Santarella, R., Lopez-Iglesias, C., Askjaer, P., & Mattaj, I. W. (2007). Caenorhabditis elegans BAF-1 and its kinase VRK-1 participate directly in post-mitotic nuclear envelope assembly. Embo J, 26(1), 132-143. doi:10.1038/sj.emboj.7601470

Greil, F., Moorman, C., & van Steensel, B. (2006). DamID: mapping of in vivo protein-genome interactions using tethered DNA adenine methyltransferase. Methods Enzymol, 410, 342-359. doi:10.1016/S0076-6879(06)10016-6

Holaska, J. M., Lee, K. K., Kowalski, A. K., & Wilson, K. L. (2003). Transcriptional repressor germ cell-less (GCL) and barrier to autointegration factor (BAF) compete for binding to emerin in vitro. J Biol Chem, 278(9), 6969-6975.

Janssen, A., Marcelot, A., Breusegem, S., Legrand, P., Zinn-Justin, S., & Larrieu, D. (2022). The BAF A12T mutation disrupts lamin A/C interaction, impairing robust repair of nuclear envelope ruptures in Nestor-Guillermo progeria syndrome cells. Nucleic Acids Res. doi:10.1093/nar/gkac726

Liu, J., Rolef Ben-Shahar, T., Riemer, D., Treinin, M., Spann, P., Weber, K., . . . Gruenbaum, Y. (2000). Essential roles for Caenorhabditis elegans lamin gene in nuclear organization, cell cycle progression, and spatial organization of nuclear pore complexes. Mol Biol Cell, 11(11), 3937-3947.

Margalit, A., Segura-Totten, M., Gruenbaum, Y., & Wilson, K. L. (2005). Barrier-to-autointegration factor is required to segregate and enclose chromosomes within the nuclear envelope and assemble the nuclear lamina. Proc Natl Acad Sci U S A, 102(9), 3290-3295. doi:10.1073/pnas.0408364102

Montes de Oca, R., Shoemaker, C. J., Gucek, M., Cole, R. N., & Wilson, K. L. (2009). Barrier-to-autointegration factor proteome reveals chromatin-regulatory partners. PLoS ONE, 4(9), e7050. doi:10.1371/journal.pone.0007050

Perez-Jimenez, M. M., Rodriguez-Palero, M. J., Rodenas, E., Askjaer, P., & Munoz, M. J. (2014). Age-dependent changes of nuclear morphology are uncoupled from longevity in Caenorhabditis elegans IGF/insulin receptor daf-2 mutants. Biogerontology, 15(3), 279-288. doi:10.1007/s10522-014-9497-0

Puente, X. S., Quesada, V., Osorio, F. G., Cabanillas, R., Cadinanos, J., Fraile, J. M., . . . Lopez-Otin, C. (2011). Exome sequencing and functional analysis identifies BANF1 mutation as the cause of a hereditary progeroid syndrome. Am J Hum Genet, 88(5), 650-656. doi:10.1016/j.ajhg.2011.04.010

Schedl, T., & Kimble, J. (1988). fog-2, a germ-line-specific sex determination gene required for hermaphrodite spermatogenesis in Caenorhabditis elegans. Genetics, 119(1), 43-61. doi:10.1093/genetics/119.1.43

Schuster, E., McElwee, J. J., Tullet, J. M., Doonan, R., Matthijssens, F., Reece-Hoyes, J. S., . . . Gems, D. (2010). DamID in C. elegans reveals longevity-associated targets of DAF-16/FoxO. Mol Syst Biol, 6, 399. doi:10.1038/msb.2010.54

Swift, J., Ivanovska, I. L., Buxboim, A., Harada, T., Dingal, P. C., Pinter, J., . . . Discher, D. E. (2013). Nuclear lamin-A scales with tissue stiffness and enhances matrix-directed differentiation. Science, 341(6149), 1240104. doi:10.1126/science.1240104

Vogel, C., & Marcotte, E. M. (2012). Insights into the regulation of protein abundance from proteomic and transcriptomic analyses. Nat Rev Genet, 13(4), 227-232. doi:10.1038/nrg3185

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 mechanisms of rare human progeria syndromes caused by mutations in nuclear lamina proteins (lamins or BANF1) are still poorly understood, mainly because these proteins are complicated: they interact and are structurally essential for mitosis and nuclear assembly; hence, disrupting either protein can (and often does) disrupt the other. Lamins and BANF1 also have multiple interwoven roles with other partners involved in 3D genome organization, chromatin regulation, and tissue-specific gene regulation during interphase. To focus on Nestor-Guillermo progeria syndrome (NGPS), caused by the homozygous A12T missense mutation in human BANF1, the authors inserted the corresponding G12T mutation in C. elegans baf-1. They tested potential phenotypes at multiple levels (molecular, transcriptional, cellular and organismal) extensively and rigorously, and did careful controls to determine whether BAF, a tiny (89-aa) protein, was disrupted by fusion to proteins such as GFP or TurboID. Animals carrying the G12T mutation exhibited reduced lifespan (Fig. 1, S1), lower responses to UV irradiation and heat-stress (Fig 7B, 7C), and revealed unexpected germline-specific defects in male worms (Fig. S2, S3), and altered gene expression in two tissues affected by human HGPS (Figs. 4 and 5). Overall, this manuscript strongly supports the major conclusion that this C. elegans line is a powerful model for human NGPS that complements a previously reported Drosophila model.

Equally importantly, from the viewpoint of fundamental discovery, this manuscript also reports major advances in understanding how BAF influences gene expression at the molecular level. Through careful attention to controls, and experimental design, the authors overcome many complications that make BAF difficult to study: its essential roles in mitosis and early embryogenesis, the 'tag'-sensitivity of endogenous BAF, and the absolute necessity to study BAF in native cell types. The authors carefully compared the impacts of tagging either baf-1 or lamin, and compared wildtype versus G12T-mutated baf-1 interactions with lamin and emerin (Fig. S5, S6, S8, S9; videos S1 and S2). DamID-Baf-1 access to chromatin was unaffected by the G12T mutation (Fig. S7), but they successfully identified subsets of genes 'occupied' by baf-1 in specific cell types, some of which were significantly affected in opposite ways by the NGPS mutation (Fig. 4, Fig. 5). However, these important new results are described too briefly, and discussion is inadequate. E.g., in hypodermal cells, the baf-1 G12T mutation dysregulated genes encoding proteins in five categories (ribosomal, proton transport, cuticle components, cell surface, lysine acetylation), by downregulating genes in three categories (ribosomal, proton transport, histone acetylation) and upregulating three other categories (cuticle components, cell surface, apical region). In intestinal cells, the mutation dysregulated genes in 8 categories (ribosomal, response to X-ray, proton transport, proteasome binding, mitochondrial protein import, endopeptidase activity, carboxy-lyase activity, ATP generation), by downregulating genes in 5 categories (ribosomal, proton transport, peptidase activity, NAD binding, metal cluster binding) and upregulating 3 categories (ribosomal, response to external stimulation, histone acetylation). Opposite results for "ribosomal genes" is confusing. Examples of genes in each affected category are shown in Fig. 6. To fully interpret this data, and address apparently-conflicting results, further analysis is needed to determine if any affected groups of genes have shared regulators. For example, Fig 5E shows "ribosomal protein genes" are both up- and down-regulated by the mutation. The authors should consider: (a) WHICH ribosomal genes are in each category, and (b) does either group of genes have known regulators that might be differentially affected by the baf-1 mutation? Similar consideration of other sets of differentially-affected genes might provide novel insight into specific chromatin-regulatory proteins (e.g., potential baf-1 partners; see next paragraph) affected by the NGPS mutation.

The current manuscript is too strictly focused on establishing C. elegans as a model for NGPS, and neglects the novel discoveries. The authors did not consider or discuss HOW a baf-1 mutation might cause such complex gene expression outcomes- given that baf-1 binds dsDNA nonspecifically. One plausible molecular explanation is that the NGPS mutation might affect baf-1 interactions with: (a) transcription factors (Requiem, RBBP4, DDB1) or chromatin-regulators (PARP1; UV-regulated interactions with DDB2 and CUL4A) identified as BAF-associated in a proteomic study (Montes de Oca et al., 2009), or (b) histone modifiers such as SET/I2PP2A (blocks H3 dephosphorylation) or H3K9 methyltransferase 'G9a' (Montes de Oca et al., 2011), or (c) other regulators that control affected genes identified in this manuscript.

Other clarifications and revisions to improve the manuscript:

Figures 1, 2, 4, 5: the graphs in Fig 1A,B,D-F and Fig 2B,D and the colorscales in Fig 4F and Fig 5E are uninterpretable when printed in black-and-white. Please fix Figs 1 and 2 using black/light-gray/white/stippled for bar graphs, and black/light-gray/solid/dotted/dashed for line graphs. Fig 2B can be fixed by direct-labeling of class numbers within each bar (instead of 'color-coding' separately).

Revise abstract lines 40-42 ("suggesting a direct relationship between BAF-1 binding [to what?] and gene expression") to reflect the deeper analysis.

Lines 132-155 (Figure 1): The impact on sperm production suggests the NGPS mutation might affect association with Germ cell-less (GCL), a transcription repressor that competes with BAF for binding to emerin in mammalian cells (Holaska et al., 2003 JBC). Lines 151-154: Did not understand the fog-2 'feminized worm' experiments. Please briefly explain for non-worm experts.

Line 190: Clarify that nuclear shapes were categorized manually by single-blind observer.

Line 237-252: Abnormal chromosome segregation and postmitotic nuclear assembly in all gfp::baf-1(G12T) embryos is fully consistent (not 'presumably causative'; line 251) with the embryonic loss-of-function phenotype for baf-1 (Margalit et al., 2005, PNAS) and is consistent with mutational disruption of binding to lamin (Liu J et al., 2000, MBC) and/or LEM-domain proteins (Liu J, Lee KK et al., 2003, PNAS).

Line 424: Agree that this new C. elegans model is important and strongly complements the Drosophila NGPS model.

Lines 463-464: Agree that future suppressor analysis in this C. elegans model will be powerfully informative.

Lines 530-533: Baf-1 localization (mobility) in intestinal cells is known to change profoundly in response to heat shock, caloric restriction or food deprivation (Bar et al., 2014, MBC). It would be worthwhile testing, in future, whether the NGPS mutation affects baf-1 localization in response to these stresses.

Referees cross-commenting

I agree with the comments from both other reviewers.

Significance

Overall, this manuscript strongly supports the major conclusion that this C. elegans line is a powerful model for human NGPS that complements a previously reported Drosophila model.

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

Romero-Bueno et al. have generated a C.elegans model of Nestor-Guillermo Progeria Syndrome (NGPS) caused by a point mutation in the BAF1 gene and they characterize the worm model. They find reduced fertility, reduced longevity and earlier aging symptoms in mutant animals compared to wild type animals. Looking at the molecular level, the authors find reduced accumulation of lamin A and emerin at the nuclear periphery in cells from mutant animals. They also find mitotic chromosome segregation defects. Using tissue-specific DamID, they show altered binding patters of the mutant protein to genome regions and they identify some gene groups which show both changes in gene expression and BAF1 binding. Finally, they show that BAF1 mutants are more resistant to oxidative stress than wild type animals.

The value of this work is two-fold: First, it is a very robust characterization of NGPS worms. Second, this will be a very useful model for the study of NGPS. Overall, the study is well-designed, technically strong, and the results are carefully and thoughtfully interpreted, which is nicely exemplified by the discussion of the relatively small number of genes which are differentially bound by BAF1 and are also differentially expressed and the authors do a good job of not overinterpreting the data, but simply state them. The results are convincing and informative.

My only minor point that may make this paper marginally better is that it would be nice to have a paragraph in the Discussing elaborating on the potential and the limitations of using the worm model to understand human NGPS, for example, humans have multiple lamin proteins etc.

Referees cross-commenting

I am glad to see that there is strong agreement that this is a valuable study.

Significance

The study is significant as it introduces a new animal model system to study an ultra-rare disease. The presented results are robust and convincing. This is the first worm model for this disease and it will be of interest to those studying laminopathies. The worm model is expected to reflect some of the human disease phenotypes but not all and a discussion of the potential and the limitation of the worm model to study NGPS would be welcomed.

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:

Specific mutations in nuclear lamina proteins such as lamin A or BAF1 can cause premature aging syndromes (Huchinson Gilford progeria or Nestor Guillermo syndrome), which show age related deterioration of nuclear morphology as a hallmark and affected individuals have a severely shortened life span. In this study Romero-Bueno et al established an animal model system for the Nestor Guillermo syndrome, by generating C. elegans strains harboring homozygous baf-1::G12A mutations. They nicely recapitulate the expected cellular and organismal phenotypes: decreased life spans of the mutant animals and faster nuclear deterioration. In addition, the authors find reduced fertility in the mutant when BAF-1 was combined with a GFP tag and synthetic lethality when the baf-1::G12T mutant is introduced into strain carrying an epitope tagged Lamin allele. At the organismal level the authors report increased resistance to oxidative stress, but reduced thermotolerance and decreased UV resistance. By conducting tissue specific DamID together with tissue specific RNA polymerase DamID, the authors find that in the baf-1::G12T mutant the overall chromatin association of chromosome arms remains largely unchanged, however a few individual loci either lost or gained BAF-1 association. The authors report that loss of BAF-1 association with chromatin correlates with gene expression in some instances, however there was no strict uniform correlation. This is not surprising, because the changes in expression could be a secondary consequence of a BAF-1-mediated change of another locus or a consequence of altered lamin nucelar envelope association. The global pattern of BAF-1 and BAF-1 G12T binding to chromatin was very similar in both genotypes. The authors find unaltered localization of BAF-1 G 12T protein at the nuclear envelope, in contrast to reduced levels of lamin and emerin. Interestingly, BAF-1 is found on sperm, in contrast to the absence of other lamina proteins, like LMN-1 or Emerin.

Major comments:

1. In the baf-1 G12T mutants the authors find reduced levels of lamin in hypodermal nuclei. It would be good to also examine the dynamics of lamin in the second tissue that was subjected to DamID (intestinal cells).
2. The authors make a statement that EMR-1 expression was reduced in the baf-1 G12T mutant, but do not comment on LMN-1 expression. Can a statement on this be made by RT PCR?
3. The authors find few alterations in gene regulation of the loci which have different occupancy WT BAF-1 versus BAF-1 G12T. It was surprising to see the DamID and RNA polymerase DamID experiments be done with worms grown at 20oC, because the more penetrant phenotypes at the organismal level were observed at 25oC. Could this be the reason for the little change of chromatin occupancy of BAF-1 and BAF-1 G12T or few changes in gene expression? Would it make sense to examine the expression of some selected BAF-1 bound loci by single molecule Fish at 25oC and compare expression wt versus baf-1 G12T?

Minor comments:

The finding that BAF-1 nuclear envelope localization remains unchanged in the mutant stems from detection of the inserted GFP epitope. Given that the tag has an influence on the BAF-1 G12 12T mutant viability, this statement should be phrased with more care. The tag could influence the turnover of the protein for example. Maybe Western blots comparing the signal of WT BAF-1 worms and BAF-1 G12T mutant worms would be instructive to compare the levels of the protein (at 20oC and at 25oC, day 1 adults and day 8 adults)

Line 105: typo: remove "s"

Line 154: A conclusion is missing for the fog-2 experiment would it make sense to discuss a possible influence of altered lamin binding to the nuclear envelope in the mutant in the context of the gene expression results?

Referees cross-commenting I also see this as a valuable study. I regretted a bit that the analysis was not done at the higher temperature when the authors saw the most prominent phenotypes--but I suppose the analysis is very expensive and time consuming.

Significance

Since a long time, it has remained a matter of debate whether the progeria T to G transition in BAF-1 reduces binding of BAF-1 to lamin or whether the mutation affects the binding of BAF-1 to chromatin and thereby alters chromatin organization. Conflicting results emerged from the studies of BAF1 mutants in tissue culture cells. For this reason, this study-conducted in the context of a whole animal-is very important: it allowed the author to do their experiments with cells with unaltered ploidy, expression from the endogenous promoters and in the context of defined tissues. A second conflicting finding concerned the localization of BAF-1 G 12 to T mutant protein at the nuclear envelope: some labs find it reduced at the nuclear envelope, others find unchanged amounts at the nuclear envelope. With this work the authors contributed novel and interesting findings to those ongoing discussions, they found both altered affinity of BAF-1 with chromatin (not on a global scale, but on a local scale) and reduced affinity to lamin.

Furthermore, this is one of the first studies mapping BAF-1 association with individual gene loci in a specific tissue and the authors showed that in a given tissue BAF-1 tends to be associated with not expressed genes. In a nutshell, the authors have established a convincing accessible model system for studying aging, ready for consecutive testing interventions to reduce the pace of premature aging. Strengths: convincing presentation of a novel genetic model system to study progeria, first study where BAF-1 bound loci were shown from the analysis of a tissue (there is a correlation of BAF-1 bound loci, which are not expressed in the examined tissue), introduction of an easy-to-handle model to search for compounds suitable for clinical intervention for progeria patients or anti-aging drugs. This study adds some clarity to conflicting views in the filed: the NGS mutation in BAF-1 both reduces the amount of lamin at the nuclear periphery and affects the affinity of BAF-1 to chromatin.

Limitations: the observed transcriptional changes in the mutant can be either a direct consequence of BAF-1 chromatin association or a consequence of an altered lamina since lamin is less stable at the nuclear envelope. The transcriptomic analysis was not conducted at the temperature at which penetrant phenotypes at the organismal level were observed.

Advance: Previous studies presented conflicting results about the nuclear envelope localization of BAF-1 G12T protein: this study clearly shows that the localization of the protein remains unaltered. This study also clearly demonstrates that there is less lamin at the nuclear envelope in the mutant, lending support to the in vitro findings that the mutant is compromised in Lamin binding.

Audience: The study will be of interest to anyone who studies the nuclear lamina, the nuclear envelope, progeria, aging and stress response of an organism. Beyond this a convincing powerful novel genetic system is being presented to study progeria, which is of interest to clinicians. It is also of interest for translational research, because the system can be used to screen for compounds, which could be useful for therapeutic intervention for progeria patients or for the identification of compounds that combat aging in general. Some of the presented synthetic effects with tagged strains open the opportunity to conduct genetic suppressor screens, which would be a wonderful entry point to collect more mechanistic insights in the phenomena of aging and stress response. This genetic system is an awesome starting point for further studies and advancements of elucidating the molecular mechanism of progeria by genetic screens.

Expertise: I am a C. elegans geneticist and appreciate that all the conflicting results from tissue culture studies can now be compared to an analysis in a more physiological setting and the context of a real tissue and a living animal. I am not really competent to judge the sensitivity of RAPID assays.

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

Learn more at Review Commons

Reply to the reviewers

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

The paper by Salinas-Rebolledo et al. describes a novel PROTAC approach in which the UBX domain of FAF1 is fused to a nanobody that recognizes the target protein. The idea is that the UBX domain will bind the fusion protein to the p97 ATPase, a major ATPase involved in the unfolding of many proteins. The target protein recognized by the UBX-nanobody fusion (UBX-Nb) is then supposed to be unfolded in a ubiquitin-independent manner and subsequently degraded by the proteasome.

The authors provide evidence that Ubx-Nb, containing a nanobody recognizing GFP, can colocalize with GFP fusion proteins in the nucleus and to liquid-liquid phase separation structures. Importantly, the fusion can reduce the cellular levels of the target proteins. The authors confirm that degradation triggered by Ubx-Nb is proteasome dependent. Ubx-Nb can also promote the degradation of model proteins that form aggregates relevant to neurodegenerative diseases.

__* The major issue with the paper is that it does not provide mechanistic insight into the degradation mechanism. First, the data implicating p97 in degradation are conflicting. On the one hand, siRNA of p97 compromises degradation (although degradation is not completely inhibited; see Figure 4G), but on the other hand, an inhibitor of p97 does not have an effect. The authors have not shown that target proteins are actually unfolded by their artificial adaptor (in vitro experiments would be required). __

__In addition, it would be important to show co-localization in vivo with p97. __

Thus, the role of p97 is not convincingly established. Another major question is how the unfolded, non-ubiquitinated proteins would be degraded by the 26S proteasome. Is there a ubiquitin ligase required after substrate unfolding?

Overall, the paper reports some intriguing effects of their designed PROTAC adaptor, but the mechanism by which it functions remains unclear. The findings of the manuscript appears too preliminary in its current version for it to be of value to the community.

Responses #1

We appreciate the detailed feedback provided and acknowledge the reviewers' concerns regarding the mechanistic insight into the degradation process. Below, we address each point raised:

We acknowledge that our current data do not fully elucidate the degradation mechanism involving p97. Our preliminary findings suggest that while siRNA of p97 compromises degradation, this effect is not absolute. Additionally, we recognize the inconsistency observed with the p97 inhibitor, which did not affect degradation. It is important to note that the selective inhibitor used in our study binds to the D2 domains of p97, as reported by Zhou et al. (J Med Chem. 2015). There are molecules that bind to the D1 domain and enhance p97-mediated degradation (Figuerola-Conchas et al., ACS Chem Biol. 2020). The primary ATPase function of p97 is associated with the D2 domain, but ATP binding to the D1 domain is crucial for hexamer formation and N-terminal domain conformation, which regulates cofactor binding and various functions of p97. Therefore, it is possible that our p97-PROTAC activates the D1 domain or interacts with cofactors enhancing protein degradation mediated by p97.

We plan to conduct in vitro experiments to demonstrate that target proteins are unfolded by our artificial adaptor. We have already cloned the degradation system into a bacterial expression vector to express and purify the system for these in vitro studies, which will be carried out in California. Furthermore, we aim to show in vivo co-localization with p97 in future studies.

*We understand the importance of establishing the role of p97 convincingly. Our preliminary data indicate the presence of residual p97 in siRNA experiments. Regarding the degradation of unfolded, non-ubiquitinated proteins by the 26S proteasome, there are studies indicating that various proteins are degraded by the proteasome independently of ubiquitin. Moreover, evidence suggests that different pools of the same protein can be directed to the proteasome via both ubiquitin-dependent and ubiquitin-independent mechanisms under the same cellular conditions. Also, prior research by Butler et al. (2016) demonstrated that fusing NbSyn87 with the mouse Ornithine Decarboxylase (ODC) PEST degron effectively reduced protein levels and this reduction was achieved by harnessing the innate cellular machinery responsible for ubiquitin-independent proteolysis. *

1. • Makaros Y, Raiff A, Timms RT, Wagh AR, Gueta MI, Bekturova A, Guez-Haddad J, Brodsky S, Opatowsky Y, Glickman MH, Elledge SJ, Koren I. Ubiquitin-independent proteasomal degradation driven by C-degron pathways. Mol Cell. 2023 Jun 1;83(11):1921-1935.e7. doi: 10.1016/j.molcel.2023.04.023. Epub 2023 May 17. PMID: 37201526; PMCID: PMC10237035.*
2. • Butler DC, Joshi SN, Genst E, Baghel AS, Dobson CM, Messer A. Bifunctional Anti-Non-Amyloid Component α-Synuclein Nanobodies Are Protective In Situ. PLoS One. 2016 Nov 8;11(11):e0165964. doi: 10.1371/journal.pone.0165964. PMID: 27824888; PMCID: PMC5100967.*
3. • Erales J, Coffino P. Ubiquitin-independent proteasomal degradation. Biochim Biophys Acta. 2014 Jan;1843(1):216-21. doi: 10.1016/j.bbamcr.2013.05.008. Epub 2013 May 14. PMID: 23684952; PMCID: PMC3770795.*
4. • Donghong Ju, Youming Xie, Proteasomal Degradation of RPN4 via Two Distinct Mechanisms, Ubiquitin-dependent and -independent*, Journal of Biological Chemistry, Volume 279, Issue 23, 2004, Pages 23851-23854, ISSN 0021-9258.*
5. *

We also speculate that the degradation may also occur via the autophagy-lysosome pathway. These speculations will be added to the discussion section, and we plan to investigate this pathway in detail in a subsequent scientific article focused on the mechanism of action of p97-PROTAC.

We recognize the need for further mechanistic studies and plan to publish these preliminary findings while continuing our research to elucidate the degradation mechanism. Our future work includes determining the crystal structure and conducting mass spectrometry to identify other proteins interacting with this complex, potentially aiding in degradation. We also plan to test the system in vivo using murine models with alpha-synuclein overexpression, in collaboration with researchers in Spain. This long-term project will form the basis of a subsequent publication.

We believe that our findings present an innovative and unique tool, and this preliminary data warrant publication. We appreciate the reviewers' comments and hope that our detailed response and future research plans address their concerns.

Minor points:

Fig. 1B: Although emerin is reported to be a nuclear envelope protein, it is not localized to the NE, but throughout the ER, likely because the protein was too highly expressed.

We agree with the reviewer, the overexpression of the NE could leak into the ER. We do not have specific Nanobodies to directly degrade Emerin. However, we would like to make the point that in both cases the protein will conserve a single transmembrane domain and even then, the GFP Nanobody fused to the UBX domain is able to trigger degradation

Fig. 1B: ETV co-localization is not obvious from the figure.

We are going to repeating the experiment and quantify the colocalization as suggested

Fig. 4: The depletion of p97 leads to cell death, so it is unclear whether the siRNA effect is specific.

We appreciate your comment and would like to clarify that there are published studies where the p97 gene has been silenced in HeLa cells without causing complete cell death. While it is true that a significant number of cells die post-transfection, we have observed that by changing the cell culture medium daily, the remaining cells start to grow again.

Studies supporting our findings include:

1. • Wójcik C, Yano M, DeMartino GN. RNA interference of valosin-containing protein (VCP/p97) reveals multiple cellular roles linked to ubiquitin/proteasome-dependent proteolysis. J Cell Sci. 2004 Jan 15;117(Pt 2):281-92. doi: 10.1242/jcs.00841. Epub 2003 Dec 2. PMID: 14657277. *
2. • Beskow A, Grimberg KB, Bott LC, Salomons FA, Dantuma NP, Young P. A conserved unfoldase activity for the p97 AAA-ATPase in proteasomal degradation. J Mol Biol. 2009 Dec 11;394(4):732-46. doi: 10.1016/j.jmb.2009.09.050. Epub 2009 Sep 24. PMID: 19782090.*
3. • Yahiro K, Tsutsuki H, Ogura K, Nagasawa S, Moss J, Noda M. Regulation of subtilase cytotoxin-induced cell death by an RNA-dependent protein kinase-like endoplasmic reticulum kinase-dependent proteasome pathway in HeLa cells. Infect Immun. 2012 May;80(5):1803-14. doi: 10.1128/IAI.06164-11. Epub 2012 Feb 21. PMID: 22354021; PMCID: PMC3347452. Additionally, although our figure does not clearly show the band corresponding to p97, upon overexposing the film, we can detect a very faint band of the protein. This indicates that there is still residual expression of p97, albeit at a lower concentration, which is consistent with a significant reduction but not a complete elimination of the protein.*

Fig. 5C: The colocalization of GFP-HTT Q23 with UBX-Nb(GFP) is not entirely convincing.

We are going to repeating the experiment and quantify the colocalization as suggested

Reviewer #1 (Significance (Required)):

Overall, the paper reports some intriguing effects of their designed PROTAC adaptor, but the mechanism by which it functions remains unclear. The findings of the manuscript appears too preliminary in its current version for it to be of value to the community.

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

Summary:

The authors have developed a p97-directed proteolysis-targeting chimera (PROTAC) that operates independently of ubiquitin. This system employs a camelid nanobody to selectively recognize target proteins, tethered to p97 through the UBX domain of the p97 adapter FAF1. The anti-GFP nanobody effectively targets various GFP-fusion proteins for degradation via the proteasome, relying on p97 for its mechanism of action. The authors validate the presence of p97 in brain tissues of Non-Human Primates (Nhp) Macaca fascicularis, rat (Sprague Dawley), and mouse (C57BL6/C), supported by proteasome inhibition and p97 RNA silencing data. Importantly, the p97-PROTAC mechanism operates independently of ubiquitination, demonstrated through degradation of clinically relevant proteins such as alpha-synuclein using a camelid nanobody (NbSyn87).

Major comments:

* Anti-GFP Nanobody clarification: Details about the original anti-GFP nanobody are unclear, which makes reproducing the current work a challenge for outside labs.

o 20. Fulcher LJ, Macartney T, Bozatzi P, Hornberger A, Rojas-Fernandez A, Sapkota GP. An affinity-directed protein missile system for targeted proteolysis. Open Biol. Oct 2016;6(10)doi:10.1098/rsob.160255

o I assume that the anti-GFP nanobody is aGFP from supplementary figure #2 in the Fulcher manuscript but is unclear as they also have used anti-GFP nanobody aGFP16.

* Amino acids from FAF1-UBX domain: Further clarity is needed regarding the amino acid details from the FAF1-UBX domain, which may have been disclosed in a patent application but should be explicitly outlined in the methods section.

We appreciate your comment regarding the need for further clarity on the amino acid details from the FAF1-UBX domain. To address this, we have added a supplementary figure which includes a table outlining the amino acid sequences of our construct and the FAF1-UBX domain. This supplementary figure provides a detailed representation of the amino acid sequences. We hope this addition meets your requirements and provides the necessary information for a comprehensive understanding of our work.

__* Degradation of Proteins of Clinical Interest: The data presented is not convincing enough to support the stated claims that the PROTAC is clearing aggregated mutant HTT. In Figure 5, there is an abundance of GFP-HTT Q74 puncta. While the western blot data suggests a reduction of soluble GFP-HTT-Q74 protein levels, it does not account for aggregated HTT. Aggregated HTT does not efficiently enter the separating gel during electrophoresis. To make these claims the authors need to 1. Show the level of mHTT Q74 aggregation in the empty control groups so that a comparison can be visually made between empty control groups and UBX-Nb(GFP) treated groups. A similar comparison would be useful with the GFP-HTT Q23 treated cells as well. __

* Visualization of Aggregated Proteins: The continued visibility of puncta raises doubts about the system's efficacy in degrading aggregated proteins. Including comparisons between untreated and treated cells for all test systems would strengthen the argument. It would be useful to show a comparison between the untreated controls and UBX-Nb (GFP) treated cells for all the test systems shown.

*We acknowledge that the data presented in Figure 5 may not be sufficient to support the claim that the PROTAC is clearing aggregated mutant HTT. The western blot data suggests a reduction in soluble GFP-HTT-Q74 protein levels, but it does not account for aggregated HTT, which does not efficiently enter the separating gel during electrophoresis. We agree that more experiments will need to be performed to evaluate the impact of the p97 PROTAC on the direct turnover of the aggregates once those are form. *

Minor comments:

* The In-text citations should be placed outside of the sentence. Example: The ubiquitin-proteasome system (UPS) regulates protein abundance by specific E3 ubiquitin ligases, which catalyze ubiquitin chain formation on the substrates, inducing their proteasome mediated degradation. 1-4

* Sentence two in the introduction is missing a period. It is unclear whether sentence three is a heading or part of paragraph one.

* There are additional formatting issues. It would be easier to read the paper if there was a space between paragraphs. Page numbers would be helpful.

* Page 2. Missing word. Inclusion body myopathy associated with Paget's disease.

* Figure 1. D, F, and E. Missing annotation to denote significance.

* Figure 2G Missing annotation to denote significance.

* Figure 4. Missing annotation to denote significance for figures 4D, 4F, 4H.

* Is Figure 4I significantly different or not? In Figure 6, you use ns to denote not significant. This feels like it is an important point that you would want to make that the effect is dependent on p97. When you knock out p97 the degradation capacity of UBX-Nb is lost.

Response

These changes are going to be apply

Reviewer #2 (Significance (Required)):

* The p97-PROTAC system is an ubiquitin-independent approach to degrade intracellular proteins. This system was able to target proteins for degradation at diverse subcellular locations, integral membrane protein residing at the inner nuclear membrane, chromatin located, and liquid-liquid phase separated compartments. The ability to clear alpha-synuclein builds on previous research suggesting that ubiquitin-independent degradation of alpha-synuclein could be a therapeutic approach to treat synucleinopathies such as Parkinson's Disease. However, the ability of this approach to clear aggregated proteins is not convincing, given the presence of visible aggregates in the treated cells.

* The investigations with NbSyn87 build upon prior research by Butler et al. (2016), who fused NbSyn87 with the mouse Ornithine decarboxylase (ODC) PEST degron. This fusion strategy not only facilitates the targeting of alpha-synuclein but also harnesses the innate cellular machinery responsible for ubiquitin-independent proteolysis. The current approach demonstrates and alternative mechanism to direct alpha-synuclein (and other proteins) into the proteasome for ubiquitin-independent clearance.

* At the current state of development, this research is of interest to specialized audience with antibody engineering backgrounds; however, it holds translational potential for clearance of toxic proteins.

* My research interests is in the development of therapeutics for the treatment of neurodegenerative diseases including Huntington's Disease, Parkinson's Disease, and Alzheimer's Disease.

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 have developed a p97-directed proteolysis-targeting chimera (PROTAC) that operates independently of ubiquitin. This system employs a camelid nanobody to selectively recognize target proteins, tethered to p97 through the UBX domain of the p97 adapter FAF1. The anti-GFP nanobody effectively targets various GFP-fusion proteins for degradation via the proteasome, relying on p97 for its mechanism of action. The authors validate the presence of p97 in brain tissues of Non-Human Primates (Nhp) Macaca fascicularis, rat (Sprague Dawley), and mouse (C57BL6/C), supported by proteasome inhibition and p97 RNA silencing data. Importantly, the p97-PROTAC mechanism operates independently of ubiquitination, demonstrated through degradation of clinically relevant proteins such as alpha-synuclein using a camelid nanobody (NbSyn87).

Major comments:

• Anti-GFP Nanobody clarification: Details about the original anti-GFP nanobody are unclear, which makes reproducing the current work a challenge for outside labs.
  • 1. Fulcher LJ, Macartney T, Bozatzi P, Hornberger A, Rojas-Fernandez A, Sapkota GP. An affinity-directed protein missile system for targeted proteolysis. Open Biol. Oct 2016;6(10)doi:10.1098/rsob.160255
  • I assume that the anti-GFP nanobody is aGFP from supplementary figure #2 in the Fulcher manuscript but is unclear as they also have used anti-GFP nanobody aGFP16.
• Amino acids from FAF1-UBX domain: Further clarity is needed regarding the amino acid details from the FAF1-UBX domain, which may have been disclosed in a patent application but should be explicitly outlined in the methods section.
• Degradation of Proteins of Clinical Interest: The data presented is not convincing enough to support the stated claims that the PROTAC is clearing aggregated mutant HTT. In Figure 5, there is an abundance of GFP-HTT Q74 puncta. While the western blot data suggests a reduction of soluble GFP-HTT-Q74 protein levels, it does not account for aggregated HTT. Aggreggated HTT does not efficiently enter the separating gel during electrophoresis. To make these claims the authors need to 1. Show the level of mHTT Q74 aggregation in the empty control groups so that a comparison can be visually made between empty control groups and UBX-Nb(GFP) treated groups. A similar comparison would be useful with the GFP-HTT Q23 treated cells as well.
• Visualization of Aggregated Proteins: The continued visibility of puncta raises doubts about the system's efficacy in degrading aggregated proteins. Including comparisons between untreated and treated cells for all test systems would strengthen the argument. It would be useful to show a comparison between the untreated controls and UBX-Nb (GFP) treated cells for all the test systems shown.

Minor comments:

• The In-text citations should be placed outside of the sentence. Example: The ubiquitin-proteasome system (UPS) regulates protein abundance by specific E3 ubiquitin ligases, which catalyze ubiquitin chain formation on the substrates, inducing their proteasome mediated degradation. 1-4
• Sentence two in the introduction is missing a period. It is unclear whether sentence three is a heading or part of paragraph one.
• There are additional formatting issues. It would be easier to read the paper if there was a space between paragraphs. Page numbers would be helpful.
• Page 2. Missing word. Inclusion body myopathy associated with Paget's disease.
• Figure 1. D, F, and E. Missing annotation to denote significance.
• Figure 2G Missing annotation to denote significance.
• Figure 4. Missing annotation to denote significance for figures 4D, 4F, 4H.
• Is Figure 4I significantly different or not? In Figure 6, you use ns to denote not significant. This feels like it is an important point that you would want to make that the effect is dependent on p97. When you knock out p97 the degradation capacity of UBX-Nb is lost.

Significance

• The p97-PROTAC system is an ubiquitin-independent approach to degrade intracellular proteins. This system was able to target proteins for degradation at diverse subcellular locations, integral membrane protein residing at the inner nuclear membrane, chromatin located and liquid-liquid phase separated compartments. The ability to clear alpha-synuclein builds on previous research suggesting that ubiquitin-independent degradation of alpha-synuclein could be a therapeutic approach to treat synucleinopathies such as Parkinson's Disease. However, the ability of this approach to clear aggregated proteins is not convincing, given the presence of visible aggregates in the treated cells.
• The investigations with NbSyn87 build upon prior research by Butler et al. (2016), who fused NbSyn87 with the mouse Ornithine decarboxylase (ODC) PEST degron. This fusion strategy not only facilitates the targeting of alpha-synuclein but also harnesses the innate cellular machinery responsible for ubiquitin-independent proteolysis. The current approach demonstrates and alternative mechanism to direct alpha-synuclein (and other proteins) into the proteasome for ubiquitin-independent clearance.
• At the current state of development, this research is of interest to specialized audience with antibody engineering backgrounds; however, it holds translational potential for clearance of toxic proteins.
• My research interests is in the development of therapeutics for the treatment of neurodegenerative diseases including Huntington's Disease, Parkinson's Disease, and Alzheimer's Disease.

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 paper by Salinas-Rebolledo et al. describes a novel PROTAC approach in which the UBX domain of FAF1 is fused to a nanobody that recognizes the target protein. The idea is that the UBX domain will bind the fusion protein to the p97 ATPase, a major ATPase involved in the unfolding of many proteins. The target protein recognized by the UBX-nanobody fusion (UBX-Nb) is then supposed to be unfolded in a ubiquitin-independent manner and subsequently degraded by the proteasome.

The authors provide evidence that Ubx-Nb, containing a nanobody recognizing GFP, can colocalize with GFP fusion proteins in the nucleus and to liquid-liquid phase separation structures. Importantly, the fusion can reduce the cellular levels of the target proteins. The authors confirm that degradation triggered by Ubx-Nb is proteasome dependent. Ubx-Nb can also promote the degradation of model proteins that form aggregates relevant to neurodegenerative diseases.

The major issue with the paper is that it does not provide mechanistic insight into the degradation mechanism. First, the data implicating p97 in degradation are conflicting. On the one hand, siRNA of p97 compromises degradation (although degradation is not completely inhibited; see Figure 4G), but on the other hand, an inhibitor of p97 does not have an effect. The authors have not shown that target proteins are actually unfolded by their artificial adaptor (in vitro experiments would be required). In addition, it would be important to show co-localization in vivo with p97. Thus, the role of p97 is not convincingly established. Another major question is how the unfolded, non-ubiquitinated proteins would be degraded by the 26S proteasome. Is there a ubiquitin ligase required after substrate unfolding?

Overall, the paper reports some intriguing effects of their designed PROTAC adaptor, but the mechanism by which it functions remains unclear. The findings of the manuscript appears too preliminary in its current version for it to be of value to the community.

Specific points:

Fig. 1B: Although emerin is reported to be a nuclear envelope protein, it is not localized to the NE, but throughout the ER, likely because the protein was too highly expressed.

Fig. 1B: ETV co-localization is not obvious from the figure.

Fig. 4: The depletion of p97 leads to cell death, so it is unclear whether the siRNA effect is specific.

Fig. 5C: The colocalization of GFP-HTT Q23 with UBX-Nb(GFP) is not entirely convincing.

Significance

Overall, the paper reports some intriguing effects of their designed PROTAC adaptor, but the mechanism by which it functions remains unclear. The findings of the manuscript appears too preliminary in its current version for it to be of value to the community.

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:

We thank the reviewer for his/her time and for the constructive comments. Below please find our detailed responses to your points.

STING is a key signalling hub in the innate immune system, receiving multiple inputs from upstream activators (such as cGAS) and in turn triggering multiple downstream events (such as IFN induction, NF-kB signalling, autophagy, cell death). Mutations in the STING gene cause a rare inflammatory disease called SAVI. Using a previously established STING ki mouse that recapitulates some of the clinical observations in SAVI patients, this manuscript tests the hypothesis that TNF signalling drives pathology. Using anti-TNF antibody and TNF receptor knockout, the authors show that TNF indeed plays important roles in causing disease in this mouse model. For example, the loss of T cells and neurons is prevented when TNF signalling is blocked, and lung pathology is rescued in STING ki mice lacking TNF receptors. Overall, the manuscript is well written and laid out, and the experimental work is of a high technical standard.

Major comments

1. • Most figures show pooled data from two independent experiments including a total of 5-8 mice. Given the variability in some of the readouts, this raises the question of whether there is sufficient statistical power to draw conclusions. For example, in Figure 2, the conclusion that "Infliximab did not alter the expression of inflammatory mediators" seems questionable given the results in Figure 2F and G. Did the authors perform a power calculation? What effect size can the authors detect given the variability and number of replicates? Similarly, in Figure 3, the authors conclude that "Disruption of TNFR signaling did not significantly prevent T cell lymphopenia"; however, with some more replicates, the data in Figure 3D would likely reach significance. Similar concerns apply to several panels in Figures 4 and 6 and to Figure S5M. Ideally, the authors should perform additional repeat experiments to increase the number of replicates. If that is not possible, power calculations need to be provided and conclusions should explicitly mention the minimum effect size that the author can detect given the small sample size (for example "Infliximab did not alter the expression of inflammatory mediators more than x-fold").* Thank you for this suggestion. However, it is not possible to repeat the treatment of mice with Infliximab for generation of more replicates. The blockade of TNF signalling by treatment with drugs did not cure the murine SAVI disease. According to animal welfare restrictions, we cannot perform additional treatment experiments with Infliximab or Etanercept.

We analysed the effect size d, f and power of all these presented results and collected them in table S4. Additional explanations about effect sizes were added in the corresponding text to Figures 2 and 3. The demonstrated results in Figure 4 and 6 already contain significant data. We did not include the calculation of effects sizes here. All effect size and power calculations are summarized in table S4.

• The authors should not make unjustified overstatements. For example, STING KI; TNFR1/2 KO mice should not be referred to as a "new mouse model". The manuscript simply tests the role of TNFR1/2 in the already published STING N153S model. In line 687, avoid using "impressively" and in line 734 avoid using "massively".*

• *

Thank you for this suggestion. We changed this sentence into:…”these newly generated mouse lines of TNFR”…., see line 796. Additionally, in line 687 (actual line 705) we omitted “impressively” and in line 734 “massively produced” into “elevated” (actual line 752).

Minor comments

• Line 767-769: The statement that spike activates cGAS is misleading, because this effect is an indirect consequence of cell-to-cell fusion (Liu et al 2022).*

• *

Thank you for this suggestion. We changed this sentence into: Cell fusion caused by the SARS-CoV-2 spike protein is a potent… (actual line 785).

Reviewer #1 (Significance (Required)):

• *

The main strengths of this study are (1) the use of complementary antibody-based and genetic methods to test the role of TNF signalling; (2) the use of multiple different readouts; and (3) the analysis of many different cell types / organ systems. The main weaknesses are (1) small sample sizes limiting statistical power (see above) and (2) the exclusive use of mouse models.

• *

Overall, my opinion is that the advance is important, both fundamentally and clinically. Studies of this and the related V154M mouse model previously showed an important role of non-IFN pathways in driving disease. This study indicates that TNF signalling may cause pathology. This not only extends our understanding of STING's role in autoinflammation but also opens a direct therapeutic avenue using approved TNF targeting drugs.

• *

This study will be primarily of interest to specialised audiences working on STING and SAVI, and secondarily to the wider innate immunity field.

• *

This reviewer has expertise in the field of nucleic acid sensing, including cGAS-STING.

• *

• *

Reviewer #2:

We thank the reviewer for his/her time and for the constructive comments. Below please find our detailed responses to your points.

*In this paper, Luksch et al (2024) examines the role of TNF signaling in STING-associated vasculopathy with onset in infancy (SAVI). By using pharmacological inhibition and genetic inactivation of TNF receptors in a murine SAVI model (STING ki), the research found that pharmacologically inhibiting TNF signaling improved T cell lymphopenia but had limited effects on lung disease. Genetic inactivation of TNFR signaling, particularly TNFR1, enhanced thymocyte survival and expanded the peripheral T cell pool, reducing inflammation and neurodegeneration. The development and progression of severe lung disease in STING ki mice are also reliant on TNFR1 signaling, while TNFR2 deletion did not alleviate lung inflammation. The authors also explored the severe inflammatory lung disease manifestation, showing that primary lung endothelial cells in STING ki mice allowed more neutrophil attachment compared to those in STING WT mice, indicating chronic STING activity in endothelial cells disrupts the endothelial barrier and promotes severe lung disease. The study highlights TNFR signaling as crucial in SAVI and COVID-19 progression and suggests blocking TNFR1 signaling as a potential therapeutic approach for both diseases. *

• *

Major comments:

The paper establishes a strong connection between TNFR1 depletion and the reduction of SAVI disease severity in lung and neuroinflammation, suggesting TNFR1 blockade as a viable therapeutic strategy for SAVI. To strengthen the arguments and improve the therapeutic potential, the authors should address the following major comments:

1. • The authors conclude that TNFR1 signaling drives murine SAVI disease, as evidenced by the reduced severity of lung disease in TNFR1 -/- mice. While the genetic model is convincing, the discrepancy between pharmacological inhibition and genetic models needs clarification. Before attributing the pharmacological failure to late administration, have the authors considered that Infliximab might not sufficiently deplete TNF to achieve therapeutic benefits? In figure 2H, serum TNF levels were not significantly altered in STING ki mice treated with Infliximab. Have the authors considered using other TNF inhibitors or alternative methods to measure TNF depletion efficacy in STING ki murine models, such as qPCR, flow cytometry, or immunohistochemistry in lymph nodes or lung tissues?* Thank you for this suggestion. In a preliminary experiment, we already treated STING WT and STING ki mice with Etanercept which is not included in the paper. 3-week-old mice were treated with subcutaneously injection of 25 mg/kg Etanercept or saline, twice per week, for 7 weeks. After treatment, all mice were euthanized and single cell suspensions of blood and spleen were used for flow cytometry analysis. Lung tissue was harvested for histological analysis. Quantification of gene expression was performed by snap frozen lung and kidney tissue and quantification of secreted proteins was analysed by snap frozen serum.

The transcription of ISGs and proinflammatory mediators in lung tissue was not significantly improved by the Etanercept treatment of mice, see additional figure below (A – D). Interestingly, the amount of secreted CXCL9 in the serum was reduced in Etanercept treated mice compared to vehicle treated mice (E). We concluded that our treatment strategy had no impact in the manifestation and progression of murine SAVI disease, in highly inflamed tissues / organs. However, we found a reduction (partially significant) of proinflammatory mediator transcriptions in the kidney of Etanercept treated mice compared to vehicle control mice. Murine SAVI disease is a systemic autoinflammatory disease without histological alteration in kidney tissue of 10 weeks old mice. Remarkably, transcription of ISGs and proinflammatory mediators is highly upregulated in SAVI mice. Treatment with Etanercept improved this aberrant gene expression in murine SAVI influenced tissue / organ (I – K). These results encouraged us to perform the treatment with infliximab because we expected a more pronounced effect since infliximab can bind the monomeric and trimeric form while etanercept can only bind to the active trimeric from of TNF.

Etanercept treatment of STING WT (in black) and ____STING ki (in red)____ mice.

(A) Relative expression level of Cxcl10, (B) Mx1, (C) Tnf and (D) Il1b in lung tissue of Etanercept or saline treated STING WT and STING ki mice. (E) Quantification of CXCL9, (F) CXCL10, (G) IL-6 and (H) TNF in serum samples from STING WT and STING ki mice after treatment. (I) Relative expression level of Cxcl10, (J) Mx1, (K) Tnf and (L) Il1b in kidney tissue of treated mice.

• The TNF pathway exhibits redundancy, as multiple signaling molecules or pathways can compensate for the loss of TNF function to maintain cellular processes and immune responses. The authors showed that thymocytes of STING ki mice lacking TNFR1/2 expressed significantly lower levels of IFN-related genes (Cxcl10, Sting1), and mice lacking TNFR1 and TNFR1/2 expressed reduced levels of NF-κB-related genes. Does this imply that IFN and NF-κB pathways are downstream of TNF signaling driving SAVI progression? It would be valuable to hear the authors' comments or postulations on the potential mechanisms of TNF driving SAVI progression in the discussion, and the methods to dissect the mechanisms further using genetic or pharmacological methods.*

Thank you for this suggestion. STING is a key player in various proinflammatory mechanism and is directly involved in IFN and NF-κB signalling. We assume that these signalling pathways are adaptable to various proinflammatory situations. Knock out of TNFR1 and TNFR1/2 results in a strong inhibition of all inflammatory reactions in the whole organisms. We think, it is not possible to conclude mechanisms of murine SAVI manifestation and progression from the results of these mouse lines only. These observations provide new hypothesis, but cannot completely explain the mechanism.

• The authors mentioned that the pharmacological inhibition of TNF by Infliximab is ineffective due to late administration compared to the onset of SAVI. How would this affect the therapeutic treatment of TNF if the treatment is going to be later than the disease onset? Can the authors elaborate on the potential ways to circumvent the timing of treatment? Would TNFR1 antagonists experience the same issue? To understand disease progression and optimal targeting times, the creation of an inducible TNFR1/2 -/- mouse model could be beneficial. This is optional, but the authors are encouraged to comment on improving TNFR1/2 -/- mouse SAVI models to further study the therapeutic potential of TNF signaling blockage in treating SAVI.*

We agree with the suggestion. In the next project, we want to generate STING ki mice with inducible knock out.

Minor comments:

• The authors separate STING WT and STING ki into different graphs, which can sometimes make it hard to compare STING WT and STING ki baseline levels. It would be beneficial to merge the two genotypes into single graphs for easier comparison.*

Thank you for this suggestion. In the first version of this manuscript, we collected results from STING WT and STING ki mice in one graph with 8 bars in different colours and textures in the case of TNFR knock out lines. These graphs were overloaded and very confusing. It is was not possible to mark statistical calculations inside these graphs without losing the focus. Hence, we created the demonstrated design of graphs. We think this is the most convincing version.

• Figure S5 lacks statistical annotations, although the legends mention them. Are the statistics usually shown when a comparison is mentioned in the text, or are they only displayed when the differences are significant? It would be helpful if the authors could clarify this and ensure that all relevant statistical comparisons are clearly reflected in the graphs, regardless of the significance level. This consistency would improve the clarity and interpretation of the data presented.*

• *

Thank you for this suggestion. We removed the significance level from the legend of Figure S5 (actually line 1199).

• *

The authors did an excellent job discussing the study's implications, but some of this content could be moved to the introduction. The hypothesis that "tumor necrosis factor (TNF) signaling is involved in the manifestation and progression of murine SAVI disease" can be introduced more naturally once the authors present previous findings on TNF's association with various autoimmune disorders. This would set a clear context for the study's objectives and rationale.

We agree with this suggestion and inserted the sentence: “In our previous investigations, we observed an elevated transcription of Tnf in spleen and thymus of STING ki mice (Siedel et al., 2020).” (actual line 89/90).

General Assessment: The study identifies enhanced TNF signaling as a driver of SAVI and specifies TNFR1 blockage as a promising treatment to reduce disease severity. It thoroughly characterizes pharmacological inhibition and genetic perturbations of TNF signaling in murine SAVI models and creates a novel mouse model for studying TNF-targeted therapies in SAVI treatment.

*However, the study is limited in characterizing the discrepancy between pharmacological inhibition and genetic depletion of TNF and understanding the underlying mechanisms of TNF driving chronic STING activation and tissue inflammation. *

Advances: The study extends knowledge in the field by demonstrating that enhanced TNF signaling drives SAVI, establishing causation rather than mere correlation. The authors provide strong rationale for treating SAVI with TNF inhibitors/blockage, previously used in other autoimmune disorders like IBD or Crohn's disease, but not in SAVI. They also present a valuable genetic model for studying TNFR signaling blockage in SAVI progression, which is important for both the field of SAVI and future therapy development.

Audience: The research provides translational and clinical insights by suggesting that targeting TNFR1 signaling could inspire novel treatments for SAVI. The study also advances basic research on SAVI disease progression. Immunologists and clinicians studying and treating autoimmune disorders are the intended audience, but the findings have broader implications. The study highlights the potential role of TNF signaling in COVID-19 disease progression and treatment, thus attracting interest beyond the field of autoimmune disorders.

• *

Field of expertise:

cGAS-STING regulation in chromosomally unstable cancers, genomic instability, nuclear envelope rupture and repair

Do not have sufficient expertise in:

Immunological underpinning of autoimmune disorders, clinical models or manifestations of SAVI

• *

• *

Reviewer #3:

We thank the reviewer for his/her time and for the constructive comments. Below please find our detailed responses to your points.

• *

Uncontrolled activation of STING is linked to autoinflammatory disease "STING-associated vasculopathy with onset in infancy (SAVI)". The authors had previously published a mouse model of SAVI, which was generated by knocking in the disease causing variant N153S into the endogenous murine Sting1 gene (STING ki) (Luksch et.al., 2019). In the current study, the author further investigated the role of tumor necrosis factor (TNF) signaling in manifestation and progression of murine SAVI disease by using the approach of pharmacologic and genetic inhibition of TNF receptors TNFR1 and TNFR2. Overall, the authors were able to demonstrate the following novel findings:

• *

1) Infliximab treatment of STING ki mice significantly increased the number of blood CD8+ T cells and thymic cells count. The authors claimed that the pharmacological inhibition of TNF signalling has a partial rescue effect of T cell lymphopenia. However, pharmacologic inhibition of TNF signalling however has no effect on lung disease.

2) On the other hand, STING ki;Tnfr1-/- (lacking TNFR1) showed the similar modest rescue of the CD8+ T and CD4+ T cells in blood compared to the WT C57BL/6 (BL6) but not with STING ki;Tnfr2-/- (lacking TNFR2). STING ki;Tnfr1-/-, STING ki;Tnfr2-/- and STING ki;Tnfr1/2-/- had modest rescue of thymic cell count and reduced spleen cell count (reduced splenomegaly). Along with the rescued thymic content and reduced splenomegaly, genetic ablation of TNF signalling (STING ki;Tnfr1-/-) also prevented manifestation of severe inflammatory lung disease.

3) To investigate the role of lung endothelial cells in the development of interstitial lung disease, primary murine lung endothelial cells from STING WT, STING ki and STING WT;Tnfr1/2-/- and STING ki;Tnfr1/2-/- mice were isolated and bulk RNAseq was performed. This showed decreased level of several proinflammatory cytokines (e.g. Tnf, Il1b) and chemokines (e.g. Cxcl1, Cxcl2, Cxcl9, Cxcl10, Ccl2, Ccl3 and Ccl4) in STING ki mice lacking TNFR1/2 compared to STING ki mice.

4) Neutrophils were isolated from bone marrow and were added to cultured primary lung endothelial cell monolayers. The experiments demonstrated that the attachment and transmigration of neutrophil cells were dependent on expression of STING gain-of-function mutation in endothelial cells.

• *

A few points require clarification before publication of this study.

• Tnfr1-/-, Tnfr2-/- and Tnfr1/2-/- did not show any statistical significant improvement of thymic cell count in STING ki mice. As such, the statement in the conclusion/summary section of discussion regarding Tnfr1 can restore thymocyte numbers should be toned-down.
• Thank you for this suggestion. In Figure 4 E, we demonstrated that knock out of TNFR1 leads to increasing of SP CD8 thymocyte count and partially of SP CD4 thymocyte count (Fig. 4 D). In agreement with this suggestion, we marked this subpopulation of thymocytes in the discussion and summary section, see actual line 684 and see actual line 794.

2) The section on Neuroinflammation and neurodegeneration and dependency of TNFR1/2 signaling is very currently difficult to follow (based on how the data are presented in figures and text). This section requires to be re-written for clarity.

• *

Thank you for this suggestion. We re-wrote this section, see line 472 - 499.

Neuroinflammation and neurodegeneration in dependency of TNFR1/2 signaling

The extent of inflammation in mouse brain resulting from constitutive activation of STING N153S was reported by quantifying the density of Iba1-positive microglia (Fig.5 A). Consistent with our previous findings (Szego et al., 2022), the density of Iba1-positive microglia in the substantia nigra was higher in STING ki;BL6 mice than in STING WT mice (Fig.5 B). TNFR deficiency did not affect neuroinflammation because there was no significant difference between the density of Iba1-positive microglia between STING ki;BL6 mice and STING ki;Tnfr1/2-/- mice (Fig.5 B). This suggests that the TNF pathway is not required for STING-induced microglia activation in the substantia nigra.

In addition, we measured the extent of STING-induced astrogliosis by quantifying the density of GFAP-positive cells (Fig. 5 A). Consistent with our previous findings, the density of GFAP-positive astroglia was higher in STING ki than in STING WT mice (Fig. 5C). Yet, as for microglia, there was no significant difference between the density of GFAP-positive astroglia between STING ki;BL6 mice and STING ki;Tnfr1/2-/- mice (Fig.5 C), suggesting that the TNF pathway is not required for STING-induced astrogliosis in the substantia nigra.

Finally, we measured the extent of STING-induced neurodegeneration by quantifying the density of TH-positive dopaminergic neurons in the substantia nigra (Fig. 5A). As in our previous findings, the density of TH-positive neurons was lower in STING ki;BL6 mice than in STING WT mice (Fig.5 D). The density of TH-positive neurons in the substantia nigra of STING ki;Tnfr1/2-/- mice was higher than the density of TH-positive neurons in the substantia nigra of STING ki;BL6 mice (Fig. 5 D), suggesting that the STING-induced degeneration of TH-positive neurons was blunted in Tnfr1/2-/- mice and that TNFR1/2 are involved in the STING-induced degeneration of dopaminergic neurons.

Hence, there is a discrepancy between STING-induced effects on glial cells as opposed to STING-induced effects on neurons. The dependence of STING-induced neurodegeneration but not glial response on TNFR1/2 suggests that the STING-induced degeneration of dopaminergic neurons is not a direct consequence of microglia or astroglia activation. This is consistent with the emerging concept of a neuron-specific inflammatory response (Welikovitch et al., 2020).

*The powerful use of in vivo genetic KO models and TNF inhibitor makes this study a valuable contribution to the field - helping further decipher the importance of the NF-KB/TNF branch of STING in SAVI (knowledge gap). The audience for this work would be specialised to STING biology and potential clinical treatments of SAVI. *

• *

Our expertise is in nucleic acids sensing (such as STING) and auto-immunity.

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

Uncontrolled activation of STING is linked to autoinflammatory disease "STING-associated vasculopathy with onset in infancy (SAVI)". The authors had previously published a mouse model of SAVI, which was generated by knocking in the disease causing variant N153S into the endogenous murine Sting1 gene (STING ki) (Luksch et.al., 2019). In the current study, the author further investigated the role of tumor necrosis factor (TNF) signaling in manifestation and progression of murine SAVI disease by using the approach of pharmacologic and genetic inhibition of TNF receptors TNFR1 and TNFR2. Overall, the authors were able to demonstrate the following novel findings:

1. Infliximab treatment of STING ki mice significantly increased the number of blood CD8+ T cells and thymic cells count. The authors claimed that the pharmacological inhibition of TNF signalling has a partial rescue effect of T cell lymphopenia. However, pharmacologic inhibition of TNF signalling however has no effect on lung disease.
2. On the other hand, STING ki;Tnfr1-/- (lacking TNFR1) showed the similar modest rescue of the CD8+ T and CD4+ T cells in blood compared to the WT C57BL/6 (BL6) but not with STING ki;Tnfr2-/- (lacking TNFR2). STING ki;Tnfr1-/-, STING ki;Tnfr2-/- and STING ki;Tnfr1/2-/- had modest rescue of thymic cell count and reduced spleen cell count (reduced splenomegaly). Along with the rescued thymic content and reduced splenomegaly, genetic ablation of TNF signalling (STING ki;Tnfr1-/-) also prevented manifestation of severe inflammatory lung disease.
3. To investigate the role of lung endothelial cells in the development of interstitial lung disease, primary murine lung endothelial cells from STING WT, STING ki and STING WT;Tnfr1/2-/- and STING ki;Tnfr1/2-/- mice were isolated and bulk RNAseq was performed. This showed decreased level of several proinflammatory cytokines (e.g. Tnf, Il1b) and chemokines (e.g. Cxcl1, Cxcl2, Cxcl9, Cxcl10, Ccl2, Ccl3 and Ccl4) in STING ki mice lacking TNFR1/2 compared to STING ki mice.
4. Neutrophils were isolated from bone marrow and were added to cultured primary lung endothelial cell monolayers. The experiments demonstrated that the attachment and transmigration of neutrophil cells were dependent on expression of STING gain-of-function mutation in endothelial cells.

A few points require clarification before publication of this study.

1. Tnfr1-/-, Tnfr2-/- and Tnfr1/2-/- did not show any statistical significant improvement of thymic cell count in STING ki mice. As such, the statement in the conclusion/summary section of discussion regarding Tnfr1 can restore thymocyte numbers should be toned-down.
2. The section on Neuroinflammation and neurodegeneration and dependency of TNFR1/2 signaling is very currently difficult to follow (based on how the data are presented in figures and text). This section requires to be re-written for clarity.

Significance

The powerful use of in vivo genetic KO models and TNF inhibitor makes this study a valuable contribution to the field - helping further decipher the importance of the NF-KB/TNF branch of STING in SAVI (knowledge gap). The audience for this work would be specialised to STING biology and potential clinical treatments of SAVI.

Our expertise is in nucleic acids sensing (such as STING) and auto-immunity.

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

Short summary

In this paper, Luksch et al (2024) examines the role of TNF signaling in STING-associated vasculopathy with onset in infancy (SAVI). By using pharmacological inhibition and genetic inactivation of TNF receptors in a murine SAVI model (STING ki), the research found that pharmacologically inhibiting TNF signaling improved T cell lymphopenia but had limited effects on lung disease. Genetic inactivation of TNFR signaling, particularly TNFR1, enhanced thymocyte survival and expanded the peripheral T cell pool, reducing inflammation and neurodegeneration. The development and progression of severe lung disease in STING ki mice are also reliant on TNFR1 signaling, while TNFR2 deletion did not alleviate lung inflammation. The authors also explored the severe inflammatory lung disease manifestation, showing that primary lung endothelial cells in STING ki mice allowed more neutrophil attachment compared to those in STING WT mice, indicating chronic STING activity in endothelial cells disrupts the endothelial barrier and promotes severe lung disease. The study highlights TNFR signaling as crucial in SAVI and COVID-19 progression and suggests blocking TNFR1 signaling as a potential therapeutic approach for both diseases.

Major comments:

The paper establishes a strong connection between TNFR1 depletion and the reduction of SAVI disease severity in lung and neuroinflammation, suggesting TNFR1 blockade as a viable therapeutic strategy for SAVI. To strengthen the arguments and improve the therapeutic potential, the authors should address the following major comments: - The authors conclude that TNFR1 signaling drives murine SAVI disease, as evidenced by the reduced severity of lung disease in TNFR1 -/- mice. While the genetic model is convincing, the discrepancy between pharmacological inhibition and genetic models needs clarification. Before attributing the pharmacological failure to late administration, have the authors considered that Infliximab might not sufficiently deplete TNF to achieve therapeutic benefits? In figure 2H, serum TNF levels were not significantly altered in STING ki mice treated with Infliximab. Have the authors considered using other TNF inhibitors or alternative methods to measure TNF depletion efficacy in STING ki murine models, such as qPCR, flow cytometry, or immunohistochemistry in lymph nodes or lung tissues? - The TNF pathway exhibits redundancy, as multiple signaling molecules or pathways can compensate for the loss of TNF function to maintain cellular processes and immune responses. The authors showed that thymocytes of STING ki mice lacking TNFR1/2 expressed significantly lower levels of IFN-related genes (Cxcl10, Sting1), and mice lacking TNFR1 and TNFR1/2 expressed reduced levels of NF-κB-related genes. Does this imply that IFN and NF-κB pathways are downstream of TNF signaling driving SAVI progression? It would be valuable to hear the authors' comments or postulations on the potential mechanisms of TNF driving SAVI progression in the discussion, and the methods to dissect the mechanisms further using genetic or pharmacological methods. - The authors mentioned that the pharmacological inhibition of TNF by Infliximab is ineffective due to late administration compared to the onset of SAVI. How would this affect the therapeutic treatment of TNF if the treatment is going to be later than the disease onset? Can the authors elaborate on the potential ways to circumvent the timing of treatment? Would TNFR1 antagonists experience the same issue? To understand disease progression and optimal targeting times, the creation of an inducible TNFR1/2 -/- mouse model could be beneficial. This is optional, but the authors are encouraged to comment on improving TNFR1/2 -/- mouse SAVI models to further study the therapeutic potential of TNF signaling blockage in treating SAVI.

Minor comments:

• The authors separate STING WT and STING ki into different graphs, which can sometimes make it hard to compare STING WT and STING ki baseline levels. It would be beneficial to merge the two genotypes into single graphs for easier comparison.
• Figure S5 lacks statistical annotations, although the legends mention them. Are the statistics usually shown when a comparison is mentioned in the text, or are they only displayed when the differences are significant? It would be helpful if the authors could clarify this and ensure that all relevant statistical comparisons are clearly reflected in the graphs, regardless of the significance level. This consistency would improve the clarity and interpretation of the data presented.
• The authors did an excellent job discussing the study's implications, but some of this content could be moved to the introduction. The hypothesis that "tumor necrosis factor (TNF) signaling is involved in the manifestation and progression of murine SAVI disease" can be introduced more naturally once the authors present previous findings on TNF's association with various autoimmune disorders. This would set a clear context for the study's objectives and rationale.

Significance

General Assessment: The study identifies enhanced TNF signaling as a driver of SAVI and specifies TNFR1 blockage as a promising treatment to reduce disease severity. It thoroughly characterizes pharmacological inhibition and genetic perturbations of TNF signaling in murine SAVI models and creates a novel mouse model for studying TNF-targeted therapies in SAVI treatment. However, the study is limited in characterizing the discrepancy between pharmacological inhibition and genetic depletion of TNF and understanding the underlying mechanisms of TNF driving chronic STING activation and tissue inflammation.

Advances: The study extends knowledge in the field by demonstrating that enhanced TNF signaling drives SAVI, establishing causation rather than mere correlation. The authors provide strong rationale for treating SAVI with TNF inhibitors/blockage, previously used in other autoimmune disorders like IBD or Crohn's disease, but not in SAVI. They also present a valuable genetic model for studying TNFR signaling blockage in SAVI progression, which is important for both the field of SAVI and future therapy development.

Audience: The research provides translational and clinical insights by suggesting that targeting TNFR1 signaling could inspire novel treatments for SAVI. The study also advances basic research on SAVI disease progression. Immunologists and clinicians studying and treating autoimmune disorders are the intended audience, but the findings have broader implications. The study highlights the potential role of TNF signaling in COVID-19 disease progression and treatment, thus attracting interest beyond the field of autoimmune disorders.

Field of expertise:

cGAS-STING regulation in chromosomally unstable cancers, genomic instability, nuclear envelope rupture and repair

Do not have sufficient expertise in:

Immunological underpinning of autoimmune disorders, clinical models or manifestations of SAVI

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

Review Commons STING SAVI May 2024

STING is a key signalling hub in the innate immune system, receiving multiple inputs from upstream activators (such as cGAS) and in turn triggering multiple downstream events (such as IFN induction, NF-kB signalling, autophagy, cell death). Mutations in the STING gene cause a rare inflammatory disease called SAVI. Using a previously established STING ki mouse that recapitulates some of the clinical observations in SAVI patients, this manuscript tests the hypothesis that TNF signalling drives pathology. Using anti-TNF antibody and TNF receptor knockout, the authors show that TNF indeed plays important roles in causing disease in this mouse model. For example, the loss of T cells and neurons is prevented when TNF signalling is blocked, and lung pathology is rescued in STING ki mice lacking TNF receptors. Overall, the manuscript is well written and laid out, and the experimental work is of a high technical standard.

Major comments

1. Most figures show pooled data from two independent experiments including a total of 5-8 mice. Given the variability in some of the readouts, this raises the question of whether there is sufficient statistical power to draw conclusions. For example, in Figure 2, the conclusion that "Infliximab did not alter the expression of inflammatory mediators" seems questionable given the results in Figure 2F and G. Did the authors perform a power calculation? What effect size can the authors detect given the variability and number of replicates? Similarly, in Figure 3, the authors conclude that "Disruption of TNFR signaling did not significantly prevent T cell lymphopenia"; however, with some more replicates, the data in Figure 3D would likely reach significance. Similar concerns apply to several panels in Figures 4 and 6 and to Figure S5M. Ideally, the authors should perform additional repeat experiments to increase the number of replicates. If that is not possible, power calculations need to be provided and conclusions should explicitly mention the minimum effect size that the author can detect given the small sample size (for example "Infliximab did not alter the expression of inflammatory mediators more than x-fold").
2. The authors should not make unjustified overstatements. For example, STING KI; TNFR1/2 KO mice should not be referred to as a "new mouse model". The manuscript simply tests the role of TNFR1/2 in the already published STING N153S model. In line 687, avoid using "impressively" and in line 734 avoid using "massively".

Minor comments

1. Line 767-769: The statement that spike activates cGAS is misleading, because this effect is an indirect consequence of cell-to-cell fusion (Liu et al 2022).

Significance

The main strengths of this study are (1) the use of complementary antibody-based and genetic methods to test the role of TNF signalling; (2) the use of multiple different readouts; and (3) the analysis of many different cell types / organ systems. The main weaknesses are (1) small sample sizes limiting statistical power (see above) and (2) the exclusive use of mouse models.

Overall, my opinion is that the advance is important, both fundamentally and clinically. Studies of this and the related V154M mouse model previously showed an important role of non-IFN pathways in driving disease. This study indicates that TNF signalling may cause pathology. This not only extends our understanding of STING's role in autoinflammation but also opens a direct therapeutic avenue using approved TNF targeting drugs.

This study will be primarily of interest to specialised audiences working on STING and SAVI, and secondarily to the wider innate immunity field.

This reviewer has expertise in the field of nucleic acid sensing, including cGAS-STING.

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

We thank the reviewers for their comments and suggestions, which we think are helpful and will improve the manuscript, and intend to address with the changes and planned revisions below.

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

Bello et al look at the SNP rs28834970 associated with Alzheimer's disease (AD), with C being the risk allele, on chromatin accessibility and expression of a nearby gene, PTK2B, in microglia. Their contention is that the single SNP affects chromatin accessibility and binding of the transcription factor CEBP[beta] in an intronic region of PTK2B and thereby affects PTKB expression. I had a few questions that I think are critical to be addressed. Please note that my numbering of panels is based on the figures, not the legends, which do not seem to quite agree with each other. There are also some figure legends that say "IFNg" while the figures say "LPS", which should be fixed.

We apologise for the mistake in the figure legend that made this confusing, which we have now revised.

The abstract says that editing a line that is homozygous for protective alleles to homozygous for risk results in "subtle downregulation of PTK2B expression". It isn't clear to me that the presented data fully supports this contention, which is central to the argument of the paper. In figure 2e, the authors show in both RNAseq and ddPCR that there is numerically lower PTK2B expression but this is not indicated to be statistically significant by one-way paired ANOVA. If there is no nominally significant difference in the edited lines, compared to the proposed significant differences in lines carrying the full risk haplotype (figure 1), then it would not seem sensible to ascribe the effects to the single edited base pair.

We agree with the reviewer that given the effect of the SNP on PTK2B expression in the edited lines is small and only significant in macrophages, we should not interpret the effects to be mediated solely through PTK2B expression, and have substantially reworded the manuscript accordingly.

Whilst the effects in the eQTL analysis are significant, it is worth noting that this is likely due to the much larger number of donors (133-217) giving greater power to detect the subtle changes in expression (~1.1 to 2 fold in eQTL). This change is of a similar magnitude in our SNP edited lines (~1.2 fold in SNP edited lines) as would be expected of most common regulatory variants so we believe that it could be the primary causal variant. However, we cannot exclude that other variants in the haplotype could contribute to the effect, so have also reworded accordingly to make this clear.

Given this uncertainty about the overall strength of effect of the single base pair change it would seem important to evaluate the proposed mechanism of CEBPb binding. It wasn't clear whether the ATAC-seq data summarized in the volcano plot in 2C is proposed to be a cause or a consequence of the CEBPb binding change. I am assuming that the 'fold change' estimate here is CC compared to TT, which would be consistent with direction of effect in figure 1, but please clarify.

We apologise for the mistake in the figure legend that made this confusing, which we have now revised along with clarification in the revised text. It is difficult to be sure whether changes in chromatin accessibility are a cause or consequence of CEBPb binding, but the fact that the binding of CEBPb is increased in the CC allele (Fig 2a, Fig 2c), that the C allele better matches the consensus sequence (Fig 2b) and there is increased chromatin accessibility (Fig 2a, Supp Fig 3b) suggests that CEBPb binding is causing the formation of the region of chromatin accessibility.

In contrast to the subtle effects at PTK2B, the global transcriptional effects in figure 3 look quite strong. Are any of these changes dependent on PTK2B, that is to say, are they mimicked by partial suppression of PTK2B expression or activity?

We agree that the downstream effects of the SNP are much stronger than the effects on PTK2B expression, and we have substantially reworded the manuscript to make it clear that we are unsure that the effects of the SNP are all mediated via PTK2B.

However, we note that there is evidence in the literature of a loss in CCL4 and CCL5 expression upon PTK2B knockout in macrophages (https://www.nature.com/articles/s41467-021-27038-5) and inhibition of PTK2B in monocytes results in a reduction in CCL5 and CXCL1 (https://www.nature.com/articles/s41598-019-44098-2) consistent with our observations.

Experiments to manipulate PTK2B expression in microglia and readout changes at the RNA level would take a few months to complete, but we would be willing to do this if the reviewer felt this was necessary.

Finally, in figure 4, it should be clarified as to why lower expression of PTK2B would be expected to have a detrimental effect on Alzheimer's risk. If understood correctly, and again fixing the figure legends would be helpful, the CC edited lines (risk) have lower chemokine induction than the unedited TT lines.

We apologise for the error in this figure which we have corrected in the revised version. You are correct that the CC lines have a lower chemokine level in both unstimulated and stimulated cells, and we have now discussed further how this may be linked to increased disease risk.

"Even though overexpression of these chemokines is characteristic of neuroinflammation, correlated with disease progression and found in late stages of AD, knockout of chemokines, such as CCL2, and chemokine receptors, such as CCR2 and CCR5, in mice is associated with increased Aβ deposition and accumulation [47,50-52,107]. It has also been found that patients carrying CCR5Δ32 mutation, which prevents CCR5 surface expression, develop AD at a younger age[108]. Therefore, we hypothesize that in individuals carrying the C/C allele of rs28834970 downregulation of these chemokines in macrophages and microglia harbouring the C/C allele of rs28834970 affects Aβ-induced microglia chemotaxis, leukocytes recruitment and clearance of Aβ, and may increase the risk of developing symptomatic AD"

Reviewer #1 (Significance (Required)):

Going from GWAS hits, which represent blocks of high LD inherited variants, to single functional variants is a difficult problem in human genetics. The current paper attempts to isolate the effect of a single variant within an LD block on IPSC derived macrophages and microglia. This idea might be useful in nominating PTK2B as a therapeutic target for AD, although there is some question in my mind as to direction of effect.

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

SUMMARY: In this manuscript the authors explore the biological effects of an intronic SNP in the PTK2B gene, previously shown to be associated with late onset Alzheimer's disease (AD) risk. Based on the likely effect of the SNP locus on PTK2B expression in the macrophage lineage, the authors explore the consequences of introducing with the Crispr/Cas9 technique the biallelic SNP base change (C/C vs T/T) in a human IPSC line that is then differentiated into macrophages or microglia. They observe that C/C increases chromatin accessibility and CEBPb binding in comparison to T/T, with a slight decrease in PTK2B expression, significant in macrophages but not in microglia. The authors then investigate the transcriptome changes induced by the C/C mutation and find alteration in many genes, including a decreased expression of a number of cytokine or receptor proteins involved in inflammatory responses. The authors also mention a decreased effect on IFNg-induced reduced mobility but the data are missing (see Figure errors below). Overall the authors propose that the risk SNP is associated with a decreased PTK2B expression and hypothesize a link between this change and a decreased function of macrophages/microglia that may contribute to AD pathology.

MAJOR COMMENTS

1- The authors claim that their results show that the investigated SNP has a causal effects in "microglial function" (Title) and in Alzheimer's disease (AD) (Abstract 2nd sentence "Here we validate a causal single nucleotide polymorphism (SNP) associated with an increased risk of Alzheimer's disease". The word "causal" is repeated many times. However the authors should qualify their claim with respect to AD. Their results do show that the SNP has an effect on chromatin accessibility, CEBP binding, PTK2B expression and transcriptome, but the link between these changes is not formally demonstrated and their potential role in AD-like phenotype is not explored. The "causal" role is not formally and logically demonstrated. It remains an interesting, plausible hypothesis and the results provide strong arguments in support of that hypothesis but do not prove it, yet.

Concerning the title, "causal effects on microglial function" is awkward, anything that has effects is logically "causal" in these effects. The title should be "... has effects on microglial functions" or "... alters microglial function".

We agree with the reviewer that given the effect of the SNP on PTK2B expression in the edited lines is small and only significant in macrophages, we should not interpret the effects to be mediated solely through PTK2B expression, or that they cause AD. We have substantially reworded the manuscript throughout to account for this.

2- One major difficulty in the results is to link the slight decrease in PTK2B transcript, which is only significant in macrophages, with the rest of the phenotype. Because what matters to make this link is not the mRNA but the protein, and because mRNA levels are often not strictly correlated with the protein levels, the authors should measure the PTK2B/PYK2 protein levels in their differentiated cell lines in basal conditions and following activation (as they do for other readouts) using immunoblotting. A robust and significant diminution in PYK2 protein would strongly support its role in linking PTK2B expression and transcriptome change.

We have performed preliminary analyses of PTK2B expression by Western blot in these cell lines after differentiation, but were unable to observe a significant change in abundance in the edited cell lines. This is not unexpected given the results at the RNA level, since the effect size of this common regulatory variant is likely very small (estimated to be ~1.2 fold from the eQTL analysis), and likely within the variability of this assay.

As mentioned above, we have reworded the manuscript to avoid interpreting that the effects of rs28834970 are mediated solely through effects on PTK2B expression. We think that an experiment to manipulate PTK2B levels (see next point) may be a better way to demonstrate whether these effects are mediated through PTK2B expression.

An optional additional key experiment would be to reverse the transcriptome phenotype by increasing the expression of PTK2B (e.g. by cDNA transfection). Note that these points are important because an alternative hypothesis to explain the effects of C/C mutation on macrophage function would be that the C/C mutation has a long distance effect on other chromatin regions with key role in regulating these cells.

We agree that this would be a valuable experiment, and are planning additional experiments to investigate the effect of manipulating PTK2B levels (through knockout) on microglia.

3- The manuscript contains several errors in the figures and figure legends. In Fig. 2 the legends for the figure items are shuffled. Figure 4 and Supplementary Figure 5 are duplicates of the same one. Consequently important data are not presented.

We apologise for the errors in these figures that were due to a mistake during uploading where the incorrect versions were used. The legends for figure 2 and panels in figure 4 have now been corrected, and show the effects of rs28834970 on microglial migration and chemokine release in the presence or absence of IFNg.

4- When the number of replicates is small (e.g. n = 3) it is preferable to use non parametric tests (rank analysis, e.g. Mann Whitney's test) rather than t test. This applies to Figures 2D (current legend 2A), 2E (current legend 2B), Figure 4A-C, Supplementary Figures 2A, 2B. In Supplementary Fig 4E (MARCO) the number of replicates (presumably 3 because based on RNAseq) and the used test are not indicated. Is it the RNAseq statistical analysis?

We thank the reviewer for this comment. We acknowledge that the t-test may lead to inflated false discovery rates. However, it has been shown that for small sample sizes parametric tests have a power advantage compared to non-parametric ones that may outweigh the possibly exaggerated false positives. See https://genomebiology.biomedcentral.com/articles/10.1186/s13059-022-02648-4#Sec3 which states:

"In conclusion, when the per-condition sample size is less than 8, parametric methods may be used because their power advantage may outweigh their possibly exaggerated false positives."

We have also modified the legend of supplementary figure 4E to clarify the number of replicates used.

5- In addition to the above comment on tests, when the number of replicates is small it is not appropriate (and misleading) to show box plots or bars with SEM. In the indicated figures the individual data points should be shown.

We now show individual replicates on box plots (Figure 2D, 2E and supp figure 4E).

MINOR COMMENTS:

a- Macrophages and microglia are very similar cell types. Could the authors comment more on the differences they observe and how they are related to those previously described?

We have now referenced the original papers and commented on the markers that we see differentially expressed, notably P2RY12 which is a key homeostatic microglia marker that distinguishes these cells from macrophages.

b- In Fig. 2A CEBPb cut and run plot, the differences are not limited to the SNP immediate vicinity, there are also visible differences between T/T and C/C plots in at least a 40-kb range. Is it due to multiple interactions of CEBPb? How can the point difference have broad consequences? Please explain this potentially interesting and relevant finding.

Whilst there may be small changes in CEBPb binding at the second intronic PTK2B chromatin peak, this is not statistically significant given the variability between repeats. In fact, the only significant change we see in CEBPb binding genome-wide is at the locus overlapping the SNP (Fig 2c).

c- Potentially cis-altered genes near the SNP include CHRNA2 and EPHX2 (see Sup. Fig. 3a). Their expression may not be detected in macrophage lineage. If this is the case please indicate in the text, otherwise please include the corresponding data in Sup. Fig. 3b to show the presence or absence of SNP-induced change.

You are correct that CHRNA2 and EPHX2 are not expressed in our macrophages or microglia, and we have now explicitly stated this in the revised text.

d- In general the Figures are not of very high quality and are difficult to read or understand without constantly going back and forth to the legends (which are mislabeled in some instances). To improve:

. Please increase font size whenever possible.

. Please improve Fig. 1d by indicating the position of the SNP, numbering the exons (an intermediate scale plot may be necessary and lines on bottom trace are hardly visible).

. Please indicate the correct color code for T/T and C/C in Fig 3a and b, left panels, which currently doesn't match.

. Please label the Venn's diagrams comparisons in Sup. Fig. 4b.

. In the text and legends the Figure items are identified with letters in upper case, in the figures they are in lower case. Please be consistent.

We have improved the resolution of the images in the pdf and Fig 1d has been revised to include the position of the SNP. The colour code for T/T and C/C is correct in fig 3a and 3b, but since the PCA plots are independently created, we would not always expect the position of the T/T and C/C alleles to be the same. The Venn diagrams in Sup Fig 4b have been updated, and the letters for the figure panels made consistently upper case throughout.

e- In Fig. 2D and 2E, the Y axes should start at zero to avoid artificially increasing the visual differences. If there is a strong reason not to do so (I don't see any here), the Y axis should be clearly interrupted to avoid confusion.

We have altered this accordingly.

f- In the introduction the authors provide some background about previous work about the potential role of PTK2B/PYK2 in AD pathophysiology. The cited preclinical results suggest that PTK2B activity could have a deleterious effect (references in the manuscript). In contrast, some other reports (PMID: 29803828, 33718872) suggest a protective effect of PTK2B/PYK2. Because the evidence in the current manuscript suggests that the risk-associated SNP results in a decreased function of PTK2B/PYK2 (through decreased levels), at least in cells of the macrophage lineage, the authors could broaden their discussion to include these results.

We have now discussed the conflicting evidence in the revised manuscript.

Reviewer #2 (Significance (Required)):

ADVANCE: Late onset Alzheimer's disease is a major medical issue. It has a complex genetic risk component with many associated loci identified in GWAS. Most of these have only a small individual impact on the risk. One of the SNPs associated with increased risk (rs28834970) is located in an intron of the PTK2B gene. Although various reports have investigated the role of the PTK2B gene product, the tyrosine kinase PYK2, in several AD models, the possible link with rs28834970, is unclear.

An important point is to determine whether TàC SNP corresponding to rs28834970 alters PTK2B expression and how it does so. An alternative hypothesis could be that the SNP has a strong linkage disequilibrium with an unidentified allele in human populations that could be responsible for AD risk. The current manuscript is a significant step forward in addressing that question. By generating a biallelic C/C SNP mutation in a human IPSC line the current study allows to eliminate such linked contribution.

The strength of the manuscript is to show an effect on chromatin accessibility, CEBP binding and possibly PTK2B transcripts. It also provides interesting evidence of a broad effect of the C/C mutation on the transcriptome of macrophage lineage cells. In its current form the manuscript presents weaknesses that could be improved. These flaws include issues with the presentation discussed above and the uncomplete demonstration that it is the decrease in PTK2B expression that causes the macrophage/microglia phenotype. If these flaws were overcome the paper would represent a significant advance.

AUDIENCE: The expected audience is specialized in AD with a possible broader range if all weaknesses are addressed.

REVIEWER EXPERTISE: Basic science close to the field.

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

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

Summary: In this manuscript the authors explore the biological effects of an intronic SNP in the PTK2B gene, previously shown to be associated with late onset Alzheimer's disease (AD) risk. Based on the likely effect of the SNP locus on PTK2B expression in the macrophage lineage, the authors explore the consequences of introducing with the Crispr/CAS9 technique the biallelic SNP base change (C/C vs T/T) in a human IPSC line that is then differentiated into macrophages or microglia. They observe that C/C increases chromatin accessibility and CEBPb binding in comparison to T/T, with a slight decrease in PTK2B expression, significant in macrophages but not in microglia. The authors then investigate the transcriptome changes induced by the C/C mutation and find alteration in many genes, including a decreased expression of a number of cytokine or receptor proteins involved in inflammatory responses. The authors also mention a decreased effect on IFNg-induced reduced mobility but the data are missing (see Figure errors below). Overall the authors propose that the risk SNP is associated with a decreased PTK2B expression and hypothesize a link between this change and a decreased function of macrophages/microglia that may contribute to AD pathology.

Major comments:

1. The authors claim that their results show that the investigated SNP has a causal effects in "microglial function" (Title) and in Alzheimer's disease (AD) (Abstract 2nd sentence "Here we validate a causal single nucleotide polymorphism (SNP) associated with an increased risk of Alzheimer's disease". The word "causal" is repeated many times. However the authors should qualify their claim with respect to AD. Their results do show that the SNP has an effect on chromatin accessibility, CEBP binding, PTK2B expression and transcriptome, but the link between these changes is not formally demonstrated and their potential role in AD-like phenotype is not explored. The "causal" role is not formally and logically demonstrated. It remains an interesting, plausible hypothesis and the results provide strong arguments in support of that hypothesis but do not prove it, yet. Concerning the title, "causal effects on microglial function" is awkward, anything that has effects is logically "causal" in these effects. The title should be "... has effects on microglial functions" or "... alters microglial function".
2. One major difficulty in the results is to link the slight decrease in PTK2B transcript, which is only significant in macrophages, with the rest of the phenotype. Because what matters to make this link is not the mRNA but the protein, and because mRNA levels are often not strictly correlated with the protein levels, the authors should measure the PTK2B/PYK2 protein levels in their differentiated cell lines in basal conditions and following activation (as they do for other readouts) using immunoblotting. A robust and significant diminution in PYK2 protein would strongly support its role in linking PTK2B expression and transcriptome change. An optional additional key experiment would be to reverse the transcriptome phenotype by increasing the expression of PTK2B (e.g. by cDNA transfection). Note that these points are important because an alternative hypothesis to explain the effects of C/C mutation on macrophage function would be that the C/C mutation has a long distance effect on other chromatin regions with key role in regulating these cells.
3. The manuscript contains several errors in the figures and figure legends. In Fig. 2 the legends for the figure items are shuffled. Figure 4 and Supplementary Figure 5 are duplicates of the same one. Consequently important data are not presented.
4. When the number of replicates is small (e.g. n = 3) it is preferable to use non parametric tests (rank analysis, e.g. Mann Whitney's test) rather than t test. This applies to Figures 2D (current legend 2A), 2E (current legend 2B), Figure 4A-C, Supplementary Figures 2A, 2B. In Supplementary Fig 4E (MARCO) the number of replicates (presumably 3 because based on RNAseq) and the used test are not indicated. Is it the RNAseq statistical analysis?
5. In addition to the above comment on tests, when the number of replicates is small it is not appropriate (and misleading) to show box plots or bars with SEM. In the indicated figures the individual data points should be shown.

Minor comments:

• a. Macrophages and microglia are very similar cell types. Could the authors comment more on the differences they observe and how they are related to those previously described?
• b. In Fig. 2A CEBPb cut and run plot, the differences are not limited to the SNP immediate vicinity, there are also visible differences between T/T and C/C plots in at least a 40-kb range. Is it due to multiple interactions of CEBPb? How can the point difference have broad consequences? Please explain this potentially interesting and relevant finding.
• c. Potentially cis-altered genes near the SNP include CHRNA2 and EPHX2 (see Sup. Fig. 3a). Their expression may not be detected in macrophage lineage. If this is the case please indicate in the text, otherwise please include the corresponding data in Sup. Fig. 3b to show the presence or absence of SNP-induced change.
• d. In general the Figures are not of very high quality and are difficult to read or understand without constantly going back and forth to the legends (which are mislabeled in some instances). To improve:
  • Please increase font size whenever possible.
  • Please improve Fig. 1d by indicating the position of the SNP, numbering the exons (an intermediate scale plot may be necessary and lines on bottom trace are hardly visible).
  • Please indicate the correct color code for T/T and C/C in Fig 3a and b, left panels, which currently doesn't match.
  • Please label the Venn's diagrams comparisons in Sup. Fig. 4b.
  • In the text and legends the Figure items are identified with letters in upper case, in the figures they are in lower case. Please be consistent.
• e. In Fig. 2D and 2E, the Y axes should start at zero to avoid artificially increasing the visual differences. If there is a strong reason not to do so (I don't see any here), the Y axis should be clearly interrupted to avoid confusion.
• f. In the introduction the authors provide some background about previous work about the potential role of PTK2B/PYK2 in AD pathophysiology. The cited preclinical results suggest that PTK2B activity could have a deleterious effect (references in the manuscript). In contrast, some other reports (PMID: 29803828, 33718872) suggest a protective effect of PTK2B/PYK2. Because the evidence in the current manuscript suggests that the risk-associated SNP results in a decreased function of PTK2B/PYK2 (through decreased levels), at least in cells of the macrophage lineage, the authors could broaden their discussion to include these results.

Significance

Advance: Late onset Alzheimer's disease is a major medical issue. It has a complex genetic risk component with many associated loci identified in GWAS. Most of these have only a small individual impact on the risk. One of the SNPs associated with increased risk (rs28834970) is located in an intron of the PTK2B gene. Although various reports have investigated the role of the PTK2B gene product, the tyrosine kinase PYK2, in several AD models, the possible link with rs28834970, is unclear.

An important point is to determine whether TC SNP corresponding to rs28834970 alters PTK2B expression and how it does so. An alternative hypothesis could be that the SNP has a strong linkage disequilibrium with an unidentified allele in human populations that could be responsible for AD risk. The current manuscript is a significant step forward in addressing that question. By generating a biallelic C/C SNP mutation in a human IPSC line the current study allows to eliminate such linked contribution.

The strength of the manuscript is to show an effect on chromatin accessibility, CEBP binding and possibly PTK2B transcripts. It also provides interesting evidence of a broad effect of the C/C mutation on the transcriptome of macrophage lineage cells. In its current form the manuscript presents weaknesses that could be improved. These flaws include issues with the presentation discussed above and the uncomplete demonstration that it is the decrease in PTK2B expression that causes the macrophage/microglia phenotype. If these flaws were overcome the paper would represent a significant advance.

Audience: The expected audience is specialized in AD with a possible broader range if all weaknesses are addressed.

Reviewer Expertise: Basic science close to the field.

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

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Bello et al look at the SNP rs28834970 associated with Alzheimer's disease (AD), with C being the risk allele, on chromatin accessibility and expression of a nearby gene, PTK2B, in microglia. Their contention is that the single SNP affects chromatin accessibility and binding of the transcription factor CEBP[beta] in an intronic region of PTK2B and thereby affects PTKB expression. I had a few questions that I think are critical to be addressed. Please note that my numbering of panels is based on the figures, not the legends, which do not seem to quite agree with each other. There are also some figure legends that say "IFNg" while the figures say "LPS", which should be fixed.

The abstract says that editing a line that is homozygous for protective alleles to homozygous for risk results in "subtle downregulation of PTK2B expression". It isn't clear to me that the presented data fully supports this contention, which is central to the argument of the paper. In figure 2e, the authors show in both RNAseq and ddPCR that there is numerically lower PTK2B expression but this is not indicated to be statistically significant by one-way paired ANOVA. If there is no nominally significant difference in the edited lines, compared to the proposed significant differences in lines carrying the full risk haplotype (figure 1), then it would not seem sensible to ascribe the effects to the single edited base pair.

Given this uncertainty about the overall strength of effect of the single base pair change it would seem important to evaluate the proposed mechanism of CEBPb binding. It wasn't clear whether the ATAC-seq data summarized in the volcano plot in 2C is proposed to be a cause or a consequence of the CEBPb binding change. I am assuming that the 'fold change' estimate here is CC compared to TT, which would be consistent with direction of effect in figure 1, but please clarify.

In contrast to the subtle effects at PTK2B, the global transcriptional effects in figure 3 look quite strong. Are any of these changes dependent on PTK2B, that is to say, are they mimicked by partial suppression of PTK2B expression or activity?

Finally, in figure 4, it should be clarified as to why lower expression of PTK2B would be expected to have a detrimental effect on Alzheimer's risk. If understood correctly, and again fixing the figure legends would be helpful, the CC edited lines (risk) have lower chemokine induction than the unedited TT lines.

Significance

Going from GWAS hits, which represent blocks of high LD inherited variants, to single functional variants is a difficult problem in human genetics. The current paper attempts to isolate the effect of a single variant within an LD block on IPSC derived macrophages and microglia. This idea might be useful in nominating PTK2B as a therapeutic target for AD, although there is some question in my mind as to direction of effect.

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 manuscript " Phosphoproteomic analysis reveals the diversity of signaling behind ErbB inhibitor-induced phenotypes" authored by Drs. Katri Vaparanta, Anne Jokilammi, Johannes Merilahti, Johanna Örling, Noora Virtanen, Cecilia Sahlgren, Klaus Elenius and Ilkka Paatero was reviewed in Review Commons, and we carried out a full revision based on the received reviewer comments.

The comments from three reviewers and our point-by-point reply is here below. After each of reviewer´s comment, our reply is formatted in bold.

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

In this study, Vaparanta and co-workers used zebrafish embryos as model to analyze the impact of ErbB tyrosine kinase inhibitors on signaling pathways at the whole organism level. Experimentally, zebrafish embryos were exposed for 1 hour to a single dose of 3 different ErbB tyrosine kinase inhibitors and the global phosphoproteome of the embryos was analyzed by MS/MS. The authors show that the 3 inhibitors differentially modulate the activity of PI3K/Akt, p38 MAPK, Notch, Hippo-YAP/TAZ and β-catenin signaling pathways, associated with different neurological and myocardial phenotypic changes. Using small molecule inhibitors of selective signaling pathways, they show that perturbation of different signaling pathways may induce similar phenotypes in zebrafish embryos.

Specific comments:

1. The observation that exposure of zebrafish embryos to lapatinib, gefitinib and AG1478 leads to different global phosphoproteomic changes and to differential modulation of cellular signaling pathways was predictable and supported by an abundant literature. These 3 inhibitors differentially inhibit ErbB homo- and heterodimers and hit many other kinases. This point should be discussed in the paper.

Indeed, the kinase inhibitors do have different selectivity for the ErbB family kinases as pointed out by the reviewer. We have now discussed this point in the manuscript (new Supplemental Table 1) and added additional data from embryos treated with different ErbB kinase inhibitors with similar selectivity profiles into the manuscript (new Supplemental Figure 3-4). The ErbB family kinase selectivity profile of the inhibitors, however, does not fully explain why treatment with lapatinib (EGFR/ErbB2 inhibitor) induced the most unique phosphoproteomic changes from AG1478 (EGFR/ErbB2/ErbB4 inhibitor) and gefitinib (EGFR inhibitor) treatment in zebrafish embryos. This point is now discussed in the manuscript.

AG1478 is a first-generation tyrphostin while gefitinib and lapatinib are FDA-approved drugs. These compounds not only have different selectivity profiles, but also different pharmacological properties. Do the authors have any information about the permeability, distribution or concentration of the compounds in zebrafish embryos? Otherwise, how can they compare their effects?

__The reviewer points out correctly, that not only selectivity but also several other parameters could differ between compounds. The logic of our experimentation was to utilize differences in the properties of inhibitors to get new insights into underlying biological processes. These utilized differences could arise from not only selectivity but also as well from pharmacokinetic and –dynamic properties. Although it can be useful to understand these differences, this information per se is not needed to identify differentially regulated pathways that could affect the studied phenotypes. This is now better clarified in the discussion section. Our data indicates that ErbB inhibition profile explains a significant proportion, but not all, of observed signaling differences (Supplemental Fig. 3C). __

One major limitation of this study is that phosphoproteomic analysis was performed at a single time point and with a single dose of inhibitor, which compromises the interpretation of the findings. How was the dose of each inhibitor selected?

The doses were chosen based on our previous work (Paatero et al, 2019; Vaparanta et al, 2023), where with these inhibitor concentrations we were able to maximize the phenotypic effects without causing significant mortality. This is now mentioned in the results section of the manuscript. Higher dosages were lethal for the embryos, especially of AG1478, which is why a lower concentration of this inhibitor was used. The higher toxicity of AG1478 at lower concentrations compared to other ERBB inhibitors has also been previously noted by another group (Pruvot et al, 2014). Similar concentrations of the inhibitors have also been previously used by other groups with zebrafish embryos (Tran et al, 2007; Gallardo et al, 2015; Zhang et al, 2021; Du et al, 2024)__. __

One approach for better exploiting the data would be to correlate changes in phosphopeptides with the kinome selectivity of the inhibitors.

Indeed, we have now correlated our results from these inhibitors with other ErbB inhibitors of similar ErbB family kinase selectivity. The phosphoproteomic changes induced by inhibitors with similar ErbB family kinase selectivity significantly correlate (P = 0.0002, r:0.80 ,R2:0.65, Supplemental Fig. 3C) indicating that the ErbB selectivity plays a major role in determining the phosphoproteomic changes induced by these inhibitors. We also performed a correlation analysis between dimensionality-reduced phosphoproteomic changes and inhibitor selectivity. There was no significant correlation between the changes in the phosphoproteome and the ERBB selectivity of the inhibitors (P=0.1551, One-tailed Pearson correlation). Taken together, these results indicate that while the phosphoproteomic changes induced by these inhibitors can be reproduced by other inhibitors with similar ERBB selectivity profiles, inhibiting only a subset of the ERBB kinases (especially EGFR and ERBB2, but not ERBB4) produces a unique signaling signature that is not recapitulated with pan-ERBB inhibitor treatment. This information may be of interest since both lapatinib (EGFR/ERBB2 inhibitor) and neratinib (pan-ERBB inhibitor) are both used in the clinic to treat HER2-positive breast cancer. Our data indicates that the administration of these inhibitors to patients will likely have a differential global effect on cell signaling.

In the same vein, the signaling inhibitors used in Fig. 4 to dissect the phenotypic impact of distinct signaling pathways are non-selective, precluding any rigorous interpretation of the data. This confounding factor should at least be discussed in the manuscript. Again, the choice of the different doses of inhibitors is not justified.

Indeed, like all inhibitors, the inhibitors we utilized in Figure 4 can have some off-target effects. We aimed to use the concentration known by previous literature to have a measurable effect on the physiology of the zebrafish embryo (Fujii et al, 2000; Geling et al, 2002; Vasilyev et al, 2012; Jiang et al, 2023)__. These concentrations for different inhibitors were different in the literature, which is why different concentrations of the different inhibitors were used. We couldn’t find a reference for the concentration for VT-103, so a 30µM concentration was selected. With this concentration, the size of the embryo hearts was significantly reduced (P

The effect of inhibitors on the motility of embryos appears variable. For example, lapatinib markedly decreases motility in Fig. 4E but has no effect in Fig. 4F. Any explanation?

Different inhibitor concentrations were used in Figure 4E and Figure 4F. This has been now more clearly indicated in the manuscript in the results section and the figure legend. The lower inhibitor dosages in Figure 4F were to reduce the mortality and allow motility analyses of the embryos treated with a combination of the inhibitors analyses to facilitate observation of potential synergistic actions of inhibitors in co-treated embryos.

The conclusion that ErbB inhibitors induce similar phenotypes by perturbing different signaling pathways is not justified.

We have now softened our conclusions in the manuscript in the results section by replacing the sentence:” Taken together, these results suggest that AG1478 and lapatinib induce similar phenotypes by partially perturbing different signaling pathways in zebrafish embryos.” With the sentences: ” Taken together, these results suggest that AG1478 and lapatinib induce similar phenotypes but perturb different signaling pathways. Inhibition of these pathways induce similar phenotypes to lapatinib or AG1478 treatment in zebrafish embryos.”.

I have a few suggestions which could enhance the study's contribution to the field-

1. The rationale for this study should be elaborated further. What new information is expected to emerge from these studies, independently of the conceptual and technical limitations outlined above?

We have now further elaborated the rationale of the study in the introduction section.

The advantage of studying the whole organism instead of selected tissues is questionable. Analyzing a mixture of organs may mask subtle and physiologically relevant alterations of signaling pathways in specific tissues.

We agree with the reviewer that if the researcher’s interests reside in a specific tissue then a more targeted approach should be applied to probe the phosphoproteome of this tissue. However, sometimes a more global view of the inhibitor effects is required especially when it is unknown which tissues are affected by the inhibitor treatment. Ideally, the global approach would be followed by a more targeted approach on the tissues that are indicated to be affected by the inhibitor. One must also consider the feasibility, time consumption and costs of probing all tissues separately. If only the targeted approach is applied, the information on what pathway activities are globally most affected in the organism by the inhibitor treatment can be hard to estimate.

Can the authors correlate neurological and myocardial phenotypes extrapolated from their study with pharmacological effects observed in mice or humans treated with these compounds?

__We have now correlated our findings in the discussion section with the previous literature on the phenotypes of ErbB inhibitor-treated and ErbB receptor knock-out mice and with the reported adverse effects of ErbB inhibitor treatment in the clinic. __

Reviewer #1 (Significance (Required)):

The authors show that the 3 inhibitors differentially modulate the activity of PI3K/Akt, p38 MAPK, Notch, Hippo-YAP/TAZ and β-catenin signaling pathways, associated with different neurological and myocardial phenotypic changes. Using small molecule inhibitors of selective signaling pathways, they show that perturbation of different signaling pathways may induce similar phenotypes in zebrafish embryos.

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

In this study, the authors assess the effects of various ErbB receptor family tyrosine kinase inhibitors on the phosphoproteome of late embryonic and early larval stages of zebrafish. MS, Western blotting, and analysis of a transgenic zebrafish Notch signaling reporter line data suggest differential but overlapping effects of treatment with gefitinib, lapatinib and AG1478. Selected deregulated pathways are further assessed using a range of candidate downstream pathway-targeting inhibitors. Inhibitor treatment followed by quantification of spontaneous larval motility and heart ventricle wall area, which were previously found by the authors to be affected by AG1478 and lapatinib treatment, identifies involved downstream signaling pathways.

Major comments:

While I do not question the validity of the presented data showing phosphoproteome perturbations resulting from the performed ErbB inhibitor treatments, the treatment regimens used to assess the differential effects of the compounds may be insuffient to substantiate general statements comparing the phenotypic and phosphorylation effects of lapatinib, gefitinib and AG1478 beyond the effects of the specific doses applied to the embryo media. Unless directly quantified, it is difficult to reliably predict the in vivo dose resulting from drug administered to the embryo medium, and therefore a dose may be too high or too low for drug-to-drug comparison. Rationale for chosen dose of drugs should be provided. If available, inclusion of quantitative data on the drug-induced change in phosphorylation status of the drug target(s) is encouraged, and the discussion of the phosphoproteomic and phenotypical data should include this information.

The reviewer points out correctly, that not only selectivity but also several other parameters could differ between compounds. The logic of our experimentation was to utilize differences in the properties of inhibitors to get new insights into underlying biological processes. These utilized differences could arise from not only selectivity but also as well from pharmacokinetic and –dynamic properties. Although it can be useful to understand these differences, this information per se is not needed for the identification of differentially regulated pathways that could affect the studied phenotypes. This is now better clarified in the manuscript.

The rationale for the chosen drug doses has now been added to the manuscript in the results section. We used drug concentrations that were known to produce a phenotypic effect without causing significant mortality in the zebrafish embryos.

The ErbB receptors themselves are expressed at low levels, and unfortunately, we couldn´t reliably observe phosphopeptides of ErbB tyrosine autophosphorylation sites. To address this issue from a different angle, we treated embryos with other ErbB inhibitors exhibiting similar ErbB inhibition profiles as AG1478, lapatinib, and gefitinib (Supplemental Figure 3-4). This data indicates that the ErbB inhibition profile correlates quite well with the observed changes in the downstream signaling pathways p38, pAkt, pErk and Notch (Figure 3C and 4C).

Husbandry: The statement that "Zebrafish were maintained (...) following standard procedures." is insufficient without a specific reference. Please provide details on water quality parameters, temperature, light/darkness cycle and feeding regimen.

The requested information has now been added to the manuscript.

Western analysis: How many embryos were pooled in each sample? Please specify standard protocol or provide reference.

We have now amended the western analysis chapter in the materials and methods section as suggested by the reviewer. Five embryos were pooled for each sample.

Ventricle growth assay: The method of ventricle wall quantification is insufficiently described and might result in unnecessarily high variation. At which stage of the cardiac contraction-relaxation cycle were ventricle wall thickness and ventricle area measured? The confounding effect of contraction could be avoided altogether by stopping the heartbeat pharmacologically e.g. by administration of blebbistatin or verapamil. Subtracting ventricle lumen area from total ventricle area seems a much more direct measure of ventricle wall area than the estimation obtained by multiplying ventricle wall thickness with ventricle area.

We apologize for the mistake in the materials methods section, where we had written area instead of perimeter. We have now amended the ventricle growth assay chapter in the materials and methods sections and added more details on the ventricle wall area estimation. The ventricle wall area was measured from high-speed movies in diastole and systole, and the average perimeter over these states was reported. The ventricle wall thickness was only measured in systole. We chose this quantification method since the lumen area is difficult to estimate in the systole.

Phosphopeptide enrichment: How many embryos per sample? Final DMSO concentration is not stated.

__Twenty embryos per sample and 1% DMSO was used. This information is now included in the materials and methods section. __

P-values are presented for comparison of select groups only and a statement that e.g. only P-values We have added the recommended statement and the mean/median value with deviation values for the data indicated by the reviewer in the figure legends.

Minor comments:

Overall, the manuscript is well written and data and methods are well presented.

The relevant targets within the ErbB family of receptors should be introduced including information on well-established functions and downstream signaling pathways to enable the non-specialist reader to place the presented data in the context of known gene and protein function. Furthermore, conservation of target proteins in zebrafish should be touched upon.

We have now rewritten the introduction and results sections to include information on the ErbB family kinase selectivity of these inhibitors, the well-established functions and the target downstream pathways of ErbB receptors. We have now performed a multiple sequence alignment on the kinase domain of the ErbB receptors in human and zebrafish to estimate the conservation of the inhibitor targets in the zebrafish model. Human ErbB kinase domains had a high 86+/-9% sequence identity with zebrafish counterparts (Supplemental Figure 2) compared to 67+/-14% identity with the other ErbB kinase domain sequences in zebrafish (P=0.012).

Given different target profiles of the tested drugs among receptors of the ErbB family, differences in protein phosphorylation perturbations and in treatment-induced phenotypes may not be unexpected. Statements such as: "An unexpectedly large cluster of phosphopeptides that were increased in lapatinib-treated embryos but reduced in AG1478 and gefitinib treated embryos was detected" and "AG1478 and lapatinib may induce similar phenotypes by partially perturbing different signaling pathways in zebrafish embryos" should be discussed in the context of known drug target(s) and their functions.

We have now rewritten these statements as suggested by the reviewer and the target profiles are now discussed in the manuscript.

**Referee Cross-commenting**

I agree with the other reviewers on almost all points.

1) While the sensitivity to smaller or highly local effects is most likely reduced using the whole organism approach compared to e.g. single tissue analysis, I do believe that it is highly relevant due to its ability to identify potential effects beyond a single tissue or organ.

2) I maintain that while the presented data nicely show the effects of each administered dose of the individual compounds, the data does not allow for meaningful drug-to-drug comparisons without quantitative information on in vivo dose or direct target effect. If such information cannot be included, cross-drug conclusions and discussion should be done very carefully.

Reviewer #2 (Significance (Required)):

The evaluation of systemic molecular and phenotypic consequences of anti-cancer drugs in a vertebrate model system represents a relevant advancement. Although drug effects are likely to differ somewhat between embryonic and larval zebrafish and human cancer patients, the authors' comparison of obtained zebrafish data with human data supports translatability of the presented phosphoproteomics data. Also, the presented data pose a relevant advancement facilitating the informed use of the tested inhibitors as tools in basic science.

Expertise: Molecular biology, signaling, zebrafish. Limited expertise in omics data analysis and pharmacology.

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

The authors evaluated selected EGFR inhibitors developed as targeted cancer therapeutics, using zebrafish embryos and larvae as an in vivo model system. They performed mass spectrometry to analyze phosphorylation levels in target proteins, in combination with western blotting and gene set enrichment analyses; using this data, they assessed overlap between the inhibitors and overlap with known human data. They also performed imaging and locomotion analyses to assess alterations in phenotypes and phosphorylation-dependent signaling due to the inhibitor(s). The study generates novel information that is potentially relevant to the toxicity and efficacy of clinically used kinase inhibitors.

• The statistical analyses are appropriate to the data and the experimental design.

• The claims made by the authors are consistent with the data. In my opinion, the following revisions are needed for the manuscript to be accepted for publication:

• There is no mention of Gefitinib in the Abstract; please include it.

Gefitinib is now included in the Abstract.

Please state the target selectivity profiles (from known preclinical and/or clinical data) of the three inhibitors used.

__These are now presented in supplemental data (Supplemental Table 1), and analysed in relation to the observed signaling changes (Supplemental Figures 3 and 4). __

Please clarify whether the residues mentioned in the phospho-specific antibody data refer to zebrafish or human proteins.

Residues refer to human proteins as they are more widely used. This is now more clearly indicated in the materials and methods section.

Please state whether the pan-antibodies corresponding to the phospho-specific antibody targets were used, and mention any problems associated with their use. This will help readers not familiar with antibody use in zebrafish experiments. It will also help emphasize the value of mass spectrometric analysis in zebrafish protein work.

__As pointed out, the target specificity of antibodies is not often defined in zebrafish models on residue level, and phospho-specific antibodies may bind several closely related targets. The availability of robustly validated antibodies for zebrafish work, especially for phosphospecific epitopes, is quite limiting and therefore other, non-antibody-based techniques would be highly useful. This is now discussed in the manuscript. The phosphorylation site-specific antibodies used in this study indeed recognize the phosphorylated residue in several protein family members which further complicates the result interpretation. This is less of a limitation in the DIA-MS based phosphoproteomics approach which is now additionally discussed in the manuscript. __

Please attempt to describe the clinically documented cardiovascular and neurological effects of the inhibitors and any correlation(s) with your data. This will enhance the impact of the study.

See our reply for reviewer#1, comment 3.

**Referee Cross-commenting**

The common points raised in all the Reviews are the following:

1. The rationale of the study should be described in more detail, especially the utility of zebrafish as an in vivo model, addressing its advantages and limitations.

This is now discussed more extensively in the manuscript.

The findings need to be described in the context of the target selectivity profiles and clinical effects of the inhibitors, especially the approved inhibitors (Gefitinib and Lapatinib).

We have added data on target selectivity profiles (Supplemental Table 1), target conservation (Supplemental Figure 2) and also compared our observations to zebrafish embryos treated with other ErbB inhibitors with similar ErbB selectivity profiles (Supplemental Figure 3 and 4).

1. In my opinion, while the comments regarding target site drug concentration (within the embryos/larvae) and dose-response are relevant, I consider these experiments to be appropriate in a more detailed follow-up study.

We agree with the reviewer that the comprehensive pharmacokinetic studies fall outside the scope of this manuscript. As discussed before, in this manuscript we utilize differential inhibitor properties to gain new insight into phenotypes and underlying biological processes. This logic works even if the differences arise from properties other than the target selectivity.

One of the main value additions of the study is that it highlights a useful alternative to conventional strategies used in preclinical cellular and mammalian model studies of kinase inhibitors. I would urge the authors to discuss specific future directions, giving due importance to all the reviewers' comments.

This is now more extensively elaborated in the discussion section.

Reviewer #3 (Significance (Required)):

The experiments are well-described and provide sufficient information and detail for readers to understand and reproduce.

The study is highly relevant to the use of zebrafish as a whole-organism model for in vivo evaluation of drugs, specifically kinase inhibitors.

References

Du K, Liu Y, Zhang L, Peng L, Dong W, Jiang Y, Niu M, Sun Y, Wu C, Niu Y et al (2024) Lapatinib combined with doxorubicin causes dose-dependent cardiotoxicity partially through activating the p38MAPK signaling pathway in zebrafish embryos. Biomed Pharmacother 175. doi:10.1016/J.BIOPHA.2024.116637.

Fujii R, Yamashita S, Hibi M, Hirano T (2000) Asymmetric p38 activation in zebrafish: Its possible role in symmetric and synchronous cleavage. Journal of Cell Biology 150. doi:10.1083/jcb.150.6.1335.

Gallardo VE, Varshney GK, Lee M, Bupp S, Xu L, Shinn P, Crawford NP, Inglese J, Burgess SM (2015) Phenotype-driven chemical screening in zebrafish for compounds that inhibit collective cell migration identifies multiple pathways potentially involved in metastatic invasion. DMM Disease Models and Mechanisms 8. doi:10.1242/dmm.018689.

Geling A, Steiner H, Willem M, Bally-Cuif L, Haass C (2002) A γ-secretase inhibitor blocks Notch signaling in vivo and causes a severe neurogenic phenotype in zebrafish. EMBO Rep 3. doi:10.1093/embo-reports/kvf124.

Jiang Y, Zhao X, Chen J, Aniagu S, Chen T (2023) PM2.5 induces cardiac malformations via PI3K/akt2/mTORC1 signaling pathway in zebrafish larvae. Environmental Pollution 323. doi:10.1016/j.envpol.2023.121306.

Paatero I, Veikkolainen V, Mäenpää M, Schmelzer E, Belting HG, Pelliniemi LJ, Elenius K (2019) ErbB4 tyrosine kinase inhibition impairs neuromuscular development in zebrafish embryos. Mol Biol Cell 30. doi:10.1091/mbc.E18-07-0460.

Pruvot B, Curé Y, Djiotsa J, Voncken A, Muller M (2014) Developmental defects in zebrafish for classification of EGF pathway inhibitors. Toxicol Appl Pharmacol 274. doi:10.1016/j.taap.2013.11.006.

Tran TC, Sneed B, Haider J, Blavo D, White A, Aiyejorun T, Baranowski TC, Rubinstein AL, Doan TN, Dingledine R et al (2007) Automated, quantitative screening assay for antiangiogenic compounds using transgenic zebrafish. Cancer Res 67. doi:10.1158/0008-5472.CAN-07-3126.

Vaparanta K, Jokilammi A, Paatero I, Merilahti JA, Heliste J, Hemanthakumar KA, Kivelä R, Alitalo K, Taimen P, Elenius K (2023) STAT5b is a key effector of NRG ‐1/ ERBB4 ‐mediated myocardial growth . EMBO Rep 24. doi:10.15252/embr.202256689.

Vasilyev A, Liu Y, Hellman N, Pathak N, Drummond IA (2012) Mechanical stretch and PI3K signaling link cell migration and proliferation to coordinate epithelial tubule morphogenesis in the zebrafish pronephros. PLoS One 7. doi:10.1371/journal.pone.0039992.

Xin M, Kim Y, Sutherland LB, Qi X, McAnally J, Schwartz RJ, Richardson JA, Bassel-Duby R, Olson EN (2011) Development: Regulation of insulin-like growth factor signaling by Yap governs cardiomyocyte proliferation and embryonic heart size. Sci Signal 4. doi:10.1126/scisignal.2002278.

Zhang Y, Cai Y, Zhang SR, Li CY, Jiang LL, Wei P, He MF (2021) Mechanism of hepatotoxicity of first-line tyrosine kinase inhibitors: Gefitinib and afatinib. Toxicol Lett 343. doi:10.1016/j.toxlet.2021.02.003.

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 evaluated selected EGFR inhibitors developed as targeted cancer therapeutics, using zebrafish embryos and larvae as an in vivo model system. They performed mass spectrometry to analyze phosphorylation levels in target proteins, in combination with western blotting and gene set enrichment analyses; using this data, they assessed overlap between the inhibitors and overlap with known human data. They also performed imaging and locomotion analyses to assess alterations in phenotypes and phosphorylation-dependent signaling due to the inhibitor(s). The study generates novel information that is potentially relevant to the toxicity and efficacy of clinically used kinase inhibitors.

• The statistical analyses are appropriate to the data and the experimental design.
• The claims made by the authors are consistent with the data.

In my opinion, the following revisions are needed for the manuscript to be accepted for publication:

1. There is no mention of Gefitinib in the Abstract; please include it.
2. Please state the target selectivity profiles (from known preclinical and/or clinical data) of the three inhibitors used.
3. Please clarify whether the residues mentioned in the phospho-specific antibody data refer to zebrafish or human proteins.
4. Please state whether the pan-antibodies corresponding to the phospho-specific antibody targets were used, and mention any problems associated with their use. This will help readers not familiar with antibody use in zebrafish experiments. It will also help emphasize the value of mass spectrometric analysis in zebrafish protein work.
5. Please attempt to describe the clinically documented cardiovascular and neurological effects of the inhibitors and any correlation(s) with your data. This will enhance the impact of the study.

Referee Cross-commenting

The common points raised in all the Reviews are the following:

1. The rationale of the study should be described in more detail, especially the utility of zebrafish as an in vivo model, addressing its advantages and limitations.
2. The findings need to be described in the context of the target selectivity profiles and clinical effects of the inhibitors, especially the approved inhibitors (Gefitinib and Lapatinib).
3. In my opinion, while the comments regarding target site drug concentration (within the embryos/larvae) and dose-response are relevant, I consider these experiments to be appropriate in a more detailed follow-up study.
4. One of the main value additions of the study is that it highlights a useful alternative to conventional strategies used in preclinical cellular and mammalian model studies of kinase inhibitors. I would urge the authors to discuss specific future directions, giving due importance to all the reviewers' comments.

Significance

The experiments are well-described and provide sufficient information and detail for readers to understand and reproduce.

The study is highly relevant to the use of zebrafish as a whole-organism model for in vivo evaluation of drugs, specifically kinase inhibitors.

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

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

In this study, the authors assess the effects of various ErbB receptor family tyrosine kinase inhibitors on the phosphoproteome of late embryonic and early larval stages of zebrafish. MS, Western blotting, and analysis of a transgenic zebrafish Notch signaling reporter line data suggest differential but overlapping effects of treatment with gefitinib, lapatinib and AG1478. Selected deregulated pathways are further assessed using a range of candidate downstream pathway-targeting inhibitors. Inhibitor treatment followed by quantification of spontaneous larval motility and heart ventricle wall area, which were previously found by the authors to be affected by AG1478 and lapatinib treatment, identifies involved downstream signaling pathways.

Major comments:

While I do not question the validity of the presented data showing phosphoproteome perturbations resulting from the performed ErbB inhibitor treatments, the treatment regimens used to assess the differential effects of the compounds may be insuffient to substantiate general statements comparing the phenotypic and phosphorylation effects of lapatinib, gefitinib and AG1478 beyond the effects of the specific doses applied to the embryo media. Unless directly quantified, it is difficult to reliably predict the in vivo dose resulting from drug administered to the embryo medium, and therefore a dose may be too high or too low for drug-to-drug comparison. Rationale for chosen dose of drugs should be provided. If available, inclusion of quantitative data on the drug-induced change in phosphorylation status of the drug target(s) is encouraged, and the discussion of the phosphoproteomic and phenotypical data should include this information.

Husbandry: The statement that "Zebrafish were maintained (...) following standard procedures." is insufficient without a specific reference. Please provide details on water quality parameters, temperature, light/darkness cycle and feeding regimen.

Western analysis: How many embryos were pooled in each sample? Please specify standard protocol or provide reference. Ventricle growth assay: The method of ventricle wall quantification is insufficiently described and might result in unnecessarily high variation. At which stage of the cardiac contraction-relaxation cycle were ventricle wall thickness and ventricle area measured? The confounding effect of contraction could be avoided altogether by stopping the heartbeat pharmacologically e.g. by administration of blebbistatin or verapamil. Subtracting ventricle lumen area from total ventricle area seems a much more direct measure of ventricle wall area than the estimation obtained by multiplying ventricle wall thickness with ventricle area.

Phosphopeptide enrichment: How many embryos per sample? Final DMSO concentration is not stated. P-values are presented for comparison of select groups only and a statement that e.g. only P-values < 0.05 are plotted would be helpful if applicable. Also, please provide mean +/- standard deviation for data presented in figures 3A, 3B, 4C, 4E, and 4F.

Minor comments:

Overall, the manuscript is well written and data and methods are well presented. The relevant targets within the ErbB family of receptors should be introduced including information on well-established functions and downstream signaling pathways to enable the non-specialist reader to place the presented data in the context of known gene and protein function. Furthermore, conservation of target proteins in zebrafish should be touched upon.

Given different target profiles of the tested drugs among receptors of the ErbB family, differences in protein phosphorylation perturbations and in treatment-induced phenotypes may not be unexpected. Statements such as: "An unexpectedly large cluster of phosphopeptides that were increased in lapatinib-treated embryos but reduced in AG1478 and gefitinib treated embryos was detected" and "AG1478 and lapatinib may induce similar phenotypes by partially perturbing different signaling pathways in zebrafish embryos" should be discussed in the context of known drug target(s) and their functions.

Referee Cross-commenting

I agree with the other reviewers on almost all points.

1. While the sensitivity to smaller or highly local effects is most likely reduced using the whole organism approach compared to e.g. single tissue analysis, I do believe that it is highly relevant due to its ability to identify potential effects beyond a single tissue or organ.
2. I maintain that while the presented data nicely show the effects of each administered dose of the individual compounds, the data does not allow for meaningful drug-to-drug comparisons without quantitative information on in vivo dose or direct target effect. If such information cannot be included, cross-drug conclusions and discussion should be done very carefully.

Significance

The evaluation of systemic molecular and phenotypic consequences of anti-cancer drugs in a vertebrate model system represents a relevant advancement. Although drug effects are likely to differ somewhat between embryonic and larval zebrafish and human cancer patients, the authors' comparison of obtained zebrafish data with human data supports translatability of the presented phosphoproteomics data. Also, the presented data pose a relevant advancement facilitating the informed use of the tested inhibitors as tools in basic science.

Expertise: Molecular biology, signaling, zebrafish. Limited expertise in omics data analysis and pharmacology.

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 study, Vaparanta and co-workers used zebrafish embryos as model to analyze the impact of ErbB tyrosine kinase inhibitors on signaling pathways at the whole organism level. Experimentally, zebrafish embryos were exposed for 1 hour to a single dose of 3 different ErbB tyrosine kinase inhibitors and the global phosphoproteome of the embryos was analyzed by MS/MS. The authors show that the 3 inhibitors differentially modulate the activity of PI3K/Akt, p38 MAPK, Notch, Hippo-YAP/TAZ and β-catenin signaling pathways, associated with different neurological and myocardial phenotypic changes. Using small molecule inhibitors of selective signaling pathways, they show that perturbation of different signaling pathways may induce similar phenotypes in zebrafish embryos.

Specific comments:

1. The observation that exposure of zebrafish embryos to lapatinib, gefitinib and AG1478 leads to different global phosphoproteomic changes and to differential modulation of cellular signaling pathways was predictable and supported by an abundant literature. These 3 inhibitors differentially inhibit ErbB homo- and heterodimers and hit many other kinases. This point should be discussed in the paper.
2. AG1478 is a first-generation tyrphostin while gefitinib and lapatinib are FDA-approved drugs. These compounds not only have different selectivity profiles, but also different pharmacological properties. Do the authors have any information about the permeability, distribution or concentration of the compounds in zebrafish embryos? Otherwise, how can they compare their effects?
3. One major limitation of this study is that phosphoproteomic analysis was performed at a single time point and with a single dose of inhibitor, which compromises the interpretation of the findings. How was the dose of each inhibitor selected?
4. One approach for better exploiting the data would be to correlate changes in phosphopeptides with the kinome selectivity of the inhibitors.
5. In the same vein, the signaling inhibitors used in Fig. 4 to dissect the phenotypic impact of distinct signaling pathways are non-selective, precluding any rigorous interpretation of the data. This confounding factor should at least be discussed in the manuscript. Again, the choice of the different doses of inhibitors is not justified.
6. The effect of inhibitors on the motility of embryos appears variable. For example, lapatinib markedly decreases motility in Fig. 4E but has no effect in Fig. 4F. Any explanation?
7. The conclusion that ErbB inhibitors induce similar phenotypes by perturbing different signaling pathways is not justified.

I have a few suggestions which could enhance the study's contribution to the field-

1. The rationale for this study should be elaborated further. What new information is expected to emerge from these studies, independently of the conceptual and technical limitations outlined above?
2. The advantage of studying the whole organism instead of selected tissues is questionable. Analyzing a mixture of organs may mask subtle and physiologically relevant alterations of signaling pathways in specific tissues.
3. Can the authors correlate neurological and myocardial phenotypes extrapolated from their study with pharmacological effects observed in mice or humans treated with these compounds?

Significance

The authors show that the 3 inhibitors differentially modulate the activity of PI3K/Akt, p38 MAPK, Notch, Hippo-YAP/TAZ and β-catenin signaling pathways, associated with different neurological and myocardial phenotypic changes. Using small molecule inhibitors of selective signaling pathways, they show that perturbation of different signaling pathways may induce similar phenotypes in zebrafish embryos.

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:

We agree with Reviewer 1 that a function of ROPGEFs in this process was expected to some degree. However, we want to point out that this manuscript focuses on the requirement of ROPGEFs and especially the spatio-temporal description of ROP signalling polarisation and activation during pollen germination. Moreover, different to the downstream ROPs, we show ROPGEFs do not act strictly redundant, confirming results from root hair initiation and providing additional evidence that multiple signalling pathways are required for pollen germination and that ROPGEFs might be essential for bringing specificity to these signals.

Major comments:

1. Only one GEF11 mutant line, gef11-t1, was analyzed for germination ratio. It is presumptuous to conclude that GEF11 has no function in the pollen germination of Arabidopsis thaliana (line 241- line 242).

After the initial negative results, we did not focus on GEF11 further. Thus, we fully agree that it is presumptuous to make such strong statements about the role of GEF11 during pollen germination. We generated additional gef11 mutant alleles for this revision plan using CRISPR/Cas9 as no other suitable lines were available. Moreover, we now have additional higher-order mutants available to demonstrate the function of GEF11 during pollen germination. These additional lines were generated and confirmed and are growing right now. Thus, we will be able to implement new results addressing this point timely, allowing us to make a more founded statement about the function of GEF11 (see Response to Reviewer #2).

Minor comments:

1. In Figure 2A, pollen germination ratio was not provided for the single mutants gef8-c△3 and gef9-c△

This is due to the generation process of the CRISPR/Cas9 alleles. These alleles were generated by a construct mutating both genes simultaneously; thus, these mutants are unavailable as single mutant lines. Instead of separating these alleles by outcrossing, we included additional single mutant alleles for both GEFs with a similar deletion. As all these CRISPR/Cas9 mutants have a complete deletion of the GEF-ORF, we are sure about the loss of the according GEF function. Additional alleles account for possible unspecific effects.

In Figure 3D, the subcellular localization of GEF12GEF8C is fuzzy. Better imaging is needed.

We agree that the quality of these images is not ideal due to this specific line having less fluorescent signal. We screened for more lines of this construct and already performed more experiments. We will provide better images for this genotype.

In Figure 3E, it is intriguing that both GEF8-S518A and GEF8-S518D are not associated with the PM in germinating pollen grains. Does it mean that phosphorylation at S518 is not relevant to polar distribution of GEF8?

We also find this very intriguing as we did not expect this result. However, we interpret it slightly differently in the way that the S518 site is relevant for GEF polarisation, which might be conferred by RLK interaction. We think both mutant forms alter this potential association with RLKs, thus losing polarisation. We will include more imaging experiments of these constructs and additional lines to strengthen our results. Moreover, we generated lines to study these lines' functionality and complementation capacity, which will be included in a revised manuscript.

T-DNA insertion lines, gef11-t1 and gef12-t1, need to be verified by PCRs in Figure S3D.

Thanks for pointing this out. This control should be provided, and we will include the verification in the supplement.

Response to Reviewer #2:

Like Reviewer #2, we are also very intrigued by the biphasic accumulation of GEFs, as this is an entirely novel feature of this process. Like Reviewer #2, we also interpret this as an exploration and establishment phase, which could help us to understand how the pollen germination site is decided in species without aperture-dependent pollen germination.

Major comments:

1. In line 241, the authors conclude that GEF11 has no function in pollen germination. However, it is likely that GEF11 also plays a redundant role as GEF12 does. I recommend the authors check the phenotypes of gef11,gef12 double mutant and gef8,gef9,gef11 triple mutant to confirm that GEF11 has indeed no function. Otherwise, this conclusion should be better rephrased.

This point is well justified and similar to the comment of Reviewer #1. As stated before, we had to generate additional lines for this. We will analyse an additional gef11 allele, gef8/gef11 and gef9/11 double mutants, and gef9/11/12 triple mutants to address the function of GEF11 in more detail. The conclusions of the original manuscript will, of course, be adjusted according to the new results.

Although GEF12 is in the cytosol, the strong pollen germination defects in gef8,gef9,gef12 triple mutants do indicate a critical role of GEF12. Is it possible that GEFs could function in the cytosol? The authors can test this possibility by examining the rescuing ability of several constructs that express, for example, GEF12, GEF12(+GEF8C), GEF8(SA), or GEF8(SD) in gef8. The authors may not perform all of these rescue experiments, but some of the mentioned lines are already in hands. They could readily check the phenotypes.

We thank the Reviewer for this great point. This information is crucial to discriminate the function of the individual GEFs. We have generated new lines expressing some of the mentioned constructs in the gef8 background to address this. We now have lines that complement gef8 with GEF12, GEF12GEF8C, GEF8S518A, GEF8S518D, and GEF8ΔC. We are currently performing experiments which determine the functionality of these constructs, which will allow us to make more conclusive statements about the function of GEFs in the cytosol and how important the PRONE domain alone, or the membrane attachment of GEFs, is for their function.

The authors conclude that the C-terminus of GEF8 and GEF9 is necessary and sufficient for membrane localization because GEF8/9C can target GEF12 PRONE domain to the membrane. It is intriguing whether the C-terminus alone could confer membrane targeting ability. Currently, it is not fully understood how GEFs localize to the membrane. Examining the localization of GEF8/9C itself would help clarify this and improve our understanding of GEF regulation. Alternatively, the authors may discuss evidence that supports or disagrees with this possibility.

This is a good suggestion by the reviewer and indeed intriguing if the C-Terminus alone could confer membrane attachment. Meanwhile, we obtained plants expressing such constructs, showing that the C-terminus alone is insufficient for membrane attachment. This is not surprising, as these domains are largely disordered, and we suspect that the context of an adjacent PRONE domain is required to carry out this function. We will include our new results in the revised manuscript.

Minor comments:

1. The N- and C-terminus of GEF8 are predicted to inhibit complex formation. How is the prediction performed? Do the authors use monomer prediction or multimer prediction? Alphafold2 has a low accuracy in predicting non-conserved regions. How confident are the predicted inhibitory contacts?

We used multimer-prediction of Alphafold2 for the shown structures. However, we fully agree that the predicted structures of Alphafold have low accuracy in that regard, especially for disordered domains like this. We will provide confidence models and predicted aligned error (PAE) plots for this structure. Additionally, we will put our conclusions in a better perspective of these structure confidences and tone down our interpretations of this section.

Localization of ROPs and calcium reporter in Figure 4 appears to be variable. It would help clarify the specific effects on each reporter if the authors present these data more quantitatively.

We agree with the reviewer that some of the observations are variable. We will provide the data more quantitatively, including overviews of which percentage we observed the described phenomena and a more quantitative analysis of the strength and timing of signal accumulation (see also Response to Reviewer #3).

Response to Reviewer #3:

Major points:

1. One of my major points is that the manuscript is now mainly based on the observations of individual pollen grains. These are then subjected to well-performed image analysis approaches but still represent somewhat anecdotal evidence (Fig 1A, B, Fig 3C-E, etc). The analysis and (numerical) presentation of a more robust data sample (which I presume the authors have acquired) would strengthen the ms considerably. This goes beyond the Figs - e.g. in l. 164-165 authors state rather vaguely, "we observed that mCit-GEF8 and mCit-GEF9 accumulated at a defined region in the cell periphery, which strongly correlated with the future germination site." Here, I would appreciate the data showing the actual correlation, if every germinated pollen grain displays GEF8/9 accumulation, whether there is a population of pollen grains showing the GEF8/9 transient but not germinating, etc...

We very much appreciate the reviewer's comment, as this version of the manuscript indeed seems like we made our conclusions based on observations made from individual pollen. However, this is not the case. As the reviewer suspected, more data is available but not included in the manuscript. We have multiple observations for each of the shown constructs and only show a representative one. Furthermore, we imaged more pollen germination events of lines that showed variability and included additional lines for some constructs. We will provide a more quantitative analysis of the results to better represent the variability of the individual constructs, and we will adjust the manuscript accordingly (see comment 2).

Where the authors analyse multiple cells, we are still missing some info - e.g. it is not stated what the error bars in Fig 1C, D represents (SD, SEM, CI?), size of the sample, etc. In any case, it is evident that there is quite substantial variability in the data, which is understandable. Maybe the authors can plot the individual profile lines along the average? Plus, GEF9 seem to have the maximum pre-germination localisation at -5 min rather than -9 min.

We agree with the Reviewer that information is missing or not obviously stated. We will correct this for the revised manuscript. Moreover, we agree that the suggested way of showing the data would provide more information and allow a better representation of the results and their variability. This can be seen in the reviewer's interpretation of the results of GEF9. In this case, we see some variability in the timing of GEF9 accumulation, leading to the peak maximum shift. In a revised manuscript, we will, as suggested, show the data as individual lines, providing a better representation of the data. Moreover, we will include such representations for other used constructs to provide a general, more quantitative data analysis (see comment 1).

I know it is very challenging, but the ms would be much stronger with the in vivo imaging of pollen germination on stigmatic papillae (i) GEF8/9 in wt, (ii) gef8/9 double mutant. This would bring crucial data about the role of the GEF polar domain and its functional relation to pollination.

This would indeed be great to see. We put an effort into establishing such in vivo imaging experiments with our fluorescent markers. However, we cannot image these events in an in vivo setup (at least with our resources). This has two reasons: 1. The events are very fast and limited to a small region at the pollen-papilla contact side, which we have issues resolving optically and timely. 2. The used marker lines only have a low fluorescent level due to the native promoter, and stronger expression would lead to overexpression artefacts. In vitro, it is difficult to see the observed signal accumulation. In the in vivo situation, we are facing additional diffraction of the papilla cells, which would make the observation of GEF accumulation impossible with our microscopes.

The phylogeny presented in Fig S1 is only rudimental and not very interesting. Given the author's results, I would love to see if GEF8/9 orthologs also exist in species with defined pollen apertures (where establishing a dynamic site makes little sense). The authors touch on this (L409-411), but it would deserve better analysis and discussion.

We agree with the reviewer that studying GEF function/accumulation in species with aperture-dependent germination would be interesting. However, we can not conclude functional orthologs in other species based on phylogeny. Such phylogenetic analyses were done, for example, by Kim et al. (BMC Plant Biology, 2020, doi: 10.1186/s12870-020-2298-5). The issue is that all Arabidopsis pollen-expressed GEFs form a closed phylogenetic group without allowing the interpretation of which rice homolog is the functional ortholog of the respective Arabidopsis GEF (this is the same for maize). Thus, such phylogenetic analyses are not conclusive, and they would require experimental data to prove orthology. However, we agree that this point can be interpreted and discussed better, and we will include this in the revised manuscript.

I am not entirely convinced by the authors' interpretation of rather strange S518 mutation data. Could S518A mutation affect overall GEF8 structure/stability?

We were also suspicious about these results, as they were unexpected (see also Response to Reviewer #1). To confirm these results, we made additional lines for these constructs, double-checked that the constructs were correct and made more observations for both GEF8S18A and GEF8S18D. Additionally, we started investigating the functionality of these constructs and have this data available timely. Preliminary results suggest that the constructs are partial to fully functional compared to the WT GEF8, arguing against these mutations' effect on structure or stability. We will include more data for these constructs in a revised manuscript to allow a more conclusive interpretation of these unexpected observations.

Although the authors cannot observe the localisation of ROPs in the plasma membrane, they see the apparent accumulation of active ROP marker CRIB4 there - implying that ROPs must localise to the pollen PM at the germination site. This discrepancy should be solved or at least discussed more.

The reviewer is correct in that we cannot observe ROP accumulation but rather the accumulation of ROP activity (as seen by CRIB4). This is in line with the observation made by Xiang et al. (2023, Plant Physiology, doi: 10.1093/plphys/kiad196), which also cannot find ROP accumulation. We are convinced that ROPs are present at the plasma membrane of the pollen germination site, but no accumulation is observable. We believe this is due to a high mobility of ROPs and that no accumulation is required, as only a few ROPs are sufficient to activate downstream signals. We will discuss these results in more detail in a revised manuscript to better explain the observed discrepancy.

Given that calcium oscillates very rapidly in pollen and pollen tubes (with frequency ~6-20s), the profound, long-term changes in calcium levels reported by the authors can hardly be referred to as oscillations. The phenomenon observed should again be analysed using a bigger sample.

We agree that the terminology is not good, as it suggests similarities to the oscillations found in pollen tubes. Thus, we will change the revised manuscript and refer to the changes in Ca2+ levels as “elevations”. Moreover, we will provide a more quantitative analysis and a bigger sample size, as stated in Response to Reviewer #2.

Minor points:

1. In Fig 1F, GEF12 also seems to be polarly localised to the future site.

The chosen sample is not ideal, as it looks like GEF12 would also slightly accumulate. However, as seen in the quantification of this cell, GEF12 does not significantly accumulate at the pollen germination site, and we never observed any accumulation of GEF12 that is comparable to GEF8 or GEF9. We will include another sample of this colocalisation in the revised manuscript to avoid misinterpretation of the data.

It is difficult to make any assumptions based on the AlphaFold2 predictions without showing their confidence assessments (e.g., PAE plots). The authors state this themselves in the discussion (L. 447-449).

As the Response to Reviewer #2 stated, we will include structures with confidence values and PAE plots in the supplement. We additionally tone down our interpretation of these structure predictions to make clear that these structures should be interpreted carefully.

On one hand the authors repeatedly state that pollen GEFs do act in a redundant manner (and provide some evidence for it), on the other hand the absence of an in vivo phenotype for single and double knockout lines and only mild phenotype for a triple ko line does suggest a level of redundancy. This should be rephrased.

We agree that this is not clearly phrased. In a revised version, we will change the manuscript to indicate which type and level of redundancy are described. We will discriminate between genetic redundancy, as seen in the mild in vivo effects, and non-redundant molecular function, as observed by protein localisation.

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 investigates the role of PRONE ROP GEFS in germinating Arabidopsis pollen. Given that the molecular mechanisms underlying cellular polarisation in pollen germinating pollen grains are still largely unknown (as opposed to the tip growth of elongating pollen tubes), this manuscript deals with an important topic. Moreover, it builds on the excellent previous research from the lead author, which uncovered ROP GEFs as principal polarisation players during root hair initiation. Here, the authors found that out of five pollen-expressed GEFS, two (GEF8 and 9) mark a future germination site with remarkable spatiotemporal dynamics. Using the genetic tools, GEF8 and 9 were shown to be important for pollen germination in vitro and participate in germination in vivo. Generally, this is an exciting topic, and I quite enjoyed reading the manuscript. However, there are several aspects of the work, which - when addressed - would significantly improve the overall message presented by the authors.

Major points

1. One of my major points is that the manuscript is now mainly based on the observations of individual pollen grains. These are then subjected to well-performed image analysis approaches but still represent somewhat anecdotal evidence (Fig 1A, B, Fig 3C-E, etc). The analysis and (numerical) presentation of a more robust data sample (which I presume the authors have acquired) would strengthen the ms considerably. This goes beyond the Figs - e.g. in l. 164-165 authors state rather vaguely, "we observed that mCit-GEF8 and mCit-GEF9 accumulated at a defined region in the cell periphery, which strongly correlated with the future germination site." Here, I would appreciate the data showing the actual correlation, if every germinated pollen grain displays GEF8/9 accumulation, whether there is a population of pollen grains showing the GEF8/9 transient but not germinating, etc...
2. Where the authors analyse multiple cells, we are still missing some info - e.g. it is not stated what the error bars in Fig 1C, D represents (SD, SEM, CI?), size of the sample, etc. In any case, it is evident that there is quite substantial variability in the data, which is understandable. Maybe the authors can plot the individual profile lines along the average? Plus, GEF9 seem to have the maximum pre-germination localisation at -5 min rather than -9 min.
3. I know it is very challenging, but the ms would be much stronger with the in vivo imaging of pollen germination on stigmatic papillae (i) GEF8/9 in wt, (ii) gef8/9 double mutant. This would bring crucial data about the role of the GEF polar domain and its functional relation to pollination.
4. The phylogeny presented in Fig S1 is only rudimental and not very interesting. Given the author's results, I would love to see if GEF8/9 orthologs also exist in species with defined pollen apertures (where establishing a dynamic site makes little sense). The authors touch on this (L409-411), but it would deserve better analysis and discussion.
5. I am not entirely convinced by the authors' interpretation of rather strange S518 mutation data. Could S518A mutation affect overall GEF8 structure/stability?
6. Although the authors cannot observe the localisation of ROPs in the plasma membrane, they see the apparent accumulation of active ROP marker CRIB4 there - implying that ROPs must localise to the pollen PM at the germination site. This discrepancy should be solved or at least discussed more.
7. Given that calcium oscillates very rapidly in pollen and pollen tubes (with frequency ~6-20s), the profound, long-term changes in calcium levels reported by the authors can hardly be referred to as oscillations. The phenomenon observed should again be analysed using a bigger sample.

Minor points

1. In Fig 1F, GEF12 also seems to be polarly localised to the future site.
2. It is difficult to make any assumptions based on the AlphaFold2 predictions without showing their confidence assessments (e.g., PAE plots). The authors state this themselves in the discussion (L. 447-449).
3. On one hand the authors repeatedly state that pollen GEFs do act in a redundant manner (and provide some evidence for it), on the other hand the absence of an in vivo phenotype for single and double knockout lines and only mild phenotype for a triple ko line does suggest a level of redundancy. This should be rephrased.

Significance

General assessment

I believe that both strenghts and limitations are evident form the list above. I feel this a study with great potential, which can be improved by textual ammendments and by several additional experiments that do not require the generation of new genetic material.

Advance

This ms builds on the results obtained previously by the lead author and does advance the knowledge of the field of plant cell polarity substantially.

Audience

The ms is targeted for the basic research audience, particularly for plant scientists.

Expertise of the reviewer

Pollen biology, membrane trafficking, phylogenetic analyses, protein biochemistry.

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

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

Summary:

In this study, Bouatta et al. report the function of RopGEFs in pollen germination. The authors analyzed all of the five RopGEFs, namely RopGEF8/9/11/12/13, that have been shown to be expressed in mature pollen tubes, and found that only GEF8/9/11/12 are detectable. In addition, GEF8 and GEF9 localize to germination sites, while GEF11 and GEF12 are cytosolic. Through a series of phenotype analyses and live-cell imaging, the authors show that GEF8, GEF9, and GEF12 are required for pollen germination while GEF11 is not. The authors also provide evidence that GEF8 and GEF9 are targeted to the membrane via the C-terminus, where they activate ROPs and calcium signaling.

Major comments:

1. In line 241, the authors conclude that GEF11 has no function in pollen germination. However, it is likely that GEF11 also plays a redundant role as GEF12 does. I recommend the authors check the phenotypes of gef11,gef12 double mutant and gef8,gef9,gef11 triple mutant to confirm that GEF11 has indeed no function. Otherwise, this conclusion should be better rephrased.
2. Although GEF12 is in the cytosol, the strong pollen germination defects in gef8,gef9,gef12 triple mutants do indicate a critical role of GEF12. Is it possible that GEFs could function in the cytosol? The authors can test this possibility by examining the rescuing ability of several constructs that express, for example, GEF12, GEF12(+GEF8C), GEF8(SA), or GEF8(SD) in gef8. The authors may not perform all of these rescue experiments, but some of the mentioned lines are already in hands. They could readily check the phenotypes.
3. The authors conclude that the C-terminus of GEF8 and GEF9 is necessary and sufficient for membrane localization because GEF8/9C can target GEF12 PRONE domain to the membrane. It is intriguing whether the C-terminus alone could confer membrane targeting ability. Currently, it is not fully understood how GEFs localize to the membrane. Examining the localization of GEF8/9C itself would help clarify this and improve our understanding of GEF regulation. Alternatively, the authors may discuss evidence that supports or disagrees with this possibility.

Minor comments:

1. The N- and C-terminus of GEF8 are predicted to inhibit complex formation. How is the prediction performed? Do the authors use monomer prediction or multimer prediction? Alphafold2 has a low accuracy in predicting non-conserved regions. How confident are the predicted inhibitory contacts?
2. Localization of ROPs and calcium reporter in Figure 4 appears to be variable. It would help clarify the specific effects on each reporter if the authors present these data more quantitatively.

Significance

Advance:

ROP GTPases and RopGEFs are critical regulators of cell polarity, but how they initiate polarity remains unclear. This study uses pollen germination as a model to address this question. It systematically analyzed all pollen-specific GEFs and found that GEF8 and GEF9 are critical regulators of pollen germination and polarity initiation. Importantly, GEF8 and GEF9 undergo biphasic accumulation, suggesting polarity is established through a transient exploration phase. This study provides a comprehensive view of the functions of GEFs in polarity initiation, which will be of interest not only to readers who work on pollen germination and growth but also to those who study cell polarity and morphogenesis in general. In my view, the most novel part of this study is that GEFs play overlapping but non-identical roles in polarity establishment and undergo transient accumulation during the polarity initiation process.

Limitations:

This study shows that GEFs use the C-terminus for membrane targeting and GEFs can activate ROPs and calcium signaling during pollen germination. These mechanisms could be largely inferred from previous studies in mature pollen tubes or others. Advancements in the regulation of GEF such as how the C-terminus mediates GEF localization, e.g. whether through direct interaction with the PRONE domain in a phosphorylation-dependent manner, would further increase the novelty of this 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

In the manuscript, the Denninger group reported the identification of ROPGEF8/9 as the key ROPGEFs for ROP activation during pollen germination, a process of polarity establishment. By examining the subcellular localization of pollen-expressed/enriched GEFs using their own promoter and fluorescence protein fusions, the authors convincingly showed the spatiotemporal distribution of GEF8 and GEF9 during pollen germination. By characterizing pollen germination of gef mutants, the authors demonstrated that GEF8 and GEF9 are critical for the process, with GEF12 playing a redundant role likely as a compensation response. The authors further showed that C-termini of GEF8/9, previously demonstrated as an inhibitory domain for GDP-GTP exchange, was critical for the polar distribution. The C-termini of GEFs interact with PRK. The authors reported that the phosphorylation of GEFs at C-termini was critical for their polar distribution. By examining the dynamic localization of active ROP biosensor CRIBRIC4, the authors demonstrated that GEF8/9 were critical for polar distribution of active ROPs at future germination sites. By introducing a calcium biosensor, the authors showed that calcium gradient, a key downstream process of ROP signaling, was compromised by functional loss of GEF8/9 during pollen germination.

Major comments

Only one GEF11 mutant line, gef11-t1, was analyzed for germination ratio. It is presumptuous to conclude that GEF11 has no function in the pollen germination of Arabidopsis thaliana (line 241- line 242).

Minor comments

In Figure 2A, pollen germination ratio was not provided for the single mutants gef8-c△3 andgef9-c△2.

In Figure 3D, the subcellular localization of GEF12GEF8C is fuzzy. Better imaging is needed.

In Figure 3E, it is intriguing that both GEF8-S518A and GEF8-S518D are not associated with the PM in germinating pollen grains. Does it mean that phosphorylation at S518 is not relevant to polar distribution of GEF8?

T-DNA insertion lines, gef11-t1 and gef12-t1, need to be verified by PCRs in Figure S3D.

Significance

The identification of ROPGEF8/9 as the key ROPGEFs for ROP activation during pollen germination is a step forward in understanding ROP signaling. Useful but not unexpected.

Pollen germination is a process of polarity establishment, similar to root hair initiation. Compared to pollen tube growth and root hair growth, processes of polarity maintenance, the role of ROP signaling was less clear. Recently, Xiang et al. (2023, Plant Physiol) reported an essential role of ROP1/3/5 and their downstream components BDR8/9 in pollen germination. Consistently with polar ROP activation, Ca2+ and post-Golgi secretion were polar. The current work is one step ahead, showing GEF8/9 as the upstream GEFs for this process, comparable to GEF3/4 during root hair initiation (Denninger et al., 2019, Curr Biol). The identification of the C-terminal phosphorylate site in GEF8/9 is informative. It was reported previously that PRKs interact with the C-termini of GEFs to release their auto-inhibition (Gu et al., 2006, Plant Cell; Zhang and McCormick, et al., 2007, Proc Nat Acad Sci USA, Zhao et al., 2013, J Exp Bot) and PRKs were reported to phosphorylate GEF (Chang et al., 2013, Mol Plant). Thus, results reported in the current work indicate that phosphorylation of GEFs likely by PRKs is a critical step for the establishment of polarity domain for pollen germination. From this perspective, it would be more mechanistically sound to investigate the role of PRKs in spatiotemporal polarization of GEFs during pollen germination.

Researchers working on cell signaling and cell morphogenesis in plants will be interested.

My lab works on cell morphogenesis and ROP signaling. This manuscript exactly falls within the expertise of my field.

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

Learn more at Review Commons

Reply to the reviewers

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

Reply to the Reviewers

We want to thank the three reviewers for their invaluable and constructive feedback. We respond to each comment individually, describing how we plan to address them in our revised manuscript.

Reviewer #1

1. Given the emphasis on super-resolution imaging deep inside a sample, we were surprised to see no mention of other forms of structured illumination that allow super-resolution imaging in samples thicker than a single cell. These include the 'spot-scanning' implementations of SIM that offer better imaging at depth by virtue of pinholes, and include MSIM, iSIM, and rescan confocal technologies. The two-photon / AO implementation of iSIM seems particularly germane, e.g. https://pubmed.ncbi.nlm.nih.gov/28628128/ Please consider citing these works, as they help place the existing work into context.

Response:

We want to thank reviewer #1 for the good point. To address this comment, we plan to add to the discussion section a description of these super resolution techniques, together with other SIM methods, explaining how they compare to our approach.

1. As we're sure the authors appreciate, besides aberrations, a major additional obstacle to 3D SIM in thick tissues is the presence of out-of-focus background. Indeed, this point was mentioned by Gustafsson in his classic 2008 paper on 3D SIM (https://pubmed.ncbi.nlm.nih.gov/18326650/): 'The application area of three-dimensional structured illumination microscopy overlaps with that of confocal microscopy, but the two techniques have different and complementary strengths. Structured illumination microscopy offers higher effective lateral resolution, because it concentrates much of the excitation light at the very highest illumination angles, which are most effective for encoding high-resolution information into the observed data, whereas confocal microscopy spreads out its illumination light more or-less uniformly over all available angles to form a focused beam. For very thick and compactly fluorescent samples, however, confocal microscopy has an advantage in that its pinhole removes out-of focus light physically. Structured illumination microscopy is quite effective at removing out-of-focus light computationally, because it is not subject to the missing-cone problem, but computational removal leaves behind the associated shot noise. Therefore confocal microscopy may be preferable on very thick and dense samples, for which the in-focus information in a conventional microscope image would be overwhelmed by out-of-focus light, whereas structured illumination microscopy may be superior in a regime of thinner or sparser samples.' This point is not mentioned at all in the manuscript, yet we are certain it is at least partially responsible for the residual image artifacts the authors mention. Please discuss the problem of out of focus light on 3D samples, particularly with an eye to the 'spot-scanning' papers mentioned above.

Response:

We appreciate this significant obstacle and we want to thank Reviewer #1 for emphasising its importance. To address the comment, we plan to add a discussion of the significance of out-of-focus light to SIM imaging to the introduction, results, and discussion sections of the manuscript.

1. The authors use a water dipping lens, yet they image into samples that are mounted on coverslips, i.e. they use a dipping lens to image through a coverslip:

This almost certainly introduces spherical aberration, which the authors seem to observe: see attached pdf for reference

We find this troubling, as it seems that in the process of building their setup, the authors have made a choice of objective lens that introduces aberrations - that they later correct. At the very least, this point needs to be acknowledged in the manuscript (or please correct us if we're wrong) - as it renders the data in Figs. 3-4 somewhat less compelling than if the authors used an objective lens that allowed correction through a coverglass, e.g. a water dipping lens with a correction collar. In other words, in the process of building their AO setup, the authors have introduced system aberrations that render the comparison with 3D SIM somewhat unfair. Ideally the authors would show a comparison with an objective lens that can image through a glass coverslip.

Response:

We want to thank Reviewer #1 for raising this point, which we did not describe clearly enough, leading to confusion. We should have made it clearer that we used a water dipping/immersion objective lens with a correction collar which extends from no coverslip (dipping) up to well beyond a standard #1.5 (170 um thick) coverslip. We adjusted this collar before each image acquisition session, to ensure that the system is optimised for each experiment individually and that the spherical aberrations are minimal before any DM-based correction. We plan to elaborate and emphasise this point in several places in the revised manuscript, including in the figure legends, materials and methods and results sections, to avoid any ambiguity and confusion about the use of the correction collar and this particular water immersion/dipping objective lens.

1. The authors tend to include numbers for resolution without statistics. This renders the comparisons meaningless in my opinion; ideally every number would have a mean and error bar associated with it. We have included specific examples in the minor comments below.

Response:

This is a good point, which we address below, in three minor comments. In summary, to address this comment, we plan to include statistical information in the revised manuscript.

1. In Fig. 5, after the 'multipoint AO SIM', the SNR in some regions seems to decrease after AO: see attached pdf for reference

Please comment on this issue.

Response:

We want to thank Reviewer #1 for the insightful comment. There are multiple phenomena in effect here, which cause the drop in intensity. The most prominent one is photobleaching, as the AO image stack (right) was acquired after the bypass one (left). To address this comment, we plan to add additional data and to include a brief discussion about this issue and other related points.

1. Please provide timing costs for the indirect AO methods used in the paper, so the reader understands how this time compares to the time required for taking a 3D SIM stack. In a similar vein, the authors in Lines 213-215, mention a 'disproportionate measurement time' when referring to the time required for AO correction at each plane - providing numbers here would be very useful to a reader, so they can judge for themselves what this means. What is the measurement time, why is it so long, and how does it compare to the time for 3D SIM? It would also be useful to provide a comparison between the time needed for AO correction at each (or two) planes without remote focusing (RF) vs. with RF, so the reader understands the relative temporal contributions of each part of the method. We would suggest, for the data shown in Fig. 5, to report a) the time to acquire the whole stack without AO (3D SIM only); b) the time to acquire the data as shown; c) the time to acquire the AO stack without RF. This would help bolster the case for remote focusing in general; as is we are not sure we buy that this is a capability worth having, at least for the data shown in this paper.

Response:

We agree that the timing (and other) costs can be an important consideration, and we want to thank Reviewer #1 for bringing up this good point. To address this issue, we plan to expand our description of the AO methods, also including numbers for the time it takes to perform the different parts. In terms of comparisons, the RF makes no contribution to the timing costs of the aberration correction, a point that we want to make clearer in the results and the methods and materials sections, as the two are independent processes in our approach. Instead, the RF can be compared to standard focusing with a piezo stage, a point which we discuss in the supplementary material. We plan to make this point clearer in the discussion section of the main manuscript, and to emphasise better the advantages of the RF in terms of imaging speed.

1. Some further discussion on possibly extending the remote focusing range would be helpful. We gather that limitations arose from an older model of the DM being used, due to creep effects. We also gather from the SI that edge effects at the periphery of the DM was also problematic. Are these limitations likely non-issues with modern DMs, and how much range could one reasonably expect to achieve as a result? We are wondering if the 10 um range is a fundamental practical limitation or if in principle it could be extended with commercial DMs.

Response:

Regrettably, we were not able to try other DMs on the Deep3DSIM system. However, Jiahe and colleagues show in [1] that similar DM-based remote focusing, even with the same model deformable mirror, can be pushed to 120 um (Strehl ratio >0.8) with a 0.42 NA dry lens (20 mm WD) and close-loop wavefront compensation operation. While this is not directly translatable to high NA 3D-SIM imaging, we expect that with a stable version of the same DM the useable RF range could be easily increased twice or even more. We thank Reviewer #1 for the good comment, which we plan to address by revising the text to make the limitations clearer and by citing relevant studies.

[1] Cui, J., Turcotte, R., Emptage, N. J., & Booth, M. J. (2021). Extended range and aberration-free autofocusing via remote focusing and sequence-dependent learning. Optics Express, 29(22), 36660-36674.

Minor comments:

1. The paper mentions Ephys multiple times, even putting micromanipulators into Fig. 1 - although it is not actually used in this paper. If including in Figure 1, please make it clear that these additional components are aspirational and not actually used in the paper.

Response:

Although not shown in the context of this paper, the Deep3DSIM system was built specifically around experiments such as electrophysiology, which can benefit from the upright configuration and the water-dipping-capable objective lens. To address this comment, we plan to clarify the role of the micromanipulators and to update Figure 1 accordingly.

1. The abstract mentions '3D SIM microscopes', 'microscopes' redundant as the 'm' in 'SIM' stands for 'microscope'.

Response:

We accept that “3D SIM microscopes” sounds repetitious and we plan to revise the wording of the abstract to “3D SIM system”.

1. 'fast optical sectioning', line 42, how can optical sectioning be 'fast'? Do they mean rapid imaging with optical sectinong?

Response:

Yes, we meant rapid imaging with optical sectioning. We plan to change the wording to make it less ambiguous.

1. line 59, 'effective imaging depth may be increased to some extent using silicone immersion objectives', what about water immersion objectives? We would guess these could also be used.

Response:

Yes, water immersion objective lenses also fall in the same category and we plan to rephrase this part to state it explicitly.

1. line 65 - evidence for 'water-dipping objectives are more sensitive to aberrations' ? Please provide citation or remove. They are certainly more prone to aberrations if used with a coverslip as done here.

Response:

The refractive index (RI) of cells and tissues [1] is closer to the RI of silicone oil (~1.4) than it is to water (~1.33). Therefore, because of the larger difference in RI, imaging with a water-dipping objective lens is more prone to aberrations from RI mismatch. We plan to rephrase this argument to make it clearer.

[1] Jacques, S. L. (2013). Optical properties of biological tissues: a review. Physics in Medicine & Biology, 58(11), R37.

1. 'fast z stacks' is mentioned in line 103. How fast is fast?

Response:

The speed would depend on the way Z-stacks are being acquired. For example, acquisitions with two channels would be at least twice as fast, because of the ability to do simultaneous imaging on the Deep3DSIM system. Likewise, experiments that can benefit from the remote focusing can be several times faster than using a Z piezo stage, and this point is discussed in the supplementary material (section “Step response”). Finally, thanks to the electronic design of the imaging system, orchestrating everything via digital logic (e.g. TTL) signals, and thanks to the elaborate control software, we can ensure that all image acquisitions are carried out as quickly as possible, operating near the limit of the underlying hardware devices. We plan to explain these points in a clear way in the discussion section, and we plan to provide more numbers in the supplementary material.

1. line 116 'we imaged 100 nm diameter green fluorescent beads'. Deposited on glass? Given that this paper is about imaging deep this detail seems worth specifying in the main text.

Response:

Yes, in this case the beads were deposited on glass. We plan to include this detail in the description of the experiment.

1. lines 127-130, when describing changes in the bead shape with numbers for the FWHM, please provide statistics - quoting single numbers for comparison is almost useless and we cannot conclude that there is a meaningful improvement without statistics.

Response:

We agree with this comment. We plan to include statistical information for all the FWHM numbers.

1. In the same vein, how can we understand that remote focus actually improves the axial FWHM of the widefield bead? Is this result repeatable, or it just noise?

Response:

The lower axial FWHM with remote focusing is likely caused by data fitting or quantification error. Together with the inclusion of statistical information, we plan to review all the resolution values and to ensure that they are accurate and sensible.

1. line 155, 'Because of the high spatial information...' -> 'Because of the high resolution spatial information...'

Response:

We agree with this comment. To address it, we plan to rephase this part.

1. When quoting estimated resolution #s from microtubules (lines 158-163) similarly please provide statistics as for beads.

Response:

We agree with this comment. To address it, we plan to include statistical information for the resolution values from microtubules.

1. It seems worth mentioning the mechanism of AO correction (i.e. indirect sensing) in the main body of the text, not just the methods.

Response:

We agree with this comment. To address it, we plan to describe briefly the aberration correction method in the introduction or the results section.

1. How long do the AO corrections take for the datasets in the paper?

Response:

The duration of the aberration correction routines is directly proportional to the number of Zernike modes, the number of iterations, the exposure time of the camera, and other parameters. In our experiments, it was usually in the order of tens of seconds. To address this comment, and in line with the sixth major comment, we plan to include more details about the timing of the different parts of the AO methods.

1. Were the datasets in Fig. 2-4 acquired with remote focusing, or in conventional z stack mode? Please clarify this point in the main text and the figure captions.

Response:

The only data acquired with RF in Fig. 2-4 are one bead in Fig. 2A and another bead in Fig. 2B, both labelled accordingly. We plan to make it clearer in the text that the rest of Figure 2, as well as Figures 3 and 4, were acquired with the piezo Z stage.

1. It would be helpful when showing z projections in Figs. 3-5 to indicate the direction of increasing depth (we assume this is 'down' due to the upright setup, but this would be good to clarify)

Response:

The direction is indicated by the arrows labelled with ‘Z’. We plan to clarify this in the figure captions.

1. line 174, 'showed significant improvements in both intensity and contrast after reconstruction' - we see the improvements in contrast and resolution, it is harder to appreciate improvements in intensity. Perhaps if the authors showed some line profiles or otherwise quantified intensity this would be easier to appreciate.

Response:

We agree with this comment. To address it, we plan to change Figure 3 to illustrate the improvement in intensity, likely with line profiles, as suggested by the reviewer.

1. line 195 'reduced artefacts' due to AO. We would agree with this statement - the benefit from AO is obvious, and yet there are still artefacts. If the authors could clarify what these (residual) artefacts are, and their cause (out of focus light, uncorrected residual aberrations, etc) this would be helpful for a reader that is not used to looking at 3D SIM images.

Response:

We agree with this comment. To address it, we plan to explain this point in both the results and the discussion sections.

1. Line 197, 'expected overall structure', please clarify what is expected about the structure and why.

Response:

We agree with this comment. To address it, we plan to describe better the Canoe (Cno) protein, including an explanation of its expression pattern, which is the honeycomb-like structure observed in the images.

1. Line 199, what is a 'pseudo structure'?

Response:

We used this expression to refer to unclear (e.g. dim, fuzzy) structures. We plan to improve the wording of that part of the results section.

1. Fig. 4B, 'a resolution of ~200 nm is retained at depth', please clarify how this estimate was obtained, ideally with statistics.

Response:

We agree with this comment. To address it, we plan to clarify this point in the results section, including statistical information.

1. Fig. 4D, please comment on the unphysical negative valued intensities in Fig. 4D, ideally explaining their presence in the caption. It would also be helpful to highlight where in the figure these plots arise, so the reader can visually follow along.

Response:

We agree with this comment. To address it, we plan to explain how negative intensities arise in SIM reconstruction, often a result of spherical aberrations, and we plan to indicate where the line profile in Figure 4D comes from.

1. Line 245, 'rapid mitosis'. What does rapid mean, i.e. please provide the expected timescale for mitosis.

Response:

The mitotic cycles at this developmental stage are short, e.g. 5 minutes per mitosis, compared to those of somatic cells where it takes several hours. We plan to include this information in the main text.

1. For the data in Fig. 6, was remote refocusing necessary?

Response:

Yes, it was necessary because the point of Figure 6 is to demonstrate the combination of remote focusing and SIM super-resolution in live samples. Drosophila embryos are a very good sample for this kind of demonstration, because they are often subject to micromanipulation (e.g. injection and electrophysiology), and these are the kind of experiments that can benefit greatly from the optical axial scanning of the remote focusing, where the sample can remain stationary. However, there is nothing preventing the imaging of this kind of sample with a piezo Z stage or with some other kind of mechanical actuator. In this sense, the remote focusing is not strictly necessary but still much more convenient in some applications. We plan to make this point clearer in the discussion section.

1. What is the evidence for 'reduced residual aberrations', was a comparative stack taken without AO? In general we feel that the results shown in Fig. 6 would be stronger if there were comparative results shown without AO (or remote focusing).

Response:

We agree with this comment. In general, it is difficult to make direct comparisons (e.g. as in Figures 3-5) with live samples, because of the dynamic character of the samples, where it is often impossible to capture the same scene more than once. To address this comment, we plan to revise the wording of the relevant part of the results section, to ensure that the data in Figure 6 is properly described.

1. Line 350, 'incorporation of denoising algorithms' - citations would be helpful here.

Response:

We agree with this comment. To address it, we plan to add references to the relevant statement, showing examples of denoising in 3D-SIM imaging and reconstruction.

1. Line 411, 'All three were further developed and improved' - vague, how so?

Response:

A detailed breakdown of all the changes is available on the respective software repositories. We also plan to add a summary in the supplementary material.

1. Sensorless AO description; how many Zernike modes were corrected?

Response:

We usually corrected 8 modes: Z5 to Z11 and Z22, using Noll indexing. We plan to add a table to the supplementary material, describing which modes were corrected for each dataset.

1. Multi-position aberration correction. Was the assumption of linearity in the Zernike correction verified or met? Why is this a reasonable assumption?

Response:

By their very definition, some aberrations, such as defocus and spherical aberrations, change linearly with depth. Others are also proportional to the imaging depth, and first-order approximation (i.e. straight line) is the most sensible for just two correction points, as is the case with the dataset presented in Figure 5. We plan to explain this point better in the results section.

1. Fig. S1B is not useful; if the idea is to give a visual impression of the setup, we would recommend providing more photos with approximate distances indicated so that the reader has a sense of the scale of the setup. As is - it looks like a photograph of some generic optical setup.

Response:

We agree with this comment. To address it, we plan on including more photos in the supplementary material, to give a better sense of the scale.

1. SI pattern generation - 'the maximum achievable reconstruction resolution was only slightly reduced to about 95% of the theoretical maximum'. We don't understand this sentence, as the resolution obtained on the 100 nm beads is considerably worse than 95% of the theoretical maximum. Or do the authors mean 95% of the theoretical maximum given their pitch size of 317 nm for green and 367 nm for red?

Response:

Limiting the stripe width to about 90% of what is achievable leads to a reduction of the theoretical maximum resolution to 95% of what it could be. We plan to rephrase this part to make it clearer.

1. SI Deformable mirror calibration 'spanning the range [0.1, 0.9]' - what are the units here?

Response:

These are normalised control amplitudes, i.e. [10%, 90%], which means that they are unitless. We plan to explain this in a clearer way.

1. What are the units in Fig. S5C, S5D?

Response:

Errors are in radians, defined by the calibration interferometric wavefront sensor. We plan on updating the figure to include this information.

1. It would be useful to define 'warmup' also in the caption of SI Fig. S6A.

Response:

We agree with this comment. We plan to change the caption of Figure S6A to clarify this point.

1. SI Remote Focusing, 'four offsets, {-5 mm, -2.5 mm, 2.5 mm, 5 mm}...' are the units mm or um?

Response:

The units are supposed to be um (micrometres). We plan on fixing this error.

1. '...whereas that of the 10 beads was...' here, do the authors mean the position of the beads derived from the movement of the piezo stage, as opposed to the remote focusing?

Response:

This is the average standard deviation between the 10 different beads, all from volumes acquired with remote focusing. We plan on rephrasing this part to make it clearer.

1. The authors refer to the 'results from Chapter 3.2'. What are they talking about? Do they mean a supplementary figure, or earlier supplementary results? In general, we found the discussion in this paragraph difficult to follow.

Response:

This is a remnant from an earlier version of the document which used numbered sectioning. Chapter 3.2 is referring to the section titled “Characterisation of drift and temperature effects”. We plan on revising this paragraph to make it clearer.

1. Supplementary Fig. 9 seems to be not referred to anywhere in the text.

Response:

We agree with this comment. To address this issue, we plan on referring to this figure in the main text.

1. Since the paper emphasizes 3D SIM, OTFs along the axial direction would also be useful to show, in addition to the lateral OTFs shown in Fig. 2D.

Response:

We agree with this comment. To address it, we plan on adding orthogonal views of the OTFs to the supplementary material.

1. When the sample is moved by the piezo, the axial phase of the 3D-SIM illumination pattern is stable as the sample is scanned through the illumination pattern. When remote focusing is performed, the sample is always stable so the axial phase of the 3D-SIM illumination pattern is presumably changing with remote focusing. Can the authors clarify if the 3D SIM illumination pattern is scanned when remote focusing is applied, or is the intensity pattern stable in z?

Response:

Yes, the illumination pattern is scanned. We plan on clarifying how the structured illumination works in the case of remote focusing in the supplementary material.

1. In Supplementary Fig. 9, primary spherical is referred to twice, both at index 11 and 22. The latter is presumably secondary spherical?

Response:

Yes, it is supposed to be secondary spherical aberrations. We plan on fixing this error.

1. we do not understand the x axis label, in Fig. S4D, is it really [0, 50, 50, 50] as written?

Response:

The labels of the x-axis are not well formatted. There are three range of [0, 50] where only the first zero is properly displayed. We will revise this part of the figure to make it clear.

Reviewer #2

1. The authors have provided an incomplete description of the structured illumination microscopy (SIM) reconstruction process. It is unclear whether the approach is based on 2D interference SIM configurations or 3D interference patterns. Furthermore, the specific algorithm utilized for image reconstruction has not been elucidated. Elaborating on these aspects is crucial as they significantly influence the interpretation of the resulting data.

Response:

We want to thank Reviewer #2 for bringing our attention to the incomplete description of the reconstruction process. Our approach was based on 3D interference patterns and it was carried out using the Gustafsson’s reconstruction techniques as implemented by the softWoRx software, designed for the OMX 3D-SIM microscopes. To address this comment, we plan to revise the manuscript and to include more details about the 3D-SIM reconstruction techniques in the methods and materials section.

1. The authors have stated that sample-induced aberrations caused by RI inhomogeneities within the specimen is another major reason for causing artifacts generation. Literature has demonstrated that RI inhomogeneities can lead to non-local distortions in the grid pattern, which suggests that applying uniform reconstruction parameters across the entire image may not be viable. Traditional artifact remediation using the classical Wiener method is likely insufficient under these conditions (PMID: 33896197). The existing adaptive optics (AO) approach, which employs a deformable mirror (DM) alongside an sCMOS camera, is inadequate for tackling the issue at hand. Actually the assertion made in the paper that "aberrations change approximately linearly with depth" is seemingly contradicted by simulations referenced in the cited literature (PMID: 33896197). Consequently, it appears that the current methodology might only achieve a partial mitigation of the problems associated with spherical aberration resulting from RI mismatches. It is advisable, therefore, that the authors explicitly acknowledge this limitation in their manuscript to prevent any potential misinterpretation by readers.

Response:

We are thankful for the thoughtful comment by Reviewer #2. The focus of our work was not the use of advanced 3D-SIM reconstruction and aberration correction methods; instead, we used standard ones which are not able to deal perfectly with anisoplanitism, i.e. when the aberrations vary laterally. As such, our approach provides an average reconstruction and correction across the field of view. In our particular setup this anisoplanitism was not very significant, but we agree that it could be an issue for optical systems with very wide field of view. To address this good point, we plan on clarifying these potential issues in the results and the discussion sections.

1. In Figure 2, the use of COS-7 cells, which are known for their relatively thin axial dimension, for the experiments raises an eyebrow. Notably, there are ample instances in existing research where both 2D-SIM and 3D-SIM, without the integration of adaptive optics, have yielded high-quality super-resolution images of structures such as tubulin and the endoplasmic reticulum. In addition, the authors did not present a direct comparison between BP-SIM and AO-SIM here. Without this comparative analysis, it remains ambiguous whether the enhancements in resolution and contrast and the reduction in artifacts can genuinely be attributed to the mitigation of spherical aberration. To clarify this, it would be beneficial for the authors to include side-by-side comparisons of these modalities to demonstrate the specific improvements attributed to AO-SIM.

Response:

We are grateful to Reviewer #2 for this helpful comment. In Figure 2, we demonstrate the performance we get out of 3D-SIM in terms of optical resolution. We do not make any statements about the impact of the aberration correction on image quality. Nevertheless, to address this comment, we plan to revise the figure to explain more clearly and explicitly this point.

1. In Figures 3 and 4, the authors have illustrated the enhancements achieved through the application of AO. However, there is a discernible presence of hammer-stroke and honeycomb artifacts in pre-AO imaged data, which seem to originate from the amplification of the incorrectly moved out-of-focal background in the frequency domain. Various strategies have been previously suggested to address these specific artifacts, encompassing methods like subtracting background noise in the raw images or employing selective frequency spectrum attenuation techniques, such as Notch filtering and High-Fidelity SIM. To facilitate a more comprehensive understanding, I would recommend that the authors incorporate into their study a comparison that includes BP-SIM data that has undergone either background subtraction or frequency spectrum attenuation. This added data would enable a more complete evaluation and comparison regarding the merits and impact of their AO approach.

Response:

We thank the reviewer for this excellent suggestion and we agree that a pre-processing step, such as background subtraction or frequency spectrum attenuation, can help with the reduction of artefacts. To address this comment, we will re-analyse our data and apply these techniques, and we will add the data to the manuscript, with an appropriate revision to the text.

Reviewer #3

1. There is an overall reference in the manuscript of the novelty possible range of applications of using an upright microscope configuration. Examples mentioned are tissue-based imaging, access to whole-mount specimens for manipulation and electrophysiology. However, authors fail to present any such applications. There is not a single example presented which could not have been obtained with an inverted microscope. Could the authors provide an example where a water-dipping is used. Expanded samples could be one case, since the thickness of the gel makes it difficult to image with an inverted microscope. Another possible example would be to label the extracellular space and do shadow imaging of the tissue (SUSHI PMID: 29474910). ExM might be simpler to do as part of revising the manuscript than SUSHI.

Response:

We are thankful to Reviewer #3 for these interesting comments. To address this comment, we will emphasise more clearly that Figure 6 of our manuscript shows a sample that is often part of live imaging experiments that require microinjection and even electrophysiology. Our aim was to show the proof of principle and the potential of such experiments, rather than to carry out real and complex experiments using electrophysiology or microinjection. Regarding providing an example where water-dipping is used, this is already present in the same Figure 6, which we will describe more explicitly and fully in the revised manuscript. The reviewer’s comments on expansion microscopy and SUSHI are interesting, but the primary purpose of our microscope system is to facilitate super resolution live cell imaging experiments. Nevertheless, to address this comment, we will add an explanation of the relevance of our approach to improving deep super resolution imaging of expanded specimens.

1. On the main text it is described a 5-fold volumetric resolution, which is confusing since authors only mention lateral and axial resolutions. Their measurements correspond to a ~1.6-fold lateral improvement and ~1.7-fold axial improvement. These are however not the 95% of the achievable resolution theoretical maximum, as stated in p7 SI (2 fold increase of 282nm), but only the 80-85%. This point should be rephrased in the manuscript.

Response:

We want to thank Reviewer #3 for bringing up this important point. To address it, we plan to make changes to the text, both in the main manuscript and in the supplementary material, to make it clearer what are the resolution improvement that we achieve and what are the limitations to our approach.

1. [OPTIONAL] p4 and related to figure 2, it would be important to report also measurements of beads with SIM but without AO, just as done for WF. Is there an improvement of using AO on SIM? This is reported for the fixed cells but not for the beads.

Response:

We found no significant improvement in resolution when AO was applied to SIM. To address this comment, we plan to add the extra data to Figure 2, demonstrating this point.

1. Figure 2, it is odd the comparison between WF+/- AO and SIM +/- AO are done using different cellular structures. Since wavelengths used are not the same it is difficult to interpret if there is any improvement of using AO on SIM compared to SIM without AO. Same questions arise as above, Is there an improvement of using AO on SIM?

Response:

We agree that the data in Figure 2C and 2D is presented in unusual way. Our intention was not to make a comparison between bypass and AO, but instead to characterise the super-resolution capabilities of the system. We use different channels because doing -/+ AO consecutively leads to noticeable intensity drop due to photobleaching. We are grateful to Reviewer #3 for the valuable comment, which we plan to address by revising Figure 2.

1. "A significant benefit and uniqueness of the Deep3DSIM design is its upright configuration, whereas commercial SIM systems are built around inverted microscopes and are usually restricted to imaging thin samples, such as cultured cells." (p5) is not correct. The commercial DeepSIM module from CREST Optics can be mounted on an inverted microscope as well as image deep into tissue (seehttps://crestoptics.com/deepsim/ and application notes therein) and be used with essentially any objective. This point should be rephrased in the text.

Response:

We want to thank Reviewer #3 for bringing our attention to this error. Of course, we meant commercial 3D-SIM systems, such as GE Healthcare DeltaVision OMX and Nikon N-SIM. To address this issue, we plan to rephrase this part of the results section. Regarding the commercial DeepSIM module from CREST Optics, as far as we can tell, it uses a different method – 2D lattice multi-spot SIM – which comes at the cost of signal loss when sample-induced aberrations are strong. This is very different from our method, which uses a deformable mirror to manipulate the phase information of both the excitation and the emission light at the back-pupil plane of the objective lens, which can theoretically provide 2× resolution enhancement with no signal lost.

1. Fig 3 reports the improvements of AO on SIM for imaging over 10um in tissue. What are the Zernike modes measured? Or how does the pupil look like before and after correction? It would be also good to report the Fourier amplitudes as done in Fig 2C as a quantitative measure of improvement. It would be good to point out the artifacts observed on the BP SIM image reconstruction (labelled with 3x, fringes are noticeable).

Response:

We thank Reviewer #3 for the good suggestions. We plan to add information about the measured Zernike modes to the results section, as well as to add a brief discussion about the noticeable reconstruction artefacts. In terms of pupil and Fourier amplitudes, we plan to change Figure 3 to include all this information or, alternatively, to include it in the supplementary material.

1. Many key details relating to image acquisition and AO correction are missing for all figures. How is the AO optimization implemented? Is it implemented via a genetic algorithm (progressive optimization of parameters) or using more clever strategies? Not clear if the optimization is implemented using images obtained with flat illumination or after SIM imaging/processing of a given dataset. How long does the AO optimization take? How sensitive to noise is the process? What metric do they use to estimate the sensorless AO correction? On pag12, they say "Fourier domain image metric" for measurements with fine details; otherwise, ISOsense when not high frequencies are present. Could the authors report the formula used to calculate the first metric? What do they consider to be low and high frequencies in this case? Is there a reason why ISOsense is not always used, or is there an automatic way to choose between the two? How many images were acquired for AO correction? Which samples were corrected with ISOsense and which ones with Fourier domain image metric? (see for example the detailed experimental reporting in the Supp Mat from Lin et al Nat Commun 2021).

Response:

We are grateful to Reviewer #3 for the extensive list of questions. The optimisation is done via non-linear least square, it uses widefield images, and it is performed before the actual image acquisition, i.e. well before any SIM reconstruction takes place. The methods used for aberration correction are described in the Methods and materials section, and further in the cited literature, e.g. Antonello et al 2020 and Hall et al 2020. ISOsense needs to be manually chosen over the Fourier image metric, and this should be done when large mode biases lead to small changes in the metric value, which is likely to happen when there are little or no sharp features in the images. One of the disadvantages of our implementation of ISOsense is that the structured illumination pattern is continuously exposed over the sample, which leads to photobleaching and phototoxicity. None of the datasets shown in the manuscript use ISOsense. To address all of the questions from this comment, we plan to significantly expand our descriptions of the AO methods, both in the main text and in the supplementary material.

1. Fig 4. Data presented for larval brain tissue is a very clear example of adding AO to image deep into tissue as the effect at ~130 cannot be understated. Here too, it would be also good to report the Fourier amplitudes as done in Fig 2C as a quantitative measure of improvement and possibly the SNR of reconstructed images. Having a way to quantitatively describe how much better are those images would be great. Also, what are the aberrations corrected? Can the wavefront or Zernike amplitude of the modes be reported? Same as for Fig 3, details about AO correction are missing.

Response:

We are grateful to Reviewer #3 for the helpful comment. We will address it by adding the Fourier amplitudes to Figure 4, as suggested, and by reporting the Zernike mode amplitudes of the aberration corrections.

1. [OPTIONAL] "It is worth noting that aberrations can differ across larger fields, and therefore, after applying an average correction, residual aberrations can still be observed in some regions of the AO-corrected bead images. However, the overall PSF shapes were still dramatically improved with AO compared to the equivalent without AO." This point is very interesting although not result either in the main text or in the SI is presented.

Response:

The residual aberrations are present in the right image of Figure 4B, although we did not highlight them specifically. We are thankful to Reviewer #3 for the good suggestion and we plan to implement it by changing Figure 4 to show a few of the beads with residual aberrations.

1. "As we found that the aberrations change approximately linearly in depth, we could measure the aberration in two planes and calculate the corrections in intermediate planes by interpolation, an approach which we termed "multi-position AO"." This is, personally, one of the major contributions of this work to the community. Unfortunately, it is not reported in detail. Not only for SIM but for imaging with WF or confocal, such linear change for aberrations with depth is not well known. Again, here the details of AO correction and image metrics are missing. To establish that for most thick biological structures 'aberrations change approximately linearly in depth' would be foundational to the widespread use of AO within standard imaging. Would it be possible for the authors to elaborate on this point and present detailed results? What is the error from measuring and correcting more than 2 planes? What is the error from just measuring and AO correcting at the deeper plane, i.e. from a single measurement? Authors could also show a case in which a linear assumption works nicely (or how well it works). For example, comparing an intermediate plane (or a plane beyond) imaged after AO optimization or after coefficient interpolation of the Zernike modes and compare it against correcting directly that plane.

Response:

Some aberrations, such as defocus and spherical aberrations, are mathematically defined as varying linearly with depth. The change in other aberrations with depth can also be estimated with a linear model, which is a standard first-order approximation in the case of two datapoints, such as corrections done in Figure 5. It is not possible to do regression analysis with just a single point, so it is impossible to apply our multi-position AO at a single plane. We are grateful to Reviewer #3 for the constructive comment. To address the questions in this comment, we plan to provide a more detailed description of the correction estimation methods to the results section, as well as a discussion on the accuracy of the linear model in the discussion section.

1. The image of the cos-7 cell in metaphase, for Fig 5 is, however, very disappointing. See Fig 1 of Novak et al Nat Commun 2018 for an example of a single z-plane of a cell in metaphase. Having the possibility to correct for the entire 3D volume, I would expect amazing 3D volumes (movies and/or projections) associated with this imaging which are not presented.

Response:

We thank Reviewer #3 for the interesting comment. The example in Novak et al 2018 was acquired with STED microscopy, which is an entirely different imaging method and thus produces different results. Nevertheless, we will revise the discussion of Figure 5 to ensure that the right expectations are set.

1. In Figure 6, they use AO in remote configuration mode to allow imaging of live specimens. It needs to be clarified if this is an a priori characterization that is then kept fixed while recording in time. The last acquired volume of fig 6A and B have a higher amount of artifacts with respect to time 00:00. Are those artifacts due to lower SNR (maybe due to sample bleaching) or due to some change in the aberrations of the specimen?

Response:

We want to thank Reviewer #3 for the valuable comment. We assume that by change in artefacts, Reviewer #3 is referring to the overall green fluorescent structure. Indeed, this last volume shows the anaphase to telophase transition where the mitotic spindle is being reorganised and disassembled. As such, the structure is much less well-defined than in the first volume. The changes in aberrations over time are not particularly significant in this case, and the photobleaching is not that impactful in such an experiment where relatively thin volumes are acquired with substantial time delay between them. To address this comment, we plan to revise the discussion of the figure and to ensure that the scene observed in the last volume is clearer.

1. "These results demonstrate that the remote focusing functionality of the system can be successfully applied for live 3D-SIM experiments, allowing four-dimensional acquisition while keeping the specimen stationary, thus avoiding the usual agitation and perturbations associated with mechanical actuation." Generally, this statement is true, but for the specific example shown of drosophila embryogenesis is it relevant? If they use piezo-driven Z-stack imaging with AO, does that lead to incorrect reconstructions or motion-induced artifacts? Related to the results shown in Fig 6, the fair comparison would be AO SIM vs SIM (without AO), not AO SIM vs AO WF.

Response:

We are grateful to Reviewer #3 for the insightful comment. Drosophila embryos are quite robust to perturbations due to their shape and size, and the restrictions imposed by SIM experiments (e.g. small Z steps and Z levels held for long periods of time) make motion-induced artefacts not very impactful. Regarding the results, the point of Figure 6 is not to demonstrate the advantages of aberration correction, which we do not claim in the caption or in the relevant part of the discussion, but to demonstrate that remote focusing works well with 3D-SIM reconstruction, which is known to have stringent requirements about the image quality. To address this comment, we plan to revise the figure and its relevant part of the results section.

1. When performing remote focusing, is the effective NA of the imaged plane changing with respect to the NA of the objective used at its focal plane?

Response:

We thank Reviewer #3 for the good question. The effective NA is not altered by the remote focusing. We plan to mention this detail in the results section.

1. [OPTIONAL] Did the authors run calculations to explore whether a commercial upright microscope could be used instead of their design? Are there any fundamental flaws that would make impossible using a commercial base? If not, could an AO SIM module be designed such that it adds on a commercial base? It would be important to discuss this point.

Response:

We thank Reviewer #3 for bringing up this interesting point. A lot of considerations, calculations, and modelling were done in the design of the Deep3DSIM system. Of course, the use of a commercial upright microscope stand was part of the deliberation. One of the obvious limitations is the difficult access to the pupil-conjugated plane. On the other hand, a commercial microscope stand is not well compatible with many of the key parts of the system, which were designed around specific biological applications, such as dual camera system for fast live simultaneous imaging and the heavy-duty Z stage intended to support two heavy micromanipulators. To address this comment, we plan to add a discussion of the compatibility of Deep3DSIM with commercial microscope stands to the discussion section and the supplementary material.

Minor comments:

1. Fig 2 lacks a color bar for D panels, which is in log scale. Authors should also show the Fourier transform along the z direction.

Response:

The colour mapping in Figure 2 uses the lookup tables called Cyan Hot and Orange Hot, as indicated in the caption, which come from the ImageJ software. To address this comment, we plan to improve the caption to reflect the fact that the plots are in log scale. We also want to include Fourier transforms along Z, either in the figure itself or in the supplementary material.

1. p4, "Such minor aberrations tend to be insignificant in conventional microscopy modalities such as widefield and confocal (Wang and Zhang, 2021). Therefore..." If optical aberrations are insignificant for single cells in widefield and confocal why do experiments here? These sentences should be rephrased to motivate better the experiments performed.

Response:

We agree with this comment. To address it, we plan to rephrase this part of the results section to motivate better the experiments.

1. Imaged microtubules look abnormal, 'dotty' (figure 2) in both WF and SIM. See https://sim.hms.harvard.edu/portfolio/microtubules/ or Fig 1 of Wegel, et al Dobbie Sci Rep 2016, for better examples of continuous microtubule structures as imaged with SIM.

Response:

The dottiness of the microtubule structures is not related to the SIM reconstruction, because the same dottiness is seen in the respective WF data, too. It is a product of the sample preparation and it has only aesthetic significance. Nevertheless, to address this comment we plan to mention the dottiness in the results section.

1. Is also the remote focusing performed via optimization of metrics similar to the one used for compensating aberrations?

Response:

Yes, as mentioned in the Methods and materials (p. 13), the calibration of the remote focusing involved sensorless aberration correction of several Zernike modes, such as defocus and spherical aberrations.

1. Figure 2, the order of names on the top right of the panel should match the order of curves presented.

Response:

We agree with this comment. To address it, we plan to reorder the curves in Figure 2.

1. I value the efforts to improve open-source tools for system and AO control and GUI. And those tools seemed to have been modified for this work, although those modifications are not described. Would it be possible for the authors to describe those modifications?

Response:

A detailed breakdown is publicly available at the respective software repositories. To address this comment, we plan to add a summary of software changes to the supplementary material.

1. Reported average values of the FWHM of imaged beads in 3D (p4) require also to report errors associated with those measurements.

Response:

We agree with this comment. To address it, we plan to add statistical information to the FWHM values on page 4.

1. Page 13, second paragraph states that "The results from chapter 3.2..." I believe that was a copy/paste from a thesis but should be corrected for a peer-reviewed publication, as there is no chapter 3.2.

Response:

This is a leftover from an older version of the document which used numbered sectioning. In this case “chapter 3.2” refers to subsection “Characterisation of drift and temperature effects”. We plan on fixing this mistake in the revised manuscript.

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

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary

The work of Wang et al entitled "Deep super-resolution imaging of thick tissue using structured illumination with adaptive optics" presents the use of a deformable mirror to simultaneously perform adaptive optics 'AO' and remote focusing 'RF' on a custom-designed upright microscope configuration. The work is novel and represents a timely application of this type of technology to imaging biological specimens. AO enables the correction of refractive index mismatch and sample-induced aberrations while remote focusing allows focusing through the sample without moving the specimen or the objective. The use of AO improves the final image reconstructed using a traditional SIM processing strategy. I greatly value the idea presented (SI pattern generation p7 Supp Inf) about maximizing the contrast of the projected structured illumination. This could be an excellent way to improve SIM imaging since image reconstructions suffer from artifacts when signal to noise ratio is low. However, since it is only one of the factors considered for reducing the stripe width it is unclear how it compares to imaging with the width that maximizes resolution.

Authors do also a very good job describing, characterizing and designing experiments to deal with instabilities exhibited by the deformable mirror.

One of the key aspects of the paper, that could be stressed more, is that including AO gives access to better-quality raw images that are then be used for standard reconstruction pipeline SIM processing. When aberrations are compensated, the illumination pattern closely matches what is expected by the SIM image formation model. Since these raw recorded images are sharper and closer to the actual assumption behind the SIM image reconstruction model, they will have a major positive impact in reducing artifacts that the inversion algorithm is returning. This is particularly evident in Figure 4.

Major comments:

• There is an overall reference in the manuscript of the novelty possible range of applications of using an upright microscope configuration. Examples mentioned are tissue-based imaging, access to whole-mount specimens for manipulation and electrophysiology. However, authors fail to present any such applications. There is not a single example presented which could not have been obtained with an inverted microscope. Could the authors provide an example where a water-dipping is used. Expanded samples could be one case, since the thickness of the gel makes it difficult to image with an inverted microscope. Another possible example would be to label the extracellular space and do shadow imaging of the tissue (SUSHI PMID: 29474910). ExM might be simpler to do as part of revising the manuscript than SUSHI.
• On the main text it is described a 5-fold volumetric resolution, which is confusing since authors only mention lateral and axial resolutions. Their measurements correspond to a ~1.6-fold lateral improvement and ~1.7-fold axial improvement. These are however not the 95% of the achievable resolution theoretical maximum, as stated in p7 SI (2 fold increase of 282nm), but only the 80-85%. This point should be rephrased in the manuscript.
• [OPTIONAL] p4 and related to figure 2, it would be important to report also measurements of beads with SIM but without AO, just as done for WF. Is there an improvement of using AO on SIM? This is reported for the fixed cells but not for the beads.
• Figure 2, it is odd the comparison between WF+/- AO and SIM +/- AO are done using different cellular structures. Since wavelengths used are not the same it is difficult to interpret if there is any improvement of using AO on SIM compared to SIM without AO. Same questions arise as above, Is there an improvement of using AO on SIM?
• "A significant benefit and uniqueness of the Deep3DSIM design is its upright configuration, whereas commercial SIM systems are built around inverted microscopes and are usually restricted to imaging thin samples, such as cultured cells." (p5) is not correct. The commercial DeepSIM module from CREST Optics can be mounted on an inverted microscope as well as image deep into tissue (see https://crestoptics.com/deepsim/ and application notes therein) and be used with essentially any objective. This point should be rephrased in the text.
• Fig 3 reports the improvements of AO on SIM for imaging over 10um in tissue. What are the Zernike modes measured? Or how does the pupil look like before and after correction? It would be also good to report the Fourier amplitudes as done in Fig 2C as a quantitative measure of improvement. It would be good to point out the artifacts observed on the BP SIM image reconstruction (labelled with 3x, fringes are noticeable).
• Many key details relating to image acquisition and AO correction are missing for all figures. How is the AO optimization implemented? Is it implemented via a genetic algorithm (progressive optimization of parameters) or using more clever strategies? Not clear if the optimization is implemented using images obtained with flat illumination or after SIM imaging/processing of a given dataset. How long does the AO optimization take? How sensitive to noise is the process? What metric do they use to estimate the sensorless AO correction? On pag12, they say "Fourier domain image metric" for measurements with fine details; otherwise, ISOsense when not high frequencies are present. Could the authors report the formula used to calculate the first metric? What do they consider to be low and high frequencies in this case? Is there a reason why ISOsense is not always used, or is there an automatic way to choose between the two? How many images were acquired for AO correction? Which samples were corrected with ISOsense and which ones with Fourier domain image metric? (see for example the detailed experimental reporting in the Supp Mat from Lin et al Nat Commun 2021).
• Fig 4. Data presented for larval brain tissue is a very clear example of adding AO to image deep into tissue as the effect at ~130 cannot be understated. Here too, it would be also good to report the Fourier amplitudes as done in Fig 2C as a quantitative measure of improvement and possibly the SNR of reconstructed images. Having a way to quantitatively describe how much better are those images would be great. Also, what are the aberrations corrected? Can the wavefront or Zernike amplitude of the modes be reported? Same as for Fig 3, details about AO correction are missing.
• [OPTIONAL] "It is worth noting that aberrations can differ across larger fields, and therefore, after applying an average correction, residual aberrations can still be observed in some regions of the AO-corrected bead images. However, the overall PSF shapes were still dramatically improved with AO compared to the equivalent without AO." This point is very interesting although not result either in the main text or in the SI is presented.
• "As we found that the aberrations change approximately linearly in depth, we could measure the aberration in two planes and calculate the corrections in intermediate planes by interpolation, an approach which we termed "multi-position AO"." This is, personally, one of the major contributions of this work to the community. Unfortunately, it is not reported in detail. Not only for SIM but for imaging with WF or confocal, such linear change for aberrations with depth is not well known. Again, here the details of AO correction and image metrics are missing. To establish that for most thick biological structures 'aberrations change approximately linearly in depth' would be foundational to the widespread use of AO within standard imaging. Would it be possible for the authors to elaborate on this point and present detailed results? What is the error from measuring and correcting more than 2 planes? What is the error from just measuring and AO correcting at the deeper plane, i.e. from a single measurement? Authors could also show a case in which a linear assumption works nicely (or how well it works). For example, comparing an intermediate plane (or a plane beyond) imaged after AO optimization or after coefficient interpolation of the Zernike modes and compare it against correcting directly that plane.
• The image of the cos-7 cell in metaphase, for Fig 5 is, however, very disappointing. See Fig 1 of Novak et al Nat Commun 2018 for an example of a single z-plane of a cell in metaphase. Having the possibility to correct for the entire 3D volume, I would expect amazing 3D volumes (movies and/or projections) associated with this imaging which are not presented.
• In Figure 6, they use AO in remote configuration mode to allow imaging of live specimens. It needs to be clarified if this is an a priori characterization that is then kept fixed while recording in time. The last acquired volume of fig 6A and B have a higher amount of artifacts with respect to time 00:00. Are those artifacts due to lower SNR (maybe due to sample bleaching) or due to some change in the aberrations of the specimen?
• "These results demonstrate that the remote focusing functionality of the system can be successfully applied for live 3D-SIM experiments, allowing four-dimensional acquisition while keeping the specimen stationary, thus avoiding the usual agitation and perturbations associated with mechanical actuation." Generally, this statement is true, but for the specific example shown of drosophila embryogenesis is it relevant? If they use piezo-driven Z-stack imaging with AO, does that lead to incorrect reconstructions or motion-induced artifacts? Related to the results shown in Fig 6, the fair comparison would be AO SIM vs SIM (without AO), not AO SIM vs AO WF.
• When performing remote focusing, is the effective NA of the imaged plane changing with respect to the NA of the objective used at its focal plane?
• [OPTIONAL] Did the authors run calculations to explore whether a commercial upright microscope could be used instead of their design? Are there any fundamental flaws that would make impossible using a commercial base? If not, could an AO SIM module be designed such that it adds on a commercial base? It would be important to discuss this point.

Minor comments

• Fig 2 lacks a color bar for D panels, which is in log scale. Authors should also show the Fourier transform along the z direction.
• p4, "Such minor aberrations tend to be insignificant in conventional microscopy modalities such as widefield and confocal (Wang and Zhang, 2021). Therefore..." If optical aberrations are insignificant for single cells in widefield and confocal why do experiments here? These sentences should be rephrased to motivate better the experiments performed.
• Imaged microtubules look abnormal, 'dotty' (figure 2) in both WF and SIM. See https://sim.hms.harvard.edu/portfolio/microtubules/ or Fig 1 of Wegel, et al Dobbie Sci Rep 2016, for better examples of continuous microtubule structures as imaged with SIM.
• Is also the remote focusing performed via optimization of metrics similar to the one used for compensating aberrations?
• Figure 2, the order of names on the top right of the panel should match the order of curves presented.
• I value the efforts to improve open-source tools for system and AO control and GUI. And those tools seemed to have been modified for this work, although those modifications are not described. Would it be possible for the authors to describe those modifications?
• Reported average values of the FWHM of imaged beads in 3D (p4) require also to report errors associated with those measurements.
• Page 13, second paragraph states that "The results from chapter 3.2..." I believe that was a copy/paste from a thesis but should be corrected for a peer-reviewed publication, as there is no chapter 3.2.

Referee Cross-Commenting

The other two reviewers raise relevant and important points that would contribute to the overall improvement of the work. I think that authors should try to address most, if not all, of the comments as long as they don't require more than 3-6 months to get done.

Significance

General assessment:

Although a very good and timely idea is presented the overall the manuscript still needs a lot of work. There is a lack of many key details of AO correction, all applications chosen could have been done in an inverted scope and some of the example images reported are suboptimal (Fig 2 and 5) that need further experimental work. Details and metrics, one example of the advantage of using an upright microscope and overall better examples of imaged cells could be provided.

This work builds upon recent work of implementing AO for 3D SIM (Lin et al Nat Commun 2021) to propose to use a deformable mirror to perfrom AO as well as remote focusing in an upright microscope configuration.

Audience: this work will be of interest for a specialized group of researchers, but it will contribute to the goal of adding AO tools to every microscope that will greatly impact the whole imaging community.

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

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

The authors want to develop a structured illumination microscopy (SIM) system for deep tissue superresolution imaging. Here they have developed a SIM system based on the upright configration, and use deformable mirror to compensate for the relection index (RI) mismatch and improve the resolution and contrast in deep tissues. They also showed examples of SR imaging of COS-7 cells and Drosophila larval brains and embryos.

However, I do have some concerns regarding the paper.

1. The authors have provided an incomplete description of the structured illumination microscopy (SIM) reconstruction process. It is unclear whether the approach is based on 2D interference SIM configurations or 3D interference patterns. Furthermore, the specific algorithm utilized for image reconstruction has not been elucidated. Elaborating on these aspects is crucial as they significantly influence the interpretation of the resulting data.
2. The authors have stated that sample-induced aberrations caused by RI inhomogeneities within the specimen is another major reason for causing artifacts generation. Literature has demonstrated that RI inhomogeneities can lead to non-local distortions in the grid pattern, which suggests that applying uniform reconstruction parameters across the entire image may not be viable. Traditional artifact remediation using the classical Wiener method is likely insufficient under these conditions (PMID: 33896197). The existing adaptive optics (AO) approach, which employs a deformable mirror (DM) alongside an sCMOS camera, is inadequate for tackling the issue at hand. Actually the assertion made in the paper that "aberrations change approximately linearly with depth" is seemingly contradicted by simulations referenced in the cited literature (PMID: 33896197). Consequently, it appears that the current methodology might only achieve a partial mitigation of the problems associated with spherical aberration resulting from RI mismatches. It is advisable, therefore, that the authors explicitly acknowledge this limitation in their manuscript to prevent any potential misinterpretation by readers.
3. In Figure 2, the use of COS-7 cells, which are known for their relatively thin axial dimension, for the experiments raises an eyebrow. Notably, there are ample instances in existing research where both 2D-SIM and 3D-SIM, without the integration of adaptive optics, have yielded high-quality super-resolution images of structures such as tubulin and the endoplasmic reticulum. In addition, the authors did not present a direct comparison between BP-SIM and AO-SIM here. Without this comparative analysis, it remains ambiguous whether the enhancements in resolution and contrast and the reduction in artifacts can genuinely be attributed to the mitigation of spherical aberration. To clarify this, it would be beneficial for the authors to include side-by-side comparisons of these modalities to demonstrate the specific improvements attributed to AO-SIM.
4. In Figures 3 and 4, the authors have illustrated the enhancements achieved through the application of AO. However, there is a discernible presence of hammer-stroke and honeycomb artifacts in pre-AO imaged data, which seem to originate from the amplification of the incorrectly moved out-of-focal background in the frequency domain. Various strategies have been previously suggested to address these specific artifacts, encompassing methods like subtracting background noise in the raw images or employing selective frequency spectrum attenuation techniques, such as Notch filtering and High-Fidelity SIM. To facilitate a more comprehensive understanding, I would recommend that the authors incorporate into their study a comparison that includes BP-SIM data that has undergone either background subtraction or frequency spectrum attenuation. This added data would enable a more complete evaluation and comparison regarding the merits and impact of their AO approach.

Significance

The authors want to develop a structured illumination microscopy (SIM) system for deep tissue superresolution imaging. Here they have developed a SIM system based on the upright configration, and use deformable mirror to compensate for the relection index (RI) mismatch and improve the resolution and contrast in deep tissues.

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

Review, 3D SIM + AO, Wang and coworkers

In this manuscript, Wang and coworkers report an upright 3D SIM system with adaptive optics (AO) correction. They demonstrate that AO improves imaging into thick 3D samples, including Drosophila larval brain. They also explore the use of remote focusing with their setup. The authors clearly demonstrate a gain with AO, and we are convinced that the microscope they build offers some utility over existing state of the art, particularly in samples thicker than a single cell. That said, we have concerns with the manuscript that we would like to see addressed before recommending publication:

• Given the emphasis on super-resolution imaging deep inside a sample, we were surprised to see no mention of other forms of structured illumination that allow super-resolution imaging in samples thicker than a single cell. These include the 'spot-scanning' implementations of SIM that offer better imaging at depth by virtue of pinholes, and include MSIM, iSIM, and rescan confocal technologies. The two-photon / AO implementation of iSIM seems particularly germane, e.g. https://pubmed.ncbi.nlm.nih.gov/28628128/ Please consider citing these works, as they help place the existing work into context.
• As we're sure the authors appreciate, besides aberrations, a major additional obstacle to 3D SIM in thick tissues is the presence of out-of-focus background. Indeed, this point was mentioned by Gustafsson in his classic 2008 paper on 3D SIM (https://pubmed.ncbi.nlm.nih.gov/18326650/): 'The application area of three-dimensional structured illumination microscopy overlaps with that of confocal microscopy, but the two techniques have different and complementary strengths. Structured illumination microscopy offers higher effective lateral resolution, because it concentrates much of the excitation light at the very highest illumination angles, which are most effective for encoding high-resolution information into the observed data, whereas confocal microscopy spreads out its illumination light more or-less uniformly over all available angles to form a focused beam. For very thick and compactly fluorescent samples, however, confocal microscopy has an advantage in that its pinhole removes out-of focus light physically. Structured illumination microscopy is quite effective at removing out-of-focus light computationally, because it is not subject to the missing-cone problem, but computational removal leaves behind the associated shot noise. Therefore confocal microscopy may be preferable on very thick and dense samples, for which the in-focus information in a conventional microscope image would be overwhelmed by out-of-focus light, whereas structured illumination microscopy may be superior in a regime of thinner or sparser samples.' This point is not mentioned at all in the manuscript, yet we are certain it is at least partially responsible for the residual image artifacts the authors mention. Please discuss the problem of out of focus light on 3D samples, particularly with an eye to the 'spot-scanning' papers mentioned above.
• The authors use a water dipping lens, yet they image into samples that are mounted on coverslips, i.e. they use a dipping lens to image through a coverslip: see attached pdf for reference

This almost certainly introduces spherical aberration, which the authors seem to observe: see attached pdf for reference

We find this troubling, as it seems that in the process of building their setup, the authors have made a choice of objective lens that introduces aberrations - that they later correct. At the very least, this point needs to be acknowledged in the manuscript (or please correct us if we're wrong) - as it renders the data in Figs. 3-4 somewhat less compelling than if the authors used an objective lens that allowed correction through a coverglass, e.g. a water dipping lens with a correction collar. In other words, in the process of building their AO setup, the authors have introduced system aberrations that render the comparison with 3D SIM somewhat unfair. Ideally the authors would show a comparison with an objective lens that can image through a glass coverslip. - The authors tend to include numbers for resolution without statistics. This renders the comparisons meaningless in my opinion; ideally every number would have a mean and error bar associated with it. We have included specific examples in the minor comments below. - In Fig. 5, after the 'multipoint AO SIM', the SNR in some regions seems to decrease after AO: see attached pdf for reference

Please comment on this issue.

• Please provide timing costs for the indirect AO methods used in the paper, so the reader understands how this time compares to the time required for taking a 3D SIM stack. In a similar vein, the authors in Lines 213-215, mention a 'disproportionate measurement time' when referring to the time required for AO correction at each plane - providing numbers here would be very useful to a reader, so they can judge for themselves what this means. What is the measurement time, why is it so long, and how does it compare to the time for 3D SIM? It would also be useful to provide a comparison between the time needed for AO correction at each (or two) planes without remote focusing (RF) vs. with RF, so the reader understands the relative temporal contributions of each part of the method. We would suggest, for the data shown in Fig. 5, to report a) the time to acquire the whole stack without AO (3D SIM only); b) the time to acquire the data as shown; c) the time to acquire the AO stack without RF. This would help bolster the case for remote focusing in general; as is we are not sure we buy that this is a capability worth having, at least for the data shown in this paper.
• Some further discussion on possibly extending the remote focusing range would be helpful. We gather that limitations arose from an older model of the DM being used, due to creep effects. We also gather from the SI that edge effects at the periphery of the DM was also problematic. Are these limitations likely non-issues with modern DMs, and how much range could one reasonably expect to achieve as a result? We are wondering if the 10 um range is a fundamental practical limitation or if in principle it could be extended with commercial DMs.

Minor comments

• The paper mentions Ephys multiple times, even putting micromanipulators into Fig. 1 - although it is not actually used in this paper. If including in Figure 1, please make it clear that these additional components are aspirational and not actually used in the paper.
• The abstract mentions '3D SIM microscopes', 'microscopes' redundant as the 'm' in 'SIM' stands for 'microscope'.
• 'fast optical sectioning', line 42, how can optical sectioning be 'fast'? Do they mean rapid imaging with optical sectinong?
• line 59, 'effective imaging depth may be increased to some extent using silicone immersion objectives', what about water immersion objectives? We would guess these could also be used.
• line 65 - evidence for 'water-dipping objectives are more sensitive to aberrations' ? Please provide citation or remove. They are certainly more prone to aberrations if used with a coverslip as done here.
• 'fast z stacks' is mentioned in line 103. How fast is fast?
• line 116 'we imaged 100 nm diameter green fluorescent beads'. Deposited on glass? Given that this paper is about imaging deep this detail seems worth specifying in the main text.
• lines 127-130, when describing changes in the bead shape with numbers for the FWHM, please provide statistics - quoting single numbers for comparison is almost useless and we cannot conclude that there is a meaningful improvement without statistics.
• In the same vein, how can we understand that remote focus actually improves the axial FWHM of the widefield bead? Is this result repeatable, or it just noise?
• line 155, 'Because of the high spatial information...' -> 'Because of the high resolution spatial information...'
• When quoting estimated resolution #s from microtubules (lines 158-163) similarly please provide statistics as for beads.
• It seems worth mentioning the mechanism of AO correction (i.e. indirect sensing) in the main body of the text, not just the methods.
• How long do the AO corrections take for the datasets in the paper?
• Were the datasets in Fig. 2-4 acquired with remote focusing, or in conventional z stack mode? Please clarify this point in the main text and the figure captions.
• It would be helpful when showing z projections in Figs. 3-5 to indicate the direction of increasing depth (we assume this is 'down' due to the upright setup, but this would be good to clarify)
• line 174, 'showed significant improvements in both intensity and contrast after reconstruction' - we see the improvements in contrast and resolution, it is harder to appreciate improvements in intensity. Perhaps if the authors showed some line profiles or otherwise quantified intensity this would be easier to appreciate.
• line 195 'reduced artefacts' due to AO. We would agree with this statement - the benefit from AO is obvious, and yet there are still artefacts. If the authors could clarify what these (residual) artefacts are, and their cause (out of focus light, uncorrected residual aberrations, etc) this would be helpful for a reader that is not used to looking at 3D SIM images.
• Line 197, 'expected overall structure', please clarify what is expected about the structure and why.
• Line 199, what is a 'pseudo structure'?
• Fig. 4B, 'a resolution of ~200 nm is retained at depth', please clarify how this estimate was obtained, ideally with statistics.
• Fig. 4D, please comment on the unphysical negative valued intensities in Fig. 4D, ideally explaining their presence in the caption. It would also be helpful to highlight where in the figure these plots arise, so the reader can visually follow along.
• Line 245, 'rapid mitosis'. What does rapid mean, i.e. please provide the expected timescale for mitosis.
• For the data in Fig. 6, was remote refocusing necessary?
• What is the evidence for 'reduced residual aberrations', was a comparative stack taken without AO? In general we feel that the results shown in Fig. 6 would be stronger if there were comparative results shown without AO (or remote focusing).
• Line 350, 'incorporation of denoising algorithms' - citations would be helpful here.
• Line 411, 'All three were further developed and improved' - vague, how so?
• Sensorless AO description; how many Zernike modes were corrected?
• Multi-position aberration correction. Was the assumption of linearity in the Zernike correction verified or met? Why is this a reasonable assumption?
• Fig. S1B is not useful; if the idea is to give a visual impression of the setup, we would recommend providing more photos with approximate distances indicated so that the reader has a sense of the scale of the setup. As is - it looks like a photograph of some generic optical setup.
• SI pattern generation - 'the maximum achievable reconstruction resolution was only slightly reduced to about 95% of the theoretical maximum'. We don't understand this sentence, as the resolution obtained on the 100 nm beads is considerably worse than 95% of the theoretical maximum. Or do the authors mean 95% of the theoretical maximum given their pitch size of 317 nm for green and 367 nm for red? SI Deformable mirror calibration

'spanning the range [0.1, 0.9]' - what are the units here?

What are the units in Fig. S5C, S5D?

It would be useful to define 'warmup' also in the caption of SI Fig. S6A. SI Remote Focusing, 'four offsets, {-5 mm, -2.5 mm, 2.5 mm, 5 mm}...' are the units mm or um? '...whereas that of the 10 beads was...' here, do the authors mean the position of the beads derived from the movement of the piezo stage, as opposed to the remote focusing? The authors refer to the 'results from Chapter 3.2'. What are they talking about? Do they mean a supplementary figure, or earlier supplementary results? In general, we found the discussion in this paragraph difficult to follow. Supplementary Fig. 9 seems to be not referred to anywhere in the text. - Since the paper emphasizes 3D SIM, OTFs along the axial direction would also be useful to show, in addition to the lateral OTFs shown in Fig. 2D. - When the sample is moved by the piezo, the axial phase of the 3D-SIM illumination pattern is stable as the sample is scanned through the illumination pattern. When remote focusing is performed, the sample is always stable so the axial phase of the 3D-SIM illumination pattern is presumably changing with remote focusing. Can the authors clarify if the 3D SIM illumination pattern is scanned when remote focusing is applied, or is the intensity pattern stable in z? - In Supplementary Fig. 9, primary spherical is referred to twice, both at index 11 and 22. The latter is presumably secondary spherical? - we do not understand the x axis label, in Fig. S4D, is it really [0, 50, 50, 50] as written? see attached pdf for reference

Referee Cross-Commenting

I don't have much to add; the other reviewers raise good points and I think it would be good if the authors could respond to their feedback in a revised manuscript.

Significance

Nearly all fluorescence images deteriorate as a function of depth. Methods to ameliorate this depth-dependent degradation are thus of great practical value, as they improve the information content of images and thus (hopefully) biological insight. In this work, the authors develop a method to improve super-resolution imaging (3D SIM) at depth, by combining it with adaptive optics.

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

Learn more at Review Commons

Reply to the reviewers

Reviewer #1:

This study provides negative in vivo evidence for the use of two PERK inhibitors and of TUDCA for the treatment of Sli1-related Marinesco-Sjögren syndrome (MSS).

Overall, the manuscript reports a substantial amount of work and the study could be published in its present format. The experiments are well described in terms of methodology and appropriate analysis has been applied. Claims are proportionate and not overstated

I would have only minor comments related to some clarifications that the authors could make in the present manuscript and a suggestion for experiments that could improve the manuscript.

First, although this is not my expertise, the in vitro analysis of CHOP luciferase assays suggests that very high concentrations, in particular of TUDCA, are needed to observe an effect. The authors may wish to clarify their opinion and whether this could be the reason why in vivo they have been unable to obtain any inhibition of the PERK pathway.

The reviewer is correct in pointing out that high concentrations of trazodone, DBM and TUDCA were required to inhibit the PERK pathway in the CHOP::luciferase reporter cell lines. However, as we state in the Discussion, we do not think that their lack of effect in vivo was due to insufficient drug levels, since woozy mice were treated with trazodone, DBM or TUDCA according to dose regimens and administration routes that have proved effective in other neurodegenerative disease mouse models. Moreover, our analysis did not find major differences in drug bioavailability between mice with the woozy genetic background (CXB5/ByJ) and C57BL/6J mice in which these drugs had shown neuroprotective effects (see also the response to the next point).

Second, it seems to me that when measuring the Trazodone metabolism there is a difference between acute and chronic treatment. It would be worth discussing what the authors make of that and what is more relevant (I assume chronic) to the disease model outcome.

We realized that the nomenclature used in Figures 6 and 7 was confusing, leading the reader to think there were differences in trazodone levels between chronically and acutely treated mice.

The experiment shown in Figure 6 was designed to test whether there were differences in trazodone pharmacokinetics and metabolism between mice of the woozy strain, which have the CXB5/ByJ genetic background, and C57BL/6J mice in which trazodone had shown neuroprotective effects in previous studies. In contrast, Figure 7 illustrates the levels of trazodone and m-CPP in control and woozy mice (both of which have the CXB5/ByJ genetic background) that had been chronically treated with trazodone for 5 weeks. These are the same animals as in Figure 3, as we state in Figure 7 legend. Therefore one should compare the levels of trazodone and m-CPP in Figure 7 with those of the "woozy" group (CXB5/ByJ genetic background) in Figure 6. This comparison shows that trazodone and m-CPP levels are comparable after chronic and acute (6h) treatment.

To avoid confusion, we have changed the mouse nomenclature. We have renamed the control group of mice as "CT" (previously "WT") throughout the text and figures. In Figure 6, we have used CXB5/ByJ instead of "woozy" to emphasize the comparison between the different genetic backgrounds (CXB5/ByJ vs C57BL/6J). Finally, we have replaced the colors of symbols in Figure 7 in order to match those of Figure 3. We have also made the description and discussion of these results clearer in the revised manuscript.

With respect to the experiments a simple and informative addition would be the evaluation of the PERK pathway in mice treated with TUDCA, as this is missing.

The effect of TUDCA treatment on the PERK pathway is shown in Figure 5, where we measured CHOP mRNA levels in Purkinje cells microdissected from mice treated with 0.4% TUDCA in the chow, and in Figure 9C and D, where we measured the percentage of CHOP-immunopositive Purkinje cells in the cerebellum of same groups of mice by immunohistochemistry.

Figure 10 illustrates the results of an additional experiment in which woozy mice were treated with 500 mg/kg TUDCA intraperitoneally (ip), to test whether this alternative dosing regimen was any better. Like the treatment per os, TUDCA ip had no beneficial effect on motor dysfunction. Therefore we deemed it unnecessary to check the effect on PERK pathway inhibition in this group of mice.

A more difficult but potentially more interesting line of investigation is that of searching for potential actions of Trazodone that are PERK independent and might be responsible for the partial rescue observed in the beam walking test, which is much more sensitive and specific than rotarod, so worth considering. Assuming authors want to go down this route and add significance to their study my suggestion would be an unbiased RNA seq from the brain samples they already have. However, this is a suggestion to steer the study towards a more positive outcome and it is not necessary to support their current conclusions.

We agree with the reviewer that it would be interesting to investigate the mechanism by which trazodone slightly ameliorated the motor performance of woozy mice in the beam walking test. In the Discussion, we speculated that this could be due to an effect of trazodone on cerebellar serotonergic neurotransmission, which would require electrophysiological investigations to demonstrate. Of course, other mechanisms may also be operative, which RNA seq may help identify, as the reviewer suggests. However, this would be a complex and lengthy investigation, the results of which would not change the main conclusions of the present paper. We plan to explore this line of investigation in a future study.

Reviewer #2:

Lavigna et al. described the effect of Trazodone in Marinesco-Sjögren syndrome model mice. Although the results are somewhat disappointing, this research has provided fundamental evidence for the future development of MSS therapeutics. There are few minor comments to further improve the manuscript

Major comment<br /> P14<br /> "Trazodone metabolism to m-CPP was slightly impaired in woozy mice compared to C57BL/6J mice. This was evident from the m-CPP/trazodone ratio, calculated on the AUC0-t in the plasma, which was 0.34 in woozy and 0.67 in C57BL/6J mice."

Why was the concentration different between WT and woozy mice? Which organ mainly contributes to the metabolism of trazodone? Is the function of this target organ different between WT and woozy mice?<br /> Similar to trazodone, m-CPP clearance from plasma was slightly faster in woozy than in C57BL/6J mice.<br /> Is m-CPP eliminated via the kidney? Or liver? Why is there a difference? Does SIL1 functions in liver or kidney? Needs discussion. This is the same for brain m-CPP levels.

As explained in the response to the second comment of reviewer #1, "woozy" in Figure 6 referred to mice with the CXB5/ByJ genetic background, and in this experiment we compared trazodone pharmacokinetics and metabolism between CXB5/ByJ and C57BL/6J mice. We have modified the nomenclature of Figure 6 and the Results to make this clear.

Trazodone undergoes extensive hepatic metabolism, and only a small percentage is excreted unchanged in the urine. Metabolism involves hydroxylation, oxidation and dealkylation reactions, forming in particular the 5HT-active metabolite m-CPP (by CYP3A4). This and other metabolites are mainly excreted in urine, as conjugates [1-3]. The slight differences in trazodone pharmacokinetics and metabolism between the CXB5/ByJ and C57BL6/J mice shown in Figure 6 is not attributable to loss of SIL1 function, since both groups of mice carried wild-type Sil1 alleles, but is most likely due to subtle differences between the two strains, for example in the binding to plasma proteins, metabolic enzymes, transporters and/or the excretion processes. The available data do not allow to clarify this issue.

The main point, however, is that no major differences were found in the plasma and brain concentrations of trazodone between these two strains of mice, which could have explained the lack of efficacy of trazodone in woozy mice, as we now further stress in the revised Discussion.

Minor comments

P3 L5 mutation should be variant.

This has been changed.

P4 L1 eIF2a-P should be phosphorylated eIF2α (p-eIF2α). The reviewer prefers (p-eIF2α) than (eIF2α-p) throughout the manuscript.

There is no standard rule for indicating phosphorylated proteins, and phosphorylated eIF2α is referred to in various ways in different papers, with the "p" in capital or lowercase, preceding or following the protein name, separated by a dash or not. We would prefer to maintain the current nomenclature for consistency with our previous publications, unless the Editor deems otherwise.

P9 L11 M-CPP should be fully spelled out the first time it appears. m-Chlorophenylpiperazine (m-CPP)

M-CPP is spelled out the first time it appears in the Material and Methods, subheading Drug treatments and bioanalysis.

Please explain the difference between the expected function of trazodone and its metabolite m-CPP. Why m-CPP is not effective.

Based on the observation that mice of the woozy strain had lower brain levels of m-CPP than C57BL6/J mice (Figure 6), we hypothesized that the lack of effect of trazodone in woozy mice could be due to m-CPP mediating the PERK signaling inhibitory activity of trazodone. However, experiments in CHOP::luciferase reporter cells demonstrated that m-CPP does not inhibit PERK signaling (Figure 2D). The precise mechanism by which trazodone inhibits PERK signaling is not known [4], which makes it difficult to speculate why its main metabolite, m-CPP, does not exhibit this activity.

P11 L3 Fig. 3 Fig. 3A and B?

Yes, we specifically refer to panels A and B of Figure 3. We have indicated this in the revised manuscript.

P11 L6 at 7 weeks of age?

We have re-done the statistical analysis by two-way ANOVA and reported the results in the legend to Figure 3. There is a significant difference between vehicle- and trazodone-treated woozy mice in the number of missteps when the two groups are compared globally. No statistically significant difference in the number of missteps is detected at specific time points by post-hoc analysis. There is no statistically significant difference between vehicle- and trazodone-treated woozy mice in the time to traverse the beam. The Results section has been revised accordingly.

P12 L17 ~4 times, 4 times? Please state the exact value.

Done.

Figure 7 Why are brain m-CPP levels higher than plasma levels? Is trazodone metabolized in brain tissue?

Trazodone is extensively metabolized in the liver through Cytochrome P450 (Rotzinger et al., 1999). It is well documented that m-CPP readily passes the blood-brain barrier, much better than the parent compound, explaining its high brain levels [2].

P19 L7 ISRIB; please fully spell out the first time it appears.

Done.

References

1. Rotzinger S, Bourin M, Akimoto Y, Coutts RT, Baker GB (1999) Metabolism of some “second”- and “fourth”-generation antidepressants: iprindole, viloxazine, bupropion, mianserin, maprotiline, trazodone, nefazodone, and venlafaxine. Cell Mol Neurobiol 19:427– 442. https://doi.org/10.1023/a:1006953923305
2. Caccia S, Ballabio M, Samanin R, Zanini MG, Garattini S (1981) (--)-m-Chlorophenyl- piperazine, a central 5-hydroxytryptamine agonist, is a metabolite of trazodone. J Pharm Pharmacol 33:477–478. https://doi.org/10.1111/j.2042-7158.1981.tb13841.x
3. DeVane CL, Boulton DW, Miller LF, Miller RL (1999) Pharmacokinetics of trazodone and its major metabolite m-chlorophenylpiperazine in plasma and brain of rats. Int J Neuropsychopharm 2:17–23. https://doi.org/10.1017/S1461145799001303
4. Halliday M, Radford H, Zents KAM, Molloy C, Moreno JA, Verity NC, Smith E, Ortori CA, Barrett DA, Bushell M, Mallucci GR (2017) Repurposed drugs targeting eIF2alpha-P-mediated translational repression prevent neurodegeneration in mice. Brain 140:1768– 1783. https://doi.org/10.1093/brain/awx074

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

Lavigna et al. described the effect of Trazodone in Marinesco-Sjögren syndrome model mice. Although the results are somewhat disappointing, this research has provided fundamental evidence for the future development of MSS therapeutics. There are few minor comments to further improve the manuscrip

Major comment

P14 "Trazodone metabolism to m-CPP was slightly impaired in woozy mice compared to C57BL/6J mice. This was evident from the m-CPP/trazodone ratio, calculated on the AUC0-t in the plasma, which was 0.34 in woozy and 0.67 in C57BL/6J mice."

Why was the concentration different between WT and woozy mice? Which organ mainly contributes to the metabolism of trazodone? Is the function of this target organ different between WT and woozy mice? Similar to trazodone, m-CPP clearance from plasma was slightly faster in woozy than in C57BL/6J mice. Is m-CPP eliminated via the kidney? Or liver? Why is there a difference? Does SIL1 functions in liver or kidney? Needs discussion. This is the same for brain m-CPP levels.

Minor comments

P3 L5 mutation should be variant. P4 L1 eIF2a-P should be phosphorylated eIF2α (p- eIF2α). The reviewer prefers (p- eIF2α) than (eIF2α-p) throughout the manuscript.

P9 L11 M-CPP should be fully spelled out the first time it appears. m-Chlorophenylpiperazine (m-CPP) Please explain the difference between the expected function of trazodone and its metabolite m-CPP. Why m-CPP is not effective.

P11 L3 Fig. 3 Fig. 3A and B? P11 L6 at 7 weeks of age? P12 L17 ~4 times, 4 times? Please state the exact value.

Figure 7 Why are brain m-CPP levels higher than plasma levels? Is trazodone metabolized in brain tissue?

P19 L7 ISRIB; please fully spell out the first time it appears.

Referees cross-commenting

Since the viewpoints of Reviewer 1 and Reviewer 2 are different, it would be a good report if both comments are satisfied.

Significance

What are the strongest and most important aspects?

The author tested three potential drug candidates for MSS in MSS model mice in vivo. In addition, the author performed a PK study.

What aspects of the study should be improved or could be developed?

The description of the PK study is not sufficient to explain why the PK is different between the WT and the woozy mice is different.

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

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

This study provides negative in vivo evidence for the use of two PERK inhibitors and of TUDCA for the treatment of Sli1-related Marinesco-Sjögren syndrome (MSS).

Overall, the manuscript reports a substantial amount of work and the study could be published in its present format. The experiments are well described in terms of methodology and appropriate analysis has been applied. Claims are proportionate and not overstated

I would have only minor comments related to some clarifications that the authors could make in the present manuscript and a suggestion for experiments that could improve the manuscript.

First, although this is not my expertise, the in vitro analysis of CHOP luciferase assays suggests that very high concentrations, in particular of TUDCA, are needed to observe an effect. The authors may wish to clarify their opinion and whether this could be the reason why in vivo they have been unable to obtain any inhibition of the PERK pathway. Second, it seems to me that when measuring the Trazodone metabolism there is a difference between acute and chronic treatment. It would be worth discussing what the authors make of that and what is more relevant (I assume chronic) to the disease model outcome.

With respect to the experiments a simple and informative addition would be the evaluation of the PERK pathway in mice treated with TUDCA, as this is missing.

A more difficult but potentially more interesting line of investigation is that of searching for potential actions of Trazodone that are PERK independent and might be responsible for the partial rescue observed in the beam walking test, which is much more sensitive and specific than rotarod, so worth considering. Assuming authors want to go down this route and add significance to their study my suggestion would be an unbiased RNA seq from the brain samples they already have. However, this is a suggestion to steer the study towards a more positive outcome and it is not necessary to support their current conclusions.

Significance

My expertise is in the characterization of preclinical models of neurodegenerative diseases. This study is significant in the field because it reveals the complications arising when searching for non toxic PERK inhibitors for MSS. There is no current treatment for MSS and this study can help directing future studies towards more promising alternatives. Of course, providing only negative results is a limitation and the study would greatly increase its overall impact and significance if the effect of Trazodone would be further investigated.

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

Learn more at Review Commons

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity):

This manuscript compiles LoF variants of M1AP and ZZS proteins (i.e., SHOC1, TEX11 and SPO16) that almost certainly underlie infertility and reports the first case of an infertile man homozygous for a variant in SPO16. The authors validated interactions between human M1AP and ZZS that were found in mice. Analyzing testicular samples from infertile men revealed that those with deficiencies in SHOC1, TEX11 or SPO16 exhibited early meiotic arrest without haploid germ cells, whereas those with M1AP variants displayed a predominant metaphase I arrest with rare haploid germ cells. Further investigations showed that disrupted SHOC1, TEX11 or SPO16 led to defective synapsis and pairing of homologous chromosomes and unpaired DNA DSBs, while M1AP mutations reduced CO events. Importantly, men with LoF variants in M1AP can father healthy children by medically assisted reproduction. Overall, the results are clear and convincing in defining likely causative variants in infertility patients.

Response: We thank reviewer #1 for the appreciation of our work. We already addressed the suggestions raised by reviewer# 1 to improve our manuscript.

I have a few minor comments for improving the manuscript:<br /> • No statistical analyses were performed. The meaning of error bars was not mentioned. It is essential to specify the minimum number of seminiferous tubules counted for each patient.

Response: We added the statistical analysis. We described now in more clarity that all round tubules in a patient's testicular section were counted (l. 646-653).

• Allele frequencies of variants are not provided.

Response: We added the allele frequencies from gnomAD v4.1.0 (SNVs) and gnomAD SVs v2.1 (CNVs) in Table 1.

• Figure 4 should clearly label the representations of each color channel.

Response: Thanks for this suggestion. We labelled each color channel accordingly.

• The authors should clearly label the bands of SPO16 in the right panel of Figure 1B.

Response: We labelled the SPO16 band in Figure 1B more clearly.

• Appendix Figure S1B and S2B, what does "rat" mean in "rat Ins2 Ex3/4/"?

Response: In the minigene assay, an artificial gene was constructed with exon 3 and 4 from the insulin 2 gene of the species rat (Rattus norvegicus). We described this in more detail in the Appendix methods section (l. 119) and in the Figure legend S1B and S2B.

Reviewer #1 (Significance):

Overall, this study significantly contributes to the understanding of some genetic causes of human infertility and offers a potential avenue to treat patients with M1AP variants/ mutants. Since no knock-in animal model was applied to mimic the subtle phenotype variations observed in patients, the functionality of truncated proteins remains unexplored. For example, it is unclear why the germ cells in patient M3260 with the SHOC1 variant can progress to round spermatids (Fig. 2C), while those in Shoc1 KO mice (10.1093/molehr/gaac015) and other patients cannot. However, this is a minor concern.

Response: Thanks for this comment. SHOC1 variant c.1939+2T>C present in M3260 is a predicted splice site variant. In vitro it results in an in-frame exon skipping as shown by the minigene assay (Appendix Figure S2) that is predicted to lead to a loss of only 4% of the protein. We assume that this does not result in a complete loss but only in an impaired protein function enabling significantly reduced progression of spermatogenesis up to the round spermatid stage in few cells (l. 354-360). We addressed this in more detail in the results section (l. 145ff and l. 189ff) and in the Appendix Figure S2 legend. Accordingly, SHOC1 variant c.1939+2T>C is not a LoF variant and we excluded it from the quantification of subsequent analyses. Immunohistological staining of this patients was excluded from Appendix Figure S6, S7, S9, S10, and S11 and incorporated into Appendix Figure S2.

In addition, the recurrent M1AP c.676dup was functionally analysed in our previous work (Wyrwoll et al., 2020, PMID: 32673564). We detected M1AP mRNA in a testicular biopsy from one patient showing that this variant leads not to degradation of the mRNA. Furthermore heterologous expression of the mutant M1AP cDNA in HEK293T cells led to the production of a truncated protein that presumably leads to loss of protein function. We added this information in l. 136. Furthermore, our preliminary experiments of co-immunoprecipitation of truncated M1AP with TEX11 hint to an abolished protein-protein interaction caused by M1AP c.676dup and thus a loss of protein function.

Our field of expertise is gametogenesis and meiosis in mice.

Reviewer #2 (Evidence, reproducibility and clarity):

Summary:<br /> This interesting manuscript provides evidence for the biological and clinical relevance in human males of mutations in genes encoding M1AP and other related proteins. In mice, M1AP, "meiosis 1 associated protein," is known to associate with several proteins (SHOC1, TEX11, and SPO16) in the ZZS complex that promotes DNA recombination and crossover formation during meiosis I prophase. Mutation of these proteins in model organisms disrupts the process of recombination and cause arrest of spermatocytes prior to the first meiotic division. Here the authors took advantage of their MERGE (Male Reproductive Genomics) cohort to screen for human loss-of-function (LoF) mutations in the relevant ZZS complex and M1AP genes and to associate these with human male reproductive phenotypes. They found that men with deficiency of ZZS proteins SHOC1, TEX11 or SPO16 genes were infertile, exhibiting arrest of germ cell development early in meiotic prophase, with aberrations of chromosome synapsis and failure to repair DNA double-strand breaks (DSBs). In interesting contrast, men with M1AP mutations exhibited metaphase arrest, and indeed, in some cases, produced haploid spermatids, which in medically assisted reproduction (ICSI), led to the birth of offspring. Because they demonstrate that M1AP interacts with the other proteins, the authors conclude that M1AP is a "catalyzer," but not essential, for the processes of synapsis, recombination, and formation of haploid gametes.

Major Comments:<br /> The work is clearly presented with detailed methods that should allow adaptation in other laboratories.<br /> Overall, this study is a tour de force with what was no doubt difficult archival samples. The histology is generally of good quality, supporting the conclusions about progress of meiotic prophase in the mutant samples. The images of H&E-stained tissue are particularly striking, especially those in supplemental figures.

Response: We thank Reviewer #2 for the appreciation of our work and the suggestions to improve our manuscript. To provide transparency of our work, we plan to upload each (immuno-) histologically stained testicular section shown in the Main and Appendix Figures in the microscopy image repository OMERO/Open Microscopy Environment (OME).

That said, and with particular reference to Fig. 3A, it is difficult to sub-stage meiotic prophase by immunocytochemistry, even in optimal samples, with only one marker (in this case gH2AX). The staging here is also at odds with the statement in the subsequent section (and Fig. 4B) on absence of pachytene cells in men with mutation of SHOC1, TEX11, or SPO16.

Because precise stages of arrest probably cannot be determined in these samples, the authors would be wiser to use phrases such as "zygotene-like"

Response: We agree with the reviewer that it is indeed difficult to sub-stage meiotic prophase based on IHC for one marker. A precise sub-staging of the meiotic prophase would require identifying the stage of the seminiferous epithelium. The cycle of the human seminiferous epithelium has been subdivided into 12 stages based on the acrosomal development made visible by immunohistochemistry for acrosin. However, in order to properly evaluate the human germ cell associations, only seminiferous tubules showing a well-preserved seminiferous epithelium with no apparent damage to the epithelium and the peritubular wall can be considered. In addition, all the different generations of germ cells have to be present as well as at least six spermatids (Muciaccia et al., 2013, PMID: 23946533). As these requirements cannot be fulfilled in the testicular tissue of men with a meiotic arrest as due to LoF variants in M1AP or the ZZS genes, we followed the reviewer's suggestion and have modified the respective phrases throughout the text, e.g. to 'zygotene-like'.

The authors should also clarify how it was confirmed that the metaphase-like cells were spermatocytes and not spermatogonia (given that gH2AX signal is weak or unclear in some such nuclei). Readers with a focus on the more regularly staged mouse or rat tubules would appreciate a few more guidelines to criteria for staging human tubules.

Response: We thank the reviewer for raising this point. In order to confirm that the metaphase-like cells were indeed spermatocytes we will perform additional IHC staining for γH2AX and MAGEA4 on sequential testis sections (distance 3 µm) on representative samples of the patient cohort as well as controls as the hosts of both antibodies are mice. For a few more guidelines on the criteria for staging human tubules, please refer to the response to the previous point.

Evidence for the birth of a (healthy) child from one individual with M1AP mutation verges on the anecdotal (N=1). It is interesting but raises multiple questions and concerns about both the frequency of chromosomal abnormalities in such individuals and the transmission of the mutant alleles.

Response: We understand very well, that the evidence based on N=1 seems to be sparse. Nevertheless, if it is in principle possible for a man affected by bi-allelic M1AP LoF variants to conceive a child by ICSI then it could be also possible for other couples with a similar genetic condition (M1AP LoF), and thus providing a proof-of-principle (l. 417f). Reviewer #2 is completely right with the concerns regarding chromosomal aberrations and the transmission of the mutant allele. Thus, it is essential for clinicians/geneticists to counsel the affected couple carefully about the small but existed chance to have a biological own child and the accompanied potential but so far unexplored risks as outlined in l. 435ff. Our future research project will address this open and highly relevant question.

The authors conclude that the M1AP protein is an essential "catalyzer" in the meiotic recombination pathway. However, it is not clear from the data presented that M1AP in fact has enzymatic catalysis activity or exactly when and how it participates. Because the word "catalyzer" is not buttressed with hard or convincing evidence, the authors should consider other ways to describe the proposed role of M1AP, perhaps as a "putative component" and/or "modifier" of the recombination pathway.

Response: We appreciate the reviewer's advice, and changed the wording to "functional enhancer".

Minor comments:<br /> Fig. 1A - these are nice illustrations, but overly simplified with respect to timing (synapsis is not completed in zygonema)

Response: We completely agree that Figure 1A is a simplified depiction that could not reflect the temporally and spatially highly complex processes of meiosis. By adding a second dotted box and describing the process in the Figure legend in more detail, we tried to reduce the simplification. Nonetheless, we believe that this simplified schematic help readers, who are less familiar with the progression of meiosis to contextualise the described processess.

Fig. 1B - greater clarity in legend would be appreciated

Response: We described Figure 1B in more detail.

Figs. 2A & 3A - colors in bar graphs are difficult to discriminate

Response: We improved the discrimination of bar graphs accordingly.

Fig. 4A - with full appreciation for the difficulty with this material, the images are of low contrast and require considerable enlargement

Response: We agree with this opinion; and we increased the contrast. In addition, we will improve the way of representation in a revised Figure 4 in the complete revision of the manuscript in accordance with the suggestions of all three Reviewers.

Reviewer #2 (Significance):

This is a very interesting paper, which I evaluated from the perspective of a reproductive geneticist with expertise in meiosis and interest in infertility. I think this report will be of interest to clinicians because it identifies a gene possibly linked to marginal fertility and establishes human protein interactions similar to those previously identified in mice. It reinforces the importance of ZZS genes in humans. The contributions of this report to the field of meiosis confirm previous evidence on M1AP, including mutant phenotypes and protein interactions, extending them to humans. We can thus appreciate the conserved function of the mammalian M1AP protein, but as yet the molecular mechanisms of M1AP are not clarified.

Response: We gratefully thank Reviewer #2 for the thorough evaluation of our work and appreciate the recognition of the significance. Indeed, it was not possible to clarify the molecular mechanisms of M1AP that, hopefully, could be identified as soon as human specific antibodies, which will function in the needed applications, will be available. Additionally, we will perform further experiments as suggested by Reviewer #3 to gain a better understanding of the processess involved. Clarifying the underlying molecular mechanism is not only one of our highest interest but will also be important for the scientific community.

Reviewer #3 (Evidence, reproducibility and clarity):

In this manuscript, Rotte et al. investigate the meiotic molecular function in human of the M1AP protein and of the ZZS complex (SCHOC1, TEX11 and SPO16 proteins). The ZSS complex is a key player of meiotic recombination. It is a sub-complex of the conserved family of the ZMM proteins, essential for the formation of class I crossovers, a proper chromosomes segregation and fertility. Understanding its mode of action, regulation and conservation in human is thus a crucial issue in the fields of meiosis and human reproduction, with potential implications for patients. In that context, the recent identification of the protein M1AP as a partner of the ZSS proteins raise the question of its role, function and conservation. The aim of this study is thus of primary importance.<br /> To perform this molecular characterization, the authors made a cohort (24 total) of men carrying LoF variants in M1AP and ZSS genes. They performed a molecular biology analysis to assess the physical interaction between the human M1AP protein and the three components of the ZSS complex. Their results confirm a previous work performed in mice, mentioned by the authors.<br /> Then, they took advantage of available biopsies from different mutant men to perform a histological and cytological analysis of the impact of the different mutations on meiosis. The main conclusions are that in human, similarly to what is known in different organisms (ranging from yeast to mice), the ZSS complex is essential for crossover formation, synapsis and spermatogenesis, and that defect in the genes is associated with a premature prophase I arrest and no sperm formation. The authors also showed that M1AP protein plays a role in meiotic progression, but to a lesser extend compare to the ZSS proteins, with a metaphase I arrest, an undetectable recombination phenotype, apart of a reduced crossover number and, spermatozoa can form in its absence.

Major points:<br /> The authors investigate the physical interaction between M1AP and the ZSS members through a single approach: Co-IP of tagged proteins after expression in human HEK293T cells. This approach is informative, but to reinforce the conclusions the authors should provide data from independent approaches: yeast two hybrid, expression of recombinant proteins followed by pull down, co-immunostaining (TEX11 antibodies were used in the study and M1AP antibody is present in the literature) are possible non-exclusive approaches to decipher, more in details, the interaction. Moreover, understanding the hierarchy of interactions appears important to understand its rational, regulation and function. What is the meaning of a M1AP interaction with all the members of the complex? Remains an open question.

Response: We thank Reviewer #3 for this comment. In an independent approach we aimed to specify the interaction of M1AP to the ZZS proteins. Thus, we already cloned truncated versions of M1AP to refine the binding site of M1AP to the ZZS proteins (Figure R1). In a preliminary experiment, we co-transfected full-length as well as truncated forms of M1AP with TEX11 and showed via Co-IP that the interaction is only possible with full-length M1AP. Within the full-revision, we plan to finalise these experiments and thus validate the specifity of the interaction between M1AP and TEX11 and thereby gain more insight into the interaction/hierarchy of the interaction of M1AP with the ZZS complex.

Figure R1 Tolerance landscape of M1AP NM_001321739.2 illustrating the respective regions selected for mutagenesis of truncated M1AP constructs. Adapted from MetaDome.

Moreover, in the last couple of years, we spent enormous resources (personnel, time, financial) to get a functional antibody against human M1AP, including testing of different commercial (and already published) antibodies, creating three customised antibodies against different M1AP polypeptides, a nanobody raised against the complete M1AP protein (failed because of the impossibility to purify the protein), and contacting the authors of previously published customised M1AP antibodies (Arango et al., 2013/PMID 23269666 and Li et al. 2023/PMID 36440627). Figure R2 recapitulates some of our attempts. Moreover, we published the initial attempts of establishing an M1AP antibody in Wyrwoll et al., 2020/PMID 32673564. Unfortunately, no human M1AP-specific antibody is available.

Additionally, we tested different TEX11, SHOC1 and SPO16 antibodies in immunohistochemistry and SHOC1 and SPO16 antibodies in immunofluorescence of spermatocyte spreads, which did not result in a specific staining (Figure R3). Due to the lack of a human specific antibody against M1AP as well as antibodies against SHOC1 and SPO16, we are not able to localise these proteins in patient testicular sections to address this highly interesting research question that remains of great interest within our work on M1AP.

Figure R2. Attempts to locate M1AP in the human testis. Previous attempts to identify a commercially available antibody that reliably detects M1AP in the human testis have not been successful (Wyrwoll et al., 2020/ PMID 32673564). Accordingly, we tried to produce a human-specific antibody in cooperation with companies specified in antibody customisation (Eurogentec, Biotem). The last attempt, conducted with Biotem, is exemplarily shown in this figure. A. Human M1AP protein sequence (NP_620159.2) highlighting the antibody epitopes (orange) that were selected so that in men carrying the M1AP LoF variant c.676dup p.Trp226Leufs*4 in a homozygous state, the respective antibody should not be able to bind due to the protein truncation. For rabbit immunisation, both epitopes were pooled. B. HEK293T cells were transfected with DYK-tagged M1AP plasmids, either expressing the wildtype (WT) or the truncated protein (W226L). Sera of day (D) 28 and 42 of the immunised rabbit as well as the purified antibody product, a commercially available anti-M1AP antibody (HPA), and anti-DYK control antibody specificity was confirmed by Western blotting. C. Customised anti-M1AP antibody validation in human testicular control and D. M1AP-deficient tissue did not yield in a reliable staining. Various protocol optimisations were tested (different antigen retrieval, adapted blocking and antibody dilution solution, various primary and secondary antibody concentrations). Date shown represents the best result, respectively. The application of both sera and the purified antibody for spermatocyte spreading was tested in parallel and has not been successful either (data not shown). SC: Sertoli cells, SPC: spermatocytes, M-I: metaphase I cells, RS: round spermatids, ES: elongated spermatids. The scale bar represents 100 µm and 10 µm.

Figure R3. Efforts to identify human-specific antibodies for ZZS localisation. A. Commercially available antibodies for ZZS were tested via Western blotting, aiming to reliably detect SHOC1, SPO16, and TEX11 in human testicular biopsies. HA-tagged wildtype plasmid DNA (WT) was transfected in HEK293T cells and the anti-HA antibody was used as a positive control. Only one antibody detected TEX11 reliable in the purified lysates (anti-TEX11: HPA002950). B.-D. Immunohistochemical staining was performed with all antibodies on human testicular and is representatively shown for anti-SHOC1: #BS155344-R, anti-SPO16: #BS15024-R, and anti-TEX11: HPA002950. Only the anti-TEX11 (#HPA002950) was found to be specific. However, presumably due to the fixation with Bouin's solution, staining could not reliably be repeated in all samples and was not implied in this study. Various protocol optimisations were tested (different antigen retrieval, adapted blocking and antibody dilution solution, various primary and secondary antibody concentrations). Date shown represents the best result, respectively. The application of all antibodies for spermatocyte spreading was tested in parallel and have not been successful (data not shown), except for anti-TEX11 (#HPA002950, Appendix Figure S13). SC: Sertoli cells, SPC: spermatocytes, RS: round spermatids, ES: elongated spermatids. The scale bar represents 100 µm and 10 µm.

The ZZS mutants have a defect in gH2AX pattern, a defect in synapsis and no MLH1 foci, associated to apoptosis and prophase I arrest. M1AP mutation has a minor impact. The characterization of the effect of the different mutations (in particular M1AP) on the recombination process should be addressed further, by cytological means. For example, effect on strand invasion and ssDNA production should be monitored using RPA, DMC1 and RAD51 antibodies. The impact on alternative resolution pathway (e.g. BLOOM dependent) should be tested as well as the effect on other ZMM proteins, in particular MSH4-5, should be investigated. These experiments are essential to characterize, at the molecular level, the function of the different proteins during recombination.

Response: We thank the reviewer for this suggestion and highly appreciate to investigate the different pathways in more depth. We plan to perform additional immunofluorescence staining of spermatocyte spreads of identified patients compared to the control in the planned revision for a better understanding of M1AP within human recombination. We already ordered the antibodies against meiotic marker proteins as suggested by the Reviewer.

We would like to take the opportunity to refer to the extremely limited access to cryopreserved testicular material of the patients presented in this manuscript: for each gene (M1AP, SHOC1, TEX11, SPO16) we were lucky to get one testicular biopsy specimen from one man for only one preparation of spermatocyte spreads. We hope for the Reviewer's understanding that we cannot address each requested staining albeit this would be of highest interest. However, we are very confident that we will provide additional staining added to the yet shown to improve the understanding of M1AP's function on human male meiotic recombination.

In the same line, TEX11 staining in M1AP mutant should be more documented and in particular the different stages shown, as well as the foci counting, to have a quantitative result, that can be compared to MLH1. Moreover, co-immunostaining of different markers with TEX11: RPA, DMC1, MSH and MLH1 are also important to understand how the pathway is perturbed and the recruitment delayed/affected.

Response: In the planned revision, we will include the TEX11 foci counting using the acquired images that will be compared to MLH1 foci quantification. In addition, we plan additional co-immunostaining of TEX11 with different markers dependent on the availability of testicular material. Due to the limited resources of cryopreserved material, we cannot repeat the TEX11 staining in the patients with M1AP LoF variant for documentation of different stages. Slides that have already been stained are unfortunately bleached and cannot be re-analysed.

The published M1AP antibody should be tested to investigate its perturbation in the absence of the ZZS proteins and the hierarchy of event.

Response: As already outlined above, we tried to get any functional M1AP antibody for several years, which was not possible (Figure R2). Thus, we unfortunately cannot address this comment via this approach albeit this research question remains of great interest within our work on M1AP.

OPTIONAL: the obligatory crossover was measured, a comment or calculation of interference would be very interesting, and it seems doable using the MLH1 counting, to test whether thses mutants have an effect on this process.

Response: We thank Reviewer #3 for the suggestion of this interesting question that was not within our focus so far. Due to the limited material and the small number of cells from which we could digitally separate the chromosomes, we believe that the sample size is insufficient to obtain a statistically significant result.

Minor comments<br /> As written, the title is misleading, the paper does not investigate the impact of M1AP in ZSS recombination. Such study implies to study genetic interactions or the genetic dependency between the different proteins, which is not the case here.

Response: Thanks for this comment. We changed the title to "Genotype-specific differences in infertile men due to loss-of-function variants in M1AP or ZZS genes".

Labelled on histological images is not clear. The authors should clearly explain to what marker each staining correspond.

Response: We changed the labelling accordingly.

L67 to 72: the authors should update and use more accurate citations for meiotic recombination.

Response: Thanks for this suggestion. In this section, we have described the fundamental processes of meiosis, which have been repeatedly reviewed by renowned scientists. We have therefore chosen four well-cited expert reviews from different groups as references (PMID: 29385397, 24050176, 27648641, 35613017).

L76: the ZMM are specifically involved in the resolution of class I crossover. Please rephrase.

Response: We rephrased the sentence and changed it throughout the manuscript.

L94: Strictly, the author identified an interaction, they didn't establish how the interaction takes place.

Response: We rephrased the sentence.

FigS13: TEX11 staining should be presented with foci counting as a main figure.

Response: We plan to restructure Figure 4 along with the new meiosis specific markers and will consider this comment.

L255: MLH1 does not quantify homologous recombination but, class I crossovers.

Response: We rephrased the sentence.

L352: The sentence is hard to understand, rephrase please.

Response: We rephrased the sentence.

Reviewer #3 (Significance):

In general, the paper is well written and easy to follow. However, in light of the importance of the questions for the field of meiosis, it currently seems a little superficial, in particular if the authors aim at addressing the molecular function of the different proteins. The role of the ZSS proteins and M1AP in the control of meiotic recombination, at the molecular level is very important to decipher and additional experiments might help to better address this question. In addition, the functional links between M1AP and ZSS remains unclear and to investigate further.<br /> This study gives information for human process, and can be compared to more advanced work done with mice.<br /> This study will be important for the community working on meiosis in mammals, but also for people interested in reproduction.

Response: We thank Reviewer #3 for the thorough evaluation and acknowledgment of the significance of our work. We appreciate the suggestion of performing additional experiments to gain a better and more in depth understanding of the molecular pathways involved. We hope for the Reviewer's understanding that we cannot address all raised comments due to the limited material and the difficulty to get human specific antibodies in a research field that primarily works with highly valuable mouse models.

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

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

In this manuscript, Rotte et al. investigate the meiotic molecular function in human of the M1AP protein and of the ZZS complex (SCHOC1, TEX11 and SPO16 proteins). The ZSS complex is a key player of meiotic recombination. It is a sub-complex of the conserved family of the ZMM proteins, essential for the formation of class I crossovers, a proper chromosomes segregation and fertility. Understanding its mode of action, regulation and conservation in human is thus a crucial issue in the fields of meiosis and human reproduction, with potential implications for patients. In that context, the recent identification of the protein M1AP as a partner of the ZSS proteins raise the question of its role, function and conservation. The aim of this study is thus of primary importance.

To perform this molecular characterization, the authors made a cohort (24 total) of men carrying LoF variants in M1AP and ZSS genes. They performed a molecular biology analysis to assess the physical interaction between the human M1AP protein and the three components of the ZSS complex. Their results confirm a previous work performed in mice, mentioned by the authors.

Then, they took advantage of available biopsies from different mutant men to perform a histological and cytological analysis of the impact of the different mutations on meiosis. The main conclusions are that in human, similarly to what is known in different organisms (ranging from yeast to mice), the ZSS complex is essential for crossover formation, synapsis and spermatogenesis, and that defect in the genes is associated with a premature prophase I arrest and no sperm formation. The authors also showed that M1AP protein plays a role in meiotic progression, but to a lesser extend compare to the ZSS proteins, with a metaphase I arrest, an undetectable recombination phenotype, apart of a reduced crossover number and, spermatozoa can form in its absence.

Major points:

The authors investigate the physical interaction between M1AP and the ZSS members through a single approach: Co-IP of tagged proteins after expression in human HEK293T cells. This approach is informative, but to reinforce the conclusions the authors should provide data from independent approaches: yeast two hybrid, expression of recombinant proteins followed by pull down, co-immunostaining (TEX11 antibodies were used in the study and M1AP antibody is present in the literature) are possible non-exclusive approaches to decipher, more in details, the interaction. Moreover, understanding the hierarchy of interactions appears important to understand its rational, regulation and function. What is the meaning of a M1AP interaction with all the members of the complex? Remains an open question.

The ZZS mutants have a defect in H2AX pattern, a defect in synapsis and no MLH1 foci, associated to apoptosis and prophase I arrest. M1AP mutation has a minor impact. The characterization of the effect of the different mutations (in particular M1AP) on the recombination process should be addressed further, by cytological means. For example, effect on strand invasion and ssDNA production should be monitored using RPA, DMC1 and RAD51 antibodies. The impact on alternative resolution pathway (e.g. BLOOM dependent) should be tested as well as the effect on other ZMM proteins, in particular MSH4-5, should be investigated. These experiments are essential to characterize, at the molecular level, the function of the different proteins during recombination.

In the same line, TEX11 staining in M1AP mutant should be more documented and in particular the different stages shown, as well as the foci counting, to have a quantitative result, that can be compared to MLH1. Moreover, co-immunostaining of different markers with TEX11: RPA, DMC1, MSH and MLH1 are also important to understand how the pathway is perturbed and the recruitment delayed/affected.

The published M1AP antibody should be tested to investigate its perturbation in the absence of the ZZS proteins and the hierarchy of event.

OPTIONAL: the obligatory crossover was measured, a comment or calculation of interference would be very interesting, and it seems doable using the MLH1 counting, to test whether thses mutants have an effect on this process.

Minor comments

As written, the title is misleading, the paper does not investigate the impact of M1AP in ZSS recombination. Such study implies to study genetic interactions or the genetic dependency between the different proteins, which is not the case here.

Labelled on histological images is not clear. The authors should clearly explain to what marker each staining correspond.<br /> L67 to 72: the authors should update and use more accurate citations for meiotic recombination.

L76: the ZMM are specifically involved in the resolution of class I crossover. Please rephrase.

L94: Strictly, the author identified an interaction, they didn't establish how the interaction takes place.

FigS13: TEX11 staining should be presented with foci counting as a main figure.

L255: MLH1 does not quantify homologous recombination but, class I crossovers.

L352: The sentence is hard to understand, rephrase please.

Significance

In general, the paper is well written and easy to follow. However, in light of the importance of the questions for the field of meiosis, it currently seems a little superficial, in particular if the authors aim at addressing the molecular function of the different proteins. The role of the ZSS proteins and M1AP in the control of meiotic recombination, at the molecular level is very important to decipher and additional experiments might help to better address this question. In addition, the functional links between M1AP and ZSS remains unclear and to investigate further.

This study gives information for human process, and can be compared to more advanced work done with mice.<br /> This study will be important for the community working on meiosis in mammals, but also for people interested in reproduction.

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

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

Summary:

This interesting manuscript provides evidence for the biological and clinical relevance in human males of mutations in genes encoding M1AP and other related proteins. In mice, M1AP, "meiosis 1 associated protein," is known to associate with several proteins (SHOC1, TEX11, and SPO16) in the ZZS complex that promotes DNA recombination and crossover formation during meiosis I prophase. Mutation of these proteins in model organisms disrupts the process of recombination and cause arrest of spermatocytes prior to the first meiotic division. Here the authors took advantage of their MERGE (Male Reproductive Genomics) cohort to screen for human loss-of-function (LoF) mutations in the relevant ZZS complex and M1AP genes and to associate these with human male reproductive phenotypes. They found that men with deficiency of ZZS proteins SHOC1, TEX11 or SPO16 genes were infertile, exhibiting arrest of germ cell development early in meiotic prophase, with aberrations of chromosome synapsis and failure to repair DNA double-strand breaks (DSBs). In interesting contrast, men with M1AP mutations exhibited metaphase arrest, and indeed, in some cases, produced haploid spermatids, which in medically assisted reproduction (ICSI), led to the birth of offspring. Because they demonstrate that M1AP interacts with the other proteins, the authors conclude that M1AP is a "catalyzer," but not essential, for the processes of synapsis, recombination, and formation of haploid gametes.

Major Comments:

The work is clearly presented with detailed methods that should allow adaptation in other laboratories.<br /> Overall, this study is a tour de force with what was no doubt difficult archival samples. The histology is generally of good quality, supporting the conclusions about progress of meiotic prophase in the mutant samples. The images of H&E-stained tissue are particularly striking, especially those in supplemental figures. That said, and with particular reference to Fig. 3A, it is difficult to sub-stage meiotic prophase by immunocytochemistry, even in optimal samples, with only one marker (in this case gH2AX). The staging here is also at odds with the statement in the subsequent section (and Fig. 4B) on absence of pachytene cells in men with mutation of SHOC1, TEX11, or SPO16. Because precise stages of arrest probably cannot be determined in these samples, the authors would be wiser to use phrases such as "zygotene-like." The authors should also clarify how it was confirmed that the metaphase-like cells were spermatocytes and not spermatogonia (given that gH2AX signal is weak or unclear in some such nuclei). Readers with a focus on the more regularly staged mouse or rat tubules would appreciate a few more guidelines to criteria for staging human tubules.<br /> Evidence for the birth of a (healthy) child from one individual with M1AP mutation verges on the anecdotal (N=1). It is interesting but raises multiple questions and concerns about both the frequency of chromosomal abnormalities in such individuals and the transmission of the mutant alleles.<br /> The authors conclude that the M1AP protein is an essential "catalyzer" in the meiotic recombination pathway. However, it is not clear from the data presented that M1AP in fact has enzymatic catalysis activity or exactly when and how it participates. Because the word "catalyzer" is not buttressed with hard or convincing evidence, the authors should consider other ways to describe the proposed role of M1AP, perhaps as a "putative component" and/or "modifier" of the recombination pathway.

Minor comments:

Fig. 1A - these are nice illustrations, but overly simplified with respect to timing (synapsis is not completed in zygonema)

Fig. 1B - greater clarity in legend would be appreciated

Figs. 2A & 3A - colors in bar graphs are difficult to discriminate

Fig. 4A - with full appreciation for the difficulty with this material, the images are of low contrast and require considerable enlargement

Significance

This is a very interesting paper, which I evaluated from the perspective of a reproductive geneticist with expertise in meiosis and interest in infertility. I think this report will be of interest to clinicians because it identifies a gene possibly linked to marginal fertility and establishes human protein interactions similar to those previously identified in mice. It reinforces the importance of ZZS genes in humans. The contributions of this report to the field of meiosis confirm previous evidence on M1AP, including mutant phenotypes and protein interactions, extending them to humans. We can thus appreciate the conserved function of the mammalian M1AP protein, but as yet the molecular mechanisms of M1AP are not clarified.

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 compiles LoF variants of M1AP and ZZS proteins (i.e., SHOC1, TEX11 and SPO16) that almost certainly underlie infertility and reports the first case of an infertile man homozygous for a variant in SPO16. The authors validated interactions between human M1AP and ZZS that were found in mice. Analyzing testicular samples from infertile men revealed that those with deficiencies in SHOC1, TEX11 or SPO16 exhibited early meiotic arrest without haploid germ cells, whereas those with M1AP variants displayed a predominant metaphase I arrest with rare haploid germ cells. Further investigations showed that disrupted SHOC1, TEX11 or SPO16 led to defective synapsis and pairing of homologous chromosomes and unpaired DNA DSBs, while M1AP mutations reduced CO events. Importantly, men with LoF variants in M1AP can father healthy children by medically assisted reproduction. Overall, the results are clear and convincing in defining likely causative variants in infertility patients.

I have a few minor comments for improving the manuscript:

• No statistical analyses were performed. The meaning of error bars was not mentioned. It is essential to specify the minimum number of seminiferous tubules counted for each patient.
• Allele frequencies of variants are not provided.
• Figure 4 should clearly label the representations of each color channel.
• The authors should clearly label the bands of SPO16 in the right panel of Figure 1B.
• Appendix Figure S1B and S2B, what does "rat" mean in "rat Ins2 Ex3/4/"?

Significance

Overall, this study significantly contributes to the understanding of some genetic causes of human infertility and offers a potential avenue to treat patients with M1AP variants/ mutants. Since no knock-in animal model was applied to mimic the subtle phenotype variations observed in patients, the functionality of truncated proteins remains unexplored. For example, it is unclear why the germ cells in patient M3260 with the SHOC1 variant can progress to round spermatids (Fig. 2C), while those in Shoc1 KO mice (10.1093/molehr/gaac015) and other patients cannot. However, this is a minor concern.

Our field of expertise is gametogenesis and meiosis in mice.

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

Author responses

__Reviewer #1 (Evidence, reproducibility and clarity (Required)): __

In their manuscript, Dutta and colleagues compared the meiotic recombination landscapes between five budding yeast species. In the first part of the work, the authors constructed a high-resolution map of meiotic recombination events in Kluyveromyces lactis supported by high-quality genome assemblies for two strains of this yeast. Then, partially repeating their CO and NCO mapping strategy, they compared a number of meiotic recombination parameters between the five species (sometimes three, depending on the quality of the data for each species). They particularly focused on key parameters for meiotic recombination, such as crossover interference and homeostasis and obligate crossover. Although the analysis is interesting, it is underdeveloped in many places and lacks the general conclusions regarding the evolution of recombination and the broader perspective that would be expected from a comparison of these phenomena in budding yeasts.

[R] Tackling the evolution of recombination is ambitious. Here, with a dataset of five species, it is hard to draw strong evolutionary conclusions besides the variations in the crossover (CO) landscapes and the control of CO formation that we observed, which is already significant. The multiple losses of CO interference that we describe here may constitute our strongest evolutionary conclusion. It potentially underscores the minor evolutionary advantage associated to CO interference at least in budding yeasts. In this context, we changed the title to be more factual and updated the text to better highlight the significance and implications of our findings.

Major comments:

The authors indicate that the distribution of hotspots and coldspots is not preserved between species, but this finding is not properly documented. I think it would be useful to include recombination maps in a main figure for all species (or at least for S. cerevisiae, K. lactis and L. waltii) with the elements highlighted. This will allow for a visual illustration of the variability in the recombination landscape between the studied species. [R] The genomes of the species show blocks of synteny but overall, they are not collinear and therefore, it is not possible to have a direct comparison of the recombination maps. In our previous work, we have highlighted the non-conservation of CO hotspots between S. cerevisiae, L. kluyveri and L. waltii (Brion et al. 2017; Dutreux et al. 2023). Briefly, we retrieved conserved syntenic blocks in L. kluyveri and L. waltii genomes containing at least two S. cerevisiae orthologs associated with one hotspot. L. waltii shares only five out of the 92 S. cerevisiae crossover hotspots (RHO5, SLS1, GYP6, OLE1 and MRPL8), while L. kluyveri shares only one. L. waltii and L. kluyveri share no crossover hotspots. In addition, our current study shows that none of the K. lactis hotspot is conserved in any of the four other species (response figure 1 and new supplementary figure S11).

Response Figure 1. Density of crossovers along the genome using a 5 kb window in the S. cerevisiae genome (Mancera et al. 2008; Oke et al. 2014; Krishnaprasad et al. 2015 combined dataset). Horizontal dotted green line represents crossover hotspot significance threshold. Solid spheres represent the conserved CO hotspots with either L. kluyveri (red) or L. waltii (blue). None of the 92 S. cerevisiae crossover hotspot is conserved in L. lactis.

Although analyses analogous to those presented in Fig. S5 had already been published in other comparisons of the recombination landscape in yeast (e.g. Dutreux et al., 2023), I think that Figs. S5A and S5B are worth to be presented in the main figures (not supplementary data). In many species of eukaryotes, the detection of NCOs is practically impossible, therefore only results for COs are presented. Therefore, it is perhaps also worth discussing the fact that the relationship applies to all recombination events and not only COs, and therefore is related to the regulation of DSBs frequency and not individual DSBs repair pathways.

[R] Figures S5A-B are now included in the main figure, Figure 2B. The association holds true for all total recombination (CO+NCO) events as well, new supplementary figure S6A.

The authors find that CO coldspots were associated with DNA repair genes. Unfortunately, an equivalent analysis was not performed for all recombination events (CO + NCO). I presume this approach is based on the belief that COs are more mutagenic than NCOs. However, recent studies in humans suggest that, at least in mammals, meiotic DSBs themselves are mutagenic, regardless of the pathway used for their repair (Hinch et al., Science 2023). Therefore, I would suggest repeating the analysis also considering NCOs (although I am aware that the picture of NCOs may be incomplete). I would also like to see some graphical representation of the analysis. Is it possible to perform a classic analysis of coldspots/hotspot enrichment in relation to gene ontology?

[R] As suggested, we performed the analysis to independently detect coldspots for all recombination events (CO+NCO). Based on a threshold of

In relation to the previous point - it may be worth repeating this type of analysis also for other yeasts used in this study, or at least for S. cerevisiae, to be able to consider the extent to which this relationship is universal and dependent on the meiotic DSB repair pathway.

[R] The analysis regarding the CO coldspots has been performed in the other species as well. As mentioned in the main text, although some overlap between CO coldspots and DNA repair genes has been observed in the other species as well, we observed a significant enrichment in K. lactis only, maybe because the dataset is larger than in the other species.

In Fig. S7, the point where WGD occurred is marked in the wrong place, or at least that is what the sentence in the text says ("The Lachancea and Kluyveromyces species branched from the Saccharomyces lineage more than 100 million years ago, before to the ancestral whole-genome duplication (WGD) event specific of the S. cerevisiae lineage").

[R] We regret the oversight and have corrected the figure.

The result presented in Fig. S8 is interesting and should be shown in the main figures. Perhaps it would be worth adding an illustration illustrating simple versus complex COs.

[R] The old Figure S8 is now a part of main Figure 2C-D with the illustrations describing the CO types.

The last part of the results includes an analysis of the evolutionary rates of the ZMM genes. In the discussion, the authors should also refer the results of this analysis to the previous analysis of the overrepresentation of DNA repair genes in recombination coldspots. I understand that ZMM are not DNA repair proteins in the strict sense, but I think it is worth familiarizing readers with the authors' view on this matter. Moreover, I would suggest showing where MLH1 and MLH3 are located on the plot in Fig. 6 (especially the meiosis-specific MLH3), whether the selection pressure acts on them as on ZMM proteins, or rather as on DNA repair proteins. Showing the SLX4 and MUS81 would also be interesting.

[R] Figure 6 has been updated as suggested and now shows the Mlh1, Mlh3, Slx4 and Mus81 dN/dS values for the three species.

I feel like the discussion is underdeveloped. I missed a deeper summary of the comparison between meiotic recombination among the tested budding yeasts in the context of the presence and absence of functional ZMM. Even the title of the work is not properly developed in the manuscript text. The analysis shows that it is not the presence of a functional ZMM pathway or its lack that introduces differences between the individual recombination landscapes, although ZMM determines the presence of proper CO interference. With the caveat that for L. kluyveri it is basically unknown whether it has a functional ZMM or not. Maybe confirming the lack of expression of some ZMM genes in meiosis of this species would answer the question of how it should be treated?

[R] We agree with this reviewer that our original title was imprecise, so we changed it to be more factual, emphasizing on the multiple losses of crossover interference in budding yeasts. As stated above, it potentially underscores the minor/negligible evolutionary advantage associated to CO interference at least in budding yeasts. From there, it is hard to draw deeper conclusions since the actual roles/functions of CO interference are still under debate, notably in yeasts where the CO frequency tends to be high. We improved the discussion to better highlight these points.

We also agree that a deeper characterization of the ZMM factors persisting in the non-Saccharomyces yeasts would be informative, but we believe it is beyond the scope of the current manuscript and more suitable for a follow up work. However, our recent publication about L. kluyveri (Legrand et al 2024) shows that Zip3 is properly expressed in meiosis and behaves as in S. cerevisiaesince it is located at DSB sites. Furthermore, we have unpublished transcriptomic data (Response Figure 2) showing that all the ZMM genes from L. kluyveri are specifically induced in meiosis (fold increase >16 at least compared to pre-sporulation conditions). Therefore, so far, although the level of CO interference in L. kluyveri is minimal, there is no indication that the ZMM genes are mis regulated.

Response Figure 2. Transcriptomic data showing that all the ZMM genes from L. kluyveri are specifically induced in meiosis (Unpublished data from Llorente Lab, CRCM, Marseille).

Minor comments:

In general, Figure captions are imprecise, many of them lack clear information explaining what is depicted. Authors should remember that figure legends should be self-sufficient. [R] The figure legends have been updated and are now self-sufficient.

In the revised manuscript, I would suggest placing figure numbers on the figures and using line numbering, which would facilitate the reception of the work and possible reference to its individual elements in the review.

[R] We regret the omission. Figure numbers, Line numbers and Page numbers have been added.

Reviewer #1 (Significance (Required)):

The study provides a new insight into the variation in recombination landscape within budding yeast species with a special emphasis on crossover control. This includes also de novo assemblies of Kluyveromyces lactis genome and high-resolution tetrad-based maps of meiotic recombination events. Previously, recombination maps of different yeast species were compared, however this study focuses on budding yeasts, some of which lost ZMM pathway and differ in some crossover parameters, like interference and homeostasis. Although the analysis is interesting, it lacks the general conclusions regarding the evolution of recombination and the broader perspective that would be expected from a comparison of these phenomena in budding yeasts.

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

This paper describes the genome-wide mapping of meiotic recombination in non-Saccharomyces yeast, Kluyveromyces lactis. By using heterologous parental strains, the authors mapped crossovers (COs) and noncrossovers (NCOs) on the genome of K. lactis which lacks proteins necessary for CO formation such as S. cerevisiae, mammals and plants. This is an extension of previous works by the authors' group which mapped CO and NCO in different yeast, Lachancea kluyveri and L. waltii by a similar approach. The authors found that CO frequencies in K. lactis are much lower than those in S. cerevisiae and COs showed weaker interference, which facilitates the non-random distribution of COs along a chromosome. Overall, the experiments and informatic analyses have been done in good quality and the results are convincing. The paper provides additional new information on the landscape of meiotic recombination in different yeast species. These results are of great interest to researchers in the field of meiotic recombination and evolution of meiosis. There are some issues that the authors may be able to address before the publication.

Major points: While the authors noted that K. lactic shows the loss of a pro-CO factors (ZMM protein), Spo16, and Msh5 (due to the introduction of an in-frame stop codon), it still possesses other proteins such as Zip1, Zip2, Zip3, Zip4/Spo22, Mer3, and Msh4. It is still likely that these pro-CO factors control CO formation (and interference) in this yeast. It would be nice for the authors to study whether the knockout of these genes is dispensable for CO formation and interference in meiosis. A similar analysis should be done for L. kluyveri which retains all ZMM genes, but this is clearly out of the scope of this paper.

[R] The question of the functions of the remaining ZMM factors is indeed interesting and related to point #8 from reviewer 1 (please see above). Although this is beyond the scope of our work, we would like to refer here to work from Amy McQueen's lab using L. lactis Zip1 in S. cerevisiae (Voelkel-Meiman 2015). This study shows that L. lactis Zip1 does not allow synaptonemal complex assembly in S. cerevisiae but allows CO formation independently of the Msh4/5 complex but that depend on Zip2/4/Spo16 and Mlh1/3 for their resolution. Overall, these results suggests that L. lactis Zip1 at least retained ancestral functions shared with S. cerevisiae Zip1. However, it is not possible to conclude if the lack of full complementation of L. lactis Zip1 in S. cerevisiae comes from functional divergence or simply by the inability of L. lactis Zip1 to function properly in a heterologous context.

Minor points:

No page number, no main Figure number. It is hard to review this paper. [R] We regret the oversight. Figure numbers, Line numbers and Page numbers have been added.

References: In some cases, in the Introduction, the authors referred to review papers such as Pyatnitskaya et al. (2019) for ZMM proteins while in the other parts, they referred to original papers; for example, three papers for Mlh1-Mlh3. If the number of references is not limited, original papers should be cited in the text.

[R] We regret this omission. Original papers have now been included in the citations.

Figure 3A, page 9, second paragraph: When the authors compared CO and NCO densities, it would be nice to show P-values for the comparison.

[R] p-values have now been added to the updated figure.

Please show a ratio of CO to NCO in each yeast in Figure 3B in the second paragraph of page 9 in the main text.

[R] The ratios have now been included in the figure for both the CO:NCO ratios and CO:corrected_NCO ratios, in the main text and figure legends.

Figure S5 and page 7, the first paragraph and page 9, third paragraph: CO/NCO densities (negative correlation to chromosome sizes) in S. cerevisiae should be checked with or without short chromosomes (I, III, and VI), which show very unique regulation of meiotic DSB formation (see Murakami et al. Nature 2020).

[R] Even excluding the small chromosomes, the size dependent trend persists for S. cerevisiae and S. paradoxus.

Table S7: Please add the S. cerevisiae gene name such as ZIP1 next to S. cerevisiae orthologs such as YDR285W. Moreover, please explain the column in detail or clarify the data. What does "meiosis" mean here? For example, YJL074C is SMC3, which is expressed in mitosis as well as in meiosis. The same is true for YGL163C, which is RAD54, which plays a minor role in meiosis, but plays a critical in mitotic DSB repair.

[R] We corrected Table S7 as desired by systematically including the standardized gene names.

The Gene Ontology (GO) annotation is a statement about the function of a particular gene. It offers a structured framework and a comprehensive set of concepts to describe the functions of gene products across all organisms. It is specifically crafted to support the computational representation of biological systems. In our specific case, we only looked at genes with the gene ontology annotation "meiosis". Together, these statements comprise a "snapshot" of current biological knowledge and is by no means absolute. This has been detailed in the supplementary Table S7.

Reviewer #2 (Significance (Required)):

This study provides the landscape of meiotic recombination in non-Saccharomyces yeast, Kluyveromyces lactis. The genome-wide recombination map in K. lactis shows lower crossover frequencies with weaker crossover interference than those in S. cerevisiae. Overall, the experiments and informatic analyses have been done in good quality and the results are convincing. The paper provides additional new information on the landscape of meiotic recombination in different yeast species, particularly in terms of the evolution of meiotic recombination. These results are of great interest to researchers in the field of meiotic recombination and evolution of meiosis.

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

Dutta et al. have compiled a genome-wide meiotic recombination map for Kluyveromyces lactis and compared it to a compilation of meiotic recombination maps for four other species, two of which (Lachancea kluyveri and Lachancea waltii), like K. lactis, predate the genome duplication event that produced the other two (Saccharomyces cerevisiae and S. paradoxus). Meiosis in many species studied (including metazoans and plants) shows control over the number and distribution of crossovers, which are critical for faithful chromosome segregation during meiosis. This takes the form of crossover interference, where crossovers are spaced more evenly than expected by chance, and crossover homeostasis, where many fewer chromosomes lack a crossover than is expected by chance. While both of the post-duplication species show both crossover interference and homeostasis, none of the pre-duplication species show crossover homeostasis, and crossover interference is very weak. In two cases (K. lactis and L. waltii), this can be explained by mutational loss of a few of the genes (called the ZMM genes) that promote meiotic crossovers in many species. However, L. kluyveribehavior cannot be explained in this way. Recombination hotspots are present but are not shared between the pre-duplication species or between the pre- and post-duplication species, perhaps not surprising for species that diverged more that 100 million years ago. Overall, this work will be a useful contribution to our understanding of the different possible flavors of meiotic recombination mechanisms and control that are possible (and, one might add, promote long-term species viability). A) Evaluation, reproducibility and clarity The work presented in this paper is straightforward and unimpeachable and will largely be of interest to those studying meiotic recombination, be it mechanistic studies or studies of the implications for population genetics. The analysis is technically correct, although there are some aspects where a slightly different emphasis should be considered (see comments below). However, the data and the analysis could stand as they currently are, without further revision.

Suggestions are below. 1. (trivial) it would have been useful if pages and lines were numbered.

[R] We regret the oversight. Figure numbers, Line numbers and Page numbers have been added.

"Across the 205 meioses...". In general, it would be desirable to apply compensation for the fact that NCOs and COs are differently detected. Since, in K. lactis, 35% of COs are not accompanied by detectable gene conversion, it seems reasonable to apply a correction to measured NCOs here and throughout the paper, regardless of the species. For example, if one assumes that 35% of NCOs are not detected, how does this affect estimates of chromosomes that do not appear to have undergone interhomolog recombination? Estimates of CO/NCO bias? In a similar vein, if the CO event is not considered (just the conversion events associated with it), how does this affect measures of conversion tract lengths in COs and NCOs?

[R] We thank the reviewer for this suggestion. We have performed the correction for the NCO estimates as described in Mancera et al. 2008, on a per tetrad basis across all the species. The fraction of missed NCOs were 7%, 34%, 30%, 23% and 25% respectively for S. paradoxus, S. cerevisiae, K. lactis, L. waltii and L. kluyveri. The fraction of missed NCOs depend upon the parental marker density. In addition, we performed the CO:NCO bias analysis both with the detected and the corrected NCO frequencies and the trends remain unchanged (Now included in figure 3). Finally, we refrain from using the corrected NCO frequencies while reporting the NCO frequencies (Table 1, main text) to maintain uniformity with our previous work and since, these corrections do not alter any results.

It might be useful to report recombination event frequencies in terms of events/chromosome, as this, rather than event/unit distance, is functionally more relevant. In the same vein, it might be useful to consider total event homeostasis, in addition to just crossover homeostasis.

[R] This has been updated as suggested. .

An interesting observation is that two of the three pre-duplication species clearly at one time had a full complement of ZMM genes but lost some due to mutation. Have there ever been attempts to detect either synaptonemal complex or axial elements in these species?

[R] This is related to point #8 from reviewer 1 and to the major point of reviewer 2 (please see above).

To our knowledge, cytological observations of synaptonemal complex (SC) or axial elements have been performed in L. kluyverionly by us and the SC is clearly visible (Legrand et al 2024).

However, it is key to remind here that K. lactis axis protein encoding genes HOP1 and RED1 have been cloned by the Roeder's lab by functional complementation of S. cerevisiae corresponding mutants, supporting the functional conservation of these genes (Smith and Roeder 2000). Finally, as mentioned above, K. lactis Zip1 retained at least some function of the ancestral Zip1 protein that are also shared by the S. cerevisiae protein (Voelkel-Meiman 2015).

The observation of elevated evolutionary rates in ZMM genes is also intriguing, but it would help if "dN/dS ratio" was defined.

[R] It is now defined in the text.

The observation of frequent E0 chromosomes is taken to suggest efficient achiasmate segregation; has the "corrected" NCO frequency been considered? Do the different frequencies of E0 chromosomes predict the different spore viabilities seen between species?

[R] E0 is not predictive at all of the spore viability as we have shown in previous studies (see L. kluyveri - Brion et al. 2017, L. waltii-Dutreux et al. 2023). In addition, this has been shown is S. cerevisiae as well (Nishant et al. 2009).

Figure 3A-what would this look like if it were plotted as "Events per chromosome" rather than per megabase?

[R] We changed the figure (now figure 2A) and plotted as events per chromosome to show the variability of events at the chromosome level.

Figure legends tend to be unreasonably terse, which makes figures more difficult to interpret.

[R] This has been updated as suggested.

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

Dutta et al. have compiled a genome-wide meiotic recombination map for Kluyveromyces lactis and compared it to a compilation of meiotic recombination maps for four other species, two of which (Lachancea kluveri and Lachancea waltii), like K. lactis, predate the genome duplication event that produced the other two (Saccharomyces cerevisiae and S. paradoxus). Meiosis in many species studied (including metazoans and plants) shows control over the number and distribution of crossovers, which are critical for faithful chromosome segregation during meiosis. This takes the form of crossover interference, where crossovers are spaced more evenly than expected by chance, and crossover homeostasis, where many fewer chromosomes lack a crossover than is expected by chance. While both of the post-duplication species show both crossover interference and homeostasis, none of the pre-duplication species show crossover homeostasis, and crossover interference is very weak. In two cases (K. lactis and L. waltii), this can be explained by mutational loss of a few of the genes (called the ZMM genes) that promote meiotic crossovers in many species. However, L. kluyveri's behavior cannot be explained in this way. Recombination hotspots are present but are not shared between the pre-duplication species or between the pre- and post-duplication species, perhaps not surprising for species that diverged more that 100 million years ago. Overall, this work will be a useful contribution to our understanding of the different possible flavors of meotic recombination mechanisms and control that are possible (and, one might add, promote long-term species viability).

A) Evaluation, reproducibility and clarity

The work presented in this paper is straightforward and unimpeachable, and will largely be of interest to those studying meiotic recombination, be it mechanistic studies or studies of the implications for population genetics. The analysis is technically correct, although there are some aspects where a slightly different emphasis should be considered (see comments below). However, the data and the analysis could stand as they currently are, without further revision. Suggestions are below.

1. (trivial) it would have been useful if pages and lines were numbered.
2. "Across the 205 meioses...". In general, it would be desirable to apply compensation for the fact that NCOs and COs are differently detected. Since, in K. lactis, 35% of COs are not accompanied by detectable gene conversion, it seems reasonable to apply a correction to measured NCOs here and throughout the paper, regardless of the species. For example, if one assumes that 35% of NCOs are not detected, how does this affect estimates of chromosomes that do not appear to have undergone interhomolog recombination? Estimates of CO/NCO bias? In a similar vein, if the CO event is not considered (just the conversion events associated with it), how does this affect measures of conversion tract lengths in COs and NCOs?
3. It might be useful to report recombination event frequencies in terms of events/chromosome, as this, rather than event/unit distance, is functionally more relevant. In the same vein, it might be useful to consider total event homeostasis, in addition to just crossover homeostasis.
4. An interesting observation is that two of the three pre-duplication species clearly at one time had a full complement of ZMM genes, but lost some due to mutation. Have there ever been attempts to detect either synaptonemal complex or axial elements in these species?
5. The observation of elevated evolutionary rates in ZMM genes is also intriguing, but it would help if "dN/dS ratio" was defined.
6. The observation of frequent E0 chromosomes is taken to suggest efficient achiasmate segregation; has the "corrected" NCO frequency been taken into account? Do the different frequencies of E0 chromosomes predict the different spore viabilities seen between species?
7. Figure 3A-what would this look like if it were plotted as "Events per chromosome" rather than per megabase?
8. Figure legends tend to be unreasonably terse, which makes figures more difficult to interpret.

Significance

This paper adds to our understanding of the spectrum of meiotic recombination behaviors that are possible, and thus is of interest primarily to those who study meiotic recombination. It expands significantly the number of species for which meiotic recombination has been analyzed, and in particular has the surprising finding that loss of crossover control by mutation of the existing crossover machinery is remarkably common, with four of the six yeast species (I include here Schizzosaccharomyces pombe) lacking crossover interference. It will be a substantial, solid contribution to the field.

My expertise: meiosis, recombination, yeast, chromatin, chromosomes

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

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

This paper describes the genome-wide mapping of meiotic recombination in non-Saccharomyces yeast, Kluyveromyces lactics. By using heterologous parental strains, the authors mapped crossovers (COs) and noncrossovers (NCOs) on the genome of K. lactics which lacks proteins necessary for CO formation such as S. cerevisiae, mammals and plants. This is an extension of previous works by the authors's group which mapped CO and NCO in different yeast, Lachancea kluyveri and L. waltii by a similar approach. The authors found that CO frequencies in K. lactics are much lower than those in S. cerevisiae and COs showed weaker interference, which facilitates the non-random distribution of COs along a chromosome. Overall, the experiments and informatic analyses have been done in good quality and the results are convincing. The paper provides additional new information on the landscape of meiotic recombination in different yeast species. These results are of great interest to researchers in the field of meiotic recombination and evolution of meiosis. There are some issues that the authors may be able to address before the publication.

Major points:

While the authors noted that K. lactics shows the loss of a pro-CO factors (ZMM protein), Spo16, and Msh5 (due to the introduction of an in-frame stop codon), it still possesses other proteins such as Zip1, Zip2, Zip3, Zip4/Spo22, Mer3, and Msh4. It is still likely that these pro-CO factors control CO formation (and interference) in this yeast. It would be nice for the authors to study whether the knockout of these genes is dispensable for CO formation and interference in meiosis. A similar analysis should be done for L. klyuveri which retains all ZMM genes, but this is clearly out of the scope of this paper.

Minor points:

1. No page number, no main Figure number. It is hard to review this paper.
2. References: In some cases in the Introduction, the authors referred to review papers such as Pyatnitskaya et al. (2019) for ZMM proteins while in the other parts, they referred to original papers; for example, three papers for Mlh1-Mlh3. If the number of references is not limited, original papers should be cited in the text.
3. Figure 3A, page 9, second paragraph: When the authors compared CO and NCO densities, it would be nice to show P-values for the comparison.
4. Please show a ratio of CO to NCO in each yeast in Figure 3B in the second paragraph of page 9 in the main text.
5. Figure S5 and page 7, the first paragraph and page 9, third paragraph: CO/NCO densities (negative correlation to chromosome sizes) in S. cerevisiae should be checked with or without short chromosomes (I, III, and VI), which show very unique regulation of meiotic DSB formation (see Murakami et al. Nature 2020).
6. Table S7: Please add the S. cerevisiae gene name such as ZIP1 next to S. cerevisiae orthologs such as YDR285W. Moreover, please explain the column in detail or clarify the data. What does "meiosis" mean here? For example, YJL074C is SMC3, which is expressed in mitosis as well as in meiosis. The same is true for YGL163C, which is RAD54, which plays a minor role in meiosis, but plays a critical in mitotic DSB repair.

Significance

This study provides the landscape of meiotic recombination in non-Saccharomyces yeast, Kluyveromyces lactics. The genome-wide recombination map in K. lactis shows lower crossover frequencies with weaker crossover interference than those in S. cerevisiae. Overall, the experiments and informatic analyses have been done in good quality and the results are convincing. The paper provides additional new information on the landscape of meiotic recombination in different yeast species, particularly in terms of the evolution of meiotic recombination. These results are of great interest to researchers in the field of meiotic recombination and evolution of meiosis.

I have been studying meiotic recombination. On the other hand, because of my limited experience, I can not evaluate bioinformatics parts in this paper.

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

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

In their manuscript, Dutta and colleagues compared the meiotic recombination landscapes between five budding yeast species. In the first part of the work, the authors constructed a high-resolution map of meiotic recombination events in Kluyveromyces lactis supported by high-quality genome assemblies for two strains of this yeast. Then, partially repeating their CO and NCO mapping strategy, they compared a number of meiotic recombination parameters between the five species (sometimes three, depending on the quality of the data for each species). They particularly focused on key parameters for meiotic recombination, such as crossover interference and homeostasis and obligate crossover. Although the analysis is interesting, it is underdeveloped in many places and lacks the general conclusions regarding the evolution of recombination and the broader perspective that would be expected from a comparison of these phenomena in budding yeasts.

Major comments:

1. The authors indicate that the distribution of hotspots and coldspots is not preserved between species, but this finding is not properly documented. I think it would be useful to include recombination maps in a main figure for all species (or at least for S. cerevisiae, K. lactis and L. waltii) with the elements highlighted. This will allow for a visual illustration of the variability in the recombination landscape between the studied species.
2. Although analyzes analogous to those presented in Fig. S5 had already been published in other comparisons of the recombination landscape in yeast (eg, Dutreux et al., 2023), I think that Figs. S5A and S5B are worth to be presented in the main figures (not supplementary data). In many species of eukaryotes, the detection of NCOs is practically impossible, therefore only results for COs are presented. Therefore, it is perhaps also worth discussing the fact that the relationship applies to all recombination events and not only COs, and therefore is related to the regulation of DSBs frequency and not individual DSBs repair pathways.
3. The authors find that CO coldspots were associated with DNA repair genes. Unfortunately, an equivalent analysis was not performed for all recombination events (CO + NCO). I presume this approach is based on the belief that COs are more mutagenic than NCOs. However, recent studies in humans suggest that, at least in mammals, meiotic DSBs themselves are mutagenic, regardless of the pathway used for their repair (Hinch et al., Science 2023). Therefore, I would suggest repeating the analysis also taking into account NCOs (although I am aware that the picture of NCOs may be incomplete). I would also like to see some graphical representation of the analysis. Is it possible to perform a classic analysis of coldspot/hotspot enrichment in relation to gene ontology?
4. In relation to the previous point - it may be worth repeating this type of analysis also for other yeasts used in this study, or at least for S. cerevisiae, to be able to consider the extent to which this relationship is universal and dependent on the meiotic DSB repair pathway.
5. In Fig. S7, the point where WGD occurred is marked in the wrong place, or at least that is what the sentence in the text says ("The Lachancea and Kluyveromyces species branched from the Saccharomyces lineage more than 100 million years ago, before to the ancestral whole-genome duplication (WGD) event specific of the S. cerevisiae lineage").
6. The result presented in Fig. S8 is interesting and should be shown in the main figures. Perhaps it would be worth adding an illustration illustrating simple versus complex COs.
7. The last part of the results includes an analysis of the evolutionary rates of the ZMM genes. In the discussion, the authors should also refer the results of this analysis to the previous analysis of the overrepresentation of DNA repair genes in recombination coldspots. I understand that ZMM are not DNA repair proteins in the strict sense, but I think it is worth familiarizing readers with the authors' view on this matter. Moreover, I would suggest showing where MLH1 and MLH3 are located on the plot in Fig. 6 (especially the meiosis-specific MLH3), whether the selection pressure acts on them as on ZMM proteins, or rather as on DNA repair proteins. Showing the SLX4 and MUS81 would also be interesting.
8. I feel like the discussion is underdeveloped. I missed a deeper summary of the comparison between meiotic recombination among the tested budding yeasts in the context of the presence and absence of functional ZMM. Even the title of the work is not properly developed in the manuscript text. The analysis shows that it is not the presence of a functional ZMM pathway or its lack that introduces differences between the individual recombination landscapes, although ZMM determines the presence of proper CO interference. With the caveat that for L. kluyveri it is basically unknown whether it has a functional ZMM or not. Maybe confirming the lack of expression of some ZMM genes in meiosis of this species would answer the question of how it should be treated?

Minor comments:

1. In general, Figure captions are imprecise, many of them lack clear information explaining what is depicted. Authors should remember that figure legends should be self-sufficient.
2. In the revised manuscript, I would suggest placing figure numbers on the figures and using line numbering, which would facilitate the reception of the work and possible reference to its individual elements in the review.

Significance

The study provides a new insight into the variation in recombination landscape within budding yeast species with a special emphasis on crossover control. This includes also de novo assemblies of Kluyveromyces lactis genome and high-resolution tetrad-based maps of meiotic recombination events. Previously, recombination maps of different yeast species were compared, however this study focuses on budding yeasts, some of which lost ZMM pathway and differ in some crossover parameters, like interference and homeostasis.

Although the analysis is interesting, it lacks the general conclusions regarding the evolution of recombination and the broader perspective that would be expected from a comparison of these phenomena in budding yeasts.

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

Learn more at Review Commons

Reply to the reviewers

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

Reply to the Reviewers

I thank the Referees for their...

Referee #1

1. The authors should provide more information when...

Responses + The typical domed appearance of a hydrocephalus-harboring skull is apparent as early as P4, as shown in a new side-by-side comparison of pups at that age (Fig. 1A). + Though this is not stated in the MS 2. Figure 6: Why has only...

Response: We expanded the comparison

Minor comments:

1. The text contains several...

Response: We added...

Referee #2

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

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

In this manuscript Ochner et al. report the 3.2 Å cryo-EM structure of the type IV pilus (minus PilY1 adhesin) from P. aeruginosa PAO1. The authors demonstrate that the conserved N-terminal helix of pilin subunits (PilA) form a tubular arrangement within the hydrophobic core of the pilus whereas the divergent C-terminal pilin globular domain decorates the periphery of the pilus. Comparisons are then made against T4P structures from other organisms, highlighting interesting differences including a shorter rod diameter and lack of solvent-accessible loops for which the authors propose reduces proteolysis of the T4P compared to other organisms.

Major comments:

The results of this manuscript are convincing. The models and cryo-EM volumes, which are already accessible from the PDB and EMDB, are of good quality with no obvious issues. The conclusions drawn from the model are not speculative. While extensive mutagenesis experiments could help delineate critical residues involved in T4P assembly and clarify involvement in adhesion/biofilm formation, these would have to be done in the native organism, would require a significant amount of time and effort, and would be beyond the scope of the current manuscript.

Minor comments:

The figures are excellent and clear, and the text is well-written, with results easy to interpret.

One of the strengths of this paper is the comparative analyses across current bacterial T4P structures. In this respect, I would have liked a more thorough analysis here: - While differences in helical parameters, rod diameter, and rod length are presented, a figure showing comparison of surface electrostatics and/or hydrophobicity could help delineate differences (if any) across these species, which may reflect the different environments these bacteria inhabit.<br /> - A consurf representation of PilA is shown in Fig. 3h. It would be helpful to include either the sequence alignment used for this analysis or a sequence alignment for all the species presented in the manuscript, to show precisely which residues are absolutely conserved across these species.<br /> - A panel showing their full T4P as a surface with Consurf coloring would be informative to show conservation across the entire pilus and not just a PilA subunit. - The authors state that the models in Fig. 3 were aligned based on the matchmaker function in Chimera. Wouldn't the poor sequence conservation of the C-terminal globular domain of PilA drive the alignment towards the N-terminal helix? In that case, wouldn't using a comparative alignment strategy that focuses on the model itself (LSQ) or secondary structure elements (SSM) which would drive the alignment more towards the globular domain be more reflective of the full pilin subunit? - Related to the point above, it would be useful to include a table highlighting pairwise RMSDs across all models presented in this manuscript.

Significance

The authors rightfully highlight the importance of P. aeruginosa T4P in the development of biofilms; structural analyses of these pili are of clinical importance and of interest to researchers involved in bacterial motility.

To date, various structures of T4P and T4P subunits across a variety of bacterial have been solved by X-ray crystallography and cryo-EM (PDB: 9EWX (this study), 6GV9, 5VXX, 6XXD, 6VK9, 8TJ2). It appears that another group has recently published a slightly lower resolution (3.6 Å vs 3.2 Å) cryo-EM structure (PDB: 8TUM) of the T4P of P. aeruginosa PAO1 (Thongchol et al., Science, 2024). The model from this latter publication appears to be identical to the model presented in this manuscript. Since this work has now been published (with models being released mid-March 2024), and since Ochner et al.'s manuscript only appeared on Biorxiv on April 9th 2024, I feel it would have been appropriate and necessary to cite this paper. And while the Thongchol publication reduces the novelty of Ochner et al.'s manuscript, there is some merit in the comparative analyses performed, which if expanded upon, could further strengthen this manuscript enough to stand on its own.

Field of expertise: cryo-EM, bacterial secretion, membrane 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 #2

Evidence, reproducibility and clarity

The manuscript by Ochner et al. reports the cryo-EM structure of the Type 4 pili of Pseudomonas aeruginosa at decent 3.2 resolution with fully resolved pilin fold. It is a straightforward report, using state-of-the-art microscopy and data processing approaches as usual for the group and the figures and data representation are clear. The main findings of the work is that it visualizes the assembled pilus of an important pathogen (Pseudomonas aeruginosa is one of the ESKAPE pathogens with particularly impressive adaptability and Type IV pili are important for substrate colonization and biofilm formation). The PilA pilin fold is not far from that of a previously crystallized isolated homolog from the PAK strain (core hydrophobic N-helix, globular b-sheet-containing exposed CTD) and presents a central melting of the core helix also observed among multiple other PilA homologs from solved G- pilin structures. The main difference for the PAO1 pilus is the tight packing in a significantly thinner filament, which lacks protruding loops or other secondary structure insertions in the core pilin fold. The authors propose that this could lead to increased stability such as to proteases.

Again, the study is quite straightforward and besides the standard and well-executed EM workflow, it uses the classical approaches for pilus overexpression and purification (a PilT mutant that cannot retract and presents more T4P; well established mechanical shearing protocol for surface release, etc.). The structure is at decent resolution allowing full backbone tracing and side-chain resolution for confident model building, etc. The figures are clear, even if I would encourage some more vivid or at least contrasting colors for the cartoon model in Fig. 2 and some more detailed surface and conservation analyses, especially in terms of packing and surface exposure.

Minor comment: The electron density maps, atomic models and validation reports should be available for the review process. The refinement statistics in the table are very good and the figures and supplementary movie present clear densities but this should be standard protocol and could help with constructive suggestions from the reviewers. Large map files, etc can be provided via a link if too big for upload directly through the manuscript tracking system.

Significance

My main concern with this work is that it is quite minimalistic in terms of biology/physiology, especially in light of the many G- pili structures available, some of which the authors nicely review in terms of specific structural parameters. The hypothesis of increased protease or perhaps mechanical resistance is tantalizing but how the compact pilus fold actually affects Pseudomonas aeruginosa in its physiology is unclear. Are there any other differences in the surface properties relative to other pili (charge, surface motif conservation, etc.) and could they have relevance in terms of interactions with the substrate, another matrix component or a peculiar niche within the host? As a PilA mutant should be easy to get from a number of laboratories, how would a Pseudomonas delta-PilA mutant behave in terms of twitching motility, surface attachment and biofilm formation if complemented with PaO1 PilA vs. other pilins from Table S2 or a pilin with an engineered hybrid architecture (e.g. a loop or b-hairpin insertion)? If such heterologous/engineered pili are incubated with a mild protease mix, would they indeed exhibit increased fragmentation relative to the wt PAO1 pili? To me, most of these assays are relatively easy to be attempted in terms of molecular biology, phenotypic assays and in vitro biochemistry (e.g. plasmid-based complementation if full genetics are judged beyond the scope/time available, twitching, pellicle formation, SDS-PAGE of protease-treated sheared pili) and could really shed more specific insights light into the peculiarities of Pseudomonas T4P function, rather than just present the next resolved filament among the multiple other T4P already out there.

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

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Summary

The study titled "Structure of the Pseudomonas aeruginosa PAO1 Type IV pilus" by Ochner and colleagues utilised cryo-electron microscopy (cryo-EM) to determine and describe the atomic structure of a complete type IV pilus (T4P) filament from Pseudomonas aeruginosa in its native state at an impressive resolution of 3.2 Å. The authors use state-of-the-art cryo-EM methodology, and the detailed description of their procedures allows for an adequate replication of the results. The T4P are essential for the virulence of P. aeruginosa, which is a clinically important human pathogen, as they play a crucial role in biofilm formation, a major factor in its resistance to antibiotics and ability to cause infections. Therefore, understanding the molecular mechanisms behind the ability of these bacteria to establish infections is vital, and this high-resolution structure of T4P provides valuable insight into this process.

Overall, this reviewer acknowledges the importance of the structure presented here and its significance to our understanding of P. aeruginosa infection and persistence, and hence is positive about the publication of the results, however, at its current state the manuscript raises the major questions outlined below, which must be addressed and corrected.

Major comments

The authors propose a model where T4P interact with the type IV secretion systems (T4SS) in the P. aeruginosa membrane. However, there is no current evidence in the literature to support a direct interaction between T4P and T4SS as these are functional and structural distinct secretion systems. T4P biogenesis is mediated by a specialised secretion complex (homologous to type II secretion systems), spanning both bacterial membranes and consisting of the outer membrane secretin subcomplex, the alignment subcomplex, and the inner membrane motor complex. This reviewer recommends that authors refer to the comprehensive review by Hospenthal and colleagues [PMID: 28496159] that details the T4P biogenesis and Craig and colleagues [PMID: 30988511] that provides an in-depth analysis of T4P secretin architecture. This reviewer recommends the authors to remove any misleading claims regarding a T4P-T4SS interaction. Furthermore, the introduction would benefit from a brief overview of the T4P biogenesis and secretin architecture to prevent any further confusion. While this study offers a higher resolution structure of P. aeruginosa T4P (3.2 Å) compared to the previously described work on the P. aeruginosa T4P (8 Å) described by Wang and colleagues [PMID: 28877506], the manuscript fails to convey the significance of this improvement. The authors should directly compare the new structure with the previously obtained cryo-EM structure, similarly to how they tackled the comparison to the X-ray crystallography structure (Figure S6). A dedicated figure visualising the key differences and benefits associated with the higher resolution is necessary to highlight the manuscript's significance. Furthermore, the authors should specify that the reported T4P belongs to the type IVa category and the "globular domain" of PilA should be further differentiated into the αβ loop and D region - widely accepted motifs present in the structures of type IV pilins [PMID: 31784891]. Highlighting them is crucial due to their roles in receptor binding, microcolony formation, and antigenic variation, warranting their inclusion in the manuscript. A more detailed display of intersubunit interactions, including the types and numbers of interactions is also recommended, however optional. Previous studies [PMID: 27698424, PMID: 28609682] hypothesize that disordered loops might be involved in significant T4P stretching, the authors should address how the lack of these structures in their model might affect the filament dynamics. Lastly, the study lacks experimental validation of the structure, either within the study or referenced from the existing literature and very weakly connects the structure to T4P's biological functions, such as twitching motility or DNA acquisition. For instance, a comparison could be drawn between the surface charge of the pili and its DNA binding capacity. Additionally, the T4P secretin complex of P. aeruginosa documented in [PMID: 27705815] should be modelled alongside the obtained T4P structure to compare the structure diameter with the PilQ secretin lumen. These revisions will strengthen the manuscript by addressing crucial points and highlighting the significance of the high-resolution T4P structure.

Minor comments

Figure 2 For consistency, the colours of PilA subunits between panels (a) and (d) should match.

Figure 3 For clarity, pilins should be coloured by domain.

L41 The word "surfaces" or "target receptors" rather than just "substrates" would be more accurate.

L87 Rather than "other bacteria" consider using "wild-type strains".

L145-147 For clarity, residues C134 and C147 that form a disulfide bond in the C-terminal loop should be displayed in the figure.

L371 For consistency, "h" should be in brackets, following the authors' style.

Significance

General assessment:

The significance of the study stems from a resolution improvement from the previously reported type IV pilus of P. aeruginosa by Wang and colleagues [PMID: 28877506] and complements well the X-ray crystallography data obtained previously by Craig and colleagues [PMID: 12769840]. Due to the role of T4P in the virulence of P. aeruginosa, the structure provides important biological information about the molecular mechanism of its niche establishment. Moreover, the structure can be used in subsequent drug design against P. aeruginosa infections.

Nature of advance:

The nature of the advance provided by this study is in the added structural detail of the T4P due to the obtained higher resolution of the map. Usually, the highest resolution structure is used to derive the conclusions about the biological functions of the filament, hence the structure provided here will be referenced as the final P. aeruginosa T4P structure in further studies.

Audience:

The higher resolution structure compared to the previously described will be interesting to the translational/clinical drug discovery audiences, which require a high-resolution structure for accurate drug design.

Field of expertise:

Type IV secretion systems, bacterial conjugation, conjugative pili.

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

Learn more at Review Commons

Reply to the reviewers

We thank the reviewers for their positive assessment of our manuscript. We agree that there are some further experiments suggested by the reviewers that would enhance our study. We have highlighted further proposed experimental work in bold for clarity.

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

1. EVIDENCE, REPRODUCIBILITY AND CLARITY Summary: The Matrix 2 (M2) protein of influenza A virus (IAV) is a single pass transmembrane protein known to act as a tetrameric ion channel that is important for both viral entry and egress. The paper by Figueras-Nova et al. entitled "Caspase cleavage of Influenza A virus M2 disrupts M2-LC3 interaction and regulates virion production" reports on the regulation of IAV virion production through a regulatory interplay between a caspase cleavage site and a LC3 interacting region (LIR) motif in M2. In its C-terminal cytoplasmic tail the IAV M2 protein contains a C-terminal LIR motif interacting with LC3. The authors show that this LIR motif is preceded by a functional caspase cleavage motif cleaved predominantly by caspase-6, with some contribution from caspase-3: The motif 82-SAVD-85 directs cleavage after the aspartate (D) at position 85. The cleavage leads to loss of the remaining C terminal sequence from amino acid 86 to 97. The core LIR motif 91-FVSI-94 LIR motif is then lost from M2 which can no longer bind LC3. As previously described by the same group using point mutations in the LIR motif (Ref 12.), loss of a functional LIR., here by caspase- mediated deletion of the LIR, affects the virion production and inhibits filamentous budding. LC3B lipidation is increased upon treatment with a caspase inhibitor. The authors show for the first time that LC3 is included into IAV virions via binding to M2. Furthermore, they also report a co-crystal structure of the M2 C terminus (aa 70-97), containing the caspase cleavage site and LIR, and LC3B (aa 3-125) adding new insights into this interaction and showing that the caspase cleavage site is in a flexible region N-terminal to the LIR. This work shows how caspase cleavage may modulate LC3B lipidation, trafficking to the plasma membrane, incorporation of LC3B in the virions, filamentous budding and virion production (viral titer).

Major comments: The findings reported here are very well supported by the data shown. This is a very clearly written paper with well described and nicely visualized results that are accompanied by adequate statistical analyses.

We thank the reviewer for their assessment of our manuscript.

The authors report a new way the LC3B binding to the C-terminal tail of the M2 proteins is regulated and suggest that this is an adaptation the virus has made to adjust virion production to host cell status by hijacking the function of host caspases. They show that the caspase cleavage motif is evolutionary conserved and use that as an argument. Perhaps it could be discussed if it also could be an argument that the host protects itself against a too massive virion production as this could be too detrimental to the host? Would it not also be an evolutionary advantage to the virus in the long run by avoiding killing the host?

This is an interesting point. We agree there could be advantage for the virus not to overproduce virions under certain circumstances. Consistent with this caspase-6 deficient mice had increased mortality in response to IAV PR8 infection, and presented and increase in viral spread in the lungs (Zheng, 2021; doi: 10.1016/j.cell.2020.03.040). This is also relevant for the comments made by Reviewer 2. The manuscript will be updated to include a discussion of this point.

A question I may raise which is optional as it may be too much work to address as part of this study is if the reported regulation of LC3B binding has any role in regulating the ion channel function of the M2 tetramer?

It is well established that there is no impact of distal C-terminal truncations on M2 ion channel activity (Cady et al., 2009, doi: 10.1021/bi9008837 Schnell and Chou, doi: 10.1038/nature06531; Nguyen et al., 2008, doi: 10.1021/bi801315m; Tobler et al., 1999, doi: 10.1128/jvi.73.12.9695-9701.1999). This is also consistent with data from our lab (Ulferts et al., 2021, doi: 10.1016/j.celrep.2021.109899, Beale et al., 2014, doi: 10.1016/j.chom.2014.01.006) as well as others (Ren et al., 2015, doi: 10.1128/JVI.00576-15) showing the effects of the LIR motif and the proton channel are distinct. We appreciate the reviewer suggesting further work here as optional, but there is already compelling evidence to show there is no substantial effect of the LIR motif on ion channel activity. (See also Reviewer 2 points 4 and 5).

Minor comments: Delete "with" in line 145.

This will be changed in the updated manuscript.

Line 217: It should be written more specifically how "cells were surface stained with M2"

The protocol for surface staining of M2 will be explained in more detail in the updated manuscript.

1. SIGNIFICANCE

This is a very well performed study with a sound experimental strategy and well performed assays with clear results increasing our insight into the interplay between the Influenza A virus and host cells. Although caspase mediated cleavage of the autophagy receptor and signaling scaffold protein p62 (Ref. 25), removing the LIR and LC3-binding, has been reported before I consider this study as novel in reporting this type of regulation of LC3 binding. The cleavage of p62 deletes a large part of the protein while here it is a "clean" deletion of the LIR sequence representing a conceptual advance of regulation of LC3 binding. The study also reports for the first time on LC3B incorporated into virions. The effects on trafficking to the plasma membrane and viral budding and virion production are similar to those reported before (Ref. 12) using viruses with point mutations crippling the LIR motif. This research will be of interested to all studying virus- host interaction and to the autophagy field both as a non autophagic role of LC3B, and as a regulatory mechanism of LIR-LC3B interactions involving the irreversible caspase cleavage-mediated deletion of the LIR motif.

We thank the reviewer for this assessment of our manuscript.

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

The influenza A virus (IAV) M2 protein is small transmembrane protein which plays a role in virus entry and egress. In a previous study, Beale et al. (2014) identified an LC3-interacting region (LIR) in the M2 cytoplasmic domain that was found to recruit the LC3B protein to the plasma membrane. Recombinant IAV harboring mutations in the LIR motif showed reduced particle stability and lost filamentous morphology.

In the present study, Figueras-Novoa et al. show that the LIR motif is removed in response to activation of cellular caspases. The authors demonstrate that in in IAV-infected THP-1 cells M2 is partially cleaved at the motif (82)SAVD(85)¯A by caspase 6. Caspase inhibitors abolished cleavage, and a mutant virus harboring the D85A substitution was found to be resistant to caspase action. A crystal structure of purified M2 C- terminus and LC3B revealed that the caspase cleavage site lies in a flexible region that is accessible to caspases.

Mutant virus encoding a truncated M2 protein (M2D86-97) was unable to interact with LC3, in accordance with the absence of the LIR motif. The M2D86-97 mutant showed reduced lipidation of LC3, while enhanced lipidation of LC3 was observed when wild-type virus-infected cells were treated with caspase inhibitors. The authors also observed that cell surface transport of M2D86-97 but not M2-D85A was impaired. However, in purified virus particles a mix of cleaved and uncleaved M2 was detected. The authors also demonstrated that lipidated LC3B was present in purified virions of wild-type virus particles but even more abundant in M2-D85A virions. Finally, M2D86-97 mutants produced significantly less infectious particles compared to wild-type virus while the D85A cleavage mutant replicated to similar titers than wt virus.

Based on these findings the authors concluded that caspases regulate the interaction of M2 protein with LC3 which impacts virion production. Specifically, they propose that caspase-mediated removal of the LIR motif may enable a switch between filamentous and non-filamentous budding in response to depletion of cellular resources. However, the authors were unable to rescue a filamentous IAV with a truncated M2 protein and therefore could not provide direct proof for their guess.

While the data are sound and presented well, they do not support the conclusions of the authors.

1. To the authors opinion, the conserved caspase cleavage site in the M2 protein might provide an evolutionary advantage for the virus. However, the M2-D85A mutation has no effect on viral replication, so the biological significance of why M2 needs to be cleaved at all is unclear. The conclusion that caspase-induced M2 cleavage is a fine-tuning mechanism of IAV has not been supported by experiments.

We thank the reviewer for the assessment of our data. We think the reviewer is specifically objecting to the phrase “We conclude that this highly conserved interaction and cleavage act as a regulatory mechanism exploited by IAV to fine-tune virion production in different cellular contexts.” This is a reasonable inference from our results, but we accept that it is not proven. We will change the wording to make it clear this has not been definitively demonstrated.

1. The finding that the permanently truncated IAV M2 mutant virus was substantially attenuated does not necessarily mean that abrogation of M2-LC3 interaction was responsible for this attenuation. As the M2 protein plays a role in virus budding at the plasma membrane (recruitment of M1 protein, induction of membrane curvature, membrane scission), the impaired transport of the truncated M2 protein might already explain that the virus was attenuated and that incorporation of the protein into the viral envelope was reduced.

We will confirm this further with additional experiments using LIR mutants. Recapitulating the plasma membrane transport defect of truncated M2 with LIR mutants including the newly characterised M2D87A and M2D88A mutants and a more severe mutant with a FVSI_AAAA substitution would strongly imply this truncation mutant phenotype is due to the lack of LIR motif.

1. It is also not clear whether the loss of the C-terminal 11 amino acids may have affected the interaction of the M2 protein with other proteins such as TRAPPC6A-delta (Zhu et al., 2017).

This is a reasonable point, however Zhu et al., 2017 (https://doi.org/10.1128/jvi.01757-16) reported that the interaction with TRAPPC6A retains M2 intracellularly. If the phenotype observed with our truncation was due to the loss of interaction with TRAPPC6A, the opposite phenotype would be observed (more M2 in the plasma membrane with the truncated M2∆86-97 mutant). To address this point directly we will attempt to rescue an M2 mutant virus that has disrupted the reported TRAPPC6A binding site and assess M2 plasma membrane localization.

The authors did not rule out whether the truncation of the M2 protein by 11 amino acids would have an effect on proton channel activity. Proton channel activity, however, might be important to preserve the metastable conformation of HA in the secretory pathway and might be also important for virus uncoating.

M2D86-97 induced less LC3 lipidation than wild-type M2 or the D85A mutant. The remaining lipidation was attributed to the ion channel activity of the M2 protein. Can the authors rule out that the truncation of the M2 protein led to reduced ion channel activity which in turn led to reduced LC3B lipidation?

We have addressed points 4 and 5 in response to Reviewer 1.

The suggested role of caspase cleavage as a regulatory switch between filamentous and spherical virions (lines 304- 313) is highly speculative as long as the authors do not provide any experimental proof for it. The authors indicated that they were unable to rescue filamentous IAV with M2D86-97. However, would it be possible to use caspase inhibitors to test their hypothesis?

We acknowledge that M2∆86-97 could not be rescued in a filamentous background. The use of caspase inhibitors would only increase the amount of full length M2 present, and does not provide an alternative strategy for increasing the proportion of truncated M2. However, since M2∆86-97 mutant could not be rescued, we will attempt to rescue additional LIR motif mutants to address this point. In particular, D87A and D88A mutants could be generated in a MUd background, as well as the F91S mutant.

The authors used only the PR8 strain for their studies, a highly cell culture-adapted strain with spherical morphology. Are the findings obtained with this strain are also valid for others IAV strains?

As we highlight in Figure 2I, both the caspase cleavage motif and LIR motif are highly conserved in human IAV strains. PR8 was used as it is the reverse genetic system in use and approved for use in the lab. We will attempt to address this by testing whether other IAV strains we are able to obtain also undergo caspase mediated cleavage of M2. If possible, we will obtain recent clinical isolates to show cleavage of M2 in a strain that has not adapted to cell culture.

1. The authors mainly used the THP-1 cells for their studies, a human macrophage-like cell line. However, human IAV mostly replicate in epithelial cells of the respiratory tract and cause only abortive infections of macrophages. Why did the authors choose this cell line? Can the findings obtained with this cell line be translated to epithelial cells of the airways?

THP-1 cells are widely used for the study of caspase activity. However, we also show M2 cleavage in MDCK cells and HAP1 cells. PR8 infection of A549 cells does not induce significant amounts of cell death in the infection time points used and, as caspase activation is linked to cell death, we did not observe M2 cleavage in this cell type. We will attempt to infect some epithelial cell types to confirm this phenotype.

1. Minor issues:

2. Fig. 1C: There seem to be quite some differences in the cleavage efficiency of M2 between panels A, B, C, and D? Any explanations?

Different cell types (THP-1 cells and HAP1 cells) are used for the experiments mentioned above, which accounts for the different amount of M2 cleavage.

• Fig. 1: Panel E: The labeling of the first amino acids as aa 76 seems to be wrong!

We thank the reviewer for pointing this out, this will be corrected in the updated manuscript.

Line 147: ...caspase mediated disruption of the M2-LC3 interaction (Fig 2A-B). Should be Fig. 2A-C.

This sentence was referring to Figure 2A-B, as it refers to LC3B lipidation and not the coIP. This sentence will be changed in the text to reflect the intended meaning.

• Growth kinetics of the various mutant viruses are missing?

__We will provide growth kinetics for the relevant mutants _(M2D85A and M2∆86-97).___

• Line 195: The authors speculate that aa85 is important for viral fitness: That should be demonstrated!

This speculation is based on the very strong conservation of D85 in human IAV strains. The importance of D85 in viral fitness (permitting cleavage of M2) is only likely to be directly demonstrable in transmission models (for example ferrets) which is not feasible or justifiable.

Reviewer #2 (Significance (Required)):

Authors concluded that caspases regulate the interaction of M2 protein with LC3 which impacts virion production. Specifically, they propose that caspase-mediated removal of the LIR motif may enable a switch between filamentous and non-filamentous budding in response to depletion of cellular resources. However, the authors were unable to rescue a filamentous IAV with a truncated M2 protein and therefore could not provide direct proof for their guess. +<br /> +

• As stated in the response to the comments above, we will attempt to rescue LIR mutant viruses (____D87A and D88A) in a MUd background which would provide further support for our hypothesis. Our data has significance for the understanding of the cell biology of influenza infection as commented on by Reviewers 1 and 3.

• • Reviewer #3 (Evidence, reproducibility and clarity (Required)): Summary : In this article, the authors identify a caspase cleavage site in the influenza A virus (IAV) Matrix 2 protein (M2) that leads to a truncated form of M2 deleted from its C-term LC3-interacting region (LIR). This cleaved form of M2 is seen and accumulates starting at 12 hours post-infection. IAV expressing M2 delta 86-97 mutant, corresponding to cleaved M2, seems to disrupt LC3B localization to cell plasma membrane upon infection. The authors also show that the IAV M2 delta 86-97 has a reduced viral titer compared to IAV WT. Overall the data are quite exciting where the authors identify the specific caspase responsible for the cleavage and show the residues of M2 necessary for LC3 interaction. However, some of the data showing the consequence of the cleavage for viral replication could be better clarified.

We thank Reviewer 3 for their kind comments and we propose further experiments to clarify the consequences of cleavage.

Major comments: - In Fig3A-B, the authors seek to demonstrate that the localization of M2 to the plasma membrane requires LIR motif. However, the representative images for cell infected with the delta 86-97 mutant show relatively few cell are expressing M2 raising questions of the infectivity of this mutant virus or if the overall expression of M2 in this assay is less for the delta 86-97 mutant. The authors should consider first quantifying the ratio of M2 cell surface staining over total M2 staining and second re-evaluate the representative images chosen.

__We will include more examples of permeabilised cells in which comparable numbers of cells are M2 positive between mutants. We will also include high-content microscopy based quantification to support this. __To clarify, we confirm that the quantification of M2 intensity in the plasma membrane is carried out relative to the number of M2 positive cells, as the reviewer agrees is the most accurate way. To avoid confusion, we will update figure legends to describe more accurately the quantification process. A comparison between surface M2 and total M2 cannot be done on an individual cell basis, as once cells are permeabilized (to look for internal M2), robust differentiation between surface and internal M2 is difficult. The above clarification and additional data should provide the necessary support for our conclusions.

• In fig3E, it is unclear what is being quantified in the graph as the legend and text lines 222-223 mention that spot intensity was measured but the y axis indicates LC3 relocalization intensity. Given LC3 is punctated particularly in the cytosol, It is unclear which spots of LC3 they are referring to. Based on the images shown, using a graph with LC3 surface staining as performed for M2 would clarify the data. The authors should clarify the reporting of these data in the results section. Additionally, the images of the control non-infected cells should be added to 3C.

We agree with the reviewer on this point. The figure will be updated to describe more accurately what is being quantified. Additionally, images for uninfected cells in 3C will be added.

• The data in Fig4 and FigS3 need to be strengthened to be conclusive. The volcano plot in FigS3A indicates that there is more LC3B and IAV proteins in M2 D85A than M2delta86-97. However in Fig4E, both LC3 I and LC3 II are increased in virions M2 delta 86-97 compared to M2 D85A which is opposite to the authors' conclusions in lines 244-245. In other words, the total amount of lipidated LC3 is higher in virions from IAV M2 without LIR motif than M2 with LIR. LC3II/I ratio in fig4F would suggest in virions containing M2 with LIR motif, LC3B II may be preferentially incorporated compared to virions containing M2 without LIR, which incorporates both LC3B I and LC3B II. Since this is a critical point made by the authors, performing a co-immunoprecipitation of M2 D58A and M2delta86-97 in the particles and then assessing for binding of LC3 I or II would bolster their conclusions.

Figure 4F quantifies the ratio of LC3II to LC3I in infectious particles. Another two repeats used to quantify this ratio will be shown in addition, with a better representation of increased amounts of lipidated LC3II in M2D85A infectious particles, as well as an increased LC3II/LC3I ration in said particles when compared to M2∆86-97. Because of the low yield acquired from the purification of IAV virions, performing an IP would be difficult. Even if this were technically feasible it would not prove that M2 is binding LC3 inside the virion – we do not make this claim in our paper, merely that LC3B can be detected in the purified viral particles. We will clarify this point in the revised manuscript.

• In Fig4J, even if statistically significant, the PFU difference between M2 D85A and M2 delta86-97 is minimal, performing growth curve assay would help appreciate this difference over time. In Thp1 cells, as the authors show caspase cleavage of M2 at time point 12h 14h 16hpi etc... (fig1), they should also show PFU data at these same time points for M2 mutant D85A compared to WT and M2 delta 86-97.

We agree with the reviewer and indeed this was a point we attempted to make in our manuscript: Figure 4J shows a statistically significant difference between the titers. However, in the text we state that, even though statistically significant, the difference is much smaller than in other titer quantifications performed. Given the nature of a plaque assay, differences of less than a log fold cannot be considered as definitively indicating biological significance. We will clarify this in a revised manuscript. We will also provide the relevant growth kinetics (as per response to Reviewer 2).

• The title of Fig4 and FigS3 and in text line 226 should be changed as M2 incorporation into virions is not shown and not described in the text. Plus, in figS3B, the authors show that between the M2 mutants, there is no difference in the abundance of M2 and other viral proteins compared to M1.

The title of Figures 4 and S3 will be changed to more accurately reflect all of the points made by the figure.

• In the image shown in Fig4H the number of plaques is higher for M2delta86-97 even though the size in smaller than M2 WT. Could the authors clarify in the text of the results section how they quantify PFU in their plaque assay and if they used a size criterion when quantifying the number of plaques?

The images of plaques are taken at different dilutions, with the M2∆86-97 image belonging to two dilutions lower than the M2WT image. We will include the calculation used for PFU/mL, which does not take into account plaque size. Furthermore, images of the whole plate, showing plaqued serial dilutions will be shown.

• In fig3B, the legend indicates 8 hpi but on the graphs it is 9 hpi.

We thank the reviewer for pointing out this mistake. Both should read 8 hpi, this will be corrected in the new manuscript.

Reviewer #3 (Significance (Required)):

The authors demonstrated that IAV M2 binding to LC3 is regulated by caspase cleavage. The authors clearly identify the cleavage site and the caspase involved: caspase 6. The cleaved form of M2 seems relevant to IAV infection as it is accumulating after 12hpi. Using a M2 mutant D85A that cannot be cleaved by caspase 6 and truncated M2 mutant delta86-97 mimicking caspase cleaved M2, the authors are able to elegantly address the role of M2 cleavage. However, the importance of M2 caspase cleavage on IAV infection is not demonstrated. Eventually, addressing the impact of the caspase cleavage of M2 LIR motif on autophagy or CASM would be interesting. - Advance: conceptual. - Audience: basic research, specialized in virology, specialized in autophagy. - Field of expertise: virology, autophagy.

We agree with the reviewer that we have made a conceptual advance in our understanding of the cell biology of influenza A virus infection. We have also determined the structure of the terminal part of the M2 tail in complex with LC3B. The biological importance of the phenotypes we show are most likely in transmission of the virus between hosts, which for IAV would require animal experiments outside the scope of this study. We have demonstrated regulation of the LIR motif by caspase cleavage in a variety of ways, using cell biological and biochemical methods. IAV is a very significant human and animal pathogen, and we believe we have made an important advance in describing a host-pathogen interaction of relevance for viral egress.

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

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary:

In this article, the authors identify a caspase cleavage site in the influenza A virus (IAV) Matrix 2 protein (M2) that leads to a truncated form of M2 deleted from its C-term LC3-interacting region (LIR). This cleaved form of M2 is seen and accumulates starting at 12 hours post-infection. IAV expressing M2 delta 86-97 mutant, corresponding to cleaved M2, seems to disrupt LC3B localization to cell plasma membrane upon infection. The authors also show that the IAV M2 delta 86-97 has a reduced viral titer compared to IAV WT. Overall the data are quite exciting where the authors identify the specific caspase responsible for the cleavage and show the residues of M2 necessary for LC3 interaction. However, some of the data showing the consequence of the cleavage for viral replication could be better clarified.

Major comments:

• In Fig3A-B, the authors seek to demonstrate that the localization of M2 to the plasma membrane requires LIR motif. However, the representative images for cell infected with the delta 86-97 mutant show relatively few cell are expressing M2 raising questions of the infectivity of this mutant virus or if the overall expression of M2 in this assay is less for the delta 86-97 mutant. The authors should consider first quantifying the ratio of M2 cell surface staining over total M2 staining and second re-evaluate the representative images chosen.
• In fig3E, it is unclear what is being quantified in the graph as the legend and text lines 222-223 mention that spot intensity was measured but the y axis indicates LC3 relocalization intensity. Given LC3 is punctated particularly in the cytosol, It is unclear which spots of LC3 they are referring to. Based on the images shown, using a graph with LC3 surface staining as performed for M2 would clarify the data. The authors should clarify the reporting of these data in the results section. Additionally, the images of the control non-infected cells should be added to 3C.
• The data in Fig4 and FigS3 need to be strengthened to be conclusive. The volcano plot in FigS3A indicates that there is more LC3B and IAV proteins in M2 D85A than M2delta86-97. However in Fig4E, both LC3 I and LC3 II are increased in virions M2 delta 86-97 compared to M2 D85A which is opposite to the authors' conclusions in lines 244-245. In other words, the total amount of lipidated LC3 is higher in virions from IAV M2 without LIR motif than M2 with LIR. LC3II/I ratio in fig4F would suggest in virions containing M2 with LIR motif, LC3B II may be preferentially incorporated compared to virions containing M2 without LIR, which incorporates both LC3B I and LC3B II. Since this is a critical point made by the authors, performing a co-immunoprecipitation of M2 D58A and M2delta86-97 in the particles and then assessing for binding of LC3 I or II would bolster their conclusions.
• In Fig4J, even if statistically significant, the PFU difference between M2 D85A and M2 delta86-97 is minimal, performing growth curve assay would help appreciate this difference over time. In Thp1 cells, as the authors show caspase cleavage of M2 at time point 12h 14h 16hpi etc... (fig1), they should also show PFU data at these same time points for M2 mutant D85A compared to WT and M2 delta 86-97.

Minor comments:

• The title of Fig4 and FigS3 and in text line 226 should be changed as M2 incorporation into virions is not shown and not described in the text. Plus, in figS3B, the authors show that between the M2 mutants, there is no difference in the abundance of M2 and other viral proteins compared to M1.
• In the image shown in Fig4H the number of plaques is higher for M2delta86-97 even though the size in smaller than M2 WT. Could the authors clarify in the text of the results section how they quantify PFU in their plaque assay and if they used a size criterion when quantifying the number of plaques?
• In fig3B, the legend indicates 8 hpi but on the graphs it is 9 hpi.

Significance

The authors demonstrated that IAV M2 binding to LC3 is regulated by caspase cleavage. The authors clearly identify the cleavage site and the caspase involved: caspase 6. The cleaved form of M2 seems relevant to IAV infection as it is accumulating after 12hpi. Using a M2 mutant D85A that cannot be cleaved by caspase 6 and truncated M2 mutant delta86-97 mimicking caspase cleaved M2, the authors are able to elegantly address the role of M2 cleavage. However, the importance of M2 caspase cleavage on IAV infection is not demonstrated.<br /> Eventually, addressing the impact of the caspase cleavage of M2 LIR motif on autophagy or CASM would be interesting.

• Advance: conceptual.
• Audience: basic research, specialized in virology, specialized in autophagy.
• Field of expertise: virology, autophagy.

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 influenza A virus (IAV) M2 protein is small transmembrane protein which plays a role in virus entry and egress. In a previous study, Beale et al. (2014) identified an LC3-interacting region (LIR) in the M2 cytoplasmic domain that was found to recruit the LC3B protein to the plasma membrane. Recombinant IAV harboring mutations in the LIR motif showed reduced particle stability and lost filamentous morphology.

In the present study, Figueras-Novoa et al. show that the LIR motif is removed in response to activation of cellular caspases. The authors demonstrate that in in IAV-infected THP-1 cells M2 is partially cleaved at the motif (82)SAVD(85)A by caspase 6. Caspase inhibitors abolished cleavage, and a mutant virus harboring the D85A substitution was found to be resistant to caspase action. A crystal structure of purified M2 C- terminus and LC3B revealed that the caspase cleavage site lies in a flexible region that is accessible to caspases.

Mutant virus encoding a truncated M2 protein (M286-97) was unable to interact with LC3, in accordance with the absence of the LIR motif. The M286-97 mutant showed reduced lipidation of LC3, while enhanced lipidation of LC3 was observed when wild-type virus-infected cells were treated with caspase inhibitors. The authors also observed that cell surface transport of M286-97 but not M2-D85A was impaired. However, in purified virus particles a mix of cleaved and uncleaved M2 was detected. The authors also demonstrated that lipidated LC3B was present in purified virions of wild-type virus particles but even more abundant in M2-D85A virions. Finally, M286-97 mutants produced significantly less infectious particles compared to wild-type virus while the D85A cleavage mutant replicated to similar titers than wt virus.

Based on these findings the authors concluded that caspases regulate the interaction of M2 protein with LC3 which impacts virion production. Specifically, they propose that caspase-mediated removal of the LIR motif may enable a switch between filamentous and non-filamentous budding in response to depletion of cellular resources. However, the authors were unable to rescue a filamentous IAV with a truncated M2 protein and therefore could not provide direct proof for their guess.

While the data are sound and presented well, they do not support the conclusions of the authors.

1. To the authors opinion, the conserved caspase cleavage site in the M2 protein might provide an evolutionary advantage for the virus. However, the M2-D85A mutation has no effect on viral replication, so the biological significance of why M2 needs to be cleaved at all is unclear. The conclusion that caspase-induced M2 cleavage is a fine-tuning mechanism of IAV has not been supported by experiments.
2. The finding that the permanently truncated IAV M2 mutant virus was substantially attenuated does not necessarily mean that abrogation of M2-LC3 interaction was responsible for this attenuation. As the M2 protein plays a role in virus budding at the plasma membrane (recruitment of M1 protein, induction of membrane curvature, membrane scission), the impaired transport of the truncated M2 protein might already explain that the virus was attenuated and that incorporation of the protein into the viral envelope was reduced.
3. It is also not clear whether the loss of the C-terminal 11 amino acids may have affected the interaction of the M2 protein with other proteins such as TRAPPC6A-delta (Zhu et al., 2017).
4. The authors did not rule out whether the truncation of the M2 protein by 11 amino acids would have an effect on proton channel activity. Proton channel activity, however, might be important to preserve the metastable conformation of HA in the secretory pathway and might be also important for virus uncoating.
5. M286-97 induced less LC3 lipidation than wild-type M2 or the D85A mutant. The remaining lipidation was attributed to the ion channel activity of the M2 protein. Can the authors rule out that the truncation of the M2 protein led to reduced ion channel activity which in turn led to reduced LC3B lipidation?
6. The suggested role of caspase cleavage as a regulatory switch between filamentous and spherical virions (lines 304- 313) is highly speculative as long as the authors do not provide any experimental proof for it. The authors indicated that they were unable to rescue filamentous IAV with M286-97. However, would it be possible to use caspase inhibitors to test their hypothesis?
7. The authors used only the PR8 strain for their studies, a highly cell culture-adapted strain with spherical morphology. Are the findings obtained with this strain are also valid for others IAV strains?
8. The authors mainly used the THP-1 cells for their studies, a human macrophage-like cell line. However, human IAV mostly replicate in epithelial cells of the respiratory tract and cause only abortive infections of macrophages. Why did the authors choose this cell line? Can the findings obtained with this cell line be translated to epithelial cells of the airways?

Minor issues:

• Fig. 1C: There seem to be quite some differences in the cleavage efficiency of M2 between panels A, B, C, and D? Any explanations?
• Fig. 1: Panel E: The labeling of the first amino acids as aa 76 seems to be wrong!
• Line 147: ...caspase mediated disruption of the M2-LC3 interaction (Fig 2A-B). Should be Fig. 2A-C.
• Growth kinetics of the various mutant viruses are missing?
• Line 195: The authors speculate that aa85 is important for viral fitness: That should be demonstrated!

Significance

Authors concluded that caspases regulate the interaction of M2 protein with LC3 which impacts virion production. Specifically, they propose that caspase-mediated removal of the LIR motif may enable a switch between filamentous and non-filamentous budding in response to depletion of cellular resources. However, the authors were unable to rescue a filamentous IAV with a truncated M2 protein and therefore could not provide direct proof for their guess.

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 Matrix 2 (M2) protein of influenza A virus (IAV) is a single pass transmembrane protein known to act as a tetrameric ion channel that is important for both viral entry and egress. The paper by Figueras-Nova et al. entitled "Caspase cleavage of Influenza A virus M2 disrupts M2-LC3 interaction and regulates virion production" reports on the regulation of IAV virion production through a regulatory interplay between a caspase cleavage site and a LC3 interacting region (LIR) motif in M2. In its C-terminal cytoplasmic tail the IAV M2 protein contains a C-terminal LIR motif interacting with LC3. The authors show that this LIR motif is preceded by a functional caspase cleavage motif cleaved predominantly by caspase-6, with some contribution from caspase-3: The motif 82-SAVD-85 directs cleavage after the aspartate (D) at position 85. The cleavage leads to loss of the remaining C terminal sequence from amino acid 86 to 97. The core LIR motif 91-FVSI-94 LIR motif is then lost from M2 which can no longer bind LC3. As previously described by the same group using point mutations in the LIR motif (Ref 12.), loss of a functional LIR., here by caspase- mediated deletion of the LIR, affects the virion production and inhibits filamentous budding. LC3B lipidation is increased upon treatment with a caspase inhibitor. The authors show for the first time that LC3 is included into IAV virions via binding to M2. Furthermore, they also report a co-crystal structure of the M2 C terminus (aa 70-97), containing the caspase cleavage site and LIR, and LC3B (aa 3-125) adding new insights into this interaction and showing that the caspase cleavage site is in a flexible region N-terminal to the LIR. This work shows how caspase cleavage may modulate LC3B lipidation, trafficking to the plasma membrane, incorporation of LC3B in the virions, filamentous budding and virion production (viral titer).

Major comments:

The findings reported here are very well supported by the data shown. This is a very clearly written paper with well described and nicely visualized results that are accompanied by adequate statistical analyses. The authors report a new way the LC3B binding to the C-terminal tail of the M2 proteins is regulated and suggest that this is an adaptation the virus has made to adjust virion production to host cell status by hijacking the function of host caspases. They show that the caspase cleavage motif is evolutionary conserved and use that as an argument. Perhaps it could be discussed if it also could be an argument that the host protects itself against a too massive virion production as this could be too detrimental to the host? Would it not also be an evolutionary advantage to the virus in the long run by avoiding killing the host? A question I may raise which is optional as it may be too much work to address as part of this study is if the reported regulation of LC3B binding has any role in regulating the ion channel function of the M2 tetramer?

Minor comments:

Delete "with" in line 145. Line 217: It should be written more specifically how "cells were surface stained with M2" In the Introduction a description of what filamentous vs "spherical" budding is, could perhaps be included as I missed that reading through, although it comes in the end of the Discussion.

Significance

This is a very well performed study with a sound experimental strategy and well performed assays with clear results increasing our insight into the interplay between the Influenza A virus and host cells. Although caspase mediated cleavage of the autophagy receptor and signaling scaffold protein p62 (Ref. 25), removing the LIR and LC3-binding, has been reported before I consider this study as novel in reporting this type of regulation of LC3 binding. The cleavage of p62 deletes a large part of the protein while here it is a "clean" deletion of the LIR sequence representing a conceptual advance of regulation of LC3 binding.

The study also reports for the first time on LC3B incorporated into virions.

The effects on trafficking to the plasma membrane and viral budding and virion production are similar to those reported before (Ref. 12) using viruses with point mutations crippling the LIR motif. This research will be of interested to all studying virus- host interaction and to the autophagy field both as a non autophagic role of LC3B, and as a regulatory mechanism of LIR-LC3B interactions involving the irreversible caspase cleavage-mediated deletion of the LIR motif.

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

This study by Mordier and colleagues represents an in depth analysis to clarify the evolutionary history and processes of the rapidly evolving Schlafen gene family with a strong focus on primates and rodents.

The study is of high quality in my opinion, though I do have some minor comments:

1. Fig 2 and Fig 4B present inferred phylogenetic trees of schalfens in primates and rodents - these trees appear to be unrooted or rooted on a single species rather than an outgroup/gene. I suggest that the authors consider whether an outgroup gene could be included or if an outgroup free approach could be used to estimate the position of the root. This is important because the use of an unrooted tree to make inferences on gene family evolution has important implications - for example, there are no clades in an unrooted tree (Wilkinson et al 2007, Trends Ecol Evol).
2. Schlafen proteins beyond mammals are referred to as SLFN11, it is not clear why this is the case because they seem to be co-orthologous to all mammal schalfen groups (except SLFNL1) based on supplementary figure S2. In this context, perhaps this image should form part of the main text?
3. For blast searches parameters should be included - what cutoffs were implied for similarity searches etc. Related to this on line 120-121 homology is described as 'significant'. Homology refers to an evolutionary relationship, sequence similarity may be significant or not based on the search performed but homology is qualitative and simply detectable or not.
4. The first results section describes the results of phylogenetic analyses, however this section relies heavily on what might better be considered interpretation of these analyses, this is great and should be included but I suggest that the branching patterns in the trees and bootstrap values supporting relationships between genes are also reported in the text to link interpretations to actual results.
5. Bustos 2009 included viral genes belonging to the family in their analyses and I think it may be pertinent to do so here also to determine if the results are consistent or not.
6. Was a rate heterogeneity (e.g. gamma rates / +G) parameter considered in phylogenetic analyses or model testing, it is not reported here and very rare for this not to improve model fit and phylogenetic accuracy.
7. The authors state that all data are available in public databases, but this is not the case for the results they generated. Making various file types produced in this study would be good - e.g. alignments, phylogenetic tree files, structures, etc.

Significance

This study is an important step forward in clarifying our understanding of schalfen evolution. I think the manuscript will be of interest to a number of research areas, including gene family evolution because of its focus on an unusually rapidly evolving gene cluster and to those working on the schalfen gene families functional importance in development and immunity. The results may also draw interest from those interested in the confluence of protein structure, function, and evolution. My expertise In the context of this study is in the phylogenetics and evolution of rapidly evolving gene families.

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 current manuscript, Mordier et al. combine bioinformatic searches, synteny, and phylogenetic analysis to reconstruct the duplicative history of the Schlafen Genes in rodents and primates and then use molecular evolution analyses in combination with structural modeling to make inferences regarding the role of natural selection in the evolution of this gene family. The study represents an update on Bustos et al. (2009), who had already presented evidence that Positive Darwinian selection was likely a factor in the diversification of these genes in mammals. In this context, the contribution of this paper is the identification of sites that are candidates to be evolving under natural selection, and the structural exploration of the location of these sites in the proteins. CODEML strength lies in the detection of signatures of positive selection at the codon level, but it is not that accurate when it comes to pinpointing the actual sites that might be under selection. Hence, without experimental data, these inferences remain speculative. The manuscript is well-written and represents an update on the evolution of this gene family.

Major Issues

The rationale for the choice of species included in the analyses is never presented, and some of it is hard to understand. Why do authors exclude the platypus but include non-mammalian lobe-finned vertebrates is not clear. If they are going to discuss the evolution of these genes outside mammals, the authors need to survey a much wider array of genomes. Even within mammals, there is little discussion on why some species were included and others not. I think that focusing the study on rodents and primates is OK, but I also think that providing a strong justification of the selection of species to include in the study and a tree that justifies splitting the focus on rodents and primates would also be important.

In the trees in Figures 2 and 4, several genes considered as orthologs are not in monophyletic groups. These pattern aligns well with the birth-and-death model of gene family evolution, and has implications for their molecular evolution analyses. The authors need to address this issue explicitly. I would use topology tests to evaluate whether these deviations from the expected topology are significant. In addition, the relevant tests to report here are M8 vs M7 and M8 vs M8a. The M0 vs M1a comparison does not provide evidence for positive Darwinian selection. If the M8 vs M7 and M8 vs M8a tests are not significant, the inferences about sites evolving with dN/dS>1 are not really valid.

CODEML can implements models that are designed to test patterns of gene family evolution, contrasting pre and post duplication branches, which I think would be of value in this family.

Some analyses are described very succinctly, which would make replication challenging.

Minor Issues

Could 2R be responsible for the emergence of SLFN and SLFNL1?

There are several minor issues authors should fix in a revised manuscript. In general, because results are presented before the materials and methods, I think it is easier for readers to have some of the information in the results section.

They need to be consistent in using italics for species names as well as for capitalization.

In the Alignment and maximum-likelihood phylogenies section the authors indicate that they used either Muscle or Mafft for the alignments. What was the rationale for picking one alignment over the other for a given gene? In this section, they also indicate the selected a best-fitting model of substitution using SMS, but then indicate that they used JTT for protein alignments and HKY for nucleotide alignments.

How did the authors ensure that nucleotide alignments remained in frame?

Significance

I think this is a significant contribution to our understanding of the evolution of the Schlafen gene family. There are two key contributions here: the demonstration that gene conversion is a factor obscuring relationships among genes in this gene family, and the mapping of amino acids inferred be evolving under positive selection to structurally important residues of the proteins. These residues should be of interest for functional assays that evaluate the functional role of these 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

Mordier et al. used in-depth phylogenomic methods to analyze the evolution of the mammalian Schlafen gene family. They identified a novel orphan Schlafen-related gene that arose in jawed vertebrates, and they assigned orthology between Schlafen cluster paralogs. This will allow for further accurate selection studies. Throughout the entire manuscript, the authors use nomenclature predating structural and biochemical studies. The nomenclature is purely based on sequence similarities, which are sometimes very weak and not convincing, and not based on the known function of the protein. In my opinion, this causes confusion and does not help scientists in the field. Especially in Figure 3, I wouldn't call it RNAse E (AlbA); instead, tRNA recognition site,endoribonuclease domain, SLFN core domain are the correct domain designations. Since SLFN11 is not a GTPase, why do the authors name the domain GTPase domain? Actually, the SWADL domain comprises a SWAVDL instead of a SWADL sequence motif. Hence, I would name the domain SWAVDL domain instead of SWADL domain, which is, in my opinion, misleading and was wrongly chosen in initial publications.

In e.g. Figure 3 SLFN11 structure it would be better if the authors illustrated the important residues concerning the known RNase active site and ssDNA binding site. Further, a close-up of the SLFN11 interface with labeled amino acids involved in the interaction and highlighting the residues undergoing positive selection would help understand the evolutionary adaptation.

Although, according to Metzner et al., the SLFN11 dimer is built up by two interfaces (I and II), where Interface I is situated in the C-terminal helicase domain and Interface II in the N-terminal SLFN11 core domain. It would be helpful for the reader if the authors stuck to this already introduced and widely accepted nomenclature in the field.

In addition to the antiviral function, SLFN11 expression levels have been reported to show a strong positive correlation with the sensitivity of tumor cells to DNA damaging agents (DDAs). Hence, SLFN11 can serve as a biomarker to predict the response to, e.g., platinum-based drugs. It was revealed that SLFN11 exerts its function by direct recruitment to sites of DNA damage and stalled replication forks in response to replication stress induced by DDAs. Could the authors include this different molecular function of SLFN11 in their discussion of SLFN11s evolution and positive selection?

Even though it seems unclear from the genetic and evolutionary aspect (Figure 4), mouse Slfn8 and Slfn9 complement human cells lacking SLFN11 during the replication stress response and seem to resemble the function of SLFN11 (Alvi et al. 2023). The authors of this study claim that Slfn8/9 genes may share an orthologous function with SLFN11. Could the authors comment on that discrepancy?

Significance

In general, the work is well conducted and provides valuable new insights in an important and growing field of research. However, there are some limitations to the study including the disregard of known protein function (e.g. SLFN11) and the usage of a purely sequence similarity based nomenclature.

Transparent Peer Review

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

Reply to the reviewers

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

#1) Summary: The transport of effector proteins across membranes from the producing bacterium into a target cell is at the core of bacterial secretion systems. How an additional layer in form of a capsule affects effector export and the susceptibility towards effector import is not fully understood. Here, Flaugnatti and colleagues combined bacterial genetics with phenotypic assays and electron microscopy to demonstrate a dual role of a bacterial capsule in preventing T6SS-mediated effector export and promoting protection from effector import by another bacterium's T6SS. The wide variety of methods used, complementation of the mutants, and validation of the findings across strains strengthen the author's conclusions. Although the main conclusions seem straight forward, the authors unravel the unexpected complexity underlying these phenotypes with strong mechanistic work. In brief, a capsule-deficient mutant (∆itra) is shown to assemble its T6SS similar to the WT, yet secretes more Hcp than the WT and is better in T6SS-mediated killing of other bacteria. A capsule-overproducing mutant (∆bfmS) shows both, a partial deficiency in T6SS assembly and an additional reduction in exported Hcp, and is worse in T6SS-mediated killing than the WT. A mutant with a capsule similar to WT and deficient in cell sensing (∆tslA) forms the least T6SS apparatuses and is yet better in T6SS-mediated killing than the overcapsulated mutant. Together, these data show an effect of the capsule on (i) T6SS apparatus assembly, (ii) effector export, (iii) effector import, and (iv) the need for clearance of accumulating non-secreted Hcp by ClpXP. The work on a clinical isolate of Acinetobacter tumefaciens and the data on an impaired T6SS activity on other cells by antibiotic-induced capsulation is a strong demonstration of the work's clinical relevance in addition to the findings' conceptual novelty.

• In my view, the manuscript is outstanding with very high quality of experimental data, very well written text and very clear presentation of the data in figures. A few minor comments and suggestions below that I think would strengthen the manuscript.*

__ Authors’ reply #1: __We thank the reviewer for their enthusiasm.

• *

Major comment:

#2) OPTIONAL: Fig. 4c/l. 320: Having an indirect effect of an antibiotic on T6SS activity by antibiotic-induced capsule formation is very intriguing and contributes to the clinical relevance of the overall findings. When I saw the data in Fig. 4c, the graph instantaneously reminded me of the panel in Fig. 2a, where a similar phenotype is observed by changing the predator:prey ratio in the absence of any antibiotic. The authors themselves comment on the possibility of antibiotic-induced, reduced predator growth (and thereby a change in predator:prey ratio) as a one factor impacting the phenotype here. I am wondering if this data could be strengthened or better disentangled to test more precisely if it is the antibiotic induced capsule formation per se that affects T6SS-mediated killing by A. baumanii in the presence of antibiotics. Would it help to take the bfmS mutant along as a control for direct comparison to see if antibiotic-induced capsule formation of the WT to similar levels of the mutant results in the same killing phenotype? Would it help to test for T6SS-mediated killing in the presence and absence of antibiotics at multiple predator:prey ratios? Could the effect of the antibiotic on A. baumanii growth be measured and considered when choosing the ratio at which the bacteria are mixed?

__ Authors’ reply #2: __The point raised by the reviewer is very important. As we have stated in the manuscript, the capsule-induced production using antibiotics impacts the growth of A. baumannii and could therefore change the predator-prey ratio, potentially affecting the observed phenotype. However, the antibiotic is expected to equally impact the non-encapsulated ΔitrA strain, yet this strain maintains very strong T6SS killing activity in the presence of chloramphenicol. Thus, we do not believe the predator-prey ratio is causing the observed effect. To address this point more directly, we nonetheless propose to: i) repeat the experiments with different predator-prey ratios (1:1, 2:1, and 5:1), and ii) include a bfmS mutant as a control.

Minor comments:

#3) Figure 1D, l. 155, I might have missed this, do the authors happen to have the numbers of E. cloacae as well? This would strengthen the claim on A. baumannii survival because of E. cloacae is being killed.

__ Authors’ reply #3: __The reviewer is correct; we did not include the survival of E. cloacae in the initial manuscript due to technical reasons (counter-selection of E. cloacae). However, we propose to repeat the experiment using an E. cloacae strain carrying a plasmid conferring kanamycin resistance. This will allow us to counter-select E. cloacae after contact with the A. baumannii predator to determine if E. cloacae is killed by A. baumannii in a T6SS-dependent manner.

#4) Figure 2, I suggest to write out the species name of the prey in the box with the ratio. With E. cloacae being referred to in the previous figure and starting with similar letters than E. coli, I wasn't sure at first sight what E. c. refers to.

__ Authors’ reply #4: __We appreciate the comment and will revise the figure as suggested.

#5) use of the term "T6SS activity" throughout the manuscript (e.g. l. 182, l. 187). I leave this up to the authors. To me, it seems like an umbrella term for the initial observation and I see that such a term can be very handy for the writing. I just would like to mention that the use of the term was not always intuitive to me and sometimes even a bit misleading. For example, l. 182 refers to "increased T6SS activity". As a reader, I only know about 'T6SS activity on other cells' or 'a T6SS-mediated effect on other cells' at this point. T6SS apparatus assembly/firing activity is tested for specifically later and it turns out to differ between mutants. By the time the term is used in the discussion, it captures multiple nuanced phenotypes described by then. The more precise definition of the term in l. 200 helped to capture what exactly is meant by the authors.

__ Authors’ reply #5: __We propose rephrasing the sentences to include the term "T6SS-secretion activity" when referring to Hcp secretion assays and "T6SS-mediated killing activity" when referring to killing experiments.

__#6) __l. 198-199 "Collectively, our findings indicate that CPS does not hinder the secretion process of the T6SS or the consequent elimination of competing cells". I might be missing something, I cannot entirely follow this sentence. Didn't the authors just show that the CPS does hinder T6SS-mediated elimination of competing cells in panel 2A and less secreted Hcp in the encapsulated WT compared to the non-encapsulated mutant in panel 2B?

__ Authors’ reply #6:__ We thank the reviewer for this comment. We realize that the sentence wasn’t well phrased, resulting in confusion. What we meant was that the T6SS is functional regarding its T6SS-mediated killing and secretion in the WT strain, while we also showed that the non-capsulated strain kills and secretes more T6SS material in the supernatant. Thus, there seems to be a balance between capsule production and T6SS activity in the WT. We will revise the sentence to better reflect this meaning.

#7) l. 224, typo, "in"

__ Authors’ reply #7:__ We will correct this typo. Thank you.

• *

#8) Two connected comments: l. 338, Just a thought, I am wondering about the title of the section. After reading it a second time, I think it is technically correct. When reading it first, I was a bit confused when getting to the data because apparatus assmebly is impaired in the capsule-overproducing strain and although "preserved", doesn't the data indicate that there is less T6SS assembly in the bfmS mutant and that this might be because of less cell sensing and isn't this a main point that there is a difference in apparatus assembly in the capsule overproducing strain compared to WT (other than no difference in apparatus assembly in the strain without capsule)? To me it seems not fully acknowledged as a finding in the interpretation of the data that less cells of the bfmS mutant have a T6SS apparatus. Isn't that interesting? A title along the lines of "Capsule-overproducing strain has preserved sensory function and assembles less T6SS apparatuses" would have been more intuitive for me. l. 352, In case I didn't miss a reference to this data earlier in the manuscript, I am wondering if it would be worth mentioning the finding on the reduced apparatus assembly of the bfmS mutant earlier, together with Figure 3 already. At least a sentence that mentions already that there is more coming later. When I got to this line in the manuscript and read the findings on the apparatus assembly, I first needed to go back to figure 3 and look at the data there again in light of this finding. It is mentioned here on the side but I think very important for the interpretation of the phenotypic data of the bfmS mutant shown earlier, isn't it? The tslA mutant is used beautifully here.

__ Authors’ reply #8:__ We thank the reviewer for the suggestion and the kind comment about the beautiful usage of the tslA mutant. We will modify the title of the corresponding paragraph as suggested to make it more intuitive.

        Regarding the comment about mentioning the T6SS apparatus assembly defect in the *bfmS* mutant earlier, we respectfully disagree. While we agree that this point is important and can partially explain the difference in killing activity, we believe that showing it together with the *tslA* mutant (Figure 5) makes more sense and is easier for the reader to understand.

#9) Discussion: optional comment. On the one hand, I like the concise discussion. On the other hand, I see more potential here for bringing it all together (potentially at the expense of shortening some of the introduction). I think the subtleties of the findings are complex. For example, I could envision a graphical summary with a working model on all the effects of a capsule on the T6SS and its potential clinical relevance making the study accessible to even more readers.

__ Authors’ reply #9: __In the revised manuscript, we will include a graphical summary/model.

Significance

#10) General assessment: I consider the story very strong in terms of novelty, experimental approaches used, quality of the data, quality of the writing and figures of the manuscript. In my view, the aspects that could be improved are optional/minor and concern only one figure and some phrasing.

• Advance: I see major advance in the findings (i, mechanistic) on the mechanism of how the capsule interferes with T6SS, (ii, fundamental) on the discovery of ClpXP degrading Hcp, and (iii, clinical) on the meaning of antibiotic treatment for the T6SS of this clinically relevant and often multi-drug resistant bacterial species, which strongly complements existing work on the T6SS and antibiotics in A. baumanii (e.g. of the Feldman group). As the authors write themselves, the starting points of the study of a capsule protecting from a T6SS and the effect of a T6SS on other cells being negatively impacted by a capsule were known, although not studied in one species and not understood mechanistically.*

• Audience: I see the result of interest to a broad audience in the fields of bacteria-bacteria interactions, Acinetobacter baumanii, type VI secretion, antimicrobial resistance, bacterial capsules.*

__ Authors’ reply #10: __We once again thank the reviewer and highly appreciate their positive and constructive feedback on our work. We hope the reviewer will be satisfied with the revised version of our manuscript.

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

#11) In the manuscript by Flaugnatti et al., the authors provide clear evidence of the interplay between capsule outer coat production and the Type VI secretion system (T6SS) in Acinetobacter baumannii. The authors demonstrate that the presence of the capsule or the activity of the T6SS enhances survival against attacking bacteria. However, they also show that in their model bacterium, the (over)production of the capsule likely hinders T6SS dynamics, thereby reducing overall killing efficiency. Additionally, they reveal that the amount of the T6SS component Hcp is regulated in cells that can no longer assemble and/or secrete via the T6SS, presumably by the ClpXP protease. Overall, the experiments are well designed, and most conclusions are supported by the data and appropriate controls. I have however some suggestions that could further strengthen the manuscript prior to publication.

__ Authors’ reply #11: __We are grateful for the reviewer’s enthusiasm and will implement their comments and suggestions in the revised version of the manuscript.

Major comments:

#12) Line 164. The authors use E. coli as prey to test the T6SS activity of A. baumannii. Why not directly use the E. cloacae strain (with or without T6SS) for this purpose? This would provide direct evidence that A. baumannii uses its T6SS to kill E. cloacae, thus confirming the authors conclusions in this section.

__ Authors’ reply #12: __We thank the reviewer for this comment. We used E. coli to assess the functionality of the T6SS in different strains of A. baumannii, as it is commonly done in the T6SS field. However, as suggested by reviewer 1 (see comment #3) and in response to this query, we will also provide survival data of E. cloacae in the revised manuscript using a plasmid-carrying E. cloacae derivative that allows direct selection.

#13) In Figure 2, the authors show that a non-capsulated strain kills more effectively and secretes more than a WT, but has a similar number of T6SS. They suggest in their conclusion that "the observed increase in T6SS activity in the non-capsulated strain suggests a compensatory mechanism for the absence of the protective capsule layer." This conclusion implies the presence of an "active" regulatory mechanism that would increase the number of successful T6SS firing events, which has not been demonstrated. Could it not simply be that the capsule blocks some shots that cannot penetrate and are therefore ineffective? This hypothesis is mentioned in lines 204-208. The authors should clarify the conclusion of this section. Given the challenge this may pose in A. baumannii, I suggest that the authors quantify the assembly/firing dynamics of the T6SS under WT and ΔitrA conditions. This would help distinguish between the two hypotheses explaining better firing in non-capsulated cells: i.e., if the number of assembled T6SS is the same in both strains (Fig 2C & 2D), do non-capsulated cells assemble/fire faster, indicating an adaptation in regulation, or do we observe the same dynamics, suggesting a simple physical barrier blocking the passage of certain T6SS firing events?

__ Authors’ reply #13:__ We realize that the sentence, and more specifically the word "compensatory," might have been misleading and thank the reviewer for bringing this to our attention. What we meant to convey is that there is a balance between capsule production and T6SS activity; if disturbed, the balance shifts in one direction or the other. Specifically, there is more protection through the production of a thicker capsule (e.g., in the ∆bfmSmutant or under sub-MIC conditions of antibiotics, regulated by the Bfm system, as mentioned in the text) or more T6SS activity when less capsule is present (e.g., in the ΔitrA mutant, which we propose is caused by the lack of the steric hindrance). We will rephrase this sentence in the revised manuscript to better convey this message.

        Regarding the quantification of T6SS dynamic assembly/firing events between the capsulated (WT) and non-capsulated (ΔitrA) strains, we do not think this is required for this study, as the amount of secreted Hcp reflects the overall activity of the system. Importantly, we also do not have the technical means to quantify assembly/firing rates under Biosafety 2 conditions, as this requires specialized microscopes with very fast acquisition options (see, for instance, Basler, Pilhofer *et al.*, 2012, *Nature*). Indeed, very few labs in the T6SS field have been able to measure such rates.

#14) Line 428-429. It is mentioned that the deletion of lon does not have a notable effect. However, I observe that the absence of Lon alone causes a more rapid degradation of Hcp in the cells compared to the WT strain (Fig 7B). How do the authors explain that the absence of this protease (whether under conditions of Hcp accumulation or not) increases the degradation of this protein in the cell? This explanation should be included in the manuscript.

__ Authors’ reply #14: __That’s a fair point. We didn’t address this point further, as the deletion of lon didn’t resolve the issue of why Hcp is degraded. In fact, the opposite seems to be the case, as there is less Hcp in the ∆lon strain compared to the WT. While this observation is not directly relevant to the question of why Hcp is degraded late during growth in secretion-impaired strains, we will properly mention it in the revised manuscript.

        Please also note that a strong growth defect of a Δ*lon*Δ*clpXP* double mutant impaired further investigation in this direction.

• *

Minor comments:

#15) Throughout the manuscript, the authors use the term "predator" to refer to A. baumannii. Predation is a specific phenomenon that involves killing for nourishment. To my knowledge, the T6SS has never been shown to be a predation weapon but rather a weapon for interbacterial competition, which is a different concept. If this has not been demonstrated in A. baumannii, the authors should replace the term "predator" with "attacker" (or an equivalent term) to clarify the context.

__ Authors’ reply #15: __We thank the reviewer for this comment. The term “predator,” as highlighted by the reviewer, typically implies killing for nourishment/cellular products. In the context of T6SS, it facilitates the killing of competitors, releasing DNA into the environment that can subsequently be acquired through natural competence for transformation, as observed in species like Vibrio cholerae (our work by Borgeaud et al., 2015, Science) or other Acinetobacter species such as Acinetobacter baylyi (Ringel et al., 2017, Cell Rep.; Cooper et al., 2017, eLife). The acquisition of DNA reflects the killing for cellular products of the prey. As most A. baumannii strains are also naturally competent, this justifies the usage of the predator and prey nomenclature.

        Apart from this fact, it seems to be a matter of nomenclature, with many papers in the field using one term or the other. Yet, ultimately, this doesn’t change any of the scientific findings. Therefore, to satisfy the reviewer, we will change “predator” to “attacker” throughout the revised manuscript.

#16) Line 274. Since the authors stated that in the Wzc mutant, the capsule is "predominantly found in the supernatant and only loosely attached to the cell," this result is not unexpected. This finding is also consistent with the previous results from Fig. 3A & B, which show sensitivity to complement-mediated killing and the weak amount of (ab)normal CPS produced in that strain, further confirmed by Fig. 3E.

__ Authors’ reply #16__: We fully agree with the reviewer’s suggestion and will remove the statement.

#17) Line 299. The authors speculate that "... T6SS may deploy through gaps akin to arrow-slits in the capsule's mesh...". However, this is very unlikely since a WT strain kills (Fig. 3C) and secretes (Fig. 2B & 3D) less effectively than the itrA mutant, suggesting that the T6SS is not assembled in the "right places" devoid of CPS; otherwise, we would expect similar T6SS activity. Based on the results in Fig. 2 (and my earlier comment), this implies that A. baumannii assembles its T6SS randomly, and in the presence of the capsule, its shots would need to be in the right place to penetrate the envelope and reach the target. Could the authors comment on this point and provide a model figure to better visualize the interplay between the capsule and T6SS under the three major conditions: WT, non-capsulated, and capsule overproduction?

__ Authors’ reply #17: __We thank the reviewer and agree with their comment. We discussed the hypothesis of T6SS deployment through gaps, drawing a parallel to what was proposed for biofilm and T6SS in V. cholerae(Toska et al., 2018, PNAS). However, as mentioned earlier, we believe that the effect of the capsule on T6SS activity is primarily due to steric hindrance, which increases the distance between the T6SS apparatus and the prey cell. To clarify our findings further, we will include a model summarizing our results, as requested by reviewer 1 (see comment #9).

---

__ #18)__ In Fig. 5A, the microscopy panels should be adjusted to the same dynamic range as the WT (which represents a true signal), which does not appear to be the case for the tlsA mutant panel for instance. The image gives the impression of a large amount of free TssB-msfGFP in the cytoplasm. However, this effect is due to the dynamic range being adjusted to display a signal. This observation is consistent with the fact that the amount of TssB-msfGFP protein is identical across all strains (Fig. S2F).

__ Authors’ reply #18: __We will adjust the images to the range of the WT in the revised manuscript, as suggested. However, regardless of how these images are presented, the enumeration of T6SS structures will remain unchanged, which was the sole point of this experiment.

• *

#19) Unless I am mistaken, the authors do not comment on the fact that in a ΔbfmS strain, the number of T6SS is halved compared to a WT or ΔitrA strain. If capsule overproduction only partially limits the TslA-dependant T6SS assembly, how can this result be explained? Is it related to the degradation of Hcp in this background, which ultimately limits the formation of T6SS? If so, it would be interesting to mention this connection in the section "Prolonged secretion inhibition triggers Hcp degradation”

__ Authors’ reply #19: __We did mention that the T6SS assembly of the ΔbfmS mutant is reduced compared to the WT (or ΔitrA), likely due to the defect in sensing the prey (lines 369-374 and 468-472 of the initial manuscript). However, we will revise the sentence to improve clarity in the revised version of the manuscript.

Significance

#20) This work is highly intriguing as it not only delves into the specific mechanisms involved but also connects fundamental elements in bacterial competition, i.e., the necessity for self-protection and aggression for survival. The manuscript offers valuable insights into cellular dynamics at a microscale level and prompts new inquiries into the regulation of these systems on a population scale. The work is well-done and the writing is also clear. I am convinced that this work represents another significant step towards understanding bacterial mechanisms and will undoubtedly spark considerable interest in the field.

__ Authors’ reply #20: __We sincerely thank reviewer #2 for their constructive inputs, which will improve our manuscript.

• *

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

#21) The manuscript by Flaugnatti et al investigates the relationship between functions of the T6SS in A. baumannii and production of capsular polysaccharide. The manuscript argues that (1) capsule protects A. baumannii against T6SS-mediated attack by other bacteria, (2) capsule also interferes with the bacterium's own T6SS activity, and (3) the T6SS inner tube protein Hcp is regulated by degradation by ClpXP. The main critiques regard the first two conclusions, which seem to be based solely on use of a mutant that has a confounding effect as described below; and to strengthen the third claim by further exploring the results of overexpressing Hcp and by determining whether there is a fitness benefit for Hcp regulation.

__ Authors’ reply #21: __We thank reviewer #3 for their relevant input. We will conduct additional experiments based on their comments, and these will be incorporated into the revised manuscript.

• *

__Main items:____ __

#22) Throughout the paper, an itrA deletion mutant is used as the capsule-deficient strain and conclusions are drawn about role of capsule based on this mutant. However, itrA deletion also eliminates the protein O-glycosylation pathway (Lees-miller et al 2013), a potential confounder. Analysis of mutants specifically deficient in the high-molecular weight capsule but not protein glycosylation, and/or mutants in the protein o-glycosylation enzyme, should be incorporated into the study to enhance the ability to make conclusions about the role of the capsule.

__ Authors’ reply #22: __Fair point. We thank the reviewer for this important suggestion. To distinguish between the O-glycosylation pathway and capsule production, we will generate a ∆pglL strain (specific to O-glycosylation), as suggested, and will repeat the key experiments (similar to Fig. 2A and 2B). We are almost done with the engineering of this mutant strain and therefore don’t expect any major delays.

#23) Evidence could be provided to support the idea raised in lines 482-483 that T6SS component accumulation is toxic ("degradation [of T6SS components] could serve as a strategy to alleviate proteotoxic stress..."). For example, growth curves of ∆clpXP strains with and without hcp could be analyzed, to determine how degrading Hcp is helping the bacteria.

__ Authors’ reply #23: __We will perform growth curves of ΔclpXP strains with and without hcp, as suggested by the reviewer. However, we are uncertain whether we will be able to observe differences between these strains, as the conditions under which such degradation is significant may be challenging to replicate under standard laboratory conditions.

__#24) __The possible ClpXP recognition sequence identified at the C terminus of Hcp is interesting-does overexpression of an Hcp variant lacking/altered in this motif alter its protein levels compared to WT Hcp?

__ Authors’ reply #24: __We thank the reviewer for this suggestion. We are in the process of performing the suggested experiment and will include the data in the manuscript.

__Minor items:____ __

#25) *A better explanation could be provided for why overexpressing hcp in WT but not in ∆hcp leads to increased Hcp protein levels. There is a statement about Hcp being regulated post transcriptionally, possibly by degradation (lines 422-423), but would that not also result in regulation in the WT strain? *

__ Authors’ reply #25: __The reviewer is absolutely correct here. Despite careful genetic engineering, we believe that the hcp mutant used may have a polar effect, causing Hcp accumulation only in the ∆hcp + p-hcp strain but not in the WT + p-hcp strain, which remains capable of secretion. The ∆hcp strain therefore mimics the secretion-impaired tssB mutant. We will clarify this in the revised manuscript.

#26) *An untreated control is needed in Fig. 4B. *

__ Authors’ reply #26: __The untreated samples were shown in all previous figures. However, we understand the reviewer's point and will repeat the experiment with the untreated control included in the same experiment.

#27) *line 179: please clarify "reflecting better invading bacteria" *

__ Authors’ reply #27: __We appreciate the reviewer mentioning this oversight. We meant to compare this to a situation where a bacterium invades an already existing community, resulting in a predator-prey ratio below 1. We will clarify this further in the revised manuscript.

#28) *line 351: consider rewording the statement that ∆tslA results in decreased in T6SS assembly and activity using the tssB-msfGFP microscopy assay; it is not clear that activity is measured in this assay. *

__ Authors’ reply #28: __The reviewer is correct. We will revise the sentence accordingly to better reflect the T6SS assembly.

#29) *lines 260-265: This experiment could use clarifying, but it would seem that it requires analysis of the secreted capsule levels in the tssB mutant to show it does not produce extracellular capsule to the same extent that ∆bfmS does. *

__ Authors’ reply #29: __We thank the reviewer for the suggestion and will include these experimental data in the revised manuscript.

#30) *Fig. 6C and 7A labelling could be improved to avoid potential confusion that the bar graphs are quantifying the western blot. E.g., could add a corresponding vertical label to the Western data, or consider changing "relative expression of hcp" to something reflecting analysis of transcript levels. *

__ Authors’ reply #30: __We will improve this figure by splitting the qPCR and Western blot data into independent panels. This will eliminate any confusion.

#31) lines 416-417 and Fig. 7A: states that "hcp mRNA levels increased significantly", but more careful wording could be used because the WT's transcript change is not significant after overexpression (though it is significant in ∆hcp).

__ Authors’ reply #31: __Point well taken. We will improve the sentence (and Figure) to make its meaning unambiguous.

• *

#32) lines 479-480 states that in secretion-impaired strains accumulation of Hcp is mitigated by ClpXP; while this was shown for ∆tssB, was this also the case for ∆bfmS?

__ Authors’ reply #32: __This is indeed an interesting suggestion. We are in the process of generating the double mutant ∆bfmS∆clpXP and will include the experimental results in the revised manuscript.

Significance

#33) *The strengths of the study are the focus on a clinically significant pathogen, the potential novel roles for the important capsule virulence factor of A. baumannii, and the identification of novel points of control of the T6SS. The analyses of T6SS function are thorough and carefully performed. *

__ Authors’ reply #33: __We thank the reviewer for their comments, which we believe will significantly strengthen our work, particularly regarding the capsule aspect.

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 Flaugnatti et al investigates the relationship between functions of the T6SS in A. baumannii and production of capsular polysaccharide. The manuscript argues that (1) capsule protects A. baumannii against T6SS-mediated attack by other bacteria, (2) capsule also interferes with the bacterium's own T6SS activity, and (3) the T6SS inner tube protein Hcp is regulated by degradation by ClpXP. The main critiques regard the first two conclusions, which seem to be based solely on use of a mutant that has a confounding effect as described below; and to strengthen the third claim by further exploring the results of overexpressing Hcp and by determining whether there is a fitness benefit for Hcp regulation.

Main items:

• Throughout the paper, an itrA deletion mutant is used as the capsule-deficient strain and conclusions are drawn about role of capsule based on this mutant. However, itrA deletion also eliminates the protein O-glycosylation pathway (Lees-miller et al 2013), a potential confounder. Analysis of mutants specifically deficient in the high-molecular weight capsule but not protein glycosylation, and/or mutants in the protein o-glycosylation enzyme, should be incorporated into the study to enhance the ability to make conclusions about the role of the capsule.
• Evidence could be provided to support the idea raised in lines 482-483 that T6SS component accumulation is toxic ("degradation [of T6SS components] could serve as a strategy to alleviate proteotoxic stress..."). For example, growth curves of ∆clpXP strains with and without hcp could be analyzed, to determine how degrading Hcp is helping the bacteria.
• The possible ClpXP recognition sequence identified at the C terminus of Hcp is interesting--does overexpression of an Hcp variant lacking/altered in this motif alter its protein levels compared to WT Hcp?

Minor items:

• A better explanation could be provided for why overexpressing hcp in WT but not in ∆hcp leads to increased Hcp protein levels. There is a statement about Hcp being regulated post transcriptionally, possibly by degradation (lines 422-423), but would that not also result in regulation in the WT strain?
• An untreated control is needed in Fig. 4B.
• line 179: please clarify "reflecting better invading bacteria"
• line 351: consider rewording the statement that ∆tslA results in decreased in T6SS assembly and activity using the tssB-msfGFP microscopy assay; it is not clear that activity is measured in this assay.
• lines 260-265: This experiment could use clarifying, but it would seem that it requires analysis of the secreted capsule levels in the tssB mutant to show it does not produce extracellular capsule to the same extent that ∆bfmS does.
• Fig. 6C and 7A labelling could be improved to avoid potential confusion that the bar graphs are quantifying the western blot. E.g., could add a corresponding vertical label to the Western data, or consider changing "relative expression of hcp" to something reflecting analysis of transcript levels.
• lines 416-417 and Fig. 7A: states that "hcp mRNA levels increased significantly", but more careful wording could be used because the WT's transcript change is not significant after overexpression (though it is significant in ∆hcp)
• lines 479-480 states that in secretion-impaired strains accumulation of Hcp is mitigated by ClpXP; while this was shown for ∆tssB, was this also the case for ∆bfmS?

Significance

The strengths of the study are the focus on a clinically significant pathogen, the potential novel roles for the important capsule virulence factor of A. baumannii, and the identification of novel points of control of the T6SS. The analyses of T6SS function are thorough and carefully performed.

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 Flaugnatti et al., the authors provide clear evidence of the interplay between capsule outer coat production and the Type VI secretion system (T6SS) in Acinetobacter baumannii. The authors demonstrate that the presence of the capsule or the activity of the T6SS enhances survival against attacking bacteria. However, they also show that in their model bacterium, the (over)production of the capsule likely hinders T6SS dynamics, thereby reducing overall killing efficiency. Additionally, they reveal that the amount of the T6SS component Hcp is regulated in cells that can no longer assemble and/or secrete via the T6SS, presumably by the ClpXP protease. Overall, the experiments are well designed, and most conclusions are supported by the data and appropriate controls. I have however some suggestions that could further strengthen the manuscript prior to publication.

Major comments:

Line 164. The authors use E. coli as prey to test the T6SS activity of A. baumannii. Why not directly use the E. cloacae strain (with or without T6SS) for this purpose? This would provide direct evidence that A. baumannii uses its T6SS to kill E. cloacae, thus confirming the authors conclusions in this section.. In Figure 2, the authors show that a non-capsulated strain kills more effectively and secretes more than a WT, but has a similar number of T6SS. They suggest in their conclusion that "the observed increase in T6SS activity in the non-capsulated strain suggests a compensatory mechanism for the absence of the protective capsule layer." This conclusion implies the presence of an "active" regulatory mechanism that would increase the number of successful T6SS firing events, which has not been demonstrated. Could it not simply be that the capsule blocks some shots that cannot penetrate and are therefore ineffective? This hypothesis is mentioned in lines 204-208. The authors should clarify the conclusion of this section. Given the challenge this may pose in A. baumannii, I suggest that the authors quantify the assembly/firing dynamics of the T6SS under WT and ΔitrA conditions. This would help distinguish between the two hypotheses explaining better firing in non-capsulated cells: i.e., if the number of assembled T6SS is the same in both strains (Fig 2C & 2D), do non-capsulated cells assemble/fire faster, indicating an adaptation in regulation, or do we observe the same dynamics, suggesting a simple physical barrier blocking the passage of certain T6SS firing events? Line 428-429. It is mentioned that the deletion of lon does not have a notable effect. However, I observe that the absence of Lon alone causes a more rapid degradation of Hcp in the cells compared to the WT strain (Fig 7B). How do the authors explain that the absence of this protease (whether under conditions of Hcp accumulation or not) increases the degradation of this protein in the cell? This explanation should be included in the manuscript.

Minor comments:

• a) Throughout the manuscript, the authors use the term "predator" to refer to A. baumannii. Predation is a specific phenomenon that involves killing for nourishment. To my knowledge, the T6SS has never been shown to be a predation weapon but rather a weapon for interbacterial competition, which is a different concept. If this has not been demonstrated in A. baumannii, the authors should replace the term "predator" with "attacker" (or an equivalent term) to clarify the context.
• b) Line 274. Since the authors stated that in the Wzc mutant, the capsule is "predominantly found in the supernatant and only loosely attached to the cell," this result is not unexpected. This finding is also consistent with the previous results from Fig. 3A & B, which show sensitivity to complement-mediated killing and the weak amount of (ab)normal CPS produced in that strain, further confirmed by Fig. 3E.
• c) Line 299. The authors speculate that "... T6SS may deploy through gaps akin to arrow-slits in the capsule's mesh...". However, this is very unlikely since a WT strain kills (Fig. 3C) and secretes (Fig. 2B & 3D) less effectively than the itrA mutant, suggesting that the T6SS is not assembled in the "right places" devoid of CPS; otherwise, we would expect similar T6SS activity. Based on the results in Fig. 2 (and my earlier comment), this implies that A. baumannii assembles its T6SS randomly, and in the presence of the capsule, its shots would need to be in the right place to penetrate the envelope and reach the target. Could the authors comment on this point and provide a model figure to better visualize the interplay between the capsule and T6SS under the three major conditions: WT, non-capsulated, and capsule overproduction?
• d) In Fig. 5A, the microscopy panels should be adjusted to the same dynamic range as the WT (which represents a true signal), which does not appear to be the case for the tlsA mutant panel for instance. The image gives the impression of a large amount of free TssB-msfGFP in the cytoplasm. However, this effect is due to the dynamic range being adjusted to display a signal. This observation is consistent with the fact that the amount of TssB-msfGFP protein is identical across all strains (Fig. S2F).
• e) Unless I am mistaken, the authors do not comment on the fact that in a ΔbfmS strain, the number of T6SS is halved compared to a WT or ΔitrA strain. If capsule overproduction only partially limits the TslA-dependant T6SS assembly, how can this result be explained? Is it related to the degradation of Hcp in this background, which ultimately limits the formation of T6SS? If so, it would be interesting to mention this connection in the section "Prolonged secretion inhibition triggers Hcp degradation."

Referee Cross-Commenting

Overall, I agree with the concerns raised by reviewers 1 and 3. This (already) very good manuscript will undoubtedly benefit from these comments.

Significance

This work is highly intriguing as it not only delves into the specific mechanisms involved but also connects fundamental elements in bacterial competition, i.e., the necessity for self-protection and aggression for survival. The manuscript offers valuable insights into cellular dynamics at a microscale level and prompts new inquiries into the regulation of these systems on a population scale. The work is well-done and the writing is also clear.

I am convinced that this work represents another significant step towards understanding bacterial mechanisms and will undoubtedly spark considerable interest in the field.

Expertise: T6SS, fluorescence microscopy, predation, interbacterial competition

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 transport of effector proteins across membranes from the producing bacterium into a target cell is at the core of bacterial secretion systems. How an additional layer in form of a capsule affects effector export and the susceptibility towards effector import is not fully understood. Here, Flaugnatti and colleagues combined bacterial genetics with phenotypic assays and electron microscopy to demonstrate a dual role of a bacterial capsule in preventing T6SS-mediated effector export and promoting protection from effector import by another bacterium's T6SS. The wide variety of methods used, complementation of the mutants, and validation of the findings across strains strengthen the author's conclusions.

Although the main conclusions seem straight forward, the authors unravel the unexpected complexity underlying these phenotypes with strong mechanistic work. In brief, a capsule-deficient mutant (∆itra) is shown to assemble its T6SS similar to the WT, yet secretes more Hcp than the WT and is better in T6SS-mediated killing of other bacteria. A capsule-overproducing mutant (∆bfmS) shows both, a partial deficiency in T6SS assembly and an additional reduction in exported Hcp, and is worse in T6SS-mediated killing than the WT. A mutant with a capsule similar to WT and deficient in cell sensing (∆tslA) forms the least T6SS apparatuses and is yet better in T6SS-mediated killing than the overcapsulated mutant. Together, these data show an effect of the capsule on (i) T6SS apparatus assembly, (ii) effector export, (iii) effector import, and (iv) the need for clearance of accumulating non-secreted Hcp by ClpXP.

The work on a clinical isolate of Acinetobacter tumefaciens and the data on an impaired T6SS activity on other cells by antibiotic-induced capsulation is a strong demonstration of the work's clinical relevance in addition to the findings' conceptual novelty.

In my view, the manuscript is outstanding with very high quality of experimental data, very well written text and very clear presentation of the data in figures. A few minor comments and suggestions below that I think would strengthen the manuscript.

Major comment:

OPTIONAL: Fig. 4c/l. 320: Having an indirect effect of an antibiotic on T6SS activity by antibiotic-induced capsule formation is very intriguing and contributes to the clinical relevance of the overall findings. When I saw the data in Fig. 4c, the graph instantaneously reminded me of the panel in Fig. 2a, where a similar phenotype is observed by changing the predator:prey ratio in the absence of any antibiotic. The authors themselves comment on the possibility of antibiotic-induced, reduced predator growth (and thereby a change in predator:prey ratio) as a one factor impacting the phenotype here. I am wondering if this data could be strengthened or better disentangled to test more precisely if it is the antibiotic induced capsule formation per se that affects T6SS-mediated killing by A. baumanii in the presence of antibiotics. Would it help to take the bfmS mutant along as a control for direct comparison to see if antibiotic-induced capsule formation of the WT to similar levels of the mutant results in the same killing phenotype? Would it help to test for T6SS-mediated killing in the presence and absence of antibiotics at multiple predator:prey ratios? Could the effect of the antibiotic on A. baumanii growth be measured and considered when choosing the ratio at which the bacteria are mixed?

Minor comments:

• Figure 1D, l. 155ff, I might have missed this, do the authors happen to have the numbers of E. cloacae as well? This would strengthen the claim on A. baumanii survival because of E. cloacae is being killed.
• Figure 2, I suggest to write out the species name of the prey in the box with the ratio. With E. cloacae being referred to in the previous figure and starting with similar letters than E. coli, I wasn't sure at first sight what E. c. refers to.
• use of the term "T6SS activity" throughout the manuscript (e.g. l. 182, l. 187). I leave this up to the authors. To me, it seems like an umbrella term for the initial observation and I see that such a term can be very handy for the writing. I just would like to mention that the use of the term was not always intuitive to me and sometimes even a bit misleading. For example, l. 182 refers to "increased T6SS activity". As a reader, I only know about 'T6SS activity on other cells' or 'a T6SS-mediated effect on other cells' at this point. T6SS apparatus assembly/firing activity is tested for specifically later and it turns out to differ between mutants. By the time the term is used in the discussion, it captures multiple nuanced phenotypes described by then. The more precise definition of the term in l. 200 helped to capture what exactly is meant by the authors.
• l. 198f "Collectively, our findings indicate that CPS does not hinder the secretion process of 199 the T6SS or the consequent elimination of competing cells". I might be missing something, I cannot entirely follow this sentence. Didn't the authors just show that the CPS does hinder T6SS-mediated elimination of competing cells in panel 2A and less secreted Hcp in the encapsulated WT compared to the non-encapsulated mutant in panel 2B?
• l. 224, typo, "in"
• Two connected comments: l. 338, Just a thought, I am wondering about the title of the section. After reading it a second time, I think it is technically correct. When reading it first, I was a bit confused when getting to the data because apparatus assmebly is impaired in the capsule-overproducing strain and although "preserved", doesn't the data indicate that there is less T6SS assembly in the bfmS mutant and that this might be because of less cell sensing and isn't this a main point that there is a difference in apparatus assembly in the capsule overproducing strain compared to WT (other than no difference in apparatus assembly in the strain without capsule)? To me it seems not fully acknowledged as a finding in the interpretation of the data that less cells of the bfmS mutant have a T6SS apparatus. Isn't that interesting? A title along the lines of "Capsule-overproducing strain has preserved sensory function and assembles less T6SS apparatuses" would have been more intuitive for me. l. 352, In case I didn't miss a reference to this data earlier in the manuscript, I am wondering if it would be worth mentioning the finding on the reduced apparatus assembly of the bfmS mutant earlier, together with Figure 3 already. At least a sentence that mentions already that there is more coming later. When I got to this line in the manuscript and read the findings on the apparatus assembly, I first needed to go back to figure 3 and look at the data there again in light of this finding. It is mentioned here on the side but I think very important for the interpretation of the phenotypic data of the bfmS mutant shown earlier, isn't it? The tslA mutant is used beautifully here.
• Discussion: optional comment. On the one hand, I like the concise discussion. On the other hand, I see more potential here for bringing it all together (potentially at the expense of shortening some of the introduction). I think the subtleties of the findings are complex. For example, I could envision a graphical summary with a working model on all the effects of a capsule on the T6SS and its potential clinical relevance making the study accessible to even more readers.

Significance

General assessment

I consider the story very strong in terms of novelty, experimental approaches used, quality of the data, quality of the writing and figures of the manuscript. In my view, the aspects that could be improved are optional/minor and concern only one figure and some phrasing.

Advance

I see major advance in the findings (i, mechanistic) on the mechanism of how the capsule interferes with T6SS, (ii, fundamental) on the discovery of ClpXP degrading Hcp, and (iii, clinical) on the meaning of antibiotic treatment for the T6SS of this clinically relevant and often multi-drug resistant bacterial species, which strongly complements existing work on the T6SS and antibiotics in A. baumanii (e.g. of the Feldman group). As the authors write themselves, the starting points of the study of a capsule protecting from a T6SS and the effect of a T6SS on other cells being negatively impacted by a capsule were known, although not studied in one species and not understood mechanistically.

Audience

I see the result of interest to a broad audience in the fields of bacteria-bacteria interactions, Acinetobacter baumanii, type VI secretion, antimicrobial resistance, bacterial capsules

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 all three reviewers for their insightful comments. Based on this feedback, we have performed additional experiments, and revised our manuscript. Below, we address each comment and describe the revisions.

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

Summary: Ponomarova et al. showed that neomorphic idh-1 mutation results in increased levels of cellular D-2HG. The authors compared the high D-2HG phenotypes by D-2HG dehydrogenase mutant and identified vitamin B12 dependent vulnerability differences. The downregulated gene function of glycine cleavage system involved in one-carbon donor units exacerbates the phenotypes while adding one-carbone donors suppresses the phenotype. They concluded that the idh-1neo mutation imposes a dependency on the one-carbon pool. The manuscript is very interesting but I think the manuscript should be modified to be more clear for broad audiences.

Concerns: The authors mention a number of examples for metabolic changes of D-2HG in the first paragraph of introduction. I think that a metabolic map explaining the changes helps readers to understand the questions proposed by the authors.

Thank you for this suggestion. A figure illustrating the contributing factors in D-2HG metabolism has been added to the manuscript (Figure 1A).

The authors say that D-2HG affects carcinogenesis in many ways, citing previous works. They should say a higher concentration of D-2HG does affect carcinogenesis or not in dhgd loss of function, if they assume the concentration is most important for carcinogenesis.

Thank you for pointing this out. We have added this information in lines 70-72 of the revised manuscript: "Increased levels of D-2HG caused by the inhibition of D-2-hydroxyglutarate dehydrogenase activity have also been associated with different cancers (PMID: 29339485, PMID: 34296423, PMID: 35007759)."

Line 110, mode should be read as model, I guess.

Thank you - we have corrected this error.

In Figure 4C, concentrations of formate are shown; 0. 20, 40, 80, 160 mM. Is this correct? the high concentration of substrates changes the osmotic pressure of the medium. Also, high concentration of formic acid is toxic to animals. Considering the concentration of vitamin B12 was 64 nM, I wonder concentration unit of formate is also nM.

We confirm that we supplemented the media with formate in the millimolar range. The highest doses of supplemented formate somewhat slowed the development of P0 animals, but they consistently produced viable progeny. To clarify this we have added the following line to the text on lines 184-187: "The highest doses of supplemented formate somewhat slowed the development of P0 animals, but restored the survival of idh-1neo embryos to wild-type levels on a regular diet of E. coli OP50 as well as the diet of RNAi-competent E. coli HT115."

Additionally, the use of sodium formate ensured that the pH of the media remained unchanged.

I could not understand how embryonic and larval lethality confer the same mechanisms on animal carcinogenesis. Could you explain the logic link between lethal mutation and carcinogenesis. Or do the two phenotypes share only a part of metabolic changes?

Thank you for this suggestion. We have added this in lines 242-246 of the Discussion:

"While our results have focused on how the neomorphic idh-1 mutation affects the developing embryo, proliferating cancer cells also have been shown to have increased demand for 1C units, for instance, to synthesize nucleosides (33)(PMID: 24657017). Thus, we can speculate that cancers with mutated IDH1 may be increasingly sensitive to depletion of the 1C pool, also."

Vitamin B12 is an essential substance and deficiency in humans results in sever diseases. Is the lethal phenotype by treatment of idh-1neo mutants comparable to humans? Is the concentration of vitamin B12 similar in humans?

The daily dose of human vitamin B12 (cobalamin) in supplements can reach 12.5 µg per kg (PMID: 18606874), while we supplement the media fed to worms with approximately 55 µg cobalamin per kg (64 nM adenosylcobalamin). No known adverse effects are associated with excessive intake of vitamin B12 by healthy individuals; therefore, no tolerable upper intake level has been set (PMID: 23193625). However, the impact of vitamin B12 on patients with IDH1neo-positive cancers has not been studied.

Reviewer #1 (Significance (Required)):

I think that the manuscript is interesting and may lead an important progress of this field. However, in general, metabolic disorders are difficult to understand for the people outside the speciality. The authors should explain carefully the structure/property, pathways, enzyme functions, and concentration effects of substances of interest.

See above, we hope these edits are sufficient.

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

Increased levels of the metabolite D-2HG (derived from alpha-KG) are associated with multiple disorders. In a previous study, the authors showed that in C. elegans dhgd-1 deletion mutants, embryonic lethality resulting from the accumulation of D-2HG in is caused by a lack of ketone bodies. In this study, the authors generated a new model of D-2HG accumulation in C. elegans, idh-1neo, in order to further understand how D-2HG exerts its toxic effects in different contexts. This allele mimics mutations found in neomorphic mutations of human IDH1 that lead to abnormal D-2HG production from alpha-KG. Interestingly, the authors find that idh-1neo mutants are distinct from animals lacking the D-2HG dehydrogenase dhgd-1 previously reported. Specifically, while vitamin B12 rescues the embryonic lethality in dhgd-1 deletion animals, it enhances the lethality of idh-1neo animals. Through an elegant genetic screen, and complementation studies with specific metabolites, they provide compelling evidence that this vitamin B12-dependent enhancement is due to depletion of the 1C pool. Specifically, a reverse genetic screen revealed that inactivation of components of the 1 C-producing glycine cleavage system (GCS) results in embryonic lethality in idh-1neo, but not wildtype animals. Complementation studies with specific metabolites show that replenishing C groups is sufficient to reverse embryonic lethality.

This is a very clear, well written paper. Experiments are well controlled and executed, figures are of the highest quality and conclusions are convincing. Prior studies are appropriately referenced. No additional experiments are required by this reviewer.

Minor points 1) In Figure 2A could authors explain how beta-alanine (increased) is different from alanine (decreased). As a non-specialist this is not clear to me.

Thank you for pointing this out. We added this explanation to the figure legend (lines 510-512).

2) Did the authors test inactivation of the lipoamide dehydrogenase (dld-1) has the same effect as the other identified components of the GCS?

The dld-1 RNAi clone was present in the metabolic library that we screened but was not identified as a "hit." We have added the following in lines 164-168 of the revised manuscript: "Two other GCS genes, gcsh-2 and dld-1 were not identified as 'hits'. gcsh-2 is associated with the same reaction as gcsh-1, indicating that the latter encodes an active enzyme (30). dld-1 functions in other metabolic processes, particularly in lactate/pyruvate metabolism, and confers embryonic lethality when knocked down in wild type animals (31)".

**Referees cross-commenting**

Comments to Reviewer #3: 1/ The authors treat the idh-1neo worms with vitamin B12 to reduce 3HP concentrations. The authors should consider conducting experiments to reduce 3HP by other means also. This would help establish a causal relationship between the D-2HG accumulation and observed phenotypes.

The authors show that adding vitamin B12 to the diet of the idh-1neo significantly increased their D-2HG levels. Furthermore, dhgd-1 RNAi drives a further increase in D-2HG in idh-1neo animals and led to 100% penetrant embryonic lethality among the F1 generation of idh-1neo animals. Together I think this provided strong evidence for a causal relationship between the D-2HG accumulation and observed phenotypes. Further characterizing these phenotypes would be interesting but is beyond the scope of this paper.

4/ The authors should clarify whether it is really vitamin B12 or any other metabolite from the bacteria (like methionine) that is bringing about the phenotypes. Have they tested metabolically inactive bacteria?

the authors show that supplementing B12-treated idh-1neo animals with formate (another 1C donor) restored the survival of idh-1neo embryos, supporting a role for B12 in depletion of the 1C pool. They also show that suppressing Met/SAM cycle genes in idh-1neo prevent 1C depletion and restore availability of 1C units. So the evidence that 1C unit depletion is at the core of the observed phenotypes is pretty convincing

7/ The authors should conduct metabolomic profiling to examine changes in metabolic pathways, including 1C, glycine metabolism, glucose metabolism etc, in idh-1neo animals subjected to GCS gene knockdown, and vitamin B12 supplementation.

Not clear how these experiments would add to this story. Open up another line of research

8/ The audience will be limited to the field although the study pertains to an oncometabolite. The study value would have improved if the authors had included cancer cell data. Also, the phenotype studied has not been mechanistically linked to the oncometabolite function, making the study academic in nature.

The intetest of this study is that it is being carried out in an organismal context.

Reviewer #2 (Significance (Required)):

As a geneticist with a general interest in metabolomics I find this an elegans study that offers new insight into how IDH-1 and -2 neomorphic mutations affect metabolic rewiring in the context of a whole animal. Although similarities are observed between idh-1neo mutants and animals lacking the D-2HG dehydrogenase dhgd-1, both of which have increased levels of the metabolite D-2HG, specific metabolic differences are observed. The identification of 1C unit deficiency as a driver of lethality in idh-1neo mutants is highly significant given the central importance of 1C metabolism. This study should therefore be of interest to a wide audience.

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

Ponomarova et al presents a short follow up of their previous study to elucidate the role of a oncogenic variant of idh-1 that increases the 3HP levels, similar to the Ddhgd-1 mutant. Using a combination of metabolomics and genetics, they show that the defect in idh-1neo worms on high vitamin B12 diet is the draining of the 1C pool, distinct from the mechanisms of lethality observed in the Ddhgd-1 mutant. While the findings are interesting, there is a lack of mechanistic understanding of the basis of the phenotype observed. Moreover, the authors do not establish the link between the oncometabolite, that should support uncontrolled cell division, with the observed phenotype. Some control experiments are missing and should be included in the revised manuscript. there could be many other The comments on the manuscript are as follows, in no particular order:

1. The authors treat the idh-1neo worms with vitamin B12 to reduce 3HP concentrations. The authors should consider conducting experiments to reduce 3HP by other means also. This would help establish a causal relationship between the D-2HG accumulation and observed phenotypes.

To further examine the link between 3HP and idh-1neo embryonic lethality, we targeted hphd-1 by RNAi, which increases 3HP levels (Ponomarova et al., 2023). Hphd-1 knockdown did not induce lethality in the wild-type or exacerbate lethality in idh-1neo animals (Figure S3), further demonstrating that lack of 3HP degradation is not linked to this phenotype (lines 143-145).

Also, see cross-comments from Reviewer #2 above.

The authors should investigate the functional impact of HPHD-1 inhibition on 3-hydroxypropionate levels and D-2HG accumulation by RNAi knockdown of HPHD-1 in idh-1neo animals.

We have now performed the suggested experiment please see response to comment 1 above.

The authors do not clearly mention clearly which diet in some of their experiments. This is imporant since the two diets used (OP50 and HT115) differ in their vitamin B12 content, and thus could have different consequences.

We added this information in figures, figure legends, and lines 259-260 of the revised manuscript.

The authors should clarify whether it is really vitamin B12 or any other metabolite from the bacteria (like methionine) that is bringing about the phenotypes. Have they tested metabolically inactive bacteria?

The reviewer correctly points out that bacterial metabolism may play a role in the effects exerted by vitamin B12. We have not tested metabolically inactivated bacteria, however, our RNAi experiments (Figure 4E) demonstrate that supplemented vitamin B12 acts through the Met/SAM cycle in idh-1neo animals. Please also see cross-comments from Reviewer #2.

The authors consistently use 64 nM of Vitamin B12. Will the hphd-1 mutant and the idh-1neo mutant have different vitamin B12 thresholds for the observed phenotypes?

Thank you for raising this interesting point. While 64 nM vitamin B12 virtually eliminates 3HP accumulation in idh-1 animals (Figure 2D), we have not tested if this dose is sufficient to eliminate 3HP accumulation in hphd-1 mutant. However, potential differences in 3HP levels in idh-1neo and hphd-1 animals treated with vitamin B12 would not contradict our conclusion that 3HP is not the cause of embryonic lethality in idh-1neo mutant animals.

Figure 3b: HT115 has inherently high levels of vitamin B12 so the RNAi effect of genes should be seen on the OP50 diet supplemented with B12.

Despite reports of elevated B12 levels in E. coli HT115, vitamin B12-induced embryonic lethality of idh-1neo on a diet of OP50 is more severe than on a diet of HT115 bacteria (Figure 4C). Therefore, it may be harder to quantify synthetic lethal interaction of idh1-neo with GCS RNAi knockdown using OP50 strains (which would need to be created).

The authors should conduct metabolomic profiling to examine changes in metabolic pathways, including 1C, glycine metabolism, glucose metabolism etc, in idh-1neo animals subjected to GCS gene knockdown, and vitamin B12 supplementation.

While these results would be interesting and further our understanding of metabolic changes that occur in idh-1neo mutant animals we think they are beyond the scope of the manuscript. Also, please see cross-comments from Reviewer #2.

Perform rescue experiments using different one-carbon donors (e.g., formate, serine) to restore embryonic viability in idh-1neo mutants under conditions of vitamin B12-induced stress. Quantify the efficacy of these interventions using developmental assays.

In addition to formate rescue experiments (Figure 4C), we supplemented idh-1neo animals with serine (Figure 4D and S7). Similar to formate, serine supplementation resulted in the rescue of idh-1neo embryonic lethality on an E. coli OP50 diet (lines 187-189). The lack of rescue on an HT115 diet could be due to HT115 bacteria containing more glycine (Gao et al., 2017), which might limit the efficiency of serine conversion to glycine needed for 1C unit production.

Provide experimental evidence to show that idh-1neo animals possess an alternative source of energy.

We have previously found that diminished production of ketone bodies in ∆dhgd-1 mutants causes embryonic lethality that can be rescued by exogenous supplementation of ketone body 3-hydroxybutyrate (Ponomarova et al., 2023). In contrast to dhgd-1 mutants, idh-1neo embryonic lethality fails to respond to supplemented 3-hydroxybutyrate (Figure S4), indicating the lethality associated with the idh-1neo mutation is caused by a different mechanism, i.e., a depletion in 1C-units.

The authors use vitamin B12 to inhibit the shunt pathway (line 127). They should explore alternate strategies to do the same, like gene knockdown.

Please see our response to comment 1 above where we discuss RNAi knock-down of the shunt pathway gene, hphd-1.

It is not clear why the authors did not follow up with the other phenotypes of the idh-1neo that were visible without the Vitamin B12 supplementation. They should follow up with this and also other phenotypes to explore the broader physiological consequences of D-2HG accumulation.

We agree that the other physiological consequences of D-2HG accumulation are interesting, and we plan to investigate them in our future studies.

The authors should include control experiments without supplementation of vitamin B12, ketone bodies etc. in each of their figures.

We thank the reviewer for this suggestion. We have added these data (Figures S5, 6, 7, and 8).

The authors posit that the idh-1neo depletes the 1C pool leading to the observed lethality. So, when they supply formate to replenish it, they rescue the lethality of the B12-treated worms. Similar results are obtained by knocking down the enzymes. So where are the 1C units going? Understanding this will provide the much-needed mechanistic understanding to this study.

We appreciate this insightful comment and expand our discussion to elaborate on this issue (lines 224-227). "We propose that a lack of 1C units in idh-1neo can impede pyrimidine biosynthesis via thymidylate synthase tyms-1, which uses 1C units to generate dTMP. Supporting this hypothesis, RNAi of tyms-1 causes embryonic lethality (36-38)."

It may be important to measure the D-2HG levels in the mitochondria vs the cytosol.

While this is an interesting point, we think that this line of inquiry is beyond the scope of this work (and is technically challenging).

The idh-1neo is an oncometabolite. The authors do not show any data to indicate whether this mutant has any defect in cell division/cell cycle in the somatic tissue or germline.

In this study we primarily focused on the molecular changes in the metabolic network that occur in idh-1neo mutant animals, which we think is an important advance in understanding the basis for how this mutation affects IDH function. Additional phenotypic outcomes of these perturbed metabolic processes will be the basis of future studies.

Reviewer #3 (Significance (Required)):

The audience will be limited to the field although the study pertains to an oncometabolite. The study value would have improved if the authors had included cancer cell data. Also, the phenotype studied has not been mechanistically linked to the oncometabolite function, making the study academic in nature.

While we agree that the link between idh-1neo, 2HG production and oncometabolite function has not been directly shown we think that our study adds important molecular understanding of metabolic changes that occur in relation to idh-1neo function which are important for future studies of how this mutation affects carcinogenesis. Also, please see cross-comments from Reviewer #2.

In addition, we specified statistical significance in Figure 2, described statistical tests used (lines 361-363) and corrected a few grammatical errors throughout the text.

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

Ponomarova et al presents a short follow up of their previous study to elucidate the role of a oncogenic variant of idh-1 that increases the 3HP levels, similar to the dhgd-1 mutant. Using a combination of metabolomics and genetics, they show that the defect in idh-1neo worms on high vitamin B12 diet is the draining of the 1C pool, distinct from the mechanisms of lethality observed in the dhgd-1 mutant. While the findings are interesting, there is a lack of mechanistic understanding of the basis of the phenotype observed. Moreover, the authors do not establish the link between the oncometabolite, that should support uncontrolled cell division, with the observed phenotype. Some control experiments are missing and should be included in the revised manuscript. there could be many other The comments on the manuscript are as follows, in no particular order:

1. The authors treat the idh-1neo worms with vitamin B12 to reduce 3HP concentrations. The authors should consider conducting experiments to reduce 3HP by other means also. This would help establish a causal relationship between the D-2HG accumulation and observed phenotypes.
2. The authors should investigate the functional impact of HPHD-1 inhibition on 3-hydroxypropionate levels and D-2HG accumulation by RNAi knockdown of HPHD-1 in idh-1neo animals.
3. The authors do not clearly mention clearly which diet in some of their experiments. This is imporant since the two diets used (OP50 and HT115) differ in their vitamin B12 content, and thus could have different consequences.
4. The authors should clarify whether it is really vitamin B12 or any other metabolite from the bacteria (like methionine) that is bringing about the phenotypes. Have they tested metabolically inactive bacteria?
5. The authors consistently use 64 nM of Vitamin B12. Will the hphd-1 mutant and the idh-1neo mutant have different vitamin B12 thresholds for the observed phenotypes?
6. Figure 3b: HT115 has inherently high levels of vitamin B12 so the RNAi effect of genes should be seen on the OP50 diet supplemented with B12.
7. The authors should conduct metabolomic profiling to examine changes in metabolic pathways, including 1C, glycine metabolism, glucose metabolism etc, in idh-1neo animals subjected to GCS gene knockdown, and vitamin B12 supplementation.
8. Perform rescue experiments using different one-carbon donors (e.g., formate, serine) to restore embryonic viability in idh-1neo mutants under conditions of vitamin B12-induced stress. Quantify the efficacy of these interventions using developmental assays.
9. Provide experimental evidence to show that idh-1neo animals possess an alternative source of energy.
10. The authors use vitamin B12 to inhibit the shunt pathway (line 127). They should explore alternate strategies to do the same, like gene knockdown.
11. It is not clear why the authors did not follow up with the other phenotypes of the idh-1neo that were visible without the Vitamin B12 supplementation. They should follow up with this and also other phenotypes to explore the broader physiological consequences of D-2HG accumulation.
12. The authors should include control experiments without supplementation of vitamin B12, ketone bodies etc. in each of their figures.
13. The authors posit that the idh-1neo depletes the 1C pool leading to the observed lethality. So, when they supply formate to replenish it, they rescue the lethality of the B12-treated worms. Similar results are obtained by knocking down the enzymes. So where are the 1C units going? Understanding this will provide the much-needed mechanistic understanding to this study.
14. It may be important to measure the D-2HG levels in the mitochondria vs the cytosol.
15. The idh-1neo is an oncometabolite. The authors do not show any data to indicate whether this mutant has any defect in cell division/cell cycle in the somatic tissue or germline.

Significance

The audience will be limited to the field although the study pertains to an oncometabolite. The study value would have improved if the authors had included cancer cell data. Also, the phenotype studied has not been mechanistically linked to the oncometabolite function, making the study academic in nature.

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

Increased levels of the metabolite D-2HG (derived from alpha-KG) are associated with multiple disorders. In a previous study, the authors showed that in C. elegans dhgd-1 deletion mutants, embryonic lethality resulting from the accumulation of D-2HG in is caused by a lack of ketone bodies. In this study, the authors generated a new model of D-2HG accumulation in C. elegans, idh-1neo, in order to further understand how D-2HG exerts its toxic effects in different contexts. This allele mimics mutations found in neomorphic mutations of human IDH1 that lead to abnormal D-2HG production from alpha-KG. Interestingly, the authors find that idh-1neo mutants are distinct from animals lacking the D-2HG dehydrogenase dhgd-1 previously reported. Specifically, while vitamin B12 rescues the embryonic lethality in dhgd-1 deletion animals, it enhances the lethality of idh-1neo animals. Through an elegant genetic screen, and complementation studies with specific metabolites, they provide compelling evidence that this vitamin B12-dependent enhancement is due to depletion of the 1C pool. Specifically, a reverse genetic screen revealed that inactivation of components of the 1 C-producing glycine cleavage system (GCS) results in embryonic lethality in idh-1neo, but not wildtype animals. Complementation studies with specific metabolites show that replenishing C groups is sufficient to reverse embryonic lethality.

This is a very clear, well written paper. Experiments are well controlled and executed, figures are of the highest quality and conclusions are convincing. Prior studies are appropriately referenced. No additional experiments are required by this reviewer.

Minor points

1. In Figure 2A could authors explain how beta-alanine (increased) is different from alanine (decreased). As a non-specialist this is not clear to me.
2. Did the authors test inactivation of the lipoamide dehydrogenase (dld-1) has the same effect as the other identified components of the GCS?

Referees cross-commenting

Comments to Reviewer #3:

1/ The authors treat the idh-1neo worms with vitamin B12 to reduce 3HP concentrations. The authors should consider conducting experiments to reduce 3HP by other means also. This would help establish a causal relationship between the D-2HG accumulation and observed phenotypes.

The authors show that adding vitamin B12 to the diet of the idh-1neo significantly increased their D-2HG levels. Furthermore, dhgd-1 RNAi drives a further increase in D-2HG in idh-1neo animals and led to 100% penetrant embryonic lethality among the F1 generation of idh-1neo animals. Together I think this provided strong evidence for a causal relationship between the D-2HG accumulation and observed phenotypes. Further characterizing these phenotypes would be interesting but is beyond the scope of this paper.

4/ The authors should clarify whether it is really vitamin B12 or any other metabolite from the bacteria (like methionine) that is bringing about the phenotypes. Have they tested metabolically inactive bacteria?

the authors show that supplementing B12-treated idh-1neo animals with formate (another 1C donor) restored the survival of idh-1neo embryos, supporting a role for B12 in depletion of the 1C pool. They also show that suppressing Met/SAM cycle genes in idh-1neo prevent 1C depletion and restore availability of 1C units. So the evidence that 1C unit depletion is at the core of the observed phenotypes is pretty convincing

7/ The authors should conduct metabolomic profiling to examine changes in metabolic pathways, including 1C, glycine metabolism, glucose metabolism etc, in idh-1neo animals subjected to GCS gene knockdown, and vitamin B12 supplementation.

Not clear how these experiments would add to this story. Open up another line of research

8/ The audience will be limited to the field although the study pertains to an oncometabolite. The study value would have improved if the authors had included cancer cell data. Also, the phenotype studied has not been mechanistically linked to the oncometabolite function, making the study academic in nature.

The intetest of this study is that it is being carried out in an organismal context.

Significance

As a geneticist with a general interest in metabolomics I find this an elegans study that offers new insight into how IDH-1 and -2 neomorphic mutations affect metabolic rewiring in the context of a whole animal. Although similarities are observed between idh-1neo mutants and animals lacking the D-2HG dehydrogenase dhgd-1, both of which have increased levels of the metabolite D-2HG, specific metabolic differences are observed. The identification of 1C unit deficiency as a driver of lethality in idh-1neo mutants is highly significant given the central importance of 1C metabolism. This study should therefore be of interest to a wide 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 #1

Evidence, reproducibility and clarity

Summary:

Ponomarova et al. showed that neomorphic idh-1 mutation results in increased levels of cellular D-2HG. The authors compared the high D-2HG phenotypes by D-2HG dehydrogenase mutant and identified vitamin B12 dependent vulnerability differences. The downregulated gene function of glycine cleavage system involved in one-carbon donor units exacerbates the phenotypes while adding one-carbone donors suppresses the phenotype. They concluded that the idh-1neo mutation imposes a dependency on the one-carbon pool. The manuscript is very interesting but I think the manuscript should be modified to be more clear for broad audiences.

Concerns:

The authors mention a number of examples for metabolic changes of D-2HG in the first paragraph of introduction. I think that a metabolic map explaining the changes helps readers to understand the questions proposed by the authors.

The authors say that D-2HG affects carcinogenesis in many ways, citing previous works. They should say a higher concentration of D-2HG does affect carcinogenesis or not in dhgd loss of function, if they assume the concentration is most important for carcinogenesis.

Line 110, mode should be read as model, I guess.

In Figure 4C, concentrations of formate are shown; 0. 20, 40, 80, 160 mM. Is this correct? the high concentration of substrates changes the osmotic pressure of the medium. Also, high concentration of formic acid is toxic to animals. Considering the concentration of vitamin B12 was 64 nM, I wonder concentration unit of formate is also nM.

I could not understand how embryonic and larval lethality confer the same mechanisms on animal carcinogenesis. Could you explain the logic link between lethal mutation and carcinogenesis. Or do the two phenotypes share only a part of metabolic changes?

Vitamin B12 is an essential substance and deficiency in humans results in sever diseases. Is the lethal phenotype by treatment of idh-1neo mutants comparable to humans? Is the concentration of vitamin B12 similar in humans?

Significance

I think that the manuscript is interesting and may lead an important progress of this field. However, in general, metabolic disorders are difficult to understand for the people outside the speciality. The authors should explain carefully the structure/property, pathways, enzyme functions, and concentration effects of substances of interest.

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)):_ _ __ In this manuscript, Jones et al. report on a potential role for fam83fa in zebrafish hatching, radiation response and autophagy. The authors are commended for generating multiple KO lines and maternal-zygotic embryos for analysis. However, important controls are lacking and the data is circumstantial throughout with very little mechanistic insight into the precise roles, if any, of fam83f in these processes.

We thank the reviewer for recognizing the strengths of our manuscript, and highlighting areas we might improve. Please see the specific comments below addressing the points raised. In respect of mechanistic insight, while we agree that our manuscript does not provide this, it was not intended to. Rather, we aim to communicate our descriptive findings on the role of Fam83fa in vivo, providing data for follow-up studies by other researchers into the mechanistic role of Fam83fa.

1. Validation of the KO phenotypes (hatching, IR sensitivity) requires rescue with WT fam83fa WT mRNA, but not 1-500 or fam83fb mRNA.

We thank the reviewer for raising the issue of rescue experiments. Such experiments are frequently used in knock-down experiments, where non-specificity may be a problem, but they are used more rarely in genetic knock-outs, where the gene defect is well defined. In the case of Fam83fa, a particular difficulty is that overexpression of fam83fa itself causes a p53-mediated DNA damage response (DDR) (Salama et al., 2019). Moreover, we have shown by both qRT-PCR and western blotting that injection of fam83fa mRNA into zebrafish embryos (the traditional technique by which rescue experiments are performed) induces a p53-mediated DDR. As a result, it would be very difficult to interpret the results of any rescue experiment, because one would have to be absolutely certain that levels of fam83fa re-expression recapitulate and do not exceed endogenous levels. As a tool for specificity, we therefore used more than one fam83fa-/- mutant line, carrying a different genomic mutation, and validated that the same phenotype was present in both. We are happy to provide the qRT-PCR and western blot data confirming the results of fam83fa mRNA injection, if required. We have included an additional section into the manuscript detailing this issue. 2.

While the hatching phenotype (Fig 3) is convincing, there is no data on HG development in the null embryos. Does the HG develop normally in the absence of fam83fb? If so, this would support the authors conclusions that the role of fam83fb is functional rather than developmental (indirect effect). In situs as in Fig.1 might be helpful here.

Thank you to the reviewer for this helpful suggestion. We agree that we did not investigate whether the hatching gland develops normally in the MZ-fam83fa-/- mutant embryos. No gross morphological differences were observed that led us to investigate this, although we agree it is an interesting question for a future project. In terms of functional vs developmental effects, we are confident that MZ-fam83fa-/- mutant embryos develop at a normal temporal rate, as evidenced by the machine learning based classifier used to assess temporal developmental trajectory (Figure S3 and Jones et al., 2022, 2024). This strongly suggests that the effect of fam83fa KO is functional rather than indirect and caused by (for example) developmental delay.

While the IR sensitivity phenotype (Fig S4) is convincing, IR-induced cell death/apoptosis was not analyzed. There is a large literature describing straightforward assays for cell death/apoptosis detection in zebrafish with assays such as acridine orange or TUNEL labeling, or active casp3 whole-mount IF. Is IR-induced cell death enhanced in fam83fa KOs?

We thank the reviewer for their positive comments and agree that investigating the nature of the cell death occurring following IR would be very interesting. We did make use of both acridine orange and TUNEL labeling following injection of fam83fa mRNA (see 1 above), and whilst the assays themselves were relatively straightforward, due to technical issues the quantification of fluorescence intensity was not. Similarly, we suspect that a significant degree of necrosis is also occurring, which further complicates the issue of data interpretation from both these approaches. We do, however, think this is an important avenue of questioning, and hope that other researchers will explore the mechanism of IR induced cell death in the MZ-fam83fa-/- mutants in the future,

Similarly, there are multiple tools to assay autophagy in zebrafish (e.g., Moss et al., Histochem Cell Biol 2020, PMC7609422; Mathai et al., Cells 2017, PMC5617967). Is autophagy affected in the KOs, with or without IR? These experiments might directly implicate fam83fa in autophagy.

We agree that there are exciting tools with which to assay autophagy in zebrafish, and although we considered some of these, including caudal fin regeneration, we deemed these experiments to be beyond the descriptive scope of this paper, given the time and resources available to us. We hope that other researchers will use our data as a basis for investigating the role of Fam83fa in autophagy further, using assays such as these suggested by the reviewer.

Figure 4: Isn't there a slight reduction in p53 induction at 10 hours?

Although the western blot in Figure 4A gives this impression, this is probably due to loading variability (see the anti-β-actin loading control band). Moreover, over three independent experiments (Figure 4B), this apparent difference is not statistically significant. Taken together with other evidence that the p53-mediated DNA damage response is not affected in MZ-fam83fa-/- mutants, we are confident there is no detectable change in the level of stabilized p53 in the MZ-fam83fa-/- mutants compared to WT.

Given the widely documented, dominant role of p53 in zebrafish IR-sensitivity, the authors should test if the IR sensitivity of fam83fa KO animals is p53-dependent, ideally via a cross into p53 null, but at least via injection of p53 morpholinos.

We agree that p53 is widely documented as playing an essential role in the IR induced DNA damage response in zebrafish. All our experiments suggest there is no difference between the levels of p53 (protein or mRNA) or any of the p53-induced downstream effectors (that we tested) in MZ-fam83fa-/- mutants compared to WT embryos. This was true whether or not the embryos were subjected to genotoxic stressors, including IR treatment. We therefore conclude that the increased sensitivity phenotype we observe as a result of loss of Fam83fa is not caused by a change in p53 activity, at least not as part of the DNA damage response.

Do autophagy inhibitors phenocopy the hatching and IR-sensitivity defects of fam83fa embryos? Do the inhibitors exacerbate the mutant phenotypes or synergize with M or Z mutant phenotypes? (I may have missed this but do M and Z fam83fa null embryos have any phenotype? Or do the phenotypes only manifest in MZ embryos?)

This is an excellent question, and indeed one we attempted to address. We tried to optimize several autophagy inhibitors including bafilomycin A1, chloroquine and wortmannin, as well as the proteasomal inhibitor MG132. In addition, we tried to optimize the autophagy promoters Torin1 and rapamycin. Unfortunately, we regularly saw global effects in zebrafish embryos that were difficult to characterize and control by dosage. At the same time, we were also working to confirm the specific effects of these drugs on autophagy using p62 and LC3-I and LC3-II western blots, which themselves were difficult to optimize. We attempted to optimize these experiments for 6 months before the COVID lockdown occurred, at which point they were abandoned. We would be delighted for future researchers to continue these experiments, as we are now unable to pursue this further due to closure of the Smith lab, but we agree that these are very pertinent questions. We hope the descriptive data provided in our paper will prompt other researchers in the autophagy field to further explore the role of Fam83fa in autophagy. In response to the zygotic phenotype question, this was something we did not investigate. As there was no immediately apparent phenotype in the zygotic generation, for ease of screening larger numbers of embryos we proceeded immediately to the maternal-zygotic (MZ) generation.

Reviewer #1 (Significance (Required)):

The role of Fam83f is not known. This study in zebrafish might be the first to clarify the function of this protein in vivo.

We thank the reviewer for this positive insight, and we agree that our work is the first do so in vivo.

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

Fam83f is one of the proteins about which little is known. The authors Jones et al., tried to shed light on Fam83f function by knocking out the gene in zebrafish. Here they found that fam83 is expressed in the hatching gland and that larvae without Fam83f hatch significantly earlier than wild-type animals. The authors furthermore investigated the response of fam83f knock-out animals to DNA damage and found increased sensitivity to ionizing radiation and MMS. In order to find out more about Fam83f function in the DNA damage response, the authors performed RNA-seq after employing DNA damage and here they saw upregulation of several autophagy/lysosome-associated proteins and downregulation of some phosphatidylinositol-3-phosphate binding proteins, among others. Finally, the authors found that Fam83f is targeted to the lysosome. The manuscript is overall well written and clear in its general statement.

We thank the reviewer for their encouraging comments.

In the manuscript, the authors describe the investigation of several aspects of Fam83f function and particularly the role in hatching seems to be important for Fam83f as the gene is strongly expressed in the hatching gland and its absence leads to a clear and considerable earlier hatching. Unfortunately, all aspects of Fam83f function that are described in the manuscript are investigated very superficially, the conclusions are not supported by data and important controls are lacking. As such, the RNA-seq results are not confirmed by qRT-PCR, the role of the Fam83f LIR domain is not confirmed by co-IPs and it has not been investigated whether the presence of Fam83f in lysosomes is due to its degradation or whether it has a function in this cellular compartment.

We thank the reviewer for their input and will address each point raised below: -

• All aspects of Fam83f function are investigated superficially.

We agree that we have not provided an in-depth analysis of the mechanistic role of Fam83fa. It was because there were so many roles that we decided to make this paper rather descriptive in nature, hoping that the observations will prove useful to other researchers who may wish to define the mechanistic roles of Fam83fa more deeply. Even without in-depth investigation, our findings are previously unreported and the phenotypes we report are clear. We have amended our manuscript to make it apparent that this paper is intended to be descriptive in nature, and we hope this addresses this issue.

• Important controls are lacking - RNA-seq results are not confirmed by qRT-PCR

We thank the reviewer for their comment. We did not include qRT-PCR data as a control for the RNA-seq data because 1) each RNA-seq experiment was repeated on three biological replicates across three independent experiments and 2) we conducted RNA-seq on two different MZ-fam83fa-/- mutant lines and only considered genes that were mis-regulated in both mutants. Taken together, we considered this to be sufficient validation for the manuscript. However, we also performed confirmatory qRT-PCR for several of the differentially expressed genes identified, including the three main PI(3)P binding genes. We have now included these data in the supplementary information as an additional control - see Figure S6G which is now also referred to in the main text, and additional primer sequences have been added to Table S1.

• The role of the Fam83f LIR domain is not confirmed by co-Ips

We agree with the reviewer that this is an important experiment, and we worked closely with Dr Brian Ludwig and Dr Karen Vousden (The Francis Crick Institute) to test this. We tried to express zebrafish Atg8 and Gabarap (the two main ATG8 proteins that bind to LIR domains) but were unable to express sufficient levels of protein to perform the co-Ips. The text in the manuscript has now been amended to reflect that this experiment is required to confirm the role of the putative LIR domain in Fam83fa.

• *it has not been investigated whether the presence of Fam83f in lysosomes is due to its degradation or whether it has a function in this cellular compartment *

Whilst we agree with the reviewer that this is an important question, we did not intend this paper to expand beyond a descriptive role of the observations we made following the loss of Fam83fa in vivo. These are important questions to follow up on to determine the mechanism of action of Fam83fa, and we hope that other researchers will pursue these avenues of investigation following the publication of our observations.

Also, there is no leading concept in the manuscript. Starting from a role in hatching, the authors go to the DNA damage response and finally to the presence of Fam83f in lysosomes. How are these different aspects linked? Is the presence of Fam83f in lysosomes important for the suppression of hatching and how does Fam83f delays this process? (One would have wished that the authors would not have been that broad and were more focused on a particular aspect which then could have been investigated in depth.)

We agree with the reviewer that the paper gives a broad overview of our observations and does not examine the underlying mechanisms in detail. However, we believe that descriptive papers such as this, where observations following genetic perturbation are reported, are equally important, providing as they do important foundational data for other researchers to take forward. We do postulate on the links between the hatching, DNA damage and lysosomal phenotypes we observe in the discussion section, and we have expanded on this following the reviewers' comments, to make our hypothesized link between these phenomena clearer.

Specific comments: - All materials should be described in material and methods including the antibodies that have been used

The antibodies used together with concentrations and catalog numbers are now in Materials and Methods

• Abbreviations should be explained

The manuscript has been revised to ensure all abbreviations are explained. We thank the reviewer for bringing this oversight to our attention.

• Figure 4A: Levels of p53 should also be shown for untreated fam83f -/-KO1 and KO2 animals

The authors thank the reviewers for raising this point. Extracts from untreated MZ-fam83fa-/- KO1 and KO2 embryos were not included on this particular blot, as p53 was observed to be undetectable in all embryos, across all our experiments (WT and both mutants) unless genotoxic stress was applied. No quantification could therefore be performed as the expression level was essentially zero. However, we have now included an example p53 western blot in Supplemental Figure 5A, which shows WT, MZ-fam83fa-/- KO1 and MZ-fam83fa-/- KO2 untreated blots for p53 (all undetectable) alongside treated embryos (detected).

• Some references are missing (e.g. page 17, lane 320/321: As this group of cells arises....)

This citation and reference have now been added; thank you to the reviewer for highlighting this omission.

• Lane 369: The authors write about 4 KO lines but only two are shown in the figure.

We thank the reviewer for this observation. In Figure 2B only KO1 and KO2 schematic diagrams are shown for simplicity (as these are the lines taken forward for further investigation). We have now amended the manuscript text to make this clear.

• Lane 374/375: The NMD is not proven

Absolutely - we have now revised the text to change this sentence accordingly and thank the reviewer for noting this.

• Lane 380: how can RNA levels of fam83fa be upregulated when the gene has been knocked out? Why are these genes only upregulated in KO1? How relevant is this?

This was a typographical error, and we are very grateful to the reviewer for picking up on this. It should have read 'fam83fb'. As nonsense-mediated decay and associated transcriptional adaptation have been previously reported in zebrafish, this finding may be of considerable interest to the community. It is a side observation, and not necessarily directly related to the role of Fam83fa in vivo, but we felt it important to include. Indeed, as a result of this observation we have recently shared our MZ-fam83fa-/- lines with another group who are planning to investigate precisely this question - why are fam83fb and fam83g only upregulated in KO1?

• Figure 3C is not mentioned in the text and lacks any labelling

Figure 3C is now clearly referred to in the text and a label added to the figure.

• Lane 434/435: all relevant data should be shown (can be done as supplementary figure)

We have now amended this to include an additional supplemental figure (Figure S5A).

• Lane 434: The reference to the figure seems to be incorrect (5A4A)

Amended accordingly - thank you for pointing out this mistake.

• Figure 4C and 4D: what is the difference?

Thank you to the reviewer for noticing this omission. These data are from t1 (+2hrs) and t2 (+10hrs) and have now been labelled accordingly.

• S5C and S5D: why are there 3 clusters?

We thank the reviewer for raising this as it has provided us with an opportunity to present our data more clearly. There are 3 clusters that represent the combination of the two first principal components, which are time and treatment. Therefore, the clusters represent i) untreated at t1, ii) treated at t1 and iii) treated at t2. However, having two plots with different color schemes made this confusing/misleading. We have now replaced the two PCA plots with one that is colored and labelled accordingly with the 3 aforementioned clusters.

• Lane 495 to 505: What does this mean that the GO analysis shows upregulation and downregulation of endopeptidases and why "in contrast"?

We thank the reviewer for this comment, and we agree that this paragraph was misleading/confusing. This has now been rewritten in the main text, clarifying that endopeptidases were consistently upregulated at both timepoints.

Reviewer #2 (Significance (Required)):

The strength of the manuscript is certainly that it provides inside into Fam83f function as there is not much known about Fam83f.

We thank the reviewer for the positive comment, and we agree that very little is known about this highly conserved protein.

These study is probably most interesting for people in the zebrafish and related fields as the authors convincingly show the expression of Fam83f in the hatching gland and also the earlier hatching in the absence of the protein is very clear.

Thank you for the positive feedback.

The weakness of the study is clearly that it does not provide an in-depth analysis. As such, it shows that Fam83f is involved in hatching and can delay the process but it remains elusive how this is achieved. (Likwise, also the investigation into the DNA damage response remains very superficial and does not prove a specific role for Fam83f in the DNA damage response or whether the increased sensitivity is more unspecifically caused by the absence of a gene or eventually even connected to the earlier hatching.

Please refer to responses above (and changes made to the manuscript) clarifying that this study is intended to be descriptive, and provides important foundational data for further in-depth mechanistic studies by other researchers interested in the role of Fam83fa in vivo.

__Reviewer #3 (Evidence, reproducibility and clarity (Required)):_ _ __ In their manuscript "Zebrafish reveal new roles for Fam83f in hatching and the DNA damage-mediated autophagic response", Jones et al. provide an interesting exploration for the function of a poorly studied protein, Fam83f in embryonic development. Using the zebrafish as a model organism, the study combines loss-of-function genetics, phenotypic analysis and RNA-sequencing to characterize and explore the result of Fam83f loss. Upon critical review of the manuscript and the results we offer suggestions to improve the manuscript (see 'minor technical issues'). Additionally, we would like to highlight a weakness of the study in making the connection between Fam83f to the observed phenotype (increased sensitivity to DNA damage), see 'major issues'.

Major issues:

Most of our concern stems from relatively incomplete connection of the loss of fam83f to increased sensitivity to DNA-damage and lysosome function.

Please refer to comments above and changes made to the manuscript to clarify this is a descriptive paper that is not intended to provide in-depth mechanistic insight into the role of Fam83fa.

Is the increased sensitivity in fam83f KO embryos a direct effect to fam83f loss? A rescue experiment (by introduction of Fam83fa mRNA into their KO2 fish line) in the presence of ionizing radiation would help us understand the functional role of this protein in this process. Furthermore, can overexpression of any of the down-regulated genes involved in lysosome function restore the early hatching phenotype or the sensitivity to DNA damage? Fam83fa rescue experiments would be very difficult to interpret - please see comments above and the corresponding changes to our manuscript.

In terms of over-expressing some of the downregulated genes identified in the RNA-seq and qRT-PCR to see if the phenotype can be rescued, we feel these are excellent suggestions and we hope other researchers in future will attempt such experiments.

Minor technical issues:

-Methods line 203, clarify how many embryos were used per sample for RNA-seq (this was only described as 15 embryos in the main body results text).

Text has been amended to clarify this. We thank the reviewer for noticing this oversight.

-Comment about the expansion of fam83f orthologs in mammals (8) as opposed to only 2 in zebrafish

We apologize for any confusion: mammals do not have 8 fam83f orthologs. Mammals and zebrafish have 8 FAM83 genes (FAM83A-FAM83H). Zebrafish, unlike mammals, have genome duplication and although mammals have only one FAM83F gene, zebrafish have two: Fam83fa and Fam83fb. We trust this clarifies this issue and believe this to be clear in our main text. However, we are happy to make any suggested amendments should the reviewer consider our wording confusing.

-Supplementary figure 1C: please include representative images of secondary axis formation in fam83fa overexpressed Xenopus embryos.

We have not included any images as these are already published in our related paper on FAM83F (Dunbar et al., 2020) which we refer to in the figure legend text. No additional images were captured specifically for this publication.

-Provide more information about the mis-regulated genes in the RNA-seq analysis, how many are up or down regulated? Perhaps a better plot than a Venn diagram can be an MA-plot with the Venn diagram moved to a supplementary figure.

The Venn diagrams in Figure 5A-C are to illustrate the number of differentially expressed genes that are shared between KO1 and KO2 (whether up or down regulated), and only those that are common to both lines are taken forward. Following the reviewer's comments, we have now displayed the behavior of the common genes across all replicates in one heatmap, with the data normalized to the WT untreated samples, and the normalized variance stabilized count indicates whether a gene is up or down regulated across each of the replicates and conditions. We believe this addresses the reviewer's comment as these data are now displayed in a more direct way and the genes that are consistently up or downregulated across all replicates (and indeed those that are not) can be clearly seen. We thank the reviewer for raising this and improving our data representation.

-A better comparison of mis-regulated genes in the fam83f knockouts would be a comparison of KO2 and perhaps KO3, as the compensatory effects in KO1 can lead to additional indirect effect on the transcriptome. We understand the time and cost involved in this experiment and suggest that the differential gene expression analysis be performed individually on up or down regulated genes from KO2, or a comparison of such analysis will be provided with the differential gene expression analysis that was performed on shared mis-regulated genes between KO1 and KO2.

The reviewer raises an excellent point. At the time of experimental design, we were concerned that omitting KO1 in favor of another line (e.g. KO3) would bias our results by excluding potentially important data. Similarly, as transcriptional adaptation occurs in a sequence specific manner, and the phenotype was present in KO1 regardless, we didn't want to exclude these data. However, with hindsight, we agree that it may have been prudent to exclude KO1 on this basis, and we may have seen an increased concordance of differentially expressed genes (DEGs) between KO2 and KO3. However, this is not possible to repeat now due to the Smith lab closing, and our documented findings are valid and important regardless. We acknowledge however that, with hindsight, what the reviewer suggests may have been better experimental design.

-Can you confirm with the RNA-seq analysis that fam83g is upregulated in KO1 as opposed to KO2? (i.e. can the compensatory analysis you have observed with qRT-PCR be confirmed with the RNA-seq data?)

This is an excellent question, and we thank the reviewer for raising this. fam83fb passed our threshold for significance to be deemed as differentially expressed (upregulated) in KO1 only, in accordance with our qRT-PCR data. fam83g did not pass the significance threshold, but perhaps this is not surprising as both fam83fb and fam83g are expressed at particularly low levels to start with and would probably require much greater sequencing depth to be detected.

Reviewer #3 (Significance (Required)):

There is fundamental value in clarifying the in vivo function of poorly characterized protein-coding genes. This study fills a gap in the literature, but the broader conceptual impact is limited. The authors do a thorough job at generating and characterizing CRISPR/Cas9 mediated knock-out zebrafish animals. It is further commended that the authors do a meticulous job in a quantitative description of the resulting phenotype. This is a thorough study, with the only major concern being the lack of rescue experiments that would be needed to substantiate the the role of fam83f in sensitivity to DNA damage and lysosome function.

We thank the reviewer for their comments and trust we have addressed the issues concerned with the changes described above.

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

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

In their manuscript "Zebrafish reveal new roles for Fam83f in hatching and the DNA damage-mediated autophagic response", Jones et al. provide an interesting exploration for the function of a poorly studied protein, Fam83f in embryonic development. Using the zebrafish as a model organism, the study combines loss-of-function genetics, phenotypic analysis and RNA-sequencing to characterize and explore the result of Fam83f loss. Upon critical review of the manuscript and the results we offer suggestions to improve the manuscript (see 'minor technical issues'). Additionally, we would like to highlight a weakness of the study in making the connection between Fam83f to the observed phenotype (increased sensitivity to DNA damage), see 'major issues'.

Major issues:

Most of our concern stems from relatively incomplete connection of the loss of fam83f to increased sensitivity to DNA-damage and lysosome function.

Is the increased sensitivity in fam83f KO embryos a direct effect to fam83f loss? A rescue experiment (by introduction of Fam83fa mRNA into their KO2 fish line) in the presence of ionizing radiation would help us understand the functional role of this protein in this process. Furthermore, can overexpression of any of the down-regulated genes involved in lysosome function restore the early hatching phenotype or the sensitivity to DNA damage?

Minor technical issues:

• Methods line 203, clarify how many embryos were used per sample for RNA-seq (this was only described as 15 embryos in the main body results text).
• Comment about the expansion of fam83f orthologs in mammals (8) as opposed to only 2 in zebrafish
• Supplementary figure 1C: please include representative images of secondary axis formation in fam83fa overexpressed Xenopus embryos.
• Provide more information about the mis-regulated genes in the RNA-seq analysis, how many are up or down regulated? Perhaps a better plot than a Venn diagram can be an MA-plot with the Venn diagram moved to a supplementary figure.
• A better comparison of mis-regulated genes in the fam83f knockouts would be a comparison of KO2 and perhaps KO3, as the compensatory effects in KO1 can lead to additional indirect effect on the transcriptome. We understand the time and cost involved in this experiment and suggest that the differential gene expression analysis be performed individually on up or down regulated genes from KO2, or a comparison of such analysis will be provided with the differential gene expression analysis that was performed on shared mis-regulated genes between KO1 and KO2.
• Can you confirm with the RNA-seq analysis that fam83g is upregulated in KO1 as opposed to KO2? (i.e. can the compensatory analysis you have observed with qRT-PCR be confirmed with the RNA-seq data?)

Significance

There is fundamental value in clarifying the in vivo function of poorly characterized protein-coding genes. This study fills a gap in the literature, but the broader conceptual impact is limited. The authors do a thorough job at generating and characterizing CRISPR/Cas9 mediated knock-out zebrafish animals. It is further commended that the authors do a meticulous job in a quantitative description of the resulting phenotype. This is a thorough study, with the only major concern being the lack of rescue experiments that would be needed to substantiate the the role of fam83f in sensitivity to DNA damage and lysosome 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

Fam83f is one of the proteins about which little is known. The authors Jones et al., tried to shed light on Fam83f function by knocking out the gene in zebrafish. Here they found that fam83 is expressed in the hatching gland and that larvae without Fam83f hatch significantly earlier than wild-type animals. The authors furthermore investigated the response of fam83f knock-out animals to DNA damage and found increased sensitivity to ionizing radiation and MMS. In order to find out more about Fam83f function in the DNA damage response, the authors performed RNA-seq after employing DNA damage and here they saw upregulation of several autophagy/lysosome-associated proteins and downregulation of some phosphatidylinositol-3-phosphate binding proteins, among others. Finally, the authors found that Fam83f is targeted to the lysosome. The manuscript is overall well written and clear in its general statement. In the manuscript, the authors describe the investigation of several aspects of Fam83f function and particularly the role in hatching seems to be important for Fam83f as the gene is strongly expressed in the hatching gland and its absence leads to a clear and considerable earlier hatching. Unfortunately, all aspects of Fam83f function that are described in the manuscript are investigated very superficially, the conclusions are not supported by data and important controls are lacking. As such, the RNA-seq results are not confirmed by qRT-PCR, the role of the Fam83f LIR domain is not confirmed by co-IPs and it has not been investigated whether the presence of Fam83f in lysosomes is due to its degradation or whether it has a function in this cellular compartment. Also, there is no leading concept in the manuscript. Starting from a role in hatching, the authors go to the DNA damage response and finally to the presence of Fam83f in lysosomes. How are these different aspects linked? Is the presence of Fam83f in lysosomes important for the suppression of hatching and how does Fam83f delays this process? (One would have wished that the authors would not have been that broad and were more focused on a particular aspect which then could have been investigated in depth.)

Specific comments:

• All materials should be described in material and methods including the antibodies that have been used
• Abbreviations should be explained
• Figure 4A: Levels of p53 should also be shown for untreated fam83f -/-KO1 and KO2 animals
• Some references are missing (e.g. page 17, lane 320/321: As this group of cells arises....)
• Lane 369: The authors write about 4 KO lines but only two are shown in the figure.
• Lane 374/375: The NMD is not proven
• Lane 380: how can RNA levels of fam83fa be upregulated when the gene has been knocked out? Why are these genes only upregulated in KO1? How relevant is this?
• Figure 3C is not mentioned in the text and lacks any labelling
• Lane 434/435: all relevant data should be shown (can be done as supplementary figure)
• Lane 434: The reference to the figure seems to be incorrect (5A<->4A)
• Figure 4C and 4D: what is the difference?
• S5C and S5D: why are there 3 clusters?
• Lane 495 to 505: What does this mean that the GO analysis shows upregulation and downregulation of endopeptidases and why "in contrast"?

Significance

The strength of the manuscript is certainly that it provides inside into Fam83f function as there is not much known about Fam83f.

These study is probably most interesting for people in the zebrafish and related fields as the authors convincingly show the expression of Fam83f in the hatching gland and also the earlier hatching in the absence of the protein is very clear.

The weakness of the study is clearly that it does not provide an in-depth analysis. As such, it shows that Fam83f is involved in hatching and can delay the process but it remains elusive how this is achieved. (Likwise, also the investigation into the DNA damage response remains very superficial and does not prove a specific role for Fam83f in the DNA damage response or whether the increased sensitivity is more unspecifically caused by the absence of a gene or eventually even connected to the earlier hatching.

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, Jones et al. report on a potential role for fam83fa in zebrafish hatching, radiation response and autophagy. The authors are commended for generating multiple KO lines and maternal-zygotic embryos for analysis. However, important controls are lacking and the data is circumstantial throughout with very little mechanistic insight into the precise roles, if any, of fam83f in these processes.

1. Validation of the KO phenotypes (hatching, IR sensitivity) requires rescue with WT fam83fa WT mRNA, but not 1-500 or fam83fb mRNA.
2. While the hatching phenotype (Fig 3) is convincing, there is no data on HG development in the null embryos. Does the HG develop normally in the absence of fam83fb? If so, this would support the authors conclusions that the role of fam83fb is functional rather than developmental (indirect effect). In situs as in Fig.1 might be helpful here.
3. While the IR sensitivity phenotype (Fig S4) is convincing, IR-induced cell death/apoptosis was not analyzed. There is a large literature describing straightforward assays for cell death/apoptosis detection in zebrafish with assays such as acridine orange or TUNEL labeling, or active casp3 whole-mount IF. Is IR-induced cell death enhanced in fam83fa KOs?
4. Similarly, there are multiple tools to assay autophagy in zebrafish (e.g., Moss et al., Histochem Cell Biol 2020, PMC7609422; Mathai et al., Cells 2017, PMC5617967). Is autophagy affected in the KOs, with or without IR? These experiments might directly implicate fam83fa in autophagy.
5. Figure 4: Isn't there a slight reduction in p53 induction at 10 hours?
6. Given the widely documented, dominant role of p53 in zebrafish IR-sensitivity, the authors should test if the IR sensitivity of fam83fa KO animals is p53-dependent, ideally via a cross into p53 null, but at least via injection of p53 morpholinos.
7. Do autophagy inhibitors phenocopy the hatching and IR-sensitivity defects of fam83fa embryos? Do the inhibitors exacerbate the mutant phenotypes or synergize with M or Z mutant phenotypes? (I may have missed this but do M and Z fam83fa null embryos have any phenotype? Or do the phenotypes only manifest in MZ embryos?)

Significance

The role of Fam83f is not known. This study in zebrafish might be the first to clarify the function of this protein in vivo.

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 reviewer for their careful evaluation and constructive criticisms of our manuscript. We also appreciate the positive review by all three reviewers. The reviewers noted:

• "The computational model in this manuscript can be a tool to discover unknown molecular pathways interactions in cardiomyocyte proliferation."
• "This is an interesting study reporting the generation of a computational model of cardiomyocyte proliferation, which predicts molecular drivers of cell cycle progression."
• "The model provides a convenient systems framework to prioritize potential signaling drivers of therapeutic modulators of cardiomyocyte proliferation." We have responded to all reviewer comments and have outlined the corresponding additions and changes to the manuscript.

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

Summary:

In the manuscript by Harris et al. titled "Dynamic map illuminates Hippo to cMyc module crosstalk driving cardiomyocyte proliferation," the authors developed a computational model of cardiac proliferation signaling that incorporates various regulatory networks (cytokinesis, mitosis, DNA replication, etc.) to predict molecular drivers (genes) that support cardiomyocyte proliferation. Published research articles on cardiomyocyte proliferation in multiple contexts (different species, ages, in vitro and in vivo, etc) were used to build and validate the computational model. The authors found using their model that different processes during cardiomyocyte proliferation may or may not be context-dependent. For example, DNA replication is regulated differently in conditions with high Neuregulin compared to high YAP, whereas mitosis and cytokinesis regulation is similar in these conditions. To experimentally validate their model, the authors used an in vitro system to test the effects of YAP on 3 connected pathways; in the context of YAP activation, inhibition of PI3K, cMyc, or FoxM1 was combined to assay cell-cycle markers in cultured neonatal rat ventricular cardiomyocytes. Cell-cycle marker expression in cardiomyocytes was attenuated by inhibition of cMyc or PI3K, suggesting that these pathways are involved in YAP-mediated cardiomyocyte proliferation. While this model can be a good tool to gain new insights on interactions between molecular pathways, there are a few questions to be addressed prior to publication.

We appreciate the Reviewer's positive remarks about important findings in our manuscript and the ability of our model to be a tool to gain insights on interactions between molecular pathways to regulate cardiomyocyte proliferation. We have strived to address their points, as shown below.

Major Comments:

1. One of the potential uses for this computational model is to discover new interactions between known pathways that are involved in cardiomyocyte proliferation. However, this would be more powerful if factors such as species, age (neonate vs. adult), and experimental design (in vivo vs. in vitro) were accounted for, as new node inputs or a combination of existing node input activity values. This is very important because cardiomyocyte proliferation can drastically vary depending on these experimental factors. We agree that future extensions of this model accounting for species, age, and experimental design may enable an understanding of how these factors regulate proliferation. While this model's predictions are most relevant to immature cardiomyocytes, we note that it is the first systems model of the molecular network regulating cardiomyocyte proliferation. We extensively validated it against neonatal cardiomyocyte literature and then made new predictions regarding Hippo-cMyc pathways, which we validated in new cardiomyocyte experiments and against data in adult mice. This provides a strong foundation for future extensions. We now address this potential in the Discussion:

"While our model's predictions are most relevant to immature cardiomyocytes, it is the first systems model of the molecular network regulating cardiomyocytes. In the future, we hope that we and others may extend this model to identify how factors like species, age, and experimental design regulate proliferation. However, these endeavors would span multiple manuscripts, and the field currently lacks sufficient stage-specific data. For example, a previous foundational computational model of cardiomyocyte electrophysiology (Luo and Rudy, Circ Res 1994) focused on adult guinea pigs. This model became the foundation for a range of developmental and species-specific models in electrophysiology (Tusscher et al, AJP 2004,; Courtemanche eta al, AJP 1998; Paci et al, ABME 2013). We believe the open availability of our code will enable similar dissemination and extension for additional factors." Line 651-661

For reference:

Luo and Rudy, Circ Res 1994, >2.1k citations; Tusscher et al, AJP 2004, >1.7k citations; Courtemanche et al, AJP 1998, 1.5K citations; Paci et al, ABME 2013, 147 citations

The finding that cardiomyocyte proliferation is context-dependent is very exciting and warrants further investigation/validation. The authors state that different sets of nodes/modules are affected by neuregulin activation compared to YAP activation. This should be experimentally validated - qPCR/Western blots on sets of genes that are predicted to be differentially regulated in the high neuregulin context vs the high YAP context.

We agree that the model's prediction of context-dependent cardiomyocyte proliferation is very exciting. To further validate these predictions, we have performed additional experiments to validate context-dependent changes of phospho-ERK treated with Nrg and TT10. Using a high throughput capillary electrophoresis western blot system, we observed that with a short treatment of 30 min, Nrg induces greater phosphor-ERK compared to TT10, which validates our model predictions at short time intervals. Additionally, the model predicted greater p-AKT with 30 min treatment with Nrg compared to TT10. To validate this prediction, we now compare to Western blots from Hara et al. examining p-AKT in Nrg and TT10-treated cells. Validating our model predictions, their data show that Nrg induces greater p-AKT than with TT10. We have added new panels C, D, and E to Figure 4.

Figure 4: Influence of node knockdowns shifts with context, revealing crosstalk from Hippo to Growth Factor modules.

(A) Total influence of node knockdowns on the DNA replication, mitosis, and cytokinesis modules, compared across multiple signaling contexts: baseline, high Nrg, and high YAP. Total influence sums the overall effect of a node knockdown on a network module. (B) The total influence of each network module varies depending on whether a basal state, high Nrg, or high YAP signaling context is applied. (C) Capillary electrophoresis western blot for phosphorylated ERK, beta-actin, and GAPDH from neonatal cardiomyocytes treated with Nrg or TT10 for 30 min. (D) Model predictions of AKT and ERK activity of acute response to Nrg or TT10 (time constants for gene expression set to 100). (E) Quantification of effects of Nrg or TT10 treatment on p-ERK (from Western blot in panel C, n = 3) or p-AKT (from Western blot from (Hara et al., 2018), n = 1).

The overall description of the model can be improved. For example, how are the input and parameters set to validate or predict different experimental observations? What is the steady state activity of each of the nodes and does this make sense biologically? Including a few more sentences to explain the model would help with overall understanding for an uninformed reader.

We have addressed the following questions provided by the reviewer in the methods and results section of the manuscript:

How are the input and parameters set to validate or predict different experimental observations?

__ __"At baseline, input reaction weight parameters (w) were set based on information from the literature describing the baseline state of these inputs in the heart (each input reaction weight can be found in Supplemental File 1). To simulate experiments with biochemical stimuli, input reaction weights were increased to 0.8 or 1. To simulate experiments with inhibition or knockdown, the corresponding maximum species value (ymax) was set to 0.1 or 0. Complete annotations for all validation simulations are provided in Supplemental File 2." Line 154-160

What is the steady-state activity of each of the nodes and does this make sense biologically?

"Steady-state activity of model nodes was obtained by running the model until there was a __ __

Minor Comments:

Line 124 - The use of "species" and "reactions" is confusing to uninformed readers. Do you mean nodes and interactions/bridges?

We now further clarify these terms in the manuscript:

"As in past network models (Zeigler et al., PMID 27017945; Tan et al., PMID 29131824; Kraeutler et al PMID 21087478), species (or nodes) refer to a small molecule, gene, protein, or process. Reactions (or edges) are activating or inhibiting relationships between network species." Line 143-146

Line 130 - I could not find Supplementary File 2, which includes the references

We apologize for the error. Supplementary File 2 references articles and resources used to build the model. These files are now attached.

Line 257 - What is the meaning of the directional arrows in Fig 1A?

We clarified the Fig 1A legend:

"Arrows between modules represent one or more reactions that link species from one module to species in another module. " Line 594-595

Line 301 - Unclear what default values mean here. Please elaborate and provide an example of how this is reasonable.

We have added further descriptions of default values in reference to the parameters to the manuscript.

"A previous study identified default values of the parameters (ymax, EC50, W, etc.) that most accurately predict the results of knockdown screens compared to a model where all biochemical parameters were measured experimentally (Kraeutler et al 2010). Subsequent studies started from these default values and further demonstrated that model accuracy was robust to random variation in the parameters (Tan et all 2017, Zeigler et al 2017). Consistent with these prior models, we performed robustness analysis that demonstrates that the CM proliferation model accuracy (compared against 78 experiments) is maintained at >80% with up to 35% variation in ymax, 30% variation with w, and a variation of >50% with EC50 (Figure S4)." Line 305-312

Supplemental FigS2 - Why would knockdown of PKA, Lats1 or SMAD3 have the exact same effects on node activation? This is seen with multiple other genes was well (IGF and FGF for example).

PKA, Lats1, and SMAD3 all inhibit cell cycle progression in part through cMyc. Therefore, their knockdown have similar effects on downstream signaling and proliferation. Similarly, IGF and FGF both stimulate Ras and PI3K via similar mechanisms, which is consistent with experimental studies of IGF- and FGF-dependent proliferation.

Reviewer #1 (Significance (Required)):

The computational model in this manuscript can be a tool to discover unknown molecular pathway interactions in cardiomyocyte proliferation. The novelty lies in the ability to adjust any parameter or the entire setting/context. While this sounds very exciting, improvement of the model to account for age, experimental conditions (in vivo vs in vitro), and species (human, pig, mouse) could lead to increase prediction accuracy. Additionally, more robust validation of context-dependent interactions between signaling pathways would also increase overall enthusiasm for the manuscript. Readers interested in a systems biology approach to cardiomyocyte proliferation, or researchers probing molecular interactions during cardiomyocyte proliferation would be interested in using such a model to discover novel contexts/combinations in which cardiomyocyte proliferation is more likely.

The reviewer comes from a varied training background and is qualified to evaluate this manuscript in full - BS in biomedical engineering and mathematics. PhD in biomedical engineering (molecular biology, cardiac electrophysiology). Postdoctoral training in cardiac regeneration and immunity.

We appreciate the positive comments about our model of the cardiomyocyte proliferation network. As described above, we believe that we have addressed the concerns with additional experimental validation.

The manuscript submitted by Harris and colleagues collates a molecular map of cardiomyocyte cell cycle activation through mathematical modeling of previously published experimental results. They attempt to validate the constructed model several ways: 1) through testing results compiled from additional literature, 2) through in vitro analysis, and 3) through in vivo supporting data. When validating through additional literature the model proves quite reliable particularly for prediction of effects on synthesis, mitosis, and cytokinetic entry, but was less reliable (or insufficiently tested) at predicting completion of these stages as determined by polyploidization and multinucleation. A potentially novel observation which arose from the model - that hippo nodule connects to the growth factor nodule through PI3K, Myc, and FoxM1 - was partially confirmed with in vitro experiments, though a few experiments are warranted.

We appreciate the reviewer's recognition of the important contributions of this model of the cardiomyocyte proliferation network. We have addressed the concerns below.

Major comments:

• The model is admittedly weakest in its handling of completion of cytokinesis resulting in new daughter cells (i.e. proliferation) versus failure to complete either M phase or cytokinesis resulting in the much more common cellular phenotypes - polyploidy and multinucleation. Notably, very few molecules were "tested" for this output (figure 2) and this proved the least reliable aspect of the model/map. I wonder if the authors consulted the literature on somatic polyploidization at all when building the model (files not provided as indicated, see minor comment 1 below )? And if not, would doing so help strengthen this arm of their map? There are some great reviews on the topic (see PMIDs 25921783, 23849927, 30021843) - while admittedly much of the work is done on other cell types (i.e. trophoblast giant cells and hepatocytes) maybe understanding the molecular intricacies in these cells could be incorporated to strengthen the predictive model in cardiomyocytes. Notably, PMID 23849927 even provides a table of citations about key nodes in the model influencing polyploidy. To validate this model, we used entirely cardiomyocyte specific studies. We appreciate the reviewer's reference to PMID 23849927, which enabled us to add two additional experiments to the validation table in Figure 2. That paper found that overexpression of either cMyc or cyclin D increases polyploidy, which both matched our new simulations in the updated Figure 2.

Motivated by the reviewer's citation of PMID 23849927, we further validated the model against polyploidization data from multiple cell types, finding an 85.7% accuracy (6 of 7 experiments) as now shown in Supplementary Figure S7.

We included an additional discussion of polyploidization in the manuscript.

"Our model validation is notably weakest in predicting experiments on polyploidization, indicating a need to better characterize polyploidy and cytokinesis pathways. Because such data are limited in cardiomyocytes, we performed an additional validation against polyploidization experiments from other cell types as summarized in Pandit et al. Our CM proliferation model predicted 85% (6 of 7) experiments. Future experiments are needed to identify conserved or differential mechanisms of polyploidization and cytokinesis in cardiomyocytes." Line 587-594

• Paragraph on the cytokinesis module (lines 364-377) is confusing - not sure what the takeaway message is. Also, while progression through G1/S and G2/M are "required" for cytokinesis they on their own are not sufficient (lines 366-368), this perhaps goes back to major comment 1. We agree this sentence was confusing, it was meant to be introductory rather than stating a particular result. We removed that sentence and further revised our description of the output module to clarify the model structure:

"The output module interlinks the phenotypic outputs of the other modules, representing how experimentally measured aspects of cell cycle activity (DNA replication by EdU or Ki67), mitosis by phospho-Histone 3 (pHH3), abscission by cytokinetic midbody converge on polyploidy, binucleation, or cytokinesis (e.g. completed proliferation) (Figure 1G)." Line 283-286

Minor comments:

• Use of the word "Proliferation" should be reserved for situations where the authors can clearly say a new daughter cell was born. In many instances, "cell cycle activation" or "cell cycle progression" might be better terms. As suggested by the reviewer, we now use "cell cycle progression" in 7 instances, reserving "proliferation" for cell cycle progression through cytokinesis. In the remaining 90 instances, we refer to proliferation based on the model's predictions of completed cell division based on the combined DNA replication, mitosis, and cytokinesis pathways in the "output module". We retain "proliferation" in the title because the model encompasses the entire proliferation process from cell cycle entry through cytokinesis.

• Supplementary Files 1 & 2 or Supplementary Document 2 were not provided or not found during review, thus we were unable to confirm which literature were used to build and validate the model. Thank you, we have included Supplementary Files 1 and 2 along with supplementary document 2 in the submission.

• Figures are too small, particular Figure 1 We have enlarged Figure 1.

• "E2F" should be specified as E2F1-3 yield quite distinct results from E2F7/8. We have changed "E2F" to "E2F123"

• Text corresponding to Figure 5 does not reference most of the panels in the Figure. i.e. figures are not "cited" in the text We have made sure that each panel in Figure 5 is referenced in the text addressing the figure. We have also bolded all references to Figure 5.

• Figure 5C - why is there no bars for PI3K. Text claims it was predicted by the model, but the data are missing? We apologize for the confusion regarding Figure 5C, in which the bar for PI3K was near zero. We now clarify this in the legend.

"Predicted DNA replication and mitosis activity is close to zero when PI3K is inhibited alone and when PI3K is inhibited in combination with TT10 treatment."

• Data provided in figure 5D & E are insufficient on their own to claim "proliferation". Perhaps adding total cardiomyocyte numbers, where one would expect expansion compared to control. We agree that Ki67 and pH3 are not sufficient to claim "proliferation", so we modified the Figure 5 legend to:

"Prediction and experimental validation of cardiomyocyte cell cycle progression mediated by the Hippo pathway via PI3K, cMyc, and FoxM1."

We previously found that cardiomyocyte numbers without live tracking are not sufficient to robustly measure proliferation (Woo et al, J Mol Cardiol, 2019).

• Consider adding a details about the p-values to the figure legend in figure 5. Thank you for this suggestion p-value information has been added to the legend of Figure 5. We use *** Our literature-based validation in Figure 2 focused on 78 experiments that examined well-established and corroborated aspects of cardiomyocyte proliferation. Later in the paper, we focused on a newly predicted mechanism of cardiomyocyte proliferation involving small number of comparisons that would naturally have a lower a priori probability of validation in vitro neonatal experiments (Figure 5) and adult mouse experiments (Figure 6). Therefore, in the revised text we focus on the specific comparisons rather than statistics.

"Based on predictions from this validated model, we hypothesized that YAP drove proliferation via PI3K, cMyc and FoxM1. To test this model-driven hypothesis, we accurately predicted TT10-induced DNA replication that is suppressed by inhibition of PI3K, cMyc and to a lesser extent FoxM1 (Figure 5D). These model predictions were further validated using RNA-seq and ATAC-seq data from adult mouse hearts showing that constitutively active YAPS5A induces expression of Myc and FoxM1 as well as increased chromatin accessibility at PI3Kca and Myc." Line 454-468

In the discussion, we add:

"Further model revision is needed based on these molecular mechanisms of YAP-TEAD-Myc interactions to distinguish between chromatin accessibility, transcription factor binding, and gene expression." Line 649-651

As it stands now, the generated map largely constitutes already known details offering few if any new insights; however, if updated as new results arise AND made available as a public tool, the model could prove to be a highly valuable resource to the field.

We thank the reviewer for recognizing our model as a valuable resource and public tool. We have made our model publicly available on GitHub at https://github.com/saucermanlab/Cardiomyocyte-Proliferation-Network.

The virtual knockdown screens in Figure 3, 4 and 5 provide a wide range of new insights, which we clarify in new text.

"Because this is a literature-based network model, each component or direct interaction has been studied individually. However, our model makes much broader predictions of how these components interact to regulate proliferation, beyond the ~30 papers available for validation on the response of this system to perturbations shown in Figure 3. For example, Supplemental Figure S3 provides ~5000 predictions of how each protein responds to knockdown of every other protein. These predictions led to new insights into how YAP regulates proliferation via cMyc (experimentally validated in Figure 5 in vitro and Figure 6 in vivo), as well as many other insights that can be validated in future studies. These future studies will be aided by the open-source availability of our model on GitHub." Line 563-571

__ I have expertise in cardiomyocyte cell cycle and polyploidization.__

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

The authors generated a computational model of cardiomyocyte proliferation, which predicts molecular drivers of cell cycle progression. Interestingly, the model correctly predicts the outcome of 95% independent experiments from the literature. The model also elucidated crosstalk between the growth factor and Hippo modules and the authors identified key hubs for which the Hippo signaling pathway regulates cardiomyocyte proliferation. The model provides a convenient systems framework to prioritize potential signaling drivers of therapeutic modulators of cardiomyocyte proliferation.

Reviewer #3 (Significance (Required)):

This is an interesting study reporting the generation of a computational model of cardiomyocyte proliferation, which predicts molecular drivers of cell cycle progression. The program may provide a convenient framework prioritizing potential signaling drivers of therapeutic modulators of cardiomyocyte proliferation. However, the overall impact of the study appears modest since it is unclear whether the study allows elucidation of the unique properties of cardiomyocyte proliferation in adult hearts (i.e. they hardly proliferate) and the validation study was conducted only in neonatal myocytes. The field has seen many studies with neonatal myocytes but the findings are not always translatable to adult cardiomyocytes.

We thank the reviewer for recognizing the importance of our work that provides a framework for prioritizing potential signaling drivers of therapeutic modulators of CM proliferation.

Neonatal studies are the most prevalent with cardiomyocyte proliferation literature, making it the most robust starting point that allows for rigorous validation. Based on the high performance of the model against neonatal data, in the future we expect this model to be a stepping stone towards adaptions to understand differences in the adult cardiomyocyte proliferation network. We have updated our model discussion on future directions on this point.

"While our model's predictions are most relevant to immature cardiomyocytes, it is the first molecular network model of cardiomyocyte proliferation. In the future, this model will enable extensions to identify how factors like species, age and experimental design regulate proliferation. However, such endeavors would span multiple manuscripts and the field currently lacks sufficient stage-specific data. For example, the highly influential computational model of Luo and Rudy focused on adult guinea-pig cardiomyocyte electrophysiology (Luo and Rudy, Circ Res 1994). That model became the foundation for a wide range of development- and species-specific models in electrophysiology (Tusscher et al, AJP 2004,; Courtemanche eta al, AJP 1998; Paci et al, ABME 2013). We believe the open availability of our code will enable similar dissemination and extension for additional factors regulating cardiomyocyte proliferation." Line 655-665

The authors described that "Literature articles used for model development came from multiple cell types due to limited CM data." It is unclear whether this would allow the identification of unique mechanisms present in cardiomyocytes. As the authors admitted, the fact that the model predictions and experimental observations for polyploidization did not match clearly suggests the complexity surrounding the possibility of cell phenotypes in cardiomyocyte populations. The authors could have addressed whether this model allows the identification of unique mechanisms mediating cardiomyocyte proliferation in the adult heart.

Although we necessarily included literature on other cell types to support network reactions, all of the experimental validation in Figure 2 was with cardiomyocyte data (~33 publications). 80% of experiments were from neonatal CMs, 10% from adult CMs, 5% from in vivo studies, and the other 5% from hiPSC-derived cardiomyocytes as annotated in Supplemental File 3.

At this time, there is insufficient data from which to make a model focused only on adult CMs. The mode's open-source availability enables future extensions that examine age and species-dependent mechanisms of cardiomyocyte proliferation. We updated the manuscript, addressing the ability of our model to adapt to new information.

"This model provides an initial network framework for integrating additional discoveries in cardiomyocyte proliferation. As more information becomes available in cardiomyocyte proliferation literature the model can be adapted. Additionally, the field can use our open-sourced model to adapt this model to other developmental stages or species." Line 671-674

Acknowledging the limited data on cardiomyocyte polyploidy, we performed a new separate validation of 7 experiments in non-myocytes from PMID 23849927, finding an 85.7% accuracy (new Supplementary Figure S7).

Please provide more information regarding the rationale for having six modules in the authors' model, including the growth factor and the Hippo pathway.

We revised the text to clarify the motivation for the six modules:

"Our initial review of the literature indicated multiple complex molecular pathways that regulate cardiomyocyte proliferation, including growth factors, Hippo signaling, G1/S transition, G2/M transition, or cytokinesis pathways (Hashmi and Ahmad, PMID: 31205684; Payan et al, PMID: 30930108; Moral et al., PMID: 35008660; Wang et al., PMID: 30111784; Johnson et al., PMID: 34360531). Several review articles (Zheng et al, PMID: 32664346; Mia and Singh, PMID: 31632964; Diaz Del Moral et al, PMID: 35008660; Besson et al, PMID: 18267085; Wang et al, PMID: 19216791)) also organized the literature based on these distinct pathways or processes, which we used to define the boundaries of the six modules. However, how these molecular pathways work together is not well characterized. Therefore, we designed the model to incorporate each of these established modules and how they work together to drive cardiomyocyte proliferation." Line 550-557

The extent of cardiomyocyte proliferation at baseline is very low in the adult heart. The model identified 25 nodes that may influence baseline proliferation. Is there any evidence to support the involvement of these mechanisms in baseline cardiomyocyte proliferation in vivo?

We agree with the reviewer that proliferation at baseline is very low in the adult heart, and also rather low in neonatal cardiomyocytes. As shown in Figure S4A, we performed a virtual knockdown screen under baseline conditions that showed that no genetic knockdowns caused a substantial decrease in DNA replication or cytokinesis, consistent with a low baseline proliferation rate.

We describe this point about baseline proliferation in revised text:

"A complete virtual knockdown screen of the model was done under baseline conditions in Figure S4A, which showed that no knockdowns caused substantial decreases in DNA replication or cytokinesis. This is consistent with a low baseline proliferation rate described in cardiomyocyte literature." Line 354-357

The validation study was conducted with neonatal rat ventricular cardiomyocytes. This study could have been repeated with adult cardiomyocytes since they are more resistant to proliferation and, thus, the Myc may not work as expected. In addition, the authors could have commented on the mechanism through which chromatin opening and YAP allow transcription of Myc in the heart.

We agree that Myc is likely less proliferative in adult hearts. While our model was extensively validated against neonatal cardiomyocytes (Figure 2 for literature, Figure 5 for new neonatal experiments), only 10% of literature-based validations in Figure 2 are from adult cardiomyocytes due to limited data. However, in Figure 6 we validate YAP-dependent signaling to Myc, PI3K, and FOXM1 using RNA-seq and ATAC-seq data from Monroe et al. from adult mouse cardiomyocytes in vivo. While molecular mechanisms of YAP regulation of Myc are not characterized in the heart, based on the reviewer's suggestion, we add new discussion on YAP-Myc interaction in other cells:

"Overexpression of Myc induces cardiomyocyte proliferation in vitro and in vivo in several contexts, with open chromatin and Myc binding near mitotic genes (PMID: 32286286). But to our knowledge, crosstalk of YAP with Myc has not been reported in the heart. Our model prediction and experiments in neonatal cardiomyocytes support a YAP-TEAD-Myc pathway for cardiomyocyte proliferation. Further, our analysis of ATAC-seq and RNA-seq data from Monroe et al. validate that YAP induces Myc chromatin availability and gene expression in adult mouse hearts.

In MDA-MB-231 breast cancer cells, YAP/TAZ/TEAD bind directly to Myc enhancers through chromatin looping, with decreased acetylation of H3K27 and cell proliferation upon YAP/TAZ knockdown (26258633). YAP-TEAD-Myc signaling regulates the proliferation of cancer cells (26258633), tumorigenesis (29416644), and the growth of Drosophila imaginal discs (20951343). In the future, computational models and experiments are needed to better resolve how YAP promotes proliferation via Myc in the adult heart, including regulation by Mycn (30315164), cyclin T1 (32286286)."Line 632-644

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 generated a computational model of cardiomyocyte proliferation, which predicts molecular drivers of cell cycle progression. Interestingly, the model correctly predicts the outcome of 95% independent experiments from the literature. The model also elucidated crosstalk between the growth factor and Hippo modules and the authors identified key hubs for which the Hippo signaling pathway regulates cardiomyocyte proliferation. The model provides a convenient systems framework to prioritize potential signaling drivers of therapeutic modulators of cardiomyocyte proliferation.

Significance

This is an interesting study reporting the generation of a computational model of cardiomyocyte proliferation, which predicts molecular drivers of cell cycle progression. The program may provide a convenient framework prioritizing potential signaling drivers of therapeutic modulators of cardiomyocyte proliferation. However, the overall impact of the study appears modest since it is unclear whether the study allows elucidation of the unique properties of cardiomyocyte proliferation in adult hearts (i.e. they hardly proliferate) and the validation study was conducted only in neonatal myocytes. The field has seen many studies with neonatal myocytes but the findings are not always translatable to adult cardiomyocytes.

The authors described that "Literature articles used for model development came from multiple cell types due to limited CM data." It is unclear whether this would allow the identification of unique mechanisms present in cardiomyocytes. As the authors admitted, the fact that the model predictions and experimental observations for polyploidization did not match clearly suggests the complexity surrounding the possibility of cell phenotypes in cardiomyocyte populations. The authors could have addressed whether this model allows the identification of unique mechanisms mediating cardiomyocyte proliferation in the adult heart.

Please provide more information regarding the rationale for having six modules in the authors' model, including the growth factor and the Hippo pathway.

The extent of cardiomyocyte proliferation at baseline is very low in the adult heart. The model identified 25 nodes that may influence baseline proliferation. Is there any evidence to support the involvement of these mechanisms in baseline cardiomyocyte proliferation in vivo?

The validation study was conducted with neonatal rat ventricular cardiomyocytes. This study could have been repeated with adult cardiomyocytes since they are more resistant to proliferation and, thus, the Myc may not work as expected. In addition, the authors could have commented on the mechanism through which chromatin opening and YAP allow transcription of Myc in the heart.

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 submitted by Harris and colleagues collates a molecular map of cardiomyocyte cell cycle activation through mathematical modeling of previously published experimental results. They attempt to validate the constructed model several ways: 1) through testing results compiled from additional literature, 2) through in vitro analysis, and 3) through in vivo supporting data. When validating through additional literature the model proves quite reliable particularly for prediction of effects on synthesis, mitosis, and cytokinetic entry, but was less reliable (or insufficiently tested) at predicting completion of these stages as determined by polyploidization and multinucleation. A potentially novel observation which arose from the model - that hippo nodule connects to the growth factor nodule through PI3K, Myc, and FoxM1 - was partially confirmed with in vitro experiments, though a few experiments are warranted.

Major comments:

1. The model is admittedly weakest in its handling of completion of cytokinesis resulting in new daughter cells (i.e. proliferation) versus failure to complete either M phase or cytokinesis resulting in the much more common cellular phenotypes - polyploidy and multinucleation. Notably, very few molecules were "tested" for this output (figure 2) and this proved the least reliable aspect of the model/map. I wonder if the authors consulted the literature on somatic polyploidization at all when building the model (files not provided as indicated, see minor comment 1 below)? And if not, would doing so help strengthen this arm of their map? There are some great reviews on the topic (see PMIDs 25921783, 23849927, 30021843) - while admittedly much of the work is done on other cell types (i.e. trophoblast giant cells and hepatocytes) maybe understanding the molecular intricacies in these cells could be incorporated to strengthen the predictive model in cardiomyocytes. Notably, PMID 23849927 even provides a table of citations about key nodes in the model influencing polyploidy.
2. Paragraph on the cytokinesis module (lines 364-377) is confusing - not sure what the takeaway message is. Also, while progression through G1/S and G2/M are "required" for cytokinesis they on their own are not sufficient (lines 366-368), this perhaps goes back to major comment 1.

Minor comments:

1. Use of the word "Proliferation" should be reserved for situations where the authors can clearly say a new daughter cell was born. In many instances "cell cycle activation" or "cell cycle progression" might be better terms.
2. Supplementary Files 1 & 2 or Supplementary Document 2 were not provided or not found during review, thus we were unable to confirm which literature were used to build and validate the model.
3. Figures are too small, particular Figure 1
4. "E2F" should be specified as E2F1-3 yield quite distinct results from E2F7/8.
5. Text corresponding to Figure 5 does not reference most of the panels in the Figure. i.e. figures are not "cited" in the text
6. Figure 5C - why is there no bars for PI3K. Text claims it was predicted by the model, but the data are missing?
7. Data provided in figure 5D & E are insufficient on their own to claim "proliferation". Perhaps adding total cardiomyocyte numbers, where one would expect expansion compared to control.
8. Consider adding a details about the p-values to the figure legend in figure 5.
9. Data presented in figure 6 do not "validate" the model. Rescue experiments as were provided in vitro would be necessary or at minimum YAP/TEAD binding to the promoters (ATAC insufficient). Alternatively, walking back these statements, might be easiest.
10. Validation studies through the literature suggested ~94% fidelity. The invitro validation suggests 66% reliability of model? The in vivo 33%? Perhaps this should be added as a discussion point - can the authors comment on the loss of fidelity as the rigor/complexity of the experiment increased?

Significance

As it stands now, the generated map largely constitutes already known details offering few if any new insights; however, if updated as new results arise AND made available as a public tool, the model could prove to be a highly valuable resource to the field.

I have expertise in cardiomyocyte cell cycle and polyploidization

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 the manuscript by Harris et al. titled "Dynamic map illuminates Hippo to cMyc module crosstalk driving cardiomyocyte proliferation," the authors developed a computational model of cardiac proliferation signaling that incorporates various regulatory networks (cytokinesis, mitosis, DNA replication, etc.) to predict molecular drivers (genes) that support cardiomyocyte proliferation. Published research articles on cardiomyocyte proliferation in multiple contexts (different species, ages, in vitro and in vivo, etc) were used to build and validate the computational model. The authors found using their model that different processes during cardiomyocyte proliferation may or may not be context-dependent. For example, DNA replication is regulated differently in conditions with high Neuregulin compared to high YAP, whereas mitosis and cytokinesis regulation is similar in these conditions. To experimentally validate their model, the authors used an in vitro system to test the effects of YAP on 3 connected pathways; in the context of YAP activation, inhibition of PI3K, cMyc, or FoxM1 was combined to assay cell-cycle markers in cultured neonatal rat ventricular cardiomyocytes. Cell-cycle marker expression in cardiomyocytes was attenuated by inhibition of cMyc or PI3K, suggesting that these pathways are involved in YAP-mediated cardiomyocyte proliferation. While this model can be a good tool to gain new insights on interactions between molecular pathways, there are a few questions to be addressed prior to publication.

Major Comments:

1. One of the potential uses for this computational model is to discover new interactions between known pathways that are involved in cardiomyocyte proliferation. However, this would be more powerful if factors such as species, age (neonate vs. adult), experimental design (in vivo vs. in vitro) are accounted for, as new node inputs or a combination of existing node input activity values. This is very important because cardiomyocyte proliferation can drastically vary depending on these experimental factors.
2. The finding that cardiomyocyte proliferation is context-dependent is very exciting and warrants further investigation/validation. The authors state that different sets of nodes/modules are affected by neuregulin activation compared to YAP activation. This should be experimentally validated - qPCR/Western blots on sets of genes that are predicted to be differentially regulated in the high neuregulin context vs the high YAP context.
3. The overall description of the model can be improved. How are the modules and overarching model built from published results? For example, how are the input and parameters set to validate or predict different experimental observations? What is the steady-state activity of each of the nodes and does this make sense biologically? Includng a few more sentences to explain the model would help with overall understanding for an uninformed reader.

Minor Comments:

Line 124 - The use of "species" and "reactions" is confusing to uninformed readers. Do you mean nodes and interactions/bridges?

Line 130 - I could not find Supplementary File 2, which includes the references?

Line 251 - "theseanal"

Line 257 - What is the meaning of the directional arrows in Fig 1A?

Line 301 - Unclear what default values mean here. Please elaborate and provide an example of how this is reasonable?

Supplemental Fig S2 - Why would knockdown of PKA, Lats1, or SMAD3 have the exact same effects on node activation? This is seen with multiple other genes as well (IGF and FGF for example).

Significance

The computational model in this manuscript can be a tool to discover unknown molecular pathway interactions in cardiomyocyte proliferation. The novelty lies in the ability to adjust any parameter or the entire setting/context. While this sounds very exciting, improvement of the model to account for age, experimental conditions (in vivo vs in vitro), and species (human, pig, mouse) could lead to increase prediction accuracy. Additionally, more robust validation of context-dependent interactions between signaling pathways would also increase overall enthusiasm for the manuscript. Readers interested in a systems biology approach to cardiomyocyte proliferation, or researchers probing molecular interactions during cardiomyocyte proliferation would be interested in using such a model to discover novel contexts/combinations in which cardiomyocyte proliferation is more likely.

The reviewer comes from a varied training background and is qualified to evaluate this manuscript in full - BS in biomedical engineering and mathematics. PhD in biomedical engineering (molecular biology, cardiac electrophysiology). Postdoctoral training in cardiac regeneration and immunity.

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

Learn more at Review Commons

Reply to the reviewers

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

Sang and colleagues present in vitro evidence for purified PARP1 forming condensates with DNA under certain conditions. They report the PARP1/DNA/NaCl concentration ranges under which these condensates form, and the impact of adding the PARP1 substrate NAD+ that allows poly(ADP-ribose) production. The results are presented clearly for the most part. One wonders if these in vitro conditions are really representative of DNA repair foci; however, the study adds new information that will be useful to the field. As noted below, there is some concern about the lack of reversibility of the condensates.

• DNA curtains assay. The imaging buffer is listed as including 25 mM NaCl and 2 mM MgCl2. Were these experiments performed in conditions of higher ionic strength? The lack of a response to the addition of NAD+ is puzzling. It seems that the condensing of the DNA is not reversible. For the data in panel C, would the DNA return to extended form upon further flow of buffer only? The data give the impression that the assay conditions promote a one-way road to an irreversible state, and it is hard to see how this should be interpreted. A different single-molecule study indicated that PARP1 condensation of long DNA is reversible with NAD+ addition (PMID 34380612). The different outcomes should be discussed.
• DNA ligase assay. Figure 5E,F. How can it be certain that the ligation is actually taking place in the condensates under these conditions? Can the Cy5 signal be shown? Is there a way to separate the condensates from the dilute phase, and then analyze the ligation state of the DNA? Also, the reactions could be performed under the same conditions (e.g. uM concentrations of LigIII and XRCC1) as presented in the rest of the figure.

The discussion mentions the impact of HPF1 on PARP1 and that the research team has used HPF1 in the past. Is there a reason for excluding it from the current study? In particular in an effort to address the reversibility/mobility of the condensates?

Supp. Figure 1F. It would be useful to convert the ng/uL concentrations to micromolar concentrations, perhaps in the legend for each nucleic acid. This would make the results easier to relate to the rest of the data that is generally listed in micromolar.

Figure 6A. ZnF3, BRCT, and WGR domains of PARP1 also bind to DNA and should probably be included, as it could help explain why full-length PARP1 is needed for the most robust condensate formation.

Supp. Figure 1G. The legend for this panel indicates absence or presence of NAD+, and that the DNA concentration is indicated, but this does not seem consistent with the figure panel.

Results, first sentence. There is a missing parenthesis.

Figure 2G,H. It would be useful to see the plot of the other replicates of the fusion experiments.

page 3. "(data not shown)" Probably worth including.

page 4. "(Supp. Fig. 3B and 3D)" Also Supp. Fig. 3C ?

Referees cross-commenting

Reviewer #1 comments are clearly stated and justified. There is good overlap in the feedback.

Significance

Strength: the study provides parameters for studying PARP1 condensate formation. Differential impact on repair factors is interesting.

Limitation: the reconstitution is missing elements that could have a very big impact (e.g. nucleosomes).

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

PARP1 plays important roles in the recognition and repair of DNA damage, primarily by catalyzing the formation of PAR at DNA breaks, which assembles various repair factors and other PAR-binding proteins. The anionic PAR scaffold was previously proposed to induce condensation of phase separating proteins including FUS (PMIDs: 26317470, 26286827), and PARP1 was recently reported to form condensates at DNA breaks to promote nucleosome dynamics and DNA end synapsis (PMIDs: 38320550, 38215753). However, several aspects of PARP1-dependent repair condensate formation and how such condensates contribute functionally to repair have remained unclear. Here, Sang et al. show that purified PARP1 forms viscous droplets in vitro in a DNA binding-dependent manner, enhanced by PAR. Interestingly, the downstream DNA repair factors XRCC1, POLB, DNA Ligase III, and FUS co-assemble with PARP1 and damaged DNA, albeit with different enrichment patterns, resulting in multiphase condensates. Functionally, the authors not only confirm a DNA end bridging function by PARP1-mediated condensation, but also report enhanced DNA end ligation. Their in vitro experiments, which are state-of-the-art and are overall well controlled, despite lacking an in vivo counterpart, provide an important step forward in the reconstitution of multi-step DNA repair reactions in repair condensates.

Major comments:

1. In addition to the short oligonucleotides that were used in this study to evaluate DNA-dependent PARP1 condensation (Table S2), the use of circular plasmid DNA (nicked or broken to resemble SSBs or DSBs, respectively) should be considered to corroborate key findings.
2. Although not essential for the main conclusions, it would be very interesting to address the role of PARG on PAR-dependent multiphase condensates. Based on the methods section, the authors have purified full-length PARG, so experiments to address the consequences of PARG-dependent PAR degradation on repair condensates and their disassembly seem feasible.
3. If PAR increases the internal dynamics and mobility of PARP1 in condensates, why does it not seem to affect the DNA end bridging function?
4. Catalytically inactive mutants of PARP1 could be employed to separate the PARP1-dependent DNA end bridging function from PAR-dependent modulation of PARP1 dynamics in condensates. Additionally, it would help to show that the DNA end bridging function depends on the ZnF domains and can be modulated by conditions that alter PARP1 condensation (see also point 6).
5. The differential organization of XRCC1, LIG3, POLB, and FUS is intriguing, but the implications of this behavior remain unclear. Can the droplet assays be adapted to inform about the sequence of events from DNA damage recognition by PARP1 and PAR induction to the handing over of the break site to LIG3 for end ligation?
6. The increase in end ligation, which correlates with PARP1- and PAR-dependent condensate formation, is very interesting. However, from the experimental setup it seems unclear if the observed effect is due to condensation or simply PARylation. Additional controls would be needed to substantiate a functional role of condensation for ligation (as implied also in the title of the manuscript). Perhaps it is feasible to modulate condensation (e.g. enhanced by crowding agents, reduced by salt or by 1,6-hexanediol) without affecting PARylation, and then reassess how this affects ligation.

Minor comments:

1. Please double-check if previous studies reaching similar conclusions are referenced appropriately.
2. Please carefully double-check if all references to figure panels are correct.
3. Please carefully double-check if the methods descriptions and discussion match the displayed experimental procedures and results.
4. Supplemental Figure 3 seems to contain only negative results. Consider showing experimental conditions with the same proteins or protein combinations that result in droplet formation as well.

Referees cross-commenting

The comments by reviewer #2 seem comprehensible and justified. Similar points are addressed, e.g. major points 2, 3, and 6 in review #1.

Significance

Several new and exciting findings on PARP1-dependent and PAR-modulated repair condensate formation are presented, including the multiphase behavior and the functional contribution to DNA compaction, end synapsis, and ligation. The study extends previous work on PARP1/PAR-triggered liquid demixing and complements very new and recently published work by the Alberti and Kay labs on PARP1 condensation at DSBs. The study makes an important step forward towards the reconstitution of DNA repair reactions inside multi-component repair condensates in vitro, which may eventually allow making testable predictions for repair condensate functions in vivo. Strengths of the current study include complementary state-of-the-art in vitro techniques such as biochemical assays, multi-component droplet assays, and single molecule experiments, which were conceived and conducted carefully. Limitations relate primarily to the undemonstrated relevance in cells, which may, however, be beyond the scope of the current study. A broad audience of basic researchers in the areas of genome stability and biomolecular condensates will likely be interested in this study.

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

Learn more at Review Commons

Reply to the reviewers

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

The majority of the conclusions are well supported by strong experimental evidence. The only area where that is not fully the case is the role of Pak1 as a downstream effector of FoxG1-FoxO6 and its effects on macropinocytosis. To further strengthen this claim, the authors should demonstrate that ablation of Pak1 can rescue the functional consequences of forced FoxO6 expression and whether overexpression of Pak1 rescues quiescence exit in FoxO6 knockout. Thank you to the reviewer for these helpful suggestions. To investigate the effects of Pak1 ablation, and therefore more directly the link between FOXG1 and FoxO6 and macropinocytosis, we tested the published Pak1 inhibitor IPA-3. Unfortunately, to distinguish the role of Pak1 in quiescence exit and macropinocytosis, we would need a dosage of IPA-3 that is efficacious but does not affect cell proliferation. It was not possible to optimise such a dosage (a dosage of 10uM is shown to be efficacious at inhibiting Pak1 (Verma et al, 2020; Wong et al, 2013) however even at 2.5uM we see significant cell death in our cells. Indeed, this is potentially due to pleiotropic roles for Pak1.

Also, it is not feasible to overexpress Pak1 in the FoxO6 KO cells with inducible FOXG1. To ensure we are investigating quiescence exit this would need to be in an inducible manner; however, re-transfecting cells using the PiggyBac system would potentially alter FOXG1 transgene levels by excising the existing transgene.

As shown in Figure S3, we do not observe clear vacuole formation in F6 (FOXG1-inducible) cells upon Dox addition. As detailed in the discussion, we hypothesise that FoxO6-induced macropinocytosis could represent a stalled state, with other pathways downstream of FOXG1 necessary to be activated concomitantly to ensure cell cycle re-entry, e.g., through increased pinocytic flux that cannot be assessed within our experimental timeframes. Indeed, active Pak1 has been found to modulate pinocytic cycling, enhancing both FITC-dextran uptake and efflux (Dharmawardhane et al, 2000). We therefore would not hypothesise that high Pak1 levels alone would be sufficient to drive quiescence exit.

Alternatively, the macropinocytosis observed may be a metabolic stress response because of the hyperactivation of signalling pathways upon FoxO6 overexpression. Hyperactivation of Ras signalling, canonical Wnt and PI3K signalling have all been shown to play roles in inducing macropinocytosis (Overmeyer et al, 2008; Tejeda-Muñoz et al, 2019; Recouvreux & Commisso, 2017).

We believe the observed macropinocytosis phenotype upon Foxo6 overexpression, and the changes in Pak1 expression upon Foxo6 loss or FOXG1 induction provide interesting insights into the function of this underexplored FoxO family member. However, currently we are unable to demonstrate a direct link between these processes and have therefore modified the text to reflect this (see lines 292-4, 330-3, 365-8).

• The manuscript stresses the role of NSC quiescence exit in GBM and demonstrates that FoxG1 KO reduces FoxO6 levels in a murine GBM cell line but a BMP4-mediated quiescence and dox-induced FoxG1 over-expression or an abolishment of cell cycle re-entry thereof by reduced FoxO6 levels in the case of FoxG1 KO is lacking. But this would significantly substantiate the relevance of the findings. *

Mouse GBM cells have elevated levels of FoxG1 and have been shown to be refractory to BMP4-mediated quiescence entry, maintaining colony formation following BMP treatment (Bulstrode et al, 2017). It is therefore challenging to specifically investigate cell cycle re-entry/ quiescence exit using these mouse GBM cells, or indeed any GBM cell line due to their inability to respond fully to BMP cues (Caren et al, 2015). It has also been shown by Bulstrode et al, 2017 that Foxg1 null mouse neural stem cells show an increased propensity to exit cycle in response to BMP treatment, and reduced colony formation on return to EGF/FGF-2 growth factors. FOXG1 null cell lines therefore show a reduced response to BMP cues, making it difficult to explore quiescence exit per se.To navigate this, instead we investigated Dox-induced FOXG1 overexpression in FoxO6 WT and KO mouse NS cells, which display similar quiescence characteristics upon BMP treatment (Figure 4).

• In the introduction and discussion, FoxO6 is mentioned for its oncogenic roles in various cancers but no reference to GBM specifically is cited. It feels like a missed opportunity to not show evidence of this in the IENS cell line that has reduced levels of FoxO6; is there an effect in their proliferative capacity? What are the expression levels of Pak1 following FoxG1 KO in IENS cells? *

Thank you for the helpful suggestion. It is indeed true the literature on FoxO6 in GBM is lacking, explaining the absence of citations on this. On investigation of expression of the proliferation marker Ki67 in these cells we found no significant difference in expression, now shown in Figure 1H. This is in fitting with previous findings of our lab (Bulstrode et al, 2017) which show that FOXG1 is dispensable for the maintenance of continued NSC or GSC proliferation in vitro. We investigated the expression levels of Pak1 following FOXG1 KO in IENS and found a decrease in both KO lines compared to parental cells (updated Figure 6F).

As explained in our discussion, these data suggest that Foxg1/FoxO6/Pak1 are not functionally important in sustaining GSC/NSC proliferation, as shown by the lack of proliferation defects upon Foxg1 or FoxO6 deletion (Bulstrode et al, 2017), but impact regulatory transitions, as cells prepare to exit quiescence into the proliferative radial-glia like state.

*Minor comments *

- Fig1A shows 4 and 2-fold respectively for the two mouse NSC lines, not 17 and 4-fold increase as written on manuscript, please adjust accordingly.

The qRT-PCR data are presented as log2(fold change) or - ddCt, where this value equals zero for the calibrator sample, as indicated in the figure legends and axes. The data are presented in this way to enable accurate visualisation of up- and down-regulation of gene expression. Data are stated as ‘fold increase’ in the text for ease of reading, which we have clarified in the text and figure legends (e.g. lines 154 and 176).

• • Fig2G manuscript reports a 235-fold upregulation, but graph looks more like a 7 or 8-fold as shown on Fig1A for the F6 NSC line. I would recommend checking the fold changes reported throughout the paper. *

See previous comment above. The qRT-PCR data are presented as log2(fold change) or - ddCt, where this value equals zero for the calibrator, as indicated in the figure legends and axes. The data are presented in this way to enable accurate visualisation of up- and down-regulation of gene expression. Data are stated as ‘fold increase’ in the text for ease of reading, which we have clarified in the text and figure legends (e.g. lines 154 and 176).

• The manuscript describes the increase of FOXG1 after BMP4-induced cell cycle exit as compared to non-BMP4 treated cells (p.8 first paragraph), but I am wondering if this expression is rather compared to dox negative and not vs BMP4 negative treatment. *

Data are presented relative to the non-BMP treated (EGF/FGF-2) control throughout the manuscript for consistency. This is to enable changes in expression between -Dox and +Dox to be visualised throughout the quiescence-exit time course relative to the initial starting population in EGF/FGF-2 growth media, prior to BMP treatment.

• 1. In Fig2G it is interesting that FoxO6 is upregulated in BMP4 treated throughout the experiment with highest values at day10 post treatment. At the same time, non-BMP4 treated cells keep decreasing their FoxO6 levels dramatically but there is no mention or reference to this effect.*

In Figure 2G, all cells have been treated with BMP4, prior to return to growth media (EGF/FGF) with or without Dox. It is true that in the +Dox condition with FOXG1 induction, FoxO6 levels continue to increase up to Day 10, perhaps reflective of the expansion of a highly proliferative radial glia-like population.

• 1. Fig2 would benefit from a western blot like Fig1D where FoxG1 and FoxO6-HA protein levels are also shown in dox-treated comparing BMP4-treated vs non-treated. *

Due to the lack of specific FoxO6 antibodies and the absence of a FoxO6-HA tag in this cell line, it is not possible to perform protein analysis of FoxO6 levels in this figure as for Figure 1D.

• The colonies in Fig3E should be quantified, as their ability to form neurospheres seems somewhat compromised upon FoxO6 KO. Fig3B and 3F could perhaps be consolidated into one panel in the interest of space and presentation. *

Good suggestion. We have now consolidated Fig 3B and 3F into one panel (now Figure 3F) as suggested by the reviewer. We performed additional replicates for Figure 3E to quantify the colony formation efficiency. This showed a small but insignificant decrease in colony forming ability in the KO cells (Figure 3E). Importantly the FoxO6 null cells do form colonies, and our results show that FoxO6 is not essential for proliferation or colony formation of NSCs in EGF/FGF-2 – this therefore does not account for the complete loss in colony formation we see the in the FoxO6 KO cells upon FOXG1 induction.

• Fig4A shows vs "parental" non-BMP on y axis but wouldn't this show fold change of dox+ parental vs parental. The authors should clarify this. *

All samples in Figure 4A are compared to parental cells in EGF/FGF-2, i.e. non-BMP treated, as the calibrator sample where log2(fold change) equals zero. We chose to set a single calibrator sample for all data (parental and FoxO6 KO cells included) to allow us to compare changes in FOXG1 transgene across the entire experiment.

• Perhaps the authors can add a non-BMP4 treated count of % FOXG1 positive cells to Fig4C for reference. *

As shown in Figure 4A, both parental and FoxO6 KO cells show similar, i.e. negligible, FOXG1 transgene expression without Dox, compared to the parental non-BMP4 treated control, therefore negligible FOXG1-V5 positive cells are seen by ICC. We have edited Figure 4A to include a non-BMP treated and BMP-treated control to show the negligible FOXG1-V5 expression by qPCR as controls.

• The sentence mentioning Fig5D for the first time (p.10 third paragraph) needs rephrasing for clarity and should also call out Fig5C for the mCherry expression live cell imaging data where appropriate. Fig5D does not appear to be live imaging as implied by the text. If vacuole formation is observed already as early as 10-11h after Dox induction, then it should be shown somewhere in Fig5. Vacuole formation is shown with a higher magnification image inset only in the 22h timepoint image. I think Fig5E should be more substantiated with some sort of quantification, e.g. % of vacuoles positive for EEA1 and/or LAMP1. *

We apologise for this. The first reference to Figure 5D one line 234 should refer to Figure 5C, this has now been corrected in the text. Vacuoles are visible in Figure 5C panel 10 h 30 min, however, to make this clearer we have also supplied an accompanying movie of the live imaging (Movie 1). The imaging in Fig 5E has not been quantified as this imaging was performed with the purpose of confirming the vacuole structures seen are not simply enlarged lysosomes, due to their similarity in appearance to those published elsewhere (Ramosaj et al, 2021; Leeman et al, 2018). Instead, we have provided Western blotting data in Figure S5E to support this conclusion that there is no clear increase in EEA1 or LAMP1 (early endosomal or lysosomal) expression upon FoxO6-HA induction.

*- Could the authors comment on the lack of proliferative advantage of the FoxO6 overexpression. FigS3 shows Edu staining, but there is no proliferation assay in either Fig5 or S3. What would be the effect of FoxO6 overexpression on BMP4-mediated quiescence with or without FoxG1 over-expression? *

Induction of FoxO6-HA overexpression does not provide a proliferative advantage to the cells. Looking at individual cells, those with high FoxO6-HA levels seem to associate with EdU negativity. In Figure S3 we provide quantitative EdU incorporation assay as a proliferation assay (quantification of the number of cells cycling, therefore incorporating EdU, within a 24h pulse period). Quantification of the EdU staining in Figure S3G is provided in Figure S3H. We have now clarified this in the text on page 11, lines 263-4.

Unfortunately, due to transgene overexpression using the PiggyBac transposon method, it is not feasible to overexpress FoxO6 and FOXG1 in the same cell line, as re-transfecting cells using the PiggyBac system would potentially alter FOXG1 transgene levels and make results difficult to interpret. Given the association of vacuolated cells with EdU negativity, we predict that FoxO6 overexpression would not give an advantage for quiescence exit. Indeed, BMP-treated cells with FoxO6 overexpression show a decrease in EdU positivity, as shown in Figure S3H. As discussed in the text, we hypothesise that cells with FoxO6 overexpression are in a stalled state, potentially due to signalling hyperactivation. While this may not be physiological, it gives us clues as to the function and downstream targets of FoxO6, which remain uncharacterised.

*- Can the authors clarify if there is a proliferation change in F6 cells in Fig6F as in Fig2F? Fig6F shows Pak1 is already upregulated in quiescent NSCs, what are the expression levels of Pak1 in FoxO6 -/- ANS4 cells upon FoxG1-mediated quiescence exit as shown in Fig4? Is there a particular reason why the F6 cell line data is shown only up to day2 post Dox-induction rather than d4 or d10? For consistency with the rest of similar experimental data this timeline should be extended. Does Pak1 remain elevated, plateaus or keeps reducing further post day2? *

The data is (previous) Figure 6F is the same assay and cell line as presented in Figure 2, but at an early timepoint (Day 2) during the quiescence exit assay. We have provided in the panel qRT-PCR analysis of Ki67 to show that cells begin to show increased proliferation at this timepoint. Due to our hypothesis that Pak1 is required at an early transition point, we decided to analyse this expression at an earlier timepoint than Figure 2. We have also repeated this at D10 (data below), showing Pak1 levels continue to increase with time, along with FoxO6 and the proliferative marker Ki67. Due to technical issues with variable FOXG1 transgene levels we were unable to analyse Pak1 expression levels in FoxO6+/- ANS4 cells upon FOXG1-mediated quiescence exit.

*15 . Reviewer #1 (Significance (Required)): *

The study provides a conceptual advance for exit from stem cell quiescence. There is strong evidence provided for murine neural stem cells, but the link to GBM cancer stem cells is less developed (but perhaps this is the subject of a separate manuscript).

While FoxG1 is a known regulator of neurodevelopment and glioblastoma, the functions of FoxO6 have not been studied in the context of neural stem cells. In my view, this study should be of high interest to audiences in both neurodevelopment and cancer research. * Expertise: glioblastoma, cancer stem cells, neurodevelopment *

We have edited the text and title to clarify that neural stem cells are used here as a model for GSCs with high levels of FOXG1 (e.g. lines 36 and 69).

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

*Major comments: *

-The choice of NSCs as a main experimental model to understand the effects of FoxG1 and FoxO6 is not fully justified. The authors had previously shown that FoxG1 is expressed at very low levels in NSCs (Fig. 1A in Bulstrode et al. 2017). FoxO6 also seems to be barely expressed in NSCs (Fig. 1 of the current manuscript) and, in addition, its levels seem to go further down as cells exit quiescence (-Dox line in Fig. 2H). Therefore, these two genes do not seem to play an important role in the normal exit from quiescence of NSCs, with FoxO6 only affecting FoxG1 overexpression-induced exit from quiescence. * * *If the aim is to mimic a GBM-like state by FoxG1 overexpression, this should be made much clearer in the text, including title and abstract. In that case, the authors should also show a direct comparison of the levels of FoxG1 in GBM and upon Dox-induced overexpression in NSCs. *

We agree with this criticism and suggestion to fix this. It is indeed our aim to mimic a GBM-like state by inducing FOXG1 overexpression and we should have made that more explicit. All experiments are performed in the context of high FOXG1 level. Like Foxg1, FoxO6’s homeostatic roles may be subtle in adulthood, and mostly involved in neural plasticity (Yu et al, 2019). This is in keeping with our finding that basal FoxO6 levels are low in adult NSCs and not required for sustained proliferation but are important for cell state transitions. If the FoxO6 levels activated by elevated FOXG1 represent an acquired dependency of GBM, there may be a therapeutic window to target this pathway. However, given the poorly understood roles of FoxO6, further work is needed to determine its specific value as a therapeutic target. We have modified the title and the text to make this clearer. This is also stated in the first paragraph of the results section on page 7 (line 148).

We have provided below a Western Blot (Bulstrode, 2016) in which FOXG1 levels in F6 cells induced with Dox (1000 ng/ml the dosage used) with the GBM cell lines G7 and G144, and the normal NS cell line U5. This shows that the FOXG1 levels induced are significantly higher than found in normal neural stem cells (mouse or human). This model has been previously used and published in Bulstrode et al, 2017, upon which this manuscript expands.

*-While the authors state that they aim to study NSC quiescence, they use a protocol that is closer to modelling astrocytic differentiation. In fact, in their previous work, they use this very same protocol (removal of growth factors and addition of BMP) to study the role of FoxG1 and Sox2 on astrocyte de-differentiation (Bulstrode et al. 2017). While there is arguably no perfect in vitro model of NSC quiescence, the current standard in the field is treatment with both BMP and FGF for 48 to 72 hours (e.g.: Mira et al., 2010, Martynoga et al., 2013, Knobloch et al., 2017, Leeman et al., 2020). BMP alone is regarded as a pro-astrocytic differentiation cue, and 24 hours might not be enough for NSCs to fully commit to either differentiation or quiescence. Therefore, either the claims in the paper are changed to match the astrocytic differentiation model, or a standard quiescence protocol should be used throughout to confirm the findings also apply to the exit from quiescence of NSCs. *

We agree with the reviewer that there is indeed no perfect in vitro model of NSC quiescence and thank the reviewer for this useful discussion. Coincident with this project, this was an active area of research from our laboratory as explored by Marques-Torrejon et al, 2021 (Nature Comms). After 24 h BMP4 treatment, we found that adult mouse NS cells: exit cell cycle, are growth factor unresponsive, obtain an astrocytic morphology, upregulate astrocytic markers such as Gfap and Aqp4, and downregulate radial glia/NS cell markers such as Nestin and Olig2 (Figure 3).

We therefore initially viewed them as terminally differentiated. However, the exact state of these cells is difficult to define due to the lack of definitive markers and transcriptional differences that can distinguish terminally differentiated GFAP-expressing astrocytes from quiescent type B SVZ NS cells (which also express GFAP) (Bulstrode et al, 2017; Doetsch et al, 1999; Codega et al, 2014). Findings from our laboratory later suggested some NS cell markers are maintained following BMP4 treatment and these cells can be forced back into cycle with combined Wnt/EGF signalling, or FGF/BMP signalling (Marques-Torrejon et al 2021). This suggests in vitro NS cells may lie along a continuous spectrum of states from dormant quiescent, activated quiescent (primed for cell cycle re-entry) to actively proliferating, similar to that observed in vivo in the mouse SVZ (Dulken et al, 2017). Indeed, after 24 h BMP4 treatment, we observe a minimal level of colony formation in no Dox controls following 10 days of exposure to the growth factors EGF/FGF-2 (Figure 2D-F).

These non-cycling BMP4-induced astrocytic cells might therefore be better viewed as dormant quiescent NSCs, hence our reference as quiescent NSCs. The assay conditions used in this manuscript differ to those of Marques-Torrejon et al, in terms of density and length of BMP4 treatment; it is therefore likely that our BMP-treated cells are at different stages along the continuum between dormancy and primed quiescent states. Importantly, regardless of the exact cell type induced by 24 h BMP4 treatment, we have considered the changes induced by FOXG1 overexpression, in comparison to the effect of NS cell media alone.

*-The FoxO6-induced vacuole formation in NSCs is a very interesting finding. However, so far it was only observed upon FoxO6 overexpression. To claim vacuolization is required for quiescence exit, the authors should show whether this phenomenon is also observed upon normal exit from quiescence and FoxG1-induced reactivation of NSCs. From the author's own data, Pak1 (which induces vacuolization) is unlikely to reactivate NSCs, as its expression is highest in BMP-treated cells (Figure 6F). The authors should show whether some vacuolization is present at these stage in NSCs and if not, discuss the possible interplay between Pak1 and FoxO6 in vacuole formation and quiescence exit. *

As detailed in the discussion, we hypothesise that FoxO6- induced macropinocytosis could represent a stalled state, with other pathways downstream of FOXG1 necessary to be activated concomitantly to ensure cell cycle re-entry, e.g., through increased pinocytic flux that cannot be assessed within our experimental timeframes. Indeed, active Pak1 has been found to modulate pinocytic cycling, enhancing both FITC-dextran uptake and efflux (Dharmawardhane et al, 2000). Alternatively, the macropinocytosis observed may be a metabolic stress response because of hyperactivation of signalling pathways upon FoxO6 overexpression Hyperactivation of Ras signalling, canonical Wnt and PI3K signalling have all been shown to play roles in inducing macropinocytosis (Overmeyer et al, 2008; Tejeda-Muñoz et al, 2019; Recouvreux & Commisso, 2017).

We do not see clear evidence of vacuoles in FOXG1-induced reactivation of NSCs – this supports that the macropinocytosis seen upon FoxO6 overexpression is a stalled state or due to hyperactivation. While this may not be physical, it gives us clues as to the function and downstream targets of FoxO6, which remain uncharacterised (such as a link of FoxO6 and FOXG1 with Pak1-related pathways). Demonstrating a requirement for vacuolisation in quiescence exit is outwidth this manuscript and therefore we are careful not to claim this. We have modified the text to clarify this.

As the reviewer noted, it is interesting that Pak1 is highest in BMP-treated cells; it seems that BMP signalling itself is triggering elevated Pak1 levels, likely as cells undergo extensive cell shape changes during the transition from proliferation to quiescence. However, in EGF/FGF-2, Pak1 levels decrease, and our data suggests that FOXG1/FoxO6 are required to increase or maintain Pak1, potentially to again enable the cell shape/metabolic changes required on quiescence exit. We have added to the text to expand upon this observation on page 14 (lines 330-333). -Finally, the data on the regulation of Pak1 expression by FoxO6 is insufficient to draw any strong conclusions. Downregulation of Pak1 in FoxO6 cells is not enough evidence to claim a direct regulation. The authors should show whether Pak1 levels are increased after FoxO6 overexpression and whether FoxG1 is downregulated in FoxO6 KO NSCs (indirectly affecting Pak1 expression).

We have performed qRT-PCR analysis of Foxg1 expression in FoxO6 KO NSCs and see no consistent difference in expression, indicating this is not indirectly affecting Pak1 expression (see below, 1). We have also investigated Pak1 levels upon FoxO6 overexpression, over a time course following Dox addition (see below, 2). Interestingly, when FoxO6 is overexpressed, Pak1 is not clearly upregulated at any time-point. It may be that as Pak1 is already expressed in the -Dox controls, due to its roles in a variety of cellular functions, that the levels are saturated already. It is clear that Pak1 expression decreases upon FoxO6 loss in EGF/FGF (without coincident Foxg1 downregulation) and in F6 cells, higher FOXG1 correlates with higher Pak1 in EGF/FGF. Together with the induction of macropinocytosis upon FoxO6 overexpression, these data provide interesting insights into the potential pathways downstream of Foxo6 in controlling quiescence exit, directly or indirectly related to Pak1 signalling. We have modified the text to reflect this on page 14 (lines 330-333).

Minor comments: * Please state in the main text that NSCs are derived from the SVZ. *

This has been added to the text on page 7 (line 149) and is in the methods ‘Cell Culture’ section.

Reviewer #2 (Significance (Required)):

As I said before, I find this work tackles a very important question, how is the exit from quiescence controlled in NSCs. This manuscript will be of interest to researchers in the fields of adult stem cell biology and adult neurogenesis. While my expertise lies mostly on NSC biology, this work is of potential great interest for the cancer field, particularly for brain cancer research. Elucidating the mechanisms GBM cells use to exit quiescence is crucial in order to avoid the relapse of this aggressive form of brain cancer. To increase the relevance of the work to the cancer community, some of the key findings should be reproduced with GBM cells. It would be particularly important to show whether Pak1 induced vacuolization and macropinocytosis can be observed in GBM cells.

As detailed in the discussion, we hypothesise that FoxO6- induced macropinocytosis could represent a stalled state, with other pathways downstream of FOXG1 necessary to be activated concomitantly to ensure cell cycle re-entry, e.g., through increased pinocytic flux that cannot be assessed within our experimental timeframes. Alternatively, the macropinocytosis observed may be a metabolic stress response because of hyperactivation of signalling pathways upon FoxO6 overexpression Hyperactivation of Ras signalling, canonical Wnt and PI3K signalling have all been shown to play roles in inducing macropinocytosis (Overmeyer et al, 2008; Tejeda-Muñoz et al, 2019; Recouvreux & Commisso, 2017). We do not see clear evidence of vacuoles in FOXG1-indued reactivation of NSCs– this supports that the macropinocytosis seen upon FoxO6 overexpression is a stalled state or due to hyperactivation. We do not therefore think macropinocytosis per se would be observed in quiescence exit of GBM cells – indeed a normal form of macropinocytosis-induced cell death called methuosis has been observed in GBM cells with hyperactivated Ras signalling (Overmeyer et al, 2008). However, this phenotype still gives us clues as to the function of FoxO6 in quiescence exit in GSCs and the downstream signalling pathways it may regulate, such as Pak1-related signalling (discussed on lines 330-3 and 366-9).

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

Summary: * The overall objective of the paper is to investigate the mechanisms by which co-option of the activity of developmental master lineage regulators by cancer cells allows them to gain fitness. To answer this question, they focus on FOXG1. This TF acts during the specification of the telecephalon. Its expression can be increased in Glioblastoma (GBM) and, more importantly for the paper, FOXG1 has previously been shown to promote exit from quiescence of glioblastoma stem cells (GSCs) and non-transformed neural stem cells (NSCs). In a previous screen, the authors identified FoxO6 as a potential direct target gene of FOXG1. In this paper, they showed that with the gain of expression for FOXG1 in NSCs and loss of FOXG1 in GSCs, FoxO6 is increased or decreased, respectively. Loss of FoxO6 in NSCs does not alter their cell cycle or cell shape and specification. Yet, loss of FoxO6 in NSCs blocks FOXG1-mediated exit from quiescence. To understand the mechanisms, they decided to overexpress FoxO6 in NSCs and demonstrated that the cells undergo macropinocytosis, a process by which cells can engulf large amount of nutriments from the external medium. It remains to be determined whether this macropinocytosis occurs in cells overexpressing FOXG1 and GSCs. The authors provide a first answer by showing that overexpression of FOXG1 induces not only FoxO6 but also the expression of PAK1, one of the key kinases that regulates the membrane engulfment of macropinocytosis in NSCs. In GSC lines, the decrease of FOXO6 decreases PAK1 levels. *

Major comments: * The paper describes interesting and convincing results (number of cell lines, repeated experiments seems sufficient) but it is difficult to reconcile them all in a single model, and this diminishes the impact of the study. Epistatic interactions between FoxG1, FoxO6, PAK1 and macropinocytosis are not always studied in the same cell models. Whether FOXG1-induced exit from quiescence of NSCs is dependent on a FOXG1-->FOXO6-->PAK1-->Macropinocytosis axis remains to be demonstrated. Also does such an axis operate in tumor cells remains to be fully assessed? In particular, if FoxO6 overexpression in NSCs can induce macropinocytosis, is this cellular process induced by FoxO6 downstream of FOXG1 activity during NSC quiescence exit? Is PAK1 a relay of FoxO6? Experiments looking at macropinocytosis and the involvement of PAK1 in the cell models of Figure 4 will definitely help to bridge the different results all together. *

We thank the reviewer for this useful insight and discussion for future work.

To directly investigate the effects of Pak1 ablation, and therefore more directly the link between FOXG1 and FoxO6 and macropinocytosis, we tested the published Pak1 inhibitor IPA-3. Unfortunately, to distinguish the role of Pak1 in quiescence exit and macropinocytosis, we would need a dosage of IPA-3 that is efficacious but does not affect cell proliferation. It was not possible to optimise such a dosage (a dosage of 10uM is shown to be efficacious at inhibiting Pak1 (Verma et al, 2020; Wong et al, 2013) however even at 2.5uM we see significant cell death in our cells. Indeed, this is potentially due to the variety of cellular functions Pak1 is involved in. Conversely, it is not feasible to overexpress Pak1 in the FoxO6 KO cells with inducible FOXG1. To ensure we are investigating quiescence exit this would need to be in an inducible manner; however, re-transfecting cells using the PiggyBac system would potentially alter FOXG1 transgene levels (through excision of the existing transgene) and therefore make results difficult to interpret.

We hypothesise that FoxO6- induced macropinocytosis could represent a stalled state, with other pathways downstream of FOXG1 necessary to be activated concomitantly to ensure cell cycle re-entry, e.g., through increased pinocytic flux that cannot be assessed within our experimental timeframes (as detailed in the text discussion). Alternatively, the macropinocytosis observed may be a metabolic stress response because of hyperactivation of signalling pathways upon FoxO6 overexpression Hyperactivation of Ras signalling, canonical Wnt and PI3K signalling have all been shown to play roles in inducing macropinocytosis (Overmeyer et al, 2008; Tejeda-Muñoz et al, 2019; Recouvreux & Commisso, 2017). We do not see clear evidence of vacuoles in FOXG1-induced reactivation of NSCs– this supports that the macropinocytosis seen upon FoxO6 overexpression is a stalled state or due to hyperactivation and therefore not a physiological process in quiescence exit. We do not therefore think macropinocytosis per se would be observed in quiescence exit of GBM cells – indeed a normal form of macropinocytosis-induced cell death called methuosis has been observed in GBM cells with hyperactivated Ras signalling (Overmeyer et al, 2008).

However, we believe the observed macropinocytosis phenotype upon Foxo6 overexpression, and the changes in Pak1 expression upon Foxo6 loss or FOXG1 induction provide interesting insights into the function of this underexplored FoxO family member, in GSCs and the downstream signalling pathways it may control, such as Pak1-related signalling. We have modified the text to reflect the limitations of our current data and discuss this (lines 330-3 and 366-9).

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

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary:

The overall objective of the paper is to investigate the mechanisms by which co-option of the activity of developmental master lineage regulators by cancer cells allows them to gain fitness. To answer this question, they focus on FOXG1. This TF acts during the specification of the telecephalon. Its expression can be increased in Glioblastoma (GBM) and, more importantly for the paper, FOXG1 has previously been shown to promote exit from quiescence of glioblastoma stem cells (GSCs) and non-transformed neural stem cells (NSCs). In a previous screen, the authors identified FoxO6 as a potential direct target gene of FOXG1. In this paper, they showed that with the gain of expression for FOXG1 in NSCs and loss of FOXG1 in GSCs, FoxO6 is increased or decreased, respectively. Loss of FoxO6 in NSCs does not alter their cell cycle or cell shape and specification. Yet, loss of FoxO6 in NSCs blocks FOXG1-mediated exit from quiescence. To understand the mechanisms, they decided to overexpress FoxO6 in NSCs and demonstrated that the cells undergo macropinocytosis, a process by which cells can engulf large amount of nutriments from the external medium. It remains to be determined whether this macropinocytosis occurs in cells overexpressing FOXG1 and GSCs. The authors provide a first answer by showing that overexpression of FOXG1 induces not only FoxO6 but also the expression of PAK1, one of the key kinases that regulates the membrane engulfment of macropinocytosis in NSCs. In GSC lines, the decrease of FOXO6 decreases PAK1 levels.

Major comments:

The paper describes interesting and convincing results (number of cell lines, repeated experiments seems sufficient) but it is difficult to reconcile them all in a single model, and this diminishes the impact of the study. Epistatic interactions between FoxG1, FoxO6, PAK1 and macropinocytosis are not always studied in the same cell models. Whether FOXG1-induced exit from quiescence of NSCs is dependent on a FOXG1-->FOXO6-->PAK1-->Macropinocytosis axis remains to be demonstrated. Also does such an axis operate in tumor cells remains to be fully assessed? In particular, if FoxO6 overexpression in NSCs can induce macropinocytosis, is this cellular process induced by FoxO6 downstream of FOXG1 activity during NSC quiescence exit? Is PAK1 a relay of FoxO6? Experiments looking at macropinocytosis and the involvement of PAK1 in the cell models of Figure 4 will definitely help to bridge the different results all together.

Minor comments:

No minor comments

Significance

Understanding how hijacking of developmental programs by tumour cells contributes to their fitness is important for the design of cancer therapies, as these programs often confer resistance to tumour cells. Although it has been shown that FOXG1, this master TF of telencephalon specification, can give cells the ability to leave quiescence, the downstream mechanisms were unknown. The identification of FoxO6 as a relay for FOXG1 and the suggestion that this may involve macropinocytosis and the PAK1 enzyme is interesting. FoxO6 acts differently from other members of the FoxO family and PAK1 could indeed be targeted. If the authors can integrate several of their findings into a single model, their paper should be of interest to oncologists, developmental biologists and cell biologists.

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

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

This manuscript by Ferguson et al. identifies FoxO6 as a FoxG1 target that promotes the reactivation of neural stem cells (NSCs). Quite remarkably, FoxO6 is dispensable for the proliferation or entry into quiescence of NSCs but required for their FoxG1-dependent reactivation. The authors claim that Pak1-induced macropinocytosis is required for quiescence exit and show that Pak1 expression depends on both FoxG1 and FoxO6. These findings are very interesting and could potentially help better understand the regulation of NSC quiescence. In addition, as the authors point out, they could shed light on the regulation of the exit of quiescence of glioblastoma multiforme (GBM) cells, which express higher levels of FoxG1 and FoxO6 than NSCs. The experiments are overall of high quality, with the authors making an appropriate and efficient use of CRISPR technologies to control the expression of their genes of interest in cultured NSCs. However, the findings are mostly the result of overexpression in NSCs (which do not seem to express FoxG1 or FoxO6) and the quiescence model used is neither the standard in the field nor appropriate to draw strong conclusions about quiescence exit in NSCs or GBM cells.

Major comments:

• The choice of NSCs as a main experimental model to understand the effects of FoxG1 and FoxO6 is not fully justified. The authors had previously shown that FoxG1 is expressed at very low levels in NSCs (Fig. 1A in Bulstrode et al. 2017). FoxO6 also seems to be barely expressed in NSCs (Fig. 1 of the current manuscript) and, in addition, its levels seem to go further down as cells exit quiescence (-Dox line in Fig. 2H). Therefore, these two genes do not seem to play an important role in the normal exit from quiescence of NSCs, with FoxO6 only affecting FoxG1 overexpression-induced exit from quiescence. If the aim is to mimic a GBM-like state by FoxG1 overexpression, this should be made much clearer in the text, including title and abstract. In that case, the authors should also show a direct comparison of the levels of FoxG1 in GBM and upon Dox-induced overexpression in NSCs.
• While the authors state that they aim to study NSC quiescence, they use a protocol that is closer to modelling astrocytic differentiation. In fact, in their previous work, they use this very same protocol (removal of growth factors and addition of BMP) to study the role of FoxG1 and Sox2 on astrocyte de-differentiation (Bulstrode et al. 2017). While there is arguably no perfect in vitro model of NSC quiescence, the current standard in the field is treatment with both BMP and FGF for 48 to 72 hours (e.g.: Mira et al., 2010, Martynoga et al., 2013, Knobloch et al., 2017, Leeman et al., 2020). BMP alone is regarded as a pro-astrocytic differentiation cue, and 24 hours might not be enough for NSCs to fully commit to either differentiation or quiescence. Therefore, either the claims in the paper are changed to match the astrocytic differentiation model, or a standard quiescence protocol should be used throughout to confirm the findings also apply to the exit from quiescence of NSCs.
• The FoxO6-induced vacuole formation in NSCs is a very interesting finding. However, so far it was only observed upon FoxO6 overexpression. To claim vacuolization is required for quiescence exit, the authors should show whether this phenomenon is also observed upon normal exit from quiescence and FoxG1-induced reactivation of NSCs. From the author's own data, Pak1 (which induces vacuolization) is unlikely to reactivate NSCs, as its expression is highest in BMP-treated cells (Figure 6F). The authors should show whether some vacuolization is present at these stage in NSCs and if not, discuss the possible interplay between Pak1 and FoxO6 in vacuole formation and quiescence exit.
• Finally, the data on the regulation of Pak1 expression by FoxO6 is insufficient to draw any strong conclusions. Downregulation of Pak1 in FoxO6 cells is not enough evidence to claim a direct regulation. The authors should show whether Pak1 levels are increased after FoxO6 overexpression and whether FoxG1 is downregulated in FoxO6 KO NSCs (indirectly affecting Pak1 expression).

Minor comments:

Please state in the main text that NSCs are derived from the SVZ.

Significance

As I said before, I find this work tackles a very important question, how is the exit from quiescence controlled in NSCs. This manuscript will be of interest to researchers in the fields of adult stem cell biology and adult neurogenesis. While my expertise lies mostly on NSC biology, this work is of potential great interest for the cancer field, particularly for brain cancer research. Elucidating the mechanisms GBM cells use to exit quiescence is crucial in order to avoid the relapse of this aggressive form of brain cancer. To increase the relevance of the work to the cancer community, some of the key findings should be reproduced with GBM cells. It would be particularly important to show whether Pak1 induced vacuolization and macropinocytosis can be observed in GBM cells.

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

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

The authors investigate mechanisms of quiescence and cell cycle entry in neural stem cells. Expanding on previous work, they show that FoxO6 is a target gene of FOXG1 in neural stem cells and glioma cancer stem cells. FoxO6 is upregulated following activation of stem cells from quiescence but is not required for proliferation. Continued over expression of FoxO6 leads to macropinocytosis through Pak1, indicating a link between FoxO6 and actin remodelling.

Major comments

The majority of the conclusions are well supported by strong experimental evidence. The only area where that is not fully the case is the role of Pak1 as a downstream effector of FoxG1-FoxO6 and its effects on macropinocytosis. To further strengthen this claim, the authors should demonstrate that ablation of Pak1 can rescue the functional consequences of forced FoxO6 expression and whether overexpression of Pak1 rescues quiescence exit in FoxO6 knockout.

The manuscript stresses the role of NSC quiescence exit in GBM and demonstrates that FoxG1 KO reduces FoxO6 levels in a murine GBM cell line but a BMP4-mediated quiescence and dox-induced FoxG1 over-expression or an abolishment of cell cycle re-entry thereof by reduced FoxO6 levels in the case of FoxG1 KO is lacking. But this would significantly substantiate the relevance of the findings. In the introduction and discussion, FoxO6 is mentioned for its oncogenic roles in various cancers but no reference to GBM specifically is cited. It feels like a missed opportunity to not show evidence of this in the IENS cell line that has reduced levels of FoxO6; is there an effect in their proliferative capacity? What are the expression levels of Pak1 following FoxG1 KO in IENS cells?

Minor comments

• Fig1A shows 4 and 2-fold respectively for the two mouse NSC lines, not 17 and 4-fold increase as written on manuscript, please adjust accordingly.
• Fig2G manuscript reports a 235-fold upregulation, but graph looks more like a 7 or 8-fold as shown on Fig1A for the F6 NSC line. I would recommend checking the fold changes reported throughout the paper.
• The manuscript describes the increase of FOXG1 after BMP4-induced cell cycle exit as compared to non-BMP4 treated cells (p.8 first paragraph), but I am wondering if this expression is rather compared to dox negative and not vs BMP4 negative treatment.
• In Fig2G it is interesting that FoxO6 is upregulated in BMP4 treated throughout the experiment with highest values at day10 post treatment. At the same time, non-BMP4 treated cells keep decreasing their FoxO6 levels dramatically but there is no mention or reference to this effect.
• Fig2 would benefit from a western blot like Fig1D where FoxG1 and FoxO6-HA protein levels are also shown in dox-treated comparing BMP4-treated vs non-treated.
• The colonies in Fig3E should be quantified, as their ability to form neurospheres seems somewhat compromised upon FoxO6 KO. Fig3B and 3F could perhaps be consolidated into one panel in the interest of space and presentation.
• Fig4A shows vs "parental" non-BMP on y axis but wouldn't this show fold change of dox+ parental vs parental. The authors should clarify this.
• Perhaps the authors can add a non-BMP4 treated count of % FOXG1 positive cells to Fig4C for reference.
• The sentence mentioning Fig5D for the first time (p.10 third paragraph) needs rephrasing for clarity and should also call out Fig5C for the mCherry expression live cell imaging data where appropriate. Fig5D does not appear to be live imaging as implied by the text. If vacuole formation is observed already as early as 10-11h after Dox induction, then it should be shown somewhere in Fig5. Vacuole formation is shown with a higher magnification image inset only in the 22h timepoint image. I think Fig5E should be more substantiated with some sort of quantification, e.g. % of vacuoles positive for EEA1 and/or LAMP1.
• Could the authors comment on the lack of proliferative advantage of the FoxO6 overexpression. FigS3 shows Edu staining, but there is no proliferation assay in either Fig5 or S3. What would be the effect of FoxO6 overexpression on BMP4-mediated quiescence with or without FoxG1 over-expression?
• Can the authors clarify if there is a proliferation change in F6 cells in Fig6F as in Fig2F? Fig6F shows Pak1 is already upregulated in quiescent NSCs, what are the expression levels of Pak1 in FoxO6 -/- ANS4 cells upon FoxG1-mediated quiescence exit as shown in Fig4? Is there a particular reason why the F6 cell line data is shown only up to day2 post Dox-induction rather than d4 or d10? For consistency with the rest of similar experimental data this timeline should be extended. Does Pak1 remain elevated, plateaus or keeps reducing further post day2?

Significance

The study provides a conceptual advance for exit from stem cell quiescence. There is strong evidence provided for murine neural stem cells, but the link to GBM cancer stem cells is less developed (but perhaps this is the subject of a separate manuscript). While FoxG1 is a known regulator of neurodevelopment and glioblastoma, the functions of FoxO6 have not been studied in the context of neural stem cells. In my view, this study should be of high interest to audiences in both neurodevelopment and cancer research.

Expertise: glioblastoma, cancer stem cells, neurodevelopment

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 this ms, the authors searched using bioinformatic approaches for the presence of natural antisense RNAs (cis-NATs) linked to LRR-RLK genes, a large gene family containing many major regulators of plant growth, to detect a potential novel mechanism for their post-transcriptional regulation. They detected a large proportion of LRR-RLK genes containing cis-NATs. TO address their potential functions, they overexpressed specific cis-NATs against Bri1, CLV1 and SOBIR1 and observed phenotypes reminiscent of the respective mutant in these genes. This means that these cis-NATs can act in trans on the endogenous loci. The authors detected reduction in expression of the cognate LRR-RLK in several transgenic lines except for SOBIR1 where a minor change in protein production was found. Then, they prepare transgenic GUS lines with the promoters of the NATs and detect variable levels of expression. In addition, expressing the NATs under epidermal specific promoters but not in a control promoter induced differences in BRI1 expression specifically in the plants. Finally, they search for the presence of NATs linked to LRR-RLK loci in other plants.

The paper is interesting but the message about NAT regulation is overstated and there are several conclusions that require major additional experiments

1. There are no experiences with mutant NATs to confirm a potential linkage to the regulation of LRR-RLK complementary transcript. Overexpressing a NAT may lead to several artefacts, including the artificial silencing of the endogenous loci (expressing a complementary RNA). Hence, it cannot be concluded that this regulation occurs in planta only based on overexpressing lines. The use of rdr6 mutants or similar may also serve to discard potential alternative NAT regulations.
2. Complementing a bri1 mutant (or in other RLK) with construct expressing the BRI1 locus with or without the NAT (e.g. with or without its promoter) is a much cleaner manner to show NAT regulation. CRISPR or other manipulation may allow to create mutants in the NAT alone to see the effects on its cis-target as well as its eventual trans action on other LRR-RLK (as apparently acts in trans).
3. The trans-NAT experiments will minimally require genome-wide RNAseq studies to see the level of cross-talk of a trans NAT. Even though the phenotype is related, there may be very general misregulation triggered by the NAT. A minimum is to pass all studied RLK in all the lines to define certain "specificity" of action.
4. To propose a translational regulation for SOBIR1 with its present data is an overstatement for me. Is the antisense nuclear or accumulate in ribosomes? Can the NAT lead to a change in the recruitment into ribosomes (without changing mRNA levels). There may be indirect effects that can explain these changes in protein accumulation
5. The GUS plants need to be shown in detail, with a more precise tissular localisation and compared to the cognate LRR-RLK. Can BRI1 expression be monitored in TRANS-NAT BRI1 vs wt using in situ or BRI1-GFP fusions. TO show the expression patterns of the antisense and the target will further support the proposed specificity regulation. What happens with other LRR-RLKs? Kinetics of expression along development in trans NATs and wt will be a possibility
6. The epidermal connection is interesting but there is any evidence about the "cell specificity" expression of BRI1 or the differential expression of BRI1 in cells containing or not the NATs?
7. The extrapolation to crops might be interesting if some features are conserved (e.g. BRI1 homologs contain NATs of the same size or similar cell-specific regulation). As many genes have NATs, this evolutionary argument of the presence of NATs is not convincing to conclude about a regulatory mechanism permitting "to engineer improved crop performances".

Significance

The paper in its present form is very limited and the major strength is the occurrence of similar phenotypes when NATs are overexpressed and the mutant cognate mRNAs. Alternative explanations of the reported data are possible and should be discarded and better presented. This paper does not advance in NAT regulatory mechanisms to be relevant for a broad audience in life sciences. However, the subject of LRR-RLK regulation is of great interest to the plant community as several critical growth regulators act through LRR-RLKs.

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 investigate the role of antisense transcription (cis-NATs) at Leucine-rich repeat receptor-like kinases (LRR-RLK) genes in plants. They find that many LRR-RLK genes are associated with cis-NATs, through data-mining and RT-PCR in Arabidopsis. For functional studies, they selected three cis-NATs for over-expression, from the BRI1, CLV1 and SOBIR genes, where over-expression has phenotypic consequences for the plants. They use reporter gene assays to study cisNAT expression is regulated across development. The authors examine the relationship between cis-NATs and LRR-RLKs in tomato and rice, where they also detect a high fraction of LRR-RLKs associated with cis-NATs.

Major comments:

• Over-expression of cis-NATs is used to support a functional role. Here, the experimental design leaves open if the effect is explained by a trans-acting role on the corresponding sense RNA (as the authors interpret), or, if the over-expression constructs trigger co-suppression of the corresponding sense RNAs. The authors should distinguish these possibilities prior to publication in a journal. It makes a big difference. Co-suppression would be a very different mechanism that is independent of the RNA functions the authors propose. The authors need to rule it out.

Minor comments:

• The authors use older resources to examine antisense transcription. In some sense, this makes it even more impressive that so many LRR-RLK genes are associated with antisense transcription, but it would be nice to include more recent data support the conclusions. In particular Araport11 is not considered a high-quality annotation (PMID: 34266383). The manuscript could benefit from the integration of some more recent genome-annotations, or genome browser screenshots for some the genes (e.g. BRI1, CLV1 and SOBIR) that shows some of the more recent methods (reviewed in PMID: 36259932). A lot of these data are even accessible on the TAIR website.

Significance

General assessment:

The authors observe cisNATs at LRR-RLK genes. These findings hint at a contribution of cisNATs to LRR-RLK regulation that has not received much attention yet. For functional characterization, the authors rely on over-expression of cis-NATs. Here, I am sceptical, this the results could also fit a model where cisNAT OE may trigger siRNA formation and silencing of the endogenous locus (TGS/co-suppression).

Advance: The manuscript uncovers cisNATs as potential large-scale regulators of LRR-RLK genes.

Audience: Plant scientists interested in gene expression and cellular roles of LRR-RLK genes.

Reviewer expertise: Plant lncRNA, genome annotation, plant gene expression, epigenetics.

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 all reviewers for their detailed and constructive feedback, which substantially helped improve the manuscript. We apologise for the time taken for the revisions, which was partially due to the first author (successfully) writing and defending her PhD thesis in the same time frame. We would like to point out already here that, based on reviewers' feedback, main figure 6 is completely redone and the conclusions of this figure have changed substantially. We no longer suggest RNA chaperoning activity (it was identified as being due to the high concentration of TEV protease, in a control suggested by the reviewers). Instead, our refined assay conditions with lower TEV protease concentration identified ribonuclease activity of membrane-bound full-length 2C, which is consistent with a publication from 2022 (PMID: 35947700).

---

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

Evidence, reproducibility, and clarity

Summary:

In this study by Shankar and colleagues, the authors aim to understand the structure and function of the enterovirus 2C protein, a putative viral helicase with AAA+ ATPase activity. Using poliovirus (as a model enterovirus) 2C, the author's propose the protein contains two amphipathic helices (AH1 and AH2) at the N-terminus that are divided by a conserved glycine. Using purified MBP-tagged 2C and N-terminal 2C truncations, their data suggests AH1 is primarily responsible for clustering at membranes, whilst AH2 is the main mediator of 2C oligmerisation and membrane binding. Furthermore, 2C was suggested to be able to recruit RNA to membranes, with a preference for dsRNA, and the author's data implies that the helicase activity of 2C is ATP-independent. Instead, the ATP activity appears to be required for 2C hexamer formation or chaperone activity. The manuscript is generally well written /presented and the author's present very interesting data which raises several questions, some of which require additional experimentation to help support the author's conclusions. Specific comments are as follows.

We thanks the reviewer for the overall positive assessment, as well as the specific comments below.

Major Comments:

1. The authors use four main constructs throughout the paper: full-length 2C, 2C with deletion of AH1 (ΔAH1), 2C with both AH1 and AH2 deleted (ΔMBP) and 2C with an extended N-terminal deletion. From this, the author's draw conclusions on the function of both AH1 and AH2. One of the author's main conclusions is that AH2 is the main mediator of 2C membrane association (e.g., in line 169). However, is it possible to conclude the relative importance of AH1 vs AH2 without testing a construct containing the deletion of AH2 only (ΔAH2)? This should be generated and used alongside this data to fully define the relative importance of AH1 and AH2 in these assay and remove the possibility that the deletion of AH1 changes the structure and/or function of AH2, which could also result in the observed differences.

This was a very good suggestion. We expressed and purified the ΔAH2 protein requested by the reviewer and characterized its oligomeric state as well as its membrane binding. It turns out, as suspected, that the ΔAH2 protein behaves very similarly to the ΔMBD protein (i.e. it does not form higher order oligomers and does not bind membranes). The changes in the manuscript due to this addition are many but can primarily be found in main figures 2-3 and their associated supplementary figures.

Previous structural predictions of 2C do not appear to have two separate AHs at the N-terminus. Are the AH1 and AH2 structures predicted to be formed in the context of the entire 2C protein, 2BC precursors and polyprotein? Are there structural approaches that could provide experimental evidence for two separate AH at the N-terminus?

This is a good point. Previous predictions were not that detailed, partially since they were done in the pre-alphafold era. Unfortunately, we cannot think of a tractable experimental method that could verify the split nature of the amphipathic helix in the only context that would matter: the protein bound to a membrane. A long-term goal would be in situ structures of full-length 2C on membranes using cryo-electron tomography, but our current sample and data sets are not sufficient for this. We added a mention of the long-term need for experimental structures of full-length 2C on lines 315-318 in the discussion.

Why are the 2C dimers (lines 137-138) not apparent on the mass photometry data presented (figure 2)?

Different constructs were measured by mas photometry and SEC-MALS. Also, the required concentration is 100-1000x lower for mass photometry which will affect a dynamic equilibrium in case the same construct were measured by the two methods.

It appeared that binding of ΔMBD-2C was better when POPS is in the membrane (line 174). What is the explanation for this and was this finding significant?

Well spotted. It may mean that 2C has a second, lower affinity membrane-binding site which is charge-dependent somewhere outside the MBD. We now added a mention of this in the discussion, lines 321-323.

From the author's data on lipid drop clustering they conclude ΔAH1 is more effective for clustering, however, the ΔAH1 construct produces pentamers not hexamers (from Figure 2). Is formation of hexamers related to or required for membrane clustering?

ΔAH1 is LESS effective at clustering, not more. As for the mention of pentamers in the original submission: we now think this was an unfortunate choice of words. The mass photometry data for 2C(ΔAH1) could more parsimoniously be interpreted as a mix of hexamers and other (unknown to us) smaller oligomers such as trimers. We have removed all mentions of pentamers.

The replicon data presented in Figure 7 should include a replication-defective control (e.g., polymerase mutant), in order to compare how defective in replication ΔAH1 and ΔMBP deletions are compared to a fully-defective construct. Likewise, deletion of ΔAH1 in this construct is likely to affect processing of the viral polyprotein where several previous studies with picornaviruses have demonstrated that the residues in the P2'-P4' positions can change cleavage efficiency (e.g., PMID: 2542331), or the structure of 2C, leading to the reduction of replication.

Thanks for these good comments. We made the polymerase-dead (GDD-to-GAA) replicon and remeasured it side by side with the 2C replicons. It has a similar luciferase activity indicating that no replication takes place in the 2C deletion replicons. This is shown in the new figure 7. As for the possibility or processing defects, we mentioned this in the original discussion and have now cited the reference suggested by the reviewer in this context (line 324).

How does the author's model of ATPase-independent helicase activity and an APT-dependent required RNA chaperone activity fit with 2 step model for RNA binding and ATPase activity suggested by Yeager et al (PMID: 36399514)?

Acting upon comments from other reviewers, we completely redid the "helicase assay" in the revised manuscript. It turns out that the ATP-independent unwinding activity in the original submission was an artefact of the assay conditions (specifically, of the TEV protease at the higher concentration we used in the old assay). In our improved assay we neither see helicase activity nor ATP-independent RNA chaperoning activity.

Optional major comments that would increase the significance of the work:

All of the optional comments below are exceptionally interesting. But given the long time needed for the several major changes to this manuscript (e.g. the ΔAH2 protein characterization and reoptimisation of the helicase assay) we believe it is more sensible to address them in future studies, for which the 2C reconstitution system can be used.

The preference for dsRNA over ssRNA appears to be quite small (Figure 5d). In the context of a viral infection where ssRNA is likely to outnumber dsRNA at different times during infection is this preference physiologically relevant? In relation to this, what size stretch of dsRNA is required for preference, and could this correspond to cis-acting RNA structural elements, dsRNA as it escapes 3D polymerase or as part of the RF and RI forms (PMID: 9343205)? What is the proposed mechanism of how dsRNA outcompetes membrane tethering of 2C? OPTIONAL The author's study has been conducted in the absence of other viral non-structural proteins. What is the physiological importance of the observations, such as membrane interaction/clustering or RNA binding when presented in the context of the other replication machinery. OPTIONAL Do 2C monomers, dimers and hexamers have different functions in viral replication perhaps at different stages of replication and which of these forms are relevant during viral infection or can they all be detected during infection? Can any suggested separate functional arrangements be separated by genetic complementation experiments? OPTIONAL

Minor comments:

1. The author's appear to interchange between naming/nomenclature of the constructs which makes it confusing to follow (for example, ΔMBD is the same as 2C(41-329) likewise, 2C(Δ115) is sometimes called 2C(116-329)). It would be much easier to follow if the naming of constructs was consistent throughout (unless I am misunderstanding some subtlety in the difference between such constructs).

Thanks very much for spotting this. We have fixed it.

The author's suggest a pentamer arrangement for the ΔAH1 construct, however in the mass photometry data (figure 2D), a hexamer is indicated with the arrow. It would be helpful to change the label to indicate the size of the pentamer where this is being generated, not the hexamer.

As mentioned above, we think the "pentamer" designation of the original manuscript was unfortunate. It is more parsimonious to interpret this as a mix of states, hexamer and undefined snaller.

In most figures, data for full-length 2C, ΔAH1 and ΔMBP is shown. However data for ΔMBP is missing in Figure 4. Using ΔMBP may demonstrate even lower clustering, hinting that AH2 is also involved in this process.

Thanks for this comment. In our view, it can be derived from figure 3 (which shows lack of binding to PC/PE membranes) that the ΔMBD construct would not cluster membranes under the conditions of the assay (clustering requires concomitant binding to two membranes). We now describe our rationale for this on lines 220-222. However, we did include the ΔMBD protein in the new negative staining TEM supplementary figure where it and ΔAH2 show no signs of clustering (figure S10).

I think it would be better for normalise the data in the flotation experiments such that the percentage of 2C in the upper faction is presented as relative to the amount of lipid in the upper fraction (presented in Figure S4).

The change suggested by the reviewer would make it impossible to show the important no-liposome control (leftmost bar in Fig. 3C) in the same plot as the other measurements. We believe that would unnecessarily complicate the figure. Thus, we opted to keep the measurement that are normalised by lipid fluorescence in the supplementary figure. Instead, we now added another mention of this supplementary figure in the legend to main figure 3.

At several places (e.g., lines 232 and 272) the author's refer to "realistic systems". I think the term "physiologically relevant" might be more appropriate.

Agreed and changed throughout.

Line 237: I think "y" is a typo and should read "by".

Thanks. This text was reworked due to the major changes to figure 6.

Reviewer #1 (Significance (Required)):

Significance

I have limited expertise with structural biology but specialise my research on positive-sense RNA virus replication, structure and function. This research is of interest to a broad audience of researchers investigating many positive-sense RNA viruses, which extends beyond the viral family studied here. The work utilises novel techniques to begin to understand the specific roles of 2C in poliovirus replication. The author's data add important incremental new insight into recent studies on viral helicase proteins as referenced in the study, however, a key limitation is understanding the importance/relevance of their observations during a viral infection.

We thanks the reviewer for this positive and nuanced appraisal of our work.

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

The authors present an alternative assay system to investigate picornavirus 2C, a protein that is tricky to analyze biochemically in its full length form because of an amphipathic helix at the N-terminus. Poliovirus 2C is expressed with an N-terminal MBP tag, a 50kD protein that helps with solubility as is commonly used for 2C investigations. A difference here is that liposomes are included to mimic membranes for 2C attachment. The key findings are that 2C induces clustering of of liposomes, that double stranded RNA binding by 2C impacts this clustering effect and that a free N-terminus (after cleavage of MBP by TEV protease) is needed for RNA binding and an ATP independent (ie non helicase) RNA duplex separation activity.

Major:

In the floatation assays in figure 3 the authors use a system where MBP-2C is fluorophore-labeled with ATTO488 on exposed cysteines. Poliovirus and other enterovirus 2C has a very well characterized zinc finger domain that has cysteines coordinating a zinc ion. Mutation experiments previously showed that these cysteines are necessary for viral replication and 2C stability. Have the authors controlled for disruption of the zinc finger domain by the labelling of cysteines with ATT0488 and checked if the protein remains folded?

We completely agree with the reviewer and apologise for the omission in the original submission. We have now included a Zn content measurement, which shows unchanged levels between labelled and unlabelled 2C protein (Figure S7). Also, we now in the revised manuscript explicitly describe our original reasoning for labelling on native cysteines: the presence of two cysteines which are not necessary for viral replication and which are more solvent exposed-exposed (and thus more likely to be labelled) in the crystal structure of the soluble fragment of 2C (lines 176-181).

In the analysis of the amphipathic helix, did the authors include membranes in their structural predictions o just the free helix? How does inclusion of membranes impact the predictions? In the predictions in Figure D, only 2 of 4 show a kink and there doesn't seem to be a correlation between those that predict a kink or not and whether the hydrophobic side is aligned in Figure S1.

Unfortunately, predicting a protein structure with the interacting membrane is beyond what is currently doable with protein prediction methods (one would have to combine protein structure predictions with molecular dynamics simulations including a membrane). Based on general principles of protein structure, it is likely that there is some flexibility around G17. Thus there may not be a single "kink angle" for any given virus, but we believe that the presence of the kink (and offset hydrophobic surfaces) for a number of viruses lends credibility and robustness to the observation. We added some descriptions of this thinking on lines 126-127.

Based on previous structures of 2C from different viruses the N-terminal amphipathic helix containing region is predicted to localize on one face of the predicted hexametric structure tethering 2C to the membrane. How does the authors hypothesized model explain 2C dependent clustering? is there evidence that 2C hexamers can oligomerize further into dodecamers for example, maintaining separate faces to enable N-terminal interaction with different membranes? What is the distance between the liposomes in figure 4 at the points of density attributed to 2C? How does this compare to the size of 2C determined in previous structural studies? Is it consistent with one hexamer/2 hexamers sitting on top of one another?

These are very interesting questions but we believe it is prudent to limit our speculation at this point. Eventually, we hope that larger data sets of cryo-electron tomography, coupled to subtomogram averaging, may provide a more definitive answer. What we managed to do with our current cryo-electron tomography data set is to estimate the volume of individual protein densities, and from the volume calculate an estimated molecular mass of the individual complexes seen in the tomograms. This correlates very well with 2C hexamers (new figure 4D).

In the Discussion lines 278-285 the authors suggest that having MBP attached may reflect the polyprotein condition. Can they make a construct with MBP-2B2C to examine interaction with liposomes and assess 2C function?

This is a highly relevant question, but the biochemistry of 2BC is even more challenging than 2C, and we are unfortunately nowhere near being able to work with purified 2BC at the moment.

Discussion lines 293-296, the possibility of two different populations of 2C, binding RNA or membranes cannot be excluded, there is much more 2C around late in infection that present in early infection- the model in figure 8 doesn't acknowledge/capture this.

We have changed the model figure such that more 2C is seen later, and the clustering function is also seen late in infection. The original discussion text referred to (which is unchanged) talks about a "preferential role in RNA replication and particle assembly at later time points" specifically for this reason. We hope the new figure 8 is better at conveying this message.

Discussion lines 313-317, the authors don't reference a study where a mutant of foot-and-mouth disease virus 2C lacking the n-terminal amphipathic helix that could bind but not hydrolyze ATP, hexamerized in the presence of RNA that seems pertinent here (PMID: 20507978).

Thanks for the suggestion. However, after the extensive changes we made to the revised to figure 6 based on excellent reviewer comments (essentially: the RNA chaperoning activity turned out to be an artefact, the improved assay shows no sign of RNA unwinding but instead of 2C-mediated ribonuclease activity), these sentence of the original discussion lost most of their context and we opted to remove them.

Some evidence of MBP-2C cleavage by TEV in the different assays used should be presented as this is a major focus of discussion and currently no gels show TEV cleavage is happening.

Thanks for the suggestion - we agree. We now show these in the new supplementary figures S5 and S12.

Reviewer #2 (Significance (Required)):

The work presents an additional methodology to investigate a a protein that has previously been difficult to study. The authors acknowledge that there is still a lot of 2C biology that remains to be discovered.

Thanks, we agree.

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

The manuscript provides insights into the role of the N-terminus in membrane binding and its importance in the various functions of 2C.

Major issues

Line 103-119. Is this novel? I thought people had done a lot of bioinformatic analysis of PV 2C (especially Wimmer) who also did mutational work to analyse the importance of various amino acids in the N-terminal helix. I feel like the paper in general, and this section in particular, underplays the large body of work that has been done on the amphipathic helix by various groups.

We apologise if our original manuscript didn't sufficiently acknowledge previous work in the field. In the first sentence of the mentioned paragraph (now lines 112-113) , we did however cite several papers that have previously addressed the amphipathic nature of the N-terminus of 2C. We have now added two more references along the same line, and changed the wording in a way that we hope better bring across that the amphipathic nature per se has been studies before. We would be happy to add more specific references if the reviewer has any suggestions. However, the rest of our analysis IS indeed novel for the following reasons: (i) we show that the amphipathic region is not a simple, single amphipathic helix, but instead has a conserved glycine (helix breaker/destabiliser residue) and two distinct amphipathic stretches before and after this region, (ii) we use alphafold2 (not available at the time of the earlier work) to provide the first reliable structural models of the membrane-binding domain. These models consistently, across several enterovirus 2C proteins, reveal that the hydrophobic surfaces of the first and second amphipathic regions, on either side of the conserved glycine 17, are offset from one another. This lends additional credibility to the distinct nature of these regions which have not previously been identified as such and which we also show in the biochemical assays to be functionally distinct. We have now also added a clarification to the Discussion that the N-terminus of 2C had previously been identified as its membrane-binding domain and we cite references for this. We hope that these changes will sufficiently acknowledge earlier work in the field while clearly pointing out the advance that our paper makes.

Line 132. Did you validate your column with known MW standards? The peak for full length and deltaAH1 look fairly standard for 2C, in that you have a mixture of species. Not sure you can say it is a hexamer when it is such a broad peak. C doesn't really help you too much since the counts at 400 (pentamer) and 480 (hexamer) are almost the same with quite large error bars. Like most people that have worked with 2C I think the best you can say is that you are making some kind of oligomerized 2C that includes hexamer, pentamer, etc. Why no dimer for MBP-2C and MBP-2C(delta AH1) when compared to the other constructs?

We did not calibrate the gel filtration column since the outcome would anyway be a more crude estimate of molecular mass than the mass photometry and SEC-MALS measurements. But we do agree with the reviewer on the broad mass photometry peaks. To address this experimentally, we compared the existing MBP-2C spectra to new recordings on apoferritin, a highly stable homomultimeric protein complex of a similar mass to aa MBP-2C hexamer. The apoferritin mass estimate is overlayed with the full-length MBP-2C in the new figure 2D and the corresponding supplementary figure S3. This indeed shows that the MBP-2C peak is broader, i.e. consistent with a mix of species which are predominantly but not only hexamers. We describe and discuss this on lines 145-149. As for the mention of pentamers in the original submission: we now think this was an unfortunate choice of words. The mass photometry data for 2C(ΔAH1) could more parsimoniously be interpreted as a mix of hexamers and other (unknown to us) smaller oligomers such as trimers. We have removed all mentions of pentamers.

Line 143. Does your data show that there are two amphipathic helices? Bioinformatics suggests it but your experiments just show the importance of the two areas in oligomerization, not that it is forming two helices.

We agree that the choice of words was not idea and have now changed it to "structure predictions indicate" (lines 162).

Figure S2. Your preps are still relatively dirty, which isn't ideal for biochemical assays. Especially lane 3, where you are looking at 50-60% purity. I don't want you to re-run experiments but I think you need to comment on the purity of the protein you are working with. Also I don't like that you removed the top and bottom of the SDS-PAGE. How much protein never entered the gel. Is there a big fat band at 20 kDa? You need to have the full gel here. Did you measure 260 nm of the preps as well to see if you had bound RNA to the 2C?

Thanks for the comment, we agree that our original submission lacked detail in the description of the protein purification. This is now addressed with the new figure S2 which shows size exclusion chromatograms of the fluorophore-labelled proteins (same chromatograms as in figure 2) and the corresponding uncropped gels imaged both in the stain-free channel (showing all proteins) and in the fluorescence channel. The A260/A280 ratio measured for all proteins shows that they are free of nucleic acids at the point of imaging. The protein preps are not 100% homogeneous but we do believe that they are more than 50-60% pure.

Lines 170. Wasn't this done in the recent "An Amphipathic Alpha-Helix Domain from Poliovirus 2C Protein Tubulate Lipid Vesicles"? I don't see it referenced. What is novel about the current work when compared to that paper? Any differences?

Thanks for pointing this out. The referenced study worked with a synthesized, isolated peptide corresponding to AH2 (i.e. not with full protein). An amphipathic peptide outside the context of its protein cannot be expected to recapitulate the properties of the entire protein, e.g. since it is not spatially constrained in how it interactis with membranes. As one example (relating to the title of that paper) we don't see full-length 2C protein tubulating membranes the way the isolated peptide does. As for the reviewer's question about novelty, the paper mentioned does not identify the split nature of the amphipathic region, does not consider the role of AH1, does not characterise the membrane-binding properties of full-length 2C with respect to liposome membrane composition and size, does not identify and characterise the membrane clustering properties of 2C, nor its interactions with nucleic acid when bound to a membrane. However, we do agree that we should have cited the paper in our manuscript. We now cite it in the discussion, lines 320-321.

I'm surprised by the lack of electron microscopy (negative stain mostly) of both the oligomerized 2C and the various liposomes. I know the Carlson group is a microscopy group so why the lack of validation using electron microscopy of the various DLS experiments? I know you did cryo-ET for one of the constructs but I think negative stain electron microscopy of other constructs would be useful.

Thanks for the suggestion. As suggested, we have now expanded the analysis with negative staining EM of several more constructs studied by DLS. It can be found in the new supplementary figure S10.

Figure 4C. What evidence is there that this is 2C apart from you added it to the liposomes? It also comes back to the relative impurity of your protein prep. Could this be E.coli contamination?

Thanks for this comment. We have now added a new supplementary figure (S5) showing SDS-PAGE gels of the reactions used for flotation and DLS assays - which are identical to the cryo-ET samples. In addition, we estimated the molecular mass of the individual, putative 2C desities in the cryo-electron tomograms by measuring their volume. This analysis, which can be found in the new figure 4D, shows that the estimated mass of individual protein densities is consistent with a hexamer of full-length 2C. In addition, we mention in the discussion the long-term need to determine high-resolution structures of membrane-bound 2C using cryo-ET and subtomogram averaging (lines 315-318).

Figure 8. Is this model supported by the data in this paper? Your cryo-ET says that 2C is there but that isn't supported by any other data. How is the dsRNA protected from the innate immune system in this model? is it just sat out in the cytosol? How is the nascent ssRNA packeged into the capsid? Is there competition between the dsRNA and capsid for 2C binding (which your model suggests)? I know it sounds like I am being overly critical of the model but in my opinion there are still too many unanswered questions in the field to come up with a half decent model.

Thanks for this comment. We are the first to agree that our understanding of the roles of 2C is far from complete! We should have been more clear that the model figure represents some of the roles of 2C identified to date, and does not claim to be complete. However we do feel that a model figure serves a purpose of putting our findings into a context, and also providing testable hypotheses for future research . As for the question, some of the roles of 2C shown in the model figure (in particular, particle assembly) are rather supported but earlier work of ourselves and others. We have now produced a new model figure and changed the figure legend to better reflect the incompleteness of the current understanding, and the origin of the different parts of the model figure. In addition, we extended the final paragraph of the discussion (which lists still-unknown aspects of 2C) with the reviewer's mention of dsRNA shielding from innate immunity (lines 374-375). The other aspects mentioned by the reviewer as not yet fully understood are already mentioned in that paragraph.

Minor issues

Lines 43-45: I feel like you underplay the success of the poliovirus vaccination program. Approximately 30 of WPV1 in 2022 and the full eradication of WPV2 and 3. Vaccine derived polio is still an issue but even that is relatively low compared to where the world was in the 1950s.

We agree that the previous wording was not ideal. We replaced it and added another recent reference - related to the type 2 vaccine switch (lines 47-49).

Line 66. I agree there are 11 individual proteins but I feel like this leaves out the fact that some of the uncleaved precursors appear to have some functions, for example 2BC.

Good point. We have now added a mention of 2BC and the fact that it has distinct functions to the introduction (lines 70-71). 2BC is also mentioned in the legend of the model figure (figure 8).

Line 56: LD needs to be defined.

Well spotted thanks. Since the abbreviation was not used anywhere else we opted to spell it out instead (line 59).

Line 75. I think you have misrepresented Xia et al here. They clearly say that in their study that they show helicase and chaperone activity. I never managed to repeat that work but you should still report what they claim. One major thing is that they used insect expressed protein, whereas most people (including myself and in the paper under review) use E.coli expressed protein. Do post translational modifications play an important role in function?

You are right that the reference to their paper for this statement was incorrect. We have now made this part of the introduction more explicit (lines 82-83) and we also in the new discussion mention the possibility of e.g. post-translational modifications affecting 2C helicase activity, under reference to Xia et al (lines 359-361)

Line 103. Need to make it clear here it is poliovirus 2C.

Thanks, we added it (line 112).

Line 135. I assume you mean kDa instead of uM?

It should actually be μM. It is the solution concentration at which the assay was performed. We added some words to clarify this (line 154).

Figure 3. What do you mean by "Only 2C"? Is that MBP-2C? Maybe I am reading the data wrong but adding TEV does nothing? How do you know TEV is removing the MBP? It looks like MBP-2C binds to the liposomes just the same as cleaved MBP-2C. I see in line 165 you acknowledge this. Could an alternative conclusion for line 168 be that MBP isn't being cleaved off but that AH2 is too small to be exposed in that construct? Did you do that construct without MBP being cleaved? I think you need to confirm that MBP is being cleaved off.

Thanks for spotting this mistake. It should indeed be MBP-2C (in the absence of liposomes). We corrected figure 3. Also, in response to this comment and similar ones, we have now added a new supplementary figure showing SDS-PAGE gels of the reaction loaded onto flotation assays and DLS (figure S5). It shows that MBP-2C is cleaved.

Line 184. Is there a reason you use the 2019 paper as a reference instead of the far earlier Bienz et al papers? I'd suggest they are the seminal papers on 2C membrane association. Once again how is this work different from the recent "An Amphipathic Alpha-Helix Domain from Poliovirus 2C Protein Tubulate Lipid Vesicles" paper?

See our response above of the paper mentioned here (which we have now cited). As for why we cite the 2019 paper here: our statement pertains specifically to the contact sites between lipid droplets and replication organelles, not to the membrane binding of 2C per se. We have now added a more general mention of membrane remodelling by non-structural proteins in the introduction, where we cite on of the Bienz papers (lines 75-77).

Figure 5D. So only 1-3% of RNA is found in the upper fraction? Is that significant enough to say that dsRNA was recruited significantly more than ssRNA? How confident are you in your quantification of the starting amounts of RNA?

We agree that the fraction is low, however, the fluorescence signal is very clearly above background. We are thus confident in the measurement. The low percentage at the end of the experiment likely has a simple physico-chemical explanation: in a dynamic equilibrium in a density gradient, whatever RNA dissociates during the run will migrate away from the 2C-vesicle fraction and not be able to rebind. We still tried to address this concern by a complementary experiment where we used fluorescence anisotropy to measure binding of RNA to 2C on vesicles. While the measurements showed the same tendency, they curves were not clean enough to be published, which we think is due to the complex system with 2C bound to vesicles and clusters of vesicles. Still, in view of the relatively low percentage of measured recruitment we opted to adjust the paper title and the title of figure 5 (including the subheading related to figure 5) to put less emphasis on the dsRNA recruitment.

Line 223. Any idea why the MBP needs to be cleaved off? Clearly the MDB is accessible or it would not bind to the liposomes.

Since we have no data directly supporting this we prefer not to speculate in the paper. But one guess would be that the NTD of 2C, as implicated by previous publications, has a dual role in membrane binding and RNA binding. It may be that it can bind membrane while conjugated to MBP, but needs MBP to be removed in order to simultaneously bind membrane and RNA.

Line 237: missing "b" in "by"

Thanks. This paragraph was rewritten in the light of the changes to figure 6.

Figure 6. I don't fully understand the results here. Earlier you showed that the delta MBD didn't really bind SUV. So presumably it isn't really membrane bound. Why does it have similar activity to full-length MBP in your helicase assay if membrane is important? Did you do SUV and TEV protease only control?

We are very grateful to this reviewer (and others) for pointing out the need for a TEV control. When performing the control, we found that the TEV protease, at the high concentrations initially used, surprisingly had an artefactual RNA chaperone-like effect on its own. We then proceeded to titrate down the TEV protease concentration to the point where it no longer interfered. At this TEV protease concentration, although 2C was substantially cleaved (see the new supplementary figure S12), we could no longer detect an RNA chaperone activity. Thus, the contents of the new figure 6, and its conclusions, have been substantially changed. We now focused our attention on the remaining effect that 2C has on RNA: single-strand ribonuclease activity. These experiments were all conducted in the presence of RNase inhibitors, and the presence of Mg2+-dependent ribonuclease activity parallels a recent publication that found this for truncated 2C from hepatitis A and several enteroviruses.

Line 257: "staring"?

Thanks, corrected. A staring glycine would indeed be something strange.

Line 336. Need to change the u to mu.

Thanks, corrected.

Any discussion on your observation in Figure 1D that EV71 and CVB3 don't appear to have AH1 and AH2 or do you think that the domains are conserved across the different viruses?

Thanks for bringing this up. Based on this and a comment from another reviewer, we have now clarified our thinking around this. Since the glycine will introduce some flexibility between AH1 and AH2, we cannot say from the single alphafold predictions that this is THE kink angle. The presence of the kink in the predictions of several MBDs lends more credibility to the robustness of the observation, but most importantly the hydrophobic surfaces in AH1 and AH2 are non-aligned for ALL sequences we looked at. This is now described on lines 126-128.

Table 1 (and possibly elsewhere): an apostrophe is not the prime symbol. 5' compared to 5′.

Thanks, we corrected this throughout.

Line 702 "and" should be "an".

Thanks, corrected.

I couldn't open one of the movies (140844_0_supp_2820374_a2g272.avi).

Sorry to hear this, we will check the movie again.

Reviewer #3 (Significance (Required)):

Overall I liked the paper and is worth publishing. One of the issues in the 2C field is the difficulty in making pure 2C and carrying out in vitro assays that correlate with what is observed in the natural infection. I think this paper suffers from similar struggles with a 2C preparation that doesn't appear that pure. I think it also suffers from not having 2C from a wild-type infection. I don't think that it is feasible to get that kind of 2C but by once again using a recombinant protein from E.coli we are left with another manuscript that provides conflicting evidence of the functions of 2C without a definitive answer. The experiments are well done, although are missing some controls and the manuscript is laid out in a logical manner and is relatively easy to follow.

We thanks the reviewer for these comments. We believe that we have now provided better information regarding the purification of the recombinant 2C protein, and we do think that the controls present in the original manuscript and the revised manuscript alleviate the concerns about lack of specificity. Of course, isolating 2C vesicles from wildtype infection would be another interesting way of approaching its function, but such an approach would come with its own set of challenges related e.g. to the presence of confounding host factors.

Reviewer #4 (Evidence, reproducibility and clarity (Required)):

This is an interesting manuscript that reports the development of an in vitro membrane assay for probing the biochemical functions of the enterovirus 2C protein. The technique is interesting because it can be applied to 2C proteins from other members of the picornavirus family, an important group of mammalian pathogens. It has the capacity to probe different functions (e.g. membrane clustering, ATPase activity, RNA-binding and manipulation activities).

Overall, the manuscript is well written and gives a clear account of the work undertaken. It adds insight to previous studies of enteroviral (and picornaviral) 2C proteins, providing confirmation of some earlier work in a more physiological context and some new insights, particularly into the membrane and RNA binding aspects of 2C.

That said, there are a number of places where some amendment of the claims made is required to provide a more precise statement of the findings of this work. These are listed below.

We thank the reviewer for this positive feedback on our work, as well as for the specific comments below.

Line 21 (Abstract) - The authors claim to have shown that a conserved glycine divides the N-terminal membrane-binding domain into 2 helices. I would suggest instead what they have produced are computational predictions that this is the case - some way short of an experimental demonstration. Sequence analysis predicts helical secondary structure in the N-terminus and indeed Alphafold2 also predicts a helical structure, but these predictions require experimental verification. The authors should therefore rewrite sections that claim to have shown the presence of 2 helices. In doing so, they should perhaps also comment on the fact that Alphafold2 does not predict 2 helices in this region for all enteroviruses (see Fig 1D). Moreover, the sequence analysis in Fig. S1 shows the presence of two Lys residues in the segment 17-38; it would be interesting for the reader to have these indicated in the figures showing the Alphafold2 prediction - do they in any way interrupt the hydrophobic face of the predicted helix?

Thanks very much for this comment, which is in line with what other reviewers also wrote. We agree, and changed the abstract sentence. We have also rewritten the manuscripts in several places to address the limits of structure predictions and the eventual need for an experimental structure of full-length membrane-bound 2C (lines 126-128 and 315-318).

Line 82 (Introduction) - The authors write that the membrane binding domain (MBD) of poliovirus has been shown to mediate hexamerisation, citing Adams et al (2009) - reference 43. However, that is not what this paper shows. Rather it provides evidence of aggregation of an MBP-2C fusion protein into forms that ranged from tetramer to octamer, but no evidence that these aggregates assume functional forms (e.g. the presumed hexameric ring structure characteristic of the AAA+ ATPase family to which 2C belongs). As far as I am aware the first demonstration of hexameric ring formation by a picornaviral 2C protein was for the 2C of foot-and-mouth disease virus (see Sweeney et al, JBC, 2010). Although this is not an enterovirus, this finding was later confirmed for Echovirus 30 (ref 51). I should declare an interest here: the Sweeney paper is from my lab. I will leave it to the editor and the authors to determine how to write a more precise account of the early observations of hexamerisation in picornaviral and enteroviral 2C proteins.

Thanks very much for this insightful comment. As a response to this and other similar comments, we are much more cautious about our wording in the revised manuscript (see also response to comment below. In the part of the introduction discussed here (now lines 89-91) we now use the original wording of the Adams paper ("oligomerization"). In the context of that new text we didn't feel that Sweeney et al paper was a suitable reference, but we now cite it in the later mention of 2C's oligomeric/hexameric state in the first part of the Results (lines 137-138 ).

Line 132 - the authors used mass photometry to investigate oligomeric forms of their MBP-2C constructs and state that for the full length 2C protein "the high-mass peak closely corresponds to a hexamer". While it is true that the peak shown in Fig 2C aligns with the expected MW for an MBP-2C hexamer, the peak is very broad, indicative of the presence of other oligomeric states with lower and higher numbers of monomers. This should be commented on. Indeed, the finding seems to echo the early findings of Adams et al (ref 43) with poliovirus MBP-2C.

Thanks for this comment, which was also made by another reviewer. We cite here what we replied to that reviewer

...we do agree with the reviewer on the broad mass photometry peaks. To address this experimentally, we compared the existing MBP-2C spectra to new recordings on apoferritin, a highly stable homomultimeric protein complex of a similar mass to aa MBP-2C hexamer. The apoferritin mass estimate is overlayed with the full-length MBP-2C in the new figure 2D and the corresponding supplementary figure S3. This indeed shows that the MBP-2C peak is broader, i.e. consistent with a mix of species which are predominantly but not only hexamers. We describe and discuss this on lines 145-149.

Line 143 - for the reasons given above, this summary paragraph represents too strong a statement of what has been observed.

We agree, and changed the paragraph. It now only refers to "oligomerization" (lines 162-164).

Line 197 - I note that the authors did not test the membrane clustering capabilities of the 2C(41-329) construct. Although the 2C(deltaAH1) construct had already shown a significant loss of activity, the shorter construct could still have been a useful control. I don't think it is necessary for this experiment to be done, but if the authors have a rationale for not performing the experiment, perhaps they could include it in a revised manuscript.

Thanks for the suggestion. The rationale is that a protein that doesn't bind a membrane in the first place will also not cluster them (an action that requires binding TWO membranes). We now describe our reasoning on lines 220-222. Nevertheless, we did test these constructs in the new supplementary figure showing negative staining TEM (figure S10).

Line 223 - typo. I think you mean MBD.

Thanks! Corrected (now line 257).

Line 215 - the authors observed that the presence of ssDNA reduced membrane clustering and conclude that "nucleic acid binding partially outcompetes membrane tethering activity". Two things: (1) although I agree is it likely that this effect is due to binding of DNA to 2C, binding has not been demonstrated experimentally so the authors should be more careful in how they describe their result; (2) there is no data presented to show that RNA binding reduces membrane tethering so at best I think the conclusion has to be that the data are consistent with the notion that DNA binding reduces membrane tethering. It would of course be interesting to see the effects of RNA and I'm curious to know why the assay was not performed.

Thanks for the comment. The honest answer is that previous publications (primarily Yeager et al, NAR 2022) convinced us that the outcome should be near-identical with DNA, so we chose DNA oligos because they are cheaper and easier to work with. But we agree with the reviewer that RNA is of course more relevant. We now present a comparison at 5 μM of ssDNA and ssRNA, which in fact shows a slightly stronger effect on membrane clustering by RNA (figure 5C). In the light of this additional experiment, we feel that some of the text changes suggested by the reviewer may no longer be necessary.

Line 237 - typo: by, not y

Thanks. In the light of the extensive changes to figure 6 this text was removed.

Line 284 - the authors claim that 2C may only bind RNA after the N-terminus is liberated from 2B in infected cells, since cleavage of the MBP tag from their construct was needed for 2C to bind RNA in their in vitro assay. However, this does not automatically follow given the large structural differences between MBP and 2B and the fact that the authors have not tested the RNA binding capacity of a 2BC fusion protein. Their claim here is too strong and should be re-written.

We agree, and have added a discussion along the lines suggested by the reviewer (line 330-332).

Line 293 - The authors speculate that RNA binding might cause a shift between the membrane clustering activities and the role of the protein in RNA replication. However, since they have not shown that RNA binding reduces membrane clustering, this is too speculative.

In our revised manuscript we have studied the effect of RNA on membrane binding, thus we feel that this text is relevant in the context of the extended experiments.

Line 299-317 - within this discussion is the assumption that in their assay system enterovirus 2C adopts the ring-like hexameric structure typical of AAA+ ATPases. While I agree this may well be the case, it has not been demonstrated in this study so the authors should make clear they are making this assumption. The same applies to the legend of Fig 8.

This part of the discussion was extensively rewritten after our changes to figure 6. We now only refer to "hexamer" once in the corresponding part of the discussion, where we talk about structural models of hexamers produced by other groups who have crystallised fragments of 2C. There we believe we should refer to hexamers to accurately cite their work.

We are not sure what the reviewer is referring to when it comes to the legend for figure 8: the original legend had no reference to the oligomeric state of 2C. We have substantially changed figure 8 and its legend and the new figure and legend make no references to hexamers/oligomers.

Line 302 - the authors claim to have shown that 2C is 'selective' for dsRNA. I think at best they have shown a preference for binding dsRNA over ssRNA.

We changed the wording (line 349). We have also changed the title of the paper where we removed "double-stranded".

Line 313 - The sentence starting "A recent study..." needs a reference.

The revised discussion no longer contains this sentence.

Line 332 - the full sequence of the synthetic gene used in this study should be made available (e.g. as supplementary information or a deposited sequence with an accession number). This is a critical point before the paper can be published.

We will of course submit the sequences as supplementary data. Thanks for the reminder.

Line 362 - the authors should describe the likely points of attachment of fluorophores and comment on how this labelling might affect 2C function.

Thanks for the comment. In response to this and a similar comment from another reviewer, we discuss the likely conjugation site of the fluorophore (lines 175-181), and also (due to the proximity to the Zn finger) provide a new measurement showing that equal amounts of Zn can be detected in the labelled and unlabelled protein (figure S7).

Line 372 - Is a single protein standard (BSA) sufficient to calibrate the SEC-MALS system?

Yes, it is the recommended procedure (note that SEC-MALS is only dependent on scattering, not elution volumes etc).

Reviewer #4 (Significance (Required)):

As stated above this is an interesting study that presents findings from a novel assay. It will be of interest to picornavirologists and the wider community interested in the mechanisms of AAA+ ATPases.

We thanks the reviewer for this positive appraisal of our work.

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

Learn more at Review Commons

Referee #4

Evidence, reproducibility and clarity

This is an interesting manuscript that reports the development of an in vitro membrane assay for probing the biochemical functions of the enterovirus 2C protein. The technique is interesting because it can be applied to 2C proteins from other members of the picornavirus family, an important group of mammalian pathogens. It has the capacity to probe different functions (e.g. membrane clustering, ATPase activity, RNA-binding and manipulation activities).

Overall, the manuscript is well written and gives a clear account of the work undertaken. It adds insight to previous studies of enteroviral (and picornaviral) 2C proteins, providing confirmation of some earlier work in a more physiological context and some new insights, particularly into the membrane and RNA binding aspects of 2C.

That said, there are a number of places where some amendment of the claims made is required to provide a more precise statement of the findings of this work. These are listed below.

Line 21 (Abstract) - The authors claim to have shown that a conserved glycine divides the N-terminal membrane-binding domain into 2 helices. I would suggest instead what they have produced are computational predictions that this is the case - some way short of an experimental demonstration. Sequence analysis predicts helical secondary structure in the N-terminus and indeed Alphafold2 also predicts a helical structure, but these predictions require experimental verification. The authors should therefore rewrite sections that claim to have shown the presence of 2 helices. In doing so, they should perhaps also comment on the fact that Alphafold2 does not predict 2 helices in this region for all enteroviruses (see Fig 1D). Moreover, the sequence analysis in Fig. S1 shows the presence of two Lys residues in the segment 17-38; it would be interesting for the reader to have these indicated in the figures showing the Alphafold2 prediction - do they in any way interrupt the hydrophobic face of the predicted helix?

Line 82 (Introduction) - The authors write that the membrane binding domain (MBD) of poliovirus has been shown to mediate hexamerisation, citing Adams et al (2009) - reference 43. However, that is not what this paper shows. Rather it provides evidence of aggregation of an MBP-2C fusion protein into forms that ranged from tetramer to octamer, but no evidence that these aggregates assume functional forms (e.g. the presumed hexameric ring structure characteristic of the AAA+ ATPase family to which 2C belongs). As far as I am aware the first demonstration of hexameric ring formation by a picornaviral 2C protein was for the 2C of foot-and-mouth disease virus (see Sweeney et al, JBC, 2010). Although this is not an enterovirus, this finding was later confirmed for Echovirus 30 (ref 51). I should declare an interest here: the Sweeney paper is from my lab. I will leave it to the editor and the authors to determine how to write a more precise account of the early observations of hexamerisation in picornaviral and enteroviral 2C proteins. Line 132 - the authors used mass photometry to investigate oligomeric forms of their MBP-2C constructs and state that for the full length 2C protein "the high-mass peak closely corresponds to a hexamer". While it is true that the peak shown in Fig 2C aligns with the expected MW for an MBP-2C hexamer, the peak is very broad, indicative of the presence of other oligomeric states with lower and higher numbers of monomers. This should be commented on. Indeed, the finding seems to echo the early findings of Adams et al (ref 43) with poliovirus MBP-2C.

Line 143 - for the reasons given above, this summary paragraph represents too strong a statement of what has been observed.

Line 197 - I note that the authors did not test the membrane clustering capabilities of the 2C(41-329) construct. Although the 2C(deltaAH1) construct had already shown a significant loss of activity, the shorter construct could still have been a useful control. I don't think it is necessary for this experiment to be done, but if the authors have a rationale for not performing the experiment, perhaps they could include it in a revised manuscript.

Line 223 - typo. I think you mean MBD.

Line 215 - the authors observed that the presence of ssDNA reduced membrane clustering and conclude that "nucleic acid binding partially outcompetes membrane tethering activity". Two things: (1) although I agree is it likely that this effect is due to binding of DNA to 2C, binding has not been demonstrated experimentally so the authors should be more careful in how they describe their result; (2) there is no data presented to show that RNA binding reduces membrane tethering so at best I think the conclusion has to be that the data are consistent with the notion that DNA binding reduces membrane tethering. It would of course be interesting to see the effects of RNA and I'm curious to know why the assay was not performed.

Line 237 - typo: by, not y

Line 284 - the authors claim that 2C may only bind RNA after the N-terminus is liberated from 2B in infected cells, since cleavage of the MBP tag from their construct was needed for 2C to bind RNA in their in vitro assay. However, this does not automatically follow given the large structural differences between MBP and 2B and the fact that the authors have not tested the RNA binding capacity of a 2BC fusion protein. Their claim here is too strong and should be re-written.

Line 293 - The authors speculate that RNA binding might cause a shift between the membrane clustering activities and the role of the protein in RNA replication. However, since they have not shown that RNA binding reduces membrane clustering, this is too speculative.

Line 299-317 - within this discussion is the assumption that in their assay system enterovirus 2C adopts the ring-like hexameric structure typical of AAA+ ATPases. While I agree this may well be the case, it has not been demonstrated in this study so the authors should make clear they are making this assumption. The same applies to the legend of Fig 8.

Line 302 - the authors claim to have shown that 2C is 'selective' for dsRNA. I think at best they have shown a preference for binding dsRNA over ssRNA.

Line 313 - The sentence starting "A recent study..." needs a reference.

Line 332 - the full sequence of the synthetic gene used in this study should be made available (e.g. as supplementary information or a deposited sequence with an accession number). This is a critical point before the paper can be published.

Line 362 - the authors should describe the likely points of attachment of fluorophores and comment on how this labelling might affect 2C function.

Line 372 - Is a single protein standard (BSA) sufficient to calibrate the SEC-MALS system?

Significance

As stated above this is an interesting study that presents findings from a novel assay. It will be of interest to picornavirologists and the wider community interested in the mechanisms of AAA+ ATPases.

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 provides insights into the role of the N-terminus in membrane binding and its importance in the various functions of 2C.

Major issues

Line 103-119. Is this novel? I thought people had done a lot of bioinformatic analysis of PV 2C (especially Wimmer) who also did mutational work to analyse the importance of various amino acids in the N-terminal helix. I feel like the paper in general, and this section in particular, underplays the large body of work that has been done on the amphipathic helix by various groups.

Line 132. Did you validate your column with known MW standards? The peak for full length and deltaAH1 look fairly standard for 2C, in that you have a mixture of species. Not sure you can say it is a hexamer when it is such a broad peak. C doesn't really help you too much since the counts at 400 (pentamer) and 480 (hexamer) are almost the same with quite large error bars. Like most people that have worked with 2C I think the best you can say is that you are making some kind of oligomerized 2C that includes hexamer, pentamer, etc. Why no dimer for MBP-2C and MBP-2C(delta AH1) when compared to the other constructs?

Line 143. Does your data show that there are two amphipathic helices? Bioinformatics suggests it but your experiments just show the importance of the two areas in oligomerization, not that it is forming two helices.

Figure S2. Your preps are still relatively dirty, which isn't ideal for biochemical assays. Especially lane 3, where you are looking at 50-60% purity. I don't want you to re-run experiments but I think you need to comment on the purity of the protein you are working with. Also I don't like that you removed the top and bottom of the SDS-PAGE. How much protein never entered the gel. Is there a big fat band at 20 kDa? You need to have the full gel here. Did you measure 260 nm of the preps as well to see if you had bound RNA to the 2C?

Lines 170. Wasn't this done in the recent "An Amphipathic Alpha-Helix Domain from Poliovirus 2C Protein Tubulate Lipid Vesicles"? I don't see it referenced. What is novel about the current work when compared to that paper? Any differences?

I'm surprised by the lack of electron microscopy (negative stain mostly) of both the oligomerized 2C and the various liposomes. I know the Carlson group is a microscopy group so why the lack of validation using electron microscopy of the various DLS experiments? I know you did cryo-ET for one of the constructs but I think negative stain electron microscopy of other constructs would be useful.

Figure 4C. What evidence is there that this is 2C apart from you added it to the liposomes? It also comes back to the relative impurity of your protein prep. Could this be E.coli contamination?

Figure 8. Is this model supported by the data in this paper? Your cryo-ET says that 2C is there but that isn't supported by any other data. How is the dsRNA protected from the innate immune system in this model? is it just sat out in the cytosol? How is the nascent ssRNA packeged into the capsid? Is there competition between the dsRNA and capsid for 2C binding (which your model suggests)? I know it sounds like I am being overly critical of the model but in my opinion there are still too many unanswered questions in the field to come up with a half decent model.

Minor issues

Lines 43-45: I feel like you underplay the success of the poliovirus vaccination program. Approximately 30 of WPV1 in 2022 and the full eradication of WPV2 and 3. Vaccine derived polio is still an issue but even that is relatively low compared to where the world was in the 1950s.

Line 66. I agree there are 11 individual proteins but I feel like this leaves out the fact that some of the uncleaved precursors appear to have some functions, for example 2BC.

Line 56: LD needs to be defined.

Line 75. I think you have misrepresented Xia et al here. They clearly say that in their study that they show helicase and chaperone activity. I never managed to repeat that work but you should still report what they claim. One major thing is that they used insect expressed protein, whereas most people (including myself and in the paper under review) use E.coli expressed protein. Do post translational modifications play an important role in function?

Line 103. Need to make it clear here it is poliovirus 2C.

Line 135. I assume you mean kDa instead of uM?

Figure 3. What do you mean by "Only 2C"? Is that MBP-2C? Maybe I am reading the data wrong but adding TEV does nothing? How do you know TEV is removing the MBP? It looks like MBP-2C binds to the liposomes just the same as cleaved MBP-2C. I see in line 165 you acknowledge this. Could an alternative conclusion for line 168 be that MBP isn't being cleaved off but that AH2 is too small to be exposed in that construct? Did you do that construct without MBP being cleaved? I think you need to confirm that MBP is being cleaved off.

Line 184. Is there a reason you use the 2019 paper as a reference instead of the far earlier Bienz et al papers? I'd suggest they are the seminal papers on 2C membrane association. Once again how is this work different from the recent "An Amphipathic Alpha-Helix Domain from Poliovirus 2C Protein Tubulate Lipid Vesicles" paper?

Figure 5D. So only 1-3% of RNA is found in the upper fraction? Is that significant enough to say that dsRNA was recruited significantly more than ssRNA? How confident are you in your quantification of the starting amounts of RNA?

Line 223. Any idea why the MBP needs to be cleaved off? Clearly the MDB is accessible or it would not bind to the liposomes.

Line 237: missing "b" in "by"

Figure 6. I don't fully understand the results here. Earlier you showed that the delta MBD didn't really bind SUV. So presumably it isn't really membrane bound. Why does it have similar activity to full-length MBP in your helicase assay if membrane is important? Did you do SUV and TEV protease only control?

Line 257: "staring"?

Line 336. Need to change the u to mu.

Any discussion on your observation in Figure 1D that EV71 and CVB3 don't appear to have AH1 and AH2 or do you think that the domains are conserved across the different viruses?

Table 1 (and possibly elsewhere): an apostrophe is not the prime symbol. 5' compared to 5′.

Line 702 "and" should be "an".

I couldn't open one of the movies (140844_0_supp_2820374_a2g272.avi).

Significance

Overall I liked the paper and is worth publishing. One of the issues in the 2C field is the difficulty in making pure 2C and carrying out in vitro assays that correlate with what is observed in the natural infection. I think this paper suffers from similar struggles with a 2C preparation that doesn't appear that pure. I think it also suffers from not having 2C from a wild-type infection. I don't think that it is feasible to get that kind of 2C but by once again using a recombinant protein from E.coli we are left with another manuscript that provides conflicting evidence of the functions of 2C without a definitive answer. The experiments are well done, although are missing some controls and the manuscript is laid out in a logical manner and is relatively easy to follow.

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 present an alternative assay system to investigate picornavirus 2C, a protein that is tricky to analyze biochemically in its full length form because of an amphipathic helix at the N-terminus. Poliovirus 2C is expressed with an N-terminal MBP tag, a 50kD protein that helps with solubility as is commonly used for 2C investigations. A difference here is that liposomes are included to mimic membranes for 2C attachment. The key findings are that 2C induces clustering of of liposomes, that double stranded RNA binding by 2C impacts this clustering effect and that a free N-terminus (after cleavage of MBP by TEV protease) is needed for RNA binding and an ATP independent (ie non helicase) RNA duplex separation activity.

Major:

In the floatation assays in figure 3 the authors use a system where MBP-2C is fluorophore-labeled with ATTO488 on exposed cysteines. Poliovirus and other enterovirus 2C has a very well characterized zinc finger domain that has cysteines coordinating a zinc ion. Mutation experiments previously showed that these cysteines are necessary for viral replication and 2C stability. Have the authors controlled for disruption of the zinc finger domain by the labelling of cysteines with ATT0488 and checked if the protein remains folded?

In the analysis of the amphipathic helix, did the authors include membranes in their structural predictions o just the free helix? How does inclusion of membranes impact the predictions? In the predictions in Figure D, only 2 of 4 show a kink and there doesn't seem to be a correlation between those that predict a kink or not and whether the hydrophobic side is aligned in Figure S1.

Based on previous structures of 2C from different viruses the N-terminal amphipathic helix containing region is predicted to localize on one face of the predicted hexametric structure tethering 2C to the membrane. How does the authors hypothesized model explain 2C dependent clustering? is there evidence that 2C hexamers can oligomerize further into dodecamers for example, maintaining separate faces to enable N-terminal interaction with different membranes? What is the distance between the liposomes in figure 4 at the points of density attributed to 2C? How does this compare to the size of 2C determined in previous structural studies? Is it consistent with one hexamer/2 hexamers sitting on top of one another?

In the Discussion lines 278-285 the authors suggest that having MBP attached may reflect the polyprotein condition. Can they make a construct with MBP-2B2C to examine interaction with liposomes and assess 2C function?

Discussion lines 293-296, the possibility of two different populations of 2C, binding RNA or membranes cannot be excluded, there is much more 2C around late in infection that present in early infection- the model in figure 8 doesn't acknowledge/capture this.

Discussion lines 313-317, the authors don't reference a study where a mutant of foot-and-mouth disease virus 2C lacking the n-terminal amphipathic helix that could bind but not hydrolyze ATP, hexamerized in the presence of RNA that seems pertinent here (PMID: 20507978).

Some evidence of MBP-2C cleavage by TEV in the different assays used should be presented as this is a major focus of discussion and currently no gels show TEV cleavage is happening.

Significance

The work presents an additional methodology to investigate a a protein that has previously been difficult to study. The authors acknowledge that there is still a lot of 2C biology that remains to be discovered.

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 study by Shankar and colleagues, the authors aim to understand the structure and function of the enterovirus 2C protein, a putative viral helicase with AAA+ ATPase activity. Using poliovirus (as a model enterovirus) 2C, the author's propose the protein contains two amphipathic helices (AH1 and AH2) at the N-terminus that are divided by a conserved glycine. Using purified MBP-tagged 2C and N-terminal 2C truncations, their data suggests AH1 is primarily responsible for clustering at membranes, whilst AH2 is the main mediator of 2C oligmerisation and membrane binding. Furthermore, 2C was suggested to be able to recruit RNA to membranes, with a preference for dsRNA, and the author's data implies that the helicase activity of 2C is ATP-independent. Instead, the ATP activity appears to be required for 2C hexamer formation or chaperone activity. The manuscript is generally well written /presented and the author's present very interesting data which raises several questions, some of which require additional experimentation to help support the author's conclusions. Specific comments are as follows.

Major Comments:

1. The authors use four main constructs throughout the paper: full-length 2C, 2C with deletion of AH1 (ΔAH1), 2C with both AH1 and AH2 deleted (ΔMBP) and 2C with an extended N-terminal deletion. From this, the author's draw conclusions on the function of both AH1 and AH2. One of the author's main conclusions is that AH2 is the main mediator of 2C membrane association (e.g., in line 169). However, is it possible to conclude the relative importance of AH1 vs AH2 without testing a construct containing the deletion of AH2 only (ΔAH2)? This should be generated and used alongside this data to fully define the relative importance of AH1 and AH2 in these assay and remove the possibility that the deletion of AH1 changes the structure and/or function of AH2, which could also result in the observed differences.
2. Previous structural predictions of 2C do not appear to have two separate AHs at the N-terminus. Are the AH1 and AH2 structures predicted to be formed in the context of the entire 2C protein, 2BC precursors and polyprotein? Are there structural approaches that could provide experimental evidence for two separate AH at the N-terminus?
3. Why are the 2C dimers (lines 137-138) not apparent on the mass photometry data presented (figure 2)?
4. It appeared that binding of ΔMBD-2C was better when POPS is in the membrane (line 174). What is the explanation for this and was this finding significant?
5. From the author's data on lipid drop clustering they conclude ΔAH1 is more effective for clustering, however, the ΔAH1 construct produces pentamers not hexamers (from Figure 2). Is formation of hexamers related to or required for membrane clustering?
6. The replicon data presented in Figure 7 should include a replication-defective control (e.g., polymerase mutant), in order to compare how defective in replication ΔAH1 and ΔMBP deletions are compared to a fully-defective construct. Likewise, deletion of ΔAH1 in this construct is likely to affect processing of the viral polyprotein where several previous studies with picornaviruses have demonstrated that the residues in the P2'-P4' positions can change cleavage efficiency (e.g., PMID: 2542331), or the structure of 2C, leading to the reduction of replication.
7. How does the author's model of ATPase-independent helicase activity and an APT-dependent required RNA chaperone activity fit with 2 step model for RNA binding and ATPase activity suggested by Yeager et al (PMID: 36399514)? Optional major comments that would increase the significance of the work:
8. The preference for dsRNA over ssRNA appears to be quite small (Figure 5d). In the context of a viral infection where ssRNA is likely to outnumber dsRNA at different times during infection is this preference physiologically relevant? In relation to this, what size stretch of dsRNA is required for preference, and could this correspond to cis-acting RNA structural elements, dsRNA as it escapes 3D polymerase or as part of the RF and RI forms (PMID: 9343205)? What is the proposed mechanism of how dsRNA outcompetes membrane tethering of 2C? OPTIONAL
9. The author's study has been conducted in the absence of other viral non-structural proteins. What is the physiological importance of the observations, such as membrane interaction/clustering or RNA binding when presented in the context of the other replication machinery. OPTIONAL
10. Do 2C monomers, dimers and hexamers have different functions in viral replication perhaps at different stages of replication and which of these forms are relevant during viral infection or can they all be detected during infection? Can any suggested separate functional arrangements be separated by genetic complementation experiments? OPTIONAL

Minor comments:

1. The author's appear to interchange between naming/nomenclature of the constructs which makes it confusing to follow (for example, ΔMBD is the same as 2C(41-329) likewise, 2C(Δ115) is sometimes called 2C(116-329)). It would be much easier to follow if the naming of constructs was consistent throughout (unless I am misunderstanding some subtlety in the difference between such constructs).
2. The author's suggest a pentamer arrangement for the ΔAH1 construct, however in the mass photometry data (figure 2D), a hexamer is indicated with the arrow. It would be helpful to change the label to indicate the size of the pentamer where this is being generated, not the hexamer.
3. In most figures, data for full-length 2C, ΔAH1 and ΔMBP is shown. However data for ΔMBP is missing in Figure 4. Using ΔMBP may demonstrate even lower clustering, hinting that AH2 is also involved in this process.
4. I think it would be better for normalise the data in the flotation experiments such that the percentage of 2C in the upper faction is presented as relative to the amount of lipid in the upper fraction (presented in Figure S4).
5. At several places (e.g., lines 232 and 272) the author's refer to "realistic systems". I think the term "physiologically relevant" might be more appropriate.
6. Line 237: I think "y" is a typo and should read "by".

Significance

I have limited expertise with structural biology but specialise my research on positive-sense RNA virus replication, structure and function. This research is of interest to a broad audience of researchers investigating many positive-sense RNA viruses, which extends beyond the viral family studied here. The work utilises novel techniques to begin to understand the specific roles of 2C in poliovirus replication. The author's data add important incremental new insight into recent studies on viral helicase proteins as referenced in the study, however, a key limitation is understanding the importance/relevance of their observations during a viral infection.

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 provides a detailed analysis of RNA and protein dynamics during transmission of the rodent malaria model P. yoelii from the mouse host to an in vitro ookinete culture setting (mimicking the mosquito midgut environment). This group and others have shown experimentally that a substantial number of mRNAs is stored in the female Plasmodium gametocyte, ready to be translated following initiation of ookinete development. The process is akin to maternal deposition of mRNA in oocytes of metazoans. With this manuscript the authors provide a significant contribution to the field of translational control in Plasmodium parasites as they explore the translational activation during the early hours of zygote-to-ookinete development. The paper presents RNAseq and mass-spec analyses of female gametocytes and for the first time for 6-hour zygotes (ie a fertilized female gamete); the zygote datasets are much improved and more comprehensive than the only other performed in 2008 in P. gallinaceum. Using comparative analyses of transcriptome and proteome data (including published datasets) the authors arrive at a list of 198 transcripts that are translationally repressed in the gametocyte and translated within 6 hours of fertilization in the zygote. Many of these mRNAs are known to be involved in zygote to ookinete transformation. BioID is finally used to explore changes in mRNP protein composition between the female gametocyte and the zygote.

The paper is generally well written. The authors present a lot of data (also in comparison with published data). Sometimes perhaps the main message could be simplified / streamlined in section titles (Quantitative Proteomics by DIA-MS is not very informative. The outcome of the analysis would be more telling).

Response: We have revised section headers to clarify the content.

A considerable proportion of the DIA mass-spec proteomics results section is very technical. The paper describes a biological phenomenon rather than a technical mass-spec advance. Can these technical details be moved to the methods section?

Response: As this is one of the first published instances of using DIA-MS to Plasmodium, we want to keep this information in the main text to help our community adopt these approaches. While these details are highly technical, they are also some of the major advances of this project.

On the other hand, a bit more detail could be provided in the main text. For example, the age of the zygotes is never mentioned. This is important, please add this. The main manuscript text has 16 mentions of the word "many". As the authors are in possession of the data, please provide, if missing, (in parenthesis) the absolute numbers, maybe in an "x out y" format. Please clearly state the number of biological and/or technical replicates used for transcriptome and proteome analyses in the main text, figures and/or figure legends. How many protein coding genes are encoded in the P. yoelii genome?

Response: Several of these requested details are noted in the materials and methods. We have added this information to the main manuscript now as well. We have also revised the manuscript to replace some instances of “many” with specific numbers unless it adversely impacted the flow of the sentence to do so.

The authors claim that only zygotes (fertilized females) have surface-exposed Pys25 (a surface protein they use to affinity-purify zygotes) but not gametocytes. I could not find the experimental data for this in the paper. The cited reference #22 also does not appear to show this. In Figure 2C Pys25 is shown to be translated in gametocytes. In this context it may be important to note that in the related P. berghei the related protein P28 is expressed even in the absence of fertilization (Billker 2004; DOI: 10.1016/s0092-8674(04)00449-0). It may not be relevant whether translation requires fertilization, but the authors claim it affects trafficking of the Pys25 protein to the surface, so it needs to be shown. A reference to an infertile P. yoelii line would be great.

Response: We have corrected the reference supporting the surface exposure of p25 on zygotes. The observation by Billker and colleagues about Pbs28 is also of interest, but outside of the scope of this study as we did not investigate the fertilization event itself here.

It is highly commendable that all data is provided throughout the manuscript. For readability, may I suggest that the authors add labels to individual sheets within an excel file from A to Z, and do so also within the manuscript. That would really help; the most relevant data sets could then be identified quickly. For example, line 184 refers to 276 zygote proteins in which sheet of which table?

Response: While this labeling system would also be effective, we have provided a README tab for our files that quickly directs the reader to the relevant tab (as we do for our previous publications).

Section 176 onwards: here the authors combine P. falciparum and P. yoelii proteomics data. Please explain why you excluded any of the available P. berghei proteome data such as the male and female gametocyte proteome? The same question applies to 294 onwards.

Response*: We compared our datasets with those of Lasonder et al. NAR 2016 because that study was also focused on translational repression of mRNAs and provided both RNA-seq and proteomic datasets of female gametocytes (although not of zygotes). *

The comparative transcriptome-proteome analysis arrives at 198 translationally repressed mRNAs. Could the authors provide one or two alternatives using less stringent parameters? The list in P. falciparum and P. berghei is considerably larger (500+ and 700+).

Response: We could have reduced the stringency of our thresholds to arrive at a far larger number, but prefer to retain higher confidence in those we are scoring as translationally repressed and then released for translation. We provide all of the pertinent data in the supplemental files if readers would like to adjust these thresholds to see which additional mRNAs may also be regulated.

The turboID data is informative but somewhat speculative in regard to spatial rearrangements within these mRNPs. Figure 6 presents the RNA helicase to bind the 5' end of mRNAs that are associated with polyribosomes and I assume being translated. Is this association realistic? The RNA helicase DOZI homolog of yeast (Dhh1) is also involved in decapping. Response: We provide Figure 6 as our working model of how the reorganization of the DOZI/CITH/ALBA complex could occur based on available data from this study and others. Future studies are warranted to determine if DOZI remains associated with monosomes vs. polysomes, but current data indicate that DOZI can bind to eIF4E when translational repression is not imposed.

Specific comments:

title Is global the appropriate word? Some transcripts appear to be translated later.

Response: We believe it does apply appropriately to these data.

Line 30/32 Please re-phrase the sentence. There is: Cell Host Microbe 2012 Jul 19;12(1):9-19. doi: 10.1016/j.chom.2012.05.014.

Response: We conclude that the sentence is correct as written, even in considering Sebastian et al. Cell Host & Microbe 2012.

30 Perhaps add ookinete that establishes infection rather than the zygote. For a general readership, a brief description of the sexual life cycle might be useful

Response: It is not possible to get into these nuances in the Abstract. This information is covered in the main text and the works that are cited.

32 DOZI/CITH/ALBA complex would require some explanation for a more general reader

Response: It is not possible to get into these nuances in the Abstract. This information is covered in the main text and the works that are cited.

36-37 I believe zygotes were collected 6 hours after fertilization. Does that qualify as soon after fertilization? Motile ookinetes are generated within 20 hours and motility can be seen before that.

Response: Yes, we think this qualifies as the process is not synchronous, but relies on when male gametes encounter and fuse with female gametes.

37 Essential functions for what?

Response: It is not possible to get into these nuances in the Abstract. This information is covered in the main text and the works that are cited.

39 Is the spatial arrangement of this mRNP known?

Response*: Some interactions of members of this complex were known (DOZI with eIF4E, ALBA4 with PABP1), but not the overall spatial arrangement. These findings are novel to this study. *

40 Can you briefly allude to the "recent, paradigm-shifting models of translational control"

Response: It is not possible to get into these nuances in the Abstract. This information is covered in the main text and the works that are cited.

44 Products = mRNA

Response: We have stated it as products because the maternal cell provides more than just mRNAs that are essential to further development post-fertilization.

45 Oocyte in metazoans ?

Response: Yes, this is the correct term. The context here is in higher eukaryotes.

60/62 Please re-phrase the sentence. There is: Cell Host Microbe 2012 Jul 19;12(1):9-19. doi: 10.1016/j.chom.2012.05.014.

Response: We conclude that the sentence is correct as written, even in considering Sebastian et al. Cell Host & Microbe 2012.

81 PbDozi Plasmodium berghei DOZI

Response: We have added this clarifying text here as suggested.

84/85 Please rephrase and cite Nucleic Acids Res. 2008 Mar;36(4):1176-86. doi: 10.1093/nar/gkm1142. Epub 2007 Dec 23. and Cell Host Microbe 2012 Jul 19;12(1):9-19. doi: 10.1016/j.chom.2012.05.014.

Response: As noted above for other comments, we hold that the current phrasing is accurate even when considering these important publications.

88 Please define the timepoints throughout this manuscript. What age are the zygotes? How many hours post-induction? Please define the time for ookinete development somewhere in the introduction

Response: The timepoint used for zygote collection is now included in the main text in addition to its previous inclusion in the Materials and Methods section. As we have not studied the ookinete stage here, we have opted to keep the introduction focused on the key details for this study.

104 Please add the age (in hours) of these zygotes from the time of starting the in vitro cultures. From the methods section it looks like 6 hours.

Response: The timepoint used for zygote collection is now included in the main text in addition to its previous inclusion in the Materials and Methods section.

103/105 I can find no evidence for P25 (Pys25) expression relying on fertilization in the cited paper (22). The SOM has no reference to Pys25 either. Please show data or reference published data that there is no translation and trafficking of Pys25 in unfertilized female gametes, ie those that are placed in ookinete medium. In this respect it may be important to note that unfertilized Plasmodium berghei females placed in ookinete medium translate P28, the P25 paralog (https://www.sciencedirect.com/science/article/pii/S0092867404004490?via%3Dihub)

Response: We have corrected the reference supporting the surface exposure of p25 on zygotes. The observation by Billker and colleagues about Pbs28 is also of interest, but outside of the scope of this study as we did not investigate the fertilization event itself here.

104 What cell line was used for the zygotes?

Response*: The PyApiAP2-O::GFP transgenic parasite line was used here. These details are included in the manuscript and supporting information. *

114 The number of transcripts detected in gametocytes is quite small compared to the twice as large proteomics dataset. See for example also Lasonder 2016 for P. falciparum detected transcripts: 4477 different sense transcripts were identified, 98% of which were shared between MG and FG.

Response: Yes, the number of mRNAs or proteins scored as detected differs based on thresholds applied. We prefer to err on the side of higher stringency as noted above.

117 Does the 194 up-in-gametocytes dataset include the 81 not found in zygotes?

Response: No, these 194 are detected in both datasets, but are more abundant in gametocytes than zygotes.

117 Could you indicate some of the genes in the plot?

Response: Several hits of special note are described in the text. We have opted to keep the figure clear and streamlined.

Fig1 How were the upregulated transcripts identified? 1647 are shown to be specific to zygotes in 1B, yet only 685 are shown in 1C to be upregulated. Do the transcripts found exclusively in zygotes not count? Are these transcripts likely the result of de novo transcription? How old are these zygotes when the libraries are made?

Response: The details of the RNA-seq processing are provided in the MakeFile, the supplementary tables, and the manuscript. The README tab provides descriptions of what processing occurred between sequential tabs. As noted above, zygotes were collected at 6 hours.

132 Many? How many? Please provide a precise number.

Response: These details are now in the revised manuscript.

134 Please explain why p28 would be differentially abundant in the zygote rather than the female gametocyte. That would require de novo transcription of this gene. If there is experimental evidence for the de novo transcription of p28 and other translationally repressed transcripts in the zygote please cite the references. Can you name a few more examples here? P25 for example, ap2-o, or anything published and experimentally validated. What about AP2-o and AP2-Z? Both are known to be translationally repressed.

Response: We state in the original manuscript that there is not a significantly different mRNA abundance of pys28.

139 Please define how many members of the IMC?

Response*: We have now replaced “many” with the number of IMC members we have detected, which is also shown in supporting tables. *

156 Can you provide a number of how many parasites were used in total or per run. And how many biological and technical replicates were analysed?

Response: These details are provided in the Materials and Methods.

169 The number of proteins detected in the gametocyte sample is twice the size of transcripts. IS this to be expected?

Response*: This reflects the sensitivity of the assays run for transcriptomics and proteomics. *

170 How many samples were analyzed? One gametocyte and one zygote sample?

Response: Yes, for the creation of the DIA-MS spectral library, a single biological replicate was used in addition to in silico library approaches. This information is provided in the next sentence.

176 Why did you not include P. berghei in the meta-analysis?

Response: We compared these results to all of the published Plasmodium proteomes in PlasmoDB.

184 Please refer to an excel table here.

Response: We have pointed to the relevant supporting files in this section.

184 145 proteins: do you mean orthologs in general or orthologs with a gene/protein annotation other than unknown function?

Response: We use the standard form of ortholog throughout the manuscript.

190 142 proteins: do they all have orthologs in P. falciparum?

Response: No, not all proteins in our dataset have unambiguous orthologues in P. falciparum, and this is accounted for in our data processing approaches.

Figure 2C P25 is not exclusive to zygotes here and also found in the gametocyte sample.

Response: That is correct. It is known that p25 is expressed in female gametocytes, but that the localization changes in the zygote.

190 shortlist

Response: The spelling of “short list” as two words is an appropriate American spelling of this term.

219 onwards Does the list of 198 transcripts exclusively arise from your RNAseq and proteomics comparison? Or does it include falciparum data as outline in section 176 onwards, ie the list of 276 proteins that only are detected in zygotes?

Response: Yes, this list of 198 mRNAs is derived from our datasets only using our defined thresholds. The details of this are provided in the manuscript.

224 Early zygote? At 6 hours do the parasites not start to transform, elongate?

Response: This process is not synchronous, as it is affected by the timing of gamete fusion.

225 >5-fold. Is this an arbitrary decision?

Response: This threshold has been used by our group and others in prior studies, and was partially informed by the behavior of previously characterized transcripts.

227 1417 mRNAs: they are from which dataset?

Response: These are from our datasets with P. yoelii, as described in the manuscript.

228/229 Please explain why DOZI and CITH are in the list of 198 repressed transcripts? They are present in the gametocyte. Are they upregulated>5 fold?

Response: Yes, they meet our criteria for this regulation, and in the manuscript we note that we believe that they are self-regulated and likely have continuing roles in early mosquito stage development.

259 ... as they are already translated in the gametocyte?

Response: Yes. Translational repression allows for the existence of some of the protein in the initial timepoint. This differs from translational silencing which does not.

295 Is this from the 198 TR list S4?

Response: No. Transcripts that remain repressed would not be in the list of 198, as the protein was not detected in zygotes.

294 onwards How many putatively falciparum transcripts are there? How many were identified in P. berghei? How many are common to all? A Venn diagram perhaps to compare the different studies

Response: There is substantial overlap between the species with respect to the presence of syntenic orthologues in this dataset. However, because we did not conduct experiments with P. falciparum or P. berghei here, we do not want to make claims that they are similarly regulated or potentially have a reader misinterpret a figure to that effect.

301 How many transcripts were found associated with Plasmodium berghei DOZI and/or CITH in female gametocytes? How many of those were abundantly detected as protein in zygotes, or had no difference in protein abundance between gametocytes and zygotes, or even greater abundance in female gametocytes?

Response: These details are now provided in the revised manuscript.

303/305 Please indicate the numbers of translationally repressed transcripts identified for P. falciparum and berghei.

Response: These data are provided in Supporting Information Table 4.

317/319 Please add the promoter used for tid-GFP

Response: We have now added this information to the Materials and Methods.

320 Please elaborate on the spatial organization of the DCA complex.

Response: This has not been previously characterized, and this entire section is dedicated to the experimental data and interpretations of how the DOZI/CITH/ALBA complex may be organized.

321/322 Have precise binding sites of DOZI and ALBA4 really been shown experimentally in the cited papers? In relation to 5' and 3' ends of the mRNA? Please cite Braks et al. paper.

Response: Yes. The association of DOZI with eIF4E and ALBA4 with PABP1 are established in the literature, in some cases by multiple independent laboratories. The Braks publication does not address the binding of these proteins, and thus is not cited.

323 What is the first generation BioID enzyme? BirA*

Response: Yes. The first generation enzyme is called BirA*

323 Please cite relevant Kyle Roux and Alice Ting for the original enzymes

Response: We have now added these citations to this sentence.

327 Could you show images of ALBA4::TurboID::GFP, DOZI::TurboID::GFP and cytosolic (free) TurboID? Perhaps stained with fluorescently labelled streptavidin and / or against GFP? In the gametocyte and zygote samples?

Response: We attempted to stain with monoclonal antibodies that are reactive against biotin and there was insufficient specificity, hence why such data is not included. We conclude that all of the other data that supports this approach suffices to demonstrate its rigor.

331 What is the age of these zygotes? Where they affinity purified?

Response: As throughout the manuscript, zygotes were collected at 6 hours. Details of experimental purifications are provided in the materials and methods.

Fig S4 Please indicate whether ALBA4 and DOZI were tagged endogenously

Response: Yes. The endogenous loci for both ALBA4 and DOZI were modified to include the C-terminal TurboID and GFP tags.

421/430 Please add a few references here

Response: We do not believe that specific references are warranted for these general statements.

429 translational repression?

Response: Yes. These statements set the stage for the use of translational repression.

445 966 proteins in gallinaceum? The zygote cultures in that study were 2-3 hours. How old were the cultures in your study?

Response: As throughout the manuscript, zygotes were collected at 6 hours.

481 Please explain / cite why repression is energetically costly.

Response: These details are provided in both the introduction and discussion sections. The energetic cost of translational repression is both the cost to produce the transcripts without immediately/fully utilizing it for translation, in addition to the energetic cost to impose the regulation.

501 Please add the time-point of RNA and protein sampling. How many hours into ookinete development? What is the time from cardiac puncture through FACS sampling of gametocytes.

Response: We have provided all of these details in the materials and methods for female gametocytes and zygotes. We did not look at ookinetes in this study.

711/713 Do you have any images that show the successful purification of zygotes away from gametocytes? Secondly, please provide a reference for the statement that unfertilized female gametocyte do not express surface exposed Pys25.

Response*: We do not have captured images of these zygotes, but confirmed them during collection using microscopy. The reference for surface exposure of Pbs25 is now provided earlier in the manuscript as well. *

711/716 Were parasites lysed and mechanically homogenised?

Response: We have provided all of these details in the materials and methods for female gametocytes and zygotes.

Figure 6 What is the evidence that DOZI stays associated with mRNA that is being translated? Rather than mRNA that is being decapped. Please add the references that unequivocally show that DOZI and ALBA4 bind to opposite ends of repressed mRNAs.

Response: This is our working model of these data. It is feasible that these complexes could form off of mRNA as well. Publications describing the interactions of DOZI with eIF4E and ALBA4 with PABP1 are provided in the manuscript. It is well established that eIF4E binds to the m7G cap of the 5’ end of mRNAs, and PABP1 binds to the poly(A) tail at the 3’ end of mRNAs.

Reviewer #1 (Significance (Required)):

The experiments in the manuscript are carefully conducted. Apart from a P. gallinaceum study from 2009 this is the first comprehensive analysis of the transcriptome and proteome of a Plasmodium zygote (developing ookinete) at 6 hours post-fertilization. The data are used to explore the temporal aspect of activation of translation during the first quarter of the 20-24 hour ookinete developmental period. The study will be of interest to the field, specifically those scientists working to understand translational control, ookinete development, and those developing intervention strategies to prevent mosquito infection and thus malaria transmission.

Response: We appreciate Reviewer 1’s extensive feedback and positive remarks about the significance of our study. We have revised our manuscript to reflect this constructive feedback.

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

Main findings

Taking a multi-omic approach, the authors provide quantitative evidence for translation repression of ~200 mRNAs in Plasmodium yoelii female gametocytes. These mRNAs are then translated, and proteins detected by 6 hours after activating gametocytes. They accomplish this by performing a comparative global analysis of the transcriptome and proteome between female gametocytes and early zygotes that provides an intresting resource. The authors also use proximity labelling of the DOZI/CITH/ALBA4 repression complex, and these data suggest the complex may disassemble in the zygote or change its composition.

Major points

Line 181-184: The authors state that there is no evidence of how the DCA complex selects specific mRNAs for translation repression. While the exact mechanisms have not been fully elucidated, Braks et al (2008, doi:10.1093/nar/gkm1142) suggested a role of the untranslated regions (UTRs) in translation repression of transcripts in Plasmodium berghei female gametocytes. They identified a uridine-rich 47-base element in the 5'UTR and or 3'UTR that was associated with translationally repressed transcripts and validated it experimentally. Considering this finding, I would recommend an amendment of the statement and to include the earlier work. I would also like to see additional analysis to check if this U-rich motif or other motifs are associated with the translationally repressed transcripts identified in the current study. The current study should be better powered to conduct such an analysis.

Response: We have now added a comment and citation in the revised text about this study in Lines 86-88. Understanding the full importance of this element is challenging, as the Plasmodium transcriptome is highly enriched in A’s and U’s due to the highly skewed A/T content of its genome. Perhaps for this reason, we did not see an association of this motif with the identified mRNAs.

The authors used zygotes that expressed GFP tagged AP2-O, however, there is no explanation of the significance of using this line.

Response: This line is described in the Materials and Methods and supporting information. It was used to provide further validation of the production of zygotes.

Minor points

In line 106-107, the authors refer to figure SI, this figure is about genomic locus and genotyping PCR for the PyApiAP2-O::GFP parasites but there is no intext description of why this specific line was used.

Response: We have provided this information in the revised manuscript.

Statement in line 122-124 "It is likely that....." should go into the discussion not results.

Response: We have placed this single sentence immediately after presenting these data here to aid reader comprehension.

Statement in line 171-175: "In addition to providing confirmatory...." Should be in the discussion not on the results.

Response: We view this sentence as a concluding remark of this section of data that also places this information in context for the reader.

In Fig. 4 A and B, could the colour scheme be changed so that the proteins that are not in both samples (and probably contain many unspecifically detected proteins) appear less prominent?

Response: We appreciate this suggestion and have adjusted these plots accordingly in the revised manuscript.

Reviewer #3 (Significance (Required)):

Why is the paper interesting. Translation repression of mRNA at a global level in the female gametocytes has been studied previously in rodent malaria parasites investigated, but prior to the current study, the release of mRNA from translation repression in the mosquito stages has only been demonstrated for specific transcripts. By characterizing and quantitating changes in protein abundance between macrogamete and zygote, coupled with transcriptomic analysis, the current work broadens our understanding of zygotic translation activation that is key to successful malaria parasite transmission to the mosquito.

This dataset provides a useful resource for the Plasmodium research community as it provides a more comprehensive view of how transcripts behave during the transitions from the mammalian host to the vector. It is one step in a broader endeavour towards finding genes crucial for parasite transmission that could be targeted for interventions.

How translational repression and derepression is regulated remains unknown, although some of the molecular players have been identified. This paper shows proximity labelling and expansion microscopy data of the ribonuclear protein complex thought to mediate repression. Although the specific mechanistic insights provided by the experiments shown here remain relatively limited, the work demonstrates interesting new avenues for how translational derepression in Plasmodium can be studied.

Response: We also appreciate Reviewer 3’s excellent feedback and positive remarks about the significance of our study. The revised manuscript addresses these comments, and we believe it is further strengthened because of it.

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