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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements

We thank the reviewers for their thorough and insightful evaluations of our manuscript and for their constructive feedback, which have significantly improved the quality of our manuscript. We were pleased to read that all three reviewers found our work novel, interesting, and relevant. In this revised manuscript, we have done our best to address all of the points raised by the reviewers by performing new experiments and revising sections of the text, as requested.

2. Point-by-point description of the revisions

Reviewer #1 (Evidence, reproducibility and clarity):

In this manuscript authors show that extracellular Mtb aggregates can cause macrophage killing in a close contact dependent but phagocytosis independent manner. They showed Mtb aggregates can induce plasma membrane perturbations and cytoplasmic Ca2+ influx with live cell microscopy. Next, the authors show that the type of cell death initiated by extracellular aggregates is pyroptosis and they partially supressed cell death with pyroptosis inhibitors. They also identified that PDIM, EsxA/EsxB and EspB all have a role in uptake-independent killing of macrophages even though their impact varies with respect membrane perturbation and Ca2+ influx. Finally, they used a small molecule inhibitor BTP15 to inhibit the effect of ESX-1 during the contact of the extracellular Mtb aggregates with the macrophages and they observed a substantial decrease in membrane perturbation and macrophage killing.<br /> The work describes a very interesting mechanism by which Mtb can kill macrophages that is possibly relevant in the context of infection.

1. In general, there are two main issues with the experiments and the interpretation: the lack of quantitative analysis showing that in a population of macrophages the ones that are in contact with the aggregates die whereas the ones that are not in contact remain alive. This is currently not shown, and it should be added in figure 1.

All our data are based on the visual inspection and annotation of time-lapse microscopy image series, from which it is conclusive that death happens more often among cells in contact with Mtb aggregates (see movies S3 and S6 for representative examples). However, we acknowledge the reviewer’s suggestion that quantitative data supporting this observation might help to convey this conclusion more effectively. Therefore, we have quantified the percentage of dead cells in: I) macrophages in uninfected controls; II) macrophages that establish contact with an Mtb aggregate; III) bystander macrophages that never contact an Mtb aggregate despite being in the same sample as the infected cells, in experiments with (figure 1D) or without (figure 1Q) cytochalasin D treatment. These data have been incorporated as two additional plots in figure 1 in the revised manuscript. We find that uninfected and bystander cells have similar survival probabilities over the time-course of an experiment, whereas most of the cells that physically interact with Mtb_aggregates die by the end of the experiment. To further validate these observations, we have also plotted the lifespans of infected cells vs. bystander cells without (figure S3A) and with (figure S3B) cytochalasin D treatment. In these plots, the lifespan of an individual cell is represented by a line; the fraction of the line coloured in black corresponds to the time spent as bystander and the fraction of the line in magenta corresponds to the time spent in contact with an _Mtb aggregate. We hope that these new data convincingly show that bystander cells (black lines) survive longer compared to cells that interact with Mtb aggregates (black-magenta lines).

1. The second is the cell death mode, as the markers used are very different and considering different outcomes (e.g., apoptosis vs. necrosis) are relevant for the infection it is unclear what is being measured here and the impact on bacterial replication.

As the reviewer points out, it has previously been shown that different cell death pathways can affect viability and propagation of intracellular bacteria (1, 2). Since in our experiments we are specifically analyzing extracellular bacteria, we cannot directly comment on how cell death affects intracellular bacterial replication. However, to address the reviewer’s comment, we have included additional data in figure S13A of the revised manuscript showing that specific inhibitors of cell death do not affect the growth or replication of extracellular Mtb. These results suggest that while these molecules do not affect Mtb growth per se, the suppression of these specific death pathways also does not significantly affect the microenvironment to alter Mtb growth (i.e., access to nutrients or molecules released by dead cells). In addition, we have included new data in figure S12 demonstrating the responsiveness of our isolated macrophages to the various cocktails of molecules typically used to induce apoptosis, pyroptosis, or necroptosis.

The authors are showing that infection with Mtb aggregates increase the rate of the macrophage killing but how does this impact infection dissemination and replication of the bacterial aggregates? Is it beneficial for the aggregates? Did the authors check the growth rate of Mtb along with cytochalasin D?

A previous study has shown that phagocytosis of Mtb aggregates leads to macrophage death more efficiently than phagocytosis of a similar number of individual bacteria (3). It has also been shown that Mtb growing on the debris of dead host macrophages forms cytotoxic aggregates that kill the newly interacting macrophages (3). These observations suggest a model in which host cell death induced by Mtb aggregates supports faster extracellular growth and propagation of infection (3). This study was cited in the Introduction section of our manuscript, and our data support these observations. In the revised manuscript, we show that single Mtb bacilli or Mtb aggregates induce macrophage death in a dose-dependent manner (figure S7A,B); however, bacterial aggregates kill more efficiently when compared to similar numbers of non-aggregated bacilli (figure S7A,B). We also show that infection with Mtb_aggregates leads to faster bacterial propagation compared to infection with similar numbers of individual bacteria (figure S7C,D). These observations, combined with our data showing that _Mtb aggregation also enhances uptake-independent killing of macrophages (figure 2), suggest that Mtb aggregates induce rapid host cell death, allowing the bacteria to escape intracellular stresses, grow faster outside host cells (figure S1B), and propagate to other cells. To address the reviewer’s concern whether cytochalasin D affects Mtb growth, the revised manuscript includes additional data confirming that cytochalasin D does not affect the growth of Mtb aggregates (figure S6).

1. How did the authors quantify the interactions of Mtb with macrophages in Figure 1D?

The interactions of Mtb with macrophages were quantified through manual annotation of the time-lapse microscopy image series. If the Mtb aggregates disaggregated upon interaction with the macrophage, resulting in redistribution of smaller aggregates of bacteria, we categorized them as “fragmented”. On the other hand, if the aggregates remained clustered, we categorized them as “not fragmented”. Representative snapshots of these two patterns are presented in figure 1E and 1F and we have included additional representative examples in movies S4 and S5 of the revised manuscript. These interactions are quantified and plotted in figure 1N of the revised manuscript (figure 1D in the original version).

1. Is it enough to conclude with one example of SEM that the mycobacteria with different fragmentation discriminates if the bacteria is intracellular or extracellularly localised? Can authors use an alternative quantitative method to confirm the localization of the bacteria by a quantification by 3D imaging of these two phenotypes with a cytoskeleton marker (or may be even with tdTomato-expressing BMDMs)?

In the revised manuscript, we provide additional examples of correlative time-lapse microscopy and SEM images (supplementary figure S5). As suggested by the reviewer, in the revised manuscript we further validate these conclusions using an alternative approach based on correlative time-lapse microscopy followed by confocal 3D imaging. After time-lapse imaging, we fixed the samples and labelled the plasma membrane of the macrophages with a fluorescent anti-CD45 antibody to define the cell boundaries and identify bacteria that are intracellular vs. extracellular. Representative images obtained using this approach have been added to figure 1 and additional examples are shown in supplementary figure S4 of the revised manuscript. The acquisition, processing, and analysis of these 3D images are time-consuming and prevent us from performing an exhaustive quantitative analysis. However, we are confident in our conclusions, since in all of the cells that we analyzed we found that aggregates that are not fragmented within 6 hours of stable interaction with macrophages are visible on the outer side of the plasma membrane.

1. How do we know if the cell is lysed at 30 h in Supplementary Figure 1, did the authors use a marker to detect the cell lysis or is it based on just the observation from the live cell imaging? Movies in supplementary are actually not very informative as there are many ongoing events and it is hard to visualise what the authors claim. A marker of cell death in the movies should be used.

In this study, we used brightfield time-lapse microscopy images to identify cell death. Dying macrophages rapidly change shape, lose membrane integrity, and stop moving. Moreover, the intracellular structures and bacteria also stop moving at the time of death of the host cell. While these events can be difficult to distinguish by examining individual snapshots, they are readily identifiable by careful frame-by-frame examination of time-lapse microscopy image series. To exemplify this process, in the revised manuscript we show in supplementary figure S2A how we identify macrophage cell death events. We also include Draq7 (a live cell-impermeable dye commonly used to identify dead cells by flow cytometry and microscopy) in the growth medium during time-lapse imaging in order to label dead macrophages. The timing of staining validates and confirms our strategy of using brightfield time-lapse images to define the time-of-death of individual cells. To further assist readers, in the revised manuscript we provide the time-lapse microscopy movie used to generate this figure (movie S4). Similar images and movies have also been added for cells treated with cytochalasin D (figure S2B; movie S7). As suggested by the reviewer, we also replaced figure S1A with a new figure that shows a representative example of an Mtb intracellular microcolony that, upon death of the host macrophage, grows and forms a large extracellular aggregate on the debris of the dead cell (Draq7-positive). Movie S2 was used to generate this figure. Finally, we replaced figures 1E,F with new figures incorporating the Draq7 staining to label macrophage cell death and we include the time-lapse microscopy movies used to generate these figures (movies S4, S5).

1. Total macrophage killing after contact in Figure 1L is around 12 hours, whereas it is observed that the macrophage death after contact with cytochalasin D treatment in Figure 1M is even longer than 24 hours. The viability at 12 hours in Figure1M is as fragmented Mtb survival in Figure1L, why there is a difference in timing with respect to macrophage killing?

We thank the reviewer for this interesting observation. Indeed, we find that macrophages treated with cytochalasin D do take longer to die upon establishing stable interaction with Mtb aggregates in comparison to untreated cells. Although we do not have a clear explanation for this difference in timing, we speculate that by inhibiting actin polymerization and consequently cell motility, cytochalasin D might slow the expansion of the macrophage plasma membrane and the establishment of a larger interface of contact between the cell and the bacterial aggregate, which could influence the timing of cell death.

1. Did authors perform statistical tests for Figure 1D and Figure 1N? p-values should be added.

Figure 1D (figure 1N in the revised manuscript) shows the percentage of interactions between macrophages and _Mtb_aggregates that do or do not lead to fragmentation of the aggregate. Each dot represents the percentage of these events in one experimental replicate. We included this plot to show that reproducibly in all our replicates approximately 20% of the interactions do not lead to fragmentation of the aggregate. Since the purpose of this plot is not to compare the “fragmented” and “non-fragmented” populations but rather to highlight the reproducibility of the phenomenon, we do not think it would be appropriate to add a p-value. However, figure 1N (figure 1Q in the revised manuscript) has been updated and modified to include statistical analysis and a p-value.

1. In Figure 3, do the observations indicated in the Figure 3 happen in all the macrophages that are in contact with aggregates? This is unclear and critical to support the conclusions. Do all the macrophages that are in contact with Mtb aggregates become Annexin-V positive? In Supplementary Figure 2 there is some information regarding this question, but it will be important to show it as a percentage.

In response to the reviewer’s suggestion, we have modified the figure to include quantitation of Annexin-V staining. Approximately 75% of the macrophages that interact with an Mtb aggregate show detectable local Annexin V-positive membrane domains at the site of contact with the aggregate during a typical 60 hour-long experiment. Since most of the macrophages show local Annexin V-positive membrane domains within the first 12 hours upon contact with an Mtb_aggregate (figure 3C), we used this criterion for comparison of different conditions or strains (for example, those shown in figure 6F). In addition, we added figure 3D, which shows the behaviour of 105 macrophages upon contact with _Mtb aggregates in a typical experiment. In this plot, each line represents the lifespan of an individual cell; the fraction of the line in black represents the time spent as bystander, the fraction of the line in magenta represents the time spent interacting with an Mtb aggregate, and the fraction in green represents the time upon formation of local Annexin V-positive membrane domains at the site of contact with the Mtb aggregate. We believe that this additional information further supports our conclusions that most of the cells in contact with an Mtb aggregate show local Annexin V-positive membrane domains and that cells that show this pattern die faster than cells that do not develop local Annexin V-positive membrane domains.

1. Did the authors try to stain Mtb aggregates alone with Annexin-V as a control over the duration of the imaging?

We thank the reviewer for suggesting this control. In supplementary figure S8C of the revised manuscript, we include a representative example of a time-lapse microscopy image series showing Mtb aggregates that never interact with a live macrophage althought they are adjacent to a dead cell. As observed in the Annexin V fluorescence images (yellow), these Mtb aggregates never become Annexin-V positive during the course of the experiment (60 hours).

1. In Figure 4, did the authors continue to image the cells interacting with Mtb aggregates that do not die after Ca2+ accumulation in Supplementary Figure 3D? Do these cells recover from the plasma membrane perturbation? Did the authors consider using another marker for plasma membrane perturbation together with BAPTA?

Unfortunately, we are not able to image macrophages stained with Oregon Green 488 Bapta-1 AM for more than 36 hours because they lose fluorescence over time, possibly due to partial dye degradation or secretion. Another issue is that macrophages do not establish synchronous interaction with Mtb aggregates (figure 3D; figure S3B). In order to pool together results from many cells, we analyze all the cells that interact with Mtb within the first 20 hours and we define as timepoint 0 the time at which each individual cell establishes interaction with the bacteria. To compare similar time windows for each cell, we use fluorescence values measured at 16 hours post-interaction with bacteria as a readout. This time window is sufficient to observe formation of local Annexin V plasma membrane domains and death in a relevant number of macrophages (figure 1P; figure 3D). Not all of the contacted cells die within the timeframe of our experiments; however, we believe that if we imaged cells that accumulate Ca2+ for longer durations, we would find that all such cells eventually die. This assumption is consistent with the observation that calcium chelation reduces inflammasome activation and death in macrophages in contact with Mtb aggregates (figure 5D; figure 4E).

With respect to the reviewer’s query whether cells recover from plasma membrane perturbation, in our time-lapse microscopy experiments, we observe that when macrophages form local Annexin V-positive plasma membrane domains at the site of contact with Mtb aggregates, they never revert to an Annexin V-negative status afterwards (figure 3D; movie S7; movie S8). Our SEM data show that Mtb aggregates colocalizing with Annexin V-positive domains are not partially covered by intact membrane, in contrast to those associated with Annexin V-negative macrophages, although they do present vesicles and membrane debris on their surface (figure 3F,G ). In the revised manuscript, we include additional fluorescence microscopy images showing that Annexin V-positive foci colocalize with markers for the macrophages’ plasma membrane (figure S8A,B) as well as with more distal areas of the bacterial aggregates, where we do not observe any positive plasma membrane staining (figure S8B). Similarly, although _Mtb_aggregates that are never in contact with macrophages never become Annexin V-positive (figure S8C), we see that upon macrophage death, aggregates in contact with dead cells retain some Annexin V-positive material on their surface (figure S8C; movie S8). Vesicle budding and shedding is a common ESCRT III-mediated membrane repair strategy that allows removal of damaged portions of the plasma membrane and wound resealing (4). Thus, we think that in our experiments the Annexin V-positive foci might represent both damaged membrane areas and released macrophage plasma membrane vesicles that stick to the hydrophobic surface of the bacterial aggregates. This means that the time of appearance of Annexin V-positive domains marks the time when the macrophage membrane experiences a damaging event. Interestingly, we do not observe a gradual increase in fluorescence intensity of the Annexin V-positive domains, but rather multiple single intensity peaks over time (movie S8). This might suggest that multiple discrete damaging events occur over time.

1. In Figure 5D-G it will be important if the authors include dots for each macrophage events for the contact conditions as well, as it was done for the bystander condition.

We apologize for using a too-pale shade of magenta in the earlier version of the manuscript, which apparently made the dots in these figures hard to visualize. In the revised manuscript, we use a darker shade of magenta to show the dots corresponding to the fluorescence values of the macrophages in contact with Mtb aggregates.

1. How did the authors discriminate between the macrophages that are in contact or not with Mtb aggregates after the staining with Casp-1, pRIP3 and pMLKL? Do the aggregates stay in contact even after the staining procedures? Representative images of the labelling should be included in this figure.

Before fixation, we make sure to remove the medium gently to avoid disrupting the interactions between cells and bacteria. This step most likely removes the floating bacterial aggregates that are not in stable contact with cells but apparently does not detach aggregates that stably interact with cells. Our correlative time-lapse microscopy and immunofluorescence images (figure 1; figure S4), as well as our correlative time-lapse microscopy and SEM images (figure S5; figure 3F,G), confirm that Mtb aggregates that interact with cells during time-lapse imaging are retained on the surface of those cells upon fixation and processing for immunofluorescence or electron microscopy. As we can observe in figure 5B (cell indicated by the white arrow), Mtb aggregates are retained on the debris of dead cells. In figure 5 we distinguish between “in contact” macrophages and “bystander” macrophages by inspecting brightfield images showing the cells and the respective fluorescence images corresponding to the bacteria. If the body of a macrophage identified in the brightfield image overlaps with a bacterial aggregate identified in the fluorescence channel, we define the macrophage as “in contact”; otherwise, it is annotated as “bystander”. We provide representative images in figure S12 and we clarify the definition of “in contact” and “bystander” in the figure legend of figure 5.

1. The labelling of Figure 5H needs to be corrected both in the text and in the figure legend.

We thank the reviewer for bringing our attention to this error, which has been corrected in the revised manuscript.

1. Pyroptosis inhibitors did reduce the percentage of cell death, but did it also reduce the number of Annexin-V positive domains? This is important as AnnexinV is a marker of apoptosis and the outcome for Mtb very different.

As pointed out by the reviewer, Annexin V staining is often used as a marker for apoptosis. Typically, apoptotic cells stain positive for Annexin V but negative for other membrane-impermeable markers such as propidium iodide, because they expose phosphatidylserine (bound by Annexin V) on the outer leaflet of the plasma membrane without losing plasma membrane integrity (5). Apoptotic cells often look round and their plasma membrane is stained homogeneously by fluorescently labelled Annexin V (5). In our experiments, we observe that macrophages in contact with Mtb aggregates become Annexin V-positive; however, this happens only at the site of contact with the bacteria (figure 3A; movie S7). Only when cells die and get stained by membrane-impermeable dies such as Draq7 do they also get stained with Annexin V over the entire membrane debris. We thus use Annexin V staining as a marker for membrane perturbation rather than for cell death. If we were using the Annexin V as a marker for cell death, we would expect to see a reduction in Annexin V-positive cells in samples treated with pyroptosis inhibitors. In these samples, we do observe a reduced percentage of cell death in comparison to untreated controls; however, we still observe a comparable percentage of macrophages that stain positive for Annexin V locally, i.e., at the site of contact with bacterial aggregates (supplementary figure S13B). In line with this observation, treated vs. untreated macrophages in contact with Mtb aggregates accumulate similar levels of intracellular calcium. These observations are consistent with our model suggesting that contact with Mtb aggregates induces membrane perturbation, calcium accumulation, inflammasome activation, and pyroptosis in contacted macrophages. Since the death inhibitors used in our study specifically target pyroptosis effectors, we do not expect them to affect upstream events such as membrane perturbation and calcium accumulation.

1. In Figure 6, The sections for Figure 6 are well described but kept relatively long with too many details, it will be helpful to the reader if the authors can combine the sections in one header.

We agree that the text linked to figure 6 is long. We tried to make these sections as concise as possible; however, we are concerned that combining all of the sections under a single header might be at the expense of clarity. Thus, unless the reviewer objects, we would prefer to maintain the use of multiple headers.

1. Figure 6F does not have a statistical test and p-value, it will be important to include the statistical test in the legend and p-values in the

As recommended by the reviewer, we have analyzed the results in figure 6F by using a one-way ANOVA test and we have added the calculated p-values to the figure.

Reviewer #1 (Significance):

Based on the literature, Mtb infection and replication can trigger different types of cell death and most of the studies have addressed cell death only as an outcome of intracellular replication. This study shows another form of host cell death, associated only to extracellular bacterial aggregates that are in contact with macrophages. Plasma membrane damage initiating pyroptosis has been defined in: "Plasma membrane damage causes NLRP3 activation and pyroptosis during Mycobacterium tuberculosis infection" by K.S. Beckwith et al. (2020). However, the effect of extracellular bacteria on plasma membrane damage was not addressed before and this paper is addressing an important observation with respect to Mtb evasion and dissemination. These observations represent a novel and interesting aspect in the induction of macrophage cell death by Mtb and potentially relevant for the disease. If the authors consider the comments listed above, this manuscript will be a novel and relevant addition to the field of host pathogen interactions in tuberculosis.

We thank the reviewer for their perspective and their positive comments about our work.

Reviewer #2 (Evidence, reproducibility and clarity):

In this work, Toniolo and coworkers use single-cell time-lapse fluorescence microscopy to show that extracellular aggregates of Mycobacterium tuberculosis can evade phagocytosis by killing macrophages in a contact-dependent but uptake-independent manner. The authors further show that this process is dependent on the functionality of the ESX-1 type VII secretion system and the presence of mycobacterial phthiocerol dimycocerosate (PDIM). In essence the authors show that M. tuberculosis can induce macrophage death from the outside of the cell, and dissect the different players that are involved in the process.

Major comments:

1. I was intrigued by all the different findings of this work, which was done by using bone marrow derived murine macrophages, however, my first question to the authors is how they imagine that this process will take under an in vivo situation? Do they have evidence that these mycobacterial clumps may form during the initial infection process in the lungs? It would be important to provide more insights and discussion into this question in order to see how relevant the described details are inside the host organism.

Formation of Mtb aggregates in tuberculosis lesions have been documented in several animal models (6, 7) and in humans (8–11). While it is unclear whether mycobacterial aggregates form during the earliest stages of infection, extracellular bacterial aggregates have been observed in animal models of infection within the first month post-infection, and they are often associated with necrotic foci. Moreover, masses of Mtb growing as pellicle-like aggregates are often observed on the surface of cavities in human tuberculosis patients. These observations confirm that Mtb aggregates can form during a tuberculosis infection and that a significant fraction of bacteria are extracellular during different stages of infection. As we observe that macrophages undergo contact-dependent uptake-independent death also in the absence of cytochalasin D in vitro, we assume that this may also happen in vivo when Mtb aggregates are formed or released outside host cells. This process may promote bacterial propagation at early stages of infection as well as at later stages when necrotic granulomas and cavities are formed.

In the revised manuscript we present and discuss our observations in the context of the in vivo phenotypes reported in the scientific literature and we include additional references showing that extracellular Mtb aggregates are often observed in vivo. We also propose this concept already in the Introduction section to better link our observations to possible in vivo scenarios.

Minor comments:

Line 91: here the authors list the different forms of cell death that is induced by MTB infection, and it would be important to add apoptosis as a reported mechanism as well (References: PMID: 23848406, PMID: 28095608)

As suggested by the reviewer, in the revised manuscript we have modified the Introduction section to include apoptosis as a Mtb-induced mechanism of macrophage death and we have cited the two publications recommended by the reviewer.

1. Line 95: The secretion of EspE was mainly described in M. marinum while in members of the M. tuberculosis complex no virulence phenotype was reported to the best of my knowledge.

In agreement with the reviewer’s comment, we have modified the sentence and removed EspE from the list of virulence factors.

1. Lines 98: In the cited papers it is described that PDIM is required for phagosomal damage/rupture, however, the methods used there do not allow to specifically report about translocation. The wording should be adapted.

We thank the reviewer for this insightful comment and we have modified the text accordingly.

1. Line 206: Here it is described that in Figure 3A the BMDMs were expressing tdTomato fluorescence and the bacteria GFP, and the same is also repeated in the Figure legend of Fig3A. However, on the images, BMDMs are shown green and bacterial clumps purple (as also indicated in the description directly on the images) Please check and explain/correct this discrepancy.

We apologize that the color scheme in figure 3A is confusing. In this figure we used tdTomato-expressing BMDMs and GFP-expressing bacteria; however, we have pseudo-colored the fluorescence images for the sake of consistency with the other figures in the manuscript, which always show bacteria in magenta. We have clarified this point in the figure legend of the revised manuscript.

1. Line 304: Here the authors could mention that this finding is similar to results found previously in reference PMID: 28095608 and opposite to the results reported previously in PMID: 28505176.

As recommended by the reviewer, we have added a sentence comparing our results with previous studies and we have cited the two references suggested by the reviewer.

1. Line 321: It should be mentioned that CFP10 (EsxB) can also be secreted without its EsxA partner (under certain circumstances, i.e. when the EspACD operon is not expressed due to a phoP regulatory mutation (PMID: 28706226)). However, in Figure S7 an EspAdeletion mutant shows loss of EsxB secretion. This should be checked and discussed how the data here compare with data and strains published previously.

We thank the reviewer for pointing out this interesting point. Our proteomics data revealed that both our esxA mutant and our espA mutants abolish secretion of both EsxA and EsxB, in line with previously published data (12–14). We do not know why the espA mutant behaves differently from the MTBVAC strain concerning secretion of EsxA and EsxB (although we note that regulatory mutations may have complex pleiotropic effects). In the revised manuscript, we have modified this section to include references highlighting that secretion of these proteins may be uncoupled in some circumstances.

1. The finding that EspB can substitute the loss of virulence due to loss of EsxA/ESAT-6 secretion is astonishing and also is different to previous observations that strain H37Ra and MTBVAC (two attenuated strains that have no or very little EsxA secretion due to a regulation defect of the espACD operon PMID: 18282096; PMID: 28706226). How does the hypothesis put forward by the authors match with these previously published data ?

We thank the reviewer for this interesting comment. We would like to clarify that we are not claiming that EspB and EsxA are in general redundant and that EspB can substitute EsxA as a virulence factor. In our experiments we show that EspB can induce contact-dependent uptake-independent death in macrophages in contact with Mtb aggregates in vitro even in the absence of EsxA; however, the precise role of EspB during infection in mice or humans remains to be elucidated and is outside the scope of this manuscript. A previous study comparing Mtb ESX-1 mutants with different secretion patterns linked EspB secretion to Mtb virulence in vivo (14); however, the behavior of an isogenic espB_deletion strain _in vivo was not reported. A M. marinum espB mutant was shown to have reduced virulence; however, in contrast to Mtb, deletion of espB also affects secretion of EsxA in this organism (15). As the reviewer points out, the Mtb strains H37Ra and MTBVAC do not secrete EsxA due to a mutated phoP gene. Previous literature has shown that espB expression is also dependent on PhoP (16). We thus speculate that these strains might behave similarly to our espA espB mutant strain in the context of contact-dependent uptake-independent induction of macrophage death, although we think that this point is outside the scope of our manuscript.

1. In the same context, it is to notice that the authors report in the paragraph between lines 310-330 about EsxA/EsxB secretion, however, looking at the Western blots of figure S7, there is no blot showing results using an antibody against EsxA. Given the previously published results that EsxA/EsxB secretion may also be disconnected (PMID: 28706226), the wording of the text in this paragraph should be adapted or the results from Western Blots using EsxA antibodies be added.

We agree with the reviewer’s comment. Unfortunately, we currently do not have access to a good antibody for EsxA. A commercial monoclonal antibody that was previously available for immunoblotting has been discontinued. We tried several other antibodies that were previously shown to work in M. marinum, but none of these antibodies were effective in M. tuberculosis. We agree that analysing secretion of EsxB alone might not be sufficient to support claims about EsxA secretion. For this reason, we performed quantitative proteome analysis of the secretome in all of the relevant mutant strains. In our revised manuscript, we are careful to make sure that whenever we refer to EsxA/EsxB secretion we always provide proteomics data to support our conclusions.

1. Line 395: Here the authors write that BTP15, a small molecule that in a previous study was shown to inhibit EsxA secretion at higher concentrations (starting from 1.5 uM and higher). However, no effect on the expression of EsxA was described for that compound in reference PMID: 25299337. Thus the corresponding sentence in line 395 needs to adapted to that situation.

We thank the reviewer for noticing this error, which we have corrected in the revised manuscript.

1. Moreover, most concentrations of the compounds used are reported in uM, except for BTP15. It would be easier for the reader if the concentration used for BTP15 could also be reported in uM.

As suggested by the reviewer, in the revised manuscript we report the concentration of BTP15 in μM.

1. Line 475 The comment on the pore forming activity has to be made with caution, as recombinant EsxA produced from E. coli cultures has been shown to often retain detergent PMID: 28119503 that may be responsible for pore forming activity of recombinant EsxA observed in quite some studies, whereas EsxA purified from M. tuberculosis cultures did not show the detergent, but still retained membranolytic activity. This point should be clarified and discussed, and the wording adapted, as EsxA is not a classical poreforming toxin, but excerts the membrane-lysing activity together with other partners (PDIM) in a yet unknown way upon cell contact.

We thank the reviewer for this comment. In the revised manuscript, we have modified the text accordingly and included the sugggested reference.

Reviewer #2 (Significance):

The findings in this work extend the current knowledge on cell infection by M. tuberculosis in a significant way and put extracellular M. tuberculosis clumps in a new context. These data obtained by single-cell time-lapse fluorescence microscopy also need to be discussed for predicting the relevance for an in vivo situation inside the host organism.

As suggested by the reviewer, in the revised manuscript we discuss additional examples from the literature showing that Mtb aggregates can form during infection and that many bacteria are extracellular and associated with necrotic foci during different stages of the disease in animal models of infection and in human patients. We believe that these previously published observations support the in vivo relevance of the process we observe in vitro.

Reviewer #3 (Evidence, reproducibility and clarity):

This is an excellent study distinguished by the volume of observations, rigor of analysis and clarity of presentation. The results are novel, biologically interesting and pathophysiologically important. The ability of aggregated M. tuberculosis to kill macrophages has been reported, but the understanding was that proliferation of Mtb within macrophages killed them. Here, the authors observe that macrophages are susceptible to pyroptotic death triggered by contact with extracellular Mtb aggregates, and that this is not recapitulated by contact with a comparable number of Mtb as single bacilli. The authors go some way to tracing the mechanism and uncover a complex inter-dependence on PDIM and on components of the mycobacterial ESX-1 secretory system.

The following comments will helpfully improve the study further.

Major points

1. The chief measurement in this study is death of individual macrophages as judged by the observer in videomicroscopy. However, the criteria for calling a macrophage "dead" are not defined with any morphological detail, beyond noting that the cell stops moving and lyses. Of course a cell will stop moving if it has lysed, but do not some if not most cells stop moving before they lyse? If so, lysis alone would seem to be the time-point marker for cell death. Yet from the images in Fig 1E and F, I cannot tell that the cells called "dead" have lysed. Watching the videos, the time of lysis is not clear to me. Eventually, shrunken cell bodies are obvious but it is not clear if these are residua of cells that had been said to "lyse" at an earlier time.

In this study, we used brightfield time-lapse microscopy images to identify cell death. Dying macrophages rapidly change shape, lose membrane integrity, and stop moving. Moreover, the intracellular structures and bacteria also stop moving at the time of death of the host cell. While these events can be difficult to distinguish by examining individual snapshots, they are readily identifiable by careful frame-by-frame examination of time-lapse microscopy image series. To exemplify this process, in the revised manuscript we show in supplementary figure S2A how we identify macrophage cell death events. We also include Draq7 (a live cell-impermeable dye commonly used to identify dead cells by flow cytometry and microscopy) in the growth medium during time-lapse imaging in order to label dead macrophages. The timing of staining validates and confirms our strategy of using brightfield time-lapse images to define the time-of-death of individual cells. To further assist readers, in the revised manuscript we provide the time-lapse microscopy movie used to generate this figure (movie S4). Similar images and movies have also been added for cells treated with cytochalasin D (figure S2B; movie S7). As suggested by the reviewer, we also replaced figures 1E,F with new figures incorporating the Draq7 staining to label macrophage cell death and we include the time-lapse microscopy movies used to generate these figures (movies S4, S5).

1. The use of BTP15 as a specific inhibitor of ESX-1 is problematic. The source of the compound is not stated.

The BTP15 molecule was kindly provided by Prof. Stewart Cole, the corresponding author of the article describing the identification of this compound and its effect on Esx-1 secretion (17). We have included this information in the Materials and Methods section.

1. The concentration used, 20 ug/mL, is well above the reported IC50 (1.2 uM) for its presumed target, a mycobacterial histidine kinase, and above the concentrations (0.3-0.6 uM) reported to inhibit Mtb's secretion of EsxA almost completely. It is concerning that the concentrations that were reported to work so well on the whole cell are lower than the IC50 for the presumed target, because uptake into Mtb and intrabacterial metabolism will typically lead to a lower potency for an inhibitor against the whole bacterium than against the isolated enzyme; and because 50% inhibition of an enzyme rarely gives a functional effect as complete as what is shown in the cited reference. In other words, it is not clear that the histidine kinase is the functionally relevant target of BTP15 in Mtb. The original report did not consider BTP15's possible effect on mammalian cells and the present authors likewise do not take that into consideration with respect to possible effects on the macrophages. No concentration-response or time course experiments with BTP15 are presented. Most important, unless I missed it, there is apparently no demonstration that the compound inhibited ESX-1-dependent secretion in the present authors' hands, no matter by what mechanism. Without this, I am reluctant to accept that the results with BTP15 demonstrate a dependence of extracellular-aggregate-induced macrophage death on ESX-1-mediated secretion from Mtb. I would recommend that the authors either provide a direct demonstration of BTP15's effect on ESX-1 dependent secretion at concentrations near those that worked on whole cells in the original report, or drop the BTP15 studies from the paper. That said, the genetic experiments remain unequivocal, so the paper's conclusions would not be affected.

We agree with the reviewer that in the original version of our manuscript we did not provide direct evidence demonstrating that BTP15 inhibits ESX-1 secretion and that it does not affect the host cells. We addressed the first issue by quantifying (by Western blot) the secretion of EsxB and EspB in Mtb cultures treated with different concentrations of BTP15. We show that BTP15 reduces secretion of these two proteins in a dose-dependent manner. These data have been included in figures S21A-B of the revised manuscript. In line with this observation, we also show that BTP15 reduces uptake-independent killing of macrophages by Mtb aggregates in a dose-dependent manner (figure 6H). To show that the dose-dependent effect observed in macrophages does not depend on a direct effect of BTP15 on the host cells, we treated Mtb with different concentrations of BTP15 for 48 hours and removed the compound by washing the bacteria prior to infection. We observe that Mtb aggregates that have been treated with BTP15 show reduced uptake-independent killing of macrophages, even when bacteria have been pre-treated and the small molecule is not present during the incubation with the cells (figure S21C). We hope that these additional results provide clear evidence that BTP15 reduces Mtb-mediated contact-dependent uptake-independent killing of macrophages by inhibiting ESX-1 secretion, consistent with our genetic data. We think these results are important because they provide a chemical validation of our genetic data. To the best of our knowledge, BTP15 is the only available compound known to inhibit ESX-1 secretion, and in the revised manuscript we confirm that this compound has the previously described effect on Mtb also in our hands. Unfortunately, we had to use concentrations higher than those previously reported to inhibit ESX-1 secretion in order to achieve the observed effects. As we had access only to prediluted aliquots that had been stored for a long time, we cannot rule out the posibility that the compound might have undergone partial degradation during storage.

1. The experiments, or at least the discussion, could consider what may distinguish single Mtb cells from aggregated Mtb in some way relevant to the present observations. The authors seem to assume that all the Mtb cells in their preparations are biochemically equivalent and that their distribution into single-cell or aggregate subpopulations is stochastic. What if it is deterministic instead? For example, what if these two subpopulations are defined by differential expression of PDIM, so that the greater macrophage-killing effect of aggregates than single cells in equivalent numbers reflects a greater amount of PDIM in the aggregates, rather than some sort of valency-of-contact effect? The authors could compare the PDIM-to-DNA ratio in the single cell and aggregated subpopulations, or at least discuss this possibility.

We thank the reviewer for proposing this extremely interesting idea. In the revised manuscript, we have added a discussion of this point (lines 487-489) and we have floated various possible explanations. However, we believe that experimental dissection of the underlying mechanism could be a very lengthy undertaking and we hope that the reviewer will agree that this is outside the scope of the current manuscript.

Minor points

1. Some of the experiments compare "low", "medium" and "high" numbers of Mtb, but I could not find a definition of these numbers.

We apologize for this oversight. In the revised manuscript, we have clarified the definition of these gates in the figure 2 legend.

1. There seem to be no positive or negative controls for any of the antibodies used for cell staining (anti-cleaved caspase 1, antiphospho RIP3, anti-phospho MLKKL).

As recommended by the reviewer, the revised manuscript includes controls for all of the antibodies used for immunostaining. In figure S12 we provide representative immunostaining images and fluorescence quantification of uninfected untreated macrophages (negative controls) and of uninfected macrophages treated with cocktails of molecules typically used to induce apoptosis, pyroptosis, or necroptosis (positive controls).

Reviewer #3 (Significance):

The results are novel, biologically interesting and pathophysiologically important.

We thank the reviewer for their appreciation of our findings.

References 1. H. Gan, et al., Mycobacterium tuberculosis blocks crosslinking of annexin-1 and apoptotic envelope formation on infected macrophages to maintain virulence. Nature Immunology 9, 1189–1197 (2008). 2. M. Divangahi, et al., Mycobacterium tuberculosis evades macrophage defenses by inhibiting plasma membrane repair. Nature Immunology 10, 899–906 (2009). 3. D. Mahamed, et al., Intracellular growth of Mycobacterium tuberculosis after macrophage cell death leads to serial killing of host cells. eLife 6, e22028 (2017). 4. A. J. Jimenez, et al., ESCRT Machinery Is Required for Plasma Membrane Repair. Science 343, 1247136 (2014). 5. M. van Engeland, L. J. W. Nieland, F. C. S. Ramaekers, B. Schutte, C. P. M. Reutelingsperger, Annexin V-Affinity assay: A review on an apoptosis detection system based on phosphatidylserine exposure. Cytometry 31, 1–9 (1998). 6. D. R. Hoff, et al., Location of Intra- and Extracellular M. tuberculosis Populations in Lungs of Mice and Guinea Pigs during Disease Progression and after Drug Treatment. PLOS ONE 6, e17550 (2011). 7. S. M. Irwin, et al., Presence of multiple lesion types with vastly different microenvironments in C3HeB/FeJ mice following aerosol infection with Mycobacterium tuberculosis. Disease Models & Mechanisms 8, 591–602 (2015). 8. Kaplan, G., et al., Mycobacterium tuberculosis Growth at theCavity Surface: a Microenvironment with FailedImmunity. Infection and Immunity 71, 7099–7108 (2003). 9. J. Timm, et al., A Multidrug-Resistant, acr1-Deficient Clinical Isolate of Mycobacterium tuberculosis Is Unimpaired for Replication in Macrophages. The Journal of Infectious Diseases 193, 1703–1710 (2006). 10. R. L. Hunter, Pathology of post primary tuberculosis of the lung: An illustrated critical review. Tuberculosis 91, 497–509 (2011). 11. G. Wells, et al., Micro–Computed Tomography Analysis of the Human Tuberculous Lung Reveals Remarkable Heterogeneity in Three-dimensional Granuloma Morphology. Am J Respir Crit Care Med 204, 583–595 (2021). 12. S. A. Stanley, S. Raghavan, W. W. Hwang, J. S. Cox, Acute infection and macrophage subversion by Mycobacterium tuberculosis require a specialized secretion system. Proc Natl Acad Sci USA 100, 13001 (2003). 13. S. M. Fortune, et al., Mutually dependent secretion of proteins required for mycobacterial virulence. Proc Natl Acad Sci U S A 102, 10676 (2005). 14. J. M. Chen, et al., Mycobacterium tuberculosis EspB binds phospholipids and mediates EsxA-independent virulence. Mol Microbiol 89, 1154–1166 (2013). 15. L.-Y. Gao, et al., A mycobacterial virulence gene cluster extending RD1 is required for cytolysis, bacterial spreading and ESAT-6 secretion. Mol Microbiol 53, 1677–1693 (2004). 16. V. Anil Kumar, et al., EspR-dependent ESAT-6 Protein Secretion of Mycobacterium tuberculosis Requires the Presence of Virulence Regulator PhoP. Journal of Biological Chemistry 291, 19018–19030 (2016). 17. J. Rybniker, et al., Anticytolytic Screen Identifies Inhibitors of Mycobacterial Virulence Protein Secretion. Cell Host & Microbe 16*, 538–548 (2014).

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

This is an excellent study distinguished by the volume of observations, rigor of analysis and clarity of presentation. The results are novel, biologically interesting and pathophysiologically important. The ability of aggregated M. tuberculosis to kill macrophages has been reported, but the understanding was that proliferation of Mtb within macrophages killed them. Here, the authors observe that macrophages are susceptible to pyroptotic death triggered by contact with extracellular Mtb aggregates, and that this is not recapitulated by contact with a comparable number of Mtb as single bacilli. The authors go some way to tracing the mechanism and uncover a complex inter-dependence on PDIM and on components of the mycobacterial ESX-1 secretory system.

The following comments will helpfully improve the study further.

Major points

The chief measurement in this study is death of individual macrophages as judged by the observer in videomicroscopy. However, the criteria for calling a macrophage "dead" are not defined with any morphological detail, beyond noting that the cell stops moving and lyses. Of course a cell will stop moving if it has lysed, but do not some if not most cells stop moving before they lyse? If so, lysis alone would seem to be the time-point marker for cell death. Yet from the images in Fig 1E and F, I cannot tell that the cells called "dead" have lysed. Watching the videos, the time of lysis is not clear to me. Eventually, shrunken cell bodies are obvious but it is not clear if these are residua of cells that had been said to "lyse" at an earlier time.

The use of BTP15 as a specific inhibitor of ESX-1 is problematic. The source of the compound is not stated. The concentration used, 20 mg/mL, is well above the reported IC50 (1.2 uM) for its presumed target, a mycobacterial histidine kinase, and above the concentrations (0.3-0.6 uM) reported to inhibit Mtb's secretion of EsxA almost completely. It is concerning that the concentrations that were reported to work so well on the whole cell are lower than the IC50 for the presumed target, because uptake into Mtb and intrabacterial metabolism will typically lead to a lower potency for an inhibitor against the whole bacterium than against the isolated enzyme; and because 50% inhibition of an enzyme rarely gives a functional effect as complete as what is shown in the cited reference. In other words, it is not clear that the histidine kinase is the functionally relevant target of BTP15 in Mtb. The original report did not consider BTP15's possible effect on mammalian cells and the present authors likewise do not take that into consideration with respect to possible effects on the macrophages. No concentration-response or time course experiments with BTP15 are presented. Most important, unless I missed it, there is apparently no demonstration that the compound inhibited ESX-1-dependent secretion in the present authors' hands, no matter by what mechanism. Without this, I am reluctant to accept that the results with BTP15 demonstrate a dependence of extracellular-aggregate-induced macrophage death on ESX-1-mediated secretion from Mtb. I would recommend that the authors either provide a direct demonstration of BTP15's effect on ESX-1 dependent secretion at concentrations near those that worked on whole cells in the original report, or drop the BTP15 studies from the paper. That said, the genetic experiments remain unequivocal, so the paper's conclusions would not be affected.

The experiments, or at least the discussion, could consider what may distinguish single Mtb cells from aggregated Mtb in some way relevant to the present observations. The authors seem to assume that all the Mtb cells in their preparations are biochemically equivalent and that their distribution into single-cell or aggregate subpopulations is stochastic. What if it is deterministic instead? For example, what if these two subpopulations are defined by differential expression of PDIM, so that the greater macrophage-killing effect of aggregates than single cells in equivalent numbers reflects a greater amount of PDIM in the aggregates, rather than some sort of valency-of-contact effect? The authors could compare the PDIM-to-DNA ratio in the single cell and aggregated subpopulations, or at least discuss this possibility.

Minor points

Some of the experiments compare "low", "medium" and "high" numbers of Mtb, but I could not find a definition of these numbers.

There seem to be no positive or negative controls for any of the antibodies used for cell staining (anti-cleaved caspase 1, antiphospho RIP3, anti-phospho MLKKL).

Significance

The results are novel, biologically interesting and pathophysiologically important.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

In this work, Toniolo an coworkers use single-cell time-lapse fluorescence microscopy to show that extracellular aggregates of Mycobacterium tuberculosis can evade phagocytosis by killing macrophages in a contact-dependent but uptake-independent manner. The authors further show that this process is dependent on the functionality of the ESX-1 type VII secretion system and the presence of mycobacterial phthiocerol dimycocerosate (PDIM). In essence the authors show that M. tuberculosis can induce macrophage death from the outside of the cell, and dissect the different players that are involved in the process.

Major comments:

I was intrigued by all the different findings of this work, which was done by using bone marrow derived murine macrophages, however, my first question to the authors is how they imagine that this process will take under an in vivo situation ? Do they have evidence that these mycobacterial clumps may form during the initial infection process in the lungs ? It would be important to provide more insights and discussion into this question in order to see how relevant the described details are inside the host organism.

Minor comments:

Line 91: here the authors list the different forms of cell death that is induced by MTB infection, and it would be important to add apoptosis as a reported mechanism as well (References: PMID: 23848406, PMID: 28095608)

Line 95: The secretion of EspE was mainly described in M. marinum while in members of the M. tuberculosis complex no virulence phenotype was reported to the best of my knowledge.

Lines 98: In the cited papers it is described that PDIM is required for phagosomal damage/rupture, however, the methods used there do not allow to specifically report about translocation.<br /> The wording should be adapted.

Line 206: Here it is described that in Figure 3A the BMDMs were expressing tdTomato fluorescence and the bacteria GFP, and the same is also repeated in the Figure legend of Fig3A. However, on the images, BMDMs are shown green and bacterial clumps purple (as also indicated in the description directly on the images) Please check and explain/correct this discrepancy.

Line 304: Here the authors could mention that this finding is similar to results found previously in reference PMID: 28095608 and opposite to the results reported previously in PMID: 28505176.

Line 321: It should be mentioned that CFP10 (EsxB) can also be secreted without its EsxA partner (under certain circumstances , i.e. when the EspACD operon is not expressed due to a phoP regulatory mutation (PMID: 28706226)). However, in Figure S7 an EspAdeletion mutant shows loss of EsxB secretion. This should be checked and discussed how the data here compare with data and strains published previously.<br /> The finding that EspB can substitute the loss of virulence due to loss of EsxA/ESAT-6 secretion is astonishing and also is different to previous observations that strain H37Ra and MTBVAC (two attenuated strains that have no or very little EsxA secretion due to a regulation defect of the espACD operon PMID: 18282096; PMID: 28706226). How does the hypothesis put forward by the authors match with these previously published data ?<br /> In the same context, it is to notice that the authors report in the paragraph between lines 310-330 about EsxA/EsxB secretion, however, looking at the Western blots of figure S7, there is no blot showing results using an antibody against EsxA. Given the previously published results that EsxA/EsxB secretion may also be disconnected (PMID: 28706226), the wording of the text in this paragraph should be adapted or the results from Western Blots using EsxA antibodies be added.

Line 395: Here the authors write that BTP15, a small molecule that in a previous study was shown to inhibit EsxA secretion at higher concentrations (starting from 1.5 uM and higher). However, no effect on the expression of EsxA was described for that compound in reference PMID: 25299337. Thus the corresponding sentence in line 395 needs to adapted to that situation.<br /> Moreover, most concentrations of the compounds used are reported in uM, except for BTP15. It would be easier for the reader if the concentration used for BTP15 could also be reported in uM.

Line 475 The comment on the pore forming activity has to be made with caution, as recombinant EsxA produced from E. coli cultures has been shown to often retain detergent PMID: 28119503 that may be responsible for pore forming activity of recombinant EsxA observed in quite some studies, whereas EsxA purified from M. tuberculosis cultures did not show the detergent, but still retained membranolytic activity. This point should be clarified and discussed, and the wording adapted, as EsxA is not a classical poreforming toxin, but excerts the membrane-lysing activity together with other partners (PDIM) in a yet unknown way upon cell contact.

Significance

The findings in this work extend the current knowledge on cell infection by M. tuberculosis in a significant way and put extracellular M. tuberculosis clumps in a new context. These data obtained by single-cell time-lapse fluorescence microscopy also need to be discussed for predicting the relevance for an in vivo situation inside the host organism.

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

n this manuscript authors show that extracellular Mtb aggregates can cause macrophage killing in a close contact dependent but phagocytosis independent manner. They showed Mtb aggregates can induce plasma membrane perturbations and cytoplasmic Ca2+ influx with live cell microscopy. Next, the authors show that the type of cell death initiated by extracellular aggregates is pyroptosis and they partially supressed cell death with pyroptosis inhibitors. They also identified that PDIM, EsxA/EsxB and EspB all have a role in uptake-independent killing of macrophages even though their impact varies with respect membrane perturbation and Ca2+ influx. Finally, they used a small molecule inhibitor BTP15 to inhibit the effect of ESX-1 during the contact of the extracellular Mtb aggregates with the macrophages and they observed a substantial decrease in membrane perturbation and macrophage killing.

The work describes a very interesting mechanism by which Mtb can kill macrophages that is possibly relevant in the context of infection. In general, there are two main issues with the experiments and the interpretation: the lack of quantitative analysis showing that in a population of macrophages the ones that are in contact with the aggregates die whereas the ones that are not in contact remain alive. This is currently not shown, and it should be added in figure 1. The second is the cell death mode, as the markers used are very different and considering different outcomes (e.g., apoptosis vs. necrosis) are relevant for the infection it is unclear what is being measured here and the impact on bacterial replication.

The authors are showing that infection with Mtb aggregates increase the rate of the macrophage killing but how does this impact infection dissemination and replication of the bacterial aggregates? Is it beneficial for the aggregates? Did the authors check the growth rate of Mtb along with cytochalasin D? How did the authors quantify the interactions of Mtb with macrophages in Figure 1D? Is it enough to conclude with one example of SEM that the mycobacteria with different fragmentation discriminates if the bacteria is intracellular or extracellularly localised? Can authors use an alternative quantitative method to confirm the localization of the bacteria by a quantification by 3D imaging of these two phenotypes with a cytoskeleton marker (or may be even with tdTomato-expressing BMDMs)?

How do we know if the cell is lysed at 30 h in Supplementary Figure 1, did the authors use a marker to detect the cell lysis or is it based on just the observation from the live cell imaging? Movies in supplementary are actually not very informative as there are many ongoing events and it is hard to visualise what the authors claim. A marker of cell death in the movies should be used.

Total macrophage killing after contact in Figure 1L is around 12 hours, whereas it is observed that the macrophage death after contact with cytochalasin D treatment in Figure 1M is even longer than 24 hours. The viability at 12 hours in Figure1M is as fragmented Mtb survival in Figure1L, why there is a difference in timing with respect to macrophage killing?

Did authors perform statistical tests for Figure 1D and Figure 1N? p-values should be added.

In Figure 3, do the observations indicated in the Figure 3 happen in all the macrophages that are in contact with aggregates? This is unclear and critical to support the conclusions. Do all the macrophages that are in contact with Mtb aggregates become Annexin-V positive? In Supplementary Figure 2 there is some information regarding this question, but it will be important to show it as a percentage. Did the authors try to stain Mtb aggregates alone with Annexin-V as a control over the duration of the imaging?

In Figure 4, did the authors continue to image the cells interacting with Mtb aggregates that do not die after Ca2+ accumulation in Supplementary Figure 3D? Do these cells recover from the plasma membrane perturbation? Did the authors consider using another marker for plasma membrane perturbation together with BAPTA?

In Figure 5D-G it will be important if the authors include dots for each macrophage events for the contact conditions as well, as it was done for the bystander condition. How did the authors discriminate between the macrophages that are in contact or not with Mtb aggregates after the staining with Casp-1, pRIP3 and pMLKL? Do the aggregates stay in contact even after the staining procedures? Representative images of the labelling should be included in this figure. The labelling of Figure 5H needs to be corrected both in the text and in the figure legend. Pyroptosis inhibitors did reduce the percentage of cell death, but did it also reduce the number of Annexin-V positive domains? This is important as AnnexinV is a marker of apoptosis and the outcome for Mtb very different.

In Figure 6, The sections for Figure 6 are well described but kept relatively long with too many details, it will be helpful to the reader if the authors can combine the sections in one header. Figure 6F does not have a statistical test and p-value, it will be important to include the statistical test in the legend and p-values in the figure.

Significance

Based on the literature, Mtb infection and replication can trigger different types of cell death and most of the studies have addressed cell death only as an outcome of intracellular replication. This study shows another form of host cell death, associated only to extracellular bacterial aggregates that are in contact with macrophages. Plasma membrane damage initiating pyroptosis has been defined in: "Plasma membrane damage causes NLRP3 activation and pyroptosis during Mycobacterium tuberculosis infection" by K.S. Beckwith et al. (2020). However, the effect of extracellular bacteria on plasma membrane damage was not addressed before and this paper is addressing an important observation with respect to Mtb evasion and dissemination. These observations represent a novel and interesting aspect in the induction of macrophage cell death by Mtb and potentially relevant for the disease. If the authors consider the comments listed above, this manuscript will be a novel and relevant addition to the field of host pathogen interactions in tuberculosis.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

Point by Point Description of Revisions

We thank the reviewers for their time, effort and constructive input. Below, our responses are bolded with yellow highlighting, while the reviewers’ comments are italicized.

---

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

Summary

The manuscript by Hays and colleagues described the spectrum of mutations that drive adaptation in nitrogen-limit environment by experimental evolution. The approach of serial transfer (fluctuating condition) allowed them to find that Ty insertion is the major mutation type for adaptive evolution. This was neither observed in nitrogen-limited condition when another experimental evolution approach, chemostat (non-fluctuating condition), was applied, nor in glucose-limited condition. The authors concluded that not only selection pressure itself but also how selection is applied are important to shape the adaptive events.

Major points

*Both serial transfer and chemostat are commonly used approaches of experimental evolution. In the manuscript, the authors refer serial transfer to "fluctuating" condition because the low nitrogen source would be consumed to none during the interval of transfers. I am wondering whether the authors have estimated the nitrogen uptake (consumption) during the transfer intervals and whether the nitrogen was exhausted within 48 hours. *

We appreciate the reviewer’s question, and although we did not directly measure nitrogen consumption throughout this specific experiment, ammonium was the limiting nutrient in the defined medium which has been previously used to achieve transient nitrogen starvation conditions in other yeast experimental evolutions (Blundell et al. 2019). In that previous work, it was confirmed that addition of ammonium above 0.04% (up to 0.15%) led to additional rounds of doubling – confirming that the amount of ammonium provided was in fact the limiting nutrient. Finally, we point out that the adaptive mutations recovered in this study predominantly impact genes known to affect nitrogen catabolism, as is expected under nitrogen-limited evolution conditions.

We’ve updated the methods section to ensure the rationale for this medium choice is clearly stated.

Since this is not precisely controlled by experiment design, the "fluctuating" condition itself may be not stable during the long-term evolution. For example, as population evolved, the rate and the amount of nitrogen uptake might change. I feel a better experiment setup for "fluctuating" condition is like 24 hour "low-nitrogen (ammonium)" - 24 hour "no ammonium" and so on. If the adaptive mutations (e.g. adaptive Ty) specifically respond to such "fluctuating" condition rather than chemostat, the authors can measure their fitness in nitrogen starvation condition, which is expected to be fitter than mutants observed only in chemostat (e.g. copy number variation of nitrogen transporters).

The reviewer correctly points out that nitrogen availability will change as the population adapts, and it is likely that some portion of the population become better at utilizing the newly available nitrogen upon transfer into fresh medium over time. This is in fact the intention of this experimental design. We have rephrased the text of the main paper to emphasize that our fluctuating conditions represent fluctuations in the nutrient availability in fresh medium upon transfer, and not strict oscillating nitrogen concentrations that cells experience locally throughout all generations.

We note that in the reviewer-proposed experimental design (using 2 stages of low- and no- nitrogen media), that the low-nitrogen condition would still exhibit the same population-dependent nitrogen usage dynamics as the population adapts over time. We chose our evolution conditions to apply a selective pressure for cells to become best adapted to the environmental fluctuations associated with this transfer regimen, and we have updated the main paper to clarify this point. We thank the reviewer for helping us clarify this important point.

The authors compared their results with published dataset using nitrogen-limitation chemostat and the mutation spectrum is different. In addition to the "fluctuating" and "non-fluctuating" difference as mentioned above, other factors need to be considered. First, the nitrogen-limited conditions in the two studies are different. The authors used 0.04% ammonium sulfate while Hong et al used "800 uM nitrogen regardless of the molecular form of the nitrogen", which may influence the mutation spectrum and need to be discussed. Second, bottlenecks were applied for each transfer in this study, in comparison with constant population size in chemostat, which will influence the efficiency of selection and further the evolutionary dynamics and outcomes. Thus, population size and bottlenecks need to take in to account to make comparisons of mutation spectrum.

We thank the reviewer for their point: we have expanded the section of the main text addressing the differences in how serial transfer and chemostat conditions are applied, the media differences necessitated by such and specifically how the conditions between our study and the Hong et al study differ. We believe the additional detail now better highlights our point that how selection is applied shapes adaptive events, and we thank the reviewer for their helpful input.

*The authors found that Ty mutagenesis accounts for a substantial number of adaptive mutations in nitrogen limitation. I am wondering for adaptive clones, whether Ty occurred independently or is more likely to co-exist with other drivers. *

We appreciate the reviewer’s question. In the clones with adaptive Ty insertions, the only co-occurring adaptive mutation is autodiploidization. There were no additional mutational classes that were adaptive and co-occur with adaptive Ty insertions in our dataset. However, many novel Ty insertions are neutral, and these DO co-occur with beneficial mutations. These data are captured in Figure 5A, and in detail in Supplemental File 1. The blue bar in the adaptive haploids reflect neutral-fitness Ty insertions that co-occur with other mutations that drive fitness increase. These are distinct from the Ty insertions that are themselves responsible for the fitness increase, which are captured in the orange bar. We have clarified the text surrounding the Fig 5A results to better emphasize these findings.

What is the distribution of number of clones with one, two, and multiple mutations? If there is co-existence of driver mutations, what is the relative contribution of each to adaptation? The phenotypic validation of Ty mutagenesis for adaptation is expected while it seems only one case was presented in Figure 2 (mep1Ty−731427).

Aside from diploidization events, only one clone with two nitrogen-adaptive mutations was identified in this study: a double mutant with mutations in both gat1 and tor1. Please see Supplemental File 1 (which is sortable) for a complete outline of all clones with mutations and fitness remeasurements. In the case of diploids that have additional beneficial mutations, those data are shown in Figure 3 with diploids indicated as well as the ploidy of the secondary beneficial mutation, and again in detail in Supplemental File 1.

The reviewer is correct in that only one Ty mutation was dissected and validated in Figure 2. However, we inferred adaptation by Ty insertion through the observation of parallel adaptation, and we fitness remeasurements of many independent Ty insertion mutants.

Statistical analysis needs to be reinforced in the manuscript, including but not limited to Figure 2 fitness comparison among clones with different genotypes, Figure 5 Ty enrichment comparison, etc.

We thank the reviewer for their helpful suggestion. We have updated figures and figure legends to more clearly include statistical comparisons between genotypes for Figures 2 and 5: specifically describing the analyses used and the associated p-values for differences between WT and adaptive alleles and significance of Ty class enrichments.

Minor points

We thank the reviewer for their detailed and careful edits below and have addressed them in the main text and figures as applicable.

"For diploids, we only sequenced those with estimated fitness greater than diploidy alone would provide." Main text clarified with additional explanation

"either through impacting alternate start (green triangle) or alternate stop sites (yellow and red triangles)." I do not see yellow and red triangles in Fig. 3. Legend updated to reflect current figure color palette.

Fig.2. FCY2 mutant fitness can be added as well?

Unfortunately, data for FCY2 backcrossed mutants were not generated

"while we found only 212 novel Ty insertions in 488 glucose evolved clones (Figure 5B)" The value in the text does not match the one in the figure.

We appreciate the reviewer’s attention to detail and have corrected the main text to match the correct value in Fig 5B.

In addition to adaptive Ty insertion, what is the genome-wide distribution or characteristics of other Ty, especially for nitrogen-limited condition? Is that distinct from glucose-limited condition?

Figure S5 addresses the major locations of Ty insertions upstream of tRNA genes, in both Glucose and Nitrogen limited evolutions, the insertion location previously published to be preferred; the only difference between glucose and nitrogen is that there are more in the nitrogen limited condition, though the profile of insertions upstream of tRNAs is essentially the same. In addition to insertions upstream on tRNAs, all other specific insertion locations are available in Supplemental File 1 and Supplemental File 4.

"Studies determining at which step(s) of the Ty life cycle nitrogen starvation shapes ty activity would be needed to determine the specific mechanism underlying the increase in transposon insertions." Here "ty" => "Ty"

Corrected! We thank the reviewer for their detailed reading.

Reviewer #1 (Significance (Required)):

The manuscript is a follow-up work of Levy et al. 2015 and Blundell et al. 2019. In general, the research is interesting and point out the important role of Ty for adaptive evolution in nitrogen-limited environment. It also compared the spectrum of adaptive mutations in response to nitrogen limitation by serial transfer (this work) and chemostat (especially the work of Gresham lab). The paper is well-written as well. Audience from the field of genetics, genomics and evolution will be interested in this work.

My field of expertise: genetics, experimental evolution, budding yeast

We thank the reviewer for their kind comments, constructive input, and generosity with their time.

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

Hays et al. sequence and analyze the mutational spectrum from a set of S. cerevisiae strains evolved in a nitrogen limiting environment, and detail genes that recurrently are found to be mutated in a fluctuating nitrogen limiting environment. These data are contrasted to evolution under glucose limited environments and non-fluctuating environments. Specifically, Hays et al. observe a high proportion of Ty element-mediated mutations arising from strains evolved under the fluctuating nitrogen limiting regime. Their fitness data are robust and clearly demonstrate that these mutations reproducibly lead to improved fitness under nitrogen limitation (based on the authors' defined criteria). Overall, the observed bias of the high proportion of Ty-mediated mutation in fluctuating nitrogen starvation is unexpected and an important finding. Further, the discussion was thoughtful and well executed in detailing interpretations of the data more broadly. We are generally positive about this work and find the analyses robust and convincing. The authors should address the concerns listed below prior to acceptance/publication.

We thank the reviewer for their kind words and enthusiasm for our study, we have worked to address their constructive feedback as detailed below.

Reviewer #2 (Significance (Required)):

Major comments to be addressed:

The claim that the 3' UTR Ty insertions in MEP1 are apparently gain of function is very interesting. The authors should consider performing RT-PCR or strand specific RNAseq to see whether the antisense transcript is reduced and the MEP1 transcript is increased in the presence of the 3' UTR insertion. This would provide much stronger support for their claim that MEP1 3' Ty insertions are gain of function. Orientation information is critical to provide!

We agree that these future directions are exciting and of extreme interest! We however believe they are out of the scope of this current study which already includes substantial data and analysis. We note that we did not claim that the 3’ UTR insertions are gain of function – instead, we suggested that “Ty insertions in the 3’ region unique to the MEP1 locus may affect fitness in nitrogen limitation via a mechanism different than the putative gain of function missense mutations in the coding region itself”. We did not speculate on the mechanism by which these insertions are adaptive, but it is an active line of research and we look forward to discovering the mechanism.

The authors seemed to miss a golden opportunity to measure Ty1 expression or transposition under fluctuating/non-fluctuating nitrogen starvation. Otherwise, the claims of increased Ty activity are unsupported. The authors measured an endpoint (Ty insertion), but this says nothing directly as to the rate of activity, although it is presumably correlated. However, based on the data one could argue activity may be equal in all environments, but the mutational events caused by Ty activity are uniquely selected for in fluctuating nitrogen starvation. As it stands, either model (increased activity vs. differential strength of selection) are equally likely. At a minimum, the authors should at least address this point.

We appreciate the reviewer bringing this concern to our attention: we address the reviewer’s concerns in 3 ways: First, we’ve rephrased to more explicitly consider the possibility that the observed difference in novel Ty insertions could be driven at the level of selection, not activity. Second, we’ve clarified the main text to greater emphasize our reasoning for why we speculate the inference of greater Ty activity under nitrogen starvation may be more likely based on the level of presumptive neutral Ty insertions being greater in nitrogen than in glucose (even after normalization for the number of evolved generations). Third, we’ve performed additional experiments that support that, at least with an artificial retrotransposition reporter construct, these starvation conditions show additional Ty activity in nitrogen compared to glucose (note, we have not carried out such experiments in chemostats, and do not currently have a functioning chemostat set up). We’re including these results below, though have not included them in the manuscript, as we intend to generate additional data for a subsequent study to make these claims more robust. We feel that adding them to this manuscript would make it less focused.

To assess Ty activity in yeast experiencing different nutrient conditions, we used a modified version of a plasmid-based Ty reporter created previously by Curcio and Garfinkel, 1991, PNAS 88(3):936-40. The original reporter construct used an inducible GAL promoter to initiate Ty transcription from the plasmid, and new Ty insertions confer the ability for the strain to grow on SC-His. To assess Ty activity induced by nitrogen limitation, we excised the GAL promoter and instead used the native Ty promoter from the insertion found at YPLWTy1-1. This Ty promoter was selected based on having recovered novel Ty insertions in evolved clones that originated from this locus.

Plasmid pGS234 was created by replacing the promoter containing XhoI fragment from pGTy1mhis3-AI with XhoI fragment containing promoter from chromosomal location of YPLWTy1-1.

Strains bearing the Ty reporter plasmid pGS234 were subjected to nitrogen limited media and glucose limited media to assess transposon activity in these conditions. We observe significantly more Ty activity from the reporter plasmid in nitrogen-limited conditions than in glucose limited conditions or in SC-ura medium (see Figure below).

Panel A: Bars represent average of three WT strains with transposon reporter plasmid; each value is number of colonies on SC-His medium with each His+ colony representing independent Ty transposition events. Strains were grown in SC-Ura and then shifted to M14, M3 or SC-Ura as a control for 48 hours and plated on SC-His plates.

Panel B. One WT strain with pGS234 was subjected to a fluctuation test (16x 5ml tubes) in M14 and M3 media. Each dot represents the number of colonies on each SC-His plate. Kruskal-Wallis chi-squared = 23.341, df = 1, p-value = 1.357e-06

In line with the above, we think the authors should soften some points in the discussion as it stands. For example: "The significant increase of Ty activity under this specific fluctuating nitrogen-starvation..." We feel the data does not exclusively support increased activity of Ty, that would require the aforementioned assays. As it stands, we feel this is more appropriate: ": "The significant increase of Ty insertions under this specific fluctuating nitrogen-starvation..."

We edited the main text to include this suggested language change.

Minor comments to be addressed:

Please provide a citation for the following statement "The single copy of Ty5 in the ancestor is known to be inactive and gives rise to no new insertions under either glucose or nitrogen limitation" - Voytas & Boeke. Nature 1992.

We appreciate the reviewer catching this, and the reference has been added.

We found the following to be a confusing sentence: "Indeed, if global Ty derepression reflects a host-parasite coevolution that minimizes host cost and maximizes potential for survival of both, the role of transposons in host evolvability is important (Levin and Moran 2011)."

We have clarified this sentence by editing it to: “Indeed, the role of transposons in host evolvability is important: global Ty derepression could reflect host-parasite coevolution towards a less parasitic lifestyle: resulting in minimal host cost and maximized potential for survival of both, especially under detrimental environmental conditions (Levin and Moran 2011)”

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

Hays et al. studied the genomic changes that lead to adaptation under fluctuating nitrogen starvation. In addition to loss of function alleles, the authors identified adaptive gain-of- function alleles. Furthermore, their results demonstrate that Ty and microhomology-facilitated mutations in several candidate genes contribute substantially (though not exclusively) to the adaptation under nitrogen-limited serial transfer. Importantly, a novel lineage tracking method provides high resolution fitness measurements.

We appreciate the reviewer’s helpful edits in clarifying and improving the manuscript, and appreciate their time and constructive input.

Despite the clear merits of the study, we also have a few relatively minor questions and suggestions

• Please elaborate on the criteria they used to identify adaptive loci. The fact that these mutations occurred repeatedly is highlighted on Table 1, but perhaps numbers could also be included in the text, to increase clarity.* We have added the pertinent numbers to the main text to accompany the values captured in Table 1 and Supplemental file 1 and further emphasize selection criteria outline in the main text.

• "Were also validated to a fitness effect of >0.01 in nitrogen-limited media". More details about the selection of this cut-off value need to be provided in either the text or the Methods section to increase clarity.*

We agree and have clarified the limit of detection used in the methods section.

• In Figure 3 it seems that the type of observed mutations was less important compared to the gene where the mutation occurred. Therefore, it seems that some genes, e.g. GAT1, contribute more to the observed fitness change. It would be beneficial if the authors discussed this observation.*

We thank the reviewer for their observation and have included some additional discussion in the main text around the per-locus fitness observations as shown in Figure 3.

• What was the reason to select samples from the 88th generation for glucose and from the 192nd generation for nitrogen, as presented in Figure 5? How does this affect the observations?*

We thank the reviewer for their question: these generations were determined to best capture peak adaptive diversity (as discussed in Blundell et al 2019), based on population barcode dynamics in the original evolutions (Levy et al 2015, Blundell et al. 2019). The challenge is balancing picking a time point late enough, such that there are sufficient numbers of adaptive clones within independent lineages, yet early enough that few mutations have occurred (ideally only a single adaptive mutation per sequenced clone) and that no very fit clones have taken over the population. Because the fitness effects of beneficial mutations in glucose limited media were larger than in nitrogen limited media it was necessary to choose a later timepoint in the Nitrogen limited evolutions, to allow for there to be a sufficient fraction of the population carrying adaptive mutations. We believe this peak diversity makes these samples the most relevant for broadly assessing the adaptive mutational spectra.

• The use of statistics is not always clear. Please provide a clear indication of the statistical methods/tests used, eg for Figure 5.*

We thank the reviewer for this important point and have updated figures 2 and 5 and their corresponding legends for clarity surrounding statistical analysis used.

• The authors could include a supplementary Table, summarising their findings on GAT1 locus, since the text is extensive and it is difficult to put all the information into perspective.*

We note that row one of Table 1 in the main text is exactly this overview of the mutations observed at the GAT1 locus. These mutations plus specific location and their fitness remeasurements are shown in Figure 3 panel A, and detailed descriptions of the mutations for each clone are also available in the sortable table in Supplemental File 1. For these reasons we’ve not included an additional GAT1-specific table.

• The introduction is extremely detailed and informative, but at the same time quite lengthy; shortening it and only keeping the most relevant parts may increase readability.*

We appreciate the reviewer’s perspective but have not made substantial changes to remove information from the introduction as we feel that each of the subsections of the introduction are necessary to provide the appropriate context to the study.

• More detailed figure legends (which should also include a brief mentioning of the statistics & sample size) would benefit comprehensibility. For example the black lines in Figure S4 are not described anywhere in the text.*

We agree and have added further description of statistics used in legends throughout. Description of the black lines in Figure S4 has been included.

• "Many of the 332 clones ... were beneficial" à rephrase.*

We have updated this sentence to clarify our intent.

Reviewer #3 (Significance (Required)):

Apart from the elegant characterization of adaptive mutations, perhaps the most important part of the study is that it highlights the importance of a particular selection regime. Together, the findings extend our knowledge on this important topic.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Hays et al. studied the genomic changes that lead to adaptation under fluctuating nitrogen starvation. In addition to loss of function alleles, the authors identified adaptive gain-of- function alleles. Furthermore, their results demonstrate that Ty and microhomology-facilitated mutations in several candidate genes contribute substantially (though not exclusively) to the adaptation under nitrogen-limited serial transfer. Importantly, a novel lineage tracking method provides high resolution fitness measurements.

Despite the clear merits of the study, we also have a few relatively minor questions and suggestions

1. Please elaborate on the criteria they used to identify adaptive loci. The fact that these mutations occurred repeatedly is highlighted on Table 1, but perhaps numbers could also be included in the text, to increase clarity.
2. "Were also validated to a fitness effect of >0.01 in nitrogen-limited media". More details about the selection of this cut-off value need to be provided in either the text or the Methods section to increase clarity.
3. In Figure 3 it seems that the type of observed mutations was less important compared to the gene where the mutation occurred. Therefore, it seems that some genes, e.g. GAT1, contribute more to the observed fitness change. It would be beneficial if the authors discussed this observation.
4. What was the reason to select samples from the 88th generation for glucose and from the 192nd generation for nitrogen, as presented in Figure 5? How does this affects the observations?
5. The use of statistics is not always clear. Please provide a clear indication of the statistical methods/tests used, eg for Figure 5.
6. The authors could include a supplementary Table, summarising their findings on GAT1 locus, since the text is extensive and it is difficult to put all the information into perspective.
7. The introduction is extremely detailed and informative, but at the same time quite lengthy; shortening it and only keeping the most relevant parts may increase readability.
8. More detailed figure legends (which should also include a brief mentioning of the statistics & sample size) would benefit comprehensibility. For example the black lines in Figure S4 are not described anywhere in the text.
9. "Many of the 332 clones ... were beneficial"  rephrase.

Significance

Apart from the elegant characterization of adaptive mutations, perhaps the most important part of the study is that it highlights the importance of a particular selection regime. Together, the findings extend our knowledge on this important topic.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Hays et al. sequence and analyze the mutational spectrum from a set of S. cerevisiae strains evolved in a nitrogen limiting environment, and detail genes that recurrently are found to be mutated in a fluctuating nitrogen limiting environment. These data are contrasted to evolution under glucose limited environments and non-fluctuating environments. Specifically, Hays et al. observe a high proportion of Ty element-mediated mutations arising from strains evolved under the fluctuating nitrogen limiting regime. Their fitness data are robust and clearly demonstrate that these mutations reproducibly lead to improved fitness under nitrogen limitation (based on the authors' defined criteria). Overall, the observed bias of the high proportion of Ty-mediated mutation in fluctuating nitrogen starvation is unexpected and an important finding. Further, the discussion was thoughtful and well executed in detailing interpretations of the data more broadly. We are generally positive about this work and find the analyses robust and convincing. The authors should address the concerns listed below prior to acceptance/publication.

Significance

Major comments to be addressed:

The claim that the 3' UTR Ty insertions in MEP1 are apparently gain of function is very interesting. The authors should consider performing RT-PCR or strand specific RNAseq to see whether the antisense transcript is reduced and the MEP1 transcript is increased in the presence of the 3' UTR insertion. This would provide much stronger support for their claim that MEP1 3' Ty insertions are gain of function. Orientation information is critical to provide!

The authors seemed to miss a golden opportunity to measure Ty1 expression or transposition under fluctuating/non-fluctuating nitrogen starvation. Otherwise, the claims of increased Ty activity are unsupported. The authors measured an endpoint (Ty insertion), but this says nothing directly as to the rate of activity, although it is presumably correlated. However, based on the data one could argue activity may be equal in all environments, but the mutational events caused by Ty activity are uniquely selected for in fluctuating nitrogen starvation. As it stands, either model (increased activity vs. differential strength of selection) are equally likely. At a minimum, the authors should at least address this point.

In line with the above, we think the authors should soften some points in the discussion as it stands. For example: "The significant increase of Ty activity under this specific fluctuating nitrogen-starvation..." We feel the data does not exclusively support increased activity of Ty, that would require the aforementioned assays. As it stands, we feel this is more appropriate: ": "The significant increase of Ty insertions under this specific fluctuating nitrogen-starvation..."

Minor comments to be addressed:

Please provide a citation for the following statement "The single copy of Ty5 in the ancestor is known to be inactive and gives rise to no new insertions under either glucose or nitrogen limitation" - Voytas & Boeke. Nature 1992.

We found the following to be a confusing sentence: "Indeed, if global Ty derepression reflects a host-parasite coevolution that minimizes host cost and maximizes potential for survival of both, the role of transposons in host evolvability is important (Levin and Moran 2011)."

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Summary

The manuscript by Hays and colleagues described the spectrum of mutations that drive adaptation in nitrogen-limit environment by experimental evolution. The approach of serial transfer (fluctuating condition) allowed them to find that Ty insertion is the major mutation type for adaptive evolution. This was neither observed in nitrogen-limited condition when another experimental evolution approach, chemostat (non-fluctuating condition), was applied, nor in glucose-limited condition. The authors concluded that not only selection pressure itself but also how selection is applied are important to shape the adaptive events.

Major points

Both serial transfer and chemostat are commonly used approaches of experimental evolution. In the manuscript, the authors refer serial transfer to "fluctuating" condition because the low nitrogen source would be consumed to none during the interval of transfers. I am wondering whether the authors have estimated the nitrogen uptake (consumption) during the transfer intervals and whether the nitrogen was exhausted within 48 hours. Since this is not precisely controlled by experiment design, the "fluctuating" condition itself may be not stable during the long-term evolution. For example, as population evolved, the rate and the amount of nitrogen uptake might change. I feel a better experiment setup for "fluctuating" condition is like 24 hour "low-nitrogen (ammonium)" - 24 hour "no ammonium" and so on. If the adaptive mutations (e.g. adaptive Ty) specifically respond to such "fluctuating" condition rather than chemostat, the authors can measure their fitness in nitrogen starvation condition, which is expected to be fitter than mutants observed only in chemostat (e.g. copy number variation of nitrogen transporters).

The authors compared their results with published dataset using nitrogen-limitation chemostat and the mutation spectrum is different. In addition to the "fluctuating" and "non-fluctuating" difference as mentioned above, other factors need to be considered. First, the nitrogen-limited conditions in the two studies are different. The authors used 0.04% ammonium sulfate while Hong et al used "800 uM nitrogen regardless of the molecular form of the nitrogen", which may influence the mutation spectrum and need to be discussed. Second, bottlenecks were applied for each transfer in this study, in comparison with constant population size in chemostat, which will influence the efficiency of selection and further the evolutionary dynamics and outcomes. Thus, population size and bottlenecks need to take in to account to make comparisons of mutation spectrum.

The authors found that Ty mutagenesis accounts for a substantial number of adaptive mutations in nitrogen limitation. I am wondering for adaptive clones, whether Ty occurred independently or is more likely to co-exist with other drivers. What is the distribution of number of clones with one, two, and multiple mutations? If there is co-existence of driver mutations, what is the relative contribution of each to adaptation? The phenotypic validation of Ty mutagenesis for adaptation is expected while it seems only one case was presented in Figure 2 (mep1Ty−731427).

Statistical analysis needs to be reinforced in the manuscript, including but not limited to Figure 2 fitness comparison among clones with different genotypes, Figure 5 Ty enrichment comparison, etc.

Minor points

"For diploids, we only sequenced those with estimated fitness greater than diploidy alone would provide." Need edits.

"either through impacting alternate start (green triangle) or alternate stop sites (yellow and red triangles)." I do not see yellow and red triangles in Fig. 3.

Fig.2. FCY2 mutant fitness can be added as well?

"while we found only 212 novel Ty insertions in 488 glucose evolved clones (Figure 5B)" The value in the text does not match the one in the figure.

In addition to adaptive Ty insertion, what is the genome-wide distribution or characteristics of other Ty, especially for nitrogen-limited condition? Is that distinct from glucose-limited condition?

"Studies determining at which step(s) of the Ty life cycle nitrogen starvation shapes ty activity would be needed to determine the specific mechanism underlying the increase in transposon insertions." Here "ty" => "Ty"

Significance

The manuscript is a follow-up work of Levy et al. 2015 and Blundell et al. 2019. In general, the research is interesting and point out the important role of Ty for adaptive evolution in nitrogen-limited environment. It also compared the spectrum of adaptive mutations in response to nitrogen limitation by serial transfer (this work) and chemostat (especially the work of Gresham lab). The paper is well-written as well. Audience from the field of genetics, genomics and evolution will be interested in this work.

My field of expertise: genetics, experimental evolution, budding yeast

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

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

Summary: The authors use an unclassified quaranjavirus, Wǔh�n mosquito virus 6 (WuMV-6), to demonstrate the possibility of orthomyxvirid global transmission dynamic analyses. The focused surface protein analysis strongly indicates a vertebrate host for WuMV-6 in addition to the insect host. The analysis is then expanded to other quaranjaviruses, which differ considerably in their surface glycoproteins, indicating a complex evolution. Finally, the authors scientifically demonstrate that orthomyxovirids are undersampled and hence that this family will have to expand considerably in the future.

Major comments: none

We thank the reviewer for a succinct summary of our study and we are very glad the key messages were sufficiently clear.

Minor comments: The article lacks precision and hence some global edits are in order. Generally:

1. For clarity to the reader, please introduce the family Orthomyxoviridae, i.e., its current official composition (i.e., 9 genera, 21 species, and 22 viruses) so the reader is not confused by terms such as "quaranjavirus" or "isavirus" etc.).

This is a fair request though we would prefer to err on the side of caution with regards to the precise number of taxonomic ranks given the flux viral taxonomy has experienced and in light of the deluge of new taxa being discovered all the time. We refer to the “traditional” view of orthomyxovirid taxonomy at the genus level, encompassing the genera described up until 2011.

After that, please clearly indicate which viruses are classified and which ones are not. For instance, the main virus dealt with in this paper is unclassified, and so are Astopletus and Ūsinis viruses.

We do not think this is reasonable since virtually all RNA viruses discussed in the text are not classified and their status as such has little bearing on any of our findings.

Please ensure correct spelling, including diacritics, of the viruses and abbreviations throughout: Wǔh�n mosquito virus 6 (WuMV-6); H�běi orthomyxo-like virus 2 [note the deletion of one "virus"]; Wēnlǐng orthomyxo-like virus 2

Thank you for the comment, we have added the diacritics where we could identify them but may have missed some.

For orientation of the reader, please refer to family groups of viruses as -virids (e.g., "orthomyxovirids", "human coronavirids", "some rhabdovirids"). This way it is clear to the reader that, for instance, "quaranjaviruses" refers to a genus-level group

Thank you, we agree that this adds much needed precision in terminology.

"influenza" is a disease. There are several viruses that can cause influenza; they belong to four different genera. Please scan for "influenza" and replace each either with a virus name (for instance, in the abstract, "...RNA viruses containing influenza A virus" or with a genus name (e.g., "alphainfluenzaviruses")

Our apologies for that misnomer. The text has been corrected.

Please ensure the differentiation of taxa (concepts), such as species, and viruses (things). Orthomyxoviridae cannot infect anything, it can also not be sampled etc. Orthomyxovirids, the physical members of Orthomyxoviridae can infect things. Most instances of "Orthomyxoviridae" should be replaced accordingly.

Thank you for the comment, this has been corrected as suggested.

In particular:

1. The title doesn't make much sense. Orthomyxovirids are not taxonomically incomplete - they are things that we simply may not have samples or may have characterized incompletely. Also, the analyses are largely restricted to quaranjaviruses. Hence, I would suggest "...genome evolution, and broad diversity of quaranjaviruses"

Our apologies for the confusion. The analyses we carried out to quantify evolutionary orthomyxovirid diversity likely waiting to be discovered was carried out on all known (at the time) members of ____Orthomyxoviridae____ and thus the title must still refer to the entire family rather than quaranjavirids. We felt that the term “taxonomic incompleteness” imparts on the reader exactly what the reviewer refers to, namely that new taxonomic ranks are likely to come as more evolutionary diversity gets uncovered. Alternative and more precise formulations, like referring to evolutionary incompleteness or something similar, would miss the fact that it is taxonomy that discretises the otherwise continuous evolutionary change.

Abstract: genomes are not employed and do not make money. Please replace "employed" with "used"

We have to respectfully disagree since the definition of the word “employ” also includes the meaning “to make use of”.

Re: point 6 above, Introduction: species/families etc. cannot be discovered. They are being established by people for viruses that may be discovered. Please fix here and elsewhere (in most cases, "species" should be replaced with "viruses")

We agree that taxonomic ranks are designated and not discovered and have changed the text accordingly.

P3, second paragraph: please place "jingmenviruses" in quotation marks as this is not an official term (yet). Please add "potentially" ("as potentially causing human disease"). Even the authors only speak of an "association" and do not fulfill Koch's postulates

We have to respectfully disagree here too. “Jingmenviruses” as a term is unambiguous in referring to a group of related segmented flaviviruses even though the groups is not officially assigned a taxonomic rank. We have altered the text to add uncertainty to the claim that jingmenviruses cause disease in humans.

P3, top right column: "e.g., the tick-borne Johnston Atoll quaranja- and thogotoviruses" is ambiguous. Please change to "e.g., the tick-borne quaranja- and thogotoviruses" or list particular viruses and clarify which belong to which genus

Apologies for the confusion. We fixed this instance.

P3, right column "smaller number" - change to "lower number"

We have altered the offending sentence in response to reviewer 2 and this combination of words is no longer present.

P3, right column "or only the polymerase" - makes no sense to the reader as it has not been introduced; and grammatically needs to be improved as the polymerase is also encoded on a segment. Likewise, PB1 makes no sense to unacquainted reader - maybe add a few sentences to the intro right after the family introduction on general genome composition and that PB1 is part of the polymerase holoenyzme?

We have altered the offending sentence in response to reviewer 2 but we take the point. We’ve added detail about the RNA-directed RNA polymerase of orthomyxovirids to the introduction.

P4: the Ebola virus glycoprotein is called GP1,2 [with 1,2 in subscript] (also Figure 2 legend)

Respectfully, while the reviewer is technically correct in that the glycoprotein of Ebola virus is referred to as GP_1,2 in proteomics literature (the 1,2 referencing the protein held together by a cysteine bridge post-cleavage), calling it GP is not out of place in evolutionary studies and the term “Ebola virus GP” is unambiguous to the reader.

P4: please change "West Africa" to "Western Africa" (the designation of the area by the UN)

Unfortunately, while we agree that the reviewer is correct in that the UN refers to the region as “Western Africa”, references to the “West African Ebola virus epidemic” are ubiquitous in the literature and thus we do not see the reason to change the term here either.

P6: change "with Rainbow / Steelhead trout orthomyxviruses" to "with mykissviruses (rainbow trout orthomyxovirus and steelhead trout orthomyxovirus)" [note that virus names are not capitalized except for proper noun components; hence also "infectious salmon anemia virus, bottom right column]

While we recognise that viruses related to infectious salmon anaemia virus discovered in trout have received a separate taxonomic designation we feel very strongly about not mentioning it in our manuscript. Our fear is that “mykissviruses” have been designated too hastily on the basis of a handful of representatives and that relatives discovered in the future may show an indiscernible continuum between “mykissviruses” and isaviruses, invalidating the former as a valid term. We would therefore strongly prefer to keep references to specific viruses rather than a taxonomic designation that may disappear so that a future reader may have an easier time with our study.

P6, right column: please change "RNA-dependent" to the IUPAC/IUB-correct "RNA-directed"

Done.

Figure 2 is too small. I could not figure out B with or without my confocals... Likewise S2, S3 are way too small. In Fig 2 legend, please place "spike" into lower case

We understand the reviewer’s concern here but Figure 2B was a compromise between vertical space available on a page, the number of taxa in the PB1 tree, and what we thought important to communicate - the variation in segment number across orthomyxoviruses and mapping of PB1 diversity to gp64 diversity. This was done at the expense of individual taxon name visibility whilst fully zoomed out. To remedy this Figure 2B was rendered in 300 dpi resolution such that zooming in will show individual taxon names clearly. We ultimately hope to publish our study in an online-only journal where printing will not present an issue. Likewise for figures S2 and S3. We have changed “Spike” to be lower case in the legend.

Figure 3: correct spelling of virus names (from top to bottom): rainbow trout orthomyxovirus, infectious salmon anemia virus, influenza C virus, influenza D virus, influenza A virus, influenza B virus, Wēnlǐng orthomyxo-like virus 2, Dhori virus, Thogoto virus, Jos virus, Aransas Bay virus, ... Johnston Atoll virus, Quaranfil virus, H�běi orthomyxo-like virus 2, Hǎin�n orthomyxo-like virus 2, Wǔh�n mosquito virus 6. Also apply to S6 and others where applicable.

The names for viruses in Figure 3 were taken directly from their NCBI records and since we do not show their accessions there is no other way to disambiguate them to the reader. We have, however, added the necessary diacritics where appropriate.

[PS: based on the somewhat backward, non-UNICODE editorial manager system, I am worried that the diacritics in virus names above are not rendered corretly. If so, please look up the Pinyin spelling of Wuhan, Hainan, Wenling etc. - easiest way is to search Wikipedia for the terns and then identify the Pinyin spelling, which is typically pointed out]

CROSS-CONSULTATION COMMENTS

I think we (all reviewers) are all largely in agreement - this is a very useful study; the manuscripts just needs various adjustments. I agree with the requests of the other two reviewers.

Reviewer #1 (Significance (Required)):

The strength of the paper is that it provides a road map on how undersampled taxa may be analyzed and which kind of information can be gleaned from these analyses. The paper also demonstrates that the analysis of seemingly "unimportant" viruses can prove important. The limitation of the paper is that there is no true novel revelation here. The sampling sites of WuMV-2 GenBank records already suggest broad distribution, which often goes along with sequence diversity; the continued discovery of orthomyxovirids in metagenomic studies clearly implied undersampling (but it is nice to have this "gut feeling" scientifically fortified now). The paper is useful for evolutionary virologists, virus taxonomists, orthomyxovirid specialists, and invertebrate virologists.

We respectfully disagree with the reviewer and believe they may have missed an important point raised by our study. We do not claim that a global distribution of WuMV6 is what makes it remarkable but that its sampled diversity is 1) sufficient to calibrate molecular clocks (in our experience this is not always the case for arthropod viruses) and 2) that WuMV6 has reached its current global distribution ____recently____.

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

This is a nice example of bringing together a variety of data from metatranscriptomic studies to answer fundamental evolutionary questions in the field of viral evolution. There is a focus on a single virus family, and although some might see this as a little restrictive, I think the 'deep-dive' presented in this paper leaves space for a relatively detailed and comprehensive analysis. No doubt, other studies will gain inspiration from the approach presented here and expand this work to other viral groups.

Overall, the paper is very well written, and the figures are of a very high quality. It is a shame that there are only 3 main figures in the paper because the supplementary figures are well presented and informative.

We thank the reviewer for the kind words.

The manuscript discusses the importance of host quite a bit, and for that reason it would have been nice to try and incorporate the host of the various viruses into the figures somehow (perhaps as a supplementary, since the trees are already quite busy). This might help orientate the reader).

While we appreciate that host information is of interest, we foresee several issues. For one, we refer to broad host classes (essentially arthropod versus vertebrate) because they are largely determined by membrane fusion protein classes, the actual focus of our study, which exhibit strong phylogenetic signal. Secondly, host information in metagenomic studies can be imprecise, incorrect or unavailable.

I have some minor comments or suggestions for the authors to consider below. Note, please use line numbers in the future for your submissions.

A paragraph in the discussion laying out the limitations of this approach would be useful to the reader and would make this excellent paper even more robust.

Thank you for the suggestion. We presume the reviewer is referring to our interpolation of orthomyxovirid diversity and included a few sentences about the limitations of this approach in the Discussion.

Pg 3. The sentence starting 'The vast majority of known orthomyxoviruses use one...' should be made into two sentences to make it easier to read. A second sentence for the arthropod description is the obvious edit.

We appreciate the suggestion and have included it in the manuscript.

Pg 3. 'The number of segments of orthomyxoviruses with genomes known to be complete varies from 6 to 8'. Rephrase to - 'Orthomyxoviruses genomes are known to have 6-8 segments, but many metagenomically discovered viruses in this group have incomplete genomes...etc...',

Thank you for the suggestion, it has been included.

Figure 1 - what do the white triangles mean? Are these the directions of reassortment? This should be explained in the legend...

We apologise for the omission, this is now explained.

New Zealand is covered up by the circular tree. It looks like there is a point which is partially obscured.

The reviewer spotted a mistake on our part here. The figure included the coordinates for Wellington, New Zealand when the detection was actually in Wellington Shire, Australia. This has been fixed.

PD analysis - t I think you assume that viruses are static in this analysis. As we all know, they continue to mutate and eventually new species will evolve. Is it possible to consider the mutation rate in this analysis and the evolution of new variants/ eventually leading to new species? It might be complicated, and maybe a matter for future work, but it might be worth discussing this as a limitation at the very least. Especially when extrapolating to the future (although you do not extrapolate too far, so maybe this is not an issue here...). You could choose to discuss this in relation to the bird analogy (which was great), and compare the rate of mutation which will lead to the evolution of new species on a totally different time scale.

We appreciate the point raised by the reviewer and while we wholly agree that the possibility of new viral taxa arising over time is an important caveat, we felt the discussion ends up being rather short. On one hand taxa definitions for different viral groups can be different, and on the other speciation in RNA viruses is difficult to place in absolute time because of a phenomenon called time-dependence of evolutionary rates. Methods accounting for the latter using sophisticated models or external calibration points would seem to imply that speciation timescales exceed those of research.

Discussion: When discussing the hypothesis that WMV6 diversity is a result of repeat exposure to vertebrate hosts, can you also describe the alternative hypothesis here, and why the evidence leads you to put more weight on the former.

This is a fair question and we have mentioned an alternative hypothesis in the discussion that’s been brought up by our colleagues before. It’s a hypothesis that alternating between different hosts induces divergent selection pressures on gp64. We contend that since gp64 proteins are thought to use a highly conserved host receptor (NPC1) we think it likely that no major changes are required when switching hosts. We are open to discussing other alternatives if the reviewer has suggestions.

CROSS-CONSULTATION COMMENTS

Seems like we are all in agreement and that after some minor adjustments this will be an excellent contribution.

Reviewer #2 (Significance (Required)):

Please see my review above. I did not use your formatting suggestions since I only saw it upon completing my review.

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

Summary

This manuscript describes the use of data from metagenomic analyses to make inferences about the evolutionary and geographic history of the Orthomyxoviridae family of viruses and their hosts. Data from Wuhan Mosquito Virus 6 (WMV6) derived from various RNA-seq analyses is used to analyse loss and gain of virus segments over time, the time since the last common ancestor of these segments and the selection pressure acting on different genes. These results are used to hypothesise about which species have vectored this virus in the past and their geographic distribution. The additional phylogenetic diversity provided by characterisation of additional viruses of this species is quantified and projected into the future to demonstrate the value of further work in this area. The study also demonstrates more generally the benefit of additional sequencing and of characterising viruses in metagenomic datasets, even in cases where novel viruses are not identified.

Major Comments

The methodology in this manuscript appears to be sound and the results support the conclusions. Appropriate and detailed analyses have been performed and are described in detail. Code is provided to allow the results to be reproduced. The figures are informative and very well presented. I do not think any additional analyses are required.

We thank the reviewer for the kind words.

Minor Comments

The manuscript is a little hard to follow in places. I think a brief introduction of WHV6 in the introduction section would help with this - where has it been isolated previously and what is known about its evolutionary history (if anything), how is it related to other Orthomyxoviruses. This information is included later but it would improve the flow of the paper to include it in the introduction.

We apologise for the inconvenience and agree with the reviewer. We have improved the flow of the manuscript per reviewer suggestion.

I think including a little more about the Method in the Results section would also be helpful, to save the reader jumping back and forth in order to understand the results. For example, at the beginning of the results section, briefly detailing how many samples were included, their broad geographic location and what the analysis is intended to show (e.g. "three full length sequences isolated from China, seven from Australia [...], between 1995 and 2019, were used to generate a reassortment network, in order to show.....") would be helpful. Each of the subsections of the Results would benefit from something similar.

Apologies for the lack of clarity on our part. We have added more methodological information to each section in the results.

Although it is clear in the Materials and Methods which datasets have been included, it is less apparent why these were selected. For example, in Figure 1A there are five countries listed - are these countries for which a particularly large amount of full length sequences were available or for which any full length sequence is available? Similarly, for Figure 1B, are these all of the countries where a dataset has originated containing any segment of WHV6?

The confusion is entirely our fault as we have clearly not provided sufficient detail. This has been fixed now by explaining this better in the methods and Figure 1 legend.

In the Discussion, it is stated that the frequency and fast evolution of WMV6 place it uniquely to enable tracking of mosquito populations, however there is no evidence presented to support this - does WMV6 evolve faster or occur more frequently than other mosquito RNA viruses?

Our apologies for the jump in logic. We now expand on what we meant by the following sentence in the discussion: “In our experience, metagenomically discovered RNA viruses can be rare or, when encountered often, do not always contain sufficient signal to calibrate molecular clocks (Webster et al. 2015).”

CROSS-CONSULTATION COMMENTS

I also agree with the requests of the other two reviewers and that the manuscript will be in great shape once these are included.

Reviewer #3 (Significance (Required)):

This manuscript is very interesting, for the specific results presented here but, more importantly, in opening up further avenues for investigation. The study provides a proof of concept for using viruses derived from metagenomic data for specific and detailed evolutionary and ecological analyses of a single species. The scope of the analysis performed on WMV6 is not particularly broad, but it differs from the typical analysis of viruses in metagenomic datasets, which tends to focus on identification and characterisation of novel viruses only. I believe that this work is valuable to others working in the field, reveals additional potential in existing data and could provide inspiration for many future studies. To my knowledge, it is one of the first studies to focus on a single, fairly under-studied virus, and draw ecological conclusions based on only bioinformatic analyses.

I think the results presented here for WMV6 may be of interest to a specialised audience, but that the manuscript overall is valuable to a broad audience, including ecologists, evolutionary biologists and virologists conducting fundamental science research.

We appreciate the reviewer’s kind words.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary:

This manuscript describes the use of data from metagenomic analyses to make inferences about the evolutionary and geographic history of the Orthomyxoviridae family of viruses and their hosts. Data from Wuhan Mosquito Virus 6 (WMV6) derived from various RNA-seq analyses is used to analyse loss and gain of virus segments over time, the time since the last common ancestor of these segments and the selection pressure acting on different genes. These results are used to hypothesise about which species have vectored this virus in the past and their geographic distribution. The additional phylogenetic diversity provided by characterisation of additional viruses of this species is quantified and projected into the future to demonstrate the value of further work in this area. The study also demonstrates more generally the benefit of additional sequencing and of characterising viruses in metagenomic datasets, even in cases where novel viruses are not identified.

Major Comments:

The methodology in this manuscript appears to be sound and the results support the conclusions. Appropriate and detailed analyses have been performed and are described in detail. Code is provided to allow the results to be reproduced. The figures are informative and very well presented. I do not think any additional analyses are required.

Minor Comments:

The manuscript is a little hard to follow in places. I think a brief introduction of WHV6 in the introduction section would help with this - where has it been isolated previously and what is known about its evolutionary history (if anything), how is it related to other Orthomyxoviruses. This information is included later but it would improve the flow of the paper to include it in the introduction. I think including a little more about the Method in the Results section would also be helpful, to save the reader jumping back and forth in order to understand the results. For example, at the beginning of the results section, briefly detailing how many samples were included, their broad geographic location and what the analysis is intended to show (e.g. "three full length sequences isolated from China, seven from Australia [...], between 1995 and 2019, were used to generate a reassortment network, in order to show.....") would be helpful. Each of the subsections of the Results would benefit from something similar.

Although it is clear in the Materials and Methods which datasets have been included, it is less apparent why these were selected. For example, in Figure 1A there are five countries listed - are these countries for which a particularly large amount of full length sequences were available or for which any full length sequence is available? Similarly, for Figure 1B, are these all of the countries where a dataset has originated containing any segment of WHV6?

In the Discussion, it is stated that the frequency and fast evolution of WMV6 place it uniquely to enable tracking of mosquito populations, however there is no evidence presented to support this - does WMV6 evolve faster or occur more frequently than other mosquito RNA viruses?

CROSS-CONSULTATION COMMENTS

I also agree with the requests of the other two reviewers and that the manuscript will be in great shape once these are included.

Significance

This manuscript is very interesting, for the specific results presented here but, more importantly, in opening up further avenues for investigation. The study provides a proof of concept for using viruses derived from metagenomic data for specific and detailed evolutionary and ecological analyses of a single species. The scope of the analysis performed on WMV6 is not particularly broad, but it differs from the typical analysis of viruses in metagenomic datasets, which tends to focus on identification and characterisation of novel viruses only. I believe that this work is valuable to others working in the field, reveals additional potential in existing data and could provide inspiration for many future studies. To my knowledge, it is one of the first studies to focus on a single, fairly under-studied virus, and draw ecological conclusions based on only bioinformatic analyses.

I think the results presented here for WMV6 may be of interest to a specialised audience, but that the manuscript overall is valuable to a broad audience, including ecologists, evolutionary biologists and virologists conducting fundamental science research.

My expertise is in computational genomics, focused on RNA virus evolution.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

1) This is a nice example of bringing together a variety of data from metatranscriptomic studies to answer fundamental evolutionary questions in the field of viral evolution. There is a focus on a single virus family, and although some might see this as a little restrictive, I think the 'deep-dive' presented in this paper leaves space for a relatively detailed and comprehensive analysis. No doubt, other studies will gain inspiration from the approach presented here and expand this work to other viral groups.

2) Overall, the paper is very well written, and the figures are of a very high quality. It is a shame that there are only 3 main figures in the paper because the supplementary figures are well presented and informative.

3) The manuscript discusses the importance of host quite a bit, and for that reason it would have been nice to try and incorporate the host of the various viruses into the figures somehow (perhaps as a supplementary, since the trees are already quite busy). This might help orientate the reader).

4) I have some minor comments or suggestions for the authors to consider below. Note, please use line numbers in the future for your submissions.

• A paragraph in the discussion laying out the limitations of this approach would be useful to the reader and would make this excellent paper even more robust.

• Pg 3. The sentence starting 'The vast majority of known orthomyxoviruses use one...' should be made into two sentences to make it easier to read. A second sentence for the arthropod description is the obvious edit.

• Pg 3. 'The number of segments of orthomyxoviruses with genomes known to be complete varies from 6 to 8'. Rephrase to - 'Orthomyxoviruses genomes are known to have 6-8 segments, but many metagenomically discovered viruses in this group have incomplete genomes...etc...',

• Figure 1 - what do the white triangles mean? Are these the directions of reassortment? This should be explained in the legend...

• New Zealand is covered up by the circular tree. It looks like there is a point which is partially obscured.

• PD analysis - t I think you assume that viruses are static in this analysis. As we all know, they continue to mutate and eventually new species will evolve. Is it possible to consider the mutation rate in this analysis and the evolution of new variants/ eventually leading to new species? It might be complicated, and maybe a matter for future work, but it might be worth discussing this as a limitation at the very least. Especially when extrapolating to the future (although you do not extrapolate too far, so maybe this is not an issue here...). You could choose to discuss this in relation to the bird analogy (which was great), and compare the rate of mutation which will lead to the evolution of new species on a totally different time scale.

• Discussion: When discussing the hypothesis that WMV6 diversity is a result of repeat exposure to vertebrate hosts, can you also describe the alternative hypothesis here, and why the evidence leads you to put more weight on the former.

CROSS-CONSULTATION COMMENTS

Seems like we are all in agreement and that after some minor adjustments this will be an excellent contribution.

Significance

Please see my review above. I did not use your formatting suggestions since I only saw it upon completing my review.

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 use an unclassified quaranjavirus, Wǔhàn mosquito virus 6 (WuMV-6), to demonstrate the possibility of orthomyxvirid global transmission dynamic analyses. The focused surface protein analysis strongly indicates a vertebrate host for WuMV-6 in addition to the insect host. The analysis is then expanded to other quaranjaviruses, which differ considerably in their surface glycoproteins, indicating a complex evolution. Finally, the authors scientifically demonstrate that orthomyxovirids are undersampled and hence that this family will have to expand considerably in the future.

Minor comments:

The article lacks precision and hence some global edits are in order. Generally:

1. For clarity to the reader, please introduce the family Orthomyxoviridae, i.e., its current official composition (i.e., 9 genera, 21 species, and 22 viruses) so the reader is not confused by terms such as "quaranjavirus" or "isavirus" etc.).

2. After that, please clearly indicate which viruses are classified and which ones are not. For instance, the main virus dealt with in this paper is unclassified, and so are Astopletus and Ūsinis viruses.

3. Please ensure correct spelling, including diacritics, of the viruses and abbreviations throughout: Wǔhàn mosquito virus 6 (WuMV-6); Húběi orthomyxo-like virus 2 [note the deletion of one "virus"]; Wēnlǐng orthomyxo-like virus 2

4. For orientation of the reader, please refer to family groups of viruses as -virids (e.g., "orthomyxovirids", "human coronavirids", "some rhabdovirids"). This way it is clear to the reader that, for instance, "quaranjaviruses" refers to a genus-level group

5. "influenza" is a disease. There are several viruses that can cause influenza; they belong to four different genera. Please scan for "influenza" and replace each either with a virus name (for instance, in the abstract, "...RNA viruses containing influenza A virus" or with a genus name (e.g., "alphainfluenzaviruses")

6. Please ensure the differentiation of taxa (concepts), such as species, and viruses (things). Orthomyxoviridae cannot infect anything, it can also not be sampled etc. Orthomyxovirids, the physical members of Orthomyxoviridae can infect things. Most instances of "Orthomyxoviridae" should be replaced accordingly.

In particular:

1. The title doesn't make much sense. Orthomyxovirids are not taxonomically incomplete - they are things that we simply may not have samples or may have characterized incompletely. Also, the analyses are largely restricted to quaranjaviruses. Hence, I would suggest "...genome evolution, and broad diversity of quaranjaviruses"

2. Abstract: genomes are not employed and do not make money. Please replace "employed" with "used"

3. Re: point 6 above, Introduction: species/families etc. cannot be discovered. They are being established by people for viruses that may be discovered. Please fix here and elsewhere (in most cases, "species" should be replaced with "viruses")

4. P3, second paragraph: please place "jingmenviruses" in quotation marks as this is not an official term (yet). Please add "potentially" ("as potentially causing human disease"). Even the authors only speak of an "association" and do not fulfill Koch's postulates

5. P3, top right column: "e.g., the tick-borne Johnston Atoll quaranja- and thogotoviruses" is ambiguous. Please change to "e.g., the tick-borne quaranja- and thogotoviruses" or list particular viruses and clarify which belong to which genus

6. P3, right column "smaller number" - change to "lower number"

7. P3, right column "or only the polymerase" - makes no sense to the reader as it has not been introduced; and grammatically needs to be improved as the polymerase is also encoded on a segment. Likewise, PB1 makes no sense to unacquainted reader - maybe add a few sentences to the intro right after the family introduction on general genome composition and that PB1 is part of the polymerase holoenyzme?

8. P4: the Ebola virus glycoprotein is called GP1,2 [with 1,2 in subscript] (also Figure 2 legend)

9. P4: please change "West Africa" to "Western Africa" (the designation of the area by the UN)

10. P6: change "with Rainbow / Steelhead trout orthomyxviruses" to "with mykissviruses (rainbow trout orthomyxovirus and steelhead trout orthomyxovirus)" [note that virus names are not capitalized except for proper noun components; hence also "infectious salmon anemia virus, bottom right column]

11. P6, right column: please change "RNA-dependent" to the IUPAC/IUB-correct "RNA-directed"

12. Figure 2 is too small. I could not figure out B with or without my confocals... Likewise S2, S3 are way too small. In Fig 2 legend, please place "spike" into lower case

13. Figure 3: correct spelling of virus names (from top to bottom): rainbow trout orthomyxovirus, infectious salmon anemia virus, influenza C virus, influenza D virus, influenza A virus, influenza B virus, Wēnlǐng orthomyxo-like virus 2, Dhori virus, Thogoto virus, Jos virus, Aransas Bay virus, ... Johnston Atoll virus, Quaranfil virus, Húběi orthomyxo-like virus 2, Hǎinán orthomyxo-like virus 2, Wǔhàn mosquito virus 6. Also apply to S6 and others where applicable.

[PS: based on the somewhat backward, non-UNICODE editorial manager system, I am worried that the diacritics in virus names above are not rendered corretly. If so, please look up the Pinyin spelling of Wuhan, Hainan, Wenling etc. - easiest way is to search Wikipedia for the terns and then identify the Pinyin spelling, which is typically pointed out]

CROSS-CONSULTATION COMMENTS

I think we (all reviewers) are all largely in agreement - this is a very useful study; the manuscripts just needs various adjustments. I agree with the requests of the other two reviewers.

Significance

The strength of the paper is that it provides a road map on how undersampled taxa may be analyzed and which kind of information can be gleaned from these analyses. The paper also demonstrates that the analysis of seemingly "unimportant" viruses can prove important. The limitation of the paper is that there is no true novel revelation here. The sampling sites of WuMV-2 GenBank records already suggest broad distribution, which often goes along with sequence diversity; the continued discovery of orthomyxovirids in metagenomic studies clearly implied undersampling (but it is nice to have this "gut feeling" scientifically fortified now). The paper is useful for evolutionary virologists, virus taxonomists, orthomyxovirid specialists, and invertebrate virologists.

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

Learn more at Review Commons

---

Reply to the reviewers

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

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary: In this study, the authors attempt to discover new effector mechanisms of IFN-gamma mediated cell autonomous immunity using a human malignant lung cell line infected with T. gondii. A genome-wide screen discovered NF2 (neurofibromatoris 2) as a transcriptional modulator of IRF-1 dependent cell autonomous response induced by IFN-gamma. To increase the chance of discovering true effectors, a focused screed was performed, which yielded the E3 ligase RNF213. This E3 ligase is constitutively expressed but its levels are further upregulated upon exposure to IFN-gamma. Functional studies indicate that RNF213 plays a role in both the basal restriction and the induced/enhanced restriction of T. gondii growth, which occurs inside a well-defined vacuole. Data further showed that RNF213 associates with the parasite vacuole, both at basal and activated states, and is associated with molecular players involved in non-canonical autophagy. However, further analysis indicated that non-canonical autophagy was itself not required for growth restriction mediated by RNF213. Additional studies also indicated a role for RNF213 in cell autonomous immunity to an intracellular bacterium and a virus. In summary, the screens identified a regulators of the antimicrobial transcriptional and effector programs induced by interferons.

Major comments:

The title of the article seems misleading as the experiments and data described in the study does not truly provide a mechanistic basis for how pathogen growth restriction occurs. A new title that better reflects the limited extent of the advance reported here should be selected.

Because RNF213 is constitutively expressed, it is possible that it could independently downregulate parasite growth without the need for other interferon-inducible effectors. Have the authors determined whether overexpression is sufficient in cells that are not exposed to interferon treatment?

Minor comments:

Figure 4F. Labelling to highlight key structures in this EM photograph would be

Referees cross commenting

The comment by Reviewer 1 regarding lack of ubiquitin staining of parasitophorous vacuole should be reconsidered, because it is shown in Figure 4 by use of FK2 antibody.

Significance

Knowledge of how human cells execute cell-autonomous growth restriction of intracellular parasites remains rudimentary. Thus, by identifying regulators of the antimicrobial and effector programs induced by interferons, this study represents an notable advance. However, it did not elucidate a novel effector mechanism of pathogen growth restriction.

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 "Molecular Basis for Interferon-mediated Pathogen Restriction in Human Cells" by Sumit Matta et al. describes the identification of RNF213 (ring finger protein 213), an E3 ubiquitin ligase, as essential for IFNg mediated control of T. gondii in human cells (A549, THP-1, HFF). RNF213 was found by a CRISPR/Cas9 based screen of IFNg stimulated genes in A549 cells. Additional data obtained from a genome wide CRISPR/Cas9 screen (using the Brunello library from Addgene) found previously known essential genes for Toxoplasma control such as IRF1, STAT1, JAK2, IFNGR1/2 as well as one novel gene, NF-2, as being important for IFNg mediated Toxoplasma control. Functional data reveal that RNF213 is recruited to the T. gondii PV and that ubiquitination is found at the RNF213 positive PVs. For RNF213 function, ATG5 appears not to be of critical importance. Finally, functional assays determined that RNF213 is also required for the IFNg mediated control of the intracellular pathogen M. tuberculosis and the IFNb mediated control of VSV.

The study is very well performed and executed, the findings are of broad interest and advance our understanding of host-pathogen relationship on a molecular level.

There are some critical points that should be addressed by the authors:

The authors use a vacuolar size growth assay. The authors should verify / compare their assay to determine Toxoplasma control, to e.g. qPCR analysis or 3H-Uracil incorporation in the RNF213ko A549 and THP-1 cells.

All experiments were conducted with the CTG strain of T. gondii which is a type III strain, the authors should investigate whether RNF213 can also restrict more virulent type II and type I toxoplasma strains.

The induction (RNA / protein) of RNF213 by titrated amounts of IFNg and IFNb should be investigated and compared in A549 cells.

Minor points:

What is the induction of RNF213 in NF-2 deficient cells after IFN stimulation?

Fig. 1E There are three genes indicated in the top left quadrant (PTEN/TSC1/TSC2) but only 2 green data points shown? Why?

Significance

The presented data corroborate and extend a study by Hernandez et al. (mBio. 2022 Oct26;13(5):e0188822. doi: 10.1128/mbio.01888-22. Epub 2022 Sep 26.) with regard to cell autonomous T. gondii defense and add information with regard to immunity against M. tuberculosis and VSV.

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

Matta et al. investigated, via an initial crispr screen, the host cellular factors involved in T. gondii growth restriction. In the past, different pathways have been implicated in parasite growth suppression including GBPs and IDO1 via tryptophan restriction. As the authors carefully note, the prior studies have notable caveats, and in human cells, other pathways must be involved. To find genes required for T. gondii growth suppression, the authors set a screen to read out parasite vacuole size after IFNg treatment, in a model cell system (A549). In this way, IFNg is the ultimate upstream cytokine that triggers parasite growth restriction and loss of components downstream of the IFNg-IFNgR-STAT1 pathway should be implicated as host anti-parasite effectors. Thus, the screen has the potential to uncover new pathways involved in host resistance.

Following the screen, the authors initially found NF2 (see below) and then refining their approach to use an ISG-targeted screen, following which they focused on RNF213, recently described as an LPS E3 ligase. The authors chose to divide their manuscript into these two parts. The main critique of the manuscript concerns the fact that neither of the two parts is fully developed.

Significance

Critique:

1. NF2 was clearly a top hit in the genome-wide screen and loss of NF2 by targeted knockout clearly recapitulated the screen result. However, (i) what the mechanism of growth restriction by NF2? After Figure 2, NF2 is dropped and the authors focus on RNF213. (ii) NF2 is not regulated (obviously) by IFNg (Fig. 2A, WT +/- IFNg). But what is the link between the IFN signaling pathway and NF2? It seems that the NF2 KO has less ISG expression (heat map, 2D) although this data is not convincingly shown: Proteomics seems essential here in addition to the transcript measurements. If NF2 regulates an "upstream" event in the IFNg pathway (implied in 2F, secondary screen), the authors should be able to track down the point at which it exerts its effect.
2. RNF213 clearly plays an unexpected and important role in parasite restriction. However, the mechanisms involved are not clear. (i) The authors state in Figure 4, that RNF213 co-localized with ubiquitinated parasite-contained vacuoles, but this is not shown (there is no Ub staining in Figure 4). (ii) The effects of RNF213 are independent of ATG5. However, what is missing is the overall quantification of Ub +/- IFNg in control, RNF213 and ATG5 KO cells (di-Gly MS seems essential here). (iii) What is the effect of parasitophorous vacuole UB in the RNF213 WT vs. KO cells? (iv) The authors explain that RNF213 is not an obvious ISG in that its transcript does not fit with canonical ISG expression. Therefore, how do the authors link RNF213 activity with the IFNg pathway? (v) Finally, since we now know that RNF213 ubiquitinates LPS, further controls using this pathway may be useful (especially as LPS activates the type 1 IFN response).

Further comments:

1. The microscopy images are too small in my view (throughout).
2. 2E should be should as bar graphs.

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

Learn more at Review Commons

---

Reply to the reviewers

---

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

All the conclusions are based on solid evidence and convincing, and the methodology are in detail to follow or repeat. The writing of the manuscript is logical and easy to follow.

We thank the reviewer for these comments

1. The mutation experiments indicated that nkd enhanced the phenotype of scr, but there is no leaf phenotype variation in nkd muations, this is some way difficult to understand, it would much better if the authors can give much more explanation in the discussion.

We have added more discussion on this point. One possibility is that collectively the four genes function redundantly, however, due to the transcriptional negative feedback loop discovered here (Figure 3B), when NKD genes are mutated then SCR expression is enhanced, hence phenotypic perturbations are less likely to be observed than when SCR genes are mutated.

2.The word green millet in the first paragraph should be changed to green foxtail. Millet means domesticated small cereal grains, such as foxtail millet, finger millet, proso millet etc.

We thank the reviewer for this feedback and have made the suggested change.

Reviewer #1 (Significance (Required)):

The manuscript, which titled Mutations in NAKED-ENDOSPERM IDD genes reveal functional interactions with SCARECROW and a maternal influence on leaf patterning in C4 grasses by Hughes et al., first reported that SCR works regulating both leaf inner pattern and epidermal stomatal patterning in the C4 model plant green foxtail. The functional difference of this gene in Setaria from that in maize and rice indicated that the inner leaf cell patterning regulation of SCR is not a characteristic of C4 Species; this gave us insight understanding of the complex of C4 leaf cell patterning. In addition to this important discover, the authors found that mutations in NKD IDD genes enhance loss of function scr phenotypes in the leaves of C4 maize and Setaria but not in the C3 rice, indicating NKD IDD was involved in the leaf cell patterning in C4 species, but no in C3. They also identified a maternal effect on cell-type patterning in maize leaves that are initiated during embryogenesis.

We thank the reviewer for their kind comments and suggestions.

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

The leaf anatomy that distinguishes C4 from C3 plants has been known for decades, with veins in C4 plants separated by 1 to 3 (generally 2) mesophyll cells whereas those in C3 plants are considerably farther apart. This anatomical pattern appears to be critical for the function of the C4 pathway, which under some environmental conditions is a more efficient way to fix carbon than the C3 pathway. Despite the obvious importance of close vein spacing, the genetic mechanisms that control it have been surprisingly difficult to untangle. The statement on the bottom on p. 2 ("To date, very few regulators of cell-patterning in inner leaf tissues have been identified...") is an understatement. The paper by Hughes et al. offers a major step in uncovering the basis of C4 vein spacing.

We thank the reviewer for this feedback and agree that this work represents a major step forward in understanding C4 vein spacing.

The authors build on their previous work in Scarecrow-like proteins in maize and rice. In maize, SCR controls patterning of the mesophyll, whereas in rice it controls development of stomata. This paper pursues the possibility that the differences in SCR roles may have to do with interacting proteins. Based on work in Arabidopsis the authors focus on proteins with an indeterminate domain (IDD) and specifically on the NAKED ENDOSPERM genes.

The paper presents an analysis of an impressive set of mutants in three species. A major step in this paper is the comparison among three species of grasses - maize, rice, and Setaria - rather than the more common two species, usually maize and rice. Maize and rice differ in photosynthetic pathway but they also differ in many other traits that reflect the ca. 50 million years since their last common ancestor. Setaria is, like maize, C4 and the two species are more closely related to each other than either is to rice, although they represent two independent acquisitions of C4. This paper shows that SCR orthologs control stomatal patterning in both rice and Setaria implying that the stomatal function of SCR may be ancestral in the grasses and also is not directly connected to photosynthetic pathway.

The availability of allelic combinations of SCR and NKD in maize in particular permits the inference of possible maternal effect on the vein spacing phenotype, although exactly how this happens remains unclear.

The discussion provides a careful and logical assessment of the state of knowledge on SCR and IDD proteins in general, and the new data on SCR and NKD in particular. Many questions remain unresolved, and many additional experiments could be suggested. However, the power of the genetics and the phenotypic analysis together provide a novel direction for research on vein spacing. I will refrain in this review from suggesting what additional information would be nice to have since I think a review should assess the quality of the paper as it stands, not as it could be with months more of work.

My only really substantive suggestion is that the micrographs of the Setaria leaves need to be improved. Specifically, in Figure 6E it is hard to see the details of the fused veins. Either the section is too thick or the camera was not focused properly. Because this image in particular is central to the entire paper I would recommend aiming for the clarity of the images of Zea cross sections, which are fine.

We thank the reviewer for this suggestion. Obtaining leaf cross section micrographs from the Setaria scr1;scr2;nkd mutants was extremely challenging as the growth phenotype is so severe (Figure 5), meaning that the available leaves are small and extremely fragile. Multiple attempts to fix and section leaves using a microtome failed, with leaves consistently collapsing. In our hands, Setaria is not as amenable to fresh vibratome sectioning as maize, and combined with the additional challenges of handling the tiny triple mutant leaves mean that the resultant images are not of the same quality as the maize figures. We have included a supplemental figure (Figure S8) with additional examples of fused veins identified in our screening.

Very minor point: p. 3 - "double Zmscr1;Zmscr1h mutants" - what does the "h" in Zmscr1h refer to?

In this context h refers to this gene being a homeologous gene duplicate, as first explained in Hughes et al. (2019). We have included an explanation in the revision.

Reviewer #2 (Significance (Required)):

Strengths of the paper are 1) the inclusion of three species to help determine which aspects of the gene function may be ascribed to C4; 2) thoughtful and comprehensive genetic analysis; 3) careful sections of leaves; 4) outlines of the limitations of the approach. Limitations (several of which the authors acknowledge in the Discussion) include a general lack of molecular genetic data (protein interactions, DNA binding sites, RNA-seq, etc.). While this information would be great to have, I think the strength of the genetics is such that the paper will be foundational for future work in any case. The one bit of additional data that would be ideal would be information bearing on the two mechanistic hypotheses laid out on p. 10. The model that SCR and NKD promote cell division and specify mesophyll identity is the opposite of the model that SCR and NKD inhibit vein formation. An experiment that helped point the reader toward one or the other of these models would be very valuable.

We agree that an experiment that could distinguish these possibilities would be extremely valuable, and will undoubtedly be the subject of future experimentation.

The paper fills a critical gap. Little to nothing is known about how the internal anatomy of leaves is patterned and the data presented provide evidence that SCR and NKD are two important players. The paper also provides a conceptual advance in offering a couple of genes and some plausible mechanisms of how they might function.

The audience will be primarily developmental geneticists and physiologists. The paper addresses an important problem that is of broad interest to developmental biologists and is potentially important for global agriculture.

We thank the reviewer for their kind comments and suggestions.

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

The manuscript of Hughes et al. aimed to demonstrate the functional interactions between Naked-Endosperm IDD genes and the transcription factor SCARECROW and a maternal effect on leaf patterning in C4 grasses. To this end, the authors conducted a greenhouse and labor experiment to create mutants of related genes and assess the expression of these genes through qRT-PCR combined with fluorescence microscopic images in Rice, Maize, and Setaria. They found an increase in the proportion of fused veins with no intervening mesophyll cells in scr;nkd mutants in C4 species (Maize and Setaria) but not in C3 species (rice). In the end, they revealed a maternal effect of derived NKD on patterning cells in leaf primordia during embryogenesis.

Major comments - Optional: the authors should have conducted a whole transcriptome experiment through RNA-seq on the mutants as compared to the controls to check if these genes were significantly up-related followed by qRT-PCR for validation. By doing so, the authors should be able to get a broad overview of all key plays involved in leaf patterning.

We agree with the reviewer that it would be useful to have this data, and such an approach will undoubtedly inform future research.

• Optional: although the authors may evoke the statistical significance of observing fused veins in mutants sr;nkd, the presence of fused veins in one mutant Svscr1;Svscr2 and Zmscr1-m2;Zmscr1h-m1 may contradict the claim that the authors made regarding the association between scr and nkd. Moreover, the sampling size is not also large enough to draw a substantial conclusion.

We disagree with the reviewer that our sampling size is not large enough to draw a substantial conclusion. In maize we surveyed 11 quadruple mutants and 588 veins. Although this phenotype is occasionally seen in Zmscr1;Zmscr1h mutants, it is far more penetrant in Zmscr1;Zmscr1h;Zmnkd1;Zmnkd2 quadruple mutants and easily distinguished by eye when viewing each mutant, the statistical analysis only serves to make this point. In Setaria we agree that the differences are less stark, and the sampling size is necessarily lower due to the challenges of working with the triple mutant leaves which are extremely small and fragile (far more so than the maize quadruple mutant leaves). We have already included discussion as to why the phenotype may be less penetrant in setaria. Together we think that the fact the direction of the phenotype matches that of maize is convincing evidence that the increase in fused veins is a real consequence of combining the scr and nkd mutations.

• There are two copies of nkd in maize but only one copy in rice and Setaria. Does the presence of two copies in maize has any evolutionary or functional meaning? Does the presence and absence of one or two copies has any effect on leaf patterning? It would be interesting to discuss this in the discussion section.

We thank the reviewer for this comment and have added discussion of this in the manuscript. This situation is common in maize, which underwent a more recent whole genome duplication since its divergence from rice and setaria. Most of these gene-pairs function redundantly, however, there is evidence of functional divergence in terms of expression in some gene-pairs. We have added a sentence in the results explaining why we think the presence of two NKD gene copies in maize is unlikely to have functional significance in this case.

• The methods section is not easy to read for a non-specialized audience. I suggest providing an explanation of the abbreviations used to describe mutants.

We thank the reviewer for this suggestion and have made the suggested change.

• For the results section, you should provide a table summarizing the differences between mutants and controls regarding the leaf structure.

We have added such a table at the end of the results section and referred to it in the discussion.

Minor comments: - "Zmscr1-m2;Zmscr1h-m1 seed were" seeds instead

We have made the suggested change.

• "Loss of NKD gene function enhances SCR mutant phenotypes in maize and setaria" This section is confusing because several perturbations were observed in triple mutants of Setaria and quadruple mutants of Maize as compared to their double mutants (Svscr1;Svscr2 and Zmscr1;Zmscr1h). You should rewrite this subtitle for clarity.

We have changed this sub title to read “In maize and setaria, but not in rice, nkd loss of function mutations enhance scr mutant phenotypes”

• "The accumulation of transcripts in the ground meristem cells" How do you estimate the accumulation of transcripts? What do you mean by the accumulation of transcripts? What do you consider transcripts?

We use this term as opposed to ‘gene expression in the ground meristem cells’ because we do not know whether the presence/absence/level of detectable RNA is regulated by transcriptional or post-transcriptional mechanisms.

Reviewer #3 (Significance (Required)):

The manuscript of Hughes et al. is very interesting in the context of C4 photosynthesis research because there are many transcription factor candidates involved in the development of C4 leaf anatomy but few of them have been validated. However, a whole comparative transcriptome of mutants and controls should provide a broad overview and probably new insight into key players involved in leaf patterning.

We agree with the reviewer that this would be of great interest, but we feel it is beyond the scope of this study and will be a productive avenue of future research.

This study goes far beyond the simple validation by outlining the potential interactions between transcription factors. The authors made a substantial effort by combining gene expression results with visual data that strengthen the quality of this manuscript. Therefore, this manuscript is very interesting for the C4 research communities and for the field of developmental biology.

We thank the reviewer for their kind comments and suggestions.

A plant biologist working on the evolution and regulation of morphological characters using transcriptomics and genomics.

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 of Hughes et al. aimed to demonstrate the functional interactions between Naked-Endosperm IDD genes and the transcription factor SCARECROW and a maternal effect on leaf patterning in C4 grasses. To this end, the authors conducted a greenhouse and labor experiment to create mutants of related genes and assess the expression of these genes through qRT-PCR combined with fluorescence microscopic images in Rice, Maize, and Setaria. They found an increase in the proportion of fused veins with no intervening mesophyll cells in scr;nkd mutants in C4 species (Maize and Setaria) but not in C3 species (rice). In the end, they revealed a maternal effect of derived NKD on patterning cells in leaf primordia during embryogenesis.

Major comments

• Optional: the authors should have conducted a whole transcriptome experiment through RNA-seq on the mutants as compared to the controls to check if these genes were significantly up-related followed by qRT-PCR for validation. By doing so, the authors should be able to get a broad overview of all key plays involved in leaf patterning.
• Optional: although the authors may evoke the statistical significance of observing fused veins in mutants sr;nkd, the presence of fused veins in one mutant Svscr1;Svscr2 and Zmscr1-m2;Zmscr1h-m1 may contradict the claim that the authors made regarding the association between scr and nkd. Moreover, the sampling size is not also large enough to draw a substantial conclusion.
• There are two copies of nkd in maize but only one copy in rice and Setaria. Does the presence of two copies in maize has any evolutionary or functional meaning? Does the presence and absence of one or two copies has any effect on leaf patterning? It would be interesting to discuss this in the discussion section.
• The methods section is not easy to read for a non-specialized audience. I suggest providing an explanation of the abbreviations used to describe mutants.
• For the results section, you should provide a table summarizing the differences between mutants and controls regarding the leaf structure.

Minor comments:

• "Zmscr1-m2;Zmscr1h-m1 seed were" seeds instead
• "Loss of NKD gene function enhances SCR mutant phenotypes in maize and setaria" This section is confusing because several perturbations were observed in triple mutants of Setaria and quadruple mutants of Maize as compared to their double mutants (Svscr1;Svscr2 and Zmscr1;Zmscr1h). You should rewrite this subtitle for clarity.
• "The accumulation of transcripts in the ground meristem cells" How do you estimate the accumulation of transcripts? What do you mean by the accumulation of transcripts? What do you consider transcripts?

Significance

The manuscript of Hughes et al. is very interesting in the context of C4 photosynthesis research because there are many transcription factor candidates involved in the development of C4 leaf anatomy but few of them have been validated. However, a whole comparative transcriptome of mutants and controls should provide a broad overview and probably new insight into key players involved in leaf patterning.

This study goes far beyond the simple validation by outlining the potential interactions between transcription factors. The authors made a substantial effort by combining gene expression results with visual data that strengthen the quality of this manuscript. Therefore, this manuscript is very interesting for the C4 research communities and for the field of developmental biology.

A plant biologist working on the evolution and regulation of morphological characters using transcriptomics and genomics.

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 leaf anatomy that distinguishes C4 from C3 plants has been known for decades, with veins in C4 plants separated by 1 to 3 (generally 2) mesophyll cells whereas those in C3 plants are considerably farther apart. This anatomical pattern appears to be critical for the function of the C4 pathway, which under some environmental conditions is a more efficient way to fix carbon than the C3 pathway. Despite the obvious importance of close vein spacing, the genetic mechanisms that control it have been surprisingly difficult to untangle. The statement on the bottom on p. 2 ("To date, very few regulators of cell-patterning in inner leaf tissues have been identified...") is an understatement. The paper by Hughes et al. offers a major step in uncovering the basis of C4 vein spacing.

The authors build on their previous work in Scarecrow-like proteins in maize and rice. In maize, SCR controls patterning of the mesophyll, whereas in rice it controls development of stomata. This paper pursues the possibility that the differences in SCR roles may have to do with interacting proteins. Based on work in Arabidopsis the authors focus on proteins with an indeterminate domain (IDD) and specifically on the NAKED ENDOSPERM genes.

The paper presents an analysis of an impressive set of mutants in three species. A major step in this paper is the comparison among three species of grasses - maize, rice, and Setaria - rather than the more common two species, usually maize and rice. Maize and rice differ in photosynthetic pathway but they also differ in many other traits that reflect the ca. 50 million years since their last common ancestor. Setaria is, like maize, C4 and the two species are more closely related to each other than either is to rice, although they represent two independent acquisitions of C4. This paper shows that SCR orthologs control stomatal patterning in both rice and Setaria implying that the stomatal function of SCR may be ancestral in the grasses and also is not directly connected to photosynthetic pathway.

The availability of allelic combinations of SCR and NKD in maize in particular permits the inference of possible maternal effect on the vein spacing phenotype, although exactly how this happens remains unclear.

The discussion provides a careful and logical assessment of the state of knowledge on SCR and IDD proteins in general, and the new data on SCR and NKD in particular. Many questions remain unresolved, and many additional experiments could be suggested. However, the power of the genetics and the phenotypic analysis together provide a novel direction for research on vein spacing. I will refrain in this review from suggesting what additional information would be nice to have since I think a review should assess the quality of the paper as it stands, not as it could be with months more of work.

My only really substantive suggestion is that the micrographs of the Setaria leaves need to be improved. Specifically, in Figure 6E it is hard to see the details of the fused veins. Either the section is too thick or the camera was not focused properly. Because this image in particular is central to the entire paper I would recommend aiming for the clarity of the images of Zea cross sections, which are fine.

Very minor point:

p. 3 - "double Zmscr1;Zmscr1h mutants" - what does the "h" in Zmscr1h refer to?

Significance

Strengths of the paper are 1) the inclusion of three species to help determine which aspects of the gene function may be ascribed to C4; 2) thoughtful and comprehensive genetic analysis; 3) careful sections of leaves; 4) outlines of the limitations of the approach. Limitations (several of which the authors acknowledge in the Discussion) include a general lack of molecular genetic data (protein interactions, DNA binding sites, RNA-seq, etc.). While this information would be great to have, I think the strength of the genetics is such that the paper will be foundational for future work in any case. The one bit of additional data that would be ideal would be information bearing on the two mechanistic hypotheses laid out on p. 10. The model that SCR and NKD promote cell division and specify mesophyll identity is the opposite of the model that SCR and NKD inhibit vein formation. An experiment that helped point the reader toward one or the other of these models would be very valuable.

The paper fills a critical gap. Little to nothing is known about how the internal anatomy of leaves is patterned and the data presented provide evidence that SCR and NKD are two important players. The paper also provides a conceptual advance in offering a couple of genes and some plausible mechanisms of how they might function.

The audience will be primarily developmental geneticists and physiologists. The paper addresses an important problem that is of broad interest to developmental biologists and is potentially important for global agriculture.

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

All the conclusions are based on solid evidence and convincing, and the methodology are in detail to follow or repeat. The writing of the manuscript is logical and easy to follow.

1. The mutation experiments indicated that nkd enhanced the phenotype of scr, but there is no leaf phenotype variation in nkd muations, this is some way difficult to understand, it would much better if the authors can give much more explanation in the discussion. 2.The word green millet in the first paragraph should be changed to green foxtail. Millet means domesticated small cereal grains, such as foxtail millet, finger millet, proso millet etc.

Significance

The manuscript, which titled Mutations in NAKED-ENDOSPERM IDD genes reveal functional interactions with SCARECROW and a maternal influence on leaf patterning in C4 grasses by Hughes et al., first reported that SCR works regulating both leaf inner pattern and epidermal stomatal patterning in the C4 model plant green foxtail. The functional difference of this gene in Setaria from that in maize and rice indicated that the inner leaf cell patterning regulation of SCR is not a characteristic of C4 Species; this gave us insight understanding of the complex of C4 leaf cell patterning. In addition to this important discover, the authors found that mutations in NKD IDD genes enhance loss of function scr phenotypes in the leaves of C4 maize and Setaria but not in the C3 rice, indicating NKD IDD was involved in the leaf cell patterning in C4 species, but no in C3. They also identified a maternal effect on cell-type patterning in maize leaves that are initiated during embryogenesis.

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer #1

Major comments: The main conclusions of this work are that promoters of the different classes of genes display differing usage of GTFs and cofactors to promote transcription and likely recruit polymerase by different mechanisms. The in vivo experiments using factor depletion offer strong evidence that certain factors including TBP/TRF2 are differentially required for transcription at the housekeeping/developmental gene classes. The in-depth analysis of different promoter types combined with the genetic approaches outlined above provide compelling mechanistic insights into promoter-specific engagement of regulatory factors. In general, the data supports the authors' suggestions.

One important shortcoming of these experiments is in the in-vitro DNA binding analysis of GTFs at differing core promoter contexts. The lack of GTFs binding to the housekeeping promoters may be a reflection of low intrinsic transcription activity. If the housekeeping promoters don't assemble active transcription complexes in this in vitro system but the developmentally-regulated promoters do, then a simple comparison of proteins bound to each promoter type is potentially misleading as to the factors required for transcription. For example, results of the in-vivo analysis suggest that the +1 nucleosome is an important factor in the positioning of the transcription start site at housekeeping promoters, therefore the use of chromatinized templates rather than naked DNA would likely better reflect the intrinsic binding properties of factors at promoters.

We thank the reviewer for highlighting that the in vivo experiments constitute strong evidence for the differential requirements of certain factors at different promoter types and that our work provides compelling mechanistic insights into promoter-specific engagement of regulatory factors. We are also grateful to the reviewer for pointing out that we had not sufficiently clearly explained the aim and rationale of the initial in vitro DNA binding analyses (Figures 1 & 2). These which were not meant to assess different factor requirements but to assess if short core-promoter DNA is sufficient recruit transcription-related proteins, as had been reported for TATA promoters, and whether different core-promoter types differ in this ability. We therefore based the in vitro DNA binding assays on the fact that 121bp-short TATA core-promoter DNA is able to recruit and assemble the PIC even in the absence of activators, i.e. when the core promoters are transcriptionally inactive, and assayed all other core-promoter types under identical conditions. Interestingly, while the TATA core promoters enrich for canonical PIC components as expected, housekeeping promoter DNA does not, suggesting that the core-promoter DNA fragments’ abilities to recruit and assemble the PIC differs.

We agree with the reviewer that one could possibly find conditions in which the different promoter types are active in vitro, e.g. by providing activators or chromatinized templates, and we hope that our explanations above clarify why this has not been the goal of these analyses. As the reviewer pointed out, we assay functional requirements of various TFs and GTFs in vivo in the remainder of the manuscript. We revised the manuscript to improve clarify the aim and scope of these sections (pages 4-9) and are grateful to the author for allowing a discussion of this topic as alternative (see below), many thanks

One way to address this issue is to test transcription activity of the promoters used in the mass spec analysis. After incubation of promoters with extract, add NTPs and quantitate the basal transcription activity of each type of promoter. If they are the ~same - great. If not, at a minimum, the authors need to acknowledge this as a limitation of the study. The suggested transcription experiment is a simple extension of the work already completed.

As outlined above, we deliberately assay all core promoter types under identical conditions, such that differences in protein binding reflect the different DNA fragments distinct functional properties. Please also note that while all core-promoter fragments are transcriptionally inactive, they can be activated by input from a strong enhancer (please see Supplementary Figure 2C; housekeeping and developmental core promoters can be induced to comparable levels, and thus weaker binding of GTFs to housekeeping promoters is not a reflection of weaker inducibility or activity).

We note that all statements and claims are strictly in line of what we tested, namely the core promoter DNA’s ability to recruit transcription-related proteins in vitro. However, we agree with the reviewer that the notion that the core promoters are assayed under identical conditions but are not active is important and discuss it in the main text (pages 8 – 9) and the ‘limitations of this study’ section.

The authors suggest from the depletion experiments of TBP/TRF2 that the factors are functionally redundant since the level of transcription for target genes recovers after prolonged depletion, though there is not specific functional evidence to support this claim. A suggested experiment to test the functional redundancy of TBP/TRF2 at subsets of genes is to assess the levels of proteins and/or protein binding to promoters after factor depletion. For instance, is there a global upregulation/stabilization of TBP after TRF2 depletion? Or is there an increase in TBP binding at promoters? These can be addressed by western blot for overall protein levels and ChIP-seq or similar method to assess binding to promoters, which are fairly straightforward experiments given that the cells lines have already been produced.

We thank the reviewer for suggesting potential compensatory mechanism regarding the redundancy of TBP and TRF2 at a subset of tested promoters. To address the question regarding the stability of TBP or TRF2 in the absence of one or the other, we have performed label-free quantitative mass spectrometry on the TRF2-AID cell line and examined TBP levels (Supplementary Figure 4E). We do not see a stabilization of TBP upon the depletion of TRF2 with auxin. The apparent functional redundancy (e.g. Fig. 4J) thus indeed suggests that there might be increased TBP binding. Unfortunately, we are not able to directly test this experimentally due to a lack of resources. We now add a discussion of the potential compensatory mechanisms to the main text (page 14), many thanks.

A discussion would be appreciated on the generality of the suggested mechanism in metazoans. For example, is DREF conserved only in insects but could other eukaryotes use a similar mechanism at housekeeping genes?

We agree that some of the specific TFs don’t have one-to-one orthologs outside insects, yet that other prominent features of Drosophila housekeeping promoters are shared more widely. We now discuss the parallel between dispersed patterns of initiation at different promoter types across species, including Drosophila housekeeping and vertebrate CpG island promoters. We also provide an outlook towards future functional, biochemical and structural studies that might reveal more diverse transcription initiation mechanisms at the different promoter types in our genomes (pages 23-24).

Minor comments: The manuscript is very difficult to read. One major problem is the large number of figures - many of which are not essential for understanding the results. I strongly suggest that the authors think carefully about which figures to include in the manuscript and keep only the most important.

We agree that the manuscript is complex with six main figures and several different approaches, including biochemistry and mass spectrometry but also genomics and bioinformatics. In addition, the manuscript includes in vitro tests of DNA-protein binding and in vivo assays to probe functional requirement (by depletion) and sufficiency (by recruitment). These different assays assess different properties and complement and validate each other, which is why we feel they are required. We hope that the clarification of the different aspects and their purpose makes the manuscript more easily accessible, many thanks.

Second, the legends on many of the graphs are very tiny and difficult to read.

We have revised the figures to improve font size and readability of the figures, many thanks.

Third, it would greatly help readability if the main figures and legends were imbedded in the manuscript and if the supplemental figures + legends were in a separate document. We have now included the main figures and legends into the manuscript, thanks.

Fig 4E: very difficult to understand what was done.

We now add further explanations to the figure legend to describe the different promoter groups compared in the analysis of ChIP-seq coverage of TBP and TRF2. Fig 4A vs G: why are ~ the same number of genes affected by TRF2 vs TBP + TRF2 depletion? I got the impression from the text that there should be a large difference in the number of affected genes.

We had the same prior expectation, but indeed observed a similar number of downregulated genes upon TRF2 depletion versus TBP and TRF2 double depletion. This may partly be technical, e.g. relating to clonal selection of the different AID-cell lines or thresholding effects, but is likely explained by the relatively few TBP dependent genes (200) that don’t contribute substantially to the larger group of TRF2 dependent genes (3826). The observed number 3935 is 98% of the sum, even ignoring potential overlap. We now clarified this in the text. Fig 5A and similar figures: include the number of affected genes in the figure.

We added the number to the figure, thanks. Fig S2C: hard to understand what was done from the legend.

We have added additional explanations to the figure legend, thanks. Fig S2F and similar figures: hard to distinguish the legend and the green colors used. Proofreading: Add citation for Cut&run in the methods.

We did not analyze CUT&RUN data, however ATAC-seq and ChIP-seq data sets are cited.

In supplemental Fig1a, the percentage of "INR only" is greater than 100%.

We thank the reviewer and fixed the typo.

Supplemental Fig 1a legend-should 170,000 protein coding genes read "17,000"? Santana et al. reference on pg 8 should read 2022.

We thank the reviewer and fixed the typos Readability: The categorizations of genes classes based on core promoter elements is somewhat unclear-from 1a, is it the case that all TATA contain INRs? A different way of representing the data to capture overlaps in motifs other than a pie chart may better convey these motif relationships. Work could be done to increase clarity in general on the promoter motif subtypes used and how mutually exclusive these elements are in the tested subsets.

We thank the reviewer for the suggestion. We have added a heatmap in Supplementary Figure 1A showing the percent match score to motif PWMs across Drosophila promoters. As the reviewer suspects, most developmental core promoters have a high-scoring INR motif and some have an additional TATA box or DPE motif. We have also revised the remainder of the text and rewritten the methods section regarding the motif analysis (pages 36 to 38) to improve clarity. Many thanks. Figure 5: authors state "all protein coding genes" are downregulated with TFIIA depletion, though it appears some transcripts are unchanged or upregulated in 5B/C. Suggest change in language.

We thank the reviewer for this comment. Less than 70 genes are not downregulated upon TFIIA depletion, and manual inspection shows that these genes include intronic non-coding RNAs such as tRNAs that hinder accurate PRO-seq quantification. However, we agree with the reviewer and revised the text to reflect that essentially all promoters are downregulated, affecting all promoter types. A discussion on the developmental context of the S2 cell line seems appropriate. If S2 cells represent a late stage developmental cell line, would the authors expect the relative utilization of cofactors to be the same/different in other cellular contexts?

We thank the reviewer for this comment. We indeed expect the relative utilization of cofactors to be the same I most cellular contexts and now added a discussion with relevant references (page 23), many thanks.

Reviewer #2

1. The DNA affinity purification method is excellent as a discovery method, but it has some potential caveats. One is that it cannot capture remodeling events that could potentially remove otherwise stably bound factors to allow for transient PIC assembly and gene activation. It is possible that some of the insulator factors such as BEAF-32 and Ibf1/2, which selectively bind housekeeping sequences, could prevent or reduce binding by PIC factors. This could occur if BEAF-32 and/or Ibf1/2 inhibit PIC assembly if bound to DNA and if these factors bind housekeeping promoters with high affinity and slow off-rates. That is, in live cells, a competition could exist between binding of these enriched housekeeping factors and PIC assembly. By contrast, this caveat is not relevant at developmental promoters due at least in part to low/sub-nM TBP binding affinity. Ultimately, this is a minor concern but the authors should address in the article to inform readers about potential limitations of the experiments.

We thank the reviewer for highlighting that DNA affinity purification is an excellent discovery method and for pointing out important differences between such in vitro assays and the in vivo situation. We agree and interpret our results from the DNA affinity purification carefully and specifically regarding differences observed for different types of core promoters under identical experimental conditions. We now highlight these differences more clearly throughout the relevant sections on pages 4-8 and expand the discussion of this issue in the ‘limitations of the study’ section. Many thanks.

1. More information about how the PRO-seq spike-ins were implemented is recommended. For example, were they fit to a linear regression of read counts/chromosome between all samples, or did they take all hg19 reads as raw fold-change of all samples compared to a control replicate?

We thank the reviewer for addressing the insufficient information provided about the spike-ins used for PRO-seq. We have added this information to the materials and methods section: We calculated the ratio of spiked-in reads representing the percentage of reads mapping to the human genome over all reads. This ratio was used to determine a scaling factor representing the fold-change of total transcriptional output between the auxin-treated sample and the control samples.

1. Figure S1C should be cited (not S1B) to support the statement "Mutating either the TATA box or DRE motifs reduced TBP or DREF binding to control levels..."

We thank the reviewer for this correction and implemented the correct panel citation.

The authors could note that TATA box mutants still show slight enrichment for TBP compared to negative controls.

We now note this in the figure legend and explain that it is consistent with TBP binding to non-TATA-box developmental core promoters (Figure 2 B & E).

In Figure 2A, it would help to remind readers here that TATA, DPE, INR = developmental and TCT, Ohler1/6, DRE = housekeeping.

We thank the reviewer for this suggestion and implement it

Figure S2A shows only 121bp and 350bp DRE core promoters but the text refers to 450bp and 1000bp sequences as well. Can the authors show representative results from these longer sequences?

We thank the reviewer for pointing out these inconsistencies, which we now fixed by revisions to the text and supplementary figures.

1. In comparing data in Fig 2B and 2E, it seems the statement "the ChIP signals reflected the differential binding preferences observed in vitro for the respective promoter subtypes" should be modified. It is true to an extent but it is more nuanced than indicated by the text.

We have reworded the section and now discuss the observed trends for GTFs and TFs.

In Fig S2I, Ohler1 + Ohler6 and TCT are difficult to distinguish because of color scheme choice.

We agree and now explain in the figure legend that the brighter green corresponds to the Ohler1/6 promoters and the darker green to the TCT promoters, we have additionally edited the legend for better color visibility, many thanks.

In Fig 3F, perhaps add that Gld has TATA and Fit2 has DRE?

We now indicate the presence of TATA-box and DRE motifs in the figure, thanks.

Fig 5D: legend is cut off in the Figure. We thank the reviewer for this comment and now fixed the cropped legend. 11. Fig S2B needs more description and clarification in the main text and the legend. We now deleted Fig.S2B. 12. Page 8, 2nd paragraph "avoiding potential" should be replaced with "minimizing" or similar. We thank the reviewer for this comment and have changed the word choice. 13. Page 16, penultimate paragraph: "Essentially" should be replaced with "Essentiality"

We thank the reviewer for this comment and correct the wording.

Reviewer #3

1. The authors perform a k-means clustering of PWM match scores within 17,000 promoter sequences. They describe in the Methods section that this data revealed 9 groups of promoters. However, although it is likely that several of these promoters contain matches for multiple core promoter motifs, the promoter classes are simply named DRE-promoters, TATA-promoters, TCT-promoters, etc., disregarding any combinatorial association. Furthermore, the clustering data is not visualized to support this naming. The authors should at least provide a heatmap showing the PWM match scores for these clusters and indicate which promoters were used. This is crucial for interpretation of results. We thank the reviewer for pointing out the description of the motif analysis lacked clarity and that the clustering of Drosophila promoters should be visualized. We agree and now provide the k-means clustering heatmap of all 17118 protein coding gene promoters, visualizing the position-weight-matrix (PWM) scores matches for the different promoter motifs in Supplementary Figure 1A. This visualization confirms the reviewer’s suspicion that core-promoter motifs often co-occur in the same core-promoter. For example, TATA promoters typically contain TATA-boxes and INR motifs, etc, which is now clearly seen in the newly provided heatmap. We have also revised the main text, figure legends and have rewritten the method section (pages 36-38) to clarify the analysis of motifs throughout the manuscript. Many thanks.

2. Relatedly, this paper uses a seemingly over-simplified terminology to describe promoters as housekeeping or developmental. While this terminology has been used in several studies from the Stark lab, this is not well supported by data and the usage of this terminology will likely lead to confusion among readers. Here, housekeeping seems to refer solely to the presence of a motif match in the promoter sequence rather than to ubiquitous expression across cell types. Similarly, developmental promoters seem to refer to anything that is not housekeeping. Are S2 cells best reflecting the activity of developmental genes? What about genes that are not expressed as part of a specific developmental trajectory, but still cell-type restricted? Since focus here is on the behavior of promoters with respect to their core promoter elements, why not just refer to them according to their promoter elements? A good example where the developmental versus housekeeping distinction is not useful is the authors' desire to generalize differences observed in Figure 2B, in which it is quite obvious that there is no clear developmental versus housekeeping split. Rather the data demonstrate that TATA-containing and DRE-containing promoters behave differently.

We thank the reviewer for raising a concern about the terminology of functionally distinct promoter types in Drosophila. The use of functionally distinct promoter types enriched in different motifs is built on extensive evidence by our lab and others (e.g. the Ohler or Kadonaga groups) that found extensive agreement between promoter sequence, promoter function, initiation pattern, gene annotation, and ubiquitous vs. cell-type-restricted activities. Ubiquitously active housekeeping promoters tend to contain the TCT, DRE and Ohler 1/6 motifs, while cell-type-restricted developmental promoters tend to contain TATA-box, DPE and INR motifs (Arnold & Zabidi, Nat Biotech 2017, Haberle et al. Nature 2019, Ngoc et al. Genetics 2019, Ohler et al. Genome Biol 2002, Ohtsuki et al. Genes & Dev 1998, Rach et al. Plos Genetics 2011).

We find that the terminology is simple and thus accessible for the non-specialist reader. We agree with the reviewer that clarity is key and revise the introduction of the terminology to clarify that it is based on multiple lines of evidence. We also clarify that Figure 2B – in contrast to the reviewer’s claim – does support a clear developmental versus housekeeping split (please see the dendrogram on top of the heatmap). We now clarified this in the main text and legend to Figure 2B, many thanks.

1. The authors state that the "prevalent model" in the community is that PIC assembly is the same at all promoters. This is not true. For instance, it is well established that certain core promoter elements have a strong positional effect on TSS selection, while dispersed promoters lack strong positional features. What is less known is how the dispersed pattern, e.g. of non-TATA promoters, arises. The authors should more clearly specify the unknowns and the novel findings of their paper.

We agree with the reviewer that certain core promoter elements have strong positioning effects on TSS selection and that these occur in promoters with focused initiation patterns such as TATA promoters and developmental non-TATA promoters (e.g. promoters with INR and/or DPE motifs). We also agree that it is unclear how dispersed patterns at housekeeping promoters arise, especially because the initiation sites don’t co-occur with the TF motifs present in these promoters (e.g. DRE or M1BP motifs; see Figure 6A).

However, the question we address goes beyond TSS selection: we have not seen any study of PIC recruitment and assembly at any promoter with dispersed initiation pattern and the idea of a single uniform Pol II PIC assembly has been the predominant view of transcription initiation during the past two decades (Schier & Taatjes, Genes & Dev 2020). Here, we provide evidence that protein recruitment and GTF usage differs between promoter types, which has mechanistic implications beyond TSS choice alone. In particular, we show that at least two modes of transcription initiation exist that differ between focused developmental and dispersed housekeeping promoters, whereby the developmental promoter DNA directly engages the Pol II PIC via TBP and TFIID, while the housekeeping promoter DNA does not and instead, housekeeping promoters recruit TFs, which recruit COFs and TFIIA. This is exciting and inconsistent with uniform GTF recruitment and assembly, and we hope that this work motivates the study of these different PIC assembly mechanisms at different promoter types.

One of the major claims made by the authors in the paper is that PIC is recruited directly or indirectly depending on the presence of TATA or DRE. However, their approach seems to pick up a lot of indirect bindings, especially for TATA. This raises concerns of potential biases, which if addressed would strengthen the author's claims. The results do not exclude that TFIIA is directly recruited to TATA but might simply reflect stronger binding to other factors compared to DRE. It is also puzzling that DRE is the only one selected for further validation as it appears to have the lowest affinity for PIC binding and the focus on Ohler1/6 motifs in the final model. Disclaimer, this reviewer is not an expert on DNA-affinity purification assays.

We thank the reviewer for pointing out that we had not sufficiently clearly explained the DNA affinity purifications. They were performed under identical conditions for all promoter types, such that the differential binding to TATA vs DRE promoters reflects the respective promoter DNA’s affinity to various transcription-related proteins – they are key results of our work. Please note that, despite the high number of TATA interactions, many of these interactors are expected and reflect the binding of multi-subunit protein complexes such as the Mediator and TFIID (please see Figure 2B) and reflect the fact that we did not purify the PIC nor reconstitute it from purified components but determine nuclear proteins that bind to TATA-box promoter DNA. We now introduce and discuss these aspects more clearly.

It is possible that the fewer interactors found for housekeeping promoters stem from lower affinity of the PIC, the lack of chromatin, or the stable binding of sequence-specific binders such as DREF, BEAF-32 and M1BP in our assay (please see our response to reviewer 2 above). As these result from identical experiments under identical conditions, the fewer interactors for housekeeping promoters are also an important result that likely reflects lower affinity or more transient binding. We now clarify these results and their interpretation in the main text and discuss differences between this assay and transcription in vivo in the “limitations of the study” paragraph.

As the reviewer might appreciate, the follow up experiments, including the creation of AID cell lines, PRO-seq, etc., are a lot of work such that we did them for promoters at the two extreme ends of the spectrum and their respective DNA-binding factors TBP and DREF identified in Figure 1. We think that these representatives sufficiently strongly demonstrate that PIC assembly and factor requirement is distinct for different promoter types, many thanks.

Their final model is supported by results by Baumann et al (2018), which directly shows binding and interactions between M1BP, putzig, gfzf and TRF2. However, these factors bind to Ohler1, while most of the work within this study (Figures 1, 3) focused on DRE. How do DRE-containing promoters fit with the final model? Currently, these promoters are not even represented in the model figure.

We thank the reviewer for pointing out that the final model highlights the Ohler 1 motif but omits the DRE motif. Based on the functional analyses shown in Figure 6 (pages 19-21), we think that the different motifs function equivalently in recruiting housekeeping cofactors and activating housekeeping transcription and have now included DRE motifs in the final cartoon. Our original choice was indeed based on the fact that previous reports from Baumann et al 2018 corroborate our findings for M1BP. As DRE promoters also recruit and depend on TRF2 (Hochheimer et al. Nature 2002), we now show a model by which housekeeping DRE promoters recruit a TRF2 containing PIC through TFIIA, but would like to stress that both likely function equivalently, leading to dispersed initiation. We also revised the data presentation and the final discussion regarding these promoters, many thanks.

Minor comments

1. The TSS patterns of promoters were evaluated using STAP-seq (in vitro data) and developmental CAGE data. For the purpose of the paper and to match the in DNA-affinity purification data better, it would be more reasonable to make use of S2 cell CAGE data (e.g. Rennie et al, 2018 PMID: 29659982).

We thank the reviewer for bringing up this point. For figure 6 we have used CAGE data from Drosophila embryos instead of S2 cells in order to capture a larger proportion of expressed developmental genes and their promoters, rather than just the ones that are expressed in S2 cells. As promoter motifs are found in stereotypical positions in relation to the TSS (Ohler et al. Genome Biol 2002) and because non-S2-cell core promoters can be activated in STAP-seq (Arnold 2017; Haberle 2019), our use of CAGE data from Drosophila embryos allows us to base all subsequent analyses on many more core promoters and also exclude any cell-type specific effects that may arise in TSS selection.

Previous models on TSS selection within non-TATA promoters have highlighted the dinucleotide frequency of +1 nucleosomal DNA as a strong positional feature. Here, the authors investigate this model using a rather weak analytical approach. We know that nucleosomes can vary between cells (fuzzy positioning). Variability across promoters may cause larger variability in relative TSS positioning. Hence, what is observed here as a TSS spread relative to the +1 nucleosome positioning might in fact be caused by averaging. A more suitable approach would be to analyze the positional cross-correlation between TSS locations (e.g. revealed by CAGE reads) and nucleosomal positions (e.g. revealed by MNase-seq reads). This would better support claims regarding specific TSS positioning with respect to nucleosome positioning.

We agree that the analysis of cross correlation between TSS locations and nucleosomal positions at individual promoters would provide a more precise measure of TSS positioning relative to the nucleosome. We had originally chosen a visualization that more directly assesses whether the +1 nucleosome determines the TSSs by centering on the predicted +1 positions. In response to this comment, we have performed two additional analyses: a cross-correlation analysis on CAGE and Mnase-seq read coverage in relation to the dominant CAGE TSS (new Supplementary Figure 6I) and a TSS-centric analysis of Mnase-seq coverage (new Supplementary Figure 7. Both analyses agree with the original analysis and we thank the reviewer for pointing out how to strengthen this analysis.

The cross-correlation analysis reveals a peak in the mean correlation score 125 base pairs downstream of housekeeping TSS (at TCT, Ohler1 and DRE) promoters but not downstream of developmental promoters (TATA-box, DPE and INR), in line with housekeeping TSS being positioned upstream of the +1 nucleosome.

The analysis assessing +1 nucleosome positions as derived from MNase-seq coverage relative to the position of the dominant TSS reveals the expected phasing of downstream nucleosomes in housekeeping promoters but not at developmental promoters. Many thanks.

It is interesting that tethering of housekeeping-associated coactivators leads to a higher positional dispersion compared to the result of developmental-associated coactivators. However, the positional TSS dispersion of housekeeping promoters seems to always be larger than that of developmental promoters regardless of coactivator recruitment. Can the authors explain these results?

We agree that CAGE data typically show TSS dispersion at housekeeping promoters, yet this reflects the promoters’ transcriptionally active states during which endogenous TFs and coactivators are present. Our analyses are based on short, transcriptionally inactive core promoters that can be activated by cofactor recruitment, leading to the observed outcomes. We now clarify this in the manuscript and highlight that the differences in focused versus dispersed patterns occur even on the very same DNA sequences upon the recruitment of developmental or housekeeping activators (e.g. Fig. 6F).

The authors seem to suggest that positional dispersion of TSSs within housekeeping promoters is due to stochastic initiation after non-positional specific PIC recruitment mediated via certain co-activators. If TSS selection is truly stochastic, why do these promoters then have dominant TSSs?

We thank the reviewer for pointing out that our phrasing might have suggested that TSS selection was entirely random or stochastic, which is neither true for STAP-seq nor for endogenous CAGE data. In fact, not all positions have the same probability to initiate transcription, but certain positions or nucleotides seem to be inherently favored. We speculate that favorable positions relate to the local DNA structure, the energy barrier landscape for both DNA helix melting to occur and for the first phospho-diester bond to form (e.g. Dineen, D. et al. NAR 2009 and Vanaja, A. et al. ACS Publications 2022). We now added this discussion and the corresponding references to our manuscript (page 21).

The authors find Chromator as a likely cofactor for indirect recruitment of TFIIA to housekeeping promoters. BEAF-32 is another factor the authors highlight as being enriched at housekeeping promoters (DRE promoters). Both of these factors have previously been considered insulator proteins or architectural proteins involved in the formation of chromatin folding (Ramirez et al, 2018, PMID: 29335486; Wang et al, 2018. PMID: 29335463). Could the authors comment on this link with their own findings?

We thank the reviewer for addressing the importance of chromatin topology in the light of our findings, which we now discuss in the main text (pages 22-23).

1. Can the authors justify PWM match thresholds used and why these were changed from Haberle et al 2019?

We thank the reviewer for pointing out that these changes had not been justified. We adjusted them to be more stringent (e.g. DPE) or sensitive (e.g. TATA-box) exclusively for the motif enrichment analysis, which we did outside the rule-based promoter-annotation effort. These adjusted thresholds reflect the motifs vastly different information contents, which is low for DPE and high for TATA-box motifs.

Figure related comments/concerns: • General: Sometimes wrong ordering of figure panels with regards to their first mention in the main text, varying font sizes, and minimal figure legends that are often inconsistent (e.g. PRO-seq is sometimes specified when used, but not always) • Typo: Supp Fig 1: INR only 121.37% • Fig 1E not explained, what does x axis describe and how is it calculated? • Figure 2C-D: The CAGE signal is poorly visualized in panel C, it also poorly describes that this is supposedly done using a pool of promoters. Where is the 450bp blot (it seems plausible that the 450bp fragment could actually facilitate a luciferase signal in Fig S2-B)? How was this pool selected, is it exclusively based on DRE-containing promoters? • Fig 2D: apparent gel leakage and loading on the second panel is low. Preferably, provide positive control on the same gel. • Figure 4C: all classes are negatively affected by TRF2 depletion, thus enrichment (4B) makes little sense here • Figure 5C: Missing axis labels • Figure 6F: A y scale would help here

We thank the reviewer for these recommendations and have implemented all of them.

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 manuscript by Serebreni and colleagues examines how core promoter elements influence the binding of general transcription factors and co-activators and the establishment of pre initiation complexes, and how the recruitment of these factors relate to the transcription initiation patterns (focused versus dispersed) within Drosophila promoters. While there is extensive literature on core promoter elements and their association with general transcription factors and promoter classes, a mechanistic link between promoter sequence and dispersed initiation patterns has been lacking. Therefore, the present study is important. Using an impressive range of well planned experiments, combining in vitro (DNA-affinity purification, STAP-seq) and in vivo (CAGE, PRO-seq, ChIP-seq) data, the authors conclude that developmental promoters directly recruit the PIC via positional core promoter elements leading to a focused transcription initiation pattern while housekeeping promoters facilitate PIC recruitment through intermediate binding of additional cofactors leading to a more dispersed promoter initiation pattern. This conclusion is strengthened by experimental data demonstrating increased TSS dispersion upon forced recruitment of cofactors naturally associated with promoters exhibiting dispersed initiation.

Major comments

1. The authors perform a k-means clustering of PWM match scores within 17,000 promoter sequences. They describe in the Methods section that this data revealed 9 groups of promoters. However, although it is likely that several of these promoters contain matches for multiple core promoter motifs, the promoter classes are simply named DRE-promoters, TATA-promoters, TCT-promoters, etc., disregarding any combinatorial association. Furthermore, the clustering data is not visualized to support this naming. The authors should at least provide a heatmap showing the PWM match scores for these clusters and indicate which promoters were used. This is crucial for interpretation of results.
2. Relatedly, this paper uses a seemingly over-simplified terminology to describe promoters as housekeeping or developmental. While this terminology has been used in several studies from the Stark lab, this is not well supported by data and the usage of this terminology will likely lead to confusion among readers. Here, housekeeping seems to refer solely to the presence of a motif match in the promoter sequence rather than to ubiquitous expression across cell types. Similarly, developmental promoters seem to refer to anything that is not housekeeping. Are S2 cells best reflecting the activity of developmental genes? What about genes that are not expressed as part of a specific developmental trajectory, but still cell-type restricted? Since focus here is on the behavior of promoters with respect to their core promoter elements, why not just refer to them according to their promoter elements? A good example where the developmental versus housekeeping distinction is not useful is the authors' desire to generalize differences observed in Figure 2B, in which it is quite obvious that there is no clear developmental versus housekeeping split. Rather the data demonstrate that TATA-containing and DRE-containing promoters behave differently.
3. The authors state that the "prevalent model" in the community is that PIC assembly is the same at all promoters. This is not true. For instance, it is well established that certain core promoter elements have a strong positional effect on TSS selection, while dispersed promoters lack strong positional features. What is less known is how the dispersed pattern, e.g. of non-TATA promoters, arises. The authors should more clearly specify the unknowns and the novel findings of their paper.
4. One of the major claims made by the authors in the paper is that PIC is recruited directly or indirectly depending on the presence of TATA or DRE. However, their approach seems to pick up a lot of indirect bindings, especially for TATA. This raises concerns of potential biases, which if addressed would strengthen the author's claims. The results do not exclude that TFIIA is directly recruited to TATA but might simply reflect stronger binding to other factors compared to DRE. It is also puzzling that DRE is the only one selected for further validation as it appears to have the lowest affinity for PIC binding and the focus on Ohler1/6 motifs in the final model. Disclaimer, this reviewer is not an expert on DNA-affinity purification assays.
5. Their final model is supported by results by Baumann et al (2018), which directly shows binding and interactions between M1BP, putzig, gfzf and TRF2. However, these factors bind to Ohler1, while most of the work within this study (Figures 1, 3) focused on DRE. How do DRE-containing promoters fit with the final model? Currently, these promoters are not even represented in the model figure.

Minor comments

1. The TSS patterns of promoters were evaluated using STAP-seq (in vitro data) and developmental CAGE data. For the purpose of the paper and to match the in DNA-affinity purification data better, it would be more reasonable to make use of S2 cell CAGE data (e.g. Rennie et al, 2018 PMID: 29659982).
2. Previous models on TSS selection within non-TATA promoters have highlighted the dinucleotide frequency of +1 nucleosomal DNA as a strong positional feature. Here, the authors investigate this model using a rather weak analytical approach. We know that nucleosomes can vary between cells (fuzzy positioning). Variability across promoters may cause larger variability in relative TSS positioning. Hence, what is observed here as a TSS spread relative to the +1 nucleosome positioning might in fact be caused by averaging. A more suitable approach would be to analyze the positional cross-correlation between TSS locations (e.g. revealed by CAGE reads) and nucleosomal positions (e.g. revealed by MNase-seq reads). This would better support claims regarding specific TSS positioning with respect to nucleosome positioning.
3. It is interesting that tethering of housekeeping-associated coactivators leads to a higher positional dispersion compared to the result of developmental-associated coactivators. However, the positional TSS dispersion of housekeeping promoters seems to always be larger than that of developmental promoters regardless of coactivator recruitment. Can the authors explain these results?
4. The authors seem to suggest that positional dispersion of TSSs within housekeeping promoters is due to stochastic initiation after non-positional specific PIC recruitment mediated via certain co-activators. If TSS selection is truly stochastic, why do these promoters then have dominant TSSs?
5. The authors find Chromator as a likely cofactor for indirect recruitment of TFIIA to housekeeping promoters. BEAF-32 is another factor the authors highlight as being enriched at housekeeping promoters (DRE promoters). Both of these factors have previously been considered insulator proteins or architectural proteins involved in the formation of chromatin folding (Ramirez et al, 2018, PMID: 29335486; Wang et al, 2018. PMID: 29335463). Could the authors comment on this link with their own findings?
6. Caan the authors justify PWM match thresholds used and why these were changed from Haberle et al 2019?
7. Figure related comments/concerns:
   • General: Sometimes wrong ordering of figure panels with regards to their first mention in the main text, varying font sizes, and minimal figure legends that are often inconsistent (e.g. PRO-seq is sometimes specified when used, but not always)
   • Typo: Supp Fig 1: INR only 121.37%
   • Fig 1E not explained, what does x axis describe and how is it calculated?
   • Figure 2C-D: The CAGE signal is poorly visualized in panel C, it also poorly describes that this is supposedly done using a pool of promoters. Where is the 450bp blot (it seems plausible that the 450bp fragment could actually facilitate a luciferase signal in Fig S2-B)? How was this pool selected, is it exclusively based on DRE-containing promoters?
   • Fig 2D: apparent gel leakage and loading on the second panel is low. Preferably, provide positive control on the same gel.
   • Figure 4C: all classes are negatively affected by TRF2 depletion, thus enrichment (4B) makes little sense here
   • Figure 5C: Missing axis labels
   • Figure 6F: A y scale would help here

Significance

The manuscript by Serebreni and colleagues examines how core promoter elements influence the binding of general transcription factors and co-activators and the establishment of pre initiation complexes, and how these factors relate to the transcription initiation patterns (focused versus dispersed) of promoters in Drosophila. While there is extensive knowledge on core promoter elements and how these relate to TSS positional dispersion within promoters, little is known about the mechanism of PIC assembly at non-TATA promoters and how this influences TSS selection. The findings will therefore be interesting for a general audience, although it is unclear how transferable results are to other organisms.

The authors use an impressive range of well planned experiments, combining in vitro (DNA-affinity purification, STAP-seq) and in vivo (CAGE, PRO-seq, ChIP-seq) data. Their main conclusion is that developmental promoters directly recruit the PIC via positional core promoter elements leading to a focused transcription initiation pattern while housekeeping promoters facilitate PIC recruitment through intermediate binding of additional cofactors leading to a more dispersed promoter initiation pattern.

While this major conclusion is of interest to the community, the manuscript unfortunately falls short in some regards, in particular in its over-generalizations and simplifications. Throughout the manuscript, the analysis is focused around specific core promoter motifs while ignoring the fact that many of these tend to co-occur within a promoter. In addition, the authors make general statements about housekeeping versus developmental promoters - a terminology based solely on the presence of core promoter elements - rather than attributing their findings to the core-promoter elements themselves. Lastly, the main figures are unpolished with minimal information provided in figure legends, making it sometimes difficult to follow the author's reasoning and raising concerns about the strength of their findings.

Fields of expertise: mammalian regulatory elements, transcription initiation, genomics

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 article from Serebreni, Stark and co-workers combines biochemical, analytical, computational and cellular methods to uncover different factor dependencies for different classes of promoters in Drosophila. The results are compelling and the data support the conclusions. Important new insights are that housekeeping and developmental promoters have different requirements for initiation factors and that TFIIA is generally required across the different promoter types. Also, the article provides evidence of potential new mechanisms that control focused vs. dispersed initiation. These are groundbreaking results and I have only a few minor comments on the article.

1. The DNA affinity purification method is excellent as a discovery method, but it has some potential caveats. One is that it cannot capture remodeling events that could potentially remove otherwise stably bound factors to allow for transient PIC assembly and gene activation. It is possible that some of the insulator factors such as BEAF-32 and Ibf1/2, which selectively bind housekeeping sequences, could prevent or reduce binding by PIC factors. This could occur if BEAF-32 and/or Ibf1/2 inhibit PIC assembly if bound to DNA and if these factors bind housekeeping promoters with high affinity and slow off-rates. That is, in live cells, a competition could exist between binding of these enriched housekeeping factors and PIC assembly. By contrast, this caveat is not relevant at developmental promoters due at least in part to low/sub-nM TBP binding affinity. Ultimately, this is a minor concern but the authors should address in the article to inform readers about potential limitations of the experiments.
2. More information about how the PRO-seq spike-ins were implemented is recommended. For example, were they fit to a linear regression of read counts/chromosome between all samples, or did they take all hg19 reads as raw fold-change of all samples compared to a control replicate?
3. Figure S1C should be cited (not S1B) to support the statement "Mutating either the TATA box or DRE motifs reduced TBP or DREF binding to control levels..."
4. The authors could note that TATA box mutants still show slight enrichment for TBP compared to negative controls.
5. In Figure 2A, it would help to remind readers here that TATA, DPE, INR = developmental and TCT, Ohler1/6, DRE = housekeeping.
6. Figure S2A shows only 121bp and 350bp DRE core promoters but the text refers to 450bp and 1000bp sequences as well. Can the authors show representative results from these longer sequences?
7. In comparing data in Fig 2B and 2E, it seems the statement "the ChIP signals reflected the differential binding preferences observed in vitro for the respective promoter subtypes" should be modified. It is true to an extent but it is more nuanced than indicated by the text.
8. In Fig S2I, Ohler1 + Ohler6 and TCT are difficult to distinguish because of color scheme choice.
9. In Fig 3F, perhaps add that Gld has TATA and Fit2 has DRE?
10. Fig 5D: legend is cut off in the Figure.
11. Fig S2B needs more description and clarification in the main text and the legend.
12. Page 8, 2nd paragraph "avoiding potential" should be replaced with "minimizing" or similar.
13. Page 16, penultimate paragraph: "Essentially" should be replaced with "Essentiality"

Significance

As noted in the prior section, the results break new ground and will be of interest to many in the field of gene regulation, broadly defined.

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:

Serebreni et al. Dissect the mechanisms of distinct transcriptional regulation patterns for the housekeeping and developmental classes of genes in Drosophila S2 cells. The authors used two primary lines of experimentation to determine the factors involved in regulation at the core promoters of the different gene classes: in vitro DNA binding with mass spectrometry, and in vivo depletion of factors with transcriptomics. The authors find that general transcription factors bind more strongly to developmental (TATA-containing) promoters and speculate that GTFs interact more transiently with housekeeping promoters. In addition, the authors find distinct preferences for TBP/TRF2 at different types of core promoters and test the roles of cofactors and promoter architecture on differing patterns of transcriptional initiation.

Major comments:

The main conclusions of this work are that promoters of the different classes of genes display differing usage of GTFs and cofactors to promote transcription and likely recruit polymerase by different mechanisms. The in vivo experiments using factor depletion offer strong evidence that certain factors including TBP/TRF2 are differentially required for transcription at the housekeeping/developmental gene classes. The in-depth analysis of different promoter types combined with the genetic approaches outlined above provide compelling mechanistic insights into promoter-specific engagement of regulatory factors. In general, the data supports the authors' suggestions. One important shortcoming of these experiments is in the in-vitro DNA binding analysis of GTFs at differing core promoter contexts. The lack of GTFs binding to the housekeeping promoters may be a reflection of low intrinsic transcription activity. If the housekeeping promoters don't assemble active transcription complexes in this in vitro system but the Developmentally-regulated promoters do, then a simple comparison of proteins bound to each promoter type is potentially misleading as to the factors required for transcription. For example, results of the in-vivo analysis suggest that the +1 nucleosome is an important factor in the positioning of the transcription start site at housekeeping promoters, therefore the use of chromatinized templates rather than naked DNA would likely better reflect the intrinsic binding properties of factors at promoters. One way to address this issue is to test transcription activity of the promoters used in the mass spec analysis. After incubation of promoters with extract, add NTPs and quantitate the basal transcription activity of each type of promoter. If they are the ~same - great. If not, at a minimum, the authors need to acknowledge this as a limitation of the study. The suggested transcription experiment is a simple extension of the work already completed. The authors suggest from the depletion experiments of TBP/TRF2 that the factors are functionally redundant since the level of transcription for target genes recovers after prolonged depletion, though there is not specific functional evidence to support this claim. A suggested experiment to test the functional redundancy of TBP/TRF2 at subsets of genes is to assess the levels of proteins and/or protein binding to promoters after factor depletion. For instance, is there a global upregulation/stabilization of TBP after TRF2 depletion? Or is there an increase in TBP binding at promoters? These can be addressed by western blot for overall protein levels and ChIP-seq or similar method to assess binding to promoters, which are fairly straightforward experiments given that the cells lines have already been produced. A discussion would be appreciated on the generality of the suggested mechanism in metazoans. For example, is DREF conserved only in insects but could other eukaryotes use a similar mechanism at housekeeping genes?

Minor comments:

The manuscript is very difficult to read. One major problem is the large number of figures - many of which are not essential for understanding the results. I strongly suggest that the authors think carefully about which figures to include in the manuscript and keep only the most important. Second, the legends on many of the graphs are very tiny and difficult to read. Third, it would greatly help readability if the main figures and legends were imbedded in the manuscript and if the supplemental figures + legends were in a separate document.

Fig 4E: very difficult to understand what was done.

Fig 4A vs G: why are ~ the same number of genes affected by TRF2 vs TBP + TRF2 depletion? I got the impression from the text that there should be a large difference in the number of affected genes.

Fig 5A and similar figures: include the number of affected genes in the figure.

Fig S2C: hard to understand what was done from the legend.

Fig S2F and similar figures: hard to distinguish the legend and the green colors used. Proofreading: Add citation for Cut&run in the methods. In supplemental Fig1a, the percentage of "INR only" is greater than 100%. Supplemental Fig 1a legend-should 170,000 protein coding genes read "17,000"? Santana et al. reference on pg 8 should read 2022.

Readability: The categorizations of genes classes based on core promoter elements is somewhat unclear-from 1a, is it the case that all TATA contain INRs? A different way of representing the data to capture overlaps in motifs other than a pie chart may better convey these motif relationships. Work could be done to increase clarity in general on the promoter motif subtypes used and how mutually exclusive these elements are in the tested subsets.

Figure 5: authors state "all protein coding genes" are downregulated with TFIIA depletion, though it appears some transcripts are unchanged or upregulated in 5B/C. Suggest change in language.

A discussion on the developmental context of the S2 cell line seems appropriate. If S2 cells represent a late stage developmental cell line, would the authors expect the relative utilization of cofactors to be the same/different in other cellular contexts?

Significance

This work is conceptually significant due to the large in gene-specific regulatory mechanisms in the field of molecular biology. In addition, the authors propose a new mechanism whereby PIC formation is substantially different at different gene classes. Much of our mechanistic understanding of the role of general transcription factors is limited to highly expressed, typically TATA-containing genes, though several lines of research have shown that not all genes are dependent on the same subsets of factors. Notably, TBP has been shown to be required for the transcription of only small subsets of genes in specific cell types, therefore investigations into the roles of general factors at diverse genes is an important step in the field. This work is also technologically significant due to its use of the auxin-inducible degron system to assess the immediate transcriptional effects of factor depletion. Prior work demonstrated that long-term loss of factors through genetic deletions can often lead to compensatory mechanisms including utilization of alternative regulatory pathways and stabilization of cellular RNAs, therefore assessment of the immediate effects of rapid factor depletion is a powerful approach to determine regulatory mechanisms. This research will be of broad interest to molecular biologists studying the basic mechanisms of transcription as well as gene-specific regulation.

Reviewer expertise:

Transcriptional regulation, biochemistry, genomics, molecular biology

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

Learn more at Review Commons

---

Reply to the reviewers

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

Summary:

In this study, mice were exposed to a specific form of so-called Intermittent Fasting (IF) and the effects of IF on adult neogenesis in the hippocampus were determined. The specific IF protocol used had no effect on activation, proliferation, or maintenance of adult Neural Stem Cells (aNSCs) and displayed a decrease in number of new neurons in the neurogenic niche but only after 1 month of the IF protocol. These results contrast previously published results from multiple studies that concluded that IF promotes survival of new neurons and by extension promote adult neurogenesis. The unresponsiveness of aNSCs or their immediate cell progeny, the Intermediate Neural Progenitors (IPCs), to IF is a novel finding. The authors make several relevant points in the discussion about the publication bias towards positive results (or omission of negative results), which may reinforce established dogmas. However, the presented results did not convincingly demonstrate that the absence of effects of IF on aNSCs or adult neurogenesis is simply not a result of a specific IF paradigm, which is not robust enough to elicit changes in adult neurogenesis. In other words, there is a lack of positive controls and alternative protocols that would rule out that the observed absence of effects is not a consequence of type II error (the error of omission), or more colloquially, a consequence of false negatives.

We thank the reviewer for acknowledging the importance and novelty of our findings. On them being the result of a specific IF paradigm, we must point out that we used the same IF paradigm as in previous studies that had shown changes in neurogenesis upon IF. We do not claim that IF is unable to increase neurogenesis in all conditions, but report that IF is not a reliable method to increase adult neurogenesis (in particular, every-other-day intermittent fasting with food re-administration in the evening). We have repeated the experiment multiple times in different strains, always with enough animals to make our experiments conclusive and we never observed an increase in adult neurogenesis, effectively ruling out that our results are a false negative. Of note, even if other protocols might indeed increase neurogenesis (which we never claimed cannot) that would not make our results a false negative.

Major Comments:

1. Protocol-driven absence of effects: The absence of IF effects on aNSCs and IPCs observed in this study does not lend it the authority to conclude that aNSCs are resilient to IF or all IF paradigms and protocols. The absence of IF effects on aNSCs and neurogenesis could be specifically related to the chosen IF paradigm. Indeed, not all previous studies that observed IF-driven effects on adult neurogenesis used the same "night-time every-other-day fasting" protocol chosen in this study. For example, Brandhorst et al., 2015 (cited in this paper) used 4 days of IF 2x per month and observed an increase of DCX+BrdU+ cells. On the other hand, certain previous studies used the same or similar IF protocol used here, but often with longer duration or with a post-fasting ad libitum feeding period, which may be responsible for the pro-neurogenic or pro-survival effects. In fact, the authors acknowledge this in the discussion (page 7, lines 289-290 and 292-294). Why would the authors then not include similar feeding/IF paradigm in their study and determine if these would generate effects on survival of new neurons but also on aNSCs and/or IPCs?

As just stated above, we never claimed that aNSCs are resilient to all IF paradigms. We refer to fasting in general in the introduction but quickly focus on every-other-day fasting throughout the paper and directly compare our results only to similar IF paradigms. We chose the most commonly used IF paradigm that had been shown to increase adult neurogenesis. As the reviewer points out, we speculate in the discussion that a refeeding period may explain the differences between our results and others. This is because a post-fasting ad libitum period was introduced in the study published in Dias et al. 2021. We are currently analysing a new experiment in which we replicate the IF protocol in that study, which we will include in our revised version.

In addition, the authors acknowledge that the chosen IF paradigm may have affected the stress levels or behaviour of mice (page 9, lines 372-378). Why did they not test if their IF protocol does not increase stress or anxiety of mice by simple behaviour tests such as open field or elevated T maze?

While testing all possible causes for the lack of positive results in our experiments is not viable, we do agree with the reviewer that stress levels might indeed influence the outcome of the experiments. We will collect blood from ad libitum-fed and fasted mice to analyse the levels of stress hormones (e.g. corticosterone). The results will be included in our revised version. These measurements will give us a more accurate reading of stress levels than behavioural tests. Of note, regardless of the outcome of this experiment, our conclusions will remain identical. We will not be able to compare stress levels with previous publications, as they were not tested. And if the protocol did increase stress levels, it would still argue that IF is not a reliable method to increase neurogenesis (as presumably might or might not increase stress to levels that affect neurogenesis).

Alarmingly, the used IF protocol does not result in changes in final weight or growth curves (S.Fig.2), which is surprising and raises a question the used IF protocol is robust enough or appropriate.

We were also surprised by the lack of change in the final weight our IF mice respect to control. Differences in final weight between different labs despite using the exact same protocol are one of the reasons why we conclude that this IF paradigm is not a robust intervention. However, we are not the first ones to report little or no difference in weight upon IF in C57BL6/J mice (Goodrick et al., 1990 and Anson et al., 2003) and this would not be a reason to dismiss the experiment since the benefits in crucial circulating factors induced by IF seem to be independent of weight loss (Anson et al., 2003).

Finally, the authors acknowledge that their own results do not support well-established findings such as aging-related reduction in number of aNSCs (page 4, lines 177-179). This again questions whether the selected protocols and treatments are appropriate.

As we already discuss, we believe this might be due to a difference between strains in the time when aNSC numbers decline. Nevertheless, we will complement our current data by counting the number of aNSCs at 1 and 3 months post-tamoxifen (3 and 5 month old mice) using GFAP, Sox2 and Nestin triple stainings (as suggested by another reviewer).

Lack of topic-specific positive controls: The authors successfully demonstrated that the used IF protocol differentially impacts the adipose tissue and liver, while also inducing body weight fluctuations synchronized with the fasting periods. However, these peripheral effects outside the CNS do not directly imply that the chosen IF protocol is robust enough to elicit cellular or molecular changes in the hippocampus. The authors need to demonstrate that their IF protocol affects previously well-established CNS parameters associated with fasting such as astrocyte reactivity, inflammation or microglia activation, among other factors. In fact, they acknowledge this systemic problem in the discussion (page 8, lines 359-360).

We fully agree with the reviewer in that even though the chosen IF protocol induces peripheral effects, it is not robust enough to elicit cellular or molecular changes in the hippocampus, and this is precisely the message of our paper. We have looked for references showing the influence of IF on astrocyte reactivity or microglia activation, but the studies we found so far look at the effects of IF and other forms of fasting in the CNS in combination with pathologies such as Alzheimer’s disease, Multiple Sclerosis, physical insults or aging (Anson et al., 2003; Chignarella et al., 2018; Rangan et al., 2022; Dai et al., 2022. Reviewed in Bok et al., 2019 and Gudden et al., 2021). Fasting seems to reduce astrocyte reactivity, inflammation or microglia activation in these pathological situations respect to the same pathology in ad libitum mice, but its effect in control, healthy mice is far less clear. In fact, the only reference that we could find where healthy mice were included in the analysis showed that these benefits only happened in the context of the injury (Song et al., 2022).

Problematic cell analyses: Cell quantification should be performed under stereological principles. However, the presented results did not adhere to stereological quantification. Instead, the authors chose to quantify specific cell phenotypes only in subjectively selected subsets of regions of interest, i.e., the Subgranular Zone (SGZ). This subjective pre-selection may have been responsible for the absence of effects, especially if these are either relatively small or dependent on anatomical sections of SGZ. For example, IF may exert effects on caudal SGZ more than on rostral SGZ. But if the authors quantified only (or predominantly) rostral SGZ, they may have missed these effects by biasing one segment of SGZ versus other. The authors should apply stereological quantification at least to the quantification of new neurons and test if this approach replicated previously observed pro-survival effects of IF. Also, the authors should describe how they pre-selected the ROI for cell quantification in greater details.

We did analyse only the more septal region of the hippocampus, which we will make clear in the text. As also suggested by other reviewers, we will include stereological counts of the neuronal output of aNSCs in the revised version. As for selecting the SGZ for aNSC counts, this is the standard in the field, as one of the criteria to identify aNSCs is precisely the location of their nucleus in the SGZ. Neuroblasts and new neurons were counted both in the SGZ and the granule cell layer. There was no subjective pre-selection of areas of interest since we counted the whole DG in each section and not a specific random region.

Alarming exclusion of data points: There appears to be different number of data points in different graphs that are constructed from same data sets. For example, in the 3-month IF data set in Figure 4, there are 14 data points for the graph of Ki67+ cells (Fig.4B), but 16 (or 17) data points for the graph of DCX+ cells (Fig.4D). How is that possible? If data points were excluded, what objective and statistical criteria were applied to make sure that such exclusion is not subjective and biased? In fact, the authors state that "Samples with poor staining quality were also excluded from quantifications" (page 12, line 528-529). Poor preparation of tissue is not only suboptimal but not a valid objective reason for data point exclusion. This major issue needs to be explained and corrected.

As we disclose in the methods, those stainings that did not work were excluded. This was done always before counting. Different samples were used in different counts because of the variability of staining quality between different antibodies. We will look back into the samples that failed in at least one of the stainings and exclude them from all counts, so that only samples for which all stainings worked are considered. These revised graphs will be provided in our revised version of the manuscript.

Different pulse-and-chase time-points: One of the reasons why this study has found that aNSCs may not be responsive to IF could be the use of less appropriate pulse-and-chase time-points either after EdU or after Tamoxifen for cell lineage tracing. The authors observed that IF has negative effects on new neurons initially (Fig.4F). Similarly, it is well established that voluntary physical exercise affects SGZ adult neurogenesis only during the first 2 weeks. After this period, the neurogenic effects of exercise are diminished beyond observational detection (i.e., van Praag's and Kempermann's papers in the past 25 years). These two arguments suggest that the observed absence of aNSC responsiveness might be a consequence of the chosen EdU administration and the EdU pulse should not be administered 15 days after Tamoxifen/IF protocol start but earlier, in the first week of the IF protocol. In fact, the decreased number of new neurons during the initial IF phase may not be only a consequence of reduced survival but of higher aNSC quiescence during the first week of the IF protocol.

We fully agree with the reviewer that BrdU or EdU pulses can give a biased view of the effects of any intervention on neurogenesis and that the EdU and Tamoxifen protocols would not allow us to detect an increase in neurogenesis during the first few days of IF. We cannot rule out that IF has a transient effect on aNSCs at some point of the treatment, but this hypothetical effect does not seem to have any consequences on neuronal output or aNSC maintenance. As for the effects on neurogenesis in the longer IF treatments, we used the same EdU protocol as in previous publications: administration after 2/3 months of IF and analysis after one month of chase.

Discussion needs more specificity and clarity: The authors claim that the absence of IF effects on neurogenesis is multi-layered (including the influence of age, sex, specific cell labelling protocols etc.) but they do not specifically address why certain studies did find IF-driven neurogenic effects while they did not. In addition, some statements and points in the discussion are not clear. For example, when the authors refer to their own experiments (page 8, lines 331-334), it is not clear, which experiments they have in mind.

We will double check our discussion and improve its clarity and direct comparison to other studies.

Minor comments:

1. Change in the title: The authors have shown that a very specific IF protocol does not affect aNSCs but initially decreases number of new neurons in SGZ. The title should reflect this. For example, it could state "Specific (night-time every-other-day) fasting does not affect aNSCs but initially decreases survival of new neurons in the SGZ".

We find our title, together with the abstract, clearly and faithfully represent our findings and would rather prefer to keep our current title unmodified.

Data depiction: Data in 3 datasets were found not normally distributed (Fig. S5A, B and S6A) and were correctly analysed with non-parametric tests. However, the corresponding graphs wrongly depict the data as mean +/- SD while they should depict median +/- IQR (or similar adequate value) because non-parametric statistical tests do not compare means but medians.

We thank the reviewer for spotting this, we will correct the graphs in Fig. S5A, B and S6A.

Statistical analysis: For ANOVA, the F and p values are not listed anywhere. The presented asterisks in the graphs are only for non-ANOVA or ANOVA post-hoc tests. This does not allow to judge statistical significance well and should be corrected.

Again, thanks for spotting this, we will include them.

Asymmetric vs Symmetric cell divisions: Representative images in Fig.2B suggest that IF may affect the plane of cell division for the Type-1 aNSCs. The plane of cell division is an indirect indicator of symmetric vs asymmetric (exhaustive vs maintaining) modes of cell division. Is it possible, IF influences this, especially during the first week of IF (see major comment 5)?

This is an interesting hypothesis. However, since we do not see any effects on aNSC maintenance, it is unlikely that IF produces any long-lasting effects on the mode of division of aNSCs. In general, we did not notice a difference in the plane of division of aNSCs between control and IF mice, although we did not systematically test for this (would require specific short EdU pulses to capture aNSCs in M-phase). In Figure 2B, the two stem cells shown in the control are unlikely to be the two daughter cells after the division of one aNSC, as one of them is positive and the other negative for Ki67. We only pointed to the second one to show a Ki67-negative aNSC. We will emphasize this in the figure legend.

Improved and more accurate citations: Some references are not properly formatted (e.g., "Dias", page 7, line 288). Some references are included in generalizing statements when they do not contain data to support such statements. For example, Kitamura et al., 2006 did not determine the number of new neurons (only BrdU+ cells) in the SGZ, yet this reference is included among sources supporting that IF "promote survival of newly born neurons" (page 2, line 60). Authors should be more careful how the cite the references.

Thanks for spotting these mistakes, we will correct them and check again all our references. As for the sentence where the Kitamura paper is cited, most of the other references also use only BrdU+ cells while concluding that IF enhances the survival of new neurons. We will change new neurons for new cells to reflect this, which we already bring up in the discussion (see also extended discussion in previous BioRxiv version).

How do the authors explain that they observe 73-80% caloric restriction and yet the final body weight is not different between IF and control animals? Would it suggest that the selected IF protocol or selected diet are not appropriate (see major point 4)?

We also found this surprising and were expecting a change in overall activity in IF mice, which we did not observe. Many factors might play a role, like, as the reviewer suggests, changes in stress levels, which we will measure and show in the revised version.

Given that aNSCs rely more on de novo lipogenesis and fatty acids for their metabolism as shown by Knobloch et al., Nature 2013 and given the interesting changes in RER with the IF shown in this study, it would be interesting to see whether there are differences in Fasn expression in aNSCs between control and IF animals (see minor point 4).

This is an interesting suggestion but given that we see no effect on aNSCs, we find it’s unlikely and unnecessary to test for Fasn expression differences in our IF protocol.

Determining apoptosis in the SGZ by picnotic nuclei (Figure S6A) should be supplemented by determining the number and/or proportion of YFP+ cells positive for the Activated Caspase 3.

We previously found that counting picnotic nuclei is a more accurate and sensitive readout of cell death in the DG, as cells positive for caspase 3 are extremely rare due to the high efficiency of phagocytosis of apoptotic cells by microglia (see Urbán et al., 2016).

Reviewer #1 (Significance (Required)):

General assessment:

This study concludes that aNSCs do not respond to the intermittent fasting. This expands and supplements previous findings that suggest that the intermittent fasting promotes adult neurogenesis by increasing survival and/or proliferation in the Subgranural Zone. The study is well designed, however, over-extends its conclusions beyond a specific fasting paradigm and does not acknowledge serious limitations in the experimental design and analyses. In fact, until major revision is done, which would rule out that the absence of effects of fasting on aNSCs is not due to false negative results, many conclusions from this study cannot be accepted as valid.

Advance:

As mentioned above, the study has a potential to advance our understanding of how fasting affects neurogenesis and fills the knowledge gap of how fasting specifically affects the stem cells. However, unless the study addresses its limitations, its conclusions are not convincing.

Audience:

This study would be particularly interesting for the niche readers from the neurogenesis field. However, the study can also be interesting for researchers in metabolomics and dietology.

My expertise:

adult neurogenesis, neural stem cells, dietology, metabolism

We disagree with the reviewer and find our conclusions well balanced, as we acknowledge our results are to be compared only with similar IF protocols. We also do not believe our results can be attributed to a false negative, as we consistently observe the same with different strains and protocols, always with sufficient animals to make our counts conclusive.

We nevertheless thank the reviewer for assessing our paper and for the advice to improve it. We hope that the reviewer will maintain the same level of scrutiny and scepticism with all IF-related papers.

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

In this manuscript, Gabarro-Solanas et al. question the suitability of IF (Intermittent fasting - non-pharmacological strategy to counteract ageing, which has been previously shown to increase the number of adult-born neurons in the dentate gyrus of mice) as a pro-neurogenic intervention, since IF treatment did not stimulate adult hippocampal neurogenesis, neither at the stem cell level nor on immature and/or dividing neurons. The Authors used a tamoxifen inducible transgenic model (Glast-CreERT2;RYFP mice) to trace neural stem cell lineage and found that IF did not enhance neural stem cell proliferation, nor the abundance of immature, DCX+ neurons. Three-months of IF failed to increase the number of new adult-born neurons (NeuN+/YFP+), while one month of IF significantly reduced the number of new adult-born neurons.

The study appears technically sound, including many different approaches in order to reach its conclusions.

For instance, tamoxifen has been reported to impair various physiological processes, including neurogenesis (Smith et al., 2022), and most studies on adult hippocampal neurogenesis use the C57BL/6J strain of mice; hence, the use of Tamoxifen or that of the GlastCreERT2;RYFP model may have underscored these observations. However, to account for this potentially confounding factor, the Authors characterised the effect of their IF treatment in C57BL/6j mice, also reporting no evident effects of IF as a pro-neurogenic intervention.

I think the study was carefully planned and the analyses well done. Several possible variables were considered, including sex, labelling method, strain, tamoxifen usage or diet length. Several controls were performed in other organs and tissues (liver, fat) to establish the fasting protocol and to check its effects.

Data are presented in a clear way. Quality of images is high level.

In general, it appears as a highly reliable paper reaching an authoritative conclusion for the absence of effect of IF on adult neurogenesis.

Major comments:

I think that the key conclusions are convincing and no further experiments are required.

The methods are presented in such a way that they can be reproduced, and the experiments adequately replicated with proper statistical analysis.

We thank the reviewer for the encouraging remarks and the appreciation of our efforts.

Minor comments:

Prior studies are referenced appropriately, both regarding the IF protocols and the adult neurogenesis modulation.

Line 288 - a reference is incomplete (Dias); integrate with: (Dias et al., 2021)

We will re-format the reference, thanks for spotting the mistake.

There is one concept that is not expressed in the manuscript. Maybe it is not strictly necessary, but I think can be useful to mention it here. It is the fact that most information currently available strongly indicates that adult neurogenesis in humans is not present after adolescence. Of course the research described here is carried out on mice, and in the manuscript it is stated many times that adult hippocampal neurogenesis is strongly decreasing with age, also due to age-related stem cell depletion. Yet, it seems that in humans the exhaustion of such a process can start after adolescence. We know that a sort of controversy is currently present on this subjects, because DCX+ neurons can be detected in adult and old human hippocampi. Yet, it is also clear that there is no substantial cell division (stem cells are depleted) to sustain such hypothetical neurogenesis. Hence, it has been hypothesized that non-newlyborn, "immature" neurons can persist in the absence of cell division, as it has been well demonstrated in the cerebral cortex (see La Rosa et al., 2020 Front Neurosci; Rotheneichner et al., 2018, Cereb Cortex).

This point can be important in the case someone want to use dietary approached such as IF (or any other pharmacological treatment) to stimulate neurogenesis in humans.

We agree with the reviewer and also find this a very interesting and timely topic. However, we find it a bit far from our results and would prefer not to comment on it in the context of the current paper.

Reviewer #2 (Significance (Required)):

The significance of this study relies on the fact that adult neurogenesis field (AN) has been often damaged by the search of "positive" results, aiming at showing that AN does occur "always and everywhere" and that most internal/external stimuli do increase it. This attitude created a bias in the field, persuading many scientists that a result in AN is worthy of publication (or of high impact factor publication) only when a positive result is found.

Personally, I found particularly meaninful the last sentences of the Discussion (reported below), which might seem "off topic" in a research paper, while - I think - underline the real significance of the manuscript:

"In addition, publication bias might be playing a role in skewing the literature on fasting and neurogenesis towards reporting positive results.

In some reviews, even studies reporting no effect are cited as evidence for improved neurogenesis upon IF. Reporting of negative results, especially those challenging accepted dogmas, and a careful and rigorous evaluation of the publications cited in reviews are crucial to avoid unnecessary waste of resources and to promote the advancement of science."

Reviewer field of expertise - keywords: adult neurogenesis, brain structural plasticity, non-newly born immature neurons, comparative neuroplasticity.

We are very happy that the reviewer shares our concern with the biased publication of positive results in the field. We hope our work (and that of Roberts et al., 2022) will encourage other labs to publish their negative results.

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

In this manuscript, Gabarro-Solanas et al. investigate the effects of intermittent fasting (IF) on adult hippocampal neurogenesis in young adult mice. IF has been reported to increase the number of adult-born neuron in the hippocampus, a region that is important for learning and memory. However, it is not well understood what stages of adult neurogenesis are regulated by IF. To address this, the authors utilized lineage tracing and label retention assays in mice undergoing an IF diet. The authors used 2 months old Glast-CreERT2;RYFP mice in combination with Edu label retention to characterize adult NSCs and placed these mice on 1 and 3 months of IF. Despite seeing a decrease in neural stem cell proliferation with age, the authors did not observe a change due to diet. The authors then used immunohistochemistry to characterize changes in cell proliferation, neuroblasts, and new neurons following 1 month and 3 months of IF. Only 1 month of IF seemed to decrease the number of new neurons; however, by 3 months the neuronal output was the same. There were no differences in neuroblasts or cell proliferation due to diet. Gabarro-Solanas et al. conclude that IF transiently and mildly inhibits neurogenesis. Due to contradicting results, the authors then try to determine what variables (sex, labeling method, strain, tamoxifen usage, or diet length) could be affecting their data. The authors saw no substantial differences due to any of their variables.

Major Points

1. The authors analyze NSCs homeostasis and neurogenesis in young adult mice and do not observe any significant changes with their chosen alternate day intermittent fasting paradigm. However, a lot of the data and cell counts appears to be highly variable between animals in the same group. At times, there is an order of magnitude difference between the highest and lowest counts (e.g. Figure 2C,E). According to the method section, it appears that the authors predominantly analyzed a single DG (section?) for most immunostainings, which may explain the large variability in their data. If this is indeed the case, it is insufficient to quantify only a single section for each animal. The authors should quantify several DG sections for each mouse from a pre-defined range along the rostral-caudal axis of the hippocampus in accordance with a standard brain reference atlas. There are also several quantifications, especially of Ki67 where several individuals appear to have no Ki67+ (Figure 3B, 6D) NSCs. These findings are surprising given the still young age of these mice and may be another reflection of the limited brain sections that were analyzed.

The counts are indeed very variable. The counts were made on 1 to 4 DG sections (counted in full), depending on the staining. We will more clearly disclose this information in the revised version. In addition, we will re-count the neuronal output after fasting using stereology. Regarding the very low number of Ki67+ aNSCs, our counts are lower than those in other publications because we are much more stringent with our aNSC identification. Instead of using merely Sox2 (which also labels IPCs), we rely on the presence of a radial GFAP+ process.

There appear to be significant cutting or imaging artifacts across most fluorescent images further raising concerns regarding the accuracy of the quantifications (e.g. Figure 3D, 4C,E, 6B) and publication quality of the images and data. Importantly, uneven section thickness, either from cutting artifacts or imaging issues, may lead to inaccurate cell quantifications a could, possibly, account for the high variability. This issue would further exacerbate concerns regarding the quantification of a single DG section for each animal.

We only processed those samples that passed our QC after sectioning, meaning any unevenly cut brains were never considered (or stained). The stitched images do show artifacts (lower signal in the image junctions), particularly in the NeuN staining. However, this did not affect quantifications, as the measured levels were always clearly above the threshold to consider a cell positive, regardless of the position within the image. The images were cropped to improve the visualisation of NSCs, and to avoid the display of empty tiles. A low magnification image will be provided in the revised version to show that there were no staining artifacts.

It is unclear how NSCs were counted in the B6 mice (Fig 6D,E). The authors only provide a description for the Glast-CRE mice, where they used YFP labeling and GFAP. We assume they performed Sox2/GFAP or Nestin labeling, however, this is not clear at all. The authors should describe their methodology and provide representative images.

We used GFAP, location and morphology to count aNSCs in non-YFP mice. We will make this clear in the text and will also add one more count using Sox2, GFAP and Nestin to identify aNSCs.

NSC populations represent a heterogenous group of stem cells with different replicative properties. As such, the Glast-Cre approach used for the majority of this study may represent a specific subset of NSCs. In line with the previous point, we recommend the authors complement their NSC counts with Sox2/GFAP and Nestin immunostainings.

aNSCs labelled with Glast-Cre are the great majority of aNSCs (>90%) in both ad libitum fed and fasted mice. The data will be included in the revised version. Nevertheless, we will add counts using Sox2, GFAP and Nestin for key experiments.

Stress is a significant negative regulator of neurogenesis. Is it possible that the IF mice display higher stress level which could counteract any beneficial effects of the IF intervention. The authors should provide some measures of stress markers to rule out this potential confounding factor in their IF paradigm.

This is a great suggestion. We will collect blood from control and fasted mice and measure the levels of stress factors (e.g. corticosterone). We will include the data in our revised version.

Minor Point

1. The authors state that "Experimental groups were formed by randomly assigning mice from different litters within each mouse strain and all experiments were conducted in male and female mice". Given that neurogenesis, especially at young ages, is highly sensitive to the exact age of the mice, the authors should provide a rationale why animals from different litters instead of littermate controls were used in these experiments.

Littermate controls were always used in the experiments. But also, more than one litter was used for each experiment, since one litter was never generating enough mice for the experiments. We will clarify this point in text.

Currently, the statistical tests are only described in the method section, however it would be helpful if this information to be integrated into the figure legend as well. Additionally, the authors provide individual data points for some but not all bar graphs (eg Figure 1D).

We will consider including the statistical information in the figure legend, provided there is not a maximum length for figure legends. In the case of figure 1D, data points are not shown because of how the food intake was calculated: as an average per cage instead of per animal (included in the materials and methods). We therefore do not consider it useful to show the datapoints in the final version of the manuscript, but will provide them for the reviewer.

Cell counts per AU is a rather unorthodox unit. With a representative selection of tissue for each animal, the authors could avoid the need to normalize to the DG length and may be able to extrapolate an estimate of cell counts for the entire DG instead.

Thanks for the suggestion. Our arbitrary units (AU) were in fact already equivalent to cells per mm of DG, and we have updated our graphs to reflect this.

In Figure 4D, the authors highlight a few NSC with arrowheads. At a quick glance this is rather confusing as it appears that the authors only counted 3 NSCs in each picture. It may be a better option to show a zoomed in picture to highlight an example of a representative NSC.

Examples of representative NSCs are already shown in Fig 2. With this image, we intended to show a larger number of NSCs. We realise the arrows only pointed to some of them, making the message confusing. We will consider removing them from the figure in the revised version.

In Supplementary Figure S6, the authors should complement the quantification of the nuclei with representative images.

We will include representative images in Figure S6.

For the daytime IF, did the authors assess weights, food intake, RER as well liver/fat measurements similar to night-time IF? If so, this data should be provided in the supplement.

We do have data for the daytime IF in the metabolic cages, which was taken from mice housed in groups (during the preliminary phase of our study). We also have the weight and data on neurogenesis, which we will show as a supplement.

Reviewer #3 (Significance (Required)):

The authors are commended for compiling a manuscript on what is commonly considered 'negative data', that, at the same time, are also contradicting independent reports on the effects of IF on neurogenesis. The studies outlined in this manuscript are comprehensive and mostly well designed. Given the broad, growing interest in dietary restriction as an aging intervention the study is timely.

We thank the reviewer for the positive assessment of the significance of our work.

Reviewer #4 (Evidence, reproducibility and clarity (Required)):

Summary:

In this manuscript, Gabarró-Solanas et al. tested the effect of intermitted fasting (IF, every-other-day fasting) on adult neural stem cells and neurogenesis. They demonstrate that the paradigm they have used does not affect NSC activation or maintenance, and does also not promote neurogenesis. As previous reports showed increased neurogenesis with IF, the authors controlled for various parameters such as mouse strain, sex, and diet length. They also used different methods of identification of newborn neurons, such as tamoxifen-induced lineage-tracing versus birth-dating with thymidine-analogues to substantiate their findings.

Major comments:

This study is very well done with carefully designed and controlled experiments. The manuscript reads nicely and the data are presented in a clear way, making it easy to follow. The authors have done a "tour-de force" to rule out confounding factors that might influence their findings that IF does not affect NSCs nor neurogenesis.

The claims and conclusions are supported by the data. The methods are clearly described and should allow to reproduce the data independently. The number of replicates (i.e. the number of mice analyzed) is impressive and statistical analysis is adequate.

The major findings, namely that the chosen IF does not affect NSCs and neurogenesis is not in line with some previous studies. Despite a careful ruling out of potentially confounding factors (see also "significance" below), it remains unclear why other studies have found an increase in neurogenesis with IF. As each of these studies has some specific experimental design, it is difficult to judge these data in the context of previous data without going through all the details of the other studies. It would thus be a great help for the reader if the authors could provide a table or schematic, which lists the major parameters of each of these studies, such as detailed paradigm of IF, age of mice at start, sex, duration of the intervention, method of identification of NSCs and neurogenesis etc.

This is a very good suggestion, and we had already created such a table. We, however, consider that it might be better suited for a review on the effects of IF on neurogenesis than for this work. We will include the table in our response to the reviewers together with our revised version.

Two points that the authors have not discussed might also be worth mentioning in the discussion part:

1.) The mice in the night-time IF were single caged, could there be a potential negative effect on neurogenesis that would mask the presumably beneficial effect of IF? Although the controls were also single caged, the stress of social isolation might play a role?

The mice were only single caged for the metabolic phenotyping, but not for the neurogenic counts. We will make this clearer in the text. In any case, we do agree that stress might play a role and we will measure stress levels in the control and fasted mice and will include this data in the revised version.

2.) The IF mice gained the same weight over time (Fig. S2), but had a ~20% reduction in overall calory intake. This would be explainable by a reduction in energy expenditure, but the overall activity was also not significantly changed (Fig. S1). Can the authors speculate why they reach the same weight with less calories?

We also found this surprising and were expecting a reduction in the overall activity of the fasted mice. We do not have an explanation for this discrepancy, but perhaps stress levels might explain part of it (we will check stress levels in the revised version). We will also look at whether energy expenditure and activity levels changed over time.

Minor comments:

1.) It would be nice to replace the arbitrary units (AU) in the graphs were this is used (e.g. Fig. 2F, 3C, 4B, D and F etc) to the actual number of cells per a certain µm DG, so that the number of cells can be put in context and compared between the figures.

Yes, our AU already corresponded to mm and we will update our figures accordingly.

2.) Fig 3 D: can the authors also show the Ki67 channel to illustrate how it looks after a 3 month IF?

We find it does not help much, as Ki67+ cells are mostly IPCs and that data is already shown in Fig. 4A. We will nevertheless include the image in our response to the reviewers together with our revised version.

3.) Fig.4E: the NeuN staining looks strangely interrupted, this might be due to tile-stitching? In that case, it would be better to either only show one segment or to try to get a better stitching algorhythm.

It is indeed because of the tile-stitching and uneven illumination. However, this did not affect the counts, as already discussed in the response to reviewer #3 (major point #2).

4.) Fig.6 D shows a minus axis in Y-axis, this should only been shown from 0 to positive values, as it is a percentage of cells and cannot be negative.

True, thanks for spotting this. We will correct the graphs in the revised version.

5.) Fig.6 B: the same problem with the NeuN staining as mentioned under point 3. This should be improved.

As with point 3, the stitching did not affect the quantification. We find it more accurate to show the image with the stitching, as that was the one used for quantification. We will provide a new picture with lower magnification to better show the quality of the staining.

6.) Fig. S6B: maybe add a comment in the result part or in the figure legend that a 10 day chase after an EdU pulse is not the classical protocol to look at mature NeuN positive neurons. But apparently enough newborn neurons were already NeuN positive for this quantification.

We fully agree 10 days is not the standard for neuronal identification. We did find neurons after the 10-day chase but in low numbers. We will add a comment in the text of the revised version to clarify this.

7.) The authors refer to personal communications with M. Mattson and S. Thuret to underline that circadian disruption is not enough to explain the differences (line 367 onwards). Can they refer the reader to published data instead?

While the results are published in their papers, the methods did not specify the time at which the food was added/removed for the IF protocol. That is why we refer to personal communication.

Further showing that disruption of circadian rhythms is not enough to explain the difference in outcome of the IF protocol, we will show the data for the 1-month daytime IF, which again does not increase adult neurogenesis (reviewer #3, minor point #6).

Reviewer #4 (Significance (Required)):

Given the great interest in the seemingly positive effects on health of IF in general, and also for increasing neurogenesis, it is important to better understand the mechanism of this intervention. The study by Gabarró-Solanas et al. clearly demonstrates that IF is not a universal, "works all the time" way of increasing neurogenesis. The study is very well done, with well controlled and measured parameters. It shows that a physiological interference such as IF might depend on many factors and might be less robust across laboratories than anticipated. This study is a very good example that all the details of the experimental settings need to be taken into consideration and are ideally reported with every IF study. It is also a good example how to follow up "no effect" data in a way that they are conclusive.

The significance of this study is to point out that IF as a strategy to increase neurogenesis needs to be reconsidered. It raises the questions how IF can be beneficial in some studies and not in others, asking for more experiments to better understand the detailed mechanisms of IF action. In a systematic approach, this study rules out some of the potentially confounding factors and shows that at least with the chosen IF paradigm, these factors are not the reason for not seeing increased neurogenesis. The study is thus of clear interest for the neurogenesis field and will also need to be considered by the broader field of IF research, although it speaks against the beneficial effects of IF. It might have the potential to bring together the different study authors who did or did not see increased neurogenesis with IF and discuss together the non-published details of their study design to advance the field.

We thank the reviewer for the positive assessment of our work and for acknowledging its importance for the broader field of IF research.

List of references used in the response to reviewers:

Anson, R. M. et al. Intermittent fasting dissociates beneficial effects of dietary restriction on glucose metabolism and neuronal resistance to injury from calorie intake. Proceedings of the National Academy of Sciences 100, 6216–6220 (2003).

Bok, E. et al. Dietary Restriction and Neuroinflammation: A Potential Mechanistic Link. International Journal of Molecular Sciences 20, 464 (2019).

Cignarella, F. et al. Intermittent Fasting Confers Protection in CNS Autoimmunity by Altering the Gut Microbiota. Cell Metabolism 27, 1222-1235.e6 (2018).

Dai, S. et al. Intermittent fasting reduces neuroinflammation in intracerebral hemorrhage through the Sirt3/Nrf2/HO-1 pathway. Journal of Neuroinflammation 19, 122 (2022).

Dias, G. P. et al. Intermittent fasting enhances long-term memory consolidation, adult hippocampal neurogenesis, and expression of longevity gene Klotho. Mol Psychiatry 1–15 (2021).

Goodrick, C. L., Ingram, D. K., Reynolds, M. A., Freeman, J. R. & Cider, N. Effects of intermittent feeding upon body weight and lifespan in inbred mice: interaction of genotype and age. Mechanisms of Ageing and Development 55, 69–87 (1990).

Gudden, J., Arias Vasquez, A. & Bloemendaal, M. The Effects of Intermittent Fasting on Brain and Cognitive Function. Nutrients 13, 3166 (2021).

Lee, J., Seroogy, K. B. & Mattson, M. P. Dietary restriction enhances neurotrophin expression and neurogenesis in the hippocampus of adult mice. Journal of Neurochemistry 80, 539–547 (2002).

Rangan, P. et al. Fasting-mimicking diet cycles reduce neuroinflammation to attenuate cognitive decline in Alzheimer’s models. Cell Reports 40, 111417 (2022).

Roberts, L. D. et al. The 5:2 diet does not increase adult hippocampal neurogenesis or enhance spatial memory in mice. 2022.10.03.510613 BioRxiv Preprint (2022).

Song, M.-Y. et al. Energy restriction induced SIRT6 inhibits microglia activation and promotes angiogenesis in cerebral ischemia via transcriptional inhibition of TXNIP. Cell Death Dis 13, 449 (2022).

Urbán, N. et al. Return to quiescence of mouse neural stem cells by degradation of a proactivation protein. Science 353, 292–295 (2016).

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

Learn more at Review Commons

---

Referee #4

Evidence, reproducibility and clarity

Summary:

In this manuscript, Gabarró-Solanas et al. tested the effect of intermitted fasting (IF, every-other-day fasting) on adult neural stem cells and neurogenesis. They demonstrate that the paradigm they have used does not affect NSC activation or maintenance, and does also not promote neurogenesis. As previous reports showed increased neurogenesis with IF, the authors controlled for various parameters such as mouse strain, sex, and diet length. They also used different methods of identification of newborn neurons, such as tamoxifen-induced lineage-tracing versus birth-dating with thymidine-analogues to substantiate their findings.

Major comments:

This study is very well done with carefully designed and controlled experiments. The manuscript reads nicely and the data are presented in a clear way, making it easy to follow. The authors have done a "tour-de force" to rule out confounding factors that might influence their findings that IF does not affect NSCs nor neurogenesis. The claims and conclusions are supported by the data. The methods are clearly described and should allow to reproduce the data independently. The number of replicates (i.e. the number of mice analyzed) is impressive and statistical analysis is adequate.

The major findings, namely that the chosen IF does not affect NSCs and neurogenesis is not in line with some previous studies. Despite a careful ruling out of potentially confounding factors (see also "significance" below), it remains unclear why other studies have found an increase in neurogenesis with IF. As each of these studies has some specific experimental design, it is difficult to judge these data in the context of previous data without going through all the details of the other studies. It would thus be a great help for the reader if the authors could provide a table or schematic, which lists the major parameters of each of these studies, such as detailed paradigm of IF, age of mice at start, sex, duration of the intervention, method of identification of NSCs and neurogenesis etc.

Two points that the authors have not discussed might also be worth mentioning in the discussion part:

1. The mice in the night-time IF were single caged, could there be a potential negative effect on neurogenesis that would mask the presumably beneficial effect of IF? Although the controls were also single caged, the stress of social isolation might play a role?
2. The IF mice gained the same weight over time (Fig. S2), but had a ~20% reduction in overall calory intake. This would be explainable by a reduction in energy expenditure, but the overall activity was also not significantly changed (Fig. S1). Can the authors speculate why they reach the same weight with less calories?

Minor comments:

1. It would be nice to replace the arbitrary units (AU) in the graphs were this is used (e.g. Fig. 2F, 3C, 4B, D and F etc) to the actual number of cells per a certain µm DG, so that the number of cells can be put in context and compared between the figures.
2. Fig 3 D: can the authors also show the Ki67 channel to illustrate how it looks after a 3 month IF?
3. Fig.4E: the NeuN staining looks strangely interrupted, this might be due to tile-stitching? In that case, it would be better to either only show one segment or to try to get a better stitching algorhythm.
4. Fig.6 D shows a minus axis in Y-axis, this should only been shown from 0 to positive values, as it is a percentage of cells and cannot be negative.
5. Fig.6 B: the same problem with the NeuN staining as mentioned under point 3. This should be improved.
6. Fig. S6B: maybe add a comment in the result part or in the figure legend that a 10 day chase after an EdU pulse is not the classical protocol to look at mature NeuN positive neurons. But apparently enough newborn neurons were already NeuN positive for this quantification.
7. The authors refer to personal communications with M. Mattson and S. Thuret to underline that circadian disruption is not enough to explain the differences (line 367 onwards). Can they refer the reader to published data instead?

Significance

Given the great interest in the seemingly positive effects on health of IF in general, and also for increasing neurogenesis, it is important to better understand the mechanism of this intervention. The study by Gabarró-Solanas et al. clearly demonstrates that IF is not a universal, "works all the time" way of increasing neurogenesis. The study is very well done, with well controlled and measured parameters. It shows that a physiological interference such as IF might depend on many factors and might be less robust across laboratories than anticipated. This study is a very good example that all the details of the experimental settings need to be taken into consideration and are ideally reported with every IF study. It is also a good example how to follow up "no effect" data in a way that they are conclusive.

The significance of this study is to point out that IF as a strategy to increase neurogenesis needs to be reconsidered. It raises the questions how IF can be beneficial in some studies and not in others, asking for more experiments to better understand the detailed mechanisms of IF action. In a systematic approach, this study rules out some of the potentially confounding factors and shows that at least with the chosen IF paradigm, these factors are not the reason for not seeing increased neurogenesis. The study is thus of clear interest for the neurogenesis field and will also need to be considered by the broader field of IF research, although it speaks against the beneficial effects of IF. It might have the potential to bring together the different study authors who did or did not see increased neurogenesis with IF and discuss together the non-published details of their study design to advance the field.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

In this manuscript, Gabarro-Solanas et al. investigate the effects of intermittent fasting (IF) on adult hippocampal neurogenesis in young adult mice. IF has been reported to increase the number of adult-born neuron in the hippocampus, a region that is important for learning and memory. However, it is not well understood what stages of adult neurogenesis are regulated by IF. To address this, the authors utilized lineage tracing and label retention assays in mice undergoing an IF diet. The authors used 2 months old Glast-CreERT2;RYFP mice in combination with Edu label retention to characterize adult NSCs and placed these mice on 1 and 3 months of IF. Despite seeing a decrease in neural stem cell proliferation with age, the authors did not observe a change due to diet. The authors then used immunohistochemistry to characterize changes in cell proliferation, neuroblasts, and new neurons following 1 month and 3 months of IF. Only 1 month of IF seemed to decrease the number of new neurons; however, by 3 months the neuronal output was the same. There were no differences in neuroblasts or cell proliferation due to diet. Gabarro-Solanas et al. conclude that IF transiently and mildly inhibits neurogenesis. Due to contradicting results, the authors then try to determine what variables (sex, labeling method, strain, tamoxifen usage, or diet length) could be affecting their data. The authors saw no substantial differences due to any of their variables.

Major Points

1. The authors analyze NSCs homeostasis and neurogenesis in young adult mice and do not observe any significant changes with their chosen alternate day intermittent fasting paradigm. However, a lot of the data and cell counts appears to be highly variable between animals in the same group. At times, there is an order of magnitude difference between the highest and lowest counts (e.g. Figure 2C,E). According to the method section, it appears that the authors predominantly analyzed a single DG (section?) for most immunostainings, which may explain the large variability in their data. If this is indeed the case, it is insufficient to quantify only a single section for each animal. The authors should quantify several DG sections for each mouse from a pre-defined range along the rostral-caudal axis of the hippocampus in accordance with a standard brain reference atlas. There are also several quantifications, especially of Ki67 where several individuals appear to have no Ki67+ (Figure 3B, 6D) NSCs. These findings are surprising given the still young age of these mice and may be another reflection of the limited brain sections that were analyzed.
2. There appear to be significant cutting or imaging artifacts across most fluorescent images further raising concerns regarding the accuracy of the quantifications (e.g. Figure 3D, 4C,E, 6B) and publication quality of the images and data. Importantly, uneven section thickness, either from cutting artifacts or imaging issues, may lead to inaccurate cell quantifications a could, possibly, account for the high variability. This issue would further exacerbate concerns regarding the quantification of a single DG section for each animal.
3. It is unclear how NSCs were counted in the B6 mice (Fig 6D,E). The authors only provide a description for the Glast-CRE mice, where they used YFP labeling and GFAP. We assume they performed Sox2/GFAP or Nestin labeling, however, this is not clear at all. The authors should describe their methodology and provide representative images.
4. NSC populations represent a heterogenous group of stem cells with different replicative properties. As such, the Glast-Cre approach used for the majority of this study may represent a specific subset of NSCs. In line with the previous point, we recommend the authors complement their NSC counts with Sox2/GFAP and Nestin immunostainings.
5. Stress is a significant negative regulator of neurogenesis. Is it possible that the IF mice display higher stress level which could counteract any beneficial effects of the IF intervention. The authors should provide some measures of stress markers to rule out this potential confounding factor in their IF paradigm.

Minor Point

1. The authors state that "Experimental groups were formed by randomly assigning mice from different litters within each mouse strain and all experiments were conducted in male and female mice". Given that neurogenesis, especially at young ages, is highly sensitive to the exact age of the mice, the authors should provide a rationale why animals from different litters instead of littermate controls were used in these experiments.
2. Currently, the statistical tests are only described in the method section, however it would be helpful if this information to be integrated into the figure legend as well. Additionally, the authors provide individual data points for some but not all bar graphs (eg Figure 1D).
3. Cell counts per AU is a rather unorthodox unit. With a representative selection of tissue for each animal, the authors could avoid the need to normalize to the DG length and may be able to extrapolate an estimate of cell counts for the entire DG instead.
4. In Figure 4D, the authors highlight a few NSC with arrowheads. At a quick glance this is rather confusing as it appears that the authors only counted 3 NSCs in each picture. It may be a better option to show a zoomed in picture to highlight an example of a representative NSC.
5. In Supplementary Figure S6, the authors should complement the quantification of the nuclei with representative images.
6. For the daytime IF, did the authors assess weights, food intake, RER as well liver/fat measurements similar to night-time IF? If so, this data should be provided in the supplement.

Significance

The authors are commended for compiling a manuscript on what is commonly considered 'negative data', that, at the same time, are also contradicting independent reports on the effects of IF on neurogenesis. The studies outlined in this manuscript are comprehensive and mostly well designed. Given the broad, growing interest in dietary restriction as an aging intervention the study is timely.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

In this manuscript, Gabarro-Solanas et al. question the suitability of IF (Intermittent fasting - non-pharmacological strategy to counteract ageing, which has been previously shown to increase the number of adult-born neurons in the dentate gyrus of mice) as a pro-neurogenic intervention, since IF treatment did not stimulate adult hippocampal neurogenesis, neither at the stem cell level nor on immature and/or dividing neurons. The Authors used a tamoxifen inducible transgenic model (Glast-CreERT2;RYFP mice) to trace neural stem cell lineage and found that IF did not enhance neural stem cell proliferation, nor the abundance of immature, DCX+ neurons. Three-months of IF failed to increase the number of new adult-born neurons (NeuN+/YFP+), while one month of IF significantly reduced the number of new adult-born neurons.

The study appers technically sound, including many different approaches in order to reach its conclusions. For instance, tamoxifen has been reported to impair various physiological processes, including neurogenesis (Smith et al., 2022), and most studies on adult hippocampal neurogenesis use the C57BL/6J strain of mice; hence, the use of Tamoxifen or that of the GlastCreERT2;RYFP model may have underscored these observations. However, to account for this potentially confounding factor, the Authors characterised the effect of their IF treatment in C57BL/6j mice, also reporting no evident effects of IF as a pro-neurogenic intervention. I think the study was carefully planned and the analyses well done. Several possible variables were considered, including sex, labelling method, strain, tamoxifen usage or diet length. Several controls were performed in other organs and tissues (liver, fat) to establish the fasting protocol and to check its effects. Data are presented in a clear way. Quality of images is high level. In general, it appears as a highly reliable paper reaching an authoritative conclusion for the absence of effect of IF on adult neurogenesis.

Major comments:

I think that the key conclusions are convincing and no further experiments are required. The methods are presented in such a way that they can be reproduced, and the experiments adequately replicated with proper statistical analysis.

Minor comments:

Prior studies are referenced appropriately, both regarding the IF protocols and the adult neurogenesis modulation. Line 288 - a reference is incomplete (Dias); integrate with: (Dias et al., 2021) There is one concept that is not expressed in the manuscript. Maybe it is not strictly necessary, but I think can be useful to mention it here. It is the fact that most information currently available strongly indicates that adult neurogenesis in humans is not present after adolescence. Of course the research described here is carried out on mice, and in the manuscript it is stated many times that adult hippocampal neurogenesis is strongly decreasing with age, also due to age-related stem cell depletion. Yet, it seems that in humans the exhaustion of such a process can start after adolescence. We know that a sort of controversy is currently present on this subjects, because DCX+ neurons can be detected in adult and old human hippocampi. Yet, it is also clear that there is no substantial cell division (stem cells are depleted) to sustain such hypothetical neurogenesis. Hence, it has been hypothesized that non-newlyborn, "immature" neurons can persist in the absence of cell division, as it has been well demonstrated in the cerebral cortex (see La Rosa et al., 2020 Front Neurosci; Rotheneichner et al., 2018, Cereb Cortex). This point can be important in the case someone want to use dietary approached such as IF (or any other pharmacological treatment) to stimulate neurogenesis in humans.

Significance

The significance of this study relies on the fact that adult neurogenesis field (AN) has been often damaged by the search of "positive" results, aiming at showing that AN does occur "always and everywhere" and that most internal/external stimuli do increase it. This attitude created a bias in the field, persuading many scientists that a result in AN is worthy of publication (or of high impact factor publication) only when a positive result is found.

Personally, I found particularly meaninful the last sentences of the Discussion (reported below), which might seem "off topic" in a research paper, while - I think - underline the real significance of the manuscript: "In addition, publication bias might be playing a role in skewing the literature on fasting and neurogenesis towards reporting positive results.

In some reviews, even studies reporting no effect are cited as evidence for improved neurogenesis upon IF. Reporting of negative results, especially those challenging accepted dogmas, and a careful and rigorous evaluation of the publications cited in reviews are crucial to avoid unnecessary waste of resources and to promote the advancement of science."

Reviewer field of expertise - keywords: adult neurogenesis, brain structural plasticity, non-newly born immature neurons, comparative neuroplasticity.

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, mice were exposed to a specific form of so-called Intermittent Fasting (IF) and the effects of IF on adult neogenesis in the hippocampus were determined. The specific IF protocol used had no effect on activation, proliferation, or maintenance of adult Neural Stem Cells (aNSCs) and displayed a decrease in number of new neurons in the neurogenic niche but only after 1 month of the IF protocol. These results contrast previously published results from multiple studies that concluded that IF promotes survival of new neurons and by extension promote adult neurogenesis. The unresponsiveness of aNSCs or their immediate cell progeny, the Intermediate Neural Progenitors (IPCs), to IF is a novel finding. The authors make several relevant points in the discussion about the publication bias towards positive results (or omission of negative results), which may reinforce established dogmas. However, the presented results did not convincingly demonstrate that the absence of effects of IF on aNSCs or adult neurogenesis is simply not a result of a specific IF paradigm, which is not robust enough to elicit changes in adult neurogenesis. In other words, there is a lack of positive controls and alternative protocols that would rule out that the observed absence of effects is not a consequence of type II error (the error of omission), or more colloquially, a consequence of false negatives.

Major Comments:

1. Protocol-driven absence of effects: The absence of IF effects on aNSCs and IPCs observed in this study does not lend it the authority to conclude that aNSCs are resilient to IF or all IF paradigms and protocols. The absence of IF effects on aNSCs and neurogenesis could be specifically related to the chosen IF paradigm. Indeed, not all previous studies that observed IF-driven effects on adult neurogenesis used the same "night-time every-other-day fasting" protocol chosen in this study. For example, Brandhorst et al., 2015 (cited in this paper) used 4 days of IF 2x per month and observed an increase of DCX+BrdU+ cells. On the other hand, certain previous studies used the same or similar IF protocol used here, but often with longer duration or with a post-fasting ad libitum feeding period, which may be responsible for the pro-neurogenic or pro-survival effects. In fact, the authors acknowledge this in the discussion (page 7, lines 289-290 and 292-294). Why would the authors then not include similar feeding/IF paradigm in their study and determine if these would generate effects on survival of new neurons but also on aNSCs and/or IPCs? In addition, the authors acknowledge that the chosen IF paradigm may have affected the stress levels or behaviour of mice (page 9, lines 372-378). Why did they not test if their IF protocol does not increase stress or anxiety of mice by simple behaviour tests such as open field or elevated T maze? Alarmingly, the used IF protocol does not result in changes in final weight or growth curves (S.Fig.2), which is surprising and raises a question the used IF protocol is robust enough or appropriate. Finally, the authors acknowledge that their own results do not support well-established findings such as aging-related reduction in number of aNSCs (page 4, lines 177-179). This again questions whether the selected protocols and treatments are appropriate.
2. Lack of topic-specific positive controls: The authors successfully demonstrated that the used IF protocol differentially impacts the adipose tissue and liver, while also inducing body weight fluctuations synchronized with the fasting periods. However, these peripheral effects outside the CNS do not directly imply that the chosen IF protocol is robust enough to elicit cellular or molecular changes in the hippocampus. The authors need to demonstrate that their IF protocol affects previously well-established CNS parameters associated with fasting such as astrocyte reactivity, inflammation or microglia activation, among other factors. In fact, they acknowledge this systemic problem in the discussion (page 8, lines 359-360).
3. Problematic cell analyses: Cell quantification should be performed under stereological principles. However, the presented results did not adhere to stereological quantification. Instead, the authors chose to quantify specific cell phenotypes only in subjectively selected subsets of regions of interest, i.e., the Subgranular Zone (SGZ). This subjective pre-selection may have been responsible for the absence of effects, especially if these are either relatively small or dependent on anatomical sections of SGZ. For example, IF may exert effects on caudal SGZ more than on rostral SGZ. But if the authors quantified only (or predominantly) rostral SGZ, they may have missed these effects by biasing one segment of SGZ versus other. The authors should apply stereological quantification at least to the quantification of new neurons and test if this approach replicated previously observed pro-survival effects of IF. Also, the authors should describe how they pre-selected the ROI for cell quantification in greater details.
4. Alarming exclusion of data points: There appears to be different number of data points in different graphs that are constructed from same data sets. For example, in the 3-month IF data set in Figure 4, there are 14 data points for the graph of Ki67+ cells (Fig.4B), but 16 (or 17) data points for the graph of DCX+ cells (Fig.4D). How is that possible? If data points were excluded, what objective and statistical criteria were applied to make sure that such exclusion is not subjective and biased? In fact, the authors state that "Samples with poor staining quality were also excluded from quantifications" (page 12, line 528-529). Poor preparation of tissue is not only suboptimal but not a valid objective reason for data point exclusion. This major issue needs to be explained and corrected.
5. Different pulse-and-chase time-points: One of the reasons why this study has found that aNSCs may not be responsive to IF could be the use of less appropriate pulse-and-chase time-points either after EdU or after Tamoxifen for cell lineage tracing. The authors observed that IF has negative effects on new neurons initially (Fig.4F). Similarly, it is well established that voluntary physical exercise affects SGZ adult neurogenesis only during the first 2 weeks. After this period, the neurogenic effects of exercise are diminished beyond observational detection (i.e., van Praag's and Kempermann's papers in the past 25 years). These two arguments suggest that the observed absence of aNSC responsiveness might be a consequence of the chosen EdU administration and the EdU pulse should not be administered 15 days after Tamoxifen/IF protocol start but earlier, in the first week of the IF protocol. In fact, the decreased number of new neurons during the initial IF phase may not be only a consequence of reduced survival but of higher aNSC quiescence during the first week of the IF protocol.
6. Discussion needs more specificity and clarity: The authors claim that the absence of IF effects on neurogenesis is multi-layered (including the influence of age, sex, specific cell labelling protocols etc.) but they do not specifically address why certain studies did find IF-driven neurogenic effects while they did not. In addition, some statements and points in the discussion are not clear. For example, when the authors refer to their own experiments (page 8, lines 331-334), it is not clear, which experiments they have in mind.

Minor comments:

1. Change in the title: The authors have shown that a very specific IF protocol does not affect aNSCs but initially decreases number of new neurons in SGZ. The title should reflect this. For example, it could state "Specific (night-time every-other-day) fasting does not affect aNSCs but initially decreases survival of new neurons in the SGZ".
2. Data depiction: Data in 3 datasets were found not normally distributed (Fig. S5A, B and S6A) and were correctly analysed with non-parametric tests. However, the corresponding graphs wrongly depict the data as mean +/- SD while they should depict median +/- IQR (or similar adequate value) because non-parametric statistical tests do not compare means but medians.
3. Statistical analysis: For ANOVA, the F and p values are not listed anywhere. The presented asterisks in the graphs are only for non-ANOVA or ANOVA post-hoc tests. This does not allow to judge statistical significance well and should be corrected.
4. Asymmetric vs Symmetric cell divisions: Representative images in Fig.2B suggest that IF may affect the plane of cell division for the Type-1 aNSCs. The plane of cell division is an indirect indicator of symmetric vs asymmetric (exhaustive vs maintaining) modes of cell division. Is it possible, IF influences this, especially during the first week of IF (see major comment 5)?
5. Improved and more accurate citations: Some references are not properly formatted (e.g., "Dias", page 7, line 288). Some references are included in generalizing statements when they do not contain data to support such statements. For example, Kitamura et al., 2006 did not determine the number of new neurons (only BrdU+ cells) in the SGZ, yet this reference is included among sources supporting that IF "promote survival of newly born neurons" (page 2, line 60). Authors should be more careful how the cite the references.
6. How do the authors explain that they observe 73-80% caloric restriction and yet the final body weight is not different between IF and control animals? Would it suggest that the selected IF protocol or selected diet are not appropriate (see major point 4)?
7. Given that aNSCs rely more on de novo lipogenesis and fatty acids for their metabolism as shown by Knobloch et al., Nature 2013 and given the interesting changes in RER with the IF shown in this study, it would be interesting to see whether there are differences in Fasn expression in aNSCs between control and IF animals (see minor point 4).
8. Determining apoptosis in the SGZ by picnotic nuclei (Figure S6A) should be supplemented by determining the number and/or proportion of YFP+ cells positive for the Activated Caspase 3.

Significance

General assessment:

This study concludes that aNSCs do not respond to the intermittent fasting. This expands and supplements previous findings that suggest that the intermittent fasting promotes adult neurogenesis by increasing survival and/or proliferation in the Subgranural Zone. The study is well designed, however, over-extends its conclusions beyond a specific fasting paradigm and does not acknowledge serious limitations in the experimental design and analyses. In fact, until major revision is done, which would rule out that the absence of effects of fasting on aNSCs is not due to false negative results, many conclusions from this study cannot be accepted as valid.

Advance:

As mentioned above, the study has a potential to advance our understanding of how fasting affects neurogenesis and fills the knowledge gap of how fasting specifically affects the stem cells. However, unless the study addresses its limitations, its conclusions are not convincing.

Audience:

This study would be particularly interesting for the niche readers from the neurogenesis field. However, the study can also be interesting for researchers in metabolomics and dietology.

My expertise:

adult neurogenesis, neural stem cells, dietology, metabolism

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

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

Marta Sanvicente-García et al and colleague developed a comprehensive and versatile genome editing web application tool and a nextflow pipeline to give support to gene editing experimental design and analysis.

The manuscript is well written and all data are clearly shown.

While I did not tested extensively, the software seems to work well and I have no reason to doubt the authors' claims.

I usually prefer ready to use web applications like outknocker, they are in general easier to use for rookies (it would be good if the author could cite it, since it is very well implemented) but the nextflow implementation is anyway well suited.

We have been able to analyze the testing dataset that they provide, but we have tried to run it with our dataset and we have not been able to obtain results. We have also tried to run it with the testing dataset of CRISPRnano and CRISPResso2 without obtaining results. The error message has been in all the cases: “No reads mapping to the reference sequence were found.”

Few minor points:

Regarding the methods to assess whether the genome editing is working or not, I would definitely include High Resolution Melt Analysis, which is by far the fastest and probably more sensitive amongst the others.

Following the Reviewer 1 suggestion, we have added this technique in the introduction: “Another genotyping method that has been successfully used to evaluate genome editing is high-resolution melting analysis (HRMA) [REFERENCE]. This is a simple and efficient real-time polymerase chain reaction-based technique.”

Another point that would important to taclke is that often these pipelines do nto define the system they are working with (eg diploid, aploid vs etc). This will change the number of reads needed ato unambigously call the genotypes detected and to perform the downstream analysis (the CRISPRnano authors mentioned this point).

In the introduction, it is already said: " it is capable of analyzing edited bulk cell populations as well as individual clones". In addition, following this suggestion we have added in the help page of CRISPR-A web application and in the documentation of the nextflow pipeline a recommended sample coverage to orient the users on that.

I am also wondering whether the name CRISPR-A is appropriate since someone could confuse it with CRISPRa.

CRISPR-A is an abbreviation for CRISPR-Analytics. Even if it is true that it can be pronounced in the same way that CRISPRa screening libraries, it is spelled differently and would be easily differentiated by context.

CROSS-CONSULTATION COMMENTS

Reviewer 2 made an excellent work and raised important concerns about the software they need to be addressed carefully.

In the meantime we had more time to test the software and can confirm some of the findings of Reviewer 1:

1) We spent hours running (unsuccessfully) CRISPR A on Nextflow. The software does not seem to run properly.

2) No manual or instruction can be found on both their repositories (https://bitbucket.org/synbiolab/crispr-a_nextflow/

https://bitbucket.org/synbiolab/crispr-a_figures/)

We have added a readme.md file to both repositories and we hope that with the new documentation the software can be downloaded and run easily. We have also added an example test in CRISPR-A nextflow pipeline to facilitate the testing of the software. Currently, the software is implemented in DLS1 instead of DLS2, making it impossible to be run with the latest version of nextflow. We are planning to make the update soon, but we want to do it while moving the pipeline to crisprseq nf-core pipeline to follow better standards and make it fully reproducible and reusable.

Few more points to be considered:

• UMI clustering is not proper terminology. Barcode multiplexing/demultiplexing (SQK-LSK109 from Oxford Nanopore).

We have added more details in the methods section “Library prep and Illumina sequencing with Unique Molecular Identifiers (UMIs)” to clarify the process and used terminology: “Uni-Molecular Identifiers are added through a 2 cycles PCR, called UMI tagging, to ensure that each identifier comes just from one molecule. Barcodes to demultiplex by sample are added later, after the UMI tagging, in the early and late PCR.”

We had already explained the computational pipeline through which these UMIs are clustered together to obtain a consensus of the amplified sequences in “CRISPR-A gene editing analysis pipeline” section in methods:

“An adapted version of extract_umis.py script from pipeline_umi_amplicon pipeline (distributed by ONT https://github.com/nanoporetech/ pipeline-umi-amplicon) is used to get UMI sequences from the reads, when the three PCRs experimental protocol is applied. Then vsearch⁴⁸ is used to cluster UMI sequences. UMIs are polished using minimap2³² and racon⁴⁹ and consensus sequences are obtained using minialign (https://github.com/ocxtal/minialign) and medaka (https://github.com/nanoporetech/medaka).”

We also have added the following in “CRISPR-A gene editing analysis pipeline” methods section to help to understand differences between the barcodes that can be used: “In case of working with pooled samples, the demultiplexing of the samples has to be done before running CRISPR-A analysis pipeline using the proper software in function of the sequencing used platform. The resulting FASTQ files are the main input of the pipeline.”

Then, SQK-LSK109 from Oxford Nanopore is followed through the steps specified in methods: “The Custom PCR UMI (with SQK-LSK109), version CPU_9107_v109_revA_09Oct2020 (Nanopore Protocol) was followed from UMI tagging step to the late PCR and clean-up step.”

Finally, we want to highlight that, as can be seen in methods as well as in discussion, UMIs are used to group sequences that have been amplified from the same genome and not to identify different samples: “Precision has been enhanced in CRISPR-A through three different approaches. [...] We also removed indels in noisy positions when the consensus of clusterized sequences by UMI are used after filtering by UBS.” As well as in results (Fig. 5C).

• Text in Figure 5 is hard to read.

We have increased the letter size of Figure 5.

• They should test the software based on the ground truth data

We have added a human classified dataset to do the final benchmarking. And we can see that for all examined samples CRISPR-A has an accuracy higher than 0.9. As has been shown in the figure with manual curated data, CRISPR-A shows good results in noisy samples using the empiric noise removal algorithm, without need of filtering by edition windows.

• The alignment algorithm is not the best one, I think minimap2 would be better for general purpose (at least it work better for ONT).

As can be seen in figure 2A, minimap is one of the alignment methods that gives better results for the aim of the pipeline. In addition, we have tuned the parameters (Figure 2B) for a better detection of CRISPR-based long deletions, which can be more difficult to report in a single open gap of the alignment.

• The minimum configuration for installation was not mentioned (for their Docker/next flow pipeline).

Proper documentation to indicate the configuration requirements for installation has been added to the readme.md of the repository·

• Fig 2: why do they use PC4/PC1?

Principal Component Analysis is used to reduce the number of dimensions in a dataset and help to understand the effect of the explainable variables, detect trends or samples that are labeled in incorrect groups, simplify data visualization… Even PC4 explains less variability than PC2 or PC3, this helps us to understand and better decipher the effect of the 4 different analyzed parameters even if the differences are not big. We have decided to include as a supplementary figure other PCs to show these.

• There are still typos and unclear statements thorughout the whole manuscript.

One more drawback is that the software seems to only support single FASTQ uploading (or we cannot see the option to add more FASTQ).

In the case of paired-end reads instead of single-end reads, in the web application, these can be selected at the beginning answering the question “How should we analyze your reads? Type of Analysis: Single-end Reads; Paired-end Reads”. In the case of the pipeline, now it is explained in the documentation how to mark if the data is paired-end or single-end. It has to be indicated in “input” and “r2file” configuration variables.

In the case of multiple samples, and for that reason multiple FASTQ files, there is the button to add more samples in the web application. In the pipeline, multiple samples can be analyzed in a single run by putting all together in a folder and indicating it with variable “input”.

Since usually people analyze more than one clone at the time (we usually analyze 96 clones together) this would mean that I have to upload manually each one of them.

All files can be added in the same folder and analyzed in a single run using the nextflow pipeline. Web application has a limit of ten samples that can be added clicking the button “Add more”.

Also, the software (the webserver, the docker does not work) works with Illumina data in our hands but not with ONT.

This should be clarified in the manuscript.

If a fastq is uploaded to CRISPR-A, the analysis can be done even if we haven't specifically optimized the tool for long reads sequencing platforms. We have checked the performance of CRISPR-A with CRISPRnano nanopore testing dataset and we have succeeded in the analysis. See results here: https://synbio.upf.edu/crispr-a/RUNS/tmp_1118819937/.

Summary of the results:

Sample

CRISPRnano

CRISPR-A

'rep_3_test_800'

42.60 % (-1del); 12.72 % (-10del)

71% (-1del);

16% (-10del)

– 36 (logo)

'rep_3_test_400'

37.50 % (-1del); 15.63 % (-10 del)

65% (-1del);

28% (-10del)

– 38 (logo)

'rep_1_test_200'

39.29 % (-1del); 8.33 % (-17del)

10del; 17del; 1del

'rep_1_test_400'

80.11 % (-17 del)

del17; del20; del18; del16;del 16

'rep_0_test_400'

80.11% (-17 del)

del17; del20; del 18; del16; del16

'rep_0_test_200'

71.91% (-17 del)

del17; del18

As we can see from these exemple, CRISPR-A reports all indels in general without classifying them as edits or noise. Since nanopore data has a high number of indels as sequencing errors the percentages of CRISPR-A are not accurate. Eventhat, CRISPR-A reports more diverse outcomes, which are probably edits, than CRISPRnano.

Therefore, we have added the following text in results:

“Even single-molecule sequencing (eg. PacBio, Nanopore..) can be analyzed by CRISPR-A, targeted sequencing by synthesis data is required for precise quantification.”

Reviewer #1 (Significance (Required)):

As I mentioned above I think this could be a useful software for those people that are screening genome editing cells. Since CRISPR is widely used i assume that the audience is broad.

There are many other software that perform similarly to CRISPR-A but it seems that this software adds few more things and seems to be more precise. It is hard to understand if everything the author claims is accurate since it requires a lot of testing and time and the reviewing time is of just two weeks. But 1) I have no reason to doubt the authors and 2) the software works

Broad audience (people using CRISPR)

Genetics, Genome Engineering, software development (we develop a very similar software), genetic compensation, stem cell biology

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

Summary:

CRISPR-Analytics, abbreviated as CRISPR-A, is a web application implementing a tool for analyzing editing experiments. The tool can analyze various experiment types - single cleavage experiments, base editing, prime editing, and HDR. The required data for the analysis consists of NGS raw data or simulated data, in fastq, protospacer sequence and cut site. Amplicon sequence is also needed in cases where the amplified genome is absent from the genome reference list. The tool pipeline is implemented in NextFlow and has an interactive web application for visualizing the results of the analysis, including embedding the results into an IGV browser.

The authors developed a gene editing simulation mechanism that enables the user to assess an experiment design and to predict expected outcomes. Simulated data was generated by SimGE over primary T-cells. The parameters and distributions were also fitted for 3 cell lines to make it more generalized (Hek293, K562, and HCT116). The process simulated CRISPR-CAS9 activity and the resulting insertions, deletions, and substitutions. The simulation results are then compared to the experimental results. The authors report the Jensen-Shannon (JS) divergence between the results. The exact distributions that served as input to the JS are not well defined in the manuscript (see below).

To clarify the used distributions in the JS divergence calculation, we have changed the following piece of text in section “Simulations evaluation” of methods:

“ Afterward, we tested the performance on the fifth fold, generating the simulated sequences with the same target and gRNA as the samples that belong to the fifth fold, in order to calculate the distance between these. The final validation, with the mean parameters of the different training interactions, was performed on a testing data set that was not used in the training. Validation was done with samples that had never taken place in the training process. Jensen distance is used to compare the characterization of real samples and simulated samples since this is the explored distance that differentiates better replicates among samples. In order to obtain the different distributions, the T cell data, including 1.521 unique cut sites, was split into different datasets based on the different classes: deletions, insertions and substitutions. For each of these classes, giving as input the datasets with only that class, we obtained the distribution for size and then for position of indels. The same was done for the other three cell lines: K562, HEK293 and HCT116, which included 96 unique cut sites, with three replicates each. The whole datasets (with 1521 and 96 unique cut sites) were split into five-folds (4 for training and one for test) and validation, in order to train and validate the simulator. Using the parameters obtained during the training-test iterations (the average value of the 5 iterations), we generate simulated sequences with the same target and gRNA as the samples that are assigned to the test subset to calculate the Jensen-Shannon (JS) divergence between the simulated and real samples of that subset. Finally, the same was performed for validation. The input for the distance calculations were the generated simulated subset and its real equivalent (same target and gRNA) distributions of the classes. ”

The authors also report an investigation of different alignment approaches and how they may affect the resulting characterization of editing activity.

The authors examine three different approaches to increase what they call "edit quantification accuracy" (aka, in a different place - "precise allele counts determination" - what is this???): (1) spike-in controls (2) UMI's and (3) using mock to denoise the results. See below for our comments about these approaches.

Moreover, the authors developed an empirical model to reduce noise in the detection of editing activity. This is done by using mock (control), and by normalization and alignment of reads with indels, with the notion and observation that indels that are far from the cut site tend to classify as noise.

The authors then perform a comparison between 6 different tools, in the context of determining and quantifying editing activities. One important comparison approach uses manually curated data. However - the description of how this dataset was created is far from being sufficiently clear. The comparison is also performed for HDR experiment type, which can be compared only to 2 other tools.

We have changed alleles by editing outcomes in the title section “Three different approaches to increase precise editing outcomes counts determination” trying to be more clear.

There is already a section in methods “Manual curation of 30 edited samples” explaining how the manual curation was done.

We see the potential contribution aspects of the paper to be the following:

1. NextFlow pipeline implementation is an important engineering contribution. Same is true for the interactive web application
2. The option to simulate an experiment to assess it is a nice feature and can help experiment design
3. Identification of amplicons when not provided as input
4. CRISPR-A seeks substitutions along the entire amplicon sequence and is less dependent on the quantification window and on the putative cutsite
5. Analysis of the difference, in edit activity, comparing different cell lines
6. CRISPR-A supports the use of UMIs
7. Interesting sequence pattern insights - like "...found certain patterns associated with low diversity outcomes: free thymine or adenine at the 3' nucleotide upstream of the cut site that leads to insertions of the same nucleotide, a free cytosine at the same place that leads to its loss, and strong micro-homology patterns that lead to long deletions " We further comment on the soundness of these contributions in our comments below and on their significance in our comments related to the general potential significance of the paper.

Major comments:

• Upon attempting to run an analysis from the web interface (https://synbio.upf.edu/crispr-a) and using: fastq of Tx and mock (control), the human genome and the gRNA sequence provided as input for the protospacer field, our run was not successful. In fact the site crashed with no interpretable error message from CRISPR-A. We have improved the error handling together with the explanations in the help page, where you will find a video. Hopefully these improvements will avoid unexpected crashings.

• Moreover, there should be more clear context. There is no information regarding the type of experiments that can be analyzed with the tool. We figure it is multiplex PCR and NGS but can the tool also be used for GUIDESeq, Capture, CircleSeq etc.? Experiments that could be analyzed are specified in Results: “CRISPR-A analyzes a great variety of experiments with minimal input. Single cleavage experiments, base editing (BE), prime editing (PE), predicted off-target sites or homology directed repair (HDR) can be analyzed without the need of specifying the experimental approach.” We have also specified this in the nextflow pipeline documentation as well as in the web application help page.

• No off target analysis. Only on-target The accuracy of the tool allows checking if edits in predicted off-target sites are produced, this being an off-target analysis with some restrictions, since just variants of the predicted off-target sites are assessed. Translocations or other structural off-targets will not be detected by CRISPR-A since the input data analyzed by this tool are demultiplexed amplicon or targeted sequencing samples.

• No translocations and long/complex deletions The source of used data as input does not allow us to do this. There are other tools like CRISPECTOR available for this kind of analysis. We have added this to supplementary table 1.

• We view the use of a mock experiment as control as a must for any sound attempt to measure edit activity. This is even more so when off-target events need to be assessed (any rigorous application of GE, certainly any application aiming for clinical or crop engineering purposes). We therefore think that all investigation of other approaches should be put in this context. We agree with the necessity of using negative controls to assess editing. For that reason we have included the possibility of using mocks in the quantification. In addition, there are few tools that include this functionality.

• It's a nice feature to have simulated data, however, it is not a good approach to rely on it. As can be seen in the manuscript we highlight the support that simulations can give without pretending to substitute experimental data by just simulated data. Simulated data has been useful in the development and benchmarking of CRISPR-A, but we are aware of the limitations of simulations. Here some examples from the manuscripts explaining how we have used or can be used simulated data:

“Analytical tools, and simulations are needed to help in the experimental design.”

“simulations to help in design or benchmarking”

“We developed CRISPR-A, a gene editing analyzer that can provide simulations to assess experimental design and outcomes prediction.”

“Gene editing simulations obtained with SimGE were used to develop the edits calling algorithm as well as for benchmarking CRISPR-A with other tools that have similar applications.”

Even simulated data has been useful for the development and benchmarking of CRISPR-A, we have also used real data and human validated data.

• In p7 the authors indicate the implementation of three approaches to improve quantification. They should be clear as to the fact that many other tools and experimental protocols are also using these approaches. for example, ampliCan, CRipresso2 and CRISPECTOR all take into account a mock experiment run in parallel to the treatment. Even in page 7 (results) we don’t mention the other tools that also use mocks for noise correction, we detail this information in Supplementary Table 1. CRISPResso2 was not included since they can run mocks in parallel but only to compare results qualitatively, i.e. there is not noise reduction in their pipeline. It has been added to the table.

• Figure1: ○ The figure certainly provides what seems to be a positive indication of the simulations approach being close to measured results. Much more details are needed, however, to fully understand the results.

We have added more details.

○ Squema = scheme ??

We have changed the word “schema” by diagram.

○ What was the clustering approach?

As is said in the caption of Figure 1 the clustering is hierarchical: “hierarchical clustering of real samples and their simulations from validation data set.” And we have added that “The clustering distance used is the JS divergence between the two subsets.”

○ What is the input to the JS calculation? What is the dimension of the distributions compared? These details need to be precisely provided.

The distribution has two dimensions, sizes and counts or positions and counts.

As said before, to clarify the used distributions in the JS divergence calculation, we have changed the following piece of text in section “Simulations evaluation” of methods:

“ Afterward, we tested the performance on the fifth fold, generating the simulated sequences with the same target and gRNA as the samples that belong to the fifth fold, in order to calculate the distance between these. The final validation, with the mean parameters of the different training interactions, was performed on a testing data set that was not used in the training. Validation was done with samples that had never taken place in the training process. Jensen distance is used to compare the characterization of real samples and simulated samples since this is the explored distance that differentiates better replicates among samples. In order to obtain the different distributions, the T cell data, including 1.521 unique cut sites, was split into different datasets based on the different classes: deletions, insertions and substitutions. For each of these classes, giving as input the datasets with only that class, we obtained the distribution for size and then for position of indels. The same was done for the other three cell lines: K562, HEK293 and HCT116, which included 96 unique cut sites, with three replicates each. The whole datasets (with 1521 and 96 unique cut sites) were split into five-folds (4 for training and one for test) and validation, in order to train and validate the simulator. Using the parameters obtained during the training-test iterations (the average value of the 5 iterations), we generate simulated sequences with the same target and gRNA as the samples that are assigned to the test subset to calculate the Jensen-Shannon (JS) divergence between the simulated and real samples of that subset. Finally, the same was performed for validation. The input for the distance calculations were the generated simulated subset and its real equivalent (same target and gRNA) distributions of the classes. ”

○ What clustering/aggregation approach did the authors use here (average dist, min dist, dist of centers?)

Hierarchical clustering.

○ 5 pairs were selected out of how many? Call that number K.

We have 100 samples in the validation set. Following the suggestion of indicating the total number of samples in the testing set, we have added this information to the figure caption.

○ What does the order of the samples in 1C mean? Is 98_real closer to 22_sim than to 98_sim? If so then state it. If not - what is the meaning of the order? Furthermore - how often, over K choose 2 pairs does this mis-matching occur for the CRISPR-A simulator??

Exactly, it is a hierarchical clustering, where samples are sorted by JS divergence. It was already stated in Results: “In addition, on top of comparing the distance between the experimental sample and the simulated, we have included two experimental samples, SRR7737722 and SRR7737698, which are replicates. These two and their simulated samples show a low distance between them and a higher distance with other samples.” As well as in Figure 1 caption: “For instance, SRR7737722 and SRR7737698, which cluster together, are the real sample and its simulated sample for two replicates.” Then, since these samples are replicates, its simulations will come from the same input and is expectable to find low distance between these two real samples as well as between both of them and their simulation. We have stated it in the discussion.

• "From the characterized data we obtained the probability distribution of each class" (page 3) - How is this done? how many guides? how many replicates? what is class? where do you elabore regarding it? how you obtain the distributions? More details of the methods need to be provided. Added in methods.

• The 96 samples used for development here - where are they taken from? This should be indicated in the first time these samples are mentioned. Namely - bottom of P6 Added: “The 96 samples, from these cell lines, are obtained from a public dataset BioProject PRJNA326019.”

• CRISPECTOR is not mentioned in the comparison in the section: "CRISPR-A effectively calls indels in simulated and edited samples" (Table S2). Is there a specific reason for having left it out? CRISPECTOR, as well as ampliCan, is not in Table S2, since in this table is shown detailed data from Figure 2. CRISPECTOR is compared with CRISPR-A in figure 5, where the different approaches to enhance precision, like using a negative control, are explored.

• In the section "Improved discovery and characterization of template-based alleles or objective modifications" - part of the analysis was made over simulated data and then over real data. The authors state "it is difficult to explain the origin of these differences...". Thus, needs to be investigated in more detail ... :) (P5) Moreover - the performance over real data is, at the end of the day, the more interesting one for comparison purposes. We have added this sample to the human validated dataset to understand better what was happening in this case and the results and pertinent discussion have been added in the manuscript: “CRISPResso2 is detecting a 2% more of reads classified as WT. These 2% correspond with the percentage classified as indels by CRISPR-A. In total, the percentage difference between CRISPResso2 and CRISPR-A template-based class is 0.6%, higher in CRISPR-A. CRISPR-A percentage is closer to the ground truth data than CRISPResso2.”

• We found no explanation of "spike-in"/"spike experimental data" across the entire article. There is some general language about lengths but the scheme is still totally unclear. We have indicated in methods section when we were talking about the spike-in controls.

• Description of the 96 gRNAs? Is this data from REF26? If so - where do you state this? If so - how do the methods described herein avoid the unique characteristics of the data of REF26? We have added the reference: “The 96 samples, from these cell lines, are obtained from a public dataset BioProject PRJNA326019.” In addition, there are other sources of data, simulations and now even human validated data.

• "distance between the percentage of microhomology mediated end-joining deletions of samples with the same target was calculated and the mean of all these distances was used to reduce the information of the 96 different targets to a single one." (P6) What is the exact calculation used? which distance? How was clustering performed? What is the connection for gene expression? The used distance was euclidean distance and the clustering was performed using hierarchical clustering. We have added this information to the manuscript. Regarding the connection of gene expression, we are exploring the correlation of two phenotypes: the gene expression of the proteins differentially related with NHEJ and MMEJ pathways, and the gene editing landscape (indel patterns that are related with MMEJ and those that are more prone to be generated with NHEJ). We have tried to improve this explanation in the manuscript.

• "we have fitted a linear model to transform the indels count depending on its difference in relation to the reference amplicon" (P7) - needs more explanation. Is this part of the pipeline? We have explained better how we have fitted the linear model in methods: “A linear regression model was fitted to obtain the parameters of Equation 1 using spike-in controls experimental data (original count, observed count and size of the change in the synthetic molecules). We have used the lm function from R. Parameter m in Equation 1 is equivalent to the obtained coefficient estimate of x which was 0.156 and n is the intercept (n=10). ”.

The model is optionally used as part of the pipeline as explained at the end of section “CRISPR-A gene editing analysis pipeline” to correct amplification biases due to differences in amplicon size. Then, what is part of the pipeline is the use of this model to make the transformation of counts from the observed counts to the predicted original counts. This is done with Equation 1 and can be found in the pipeline (VC_parser-cigar.R).

• What is it "...manually curated data set"? (page 8) This is explained in “Manual curation of 30 edited samples” in methods.

• Section "CRISPR-A empiric model removes more noise than other approaches" - with what data were the comparisons performed? Moreover, how were the comparison criteria selected (efficiency and sensitivity)? The literature already used several approaches to compare data analysis tools for editing experiments. See for example ampliCan, Crispresso (1 and 2) and CRISPECTOR. Maybe the authors should follow similar lines. The data used in this comparison comes from the reference 26:“26. van Overbeek, M. et al. DNA Repair Profiling Reveals Nonrandom Outcomes at Cas9-Mediated Breaks. Mol. Cell 63, 633–646 (2016).” We have added it to the manuscript.

The values of efficiency and sensitivity were not used directly for the comparison. We wanted to firstly evaluate our own algorithm. For that we obtained the values of efficiency and sensitivity for the previous mentioned dataset. These values were chosen to firstly have an idea of firstly, how much noise the algorithm is able to detect, and secondly, how much of it is able to be reduced after the Tx vs M process. That established a framework of comparison in which we can then compare directly the reported percentage of edition of the different tools.

Regarding the approaches used to compare data analysis tools for editing experiments, we are going to explain why we haven’t followed similar lines or how we have now included it:

In the case of ampliCan, the comparison that they do is with a synthetic dataset with introduced errors:

"synthetic benchmarking previously used to assess these tools (Lindsay et al. 2016), in which experiments were contaminated with simulated off-target reads that resemble the real on-target reads but have a mismatch rate of 30% per base pair".

In CRISPResso2, they benchmarked the efficiency against an inhouse dataset but this dataset is not published. Finally, for the benchmarking of CRISPECTOR, a manual curated dataset is used as a standard: "Assessment of such classification requires the use of a gold standard dataset of validated editing rates. In this analysis, we define the validated percent indels as the value determined through a detailed human investigation of the individual raw alignment results". In this sense, we have added a human validated dataset to do something similar to complement the analysis that we had already done.

In the end, we consider that simulated or synthetic datasets, as those used by ampliCan or CRISPResso2, does not capture the complete landscape of confounding events that can be detrimental to the analysis results. Similar limitations are found in the use of a gold standard dataset of validated editing rates, since the amount of reads or samples that can be validated by humans is not big since it is time consuming. In addition, humans can also make errors and have biases. Eventhogh, we have found very valuable talking into consideration adding a human validated dataset to complete our exploration.

• In the section "CRISPR-A empiric model removes more noise than other approaches" the authors state, incorrectly, that CRISPECTOR only reports the percentage of editing activity per site (there is much more information reported in the HTML report, including the type of edit event detected - deletion, of various lengths, insertions, substitutions etc). (P8) We thank the reviewer for the observation, as indeed the state is incorrect. What we wanted to express is that with CRISPECTOR we cannot trace individually each of the called indels, as any sort of excel or file with this content is given in the output. Therefore we cannot investigate which events have been corrected. To be precise in our statement we changed this sentence to the following:

“CRISPECTOR, although providing extensive information on the statistics and information about the indels, is not possible to track the reads along their pipeline, thus we cannot know which have been corrected and which have not.”

• Section "CRISPR-A noise subtraction pipeline" describes a pretty naive method for noise subtraction (P12). Should be rigorously compared, for Tx vs Mock experiments, to CRISPECTOR and to CRISPResso2. In the section "CRISPR-A empiric model removes more noise than other approaches", we perform an exhaustive comparison with a dataset that contains 288 Mock Files vs 864 Tx files. This can be better appreciated in the, now included, figure Sup. 13A. CRISPResso2 was intentionally left out since their pipeline does not use a model to reduce noise but other approaches like reducing the quantification window.

• "recalculated using a size bias correction model based on spike-in controls empiric data.." (P14). Where is the formula? The formula comes from Equation 1. Now it is correctly referenced.

• Section "Noise subtraction comparison with ampliCan and CRISPECTOR" - fake mock was generated for comparison. We consider the avoidance of a Mock control in experiments designed to measure editing activity to not be best practice. It is OK to support this approach in CRISPR-A. However - the comparison to tools that predominantly work using a Mock control (including ampliCan and CRISPECTOR) should be done with actual Mock. Not with fake Mock .... (P15) We understand the claims of the reviewer for this point as the use of a “fake” mock may not be the best practice for general comparisons. Nevertheless here what we wanted to compare is the difference in the edition percentages using mock and not using it. Since to make a run for on-target data CRISPECTOR requires a mock, the only way to replicate the conditions of “no mock” was to use a synthetic file with the same characteristics of the treated files in terms of depth, but with no edition/noise events to avoid any correction outside this framework. The other run was made with the 288 real Mocks. This was a solution ad Hoc for CRISPECTOR, with ampliCan we used only real mock since they allow to make runs without a mock for on-target.

We changed the word fake for synthetic in the Noise subtraction comparison with ampliCan and CRISPECTOR section:

“As for CRISPECTOR, since it requires a mock file to perform on-target analysis, synthetic mock files were generated”.

Minor comments:

• "Also, most of these tools lack important functionalities like reference identification, clustering, or noise subtraction" - bold part incorrect for CRISPECTOR, although it is not aiming only for CRISPECTOR In supplementary table 1, it is already elucidated which are the functionalities that each tool has. We have also added more context to that statement to highlight the differences between different tools:

“Even not all of them have the same missing functionalities, as can be seen in the Supplementary table 1, CRISPR-A is the only tool that can identifies the amplicon reference from in a reference genome, correct errors through UMI clustering and sequence consensus, correct quantification errors due to differences in amplicon size, and includes interactive plots and a genome browser representation of the alignment.”

• "Same parameters and probability distributions were fitted for three other cell lines: Hek293, K562, and HCT11626, to make SimGE more generalizable and increase its applicability" (page 3) - how was fitted? It was fitted in the same way as the t-cell samples as specified in methods. We have detailed more methods explaining how SimGE is built.

• What is the "nature of modification"? (P5) We have changed nature by type for a better understanding.

• In the section "CRISPR-A effectively calls indels in simulated and edited samples" (P5) towards the end, the authors write that the CRISPR-A algorithm did not give good results for a few examples. They then state that this was corrected and then yielded good results. There is no explanation of what correction was done, if it was implemented in the code and how to avoid/detect it in further cases. The problem was that the used reference sequence was too short. There is no modification in CRISPR-A code, we have just used the whole amplicon reference sequence obtained with the amplicon reference identification functionality of CRISPR-A. We have tried to explain it better in the manuscript: “Once the reference sequence is corrected used is the one corresponding to the whole reference amplicon, obtained with CRISPR-A amplicon sequence discovery function, CRISPR-A shows a perfect edition profile”

• Cell culture, transfection, and electroporation - explanation only for HEK293, what about the others? (P15) We already had explained it for HEK293 and for C2C12, that are the experiments done by use. In the case of the analysis of the three cell lines and 96 targets we reference the source of the data as this data was not produced in our lab.

• Typos and unclear wording: ○ "obtention" (P8) → changed by obtaining

○ "mico" >> micro (P 7,10) → changed

○ "Squema" >> scheme (Fig.1) → changed

○ "decombuled" (P10) → changed by separated

○ "empiric" >> empirical (P8 and other places) → changed

○ "Delins" (P14) → this is not a typo, it is used to indicate that a deletion and insertion has take place (http://varnomen.hgvs.org/recommendations/DNA/variant/delins/)

○ "performancer" (P9) → Change to performance

○ Change word across all article - "edition" to "editing" → changed. In the case of edition windows it has been changed by quantification windows.

○ "...has enough precision to find" (P6) not related to "results" section → We have moved to discussion.

• Comments on figures: ○ Fig. 2C:

■ No CRISPECTOR in the analysis

It is not included because for on-target analysis this tool requires a mock control sample. For this reason, it is compared in Figure 5D, where samples using negative controls are compared, and in Figure 5E where all tools and their different analysis options are compared.

■ It is simulated data only

Yes, it is. Comparison with real data is done in Figure 2D and 2E. And now we also have added a ground truth data in our comparisons obtained from human validation of the classification of more than 3,000 different reads.

■ It is not violin plot as mentioned in the description

It is a violin plot, but in general there is not much dispersion of the data points making the density curves flat.

○ Fig 3A - Is it significant? Yes, it is. We have added this information in the caption of the figure.

○ Fig. 4:

■ A

• Each row/column is a vector of 96 guides? No, as it is said in the caption of the figure, it is the “mean between the distances calculated for each of the 96 different targets.”

• How is the replicate number decided? Is it a different experiment by date? What is separating between experiments? Rep numbers? All this information should be found in the referenced paper from which this dataset comes from as already referenced.

■ B - Differential expression:

We have realized that the caption was not correct, missing the explanations for Fig. 4B and all the following ones moved to a previous letter.

• How? did you measure RNA? It is already stated in methods that RNAseq data was obtained from SRA database and the analysis was done using nf-core/rnaseq pipeline: “RNAseq differential expression analysis of samples from BioProject PRJNA208620 and PRJNA304717 was performed using nf-core/rnaseq pipeline⁵².”

• Is the observed data in the figure sufficiently strong in terms of P-value? Yes, at is it is highlighted in the plot with ** and ***. We have also added the p-value in the cation of the figure.

• Where is the third cell-line? As mentioned in the text, we have just chosen the cell lines that show us higher differences in the the percentage of MMEJ: “HCT116 than in K562, which are the cell lines with the major and minor ratios of MMEJ compared with NHEJ, respectively”.

○ Fig.13 - There is no A and B as mentioned in the text

We thank the reviewer for the observation as we mistakenly uploaded the wrong figure. We corrected it.

Reviewer #2 (Significance (Required)):

We repeat the aspects of contribution, as listed in the first part of the review, and comment about significance:

• NextFlow pipeline implementation is an important engineering contribution. Same is true for the interactive web application

  Significant engineering contribution. Nonetheless, we were not able to run the analysis. So - needs to be checked.

Hopefully now that the documentation is properly added to the repository it will be easier to run analysis.

• The option to simulate an experiment to assess it is a nice feature and can help experiment design

  An important methodology contribution

• Identification of amplicons when not provided as input

  Not important in the context of multiplex PCR and NGS measurement assays, as amplicons will be known. Not clear what other contexts the authors were aiming at.

It is useful to save time, no need to look for the sequence of each amplicon and add it as input. Also, it can help to detect unspecific amplification, since all amplicons of the same genome can be retrieved from the discovery amplicon process. In addition, we have already found one example where this avoids getting incorrect results: “Once the reference sequence used is the one corresponding to the whole reference amplicon, obtained with CRISPR-A amplicon sequence discovery function, CRISPR-A shows a perfect edition profile”. We have added this to the discussion of the manuscript.

• CRISPR-A seeks substitutions along the entire amplicon sequence and is less dependent on the quantification window and on the putative cutsite

  Importance/significance needs to be demonstrated

In figure 3 are shown the results of template-based and substitutions detection. CRISPR-A is a versatile and agnostic tool for gene editing analysis. This means that it can be prepared for the analysis of gene editing of future tools, since the cut site or other elements of experiment design are not required. In addition, it has been shown that when a mock is used its performance is comparable to filtering by edition windows, avoiding the loss of edits when the cut site is slided.

• Analysis of the difference, in edit activity, comparing different cell lines

  Significant contribution. However - the methods need to be much better explained and the results better described in order for this to be useful to the community.

We have made an effort to try to be more clear in the description of the results.

• CRISPR-A supports the use of UMIs

  Mildly significant technical contribution. However - only addresses on-target. Also addressing off-target would have been significant.

The use of UMIs is something that has never been done before in this context. Sequencing biases are not taken into account and editing percentages are reported as observed. Being able to differentiate between different molecules at the beginning of the amplification sequence, allows a higher precision avoiding under or overestimation of each of the species in a bulk of cells.

In the case of off-targets, can be for sure done using sequencing the predicted off-target sites. In addition, there are other methods, like GuideSeq that can be used to discover off-targets, but this kind of data is out of the scope of CRISPR-A. Even that, we are aware of the importance of being able to analyse off-targets when in a context of a broad analysis platform and we will take these into consideration when participating in the building of crisprseq pipeline from nf-core.

• Interesting sequence pattern insights - like "...found certain patterns associated with low diversity outcomes: free thymine or adenine at the 3' nucleotide upstream of the cut site that leads to insertions of the same nucleotide, a free cytosine at the same place that leads to its loss, and strong micro-homology patterns that lead to long deletions "

  As stated - interesting.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

CRISPR-Analytics, abbreviated as CRISPR-A, is a web application implementing a tool for analyzing editing experiments. The tool can analyze various experiment types - single cleavage experiments, base editing, prime editing, and HDR. The required data for the analysis consists of NGS raw data or simulated data, in fastq, protospacer sequence and cut site. Amplicon sequence is also needed in cases where the amplified genome is absent from the genome reference list. The tool pipeline is implemented in NextFlow and has an interactive web application for visualizing the results of the analysis, including embedding the results into an IGV browser. The authors developed a gene editing simulation mechanism that enables the user to assess an experiment design and to predict expected outcomes. Simulated data was generated by SimGE over primary T-cells. The parameters and distributions were also fitted for 3 cell lines to make it more generalized (Hek293, K562, and HCT116). The process simulated CRISPR-CAS9 activity and the resulting insertions, deletions, and substitutions. The simulation results are then compared to the experimental results. The authors report the Jensen-Shannon (JS) divergence between the results. The exact distributions that served as input to the JS are not well defined in the manuscript (see below).

The authors also report an investigation of different alignment approaches and how they may affect the resulting characterization of editing activity. The authors examine three different approaches to increase what they call "edit quantification accuracy" (aka, in a different place - "precise allele counts determination" - what is this???): (1) spike-in controls (2) UMI's and (3) using mock to denoise the results. See below for our comments about these approaches. Moreover, the authors developed an empirical model to reduce noise in the detection of editing activity. This is done by using mock (control), and by normalization and alignment of reads with indels, with the notion and observation that indels that are far from the cut site tend to classify as noise. The authors then perform a comparison between 6 different tools, in the context of determining and quantifying editing activities. One important comparison approach uses manually curated data. However - the description of how this dataset was created is far from being sufficiently clear. The comparison is also performed for HDR experiment type, which can be compared only to 2 other tools. We see the potential contribution aspects of the paper to be the following:

1. NextFlow pipeline implementation is an important engineering contribution. Same is true for the interactive web application
2. The option to simulate an experiment to assess it is a nice feature and can help experiment design
3. Identification of amplicons when not provided as input
4. CRISPR-A seeks substitutions along the entire amplicon sequence and is less dependent on the quantification window and on the putative cutsite
5. Analysis of the difference, in edit activity, comparing different cell lines
6. CRISPR-A supports the use of UMIs
7. Interesting sequence pattern insights - like "...found certain patterns associated with low diversity outcomes: free thymine or adenine at the 3' nucleotide upstream of the cut site that leads to insertions of the same nucleotide, a free cytosine at the same place that leads to its loss, and strong micro-homology patterns that lead to long deletions " We further comment on the soundness of these contributions in our comments below and on their significance in our comments related to the general potential significance of the paper.

Major comments:

• Upon attempting to run an analysis from the web interface (https://synbio.upf.edu/crispr-a) and using: fastq of Tx and mock (control), the human genome and the gRNA sequence provided as input for the protospacer field, our run was not successful. In fact the site crashed with no interpretable error message from CRISPR-A.
• Moreover, there should be more clear context. There is no information regarding the type of experiments that can be analyzed with the tool. We figure it is multiplex PCR and NGS but can the tool also be used for GUIDESeq, Capture, CircleSeq etc.?
• No off target analysis. Only on-target
• No translocations and long/complex deletions
• We view the use of a mock experiment as control as a must for any sound attempt to measure edit activity. This is even more so when off-target events need to be assessed (any rigorous application of GE, certainly any application aiming for clinical or crop engineering purposes). We therefore think that all investigation of other approaches should be put in this context.
• It's a nice feature to have simulated data, however, it is not a good approach to rely on it.
• In p7 the authors indicate the implementation of three approaches to improve quantification. They should be clear as to the fact that many other tools and experimental protocols are also using these approaches. for example, ampliCan, CRipresso2 and CRISPECTOR all take into account a mock experiment run in parallel to the treatment.
• Figure1:
  • The figure certainly provides what seems to be a positive indication of the simulations approach being close to measured results. Much more details are needed, however, to fully understand the results.
  • Squema = scheme ??
  • What was the clustering approach?
  • What is the input to the JS calculation? What is the dimension of the distributions compared? These details need to be precisely provided.
  • What clustering/aggregation approach did the authors use here (average dist, min dist, dist of centers?)
  • 5 pairs were selected out of how many? Call that number K.
  • What does the order of the samples in 1C mean? Is 98_real closer to 22_sim than to 98_sim? If so then state it. If not - what is the meaning of the order? Furthermore - how often, over K choose 2 pairs does this mis-matching occur for the CRISPR-A simulator??
• "From the characterized data we obtained the probability distribution of each class" (page 3) - How is this done? how many guides? how many replicates? what is class? where do you elabore regarding it? how you obtain the distributions? More details of the methods need to be provided.
• The 96 samples used for development here - where are they taken from? This should be indicated in the first time these samples are mentioned. Namely - bottom of P6
• CRISPECTOR is not mentioned in the comparison in the section: "CRISPR-A effectively calls indels in simulated and edited samples" (Table S2). Is there a specific reason for having left it out?
• In the section "Improved discovery and characterization of template-based alleles or objective modifications" - part of the analysis was made over simulated data and then over real data. The authors state "it is difficult to explain the origin of these differences...". Thus, needs to be investigated in more detail ... :) (P5) Moreover - the performance over real data is, at the end of the day, the more interesting one for comparison purposes.
• We found no explanation of "spike-in"/"spike experimental data" across the entire article. There is some general language about lengths but the scheme is still totally unclear.
• Description of the 96 gRNAs? Is this data from REF26? If so - where do you state this? If so - how do the methods described herein avoid the unique characteristics of the data of REF26?
• "distance between the percentage of microhomology mediated end-joining deletions of samples with the same target was calculated and the mean of all these distances was used to reduce the information of the 96 different targets to a single one." (P6) What is the exact calculation used? which distance? How was clustering performed? What is the connection for gene expression?
• "we have fitted a linear model to transform the indels count depending on its difference in relation to the reference amplicon" (P7) - needs more explanation. Is this part of the pipeline?
• What is it "...manually curated data set"? (page 8)
• Section "CRISPR-A empiric model removes more noise than other approaches" - with what data were the comparisons performed? Moreover, how were the comparison criteria selected (efficiency and sensitivity)? The literature already used several approaches to compare data analysis tools for editing experiments. See for example ampliCan, Crispresso (1 and 2) and CRISPECTOR. Maybe the authors should follow similar lines.
• In the section "CRISPR-A empiric model removes more noise than other approaches" the authors state, incorrectly, that CRISPECTOR only reports the percentage of editing activity per site (there is much more information reported in the HTML report, including the type of edit event detected - deletion, of various lengths, insertions, substitutions etc). (P8)
• Section "CRISPR-A noise subtraction pipeline" describes a pretty naive method for noise subtraction (P12). Should be rigorously compared, for Tx vs Mock experiments, to CRISPECTOR and to CRISPResso2.
• "recalculated using a size bias correction model based on spike-in controls empiric data.." (P14). Where is the formula?
• Section "Noise subtraction comparison with ampliCan and CRISPECTOR" - fake mock was generated for comparison. We consider the avoidance of a Mock control in experiments designed to measure editing activity to not be best practice. It is OK to support this approach in CRISPR-A. However - the comparison to tools that predominantly work using a Mock control (including ampliCan and CRISPECTOR) should be done with actual Mock. Not with fake Mock .... (P15)

Minor comments:

• "Also, most of these tools lack important functionalities like reference identification, clustering, or noise subtraction" - bold part incorrect for CRISPECTOR, although it is not aiming only for CRISPECTOR
• "Same parameters and probability distributions were fitted for three other cell lines: Hek293, K562, and HCT11626, to make SimGE more generalizable and increase its applicability" (page 3) - how was fitted?
• What is the "nature of modification"? (P5)
• In the section "CRISPR-A effectively calls indels in simulated and edited samples" (P5) towards the end, the authors write that the CRISPR-A algorithm did not give good results for a few examples. They then state that this was corrected and then yielded good results. There is no explanation of what correction was done, if it was implemented in the code and how to avoid/detect it in further cases.
• Cell culture, transfection, and electroporation - explanation only for HEK293, what about the others? (P15)
• Typos and unclear wording:
  • "obtention" (P8)
  • "mico" >> micro (P 7,10)
  • "Squema" >> scheme (Fig.1)
  • "decombuled" (P10)
  • "empiric" >> empirical (P8 and other places)
  • "Delins" (P14)
  • "performancer" (P9)
  • Change word across all article - "edition" to "editing"
  • "...has enough precision to find" (P6) not related to "results" section
• Comments on figures:
  • Fig. 2C:
• No CRISPECTOR in the analysis
• It is simulated data only
• It is not violin plot as mentioned in the description
  • Fig 3A - Is it significant?
  • Fig. 4:
• A
• Each row/column is a vector of 96 guides?
• How is the replicate number decided? Is it a different experiment by date? What is separating between experiments? Rep numbers?
• B - Differential expression:
• How? did you measure RNA?
• Is the observed data in the figure sufficiently strong in terms of P-value?
• Where is the third cell-line?
  • Fig.13 - There is no A and B as mentioned in the text

Significance

We repeat the aspects of contribution, as listed in the first part of the review, and comment about significance:

• NextFlow pipeline implementation is an important engineering contribution. Same is true for the interactive web application
  • Significant engineering contribution. Nonetheless, we were not able to run the analysis. So - needs to be checked.
• The option to simulate an experiment to assess it is a nice feature and can help experiment design
  • An important methodology contribution
• Identification of amplicons when not provided as input
  • Not important in the context of multiplex PCR and NGS measurement assays, as amplicons will be known. Not clear what other contexts the authors were aiming at.
• CRISPR-A seeks substitutions along the entire amplicon sequence and is less dependent on the quantification window and on the putative cutsite
  • Importance/significance needs to be demonstrated
• Analysis of the difference, in edit activity, comparing different cell lines
  • Significant contribution. However - the methods need to be much better explained and the results better described in order for this to be useful to the community.
• CRISPR-A supports the use of UMIs
  • Mildly significant technical contribution. However - only addresses on-target. Also addressing off-target would have been significant.
• Interesting sequence pattern insights - like "...found certain patterns associated with low diversity outcomes: free thymine or adenine at the 3' nucleotide upstream of the cut site that leads to insertions of the same nucleotide, a free cytosine at the same place that leads to its loss, and strong micro-homology patterns that lead to long deletions "
  • As stated - interesting.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

Marta Sanvicente-García et al and colleague developed a comprehensive and versatile genome editing web application tool and a nextflow pipeline to give support to gene editing experimental design and analysis.

The manuscript is well written and all data are clearly shown. While I did not tested extensively, the software seems to work well and I have no reason to doubt the authors' claims. I usually prefer ready to use web applications like outknocker, they are in general easier to use for rookies (it would be good if the author could cite it, since it is very well implemented) but the nextflow implementation is anyway well suited.

Few minor points:

Reagording the methods to assess whether the genome editing is working or not, i would definetely include High Resolution Melt Analysis, which is by far the fastest and probably more sensitive amongst the others.

Another point that would important to taclke is that often these pipelines do nto define the system they are working with (eg diploid, aploid vs etc). This will change the number of reads needed ato unambigously call the genotypes detected and to perform the downstream analysis (the CRISPRnano authors mentioned this point).

I am also wondering whether the name CRISPR-A is appropriate since someone could confuse it with CRISPRa.

Referees cross-commenting

Reviewer 2 made an excellent work and raised important concerns about the software they need to be addressed carefully.

In the meantime we had more time to test the software and can confirm some of the findings of Reviewer 1:

1. We spent hours running (unsuccessfully) CRISPR A on Nextflow. The software does not seem to run properly.
2. No manual or instruction can be found on both their repositories (https://bitbucket.org/synbiolab/crispr-a_nextflow/ https://bitbucket.org/synbiolab/crispr-a_figures/)

Few more points to be considered

• UMI clustering is not proper terminology. Barcode multiplexing/demultiplexing (SQK-LSK109 from Oxford Nanopore).
• Text in Figure 5 is hard to read.
• They should test the software based on the ground truth data
• The alignment algorithm is not the best one, I think minimap2 would be better for general purpose (at least it work better for ONT).
• The minimum configuration for installation was not mentioned (for their Docker/next flow pipeline).
• Fig 2: why do they use PC4/PC1?
• There are still typos and unclear statements thorughout the whole manuscript.

One more drawback is that the software seems to only support single FASTQ uploading (or we cannot see the option to add more FASTQ). Since usually people analyze more than one clone at the time (we usually analyze 96 clones together) this would mean that I have to upload manually each one of them.

Also, the software (the webserver, the docker does not work) works with Illumina data in our hands but not with ONT. This should be clarified in the manuscript.

Significance

As I mentioned above I think this could be a useful software for those people that are screening genome editing cells. Since CRISPR is widely used i assume that the audience is broad.

There are many other software that perform similarly to CRISPR-A but it seems that this software adds few more things and seems to be more precise. It is hard to understand if everything the author claims is accurate since it requires a lot of testing and time and the reviewing time is of just two weeks. But 1) I have no reason to doubt the authors and 2) the software works

Broad audience (people using CRISPR)

Genetics, Genome Engineering, software development (we develop a very similar software), genetic compensation, stem cell biology

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

Learn more at Review Commons

---

Reply to the reviewers

Dear Editor,

Please find below our detailed responses (in black font) to the Reviewer's comments (in blue). In addition, and to the request of Reviewer #1, we added a PDF file called “Reply to the reviewers MS data” that shows MS/MS and quantification information of representative peptides which were selected based on their (different) caspase/control abundance ratios. We thank the reviewers for their time and valuable comments.

NOTE: our original reply includes several tables and graphs that were not incorporated into our reply shown below

Reviewer #1

Page 4 - In contrast to the hindrance of N-terminal amine ionization by Nt-acetyl groups concluded by the authors, previous studies reported an improved MS-scoring if α-amino-acetylated (tryptic) peptides by the higher numbers of b and y fragment ions observed as compared to α-amino-free (tryptic) peptides (e.g. (Staes et al., 2008)). It is rather the lack of any N-/C-terminal charged residue in case of Lys-N type N-termini which makes LATE less suitable for studying N-terminal protein acetylation.

We thank the reviewer for this comment. In the HYTANE and LATE workflows, only peptides with modified N-termini (by dimethylation or acetylation) are observed after negative selection, hence we argue that the important comparison here is between Nt-acetylated peptides and Nt-dimethylated peptides with (as in HYTANE) or without basic residue (as in LATE). While we are aware of the study by Staes et al 2008 (PMID: 18318009), we do not believe it contradicts our claim as it discusses the difference between Nt-acetylated peptides and peptides with free N-termini.

As we indicated in the manuscript (page 5 in the last sentence of 1st paragraph), we observed less overall peptide identifications in LATE, which was expected due the lack of basic C-term residue. The reduction of identification was more pronounced for Nt-acetylated peptides. However, this still does not exclude LATE as a useful tool for identification of such peptides.

Of note, the overall fragmentation coverage we obtained by LATE and HYTANE for Nt-acetylated and Nt-dimethylated peptides was similar. See the figure below.

Hence, following Cho et al 2016 (PMID: 26889926), we suggest that the difference in ionization of Nt-dimethylated peptides vs Nt-acetylated peptides is the more dominant factor in peptide identifications.

Figure 1:relative Ion coverage for modified peptides in LATE and HYTANE

Page 4 - Besides indication the retained N-termini with high relative caspase-3/control abundance ratio's as putative caspase-3 proteolytic products, also indicate that unique peptides were retained, as many such singletons were reported in previous (caspase-focussed) degradomics studies making use of differential proteomics (e.g. (Van Damme et al., 2005)). Therefore the cut-off ratio of 2 rather seems unsubstantiated, unless the cellular proteomes of so-called control cells were affected by caspase activation. As such, showing some representative MS-spectra of neo-N-termini would be informative.

We thank the reviewer for this comment. We agree that caspase-3 cleavage generates many singletons. This is indeed what we observed in the in vitro experiment as shown in Figure 2B by the long straight lines at Log2(caspase-3/control) >10. We also add here histograms of the obtained ratios that we hope will make this clearer. We chose a cut-off of 2 due to the basal activity of proteases (including caspase-3) as we did not add caspase-3 inhibitors to the cell lysate. In addition, peptides derived from the putative caspase-3 cleavages in the in vitro experiment were required to be identified only in the caspase-3-treated samples (i.e. to appear only with the heavy labeling). Minor changes to Figure 3 legend have been introduced accordingly. As can be seen in the table below, with a cut-off ratio of 2 (Log2=1) and selection of cleavage sites after D or E we identified >98% of the cleavage sites that were identified only in the caspase-3 treated samples (column text in blue). This rate did not change when the cut-off was set to 8 (Log2=3). Therefore, we have chosen to maintain our selection criteria and cut-off ratio as used before for both experiments.

Figure 2: Histograms of Log2(Caspase/control) ratios indicating the large number of singleton peptides (marked with arrows)

Table 1: In vitro experiment selection ratio

Method

Cutoff

Time

Sites

Sites identified only caspase-3 treated samples

% of caspase-treated only sites (singleton)

Sites D/E with light

Sites after D/E no light

% of singleton

LATE

Log2=1

18H

936

906

96.8%

798

786

98.5%

LATE

Log2=2

18H

884

866

98.0%

767

759

99.0%

LATE

Log2=3

18H

819

810

98.9%

722

716

99.2%

HYTANE

Log2=1

18H

1186

1159

97.7%

1037

1032

99.5%

HYTANE

Log2=2

18H

1128

1110

98.4%

998

993

99.5%

HYTANE

Log2=3

18H

1035

1025

99.0%

924

919

99.5%

LATE

Log2=1

6H

755

732

97.0%

656

645

98.3%

LATE

Log2=2

6H

711

700

98.5%

630

623

98.9%

LATE

Log2=3

6H

671

666

99.3%

601

597

99.3%

HYTANE

Log2=1

6H

1028

988

96.1%

899

890

99.0%

HYTANE

Log2=2

6H

955

931

97.5%

851

844

99.2%

HYTANE

Log2=3

6H

882

871

98.8%

795

791

99.5%

LATE

Log2=1

1H

445

423

95.1%

380

372

97.9%

LATE

Log2=2

1H

411

402

97.8%

361

355

98.3%

LATE

Log2=3

1H

386

380

98.4%

344

338

98.3%

HYTANE

Log2=1

1H

593

559

94.3%

513

506

98.6%

HYTANE

Log2=2

1H

544

532

97.8%

488

482

98.8%

HYTANE

Log2=3

1H

508

498

98.0%

461

455

98.7%

In the cell-based experiments of caspase-3, we induced apoptosis on both cell types (over-expressing caspase-3 and the control). Therefore, in this case, as the reviewer has also mentioned, a cut-off of 2 is appropriate because the control cells are also affected by caspase activation. Following the reviewer’s request we have added (in a separate PDF file) several representative MS/MS spectra of neo-N-term peptides and their corresponding quantification data.

Page 4 - replace 'without labelling of lysine residues (epsilon-amines)' to 'without notable labelling of lysine residues (epsilon-amines)', as residual labelling of lysine side-chains was observed. Also in case of the latter, do note that reduced MS-ionization potential might impact labelling efficiency calculation, and chromatographic detection of labelling efficiency should be considered to conclusify this finding.

We thank the reviewer for this comment. We have changed the sentence as requested (Page 4 marked in red). Regarding the labeling efficiency calculations, it is true that ionization potential might affect them. We used a common way to test this aspect (see e.g. Hurtado Silva et al 2019 (PMID: 30934878)) and we are not aware of any reduction in ionization potential following lysine dimethylation. Although we did not study this aspect thoroughly, we frequently observe the opposite: that dimethylation improves MS detections.

Page 6 - The experimental setup comparing caspase-3 overexpressing and ABT-199 induced versus ABT-199 induced cells will be highly biased as it will not be able to detect efficient caspase-3 cleavages (Plasman et al., 2011), as such cleavage events are complete and thus do not require any additional overexpressed capase-3. I see this as an important flaw and the authors should demonstrate that the list also includes efficient caspase-3 cleavages.

We thank the reviewer for highlighting this important aspect. We agree that with our setup, we can miss some efficient cleavages of caspases-3. We acknowledged this caveat in the original text (page 6), but chose to perform our experiments this way in order to highlight cleavages that are affected by caspase-3 expression. To address the reviewer’s comment we have added new experiment and data on caspase cleavages that occur following ABT-199 treatment in HCT116 cells without overexpression of caspase-3. The focus of this experiment was on the relatively short time points following the ABT-199 treatment when no cell death is observed based on XTT assay (see Supplement Figure 6B). This experiment was used to prove that neo-Nt-acetylation of NACA is an early event in apoptosis (Figure 5 E-F page 12). We also used this experiment as an indication of the appearance of efficient cleavages. As can be seen from Supplement Table S10, if we consider all 3 time points of the ABT-199 treatment, we quantified 106 cleavages with free neo-Nt that were cleavages after D and were identified only in the ABT-treated samples. We refer to such cleavages, which appeared prior to noticeable cell death, as "efficient cleavages". Out of these efficient cleavages, 82 were also identified and quantified in the cell-based experiment with overexpression of caspase-3. Twenty efficient cleavages show a high ratio (≥2) in both experiments. Fifty six efficient cleavages had a high ratio in the new experiment and ratio below 2 in the cell-based experiment with overexpression of caspase-3. This supports our original claim regarding efficient cleavages and addresses the reviewer’s concern regarding our ability to identify efficient caspase-3 cleavages with the experimental setup of HCT116 cells overexpressing caspase-3.

Page 12 - The setup doesn't permit ORF N-terminal stability per se, rather the cleavage susceptibly of N-termini holding (a) putative caspase-3 cleavage site(s). Please adjust accordingly. Again since the setup might have missed efficient cleavages, the assessment might be biased.

Thanks for the comment. As requested, the word “stability” has been deleted. As discussed above, we demonstrate that our setup does allow the identification of efficient cleavages and hence our basis for believing that the assessment is not biased. Please also refer to our reply to the next comment.

The claim that Nt-acetylation is protective for caspase-3 cleavage should be validated by monitoring cleavage efficiency of an Nt-acetylated versus an Nt-free variant (e.g. by introducing a Pro residue at AA position 2, or comparing cleavage efficiencies in corresponding NAT knockdown versus control cells) of an identified caspase substrate (i.e. a substrate holding a caspase-3 cleavage site in its N-terminal sequence) versus its Nt-free counterpart

Thanks for raising this point. The reviewer's suggestions have some caveats: a mutation at a protein’s N-terminus in order to generate an Nt-free variant can alter its stability or function and NAT knockdown might have a profound biological impact on the cells. Therefore we chose a different way to study this aspect by selecting from our data ORF N-terminal peptides that were identified with both free N-termini and acetylated N-termini (i.e. the same peptide was identified in some PSMs as acetylated and in other as dimethylated). We managed to find 136 ORF N-terminal peptides that were quantified in both forms, and out of these, 122 contained Asp or Glu residues (the putative caspase cleavage motifs). We added the comparison of the abundance ratios of these peptides in Figure 4C (see also below). It shows a remarkable difference between the groups when the Nt-acetylated peptides ratios did not change as a result of caspase-3 overexpression while the peptides with free Nt were identified mostly in the control cells (negative Log2(caspase-3/control)). Comparison of the 14 ORF Nt-peptides that do not have Glu or Asp in their sequence shows no difference (see below).

Figure 3: Abundance ratio distributions of the ORF Nt peptides identified with both Nt-acetylated and free Nt in HCT116 cells overexpressing caspase-3 and in the control. A. Comparison of peptides that contain putative caspase cleavage in their sequence (D or E) B. comparison of peptides without putative caspase cleavage

These results provide additional support for the notion of the protective or shielding effect of Nt-acetylation against caspase-3 cleavage.

Page 12 - Since post-translational Nt-acetylation of neo-N-termini could be reproduced in vitro in the non-dialyzed sample, enzymatic over chemical Nt-acetylation should be demonstrated (e.g. by the use of a (bisubstrate) NAT inhibitor).

We think this is an interesting idea for future work. Yet, in our opinion, the fact that only very few neo-Nt-acetylated peptides were affected in vitro and that a similar trend of few selected neo-Nt-acetylation targets was shown in the cell-based experiments indicates that this process is enzymatic and not chemical in nature.

Other concerns:

Abstract - The abstracts holds complex/incorrect sentence constructions (e.g. simply indicate 'Protein N-termini', '... undergo ... processing by proteases' (currently: 'not be processed by proteases').

Thanks for pointing this out. We have changed the abstract accordingly.

Abstract - 'To expand the coverage of the N-terminome' only applies when this is used in conjunction with other negative enrichment strategies as by itself, LATE doesn't intrinsically provide a better coverage of the N-terminome (this is also noted at page 2).

We thank the reviewer for pointing this out. We have changed the abstract accordingly.

Change 'that cannot be identified by other methods' to 'that cannot be identified by other negative selection methods'

Thanks for pointing this out. We believe that our description here is appropriate as we explicitly state “some of which cannot be identified by other methods”.

Page 1 - Suggestion to change 'Proteases are typically described as degradative enzymes' to 'Proteases used to be described as degradative enzymes'

Changed as suggested.

Page 1 - Not really correct how written; 'N-terminomics methods highlight the N-terminal fragment of every protein (N-terminome)'

Changed as suggested.

Page 2 - Positive selection techniques .... Enrichment of unblocked (or Nt-free) N-termini

We are not sure what the reviewer had in mind here but have added the text in the brackets

Page 2 - Besides altering charge, Nt-acetylation also alters hydrophobicity ...

Changed as suggested.

Page 2 - remove 'to better chart'

Changed as suggested.

Page 2 etc. - Do note that caspase-3 can potentially activate downstream caspases in vitro

Following this comment, we have added a sentence on Page 5 with this reservation

Page 3 - functional crosstalk between proteolysis and neo-Nt-acetylation has already been demonstrated in the case of co-translational acting methionine aminopeptidases and chloroplast N-terminal acetyltransferases. Adjust accordingly.

We thank the reviewer for highlighting this aspect, although we used the term “neo-Nt-acetylation” which we used to mark that this is not the common (co-translational) acetylation. To assure that this is more clear we have added the words “post-translational” to better define the novelty of our findings.

Page 3 - when discussing the identification of ORF N-termini, note that some of the strategies of which note when used to enrich for in vivo blocked N-termini, can also be used without blocking/labelling of Lys residues, and thus trypsin will also result in Lys-ending peptides. This is important to consider in this context.

Following the reviewer's remark we have changed the sentence so it now states: “Many of these N-terminomics methods……”

Page 3 - remove the following sentence part; '... or run individually which can be useful for quantifying naturally modified N-termini.', since also a differential/labelled proteomics setup enables such assessment. Related to this, the authors should comment on the observation that much fewer (i.e. less than 40%) Nt-acetylated N-termini were identified by LATE as compared to HYTANE. How is this reflected in the number of PSMs? Probably these difference are further intensified when considering PSMs.

We have changed the sentence as suggested.

Regarding the reduction of Nt-acetylation, we thank the reviewer for this question as it led us to find typos in the numbers in Figure 1E which are now corrected. These typos did not change the overall observation that with LATE we identify fewer Nt-acetylated peptides than Nt-free (dimethylated) peptides. As the reviewer anticipated (see below), the reduction in LATE-based “contribution” to the identification of Nt-acetylated peptides as opposed to the identification of dimethylated peptides, is pronounced when considering PSMs but this is not much different than the peptide-based data. Therefore, we prefer to keep the current presentation of Figure 1E.

Figure 4: Comparison of HACAT cells N-terminal peptides identification with LATE and HYTANE when considering peptide sequences and PSMs. Peptides identified with both methods are in green and those that are unique to one method are in blue. Shared peptides were determined based on the sequence of the first 7 amino acids of the identified peptides. A. comparison for peptides with dimethylated N-terminal (free Nt) B. comparison for Nt-acetylated peptides.

Page 6 - Informative to indicate how many of the in silico predicted putative DEVD P4-P1 cleavages were actually present in the list of 2049 putative cleavages identified.

In our dataset, we identified 17 cleavages after DEVD motif. 11 were identified only with HYTANE, 3 were identified by both methods, and 3 more were identified only with LATE. Of note, it seems that in large-scale proteomic studies of apoptosis, the number of caspase cleavages after DEVD motif is quite low. For example, in the CASBAH database (PMID: 17273173__) __there are 10 reports of such cleavage out of 391 reported sites, and in DegraBase (PMID: 23264352) that combined many different apoptotic experiments there are 64 reported DEVD sites out of a total of 6896 P1-Asp sites.

Page 6 - Unclear if any of the of 2049 putative cleavages, included non-canonical P1 cleavages besides the P1 Asp and Glu cleavages identified.

These are 2049 putative cleavage sites with P1 Asp or Glu. We have changed the text to make it clearer.

Page 6 - Were the 'regular' cells mock transfected?

No. The control cells used for the cell-based experiments were the non-transfected cells from the same culture of HCT166. We chose this option to guarantee that exactly the same cells that were grown in the same dish went through the same FACS sorting as a control.

Page 6 -Important to note that an ORF can have multiple N-termini besides neo-N-termini (e.g. in the case of alternative translation initiation)

Thanks for the great point. We have added an indication if the neo-N-termini site has been reported as an alternative translation initiation site to all of the results of the cell-based experiments (Supplementary Tables S4, S5, S6, S9). We also changed the Figures and text accordingly. Our analysis of reported/unreported neo-N-temini is based on the TopFind database which includes information about alternative translation initiation sites from TISdb. Of note, since our focus is on caspase cleavages and we further select putative cleavages based on D/E in P1 and fold change, out of 973 peptides that we reported as putative caspase cleavage (Table S6) only one is in the vicinity of an alternative initiation site.

Page 6 - The authors should be more careful with generalization when comparing LATE and HYTANE (and other degradomics approaches) as in this study LATE was only applied for the identification of caspase-3 neo-N-termini, which by its extended substrate specificity might hold specific features enabling the preferred detection by one technique over the other. Also note that as compared to less recent studies, evidently the MS instrument used is a key factor in the increase in cleavages reported in the current study.

It is conceivable that caspase cleavage may differ from other proteases and thus theoretically work better with LATE, but we fail to see why this would also be the case for other N-terminomics method (like TAILS, Subtiligase, CoFRADIC, ChaFRADIC etc). We showed that LATE provides additional ORF Nt peptides identifications and demonstrated its effectiveness in E. coli (Supplement Figure S2) also, which has a proteome with a different amino acid composition to the human proteome. Furthermore, using LATE in the cell-based experiment led to the identification of many neo-Nt-peptides that do not match caspase cleavage patterns (as indicated for both HYATNE and LATE in Figures 3E and 3F). We reviewed the text again, and believe that we have used a fair description of the results especially when we compared them to previous studies.

Page 9 - The authors should provide some info/supporting statistics in the text regarding the new putative substrates showing GO-enrichments (compared to which control?) similar to previously reported caspase-3 substrates.

The results of the GO enrichment analysis are presented in Fig. S8 and details about how the test was performed are provided in the Materials & Methods. In the revised version, we are including the numerical data that include results of the statistical tests per GO term as Table S12. The enrichment analysis was performed with respect to the whole human proteome.

Page 11 - Indicate that the 11 neo-N-terminal peptides of which note are the neo-Nt-peptides matching (signal peptide) cleavages indicated in the Uniprot database. Were any corresponding di-methylated neo-N-termini of these cleavages identified? In case of the 'other' proteolytic cleavages of which note, refer to these as not-annotated in UniProt.

We thank the reviewer for pointing this out. We have added an indication that this analysis is based on UniProt annotations. Yes, all of the reported 11 neo-Nt-Acet peptides shown in Figure 4 were also found as neo-Nt-DiMet peptides.

Page 11 - post-translational Nt-acetylation is abundant in plant and the responsible NAT has been identified, please reference these studies as well.

We thank the reviewer for pointing this out regarding page 11. A relevant reference has been added in Page 11. In the discussion, we already referenced Nt-acetylation in plants in the discussion as well (see page 14).

Page 12 - Define 'undoubtedly dependent on caspase-3 cleavage'

We thank the reviewer for pointing this out. The word ‘undoubtedly’ has been deleted.

Page 14 - The NAA30 discussion is not really relevant for the discussion of the post-translational Nt-acetylation of mitochondrial neo-N-termini.

We thank the reviewer for pointing this out. This sentence has been deleted.

Viewing the harsh in vitro caspase-3 cleavage condition used, namely 1 µg caspase 3 over 20 µg protein, the P1 specificities of all identified neo-N-termini should clearly be shown.

The P1 specificities of all neo-N-termini found in the in vitro experiment are listed in the supplementary tables S2 and S3. For the reviewer’s convenience, we are providing the table with the P1 specificities below:

Since acetylation of serine and threonine residues are reported forms of post-translational modification, and many so-called past-translational Nt-acetylated neo-N-termini harbour such AA residues in their N-terminal sequence, b-ion coverage for these neo-N-termini should be provided/inspected.

We are not sure that we understand this comment. O-Acetylation of amino acids refers to their side chain. Since we are using Di-methylation labeling in both HYTANE and LATE, if we have a peptide with O-acetylated Ser or Thr at its first position, it is possible to distinguish it from the same peptide with Nt-acetylation by MS1 as illustrated in the following table for a hypothetical peptide SAAANPELKR (mass is MH+1)

Regardless we include in the manuscript MS/MS spectra of NACA Neo-Nt-acetylated peptide by HYTANE and LATE

Reviewer #2

Major suggestions:

• The LATE method relies on digestion with LysN. Can the authors comment on the digestion efficiency of the samples where the LATE workflow was applied?

The LysN digestion details that we used were based on vendor (Promega) instructions combined with details from the Nature Protocol paper by Giansanti et al 2016 (PMID: 27123950__)__ that describes optimized digestion protocol for LysN. We tested LysN efficiency in terms of the identification of missed cleavage and found that it performed very well with a missed-cleavage rate of

• The authors state that the number of peptides with acetylated N-termini was lower compared with HYTANE. Yet, the Nt-acetylation can occur co-translationally in approximately 85% of human proteins.

Did the authors consider optimizing the method (e.g. by fractionating the sample) for better identification of such peptides?

We thank the reviewer for this important comment. We are certain that it is possible to improve the output of LATE by fractionation and/or optimization by changes to the LC gradient as it is well established for most, if not all, bottom-up proteomics methods. In this work, we concentrated more on the proof of concept of the methodology and hence chose to work without fractionation. We performed one attempt to optimize the LC gradient but found that the results were not significantly different, and we thus used the same LC-MS methods that have been optimized for trypsin.

Regarding the reduced identification of Nt-acetylated peptides, as we state in the manuscript following Cho et al 2016 (PMID: 26889926), we believe that this is mainly due to the reduced ionization efficiency of Nt-acetylated peptides compared to Nt-dimethylated peptides which is more pronounced when a C-terminal positive charge is missing (due to LysN digestion).

Also, were the results of the study compared with searches done using other proteomic pipelines (e.g. FragPipe)?

Unfortunately, when we started this project, MS-Fragger did not support LysN as the digesting enzyme. At the time TPP also provided better visualization and quantification support than FragPipe. Recently, we found that MSFragger is faster while providing similar identifications but we are not convinced of the quantification output via FragPipe. In addition, we performed comparisons of Comet to X!Tandem and while the searches took longer than with Comet, the total number of IDs did not improve significantly.

Can the authors provide details on the settings used for searches done in COMET, especially for the samples treated with LysN?

The settings are provided in Table S10 in the supplementary information (Page 14 of the PDF file).

"Fractions containing relatively pure caspase-3 were pooled together and dialyzed against 20 mM HEPES 7.5, and 80 mM NaCl. Aliquots of the protein were stored at -80{degree sign}C"

o What exactly is meant by 'relatively pure'?

We apologize for the inaccurate description. The relevant text has been updated (Page 17) and now explains that this was based on Coommasse stain SDS-PAGE.

Minor suggestions:

• Please check the link for the Github as this reviewer could not open it.

We thank the reviewer for pointing this out. We corrected the link. In any case, the relevant scripts can be found here: https://github.com/OKLAB2016

• Please correct the spelling.

The manuscript was proofread.

Comments regarding figures:

• Figure 2:

o All figures comparing LATE and HYTANE utilize color green for LATE. Yet, in figure 2G, HYTANE is depicted in green-like color. Please consider staying consistent with the color scheme.

We thank the reviewer for this comment. Done as suggested.

Reviewer #2 (Significance (Required)):

Significance:

• The LATE method provides an excellent way to study proteases in vitro or in cell-based experiments. It enables deep investigation of N-terminome based on a simple and cost-effective workflow that utilizes digestion with LysN followed by chemical derivatization of α-amines. This approach allows for the identification of N-terminal peptides that may escape detection by other N-terminomics methods. With LATE, proteases' cleavage sites that might not so far be reporter due to technical limitations, can be studied and characterized. Hence, LATE is a useful addition to the N-terminomic toolbox.

We thank the reviewer for the positive comments and general assessment of LATE.

Reviewer #3

In this manuscript, Hanna et al. report LATE, an N terminomics method similar to N-TAILS and HYTANE, with modifications that enhance or change coverages of the N-terminal proteome in proteomics datasets. LATE relies on selective N-terminal modification of protease-treated, LysN digested samples, enabling internal peptides to be depleted based on the presence of the unblocked lysine epsilon amine. Using LATE in comparison with HYTANE, the authors identified a large number of both known and unknown caspase-3 cleavage sites, both in vitro and in vivo. Because LATE enables identification of both proteolytic neo-N termini and natively blocked N termini such as those that are acetylated, the authors were able to discover a number of post-translationally acetylated proteolytic neo-N termini. This finding points to potential functional cross talk between apoptotic proteolysis and Nt-acetylation. Overall, this is a very nice manuscript that adds a valuable new tool to the N-terminal proteomics toolbox. However, the manuscript could be improved by addressing the following questions and comments.

We thank the reviewer for this assessment.

1. One of the benchmark points used to describe the need for a new technology such as LATE is the idea that there are 134 putative caspase-3 substrates in the human proteome, of which only about half can be identified based on ArgC cleavages. However, the 134 substrates seem to include only those that have the exact canonical DEVD motif. Many more substrates than this are already known for caspase-3. For example, >900 caspase-3 substrates were identified by Araya et al. alone. It might make more sense to apply a position-specific scoring matric to the human proteome to predict a maximum number of possible caspase-2 cleavage sites and how many would be expected to be identified using other technologies. Otherwise, please provide a rationale for why these 134 putative caspase-3 sites are representative.

The reviewer is correct. Indeed, most of the identified caspase-3 cleavage are not exact matches to the DEVD motif. We used the DEVD as an example to illustrate the added value of using lysine-based digestion together with ArgC. We obtained a similar trend with some variations when we tested the feasibility of the identification of the human ORF Nt-peptides, E. coli ORF Nt-peptides and more. We are quite confident that any prediction will show a relatively similar distribution. To demonstrate this, we show here the relative contribution of each method for the identification of any peptide that begins after Asp in the human proteome.

While the distributions are not identical, they are very similar, and the specific additions from LATE (LysN) are between 20% to 22% out of the total and it can help to expand the coverage by 42% to 45%.

It is definitely plausible and have been previously demonstrated that selective N-terminal demethylation can be achieved under the right reaction conditions, and I do not doubt that it has been achieved here. However, I do not understand how the authors were able to conclude that alpha-amines are blocked with 95% efficiency and lysines are blocked at

This is a very good point. The reviewer is correct and indeed we don’t have a way to establish if the dimethylation is on the side chain amine of lysine or on its N-terminal amine. A partial support for our claim is from labeling experiments that we (and others) conducted with tryptic and LysC peptides that clearly demonstrate that under the specified labeling conditions, 95% of the N-terminal amines are labeled and not the lysine side chain amines. However, at the end of the day, this does not change the outcome of LATE.

Related to the above comment, Table S10 seems to indicate that MS/MS data from LATE were searched with dimethylation as a fixed modification at the N terminus. Were LATE samples searched with different parameters to generate Figure 1C? Are the dimethylated Ks identified mostly from missed cleavages and therefore not at the N terminus?

We thank the reviewer for pointing this out. The search parameters used for the generation of Figure 1C have been added to Table S10. The reviewer is correct, the few dimethylated Ks identified in the search used for Figure 1C are mostly from missed cleavages.

For both the in vitro and in vivo experiments, how many of the new caspase-3 cleavage sites occurred in proteins that were not previously known to be caspase substrates?

In the in vitro experiments, we identified cleavages of 372 proteins that were not reported as caspase-3 substrates based on the databases we used as references. A line specifying this number was also added to text on page 7. In the cell-based experiment, we identified putative caspase-3 cleavages of 67 proteins that were not reported so far as caspase-3 substrates. This information has been added to the main text on page 10. We have added columns indicating the known/unreported protein substrates to Tables S2, S3, S4, S5, and S6.

For the experiment in cells, can the authors explain the rationale for comparing cells in which apoptosis is induced with ABT-199 to ABT-199-treated cells with caspase-3 overexpression? What is the advantage over comparing ABT-199 treated cells to untreated cells

Great question. An N-terminomics study of “common” apoptosis would lead mainly to the identification of effector caspases (caspase-3 and -7) substrates. Our aim was to focus mostly on the caspase-3 cleavages that occur in the cell during apoptosis. In choosing this gain-of-function approach we were motivated by the idea that it couldprovide new insights that would otherwise go undetected when using knockout or other loss-of-function approaches. The advantage of this system over comparing ABT-199 treated to non-treated cells (which we have now added as well) is that it can enhance the identification of caspase-3 specific cleavages.

Can the authors discuss the timescale of cell death in ABT-199 treated cells vs. ABT-199 treated caspase-3-overexpressing cells. Ideally, data showing cell viability over time (e.g. Cell Titer Glo or MTT assays) would be presented, but if the authors could at least describe whether apoptosis is accelerated in the caspase-3 overexpressing cells, it would be helpful.

Great suggestion. Following the reviewer’s suggestion we have characterized the effect of caspase-3 overexpression of the cells by XTT assay, and indeed caspase-3 overexpressing cells do show accelerated cell-death in response to ABT199 compared to non-transfected cells. These results are now presented as Supplement Figure S6B and are mentioned in the results section.

The authors say that in their experimental design, they expect to see no difference between ABT-199 only and ABT-199/caspase-3 overexpression for substrates that are cleaved efficiently by endogenous caspases. If the new caspase-3 substrates are not cleaved efficiently by endogenous caspase-3, this seems to call into question their physiological relevance. Can the authors explain more thoroughly how these new substrates fit into the apoptotic program?

We thank the reviewer for raising this issue. We are aware that our original cell-based experimental design may have some limitations, yet we chose this gain-of-function setup in order to identify caspase-3 substrates in a cell-based system. We believe that this setup does allow identification of substrates that are efficiently cleaved by endogenous caspase-3, such as cleavage and acetylation of NACA at Ser34 (and neo-Nt-acetylation after caspase-3 cleavage in general). To study the physiological relevance of the neo-Nt-acetylation, we have added to the revised manuscript a time-course N-terminomics characterization of early apoptosis events conducted in HCT116 cells (without caspase-3 overexpression). The results of these experiments are now shown in Figure 5C and also in the Supplementary Table

The authors convincingly show that cleaved NACA is a neo-substrate for Nt-acetylation, suggesting functional crosstalk between proteolysis and acetylation. However, it is not clear if this acetylation event has a functional consequence, so it seems inaccurate to say at the top of page 3 that "This is the first demonstration of functional crosstalk between neo-Nt-acetylation and proteolytic pathways."

The author is correct. We changed the text accordingly.

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, Hanna et al. report LATE, an N terminomics method similar to N-TAILS and HYTANE, with modifications that enhance or change coverages of the N-terminal proteome in proteomics datasets. LATE relies on selective N-terminal modification of protease-treated, LysN digested samples, enabling internal peptides to be depleted based on the presence of the unblocked lysine epsilon amine. Using LATE in comparison with HYTANE, the authors identified a large number of both known and unknown caspase-3 cleavage sites, both in vitro and in vivo. Because LATE enables identification of both proteolytic neo-N termini and natively blocked N termini such as those that are acetylated, the authors were able to discover a number of post-translationally acetylated proteolytic neo-N termini. This finding points to potential functional cross talk between apoptotic proteolysis and Nt-acetylation. Overall, this is a very nice manuscript that adds a valuable new tool to the N-terminal proteomics toolbox. However, the manuscript could be improved by addressing the following questions and comments.

1. One of the benchmark points used to describe the need for a new technology such as LATE is the idea that there are 134 putative caspase-3 substrates in the human proteome, of which only about half can be identified based on ArgC cleavages. However, the 134 substrates seem to include only those that have the exact canonical DEVD motif. Many more substrates than this are already known for caspase-3. For example, >900 caspase-3 substrates were identified by Araya et al. alone. It might make more sense to apply a position-specific scoring matric to the human proteome to predict a maximum number of possible caspase-2 cleavage sites and how many would be expected to be identified using other technologies. Otherwise, please provide a rationale for why these 134 putative caspase-3 sites are representative.
2. It is definitely plausible and have been previously demonstrated that selective N-terminal demethylation can be achieved under the right reaction conditions, and I do not doubt that it has been achieved here. However, I do not understand how the authors were able to conclude that alpha-amines are blocked with 95% efficiency and lysines are blocked at <5%. This claim seems to be based on PSMs for each type of modification. However, in a LysN digested sample, we would expect the vast majority of peptides to begin with K and the vast majority of Ks to be found at the N terminus of a peptide. In this situation, how is it possible to distinguish whether demethylation has occurred on the alpha-amine or the epsilon-amine? With N-terminal K, all of the MS2 fragments containing the N-terminal a-amine would also contain the lysine epsilon-amine. The m/z values for the y-ions, b-ions, and a-ions containing this residue would be the same. I may be misunderstanding, so it would be helpful if the authors could explain how they are able to distinguish these.
3. Related to the above comment, Table S10 seems to indicate that MS/MS data from LATE were searched with dimethylation as a fixed modification at the N terminus. Were LATE samples searched with different parameters to generate Figure 1C? Are the dimethylated Ks identified mostly from missed cleavages and therefore not at the N terminus?
4. For both the in vitro and in vivo experiments, how many of the new caspase-3 cleavage sites occurred in proteins that were not previously known to be caspase substrates?
5. For the experiment in cells, can the authors explain the rationale for comparing cells in which apoptosis is induced with ABT-199 to ABT-199-treated cells with caspase-3 overexpression? What is the advantage over comparing ABT-199 treated cells to untreated cells
6. Can the authors discuss the timescale of cell death in ABT-199 treated cells vs. ABT-199 treated caspase-3-overexpressing cells. Ideally, data showing cell viability over time (e.g. Cell Titer Glo or MTT assays) would be presented, but if the authors could at least describe whether apoptosis is accelerated in the caspase-3 overexpressing cells, it would be helpful.
7. The authors say that in their experimental design, they expect to see no difference between ABT-199 only and ABT-199/caspase-3 overexpression for substrates that are cleaved efficiently by endogenous caspases. If the new caspase-3 substrates are not cleaved efficiently by endogenous caspase-3, this seems to call into question their physiological relevance. Can the authors explain more thoroughly how these new substrates fit into the apoptotic program?
8. The authors convincingly show that cleaved NACA is a neo-substrate for Nt-acetylation, suggesting functional crosstalk between proteolysis and acetylation. However, it is not clear if this acetylation event has a functional consequence, so it seems inaccurate to say at the top of page 3 that "This is the first demonstration of functional crosstalk between neo-Nt-acetylation and proteolytic pathways."

Significance

See above.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary:

The authors present a novel N-terminal enrichment method named LATE (LysN Amino Terminal Enrichment) that utilizes chemical derivatization of α-amines that enables characterization of the N-terminome. Using LATE as well as the already established HYTANE method, Hanna et al conducted a study of caspase-3 mediated proteolysis both in vitro and in cell-based apoptosis experiments, which led to the discovery of new potential caspase-3 cleavages. The results are well presented and nicely highlight that LATE is an efficient and inexpensive method that can be used to identify cleavage sites that cannot be found by other N-terminomics workflows.

Major suggestions:

• The LATE method relies on digestion with LysN. Can the authors comment on the digestion efficiency of the samples where the LATE workflow was applied?
• The authors state that the number of peptides with acetylated N-termini was lower compared with HYTANE. Yet, the Nt-acetylation can occur co-translationally in approximately 85% of human proteins. Did the authors consider optimizing the method (e.g. by fractionating the sample) for better identification of such peptides? Also, were the results of the study compared with searches done using other proteomic pipelines (e.g. FragPipe)?
• Can the authors provide details on the settings used for searches done in COMET, especially for the samples treated with LysN?
• "Fractions containing relatively pure caspase-3 were pooled together and dialyzed against 20 mM HEPES 7.5, and 80 mM NaCl. Aliquots of the protein were stored at -80{degree sign}C"
  • What exactly is meant by 'relatively pure'?

Minor suggestions:

• Please check the link for the Github as this reviewer could not open it.
• Please correct the spelling. Comments regarding figures:
• Figure 2:
  • All figures comparing LATE and HYTANE utilize color green for LATE. Yet, in figure 2G, HYTANE is depicted in green-like color. Please consider staying consistent with the color scheme.

Significance

• The LATE method provides an excellent way to study proteases in vitro or in cell-based experiments. It enables deep investigation of N-terminome based on a simple and cost-effective workflow that utilizes digestion with LysN followed by chemical derivatization of α-amines. This approach allows for the identification of N-terminal peptides that may escape detection by other N-terminomics methods. With LATE, proteases' cleavage sites that might not so far be reporter due to technical limitations, can be studied and characterized. Hence, LATE is a useful addition to the N-terminomic toolbox.

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

Manuscript Reference: RC-2022-01676

TITLE: In-depth characterization of apoptosis N-terminome reveals a link between caspase-3 cleavage and post-translational N-terminal acetylation By Rawad Hanna, Andrey Rozenberg, Daniel Ben-Yosef, Tali Lavy, and Oded Kleifeld

Summary of key results:

The manuscript "In-depth characterization of apoptosis N-terminome reveals a link between caspase-3 cleavage and post-translational N-terminal acetylation" by Rawad and co-authors reports on a negative enrichment strategy, named LysN Amino Terminal Enrichment (LATE) to perform N-terminome analysis, a strategy which complements the cohort of existing negative enrichment strategies thereby jointly permitting a more comprehensive capture of the (neo-)N-terminome by additionally enabling the capture of (neo-)N-termini with (semi-)Lys-N specificity. The authors provide preliminary evidence that Nt-acetylation is protective for a proteins' N-terminus to be cleaved by caspase-3 besides the occurence of putative post-translational Nt-acetylation occurring on neo-N-termini generated upon caspase-3 cleavage.

Concerns:

Page 4 - In contrast to the hindrance of N-terminal amine ionization by Nt-acetyl groups concluded by the authors, previous studies reported an improved MS-scoring if α-amino-acetylated (tryptic) peptides by the higher numbers of b and y fragment ions observed as compared to α-amino-free (tryptic) peptides (e.g. (Staes et al., 2008)). It is rather the lack of any N-/C-terminal charged residue in case of Lys-N type N-termini which makes LATE less suitable for studying N-terminal protein acetylation.

Page 4 - Besides indication the retained N-termini with high relative caspase-3/control abundance ratio's as putative caspase-3 proteolytic products, also indicate that unique peptides were retained, as many such singletons were reported in previous (caspase-focussed) degradomics studies making use of differential proteomics (e.g. (Van Damme et al., 2005)). Therefore the cut-off ratio of 2 rather seems unsubstantiated, unless the cellular proteomes of so-called control cells were affected by caspase activation. As such, showing some representative MS-spectra of neo-N-termini would be informative.

Page 4 - replace 'without labelling of lysine residues (epsilon-amines)' to 'without notable labelling of lysine residues (epsilon-amines)', as residual labelling of lysine side-chains was observed. Also in case of the latter, do note that reduced MS-ionization potential might impact labelling efficiency calculation, and chromatographic detection of labelling efficiency should be considered to conclusify this finding.

Page 6 - The experimental setup comparing caspase-3 overexpressing and ABT-199 induced versus ABT-199 induced cells will be highly biased as it will not be able to detect efficient caspase-3 cleavages (Plasman et al., 2011), as such cleavage events are complete and thus do not require any additional overexpressed capase-3. I see this as an important flaw and the authors should demonstrate that the list also includes efficient caspase-3 cleavages.

Page 12 - The setup doesn't permit ORF N-terminal stability per se, rather the cleavage susceptibly of N-termini holding (a) putative caspase-3 cleavage site(s). Please adjust accordingly. Again since the setup might have missed efficient cleavages, the assessment might be biased.

The claim that Nt-acetylation is protective for caspase-3 cleavage should be validated by monitoring cleavage efficiency of an Nt-acetylated versus an Nt-free variant (e.g. by introducing a Pro residue at AA position 2, or comparing cleavage efficiencies in corresponding NAT knockdown versus control cells) of an identified caspase substrate (i.e. a substrate holding a caspase-3 cleavage site in its N-terminal sequence) versus its Nt-free counterpart

Page 12 - Since post-translational Nt-acetylation of neo-N-termini could be reproduced in vitro in the non-dialyzed sample, enzymatic over chemical Nt-acetylation should be demonstrated (e.g. by the use of a (bisubstrate) NAT inhibitor).

Other concerns:

Abstract - The abstracts holds complex/incorrect sentence constructions (e.g. simply indicate 'Protein N-termini', '... undergo ... processing by proteases' (currently: 'not be processed by proteases').

Abstract - 'To expand the coverage of the N-terminome' only applies when this is used in conjunction with other negative enrichment strategies as by itself, LATE doesn't intrinsically provide a better coverage of the N-terminome (this is also noted at page 2).

Change 'that cannot be identified by other methods' to 'that cannot be identified by other negative selection methods'

Page 1 - Suggestion to change 'Proteases are typically described as degradative enzymes' to 'Proteases used to be described as degradative enzymes'

Page 1 - Not really correct how written; 'N-terminomics methods highlight the N-terminal fragment of every protein (N-terminome)'

Page 2 - Positive selection techniques .... Enrichment of unblocked (or Nt-free) N-termini

Page 2 - Besides altering charge, Nt-acetylation also alters hydrophobicity ...

Page 2 - remove 'to better chart'

Page 2 etc. - Do note that caspase-3 can potentially activate downstream caspases in vitro

Page 3 - functional crosstalk between proteolysis and neo-Nt-acetylation has already been demonstrated in the case of co-translational acting methionine aminopeptidases and chloroplast N-terminal acetyltransferases. Adjust accordingly.

Page 3 - when discussing the identification of ORF N-termini, note that some of the strategies of which note when used to enrich for in vivo blocked N-termini, can also be used without blocking/labelling of Lys residues, and thus trypsin will also result in Lys-ending peptides. This is important to consider in this context.

Page 3 - remove the following sentence part; '... or run individually which can be useful for quantifying naturally modified N-termini.', since also a differential/labelled proteomics setup enables such assessment. Related to this, the authors should comment on the observation that much fewer (i.e. less than 40%) Nt-acetylated N-termini were identified by LATE as compared to HYTANE. How is this reflected in the number of PSMs? Probably these difference are further intensified when considering PSMs.

Page 6 - Informative to indicate how many of the in silico predicted putative DEVD P4-P1 cleavages were actually present in the list of 2049 putative cleavages identified.

Page 6 - Unclear if any of the of 2049 putative cleavages, included non-canonical P1 cleavages besides the P1 Asp and Glu cleavages identified.

Page 6 - Were the 'regular' cells mock transfected?

Page 6 -Important to note that an ORF can have multiple N-termini besides neo-N-termini (e.g. in the case of alternative translation initiation)

Page 6 - The authors should be more careful with generalization when comparing LATE and HYTANE (and other degradomics approaches) as in this study LATE was only applied for the identification of caspase-3 neo-N-termini, which by its extended substrate specificity might hold specific features enabling the preferred detection by one technique over the other. Also note that as compared to less recent studies, evidently the MS instrument used is a key factor in the increase in cleavages reported in the current study.

Page 9 - The authors should provide some info/supporting statistics in the text regarding the new putative substrates showing GO-enrichments (compared to which control?) similar to previously reported caspase-3 substrates.

Page 11 - Indicate that the 11 neo-N-terminal peptides of which note are the neo-Nt-peptides matching (signal peptide) cleavages indicated in the Uniprot database. Were any corresponding di-methylated neo-N-termini of these cleavages identified? In case of the 'other' proteolytic cleavages of which note, refer to these as not-annotated in UniProt.

Page 11 - post-translational Nt-acetylation is abundant in plant and the responsible NAT has been identified, please reference these studies as well.

Page 12 - Define 'undoubtedly dependent on caspase-3 cleavage'

Page 14 - The NAA30 discussion is not really relevant for the discussion of the post-translational Nt-acetylation of mitochondrial neo-N-termini.

Viewing the harsh in vitro caspase-3 cleavage condition used, namely 1 µg caspase 3 over 20 µg protein, the P1 specificities of all identified neo-N-termini should clearly be shown.

Since acetylation of serine and threonine residues are reported forms of post-translational modification, and many so-called past-translational Nt-acetylated neo-N-termini harbour such AA residues in their N-terminal sequence, b-ion coverage for these neo-N-termini should be provided/inspected.

References

Plasman, K., Van Damme, P., Kaiserman, D., Impens, F., Demeyer, K., Helsens, K., . . . Gevaert, K. (2011). Probing the efficiency of proteolytic events by positional proteomics. Mol Cell Proteomics, 10(2), M110 003301. doi:M110.003301 [pii] 10.1074/mcp.M110.003301

Staes, A., Van Damme, P., Helsens, K., Demol, H., Vandekerckhove, J., & Gevaert, K. (2008). Improved recovery of proteome-informative, protein N-terminal peptides by combined fractional diagonal chromatography (COFRADIC). Proteomics, 8(7), 1362-1370. doi:10.1002/pmic.200700950

Van Damme, P., Martens, L., Van Damme, J., Hugelier, K., Staes, A., Vandekerckhove, J., & Gevaert, K. (2005). Caspase-specific and nonspecific in vivo protein processing during Fas-induced apoptosis. Nat Methods, 2(10), 771-777. doi:nmeth792 [pii] 10.1038/nmeth792

Significance

The manuscript "In-depth characterization of apoptosis N-terminome reveals a link between caspase-3 cleavage and post-translational N-terminal acetylation" by Rawad and co-authors reports on a negative enrichment strategy, named LysN Amino Terminal Enrichment (LATE) to perform N-terminome analysis, a strategy which complements the cohort of existing negative enrichment strategies thereby jointly permitting a more comprehensive capture of the (neo-)N-terminome by additionally enabling the capture of (neo-)N-termini with (semi-)Lys-N specificity. The authors provide preliminary evidence that Nt-acetylation is protective for a proteins' N-terminus to be cleaved by caspase-3 besides the occurence of putative post-translational Nt-acetylation occurring on neo-N-termini generated upon caspase-3 cleavage.

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

Learn more at Review Commons

---

Reply to the reviewers

Manuscript number: RC-2022-01588

Corresponding author(s): Erh-Min, LAI

[Please use this template only if the submitted manuscript should be considered by the affiliate journal as a full revision in response to the points raised by the reviewers.

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

1. General Statements [optional]

This section is optional. Insert here any general statements you wish to make about the goal of the study or about the reviews.

The authors thank the reviewers for the positive and valuable comments, which have helped us to improve the quality of this work. We have addressed all comments by providing additional data and/or explanation with a detailed point-by-point response. The revised manuscript included new data: 1) viable cell counts of growth inhibition assay (Fig. 2A), 2) Quantitative data of microscope data (Fig. 2C, Fig. 4), 3) quantitative data of interabacterial competition (Fig. 5A, 5B), western blotting data of growth inhibition (Fig. S1A and S1B), secretion assay of single glycine-zipper mutants (Fig. 5C), and inclusion of full gel of western blot results (Fig. S3 and S5). By integrating these new results, we have substantially strengthened the findings that a glycine zipper motif of a type VI secretion effector T6SS Tde1contributes to its translocation across the cytoplasmic membrane of target cells.

2. Point-by-point description of the revisions

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

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

Summary: In this manuscript, Ali et al. propose that a glycine zipper motif located at the N-terminus of the Agrobacterium tumefaciens T6SS DNase effector, Tde1, can transport the toxin across the cytoplasmic membrane and into the cytoplasm, where its target is found. To support these claims, they perform a series of secretion, competition, toxicity, and fluorescence microscopy assays showing that a mutation in two glycine residues affects toxicity of the effector during competition and its ability to enter a target cell, but not its secretion through the T6SS or its binding to the adaptor protein Tap1. The concept brought forth in this study is quite interesting and important - the notion that T6SS effectors have domains that aid in their transport into the cytoplasm of the target cell. This is similar to a recent finding that a domain common to bacterial pyocins and T6SS effectors can mediate DNase toxin transport through the target cell's cytoplasmic membrane (Atanaskovic et al., mBio, 2022); the authors should mention and discuss this recent work. Nevertheless, it is my impression that the results do not fully support the conclusions and proposed mechanism, even though the general idea seems correct.

Ans: We thank this reviewer found this work interesting and important. We hope the revised manuscript including the new data and careful interpretation have substantialized the conclusions and proposed mechanisms. We also included the excellent work by Atanaskovic et al., 2022 and discussed the findings in the revision (see lines 344-349).

Major comments:

• An experiment that directly demonstrates the ability of the glycine zipper to mediate transport of a toxin across a membrane would greatly support and solidify the conclusions of this work. For example, showing the ability of a purified protein to enter spheroplasts or liposomes in a glycine zipper dependent fashion. Currently, the authors perform experiments that can only indirectly support the proposed function of the glycine zipper to enable the effector to cross the membrane, and as detailed below, some of these experiments are over-interpreted in my opinion. Ans: We agree that the direct evidence for the ability of the glycine zipper to mediate Tde1 transport across target cell membrane is to perform the in vitro translocation assay. Unfortunately, the attempts to purify sufficeint amounts of full-length or N-termial version of Tde1 have not been successful. Therefore, we are unable to perform this experiment. Accoringly, we have tried our best to carefully interpret the data and rephrase the statements accordingly.

• Lines 153-159: It is not clear how much these results are relevant to the activity of the glycine zipper motif during effector delivery by the T6SS. If I understand correctly, the described experiments are of over-expression of the proteins in the E. coli cytoplasm, where glycine zipper-dependent membrane permeability and toxicity are detected. However, one would expect that if the effector is to be transported from the periplasm to the cytoplasm during T6SS delivery, then the glycine zipper should function from the periplasmic face of the cytoplasmic membrane, and not from its cytoplasmic face, as is the case in these experiments. Is it possible that the observed toxicity and membrane permeability be the result of over-expression in the "wrong place"? Ans: The reviewer is right that Tde1 should permealize cytoplasmic membrane from periplasmic side upon injection from the attacker based on our proposed model. The purpose of ectopic expression of Tde1 and its variant in E. coli is to dissect the region and motif of Tde1 DNase-independent toxicity and the ability in enhancing membrane permeability regardless of which sides of cytoplasmic membrane the Tde1 mediates toxicity and permeability. The results of glycine zipper-dependent toxicity and membrane permeability provide a ground work for the experiments of secretion and interabacterial compeittion in the context of active T6SS action to determine the role of glycine zipper in Tde1 export and translocation.

• Fig. 4B: This figure appears to be very important, and the authors base a large part of their main conclusion regarding the role of the glycine zipper in membrane crossing on it. However, some controls are missing and part of the results observed in the figure do not match their description in the text. • Lines 233-237 - While the authors state in the text that GFP and mCherry signals did not overlap in E. coli cells co-cultured with Agrobacterium cells expressing Tde1(M)-GLGL, I see many double-colored cells in this sample (bottom panels in Fig. 4B). Actually, all cells appear to have both green and blue colors, except for a few cells that are only green but that also seem to be dead judging by their ghostly appearance in the phase contrast channel.

Ans: We thank the reviewer pointed this out. By looking at this particular image more carefully, it is striking that the majority of cells seem to emit both green and blue colors from this Tde1(M)GLGL sample. We have performed a total of three indepenent experiments for this translocation assay and all results except this particular sample in this particular experiment are consistent in all three independent experiments. Honestly, we could not explain this result and a possibility is this sample might be accidentally mixed with another sample. Because this is the only sample with inconsistent result with another two independent experiments, we decided NOT to use the results from this independent experiment and instead performed another independent experiment. We now have included the quantitative data from three effective independent experiments and show the representative images in Figure 4.

How is it that all cells in the bottom panels are blue (indicating that they are E. coli target cells)? Shouldn't a large portion of the cells be Agrobacterium cells that should not be blue, since these are added at the beginning of the competition assay at a 10:1 ratio in their favor? Ans: As explained above, we have no defined answer and decided to perform additional repeats, which are consistent with results of another two independent experiments.

It is quite remarkable that so much GFP signal is transported into the E. coli target cells so that it is so clearly visible under the microscope. How do the authors know that the GFP signal overlapping with the mCherry is really inside the cell and not outside (for example, proteins secreted to the media that attach to the cell envelope)? Will the GFP signal remain if trypsin is added to the media before visualization under the microscope? Ans: Indeed, our quantitative data show there are ~50% cells have GFP overlapping with mCherry in the translocation positive samples. The signals should be inside the cells because no overlay signals were observed from N-Tde1GLGL or Tde1(M)GLGLeven though they are secreted.

Can the authors quantify the ratio of E. coli cells that have overlapping green and blue colors over several experiments for each sample, to show that this phenomenon repeats and is statistically significant? Ans: Yes, see quantitative data in Figure 4.

Can the authors explain why at least some of these E. coli cells should not be dead due to the toxicity mediated by the third effector of the Agrobacterium T6SS, Tae? Ans: In Agrobacterium tumefaciens C58, Tde1/2 are the major effectors contributing to antibacterial activity. Tae effector has little impact on interbacterial competition outcome (see previous publications Ma et al., 2014 doi: 10.1016/j.chom.2014.06.002.; Yu et al., 2020 doi.org/10.1128/JB.00490-20)

Why were the microscopy competitions performed differently than the regular competition assays? Why wasn't AK media used in these competitions? How active is the T6SS under these conditions compared to the AK media? Ans: We have tried to use AK medium for the translocation assay but only very weak fluorescent signals can be observed likely due to the low expression when grown on this nutrient poor medium. In order to correlate the results of the compeittion assay with translcoation experiment, we have performed E. coli killing assay using LB medium that is used for translocation experiment now. For the interbacterial competition against agrobacterial siblings, we still used AK medium for competition because no detectable interbacterial compettion activity could be observed between two A. tumefaciens strains on LB agar. As reported earlier, stronger interbacterial competition outcome was detected from co-culture on AK than other nutrient rich medium while the secretion activity grown in AK medium is lower (Yu et al., 2020 doi.org/10.1128/JB.00490-20). These results indicate that the factors other than secretion activity also impacted recipient cell susceptibity, which however is not the main focous of this work.

In the N-Tde1 sample, many Agrobacterium cells appear to have the GFP signal in foci rather than distributed throughout the cell (as it is in other samples), while the E. coli cells have a uniform and strong GFP signal. Can the authors comment on that? Ans: Thanks the reviewer for raising this question.We are also curious about the Tde1 glycine zipper-dependent GFP foci and now include this potential explanation in the Discussion of revised manuscript (line 387-406). To this end, we do not have an answer for it. Because glycine zipper repeats are known to interact with membrane, it is possible that Tde1 proteins may preferntially bind to microdomain of cytoplasmic membrane, which was recently found in A. tumefaciens (Czolkoss et al., 2021). We also found that Tde1 proteins (either tagged with HA or GFP) are proned for truncation when they are ectopically expressed in E. coli or when Tdi1 is absent or not equivalent. Thus, it is possible that Tde1-GFP proteins are truncated after translocation into E. coli cells, in which most GFP signals are emitted from free GFP instead of Tde1-GFP. The stability of free GFP derived from translocated Tde1-GFP may also explain the high percentage of E. coli cells exhibiting overlayed GFP/mCherry signals.

It might be easier for readers to visualize this figure and see the signal distribution in the different cells if the authors show a zoomed in version in the main text, and provide the wide field images as a supplementary figure. Ans: We have tried to include zoom-in images but the resolution is not good. We have improved the quality of images in the Figure 4 and believe the images are clear to see individual and overlayed fluorescence signals.

• Fig. 5C-D: The reduced expression and secretion of the GLGL mutant is considerable. How can the authors rule out that this reduction was the cause for the reduced observed toxicity of the mutant in 5A-B? Moreover, the results show that the GLGL double mutant is hampered in expression, secretion, and DNase activity, and it negatively affects overall T6SS activity. Since this mutant was used throughout the paper, and in the absence of a direct assay showing membrane transport mediated by the glycine zipper motif, the claim of the role of this motif in membrane crossing is not well substantiated by the results. If the authors were to show that the single glycine mutants used in Fig. 5D, which are stable and have an intact DNase activity, behave as claimed in the final conclusion sentence (lines 279-283), then the conclusions will be better substantiated by the results. Ans: Thank you very much for suggesting this important experiment. We have now constructed the single G39L and G43L variants expressed together with Tdi1 in A. tumefaciens tdei mutant for both secretion and interbacterial competition assays (see description in lines 259-280 and Fig. 5). As shown in Figure 5, both G39L and G43L variants are expressed and secreted at similar or even higher levels than wild type Tde1 but have no detectable antibacterial activity against either E. coli or A. tumefaciens 1D1609. This result substantializes the role of this glycine zipper motif in translocation.

Minor comments:

• Line 93: I am not sure that Ntox15 should still be referred to as a "novel" domain.

Ans: despite the evidence of this domain as DNase, the name of Ntox15 is used. We think to keep this nomenclatture as it will be easier to be ditinquished from other nuclease or toxin domain.

• Line 105: The section's heading does not actually describe its content. The results here only show toxicity upon over-expression of the effector or its mutant forms in E. coli. Therefore, this cannot be referred to as a "prey cell" since the effector was not transported into it during competition. Moreover, the results in Fig. 5A do not support DNase-independent toxicity during competition. Ans: The heading is changed to “Tde1 exhibits DNase-independent growth inhibition in E. coli” (line 115).

• Please consider making all of the symbols in the growth assays semi-transparent. It is impossible to discern between the different, overlapping curves. Ans: The growth curve results are improved by changing line colors and reducing size bars (Fig. 1B, 1C; Fig. 2A, 2D)

• Please consider making the size bars in all microscopy images more pronounced. They are barely visible in their current form. Also, it would be better to show images of the same magnification/zoom for the different samples, since the current presentation shows cells from different samples at different sizes, and it can be confusing to the readers. Ans: Amended (Fig. 2C; Fig. 4).

• In Fig. 1B and in Fig. 2A the authors show that expression of Tde1(M) in cells is toxic, yet in Fig. 2D they see no toxicity. Can the authors please comment on this discrepancy? Ans: Fig. 2D showed the viability of E. coli cells after Tde1 variants were induced for 1 hr before ONPG uptake assay to indicate the increased membrane permeability is not due to cell death. In Fig. 1B, the growth inhibition of Tde1(M) is also not evident at 1 hr. So, the results are consistent.

• I am not convinced that the assay in Fig. 2E can be used to determine bacteriostatic/bacteriolytic effect. It is not clear how such a distinction can be made from OD measurements, since an increase in OD can result from the entire population growing after removal of the stressor, or just part of the population that did not lyse/die. To make such a claim, the authors can spot bacteria on repressing media at different timepoints after protein induction, and then determine CFU.

Ans: The increased OD600 value during recovery could be caused by either resumed cell division or cell elongation. Based on the newly added growth inhibition assay of all Tde1 variants which we showed nice correlation between CFU counting and OD600value (Fig. 2A, S2) and no increased cell size/length of E. coli cells expressing N-Tde1 or Tde1(M), we think the recovered OD600 value is supportive of N-Tde1 or Tde1(M) exhibiting bacteriostatic toxicity. In addition to that, our interbacterial competition data showed that Tde1(M)-Tdi1 which is still having intact glycine zipper doesn’t show significant detectable killing, supporting the bacteriostatic function of Tde1 glycine zippers. In fact, we performed this experiment based on Mariano et al.(Nat. Commun. 2019 doi: 10.1038/s41467-019-13439-0), which showed the recovery of OD600 value after removal of inducer as the evidence that the Ssp6 toxin is not bacteriolytic.

• Fig. 3A: A control is missing. To verify that the N-terminal part of Tde1 is not promiscuously interacting with proteins, the authors should include a control sample testing its inability to precipitate a protein other than Tap1 in the same experiment. Ans: Our previous study has showed that Tde1 can co-immunprecipiate Tap1 but not a non-T6SS protein RpoA (Bondage et al., 2016 doi:10.1073/pnas.1600428113), indicating that Tde1 is not promiscuously interacting with proteins. Considering the tight biochemical interaction between Tap1 with N-Tde1 but not C-Tde1 that correlate with their ability for secretion upon loading onto VgrG1, N-Tde1 is unlikely to bind proteins non-specifically. This is also supported by the non-specific protein bands from cellular fractions recognized by anti-Tap1 are not co-immunoprecipitated by any of Tde1 variants (Fig. S3). We could repeat the experiments to include additional proteins as negative controls but we chose to use time for other more critical experiments during the limited revision time.

• Fig. 3B: the blots are very "dirty". It is not clear how the authors were able to determine expression and precipitation of some truncations (for example, C2-Tde1 in the E. coli IP panel looks like a background band found in other lanes too).

Ans: We agree that western blots of co-IP experiments in E. coli are not very clear due to the weak signals of some Tde1 variants and background. As pointed out by the reviewer 3, this result is not conclusive and rovide little additional information other than the co-IP results from A. tumefaciens. Because the interaction between Tde1 variants and Tap1 when expressed in E. coli are not physiologically relevant and not the main focus of this work, we have removed the E. coli co-IP results from this manuscript as suggested by the reviewer 3.

• Lines 222-225 (Fig. 4A): I can't see C-1-Tde1(M)-sfGFP in the cellular blot. All the bands in this lane look like background bands that are also present in all other lanes. Therefore, I am not sure how the conclusion regarding this truncation's ability to be secreted was reached. Ans: We agree that C1-Tde1(M)-sfGFP is barely detectable due to its weak signal overlapping with cross-reacted bands. Since several attemps to improve the western blot quality by changing antibody and pre-blocking with protein lysages of vector control strain did not produce convincing results for detection of C1-Tde1(M)-sfGFP, we have rephrased the description of this result as “However, C1-Tde1(M)-sfGFP protein signal could not be unambiguously determined in the cellular fraction due to the overlapping of its predicted protein band with cross-reacted proteins, and no corresponding C1-Tde1(M)-sfGFP band was detected in the extracellular fraction.” (line 234-237).

• Fig. 4A: the protein names above the lanes should include the sfGFP that is fused to them. Ans: Amended.

• It would be preferable to show quantitative competition assays with statistics rather than pictures of a plate showing a single competition result, if conclusions or observations on minor differences in toxicity are made (for example, line 253: "The killing activity of Δtdei(Tde1GLGL-Tdi1) was largely compromised"). Since the authors performed each competition assay more than once, these data should be available to them. Ans: Amended. We have repeated the interbacterial competition experiments including single G39L and G43L variants for multiple biological repeats (see detailed in legends of Fig. 5A, 5B). The quantitative data with statistical analysis were added, which show no statistical difference of any glycine zipper mutants as comapred to Tde1(M) or when expressed in the T6SS mutant. Thus, there are no detectable antibacterial activity of glycine zipper mutants against either E. coli or A. tumefaciens siblings.

• Fig. 5A: The author claim at the beginning of the manuscript (first results section heading: "Tde1 can cause DNase-independent growth inhibition of prey cells") that the N-terminal region of Tde1 is toxic on its own in the prey cell, yet in this competition assay Tdi1(M) shows no toxicity against the E. coli target cells. In the microscopy assay (Fig. 4B), it appears that a lot of Tdi1(M) enters the prey cell, since we can visualize it under the microscope. Can the authors clarify this discrepancy and explain why they do not expect to see target killing by this mutant even though they claimed it is toxic earlier? Ans: As describbed in earlier response, N-Tde1 amd Tde1(M) toxicity can exhibit toxicity by ectopic expression in E. coli. We mainly used this ectopic expression assay to dissect the region and motif contributing the toxicity. Compared to the interbacterial competiton process where Tde1(M) may only transiently permealze cytoplasmic membrane transiently as the final destination is cytoplasm where wild type Tde1 but not Tde1(M) exerts DNase toxicity. Thus, the toxicity of N-Tde1 and Tde1(M) can be only observed when the proteins are continuously produced in the cytoplasm. The role of N-Tde1, specifically the glycine zipper motifs, is to mediate Tde1 translocation across inner membrane, instead of exerting toxicity during the context of interbacterial competition.

• Fig. 5B: the GLGL mutant seems to have some residual toxicity, not dissimilar to what is shown in 5A. Why are these similar results interpreted differently (in 5A they are "largely compromised", while in 5B "killing activity... was not detectable")? Also, why was Tde(M)1-Tdi1 used in Fig. 5A but Tdi1(M) without the immunity gene used in Fig. 5B? Ans: As described above, to better quantify the interbacterial competition outcomes, we have repeated the interbacterial competition experiments and used Tde(M)1-Tdi1 instead of Tdi1(M) for at least three biological replicates. The quantitative data with statistical analysis were carried out to clarify this ambiguity (Fig. 5A, 5B).

• Fig. 5: Does the remaining third effector, Tae, not play a role in these competition assays? If, as shown in Fig. 5C, the entire T6SS is less active when a GLGL mutant is expressed, couldn't the different in toxicity shown in Figs. 5A-B be the result of lack of Tae secretion and toxicity?

Ans: As decribed above, Tae effector has little impact on interbacterial competition outcome. The quantatitive interbacterial competition results (Fig. 5A, 5B) also clarify the ambiguity because single G39L and G43L variants are expressed and secreted at similar or even higher levels than wild type Tde1 but have no detectable antibacterial activity against either E. coli or A. tumefaciens 1D1609.

• Lines 359-362: T6SS effectors that bind the inner Hcp tube were suggested to be only partially folded. Ans: Amended.

Reviewer #1 (Significance (Required)):

The concept of T6SS effectors providing their own mechanism of transport from the cytoplasm to the periplasm is very interesting. It will appeal to audience in a wide range of microbiology disciplines, including those interested in toxins, membrane transport, and even translational applications. A similar concept was recently proposed and demonstrated for a domain that is also found in T6SS effectors (Atanaskovic et al., mBio, 2022).

Expertise: I have been studying the different aspects of T6SS for the past decade.

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

This manuscript is focused on understanding how the Agrobacterium tumefaciens T6SS effector, Tde1, is translocated across the cell envelope of target cells and how this effector binds to the adapter Tap1. The authors show that GxxxG motifs in the N terminal region of Tde1 are required for delivery into the cytoplasm of target cells and permeabilising the cytoplasmic membrane. Given that these GxxxG motifs resemble glycine zipper structures that are found in proteins involved in membrane channel formation, the authors propose that these Tde1 motifs are involved in channel formation in the target cell. The authors also show that the N terminal region of Tde1 binds to Tap1 to facilitate loading onto the T6SS machinery but that the GxxxG motifs are not involved in this binding. Overall the manuscript was easy to read and followed a logical presentation of the findings. There are a few major comments that this reviewer has below - addressing these would allow the authors' claims to be more robustly supported. Ans: Thank you very much for the positive comments and valuable suggestion. We hope the revised manuscript including the new data and careful interpretation have substantialized the conclusions and proposed mechanisms.

Major comments:

1. Fig 1B: Why is this such a short growth experiment (5 hrs total with 2 hr pre and 3 hrs post induction)? Reporting on a growth experiment would normally be at least until the cells reach stationary phase but here the cells are still clearly in exponential phase. This reviewer would query what happens to growth rate in later exponential growth and into stationary phase? Is the toxic effect lessened in later stages of growth? Ans: We have indeed performed the growth curve analysis with longer time period. However, we noted that the growth at later time points are not always consistent and our interpretation is that the continuous expression of toxins may lead to the selection of mutants. Since the 3 or 4 hr time period already showed the toxicity phenotype, we have focused on this time frame for the growh experiments.

2. It is indeed surprising that C2-Tde1(WT) does not inhibit growth despite it having a functional DNase domain and being expressed in the cytoplasm. Did the authors confirm that this protein variant was expressed by Western blot or other means? This should be done to confirm that this variant is indeed not impacting upon growth instead of it not impacting growth simply because it is not being expressed.

Ans: Amended. All Tde1 variants including C2-Tde1 are expressed (data included in Fig S1)

1. The letters used to report significance are not clear to this reviewer. The authors say that "The significant differences were shown by the different letters (p value

For all fluorescence microscopy experiments how many fields of view were imaged for each biological replicate? Were the fields selected at random or was the field selection biased to what was present in the field before taking the image? The answers to all of these questions should be stated in the methods. Also the microscopy data presented in the manuscript is not quantitative. Quantification of the number of cells with PI vs Hoechst signal (in Fig 2C) and mcherry vs gfp signal (in Fig 4B) for all fields of view and for all biological replicates would be very informative and convince the reader that the authors have not just "cherry picked" the images they are showing in the manuscript. This could be performed manually or the authors could use the freely available image analysis program Fiji (https://imagej.net/software/fiji/) to perform these analysis in a semi-automated manner.

Ans: The number of images and experiments were now described in the figure legends and the quantititive data are included (Fig. 2C).

1. For the co-IP experiments in Fig 3 where interaction between HA tagged Tde1 and Tap1 is demonstrated the authors should also show that Tap1 does not interact with a different HA-tagged protein i.e. that the interaction is specific to Tde1 and not the HA motif. Ans: All Tde1 variants were tagged with HA. As shown in Fig. 3A, Tap1was not co-precipitated by C2-Tde1 and C1-Tde1(M), indicating that Tap1 specifically interacts with N-terminal region of Tde1.

For all Western blot images there should be at least 2 protein standard markers present in each individual blot - i.e. for Fig 3A and B the bottom panel showing Tap1 detection only has the 35 kDa marker, it should have at least one more marker in it. The same is true for other panels in Fig 2, 3 and 4. Having at least two molecular weight markers in a panel is now standard for most journals when presenting Western blot images. Ans: Amended. We have now included the full gel of western blot results in Fig. S3 and S5 of those shown in main figures.

For the competition assay serial dilution images in Fig 5A-B the images are a nice way to visually represent the experimental outcome but they should accompany graphs showing the competitive index of CFU/ml of the input prey and attacker vs the output prey and attacker for all biological replicates. This will convince the reader that the authors had equivalent amounts of the prey and the attacker going into the experiment and also that all attackers grew at the same rate and so were equally able to target the prey cell. This quantification could also provide more convincing out competition of ID1609 prey by C58 attacker (Fig 5B). Ans: Amended. As indicated above, we have repeated the interbacterial competition experiments for at leaset three biological replicates and show that quantitative data with statistical analysis (Fig. 5A, 5B).

Minor comments:

Line 40: should read "...demonstrate that the effector itself..." Ans: The sentence has been rephrased (line 40) .

Line 41: "...we propose..." instead of "...we proposed..." since present tense makes more sense for this statement.

Ans: Amended (line 42).

Line 51: "Each specialized protein secretion system" instead of "Each of...." Ans: Amended (line 52).

Line 76: "A glycine zipper structure..."

Ans: Amended (line 83).

Line 79: "For example..."

Ans: Amended (line 86).

Lines 96-100: The present tense should be used here as the current usage of past tense implies that this has been done in previous work and not in the current study - eg "we revealed", "we showed" would be better as "we reveal", "we show".

Ans: Thanks for the advice. We have made changes throughout the manuscript.

Fig 5B - The competition assay serial dilution images look a bit blurry, are there images the authors could use that are not blurry?

Ans: Amended. As indicated above, we now show quantitative data with statistical analysis (Fig. 5A, 5B).

Reviewer #2 (Significance (Required)):

This work is significant in as while there is a great deal known about how T6SS effectors cause toxicity there is less known about how these effectors are loaded onto the T6SS machinery and very little known about how T6SS effectors are able to translocate across the cytoplasmic membrane of target cells to reach a cellular component that is in the cytoplasm. This work would be of wide general interest to researchers in the T6SS field as well as those interested in bacterial secretion systems.

Reviewer expertise key words: Molecular microbiology, T6SS, interbacterial competition

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

EVIDENCE, REPRODUCIBILITY AND CLARITY

Summary:

In this work, Ali et al. demonstrate that the N-terminal GxxxG motif of the T6SS DNase effector Tde1 of Agrobacterium tumefaciens is required for interbacterial intoxication. Using a combination of cell viability, reporter, and microscopy assays, the authors demonstrate that over-expression of the N-terminus of Tde1 results in inner membrane permeability. Moreover, the authors show that both the interaction between Tde1 and its adaptor Tap1 as well as the T6SS-mediated secretion of Tde1 are dependent on the N-terminus of Tde1. Finally, using a combination of in vitro and in vivo experiments, the authors determine that the N-terminal GxxxG motif is essential to Tde1-dependent interbacterial killing by enabling effector entry into competing bacterial cells.

Major comments:

If N-tde1 is 1-97 aa, the predicted size is 9 kDa, but it shows up as ~17 kDa? Can the authors comment on this? Does N-tde1 or tde1 dimerize? Ans: The theoretical Mw of N-Tde1-HA is 10.64 KDa, which indeed migrated at higer position ~17 kDa. It is notable that full-length Tde1 with theoretical 29.5-kDa migrated slower in SDS-PAGE with a observed size ~36 kDa as observed previously (Ma et al., 2014 doi:10.1016/j.chom.2014.06.002). Similarly, the full-length HA-tagged Tde1(M) with theoretical 30.89 kDa migrated at a position ~38 kDa. Since the protein samples analyzed by SDS-PAGE including reducing agent, we cannot exclude the possibility that Tde1 or N-Tde1 may form dimer or oligomer that was disrupted by SDS-PAGE but it appears not forming dimer on SDS-PAGE.

I have many concerns with the data and conclusions drawn from the data in Fig. 3B. I recommend removing it since (1) the data are not accurately represented in the text and (2) it is difficult to ascertain whether biologically relevant conclusions can be drawn from what happens with Agrobacterium proteins in E. coli. Below is a summary of my concerns regarding this section: I disagree with the authors' statements in lines 191-198. Their pulldown with E. coli is not consistent with their pulldown in C58. In fact, given the expression problems of some of the constructs in E. coli, I believe the data shown in Fig. 3B is inconclusive. The amount of Tap1 that co-IP'ed with N-Tde1GLGL and Tde1(M) is very low even though the expression levels of N-Tde1GLGL and Tde1(M) were relatively strong. Therefore, I do not feel confident concluding that these proteins "interact". Secondly, Tde1(M)GLGL was not expressed in E. coli, so no conclusions can be drawn. Moreover, the C1 and C2 variants were also not expressed well, so I believe the authors' statement in line 191-192: "Similar to the results in A. tumefaciens, the N-Tde1 and Tde1(M) interacted with Tap1 but not the C-terminal variants", is unjustified. You cannot rule out that C1 and C2 do not interact with Tap1 because C1 and C2, like Tde1(M)GLGL, were not expressed well in E. coli. Ans: We agree with the reviewer that the E. coli co-IP result is not conclusive due to the low expression and instability of proteins mostly during the process of cell lysis and purification, and it provides little additional information other than data from co-IP in A. tumefaciens. Because the interaction between Tde1 variants and Tap1 when expressed in E. coli are not physiologically relevant and not the main focus of this work, we have removed the E. coli co-IP results from this manuscript.

Lines 211-214: It looks like C1-Tde1(M) inhibits T6SS secretion. I am aware that in Agrobacterium, it has been shown that effector loading is essential for secretion, but then why does the pTrc200 secrete Hcp? Also, in Fig. 4B, a strain expressing C1-Tde1(M) now secretes Hcp. Ans: Thanks for noting our previous finding that Tde loading is critical for secretion. Our data are indeed supportive of the effector loading in activating T6SS as only very low levels of Hcp secretion could be detected from the strain containing vector only or C1-Tde1(M). In our previous paper (Wu et al., 2020 https://doi.org/10.15252/embr.201947961), there is either little or no detection of Hcp secretions when effectors are not loaded, indicating that effector loading is important but not essential for Hcp secretion. Because overexpression of VgrG can also activate T6SS secretion in the absence of effector loading (Bondage et al., 2016 doi:10.1073/pnas.1600428113), we think the low level secretion under certain conditions could be caused by some cells with higher levels of VgrG protein concentration but more work is required to elucidate the underlying mechanisms.

Minor comments:

Fig. 2B could benefit from better labeling to indicate that most strains lack lacY. Also, why is BW25113 WT showing such a low OD420 if it has LacY? Or is WT without lacZ? Please clarify.

Ans: We apologize for not labeling clearly. The BW25113 strain lacks lacZ, therefore all the ∆lacY strains were complemented with a plasmid encoding lacZ (pYTA-lacZ). We have now added the labels to avoid confusion (Fig. 2B).

Reviewer #3 (Significance (Required)):

SIGNIFICANCE

It has been known for over a decade that T6SS effectors have both periplasmic and cytosolic targets (e.g., cell wall and DNA). However, it remains unclear (1) where within the target cell are T6SS effectors are delivered and (2) once delivered, how do effectors reach their intracellular target site. In this work, Ali et al. demonstrate that for Tde1, the N-terminal GxxxG motif is essential for Tde1 to reach its target (DNA). The authors identified Tde1 homologs in several bacteria, suggesting that this model may be relevant across a wide range of bacteria. Additional research is needed to (1) determine whether Tde1 is originally secreted into the periplasm and (2) understand how non-Tde1/non-GxxxG effectors reach their target site.

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 work, Ali et al. demonstrate that the N-terminal GxxxG motif of the T6SS DNase effector Tde1 of Agrobacterium tumefaciens is required for interbacterial intoxication. Using a combination of cell viability, reporter, and microscopy assays, the authors demonstrate that over-expression of the N-terminus of Tde1 results in inner membrane permeability. Moreover, the authors show that both the interaction between Tde1 and its adaptor Tap1 as well as the T6SS-mediated secretion of Tde1 are dependent on the N-terminus of Tde1. Finally, using a combination of in vitro and in vivo experiments, the authors determine that the N-terminal GxxxG motif is essential to Tde1-dependent interbacterial killing by enabling effector entry into competing bacterial cells.

Major comments:

If N-tde1 is 1-97 aa, the predicted size is 9 kDa, but it shows up as ~17 kDa? Can the authors comment on this? Does N-tde1 or tde1 dimerize?

I have many concerns with the data and conclusions drawn from the data in Fig. 3B. I recommend removing it since (1) the data are not accurately represented in the text and (2) it is difficult to ascertain whether biologically relevant conclusions can be drawn from what happens with Agrobacterium proteins in E. coli. Below is a summary of my concerns regarding this section: I disagree with the authors' statements in lines 191-198. Their pulldown with E. coli is not consistent with their pulldown in C58. In fact, given the expression problems of some of the constructs in E. coli, I believe the data shown in Fig. 3B is inconclusive. The amount of Tap1 that co-IP'ed with N-Tde1GLGL and Tde1(M) is very low even though the expression levels of N-Tde1GLGL and Tde1(M) were relatively strong. Therefore, I do not feel confident concluding that these proteins "interact". Secondly, Tde1(M)GLGL was not expressed in E. coli, so no conclusions can be drawn. Moreover, the C1 and C2 variants were also not expressed well, so I believe the authors' statement in line 191-192: "Similar to the results in A. tumefaciens, the N-Tde1 and Tde1(M) interacted with Tap1 but not the C-terminal variants", is unjustified. You cannot rule out that C1 and C2 do not interact with Tap1 because C1 and C2, like Tde1(M)GLGL, were not expressed well in E. coli.

Lines 211-214: It looks like C1-Tde1(M) inhibits T6SS secretion. I am aware that in Agrobacterium, it has been shown that effector loading is essential for secretion, but then why does the pTrc200 secrete Hcp? Also, in Fig. 4B, a strain expressing C1-Tde1(M) now secretes Hcp.

Minor comments:

Fig. 2B could benefit from better labeling to indicate that most strains lack lacY. Also, why is BW25113 WT showing such a low OD420 if it has LacY? Or is WT without lacZ? Please clarify.

Significance

It has been known for over a decade that T6SS effectors have both periplasmic and cytosolic targets (e.g., cell wall and DNA). However, it remains unclear (1) where within the target cell are T6SS effectors are delivered and (2) once delivered, how do effectors reach their intracellular target site. In this work, Ali et al. demonstrate that for Tde1, the N-terminal GxxxG motif is essential for Tde1 to reach its target (DNA). The authors identified Tde1 homologs in several bacteria, suggesting that this model may be relevant across a wide range of bacteria. Additional research is needed to (1) determine whether Tde1 is originally secreted into the periplasm and (2) understand how non-Tde1/non-GxxxG effectors reach their target site.

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 is focused on understanding how the Agrobacterium tumefaciens T6SS effector, Tde1, is translocated across the cell envelope of target cells and how this effector binds to the adapter Tap1. The authors show that GxxxG motifs in the N terminal region of Tde1 are required for delivery into the cytoplasm of target cells and permeabilising the cytoplasmic membrane. Given that these GxxxG motifs resemble glycine zipper structures that are found in proteins involved in membrane channel formation, the authors propose that these Tde1 motifs are involved in channel formation in the target cell. The authors also show that the N terminal region of Tde1 binds to Tap1 to facilitate loading onto the T6SS machinery but that the GxxxG motifs are not involved in this binding. Overall the manuscript was easy to read and followed a logical presentation of the findings. There are a few major comments that this reviewer has below - addressing these would allow the authors' claims to be more robustly supported.

Major comments:

1. Fig 1B: Why is this such a short growth experiment (5 hrs total with 2 hr pre and 3 hrs post induction)? Reporting on a growth experiment would normally be at least until the cells reach stationary phase but here the cells are still clearly in exponential phase. This reviewer would query what happens to growth rate in later exponential growth and into stationary phase? Is the toxic effect lessened in later stages of growth?
2. It is indeed surprising that C2-Tde1(WT) does not inhibit growth despite it having a functional DNase domain and being expressed in the cytoplasm. Did the authors confirm that this protein variant was expressed by Western blot or other means? This should be done to confirm that this variant is indeed not impacting upon growth instead of it not impacting growth simply because it is not being expressed.
3. The letters used to report significance are not clear to this reviewer. The authors say that "The significant differences were shown by the different letters (p value <0.01)" and then have a, b, c etc next to lines of the growth experiments in Figure 1B, C and Fig 2A, B, E etc. Which comparisons have a p value <0.01? this is not clear.
4. For all fluorescence microscopy experiments how many fields of view were imaged for each biological replicate? Were the fields selected at random or was the field selection biased to what was present in the field before taking the image? The answers to all of these questions should be stated in the methods. Also the microscopy data presented in the manuscript is not quantitative. Quantification of the number of cells with PI vs Hoechst signal (in Fig 2C) and mcherry vs gfp signal (in Fig 4B) for all fields of view and for all biological replicates would be very informative and convince the reader that the authors have not just "cherry picked" the images they are showing in the manuscript. This could be performed manually or the authors could use the freely available image analysis program Fiji (https://imagej.net/software/fiji/) to perform these analysis in a semi-automated manner.
5. For the co-IP experiments in Fig 3 where interaction between HA tagged Tde1 and Tap1 is demonstrated the authors should also show that Tap1 does not interact with a different HA-tagged protein i.e. that the interaction is specific to Tde1 and not the HA motif.
6. For all Western blot images there should be at least 2 protein standard markers present in each individual blot - i.e. for Fig 3A and B the bottom panel showing Tap1 detection only has the 35 kDa marker, it should have at least one more marker in it. The same is true for other panels in Fig 2, 3 and 4. Having at least two molecular weight markers in a panel is now standard for most journals when presenting Western blot images.
7. For the competition assay serial dilution images in Fig 5A-B the images are a nice way to visually represent the experimental outcome but they should accompany graphs showing the competitive index of CFU/ml of the input prey and attacker vs the output prey and attacker for all biological replicates. This will convince the reader that the authors had equivalent amounts of the prey and the attacker going into the experiment and also that all attackers grew at the same rate and so were equally able to target the prey cell. This quantification could also provide more convincing out competition of ID1609 prey by C58 attacker (Fig 5B).

Minor comments:

Line 40: should read "...demonstrate that the effector itself..."

Line 41: "...we propose..." instead of "...we proposed..." since present tense makes more sense for this statement.

Line 51: "Each specialized protein secretion system" instead of "Each of...."

Line 76: "A glycine zipper structure..."

Line 79: "For example..."

Lines 96-100: The present tense should be used here as the current usage of past tense implies that this has been done in previous work and not in the current study - eg "we revealed", "we showed" would be better as "we reveal", "we show".

Fig 5B - The competition assay serial dilution images look a bit blurry, are there images the authors could use that are not blurry?

Significance

This work is significant in as while there is a great deal known about how T6SS effectors cause toxicity there is less known about how these effectors are loaded onto the T6SS machinery and very little known about how T6SS effectors are able to translocate across the cytoplasmic membrane of target cells to reach a cellular component that is in the cytoplasm. This work would be of wide general interest to researchers in the T6SS field as well as those interested in bacterial secretion systems.

Reviewer expertise key words: Molecular microbiology, T6SS, 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:

In this manuscript, Ali et al. propose that a glycine zipper motif located at the N-terminus of the Agrobacterium tumefaciens T6SS DNase effector, Tde1, can transport the toxin across the cytoplasmic membrane and into the cytoplasm, where its target is found. To support these claims, they perform a series of secretion, competition, toxicity, and fluorescence microscopy assays showing that a mutation in two glycine residues affects toxicity of the effector during competition and its ability to enter a target cell, but not its secretion through the T6SS or its binding to the adaptor protein Tap1. The concept brought forth in this study is quite interesting and important - the notion that T6SS effectors have domains that aid in their transport into the cytoplasm of the target cell. This is similar to a recent finding that a domain common to bacterial pyocins and T6SS effectors can mediate DNase toxin transport through the target cell's cytoplasmic membrane (Atanaskovic et al., mBio, 2022); the authors should mention and discuss this recent work. Nevertheless, it is my impression that the results do not fully support the conclusions and proposed mechanism, even though the general idea seems correct.

Major comments:

• An experiment that directly demonstrates the ability of the glycine zipper to mediate transport of a toxin across a membrane would greatly support and solidify the conclusions of this work. For example, showing the ability of a purified protein to enter spheroplasts or liposomes in a glycine zipper dependent fashion. Currently, the authors perform experiments that can only indirectly support the proposed function of the glycine zipper to enable the effector to cross the membrane, and as detailed below, some of these experiments are over-interpreted in my opinion.
• Lines 153-159: It is not clear how much these results are relevant to the activity of the glycine zipper motif during effector delivery by the T6SS. If I understand correctly, the described experiments are of over-expression of the proteins in the E. coli cytoplasm, where glycine zipper-dependent membrane permeability and toxicity are detected. However, one would expect that if the effector is to be transported from the periplasm to the cytoplasm during T6SS delivery, then the glycine zipper should function from the periplasmic face of the cytoplasmic membrane, and not from its cytoplasmic face, as is the case in these experiments. Is it possible that the observed toxicity and membrane permeability be the result of over-expression in the "wrong place"?
• Fig. 4B: This figure appears to be very important, and the authors base a large part of their main conclusion regarding the role of the glycine zipper in membrane crossing on it. However, some controls are missing and part of the results observed in the figure do not match their description in the text.
  • Lines 233-237 - While the authors state in the text that GFP and mCherry signals did not overlap in E. coli cells co-cultured with Agrobacterium cells expressing Tde1(M)-GLGL, I see many double-colored cells in this sample (bottom panels in Fig. 4B). Actually, all cells appear to have both green and blue colors, except for a few cells that are only green but that also seem to be dead judging by their ghostly appearance in the phase contrast channel.
  • How is it that all cells in the bottom panels are blue (indicating that they are E. coli target cells)? Shouldn't a large portion of the cells be Agrobacterium cells that should not be blue, since these are added at the beginning of the competition assay at a 10:1 ratio in their favor?
  • It is quite remarkable that so much GFP signal is transported into the E. coli target cells so that it is so clearly visible under the microscope. How do the authors know that the GFP signal overlapping with the mCherry is really inside the cell and not outside (for example, proteins secreted to the media that attach to the cell envelope)? Will the GFP signal remain if trypsin is added to the media before visualization under the microscope?
  • Can the authors quantify the ratio of E. coli cells that have overlapping green and blue colors over several experiments for each sample, to show that this phenomenon repeats and is statistically significant?
  • Can the authors explain why at least some of these E. coli cells should not be dead due to the toxicity mediated by the third effector of the Agrobacterium T6SS, Tae?
  • Why were the microscopy competitions performed differently than the regular competition assays? Why wasn't AK media used in these competitions? How active is the T6SS under these conditions compared to the AK media?
  • In the N-Tde1 sample, many Agrobacterium cells appear to have the GFP signal in foci rather than distributed throughout the cell (as it is in other samples), while the E. coli cells have a uniform and strong GFP signal. Can the authors comment on that?
  • It might be easier for readers to visualize this figure and see the signal distribution in the different cells if the authors show a zoomed in version in the main text, and provide the wide field images as a supplementary figure.
• Fig. 5C-D: The reduced expression and secretion of the GLGL mutant is considerable. How can the authors rule out that this reduction was the cause for the reduced observed toxicity of the mutant in 5A-B? Moreover, the results show that the GLGL double mutant is hampered in expression, secretion, and DNase activity, and it negatively affects overall T6SS activity. Since this mutant was used throughout the paper, and in the absence of a direct assay showing membrane transport mediated by the glycine zipper motif, the claim of the role of this motif in membrane crossing is not well substantiated by the results. If the authors were to show that the single glycine mutants used in Fig. 5D, which are stable and have an intact DNase activity, behave as claimed in the final conclusion sentence (lines 279-283), then the conclusions will be better substantiated by the results.

Minor comments:

• Line 93: I am not sure that Ntox15 should still be referred to as a "novel" domain.
• Line 105: The section's heading does not actually describe its content. The results here only show toxicity upon over-expression of the effector or its mutant forms in E. coli. Therefore, this cannot be referred to as a "prey cell" since the effector was not transported into it during competition. Moreover, the results in Fig. 5A do not support DNase-independent toxicity during competition.
• Please consider making all of the symbols in the growth assays semi-transparent. It is impossible to discern between the different, overlapping curves.
• Please consider making the size bars in all microscopy images more pronounced. They are barely visible in their current form. Also, it would be better to show images of the same magnification/zoom for the different samples, since the current presentation shows cells from different samples at different sizes, and it can be confusing to the readers.
• In Fig. 1B and in Fig. 2A the authors show that expression of Tde1(M) in cells is toxic, yet in Fig. 2D they see no toxicity. Can the authors please comment on this discrepancy?
• I am not convinced that the assay in Fig. 2E can be used to determine bacteriostatic/bacteriolytic effect. It is not clear how such a distinction can be made from OD measurements, since an increase in OD can result from the entire population growing after removal of the stressor, or just part of the population that did not lyse/die. To make such a claim, the authors can spot bacteria on repressing media at different timepoints after protein induction, and then determine CFU.
• Fig. 3A: A control is missing. To verify that the N-terminal part of Tde1 is not promiscuously interacting with proteins, the authors should include a control sample testing its inability to precipitate a protein other than Tap1 in the same experiment.
• Fig. 3B: the blots are very "dirty". It is not clear how the authors were able to determine expression and precipitation of some truncations (for example, C2-Tde1 in the E. coli IP panel looks like a background band found in other lanes too).
• Lines 222-225 (Fig. 4A): I can't see C-1-Tde1(M)-sfGFP in the cellular blot. All the bands in this lane look like background bands that are also present in all other lanes. Therefore, I am not sure how the conclusion regarding this truncation's ability to be secreted was reached.
• Fig. 4A: the protein names above the lanes should include the sfGFP that is fused to them.
• It would be preferable to show quantitative competition assays with statistics rather than pictures of a plate showing a single competition result, if conclusions or observations on minor differences in toxicity are made (for example, line 253: "The killing activity of Δtdei(Tde1GLGL-Tdi1) was largely compromised"). Since the authors performed each competition assay more than once, these data should be available to them.
• Fig. 5A: The author claim at the beginning of the manuscript (first results section heading: "Tde1 can cause DNase-independent growth inhibition of prey cells") that the N-terminal region of Tde1 is toxic on its own in the prey cell, yet in this competition assay Tdi1(M) shows no toxicity against the E. coli target cells. In the microscopy assay (Fig. 4B), it appears that a lot of Tdi1(M) enters the prey cell, since we can visualize it under the microscope. Can the authors clarify this discrepancy and explain why they do not expect to see target killing by this mutant even though they claimed it is toxic earlier?
• Fig. 5B: the GLGL mutant seems to have some residual toxicity, not dissimilar to what is shown in 5A. Why are these similar results interpreted differently (in 5A they are "largely compromised", while in 5B "killing activity... was not detectable")? Also, why was Tde(M)1-Tdi1 used in Fig. 5A but Tdi1(M) without the immunity gene used in Fig. 5B?
• Fig. 5: Does the remaining third effector, Tae, not play a role in these competition assays? If, as shown in Fig. 5C, the entire T6SS is less active when a GLGL mutant is expressed, couldn't the different in toxicity shown in Figs. 5A-B be the result of lack of Tae secretion and toxicity?
• Lines 359-362: T6SS effectors that bind the inner Hcp tube were suggested to be only partially folded.

Significance

The concept of T6SS effectors providing their own mechanism of transport from the cytoplasm to the periplasm is very interesting. It will appeal to audience in a wide range of microbiology disciplines, including those interested in toxins, membrane transport, and even translational applications. A similar concept was recently proposed and demonstrated for a domain that is also found in T6SS effectors (Atanaskovic et al., mBio, 2022).

Expertise: I have been studying the different aspects of T6SS for the past decade.

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer 1

Part of major comment 1. Unfortunately, not all the claims made are adequately supported by the data presented. In many experiments, the number of biological replicates is insufficient (sometimes n=1). This would have to be remedied prior to publication, to ensure the data can be properly interpreted.

The reviewer is specifically referring to our cell cycle analysis (as these are the only experiments where some of them only contain one replicate; e.g., figure 1D and figure 2E). In the coming months we will perform for each biological replicate two additional technical replicates to increase the number of measurements. Additionally, we will also perform three technical replicates for an additional GIMEN ATRX exon 2-10 clone and three technical replicates for two additional GIMEN ATRX exon 2-13 clones. Also, for SKNAS (Figure 2F) we will perform measurements for two more wildtype clones in triplicate. This will also increase the number of biological replicates where possible.

Reviewer 1

Part of major comment 1. Each data point should be indicated in bar graphs (for example in Figure 5, especially given the variability observed).

In figure 5 we now added the data points in the figures.

Reviewer 1

Major comment 2. The Western blot data is often very difficult to interpret, given that many bands are present in addition to the specific ones for the WT and FTT bands. Even for some controls presented as WT, the full-length protein is very faint while other bands predominate. This should be explained in the text. If no explanation is available, I would recommend confirming the results with other ATRX antibodies.

There are several other known isoforms of ATRX, namely around 250, around 200 and 150 kDa, we now mention them in the text (page 4) and in the main and supplementary figures we made the panels smaller (figure 1B and S3A), to remove all the non-important bands below the IFFs. A possible explanation for the faint full-length bands is that the mutant ATRX protein products are much stronger expressed and therefore during blot development the wild-type bands become faint. Previously, we already tested other ATRX antibodies, and they either showed similar patterns in bands (also many bands observed) or performed much worse.

Reviewer 1

Major comment 3. While the Western blot data suggests that ATRX protein products from MEDs are largely retained in the cytoplasm, this is not observed in the immunofluorescence pictures shown in supplementary figures. The authors should make a decision whether to provide more convincing and clear data, or to remove the immunofluorescence data.

We agree with the reviewer that the ATRX fractionation western blot have superior resolution over the stainings and therefore we decided to remove these stainings from the manuscript.

Reviewer 1

Major comment 4. The immunofluorescence data shown in supplementary figures are not of adequate quality. It is impossible to see much of what the authors are claiming. The Telomere and PML images are especially problematic.

We agree with the reviewer that without zooming in in the word file it might be hard to detect the co-localizations for the telomere and PML images. To resolve this, we made zoom-ins for single cells for the merged panels (only for CHLA-90, SK-N-MM and AMC772T2 co-localization can be observed; many studies only use the TelO staining as telomeric dots are often exclusively observed in ALT lines, however sometimes false positive can be observed and therefore including PML reduces the rate of false positives). Additionally, we also performed southern blots, which confirmed the telomere and PML stainings.

Reviewer 1

Major comment 5. More generally, the data is presented in a disorganized way, making it difficult to follow. Some are in main figures, some in supplementary, some experiments are done on only a subset of clones (i.e. cytoplasmic vs nuclear distribution). The authors should try to show all relevant results (for example western, facs data) for all their lines in the main figure, so that they can be compared, with adequate number of replicates and statistical analysis.

To improve on these points raised by the reviewer, we will perform additional cell cycle analyses to get an adequate number of replicates and we will perform statistical analyses. We also added the cell cycle analysis of SKNAS (old figure S5A) to main figure 2 and we added the western blot for yH2A.X of SKNAS and NB139 (old figure S4C) to main figure 2. Regarding the cytoplasmic vs. nuclear distribution, these experiments will be added for the NB139 models. We won’t perform these experiments for the SKNAS models since they are ATRX knockout.

Reviewer 1

Minor comment 1. Some grammatical errors should be corrected throughout.

We re-read the entire manuscript and changed the grammatical mistakes that we could detect.

Reviewer 1

Minor comment 2. Supplementary Table 1 was mislabeled as "Supplementary Figure 1"

We corrected this.

Reviewer 1

Extra comment. The authors should comment on the differences between the protein products (MED exon 2-10 vs MED exon 2-13) that could cause opposite transcriptional effects. What are the protein motifs that will be affected in one but not the other, and could this explain different effects on transcription, especially considering their claim that the majority of these protein products remain in the cytoplasm.

We added a paragraph to the discussion addressing this (page 18), but unfortunately no domains are currently known for the region of exon 11-13. Our claim that the majority of the protein resides within the cytoplasma is supported by the paper Qadeer et al., 2019.

Reviewer 2

Minor comment 1. Can the authors generalize these observations to other cancers with ATRX mutations?

In our discussion, we already mentioned increased ribosome biogenesis in glioma tumors with nonsense mutations, but we have included an additional sentence about these observations after that sentence (page 18).

Reviewer 2

Minor comment 2. RNA-Seq data for many cancers are now available, and so the authors could perform RNA-Seq analysis across ATRX mutant tumors and correlate with the type of ATRX mutation to see if the dichotomy they observed is present in patient data. This could be done for neuroblastoma and other tumors. The authors state that other tumors do not typically contain multi-exon deletions, but the effect of point mutations on the ATRX protein could similarly be non-uniform.

This is a nice suggestion but is beyond the scope of this study. Our manuscript already contains RNA sequencing data from neuroblastoma tumours (iTHER data), where we find decreased ribosome biogenesis for the two iTHER patients with an exon 2-10 deletion compared to ATRX wild-type neuroblastoma tumors. Nonsense mutations are rare in neuroblastoma, only ~20 patients with such a mutation have been reported in the literature, and our iTHER cohort does not contain any neuroblastoma tumors with ATRX nonsense mutations. More extensive analyses across tumor types might be difficult since many RNA-sequencing data sets only contain a few ATRX aberrant tumors (and combining distinct data sets is very challenging due to potential batch effects) and for the majority of the rare point mutations (nonsense and missense) and rare deletions no (RNA) sequencing data is available and therefore there will not be enough statistical power.

Reviewer 3

Major comments 1. In Figures 3 and 4, the authors showed two distinct gene set enrichment profiles in the ATRX deletion constructs ATRXΔ2-13 and ATRXΔ2-10. They used GI-ME-N WT clones C1 and C2 for Figure 4D, whereas in Figure 4E, they utilized WT clones C3 and C4. It is not clear from the above two Figures how WT C1, C2 are different with WT GI-ME-N C3 and C4 and share distinct gene signatures. The authors should put the supplementary Figure 15A into the main Figure 4 and use the same WT GI-ME-N clones while comparing the gene expression with ATRX KO or ATRXΔ2-13, or ATRXΔ2-10. Is the difference in gene signature between ATRXΔ2-13 and ATRXΔ2-10 due to the heterogeneity present in the WT GI-ME-N cells?

This might indeed be confusing. In our material and methods section, we addressed this under header: RNA sequencing analysis. Here we mentioned: “For the GI-ME-N clones, we observe a batch effect in the wild-type clones. Therefore, we decided to compare the different GI-ME-N ATRX aberrant models only with their corresponding wild-type clones (generated by same person).” To make this more clear, we now mention this in the result section (page 9). Our GI-ME-N models were generated in two batches (each batch by a different person, while the harvest and work-up of the RNA samples was performed by the same person on the same day for all samples) and therefore we decided to send two wild-type clones belonging to one batch and two clones belonging to the other batch (wildtype clone 3 and 4 were created by the same person as the GI-ME-N ATRXΔ2-10, while clone 1 and 2 were created by the same person as all the other GI-ME-N models). In our PCA plot for the GI-ME-N models (Supplementary figure 8B) we observe separation between wildtype clones 1+2 and wildtype clones 3+4 especially on PCA1, which explains the largest proportion of the variance. This clearly indicated a batch effect and therefore we compared the GIMEN ATRX aberrant clones with the batch corresponding wild-type clones. To exclude that the difference in ribosome biogenesis gene signatures between the different models was due to the heterogeneity present in wildtype GI-ME-N cells we also conducted the RNA analysis and GSEA for the distinct isogenic GI-ME-N models using all four GI-ME-N wild-type clones. This GSEA also showed ribosome biogenesis among the enriched gene sets (again down in ATRXΔ2-10 and up in ATRXΔ2-13 and knock-out). Nevertheless, the batch effect could have influence on other terms or single genes and therefore we needed to correct for this in all our analyses. Lastly, we now included supplementary figure 15A in main figure 4F.

Reviewer 3

Major comments 2. In Figure 3, the authors compare the differential gene expression and gene ontology analysis with ATRX deletion conditions. The authors should do the gene set enrichment analysis/Gene ontology term with WT vs ATRX KO or ATRXΔ2-10 or ATRXΔ2-13 and see whether the ribosome biogenesis pathway shows up. It is unclear from Figure 4B-E why authors have used two different cell lines for GO term comparison as their genetic background is different.

If we understand the reviewer correctly, the reviewer wants us to determine the differentially expressed genes (DEGs) by comparing wildtype versus the distinct ATRX mutant. This is what we already performed as DEGs are determined by comparing WT versus the distinct ATRX mutant. To make this clearer we included a new figure explaining how DEGs are determined in figure 3. Similarly, if we understand the reviewer correctly, the reviewer suggests us to perform gene set enrichment analysis (GSEA) by comparing the WT versus the different mutants, this we already did as GSEA is performed on the stat values (a value that both represents the fold change and significance and is advised for GSEA. These stat values are acquired by performing DEG analysis on wildtype versus mutant). An overview of how we used the DEG lists for GSEA is shown in figure 4A. For the last comment, we added a new sentence explaining why we compared two different cell lines for GO term analysis (page 11) in Figures 4B-E. This we specifically did because of the different genetic background, as we are only interested in DEGs that always change in ATRX aberrant models irrespective of their (epi)genetic background. The changes in expression of those overlapping genes are more likely the direct result of the ATRX aberrations.

Reviewer 3

Major comments 3. The ribosome biogenesis pathway is up-regulated in the ATRXΔ2-13 model. It is better to test their hypothesis in mice with a xenograft model with WT and ATRXΔ2-13 cell lines in combination with Pol I inhibitor or other well-known drugs which will inhibit the ribosome biogenesis and determine the effects on the growth of the tumor.

This is an interesting suggestion for a follow-up study but would take too much time to perform within the scope of this revision. Additionally, it would be of limited added value to the patients as only two patients with an ATRXΔ2-13 have been reported world-wide. Lastly, the most commonly used RNA polymerase I inhibitor Pidnarulex was recently discovered to inhibit topoisomerase 2B (TOP2B) instead of RNA polymerase 1 (DOI:10.1038/s41467-021-26640-x).

Reviewer 3

Major comment 4. In Figure 2D-E, cell cycle analysis was performed with ATRX WT and multi-exon ATRX deletions and there is an increased percentage of cells visible in the S phase compared to WT cells. Still, it is not clear from the Figure whether the result is statistically significant. The experiment should be repeated one more time and a statistical test should be done.

These comments were also postulated by reviewer 1. We will increase the number of measurements for our FACS experiments and include the statistical analysis (for more detail see the response to the comments of reviewer 1).

Reviewer 3

Major comments 5. As the ribosome biosynthesis was increased in ATRX KO/ ATRXΔ2-13 compared to ATRXΔ2-10. ATRXΔ2-13 deletion was generated only in GI-ME-N cell line model. To bypass the cell line-specific effect, it is necessary to prepare ATRXΔ2-13 deletion in other cell lines and validate whether the ribosome biogenesis pathway is still induced in another cell line.

Originally, we attempted to also create the ATRXΔ2-13 model in the NB139 cell line (we screened more than 70 clones, which were all wild-type), however generating such large deletion is extremely difficult as the efficiency is very low (several 100 kbs have to be removed). It would take too much time to generate such clones for the revision. Additionally, as mentioned above this ATRX aberration is less relevant to patients as it is very rare. However, the reviewer does have a good point about this limitation and therefore we have included a part in the discussion regarding this limitation (page 18, included in the new paragraph).

Reviewer 3

Part of major comment 6. Statistical analyses is missing from almost every Figure.

As mentioned above we will include the missing statistical analyses on the cell cycle analysis. Due to this comment, we noticed that we forgot to include the text about the statistical analyses in the legend of figure 5 (the p-values and statistics were mentioned in the result section), which we now changed in this revision.

Reviewer 3

Part of major comment 6. Statistical analyses is missing from almost every Figure.

This was also mentioned by reviewer 1. For the FACS measurements, we will first perform additional experiments before we add the statistical analysis. For figure 5 the p-values and statistics were mentioned in the result section, but we now also added this information to the figure legend.

Reviewer 3

Minor comment 1. The immunofluorescence labeling text in the supplementary Figures is not visible. The imaging should be done with confocal microscopy to avoid the background signal and to get a better resolution.

We changed these images and improved the labeling text. Additionally, we show a zoom-in for the merged figures to increase the interpretability. We decided not to perform confocal microscopy as we also performed telomere southern blots which are in agreements with our microscopy pictures.

Reviewer 3

Minor comment 2. Please put the appropriate color symbol in supplementary Figure 12A. Currently, the color symbol in the Figure panel does not match the Figure.

We understand that the current labeling of the ATRX status using circles might have led to confusion to the reviewer, therefore we changed the depiction of the ATRX status and also changed the order of the two legends.

Reviewer 3

Minor comments 3. The WT GI-ME-N clone should be consistent in all supplementary western blots.

The wild-types samples are only included in the western blots as a positive control and as reference to compare the mutant with. Re-performing all those western blots with the samen WT GI-M-EN would not lead to any changes in the conclusions. Therefore, we think it is not of added value to repeat these western blots.

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

Michael R. van Gerven and his co-workers studied the role of ATRX in neuroblastoma. Specifically, they focused on multi-exon deletions, the most frequent aberrations found in neuroblastoma. The multi-exon deletions often generate in-frame fusion protein with the potential gain-off function generation. To understand the importance of ATRX multi-exon deletions and mutations, the authors generated several isogenic cell lines using the CRISPR/Cas9 system and performed RNA-seq analysis. The gene set enrichment analysis showed increased gene expression in ribosome biosynthesis and metabolic processes in ATRX KO and ATRXΔ2-13 and deceased expression in ATRXΔ2-10 model. They validated the expression of ribosome biosynthesis genes with qPCR. They concluded that this study suggests the need for different therapeutic options for neuroblastoma patients.

Major Comments:

1. In Figures 3 and 4, the authors showed two distinct gene set enrichment profiles in the ATRX deletion constructs ATRXΔ2-13 and ATRXΔ2-10. They used GI-ME-N WT clones C1 and C2 for Figure 4D, whereas in Figure 4E, they utilized WT clones C3 and C4. It is not clear from the above two Figures how WT C1, C2 are different with WT GI-ME-N C3 and C4 and share distinct gene signatures. The authors should put the supplementary Figure 15A into the main Figure 4 and use the same WT GI-ME-N clones while comparing the gene expression with ATRX KO or ATRXΔ2-13, or ATRXΔ2-10. Is the difference in gene signature between ATRXΔ2-13 and ATRXΔ2-10 due to the heterogeneity present in the WT GI-ME-N cells?
2. In Figure 3, the authors compare the differential gene expression and gene ontology analysis with ATRX deletion conditions. The authors should do the gene set enrichment analysis/Gene ontology term with WT vs ATRX KO or ATRXΔ2-10 or ATRXΔ2-13 and see whether the ribosome biogenesis pathway shows up. It is unclear from Figure 4B-E why authors have used two different cell lines for GO term comparison as their genetic background is different.
3. The ribosome biogenesis pathway is up-regulated in the ATRXΔ2-13 model. It is better to test their hypothesis in mice with a xenograft model with WT and ATRXΔ2-13 cell lines in combination with Pol I inhibitor or other well-known drugs which will inhibit the ribosome biogenesis and determine the effects on the growth of the tumor.
4. In Figure 2D-E, cell cycle analysis was performed with ATRX WT and multi-exon ATRX deletions and there is an increased percentage of cells visible in the S phase compared to WT cells. Still, it is not clear from the Figure whether the result is statistically significant. The experiment should be repeated one more time and a statistical test should be done.
5. As the ribosome biosynthesis was increased in ATRX KO/ ATRXΔ2-13 compared to ATRXΔ2-10. ATRXΔ2-13 deletion was generated only in GI-ME-N cell line model. To bypass the cell line-specific effect, it is necessary to prepare ATRXΔ2-13 deletion in other cell lines and validate whether the ribosome biogenesis pathway is still induced in another cell line.
6. Statistical analyses is missing from almost every Figure.

Minor Comments:

1. The immunofluorescence labeling text in the supplementary Figures is not visible. The imaging should be done with confocal microscopy to avoid the background signal and to get a better resolution.
2. Please put the appropriate color symbol in supplementary Figure 12A. Currently, the color symbol in the Figure panel does not match the Figure.
3. The WT GI-ME-N clone should be consistent in all supplementary western blots.

Significance

The ATRX multi-exon deletions people have studied before in the context of neuroblastoma. But, in this manuscript, the authors showed for the first time the in-frame multi-exon deletions and their involvement in ribosome biogenesis using isogenic cell lines.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

This is an important study investigating the roles of ATRX mutations in neuroblastoma using isogenic CRISPR models. The authors introduced ATRX multi-exon deletions into neuroblastoma cell lines and characterized those cell lines and tumoroids using RNA-Seq, ALT assays, Western blot, and rRNA assays. The study found two different patterns of gene expression and a potential role for ATRX in ribosome biogenesis. The authors state that these findings are potentially very important for the clinic, as patients with the different types of ATRX mutations should be treated differently.

I found the study well-written and well-thought-out. I recommend the manuscript for publication.

Significance

This is a very important study for the field of neuroblastoma, but also for the pediatric field more broadly, as many tumors harbor mutations in ATRX.

Minor comments:

Can the authors generalize these observations to other cancers with ATRX mutations? RNA-Seq data for many cancers are now available, and so the authors could perform RNA-Seq analysis across ATRX mutant tumors and correlate with the type of ATRX mutation to see if the dichotomy they observed is present in patient data. This could be done for neuroblastoma and other tumors. The authors state that other tumors do not typically contain multi-exon deletions, but the effect of point mutations on the ATRX protein could similarly be non-uniform.

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 report, van Gerven et al have characterized several available neuroblastoma cell lines with or without ATRX multi exon deletions (MEDs) that produce in-frame fusions, as well as comparable CRISPR-Cas9 generated ATRX MEDs in diverse cell lines. They examined the length and cellular localization of ATRX proteins produced by Western blot, the ALT status by immunofluorescence staining and southern blot, proliferation with the violet trace approach and co-localization with HP1alpha (heterochromatin) by immunofluorescence staining. Finally, they compared transcriptional profiles of the different original and modified neuroblastoma cell lines. They observed that several MED products appeared to be largely cytoplasmic by Western blot. They observed no consistent changes in proliferation or S and G2/M phase length. Transcriptional profiling demonstrated that while MED exon2-10 resulted in a prolife most similar to ATRX-null cells, MED exons2-13 had a very different profile. Importantly, the effect on genes involved in ribosomal and metabolic processes were opposite between these two types of deletions.

Major comments:

1.Unfortunately, not all the claims made are adequately supported by the data presented. In many experiments, the number of biological replicates is insufficient (sometimes n=1). This would have to be remedied prior to publication, to ensure the data can be properly interpreted. Each data point should be indicated in bar graphs (for example in Figure 5, especially given the variability observed). 2.The Western blot data is often very difficult to interpret, given that many bands are present in addition to the specific ones for the WT and FTT bands. Even for some controls presented as WT, the full length protein is very faint while other bands predominate. This should be explained in the text. If no explanation is available, I would recommend confirming the results with other ATRX antibodies.<br /> 3.While the Western blot data suggests that ATRX protein products from MEDs are largely retained in the cytoplasm, this is not observed in the immunofluorescence pictures shown in supplementary figures. The authors should make a decision whether to provide more convincing and clear data, or to remove the immunofluorescence data.<br /> 4. The immunofluorescence data shown in supplementary figures are not of adequate quality. It is impossible to see much of what the authors are claiming. The Telomere and PML images are especially problematic. 5. More generally, the data is presented in a disorganized way, making it difficult to follow. Some are in main figures, some in supplementary, some experiments are done on only a subset of clones (i.e. cytoplasmic vs nuclear distribution). The authors should try to show all relevant results (for example western, facs data) for all their lines in the main figure, so that they can be compared, with adequate number of replicates and statistical analysis.

Minor comments:

Some grammatical errors should be corrected throughout.

Supplementary Table 1 was mislabelled as "Supplementary Figure 1"

Significance

A major finding from this study is that there is an opposite effect of MED exon 2-10 vs MED exon 2-13 on expression of genes involved in ribogenesis and metabolic processes. While a role of ATRX in ribogenesis is not new, as pointed out by the authors, it indicates that tumor states could be quite different depending on the type of ATRX MED fusion products and could potentially require very different therapeutic approaches. The authors should comment on the differences between the protein products (MED exon 2-10 vs MED exon 2-13) that could cause opposite transcriptional effects. What are the protein motifs that will be affected in one but not the other, and could this explain different effects on transcription, especially considering their claim that the majority of these protein products remain in the cytoplasm. It will be interesting to start exploring the location of these ATRX mutants on chromatin, chromatin structure changes and histone modifications, histone variants by ATAC-seq and ChIP-seq to better under understand the underlying mechanisms.

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

Learn more at Review Commons

---

Reply to the reviewers

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

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

This study by Ghosh et al. proposes a role for phosphatidylinositol 5-phosphate 4-kinase (PIP4K) in regulating PI3P levels in vivo. They use loss-of-function Drosophila model of the only PIP4K gene (dPIP4K29) to investigate the PI3P and PI(3,5)P2 metabolizing enzymes. First, they showed that loss of function of PIP4K leads to reduced cell size in larval salivary glands and this was attributed to the elevated level of PI3P. Then, they modulated enzymes involved in PI3P metabolism (kinases and phosphatases) and propose the implication of the PI3P phosphatase myotubularin (Mtm) and the Pi3k Class III (PI3K59F) in PIP4K-dependent cell seize control. Finally, as PI3P has an established role in autophagy, they modulate the autophagy related gene (atg1) and connect the observed increase of PI3P level to the upregulation of autophagy in dPIP4K29 model. The authors used genetic manipulations of dPIP4K29 models as well as specialized lipidomic expertise (phosphoinositide measurement using mass spectrometry and PI-kinase/phosphatase assays) to address their conclusions. The experimental strategies were well designed and major conclusions were in line with the obtained results.

Major comments:

• Are the key conclusions convincing?

Almost yes, however there is two major concerns for me: Concern 1 is about the level of PIP2/PI4,5P2, the product of PIP4K, in the dPIP4K29 model. This was not measured in the study. The authors claim page 5 that: "This observation suggests that the ability of dPIP4K to regulate cell size does not depend on the pool of PI(4,5)P2 that it generates... based on the fact that re-expression a mutation that hPIP4Kβ[A381E] in the salivary glands of dPIP4K29 (AB1> hPIP4Kβ[A381E]; dPIP4K29) (Figure S1A) did not rescue the reduced cell size. This mutation hPIP4Kβ[A381E] was generated in a study by Kunz et al. (2002) where they demonstrated by in vitro kinase assay that it cannot utilize PI5P as a substrate but can produce PI(4,5)P2 using PI4P as a substrate. In the same study, using MG-63 cells, Kunz et al. propose that the A381E mutation did not metabolize PI5P as it lost its plasma membrane localization. In my opinion the author should strength their claim about the role of dPIP4K independently of PI(4,5)P2 by addressing the level of PI(4,5)P2 in their model biochemically by mass spectrometry as they have this powerful tool and support this by using PH-PLCd probe to detect PI(4,5)P2. Also, as they use completely different model as Kunz et al. they should verify, if possible, the localization of hPIP4Kβ[A381E] vs WT PIP4Kβ in salivary glands.

Concern 2: Page 7: The author used Mtm tagged constructs (mCherry and GFP) and measure its phosphatase activity toward PI(3,5)P2 and they did not show any obvious activity. I would like to suggest the use of untagged (or small tag construct, Flag or HA) for the expression experiment in S2R+ cell as it is known that active myotubularins in other cell model as well as in vitro have a strong 3-phosphatase activity toward PI(3,5)P2. By looking at the graph FigS2 Bii, we could clearly see a big disparity within mCherry-Mtm data points. This experiment should be more strengthen by additional experimental points but also by using a positive CTRL where PI(3,5)P2 level drops (inhibition of PIKfyve by Apilimod).

Concern 3: Page 10: "we tagged dPIP4K with the tandem FYVE domain at the C-terminus end of the protein (dPIP4K2XFYVE) to target it to the PI3P enriched endosomal compartment and reconstituted this in the background of dPIP4K29. We did not observe a significant change in the cell size of dPIP4K29" I really don't understand the relevance of this experiment. FYVE tandem will bind to PI3P whenever it was in the cell (Lysosomes, autophagosome). Why the authors claim that the expression of restricted dPIP4K2XFYVE will be restricted to the endosomes. I think that this experiment is confusing and should be removed. - Should the authors qualify some of their claims as preliminary or speculative, or remove them altogether?

See concern 1 to 3. - Would additional experiments be essential to support the claims of the paper? Request additional experiments only where necessary for the paper as it is, and do not ask authors to open new lines of experimentation.

Yes, the proposed experiments in concern 1-3 are not difficult to address as the authors have all the appropriate tools to manage this. - Are the suggested experiments realistic in terms of time and resources? It would help if you could add an estimated cost and time investment for substantial experiments.

Yes. It is not time consuming and not costly according to their expertise, available tools and materials that they used through the study. - Are the data and the methods presented in such a way that they can be reproduced?

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

Yes

Minor comments:

• Specific experimental issues that are easily addressable.

  1. Address the level of PI(4,5)P2 in dPIP4K29 model by mass spectrometry.
  2. Address the localization of hPIP4Kβ[A381E] vs WT PIP4Kβ in salivary glands.
  3. Test the Mtm phosphatase activity toward PI(3,5)P2 using untagged or small tagged (HA or Flag) Mtm and repeat/homogenize the PI(3,5)P2-phosphatase assay (FigS2ii).
  4. Are prior studies referenced appropriately?

Yes - Are the text and figures clear and accurate?

The figures needsmore organization. - Do you have suggestions that would help the authors improve the presentation of their data and conclusions?

NO

Referees cross-commenting

Overall, Reviewer #1 and #2 found the study by Ghosh et al interesting well designed and written providing insights into the role of PIP4K in regulating cell seize. However, they comment few points that would be very helpful to improve the study. I am agreeing with both reviewers for the raised comments.

Significance

The author addressed how elevated PI3P in dPIP4K29 model impacted cell seize. Indeed, they connected this cell phenotype to the autophagy where PI3P plays a crucial role. However, I am still questioning how deletion of PIP4K enhances PI3P level.

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

The role of PIP4K in cellular homeostasis and organismal physiology is still unclear. This study brings additional insights into how PIP4K could be involved in important cellular process such as autophagy by regulating additional phsophoinositides.<br /> - State what audience might be interested in and influenced by the reported findings.

Phosphoinositide metabolism<br /> - Define your field of expertise with a few keywords to help the authors contextualize your point of view. Indicate if there are any parts of the paper that you do not have sufficient expertise to evaluate.

Phosphoinositides, Myotubularin, endolysosomal trafficking, skeletal muscle.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

The authors utilise a drosophila model to investigate the molecular mechanisms underlying the role of dPIP4K in regulating cell size. They suggest that PIP4K directly regulates PI3P levels in cells, which through upregulating autophagy, can reduce cell size. Overall, this is an interesting and well-designed study. The main downfall of this study is whether the regulation of PI3P levels by dPIP4K is occurs via direct or indirect mechanisms, which is unclear in the data provided in this study.

More specific comments are as follows:

Major Comments:

It's not clear why there are no differences in PI3P/PIP2 levels in Figure 4B, but this is overcome by normalising to organic phosphate levels (4C)? Can differences in PI3P/PIP2 levels be seen in Figure 4B without normalisation if additional controls such as PI3K59F/VPS34 KD were used (as done in figure 5B)? A discussion of this could be useful.

Figure 4D: Does the A381E mutant of PIP4K affect PI3P levels in cells as it cannot reverse the cell size phenotype in Figure S1B?

Figure 4G: The conclusion on line 255 that all phosphatase transcripts are unchanged in this figure when two of them appear to have significant reduction appears inaccurate. In addition, changes of transcript levels of these enzymes may not necessarily reflect their overall activity in cells. A localised reduction in MTM levels or activity may well play a role in dPIP4K29 cells even though an overall phosphatase activity is seen increased in the in vitro assay in Figure 4F. Similarly, not clear that the authors can completely rule out a potential activation of PIP3K59/vps34 and subsequent increase in PI3P levels in cells by simply looking at RNA levels. Is there a reason why the authors could not measure the enzyme levels in cells as mentioned in the text? VPS34 activity can be measured in mammalian systems. This is important as PI3PK59 KD does seem to reverse change in cell size (Figure 5A).

Another method to test the involvement of PI3K59/Vps34 is to target its adaptor proteins. Can the authors distinguish the endosomal and autophagosomal PIP3K59/vps34 complex and PI3P production by looking at drosophila homologues of Atg14 and UVRAG? The majority of PI3P in mammalian cells is found in the endosomal compartment rather than autophagosomal vesicles. If the authors predict that only autophagosomal PI3P levels are changed, then an overall change in enzymatic activity required for PI3P accumulation may not be easy to detect in total cell extracts.

Figure 5C&D: how specific is the FYVE domain fused probes to endosomal PI3P? Such probes are used in mammalian cells to measure overall PI3P, whether endosomal or autophagosomal. In addition, such probes when expressed in live cells can alter PI3P generation. In line with this comment, FYVE-domain probes can be used to quantify PI3P levels in fixed cells, this method could be used to verify changes in PI3P levels seen in PIP4K mutant flies.

Minor Comments:

Fig 1A: this is a slightly confusing diagram and could perhaps be made a little clearer. For an example, the arrows are not clearly differentiating phosphorylation from dephosphorylation events. Also, the choice of colour for the phosphatase arrows (brown-red) and kinases (also appearing brown-red) makes it harder to follow this figure.

Similar comment applies to S4B: PI could be depicted as an unphosphorylated version of PI3P/PI5P and drown in the centre.

Line 301: "lipidated Atg8a fuses with the formed omegasome" Atg8a fusion with omegasome is not an accurate description of the early autophagosome biogenesis events.

A new image (similar to Fig 1A) depicting how PIP4K affect PI3P levels to summarise the findings of this manuscript would be helpful.

The material and methods is an important section in this paper: a more thorough description of the methods, especially those referred to previous publications would be very helpful. The authors can at least add a brief outline of the methods they followed and include contents of buffers used.

Significance

Overall, this is a well designed and written study providing insights into the role of PIP4K in regulating PI3P levels and cell size in Drosophila. The authors develop interesting methods to measure endogenous levels of PI species, which can be useful for the wide research community. As I am not an expert in these Mass Spec analyses, it would be important for these assays to be thoroughly reviewed by a specialist to ensure that the methods used to quantify these phospholipids have been carefully controlled.

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:

Here Ghosh et al describe a potentially novel role for PIP4K in controlling PI3P in Drosophila tissues. Using PIP4K loss of function mutants and innovative lipid measurement techniques the authors try to address how cell size and autophagy are affected and report it may be through a PIP4K-regulated PI3P pool in flies.

Comments:

Overall, the paper is interesting, pretty well written and it has a lot of details in it that have addressed most of the questions, however they do refrain from stating that actually the PIP4Ks phosphorylate PI3P. Is it possible to measure the product PI3,4P2 if this were true? The authors claim that the regulation is direct but never show if by expressing dPIP4K it can phosphorylate PI3P to PI34P2. Using their optimized label-free LC-MS/MS methods this should not be trivial or by performing an invitro kinase assay.

Further, the authors claim there is an increase in autophagy in the dPIP4K animals however they only measured autophagosome numbers. Autophagy flux and lysosome functional assays need to be performed to accurately show this, as it has been demonstrated that the inhibition of PIP4kinases in mammalian cells does indeed cause an increase in the autophagosome pools but because of an autophagosome-lysosome fusion defect which ultimately impairs autophagy not increasing autophagy. This needs to be addressed in the fly system.

Also, localization studies with PIP4K in Drosophila should be performed to explain the role in autophagy or see if they localize at the same compartments as the enzymes that have been shown to regulate PI3P levels in flies.

Also of note, PI3P in mammalian epithelial cells has been shown to control cell size through regulation of autophagy (https://pubmed.ncbi.nlm.nih.gov/31941925/), but I guess it's novel in flies.

Significance

Overall, as it has already been shown that the PIP4K can regulate PI3P levels in vitro as well as PI3P has been shown to control cell size in mammalian cells so the novelty is diminished as well as how their results really impact autophagy are not complete as the authors only quantified Atg8a puncta. If the authors can show the activity in flies is real by measuring the product PI34P2 this would be compelling evidence. Also, they need to complete localization, autophagy flux assays, westerns of LC3 or p62, etc to accurately state that autophagy is enhanced.

Transparent Peer Review

---

Download the complete Review Process [PDF] including:

• reviews
• authors' reply
• editorial decisions

</br>

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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements

We want to thank the three reviewers for their thorough revisions provided, which have helped us improve the manuscript. Following their comments, we have pursued two new main lines of analysis, the results of which are included in the new version of the manuscript.

• We have now validated the signal of platinum CN footprints observed across tumor types in the Hartwig Medical Foundation cohort in an independent cohort of metastatic tumors (POG570). Despite differences between these two cohorts, the same signal of increased chromosomal fragments of size below 10 Mb is observed in two tumor types.
• We have extended the analysis of CN signatures to those recently identified by Steele et al., 2022 (https://www.nature.com/articles/s41586-022-04738-6) across primary tumors. This new analysis revealed that no CN signature shows significant different activity between tumors exposed or unexposed to most frequent anticancer therapies. A simple measurement –i.e., the number of CN fragments of a range of lengths– proves in this case more effective than more complex CN signatures that capture combinations of CN features to identify the signal of exposure to platinum.

Other minor points raised by the three reviewers have also been addressed in the new version of the manuscript (see below).

Reviewer #1 (Evidence, reproducibility and clarity):

In this study, the author examine WGS data from two cohorts of cancer samples from previous studies: PCAWG, mostly representing primary tumors, and HMF, representing metastatic tumors. The HMF dataset represented 4902 metastatic tumors (including 2709 whole-genome doubling, i.e., WGD, metastatic tumors) from patients who had been exposed to 85 anticancer therapies. The study identified a pattern of large LOH events associated with the exposure of tumors of some cancer types to platinum-based therapies. This pattern is characterized by a significant increase in the number of chromosomal fragments in exposed tumors with respect to their unexposed counterparts. The findings could support the hypothesis that WGD may provide tumors with an advantage to withstand the effects of structural variation.

We thank the reviewer for their accurate summarization of our manuscript.

Specific comments:

1. There are a number of statements that would suggest that there is some uncertainty regarding the robustness of results, and that the analysis of additional cohorts may be needed to substantiate the overall findings. For example, page 4: "It is also plausible that more numerous cohorts of exposed tumors are required to understand whether the observed differences are indeed robust." Page 5: "Further analysis with larger cohorts are required to clarify this point, which appears especially to clarify whether a significant imbalance in favor of deleted chromosomal fragments does occur across platinum-exposed lung tumors." However, the abstract does not seem to reflect this level of uncertainty in reporting the main findings.

We thank the reviewer for pointing this out. It is important to highlight that the first statement cited by the reviewer refers to the potential taxanes-related CN pattern, which is not mentioned in the abstract. The second statement refers to the fact that while we observe no significant differences of ploidy between platinum exposed and not exposed WGD tumors, we caution that this may change in larger cohorts. Following the reviewer’s observation, we have now deleted this sentence from the abstract. Therefore, in the current version of the manuscript, all statements presented in the abstract are robustly observed across tumors.

Moreover, we have now replicated the finding of the platinum CN footprint across the POG570 cohort (https://www.nature.com/articles/s43018-020-0050-6), the second largest cohort of whole-genome sequenced metastatic tumors available. (See answers below for details.)

1. Many of the findings made appear to apply not to all tumors but are found within tumors of specific cancer types. However, the abstract does not appear to note this.

As stated in the previous point, all statements contained in the abstract in the current version of the manuscript are replicated across platinum-exposed tumors of different cancer types across the Hartwig Medical Foundation metastatic cohort. Results on different tumor types with WGD samples exposed to platinum treatment are presented in Figures 2d and 3c and Supplementary Figure 5. A summary of the platinum CN footprint resulting from the aggregation of platinum CN patterns observed across all WGD exposed tumors is presented in Figure 3a.

We consider precisely this robustness of the CN pattern observed across platinum-exposed tumors of different cancer types as evidence supporting the proposed universal platinum CN footprint. Conversely, we do not propose a taxane-related CN footprint, precisely due to the lack of such robustness, as explained in the manuscript. We haver rewritten this part of the manuscript to make this point clearer.

Finally, as mentioned above, we have now replicated the platinum CN footprint across platinum-exposed tumors in the POG570 cohort (https://www.nature.com/articles/s43018-020-0050-6). Despite the limitations in sample size and the overall shorter time elapsed between the end of the treatment and the biopsy of the metastasis compared to the HMF cohort, as well as a different method for CN calling 2 out of 3 tumor types (breast and lung) with at least 10 platinum exposed and unexposed samples show exactly the same footprint while comparing platinum treated vs untreated. (See details below.)

1. With regards to additional cohorts, there is a POG570 cohort of WGS data on 570 recurrent or metastatic tumors (Nature Cancer 2020, PMID: 35121966), some 82% of which were from patients receiving systemic therapy before biopsy. Is it possible that some of the patterns identified using the HMF datasets could be validated in the POG570 datasets? If not, what numbers of tumors would be needed for the patterns of interest to be reliably identified?

We thank the reviewer for pointing us in the direction of this very interesting dataset. We have now analyzed the POG570 cohort (https://www.nature.com/articles/s43018-020-0050-6) using the clinical and treatment data and chromosomal fragment calls provided by the authors. Briefly, for each tumor, we computed the length of each chromosomal fragment identified and then counted the number of fragments with copy number 1-4 and length below 10 Mb, which constitutes our measurement of the intensity of the platinum CN footprint. Then, we compared the number of these fragments across platinum exposed and unexposed colon, breast and lung tumors. We selected these tumor types because only for them are there at least 10 exposed and unexposed tumors in the POG570 cohort.

The results of this comparison are shown in the new Figure S7a of the manuscript, which we reproduce here.

(See PDF version included as Supplemental Material)

We found that platinum-exposed breast and lung tumors have a significantly higher number of chromosomal fragments of CN 1-4 and length below 10Mb than their unexposed counterparts, thus replicating our observation across HMF tumors. In the case of colon tumors the difference is not significant. Several differences in the composition of the cohorts and the analysis of the data must be taken into account in the analysis of these results. On the side of the data, the calls of chromosomal fragments have been done using different algorithms. On the side of cohorts composition, one important difference is the number of days elapsed between the end of the exposure of the patients to platinum and the moment of the biopsy of the metastatic or recurrent tumor. The differences between the three tumor types analyzed across both cohorts are now represented in Figure S7b of the manuscript, reproduced here.

(See PDF version included as Supplemental Material)

Across the HMF cohort, colorectal tumor patients exposed to platinum have a median of 300 days between end of treatment and biopsy of the metastasis, but only 178 across POG570 colon tumors (difference close to significance). The same gap is appreciable across breast tumors (289 days vs 41 days; significantly higher across HMF patients) and lung tumors (242 days vs 187.5 days; borderline significantly higher across HMF patients).

In a previous work (https://www.nature.com/articles/s41467-021-24858-3) we demonstrated that longer time elapsed between the end of the treatment and the biopsy accounted for a higher likelihood of a full clonal expansion upon treatment and, as a consequence a higher probability to detect the SBS platinum footprint through bulk sequencing. Since the same must apply to the CN footprint, the significantly shorter time between the end of treatment and biopsy across POG570 tumors would make its detection more difficult. Nevertheless, it is still nicely reproduced at least across breast and lung tumors.

In summary, despite the differences between the HMF and POG570 cohorts, the platinum CN footprint is replicated across tumors of the latter, thus providing independent support to its presence in platinum-exposed tumors.

1. The PCAWG cohort is described as comprising all primary tumors, but in fact there are some metastatic tumors in PCAWG cohort. In particular, most of the TCGA melanoma (SKCM) samples are metastatic (PMID: 30401717). This may have bearing on using comparisons between PCAWG and HMF as a surrogate for primary versus metastases.

The reviewer is correct that there is a very minor representation of metastatic tumors in the PCAWG cohort, specifically, across melanomas. It is nevertheless a good choice for a whole-genome sequenced cohort of tumors, which has been used repeatedly in primary-to-metastatic comparisons (see, for example, PMID: 31645765; https://www.biorxiv.org/content/10.1101/2022.06.17.496528v1). Since the vast majority of tumors in the PCAWG cohort are primary, and the comparisons presented in Figure 1 encompass the entire cohort (with the exception of Fig. 1b, where melanomas are not included), the influence of a few tumors are not expected to confound the results. In any case, differences between primary and metastatic tumors are still very apparent in Figure 1c; exclusion of PCAWG melanomas, if anything, would make them still more apparent.

1. For each boxplot, the number of tumors represented in each group should be indicated somewhere (e.g., along the bottom).

We thank the reviewer for bringing this oversight to our attention. We have now added the number of tumors in each group in all relevant figures. Specifically, the number of tumors in each group appear now in Figures 1c and d, 2d and e, 3 c and d, and 4 a and b, as well as in relevant Supplementary Figures.

1. For Figure 1a, is a color legend needed here?

Thanks for bringing this to our attention. We have now added a color legend to represent WGD and near diploid (non-WGD) tumors in this panel.

1. For analyses comparing HMF to PCAWG (e.g., Figure 1c), the p-values ought to corrected for cancer type (e.g., using a linear regression model with cancer type as a factor).

We thank the reviewer for the suggestion to look at the influence of tumor types in the observed differences in the fraction of LoH tumors in primary and metastatic cohorts. Following their suggestion, we have now carried out separate comparisons of the overall ploidy and fraction of genome LoH of tumors per cancer type (Fig. S1b-i in the reviewed version of the manuscript). This analysis is, of course, limited to tumor types represented in both PCAWG and HMF cohorts. As in the pan-cancer analysis, no significant differences in ploidy are observed in any cancer type. Conversely, significant differences in LoH fraction appear in colorectal, prostate and kidney WGD primary and metastatic tumors. Regardless of statistical significance, in the majority of cancer types, the fraction of LoH genome across tumors appears greater in metastasis than in primaries. Therefore, our starting observation that more LoH is observed across metastatic than primary WGD tumors and that this could be related with the former’s exposure to anticancer therapies holds in this per-tumor-type analysis.

1. For Figure 1d, are the numbers of tumors in each category indicated in parentheses?

The reviewer is correct. We have clarified this in the caption of Figure 1d.

1. For figures 2d and 2e legend, the numbers of tumors in exposed vs unexposed groups for each category should be indicated. Similar for Figures 3a, 3c, 3d.

We thank the reviewer for pointing out this oversight. We have now added the numbers of exposed and unexposed tumors in the relevant plots in Figures 2 and 3. Moreover, we have added a new Supplementary Table (Table S2) with the numbers of tumors of each cancer type exposed and unexposed to each treatment across the HMF cohort.

1. For figure 2c, what is the statistical test used and multiple testing correction applied? Could this be noted in the figure legend?

Following the reviewer’s observation, we have included the name of the test (Wilcoxon signed-rank test) in the caption of Figure 2c. The p-values shown are corrected for False Discovery Rate: this is now indicated in the Figure caption.

Reviewer #1 (Significance):

The study makes effective use of public genomic resources to make new observations regarding platinum-based anticancer therapies. The observations identify patterns within specific cancer types. The analysis is exploratory in nature and would benefit from independent observation in an independent cohort, though it is not clear whether such cohorts may exist in sufficient numbers.

As explained above in detail, motivated by this comment by the reviewer, we have now validated the platinum CN footprint across an independent cohort of metastatic tumors (POG570).

Reviewer #3 (Evidence, reproducibility and clarity):

Analyzing the genome wide copy number patterns across publicly available ~2700 primary and ~5000 metastatic tumors treated by a number of different classes of chemotherapy agents, the authors find a distinct signature of CNVs in tumors treated with platinum-based agents. These platinum-exposed tumors are characterized by a significant increase in the number of chromosomal fragments of lengths between 10 Kb-10 Mb, and this tendency correlates with dosage (approximated by previously published platinum induced mutational signatures). Also, it is interesting that comparison of WGD with non-WGD treated-vs-untreated samples shows that WGD samples tolerate larger CNVs, suggesting relaxed selection against large CNVs in WGD (or WGD as a mechanism to accumulate large CNVs).

Previous works have focused on mutational signatures of various environmental exposures and drugs. This paper attempts to extend the previous research by looking at patterns of copy number variations. The work is somewhat motivated (see comment below) and the experimental design and execution are reasonable.

The manuscript is well written. The method section could be elaborated more for reproducibility.

We thank the reviewer for their appreciative comments on our manuscript. Following their observation, we have carefully reviewed the methods section thinking on the reproducibility of our results.

Major comments:

1. Overall, the results are modest. Although statistically significant, the increase in specific classes of CNVs in treated v untreated WGD mets is modest (Fig 2d), casting a doubt on clinical significance.

While the statistical significance of the association of independent CN features to the exposure to platinum-based drugs is not as high as that observed for platinum-related single nucleotide variants (https://www.nature.com/articles/s41588-019-0525-5), this does not mean that the signal is modest. When the number of CN fragments of length below 10Mb with CN 1-4 of tumors exposed and unexposed to platinum (across three cancer types) are compared, the signal is very clear (Figure 4a). Across cancer types, these particular chromosomal fragments are more abundant (significantly in most cases) in platinum-exposed tumors than in their unexposed counterparts. The increase across exposed tumors is between 13% and 387%.

The number of CN fragments in this range of sizes that may be identified due to the exposure to platinum is much more stringently limited by the size of the genome than the corresponding number of SNVs. This is why while we observe thousands of platinum-related SNVs in exposed tumors(https://www.nature.com/articles/s41588-019-0525-5), the numbers of observed CN fragments are smaller, making the signal less strong. However, while each single nucleotide variant affects a single nucleotide, a chromosomal fragment in the middle of the range observed would affect thousands of base pairs. This makes the cumulative effect of the platinum CN footprint on exposed tumors and normal cells is much larger than that of single nucleotide variants. In other words, the effect of the exposure to platinum on the landscape of CN fragments, far from modest is more consequential than that of SNVs. Thus, while a clinical application of the identified platinum CN footprint is out of the scope of this work, we do believe that, like its mutational footprint counterpart (described in the abovementioned papers) it does have implications for chemotherapy survivors.

To clarify further the effect size of the exposure to platinum, following the reviewer’s comment, we have now added a fold-change to the comparisons between exposed and unexposed tumors presented in figure 4a. Furthermore, we have added the following consideration to the last paragraph of the Discussion section:

Moreover, these SVs –as described by the platinum CN footprint– are bound to affect much larger genomic portions than platinum-contributed point mutations (6, 11). Therefore, their impact on exposed healthy cells and on the development of late effects of the chemotherapy could potentially be greater than those caused by previously recognized platinum-related SBS footprints.

1. Three of the four drugs that yielded significant patterns seems to have largest sample sizes (Fig 2a). Is there a link between sample size and detection power? In general, robustness of the signals is not analyzed, relative to subsampling of tumors or genomic regions etc. Indeed, the authors have noted the potential lack of robustness somewhere in the manuscript.

The reviewer is right that the sample size is an important limiting factor for the detection of CN patterns related to anticancer therapies. This is much more acute than in the case of footprints of single base substitutions, thousands of which are contributed by platinum (for example) to the genome of exposed cells. In contrast, limited by their sheer size, only a few dozen extra chromosomal fragments are contributed by the same treatment to metastatic tumors. This is the reason why, rather than carrying out a subsampling exercise, we have resorted to identifying CN patterns in the tumors of different tumor types separately. As suggested by the reviewer, we have only considered robust the CN pattern that is detected across all cancer types with exposed samples, that is, that of platinum-based therapies.

Adding robustness to the detected platinum CN footprint, we have now replicated its finding in a totally independent cohort of tumors, the POG570 cohort. See above answer to point 3 raised by reviewer 1.

Moreover, we include a paragraph in the Discussion section dedicated to comment on the question of power for the detection of the CN footprints associated to other therapies.

1. Given inter-individual heterogeneity, analyzing longitudinal data of pre-treated primary and post-treated mets rom same individuals would really help strengthen the findings.

We agree with the reviewer that such a comparison would be very interesting. Unfortunately, pretreatment samples of the primary tumors of patients in the Hartwig Medical Foundation cohort are not available.

1. The authors show that several CN features show and increase upon platinum treatment. Are these independent observations? A global correlation among the 48 features in various samples classes (WGD/non-WGD, Primary/Met, treated/untreated) should be done and equivalence class of features defined. Otherwise, the biological significance of these observations could be overstated.

The reviewer is right that there is correlation between the CN features used to identify the CN footprint. Nevertheless, these features, which have been defined elsewhere (https://www.nature.com/articles/s41586-022-04738-6) for the identification of CN signatures are only used in the context of our analysis to determine that some of them (despite their potential correlation) are different between platinum exposed and unexposed tumors. Precisely, taking into account the correlations between CN features, our conceptualization of the platinum CN footprint is the number of chromosomal fragments with copy number between 1 and 4 with length below 10 Mb. Moreover, using this definition and not the original CN features, we are able to replicate the observation of the platinum CN footprint across an independent cohort, which provides further robustness to its identification.

In our work, in summary, the CN features are only a means to the end of identifying a quantitative difference in the structure of chromosomal fragments between tumors exposed or unexposed to a certain anticancer therapy.

1. The only mechanistic link between platinum treatment and the observed CNV patterns is speculated to be via platinum-induced DNA breaks and errors during correction. This seems like a very general mechanisms applicable to any exposures (environmental or drug) that induces breaks. This lack of specificity makes it hard for me to understand the rationale to study CNV patterns - why, after all, should one expect to see a CNV signature?

The question posed by the reviewer –are there any treatment related CN footprints?– is precisely the starting point of our study. We thus carried out an unbiased discovery of CN patterns related with the exposure to different treatments. Upon identification of the association between the exposure to platinum and the increase of LoH chromosomal fragments of 10kb-10Mb signatures with different copy number, we hypothesize that the platinum-induced increase of double strand breaks and their faulty repair may be the underlying mechanism. We absolutely agree with the reviewer that other therapies inducing double strand breaks could lead to a similar –or other– CN footprint. Nevertheless, we have not been able to detect other consistent CN footprint associated with any anticancer therapy across tumors in the Hartwig Medical Foundation cohort. Whether this is due to the lack of statistical power or some underlying mechanistic difference between platinum-based and other drugs (see for example the causes underlying differences in the detectability of platinum and 5FU-related footprints; https://www.nature.com/articles/s41467-021-24858-3) we are not currently able to answer.

1. Authors should contrast their findings with those in https://www.nature.com/articles/s41586-022-04738-6.pdf

We thank the reviewer for this suggestion. Actually, taking advantage of the fact that the tool employed to extract CN signatures de novo (the original SigProfiler aimed at mutational signatures extended to CN features) was available prior to the publication of this article, we had already carried out a CN signatures extraction de novo from the HMF cohort. We then asked if any of these CN signatures (their activity across tumors) is significantly associated with the treatments in the cohort, and found none (current Fig. S3a). In the manuscript we hypothesize that this is due to the intrinsic difficulties in defining CN signatures, as opposed to SBS and DBS signatures. This is why we decided in our study to focus on a collection of individual CN features that show differences between platinum-exposed and unexposed tumors to define the platinum CN footprint.

Following the suggestion by this (and other) reviewer, we have now carried out the same analysis (identification of CN signatures potentially related to exposure to anticancer therapies) using the set of CN signatures originally defined by Steele et al in their paper (reference 21 in our manuscript). This analysis yields negative results too (current Fig. S3c). We also now include the equivalence (established through linear reconstruction) between the CN signatures extracted de novo from the HMF cohort and the CN signatures originally defined by Steele et al. across primary tumors. (This equivalence is provided by the SigProfiler upon extraction.) In general, the signatures extracted across HMF tumors bear little resemblance to those extracted from primary tumors (highest cosine similarity of a linearly reconstructed signature, 0.775). This is presented in current Figure S3b.

Taken together, these results further strengthen our point that a CN footprint defined from differences in individual CN features are probably more appropriate than CN signatures in their current format to identify the effects of anticancer therapies on the CN landscape of exposed cells.

1. For pyrimidine treated samples... "significance is lost across non-overlapping tumors". Authors should ascertain that this is not simply a matter of power. Also, would the significance not be lost for non-overlapping platinum-treated samples?

We thank the reviewer for pointing out the lack of clarity in our statement. To solve it, we have included a new Supplementary Table (Table S4) containing the number of WGD tumors exposed to different pairs of anticancer therapies across cancer types in the HMF cohort.

Let’s look at three cancer types showing the platinum CN footprint with different degrees of overlap of platinum and pyrimidine analogs exposed tumors. In the case of colorectal tumors, 194 out of 220 pyrimidine analogs exposed WGD tumors are also exposed to platinum. No signal is observed when the 26 tumors solely exposed to pyrimidine analogs are compared to tumors that are unexposed to pyrimidine or platinum (as shown in the Figure below, the p-values of which correspond to a two-tailed Wilcoxon-Mann-Whitney test).

(See PDF version included as Supplemental Material)

The reverse analysis is impossible, as only 4 WGD tumors are exposed to platinum but not to pyrimidine analogs. In the case of lung tumors, the 22 tumors exposed to platinum and pyrimidine analogs constitute the entire pyrimidine analogs exposed set. When only the 84 tumors exposed solely to platinum are compared to tumors unexposed to platinum or pyrimidine analogs, CN features associated to platinum exposure still appear different. Finally, in the case of ovarian tumors, no exposure to pyrimidine analogs is recorded, as it is not employed in the treatment of this malignancy.

In summary, the significance of platinum-related CN features is present when only platinum-exposed tumors are included in the comparison. The signal observed is thus attributable to the exposure to platinum-based drugs. The number of exclusively pyrimidine-exposed tumors are few across tumor types, and thus at this stage we are not able to rule out the existence of a pyrimidine associated footprint.

1. Interpretation of Fig 3b. "Had this increase in the number of 10Kb-10Mb chromosomal fragments across exposed tumors arisen through positive selection, we would expect to observe a concentration at specific genomic regions containing resistance genes." This needs to be tested specifically. A "concentration" perhaps would not jump out in a visual inspection of the global pattern, which does seem to show variability.

Following the reviewer’s suggestion, we have now compared the number of chromosomal fragments of CN 1-4 and size smaller than 10 Mb observed in each chromosome across platinum exposed or unexposed lung and colorectal tumors. The results of these comparisons are presented in Figure S4a,b. This figure shows that more fragments of this size range are observed for all chromosomes across exposed tumors than across their unexposed counterparts. In most cases the differences are significant. This means that the increase of chromosomal fragments of size below 10Mb is not restricted to one or few chromosomes. It is rather a general effect distributed along the entire genome.

Minor comments:

1. "the ploidy of tumors with WGD varies in a range between 2.9 and 3.6 (quartiles 1 and 3, Fig. 1a)". I am not sure, there seem to several orange (WGD) points with ploidy below 2.9.

The reviewer is correct that several WGD tumors possess a ploidy below 2.9. This is because the cited values 2.9 and 3.6 correspond, respectively to the lower and upper limit of the first and third quartiles. In other words, 25% of all WGD tumors possess ploidy below 2.9. Following the reviewer’s comment, we have clarified this statement.

Reviewer #3 (Significance):

The novelty is to look at CNV signatures upon drug treatment (beyind mutational signatures). However, as mentioned above, it is not clear how different exposures that ultimately cause DNA break would have distinct CNV pattern. Overall, the results seem modest to me. Although statistically significant, the increase in specific classes of CNVs in treated v untreated WGD mets is modest (Fig 2d), casting a doubt on clinical significance. This work could still be of interest to some researchers, in particular, those interested in mutational signatures of environmental exposure. This work should be interpreted in the context of pan-cancer signatures of CNVs very recently published https://www.nature.com/articles/s41586-022-04738-6.pdf.

The increase of chromosomal fragments below 10 Mb across platinum-exposed tumors is between 13% and 387% with respect to unexposed tumors (Fig. 4a). The statistical significance of the signal of platinum exposure on the number of CN fragments is smaller than that observed for single nucleotide variants produced by the exposure to platinum (https://www.nature.com/articles/s41588-019-0525-5). However, while each single nucleotide variant affects a single nucleotide, a chromosomal fragment in the middle of the range observed would affect thousands of base pairs. In other words the cumulative effect of the platinum CN footprint on exposed tumors and normal cells is much larger than that of single nucleotide variants. (See also response to point 1.)

With respect to CN signatures, motivated by the reviewer’s comments we now demonstrate, using the activity of CN signatures extracted from the HMF cohort (using the methodology presented in https://www.nature.com/articles/s41586-022-04738-6) that none of them is significantly different between tumors exposed or unexposed to major anticancer therapies. Note are there any significant differences in the activities of the original CN signatures extracted in the aforementioned paper across primary tumors between exposed and unexposed tumors in the HMF cohort.

My background is in computational biology, working on transcriptional regulation for decades and more recently in cancer systems biology. I am comfortable with the techniques employed in this work but not so much with the mechanisms linking a specific drug to specific copy number signatures and also with the clinical significance of this problem. Keywords: "Computational biology", "Bioinformatics", "Transcriptional regulation", "NGS", "Omics", "Cancer systems biology"

Reviewer #4 (Evidence, reproducibility and clarity):

Summary:

This paper by Gonzalez et al attempts to identify copy number footprints of anti-cancer therapies. It follows previous work by the group looking at single base mutational footprints of anti-cancer therapies, which provides clear evidence of the effect of these drugs on the genome. This study into copy number footprints is less convincing, mainly due to the challenges in identifying these low frequency copy number signatures. The authors present weak evidence using CN signatures for an increase in the number of chromosomal fragments less than 10 Mb in size. However, this is not consistently significant between different cancer types treated with the same drugs. The interesting finding of an effect of platinum treatment intensity on copy number is seen quite nicely in Fig 4b when pooled into one simple effect. Signature analysis in this case seems unnecessary as the main finding is that platinum treatment results in increased 10 kb- 10 Mb fragments but only when pooled in this way. The paper is otherwise nicely written, although some clarity adjustments are required in the Figures and Figure legends.

We thank the reviewer for their appreciative summarization of our work.

Major comments:

Whilst a commendable effort has been made to identify copy number footprints, the evidence presented here for the identification of CN signatures is not so convincing. The main focus of the paper is the effect of Platinum based therapies and yet the two featured cancer types lung non-small cell and colorectal do not have consistent significant effects in the signature analysis.

The reviewer is correct that we don’t identify a CN signature (in the sense understood in recently published manuscript by Steele et al. https://www.nature.com/articles/s41586-022-04738-6) associated with platinum treatment. A clearer statement to this effect has now been added to the manuscript as a result of novel analyses of these CN signatures in the cohort studied in our work.

What we identify (as the reviewer states in their summary of our work) is a general increase of chromosomal fragments below 10Mb among platinum-exposed tumors. This is consistent across tumor types as shown in Figure 4a, and it is what we describe in the manuscript as the platinum CN footprint. We precisely avoid the term signature in an effort to prevent confusion with the canonical usage of this term.

The referenced BioRXiv paper by Steele et al. is now published https://www.nature.com/articles/s41586-022-04738-6 and one wonders whether additional methods and analyses performed during their peer review may be useful in this paper as well. Can reanalyse using the predefined 21 CN signatures from Steele et al?

We thank the reviewer for this suggestion. Actually, taking advantage of the fact that the tool employed to extract CN signatures de novo (the original SigProfiler aimed at mutational signatures extended to CN features) was available prior to the publication of this article, we had already carried out a CN signatures extraction de novo from the HMF cohort. We then asked if any of these CN signatures (their activity across tumors) is significantly associated with the treatments in the cohort, and found none (current Fig. S3a). In the manuscript we hypothesize that this is due to the intrinsic difficulties in defining CN signatures, as opposed to SBS and DBS signatures. This is why we decided in our study to focus on a collection of individual CN features that show differences between platinum-exposed and unexposed tumors to define the platinum CN footprint.

Following the suggestion by this (and other) reviewer, we have now carried out the same analysis (identification of CN signatures potentially related to exposure to anticancer therapies) using the set of CN signatures originally defined by Steele et al in their paper (reference 21 in our manuscript). This analysis yields negative results too (current Fig. S3c). We also now include the equivalence (established through linear reconstruction) between the CN signatures extracted de novo from the HMF cohort and the CN signatures originally defined by Steele et al. across primary tumors. (This equivalence is provided by the SigProfiler upon extraction.) In general, the signatures extracted across HMF tumors in general bear little resemblance to those extracted from primary tumors (highest cosine similarity of a linearly reconstructed signature, 0.775). This is presented in current Figure S3b.

Taken together, these results further strengthen our point that a CN footprint defined from differences in individual CN features are probably more appropriate than CN signatures in their current format to identify the effects of anticancer therapies on the CN landscape of exposed cells.

From Fig 2a there are 5 cancer types in HMF with Platinum treatment: Lung non-small cell, colorectal, esophagus, urothelial and ovary and it is not clear if all these cancer types are combined or separated in the final analysis. All but urothelial are featured at some point though, but e.g. ovary has no significant differences between treated and untreated. Are samples from all 5 cancer types combined in the "treated versus untreated" analyses?

We thank the reviewer for pointing this out. The analysis is carried out separately by cancer type. We have now included a statement in the Profiles of chromosomal fragments associated with anticancer therapies section clarifying this.

The strongest evidence for a real effect on copy number for platinum treatment comes in Figure 4, where there is a significant increase in LoH segments CN 1-4 with samples showing high SBS 35 mutations (a clever idea!). Attempting to separate out the samples into a "Copy number signature" in Figures 2 and 3 seem a bit like fillers to get to this actual potentially interesting finding. What is the benefit of separating out the copy number and zygosity when the real effect is much clearer when you pool everything and simplify it?

As the reviewer points out, and we highlight above, it is this type of chromosomal fragments that we conceptualize as the platinum CN footprint. This, however, is a discovery that stems from the unbiased analysis carried out across tumor types and treatments, the results of which are presented in Figure 2 and which is further characterized in Figure 3. It would have been impossible to identify this footprint from the outset. We reasoned that the CN features defined by Steele et al. were a good starting point to capture differences in the overall landscape of chromosomal fragments of tumors exposed or unexposed to DNA damaging drugs.

Can you investigate other drug treatments using this bulk approach using the proxy of the SBS drug mutations to indicate the "strength" of the mutational process of the drug.

In theory this is possible for treatments that leave both discernible mutational and CN footprints. So far, only platinum-based drugs fulfill this criteria. In the case of 5-FU a salient mutational footprint is associated with the exposure to the drug, but we were unable to identify any discernible CN footprint.

Another point of interest is the overlap between treatments. A judgement call is made as to which is the overriding drug corresponding to the effect. Is it possible to separate these effects with NMF as per SBS? Or could a combination effect be detected? Likely the numbers would be too low for this separated analysis. But just looking at LoH 10kb to 10 Mb might show something?

This is a very interesting suggestion. Several lines of evidence support the idea that the observed CN footprint is associated with exposure to platinum and not 5-FU and that it is not a combination of both drugs (see above response to point 7 raised by reviewer 3). The most important is that the CN footprint is also observed across tumors that are not exposed to 5-FU. For example, in the case of the ovarian tumor cohort, which are not exposed to 5-FU the CN footprint is recovered (Fig. 3c; 4a), although the individual CN features are not significant, due to the low numbers.

With respect to the overlap specifically between platinum-based therapies and pyrimidine analogs, we have included in the manuscript a new Supplementary Table (Table S4) presenting the numbers of WGD tumors exposed to different pairs of drugs across cancer types. We have also extended the statements about the overlaps in the manuscript to further clarify the decisions made. (See above our reply to point 7 by reviewer 3.)

A table summarizing the included samples, treatments, overlap, etc per cancer type is missing.

Following the reviewer’s suggestion, we have prepared and included in the manuscript two new Supplementary Tables. Table S2 details the number of WGD and non-WGD metastatic tumors of each tumor type across the HMF cohort exposed to different anti-cancer therapies. Table S4 presents the number of tumors exposed to different pairs of treatments across tumor types in the cohort.

Your study may also benefit from a comparison to the latest Hartwig cohort paper https://www.biorxiv.org/content/10.1101/2022.06.17.496528v1.full particularly focusing on the suggestion of Treatment enriched drivers (TED) some of which are infact copy number driven.

We thank the reviewer for this interesting suggestion. The only TED identified by the authors, which is related to platinum-based drugs (with the criteria described in the Supplementary Table 8 of their manuscript) concerns point mutations of TP53 across metastatic stomach adenocarcinomas (where we are unable to identify the CN footprint due to a low sample size). Although some driver amplification or deletion events do appear significantly enriched across platinum exposed tumors of different cancer types, they are discarded by the authors due to lack of orthogonal evidence of being associated with the specific mechanism of action of the drug.

We now include this observation in the revised manuscript:

As anticipated, we observed that chromosomal fragments smaller than 10 Mb (representative of the platinum CN signature) are evenly distributed along the genomes of WGD colorectal and lung tumors (Fig. 3b; Fig. S4a,b). Had this increase in the number of platinum-related chromosomal fragments across exposed tumors been constrained to one or few genomic regions, it would point to positive selection of one or more resistance-associated genes. A recent systematic analysis of the HMF cohort revealed that only mutations in TP53 across stomach adenocarcinomas appear as a potential bona fide driver event associated with the exposure to platinum in the HMF cohort (Martínez-Jiménez et al, 2022).

Minor comments:

The Figure legends and Figures themselves need to be altered for clarity. Axis should be labelled more specifically, e.g Fig 1d axis currently reads "percentage". Fig 3b says left and right and there is no such thing. What do the sizes of the circles in 2c represent? Can you indicate cancer type as well in this plot (shading or line type) or are all treated samples pooled- this is not clear?

We thank the reviewer for this suggestion. We have checked all figures and figure legends to enhance their clarity.

It is not clear why Fig 2a only includes HMF samples and not PCAWG- PCAWG could be in supplement?

Figure 2a describes the types of anticancer treatments received by patients bearing different types of malignancies. PCAWG tumors are primary and treatment-naive; this is why all the study to identify treatment-related CN features focuses on the HMF cohort.

Be consistent with labelling as well, in the text everything is referred to as 10 kb -10 Mb and some Figures labeled as such but others with 10^4- 10^7. How is the size 10 kb established? All the plots show 0-100 kb, where did the 10 kb limit come from?

We thank the reviewer for this recommendation. It actually led us to review our definition of the platinum CN footprint and to realize that, indeed, fragments smaller than 10 kb are part of this footprint. All analyses (and relevant figures and supplementary tables) have been updated accordingly. The rationale to define the CN footprint is now more thoroughly explained (Fig. 3a).

Why is ovary not in 4b?

There are very few platinum-exposed ovarian tumors with WGD and activity of SBS35. Therefore, the groups of tumors with high and low activity of SBS35 are too small to carry out a meaningful comparison of the platinum CN footprint in Fig. 4b.

Methods needs clarification. Are the visualized samples the average of the cancer types in each of the two groups (untreated vs treated) how many samples in each group? The table suggested above would help a lot with understanding what is actually being compared. How reliable is the calculation of WGD status? Some explanation into the values used in the calculation "WGD: 2.9 -1.7*LoH <= Ploidy" is warranted.

Following the reviewer’s suggestion, we have added two new Supplementary Tables (Tables S2 and S4) presenting a more thorough description of the subset of HMF tumors employed in the detection of treatment-related CN features (Table S2) and the overlap between treatments in terms of numbers co-treated tumors (Table S4). We have also expanded the rationale behind the inequality used to separate WGD and non-WGD tumors (see new version of Methods).

CROSS-CONSULTATION COMMENTS

Everyone seems to be in relative agreement that the results are modest, should be compared to https://www.nature.com/articles/s41586-022-04738-6.pdf and require some clarity throughout the manuscript.

Additional analyses, such as comparing to other datasets such as POG570 would benefit the paper.

Reviewer #4 (Significance):

Whilst previous studies have looked at the effect of anti-cancer drugs on single base mutations, owing to the challenges also seen here, no thorough investigation of the effect of these drugs on copy number has been performed. Therefore, this is an advance, albeit minor as the "copy number signature" of the exposed cancers was not particularly clear.

The use of WGD samples was a clever step forward for the analysis of copy number, as the effects of selection are weakened with an extra copy of the genome.

The finding that increased treatment with platinum results in increased copy number changes of size 10 kb to 10 Mb is an interesting finding, and something that could be considered when looking at treatment options in the future, particularly if it is shown to also affect normal cells in this way.

The cancer genomics field in which I am a part, would be interested in this finding.

We thank the reviewer for their appreciative comment 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

Summary:

This paper by Gonzalez et al attempts to identify copy number footprints of anti-cancer therapies. It follows previous work by the group looking at single base mutational footprints of anti-cancer therapies, which provides clear evidence of the effect of these drugs on the genome. This study into copy number footprints is less convincing, mainly due to the challenges in identifying these low frequency copy number signatures. The authors present weak evidence using CN signatures for an increase in the number of chromosomal fragments less than 10 Mb in size. However, this is not consistently significant between different cancer types treated with the same drugs. The interesting finding of an effect of platinum treatment intensity on copy number is seen quite nicely in Fig 4b when pooled into one simple effect. Signature analysis in this case seems unnecessary as the main finding is that platinum treatment results in increased 10 kb- 10 Mb fragments but only when pooled in this way. The paper is otherwise nicely written, although some clarity adjustments are required in the Figures and Figure legends.

Major comments:

Whilst a commendable effort has been made to identify copy number footprints, the evidence presented here for the identification of CN signatures is not so convincing. The main focus of the paper is the effect of Platinum based therapies and yet the two featured cancer types lung non-small cell and colorectal do not have consistent significant effects in the signature analysis.<br /> The referenced BioRXiv paper by Steele et al. is now published https://www.nature.com/articles/s41586-022-04738-6 and one wonders whether additional methods and analyses performed during their peer review may be useful in this paper as well. Can reanalyse using the predefined 21 CN signatures from Steele et al?

From Fig 2a there are 5 cancer types in HMF with Platinum treatment Lung non-small cell, colorectal, esophagus, urothelial and ovary and it is not clear if all these cancer types are combined or separated in the final analysis. All but urothelial are featured at some point though, but e.g. ovary has no significant differences between treated and untreated. Are samples from all 5 cancer types combined in the "treated versus untreated" analyses?

The strongest evidence for a real effect on copy number for platinum treatment comes in Figure 4, where there is a significant increase in LoH segments CN 1-4 with samples showing high SBS 35 mutations (a clever idea!). Attempting to separate out the samples into a "Copy number signature" in Figures 2 and 3 seem a bit like fillers to get to this actual potentially interesting finding. What is the benefit of separating out the copy number and zygosity when the real effect is much clearer when you pool everything and simplify it?

Can you investigate other drug treatments using this bulk approach using the proxy of the SBS drug mutations to indicate the "strength" of the mutational process of the drug.

Another point of interest is the overlap between treatments. A judgement call is made as to which is the overriding drug corresponding to the effect. Is it possible to separate these effects with NMF as per SBS? Or could a combination effect be detected? Likely the numbers would be too low for this separated analysis. But just looking at LoH 10kb to 10 Mb might show something?

A table summarizing the included samples, treatments, overlap, etc per cancer type is missing.

Your study may also benefit from a comparison to the latest Hartwig cohort paper https://www.biorxiv.org/content/10.1101/2022.06.17.496528v1.full particularly focusing on the suggestion of Treatment enriched drivers (TED) some of which are infact copy number driven.

Minor comments:

The Figure legends and Figures themselves need to be altered for clarity. Axis should be labelled more specifically, e.g Fig 1d axis currently reads "percentage". Fig 3b says left and right and there is no such thing. What do the sizes of the circles in 2c represent? Can you indicate cancer type as well in this plot (shading or line type) or are all treated samples pooled- this is not clear?<br /> It is not clear why Fig 2a only includes HMF samples and not PCAWG- PCAWG could be in supplement?<br /> Be consistent with labelling as well, in the text everything is referred to as 10 kb -10 Mb and some Figures labeled as such but others with 10^4- 10^7. How is the size 10 kb established? All the plots show 0-100 kb, where did the 10 kb limit come from?<br /> Why is ovary not in 4b?<br /> Methods needs clarification. Are the visualized samples the average of the cancer types in each of the two groups (untreated vs treated) how many samples in each group? The table suggested above would help a lot with understanding what is actually being compared. How reliable is the calculation of WGD status? Some explanation into the values used in the calculation "WGD: 2.9 -1.7*LoH <= Ploidy" is warranted.

Referees cross-commenting

Everyone seems to be in relative agreement that the results are modest, should be compared to https://www.nature.com/articles/s41586-022-04738-6.pdf and require some clarity throughout the manuscript.<br /> Additional analyses, such as comparing to other datasets such as POG570 would benefit the paper.

Significance

Whilst previous studies have looked at the effect of anti-cancer drugs on single base mutations, owing to the challenges also seen here, no thorough investigation of the effect of these drugs on copy number has been performed. Therefore, this is an advance, albeit minor as the "copy number signature" of the exposed cancers was not particularly clear.

The use of WGD samples was a clever step forward for the analysis of copy number, as the effects of selection are weakened with an extra copy of the genome.

The finding that increased treatment with platinum results in increased copy number changes of size 10 kb to 10 Mb is an interesting finding, and something that could be considered when looking at treatment options in the future, particularly if it is shown to also affect normal cells in this way.<br /> The cancer genomics field in which I am a part, would be interested in this finding.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Analyzing the genome wide copy number patterns across publicly available ~2700 primary and ~5000 metastatic tumors treated by a number of different classes of chemotherapy agents, the authors find a distinct signature of CNVs in tumors treated with platinum-based agents. These platinum-exposed tumors are characterized by a significant increase in the number of chromosomal fragments of lengths between 10 Kb-10 Mb, and this tendency correlates with dosage (approximated by previously published platinum induced mutational signatures). Also, it is interesting that comparison of WGD with non-WGD treated-vs-untreated samples shows that WGD samples tolerate larger CNVs, suggesting relaxed selection against large CNVs in WGD (or WGD as a mechanism to accumulate large CNVs).

Previous works have focused on mutational signatures of various environmental exposures and drugs. This paper attempts to extend the previous research by looking at patterns of copy number variations. The work is somewhat motivated (see comment below) and the experimental design and execution are reasonable.

The manuscript is well written. The method section could be elaborated more for reproducibility.

Major comments:

1. Overall, the results are modest. Although statistically significant, the increase in specific classes of CNVs in treated v untreated WGD mets is modest (Fig 2d), casting a doubt on clinical significance.
2. Three of the four drugs that yielded significant patterns seems to have largest sample sizes (Fig 2a). Is there a link between sample size and detection power? In general, robustness of the signals is not analyzed, relative to subsampling of tumors or genomic regions etc. Indeed, the authors have noted the potential lack of robustness somewhere in the manuscript.
3. Given inter-individual heterogeneity, analyzing longitudinal data of pre-treated primary and post-treated mets rom same individuals would really help strengthen the findings.
4. The authors show that several CN features show and increase upon platinum treatment. Are these independent observations? A global correlation among the 48 features in various samples classes (WGD/non-WGD, Primary/Met, treated/untreated) should be done and equivalence class of features defined. Otherwise, the biological significance of these observations could be overstated.
5. The only mechanistic link between platinum treatment and the observed CNV patterns is speculated to be via platinum-induced DNA breaks and errors during correction. This seems like a very general mechanisms applicable to any exposures (environmental or drug) that induces breaks. This lack of specificity makes it hard for me to understand the rationale to study CNV patterns - why, after all, should one expect to see a CNV signature?
6. Authors should contrast their findings with those in https://www.nature.com/articles/s41586-022-04738-6.pdf
7. For pyrimidine treated samples... "significance is lost across non-overlapping tumors". Authors should ascertain that this is not simply a matter of power. Also, would the significance not be lost for non-overlapping platinum-treated samples?
8. Interpretation of Fig 3b. "Had this increase in the number of 10Kb-10Mb chromosomal fragments across exposed tumors arisen through positive selection, we would expect to observe a concentration at specific genomic regions containing resistance genes." This needs to be tested specifically. A "concentration" perhaps would not jump out in a visual inspection of the global pattern, which does seem to show variability.

Minor comments:

1. "the ploidy of tumors with WGD varies in a range between 2.9 and 3.6 (quartiles 1 and 3, Fig. 1a)". I am not sure, there seem to several orange (WGD) points with ploidy below 2.9.

Significance

The novelty is to look at CNV signatures upon drug treatment (beyind mutational signatures). However, as mentioned above, it is not clear how different exposures that ultimately cause DNA break would have distinct CNV pattern. Overall, the results seem modest to me. Although statistically significant, the increase in specific classes of CNVs in treated v untreated WGD mets is modest (Fig 2d), casting a doubt on clinical significance. This work could still be of interest to some researchers, in particular, those interested in mutational signatures of environmental exposure. This work should be interpreted in the context of pan-cancer signatures of CNVs very recently published https://www.nature.com/articles/s41586-022-04738-6.pdf.

My background is in computational biology, working on transcriptional regulation for decades and more recently in cancer systems biology. I am comfortable with the techniques employed in this work but not so much with the mechanisms linking a specific drug to specific copy number signatures and also with the clinical significance of this problem. Keywords: "Computational biology", "Bioinformatics", "Transcriptional regulation", "NGS", "Omics", "Cancer systems biology"

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In this study, the author examine WGS data from two cohorts of cancer samples from previous studies: PCAWG, mostly representing primary tumors, and HMF, representing metastatic tumors. The HMF dataset represented 4902 metastatic tumors (including 2709 whole-genome doubling, i.e., WGD, metastatic tumors) from patients who had been exposed to 85 anticancer therapies. The study identified a pattern of large LOH events associated with the exposure of tumors of some cancer types to platinum-based therapies. This pattern is characterized by a significant increase in the number of chromosomal fragments in exposed tumors with respect to their unexposed counterparts. The findings could support the hypothesis that WGD may provide tumors with an advantage to withstand the effects of structural variation.

Specific comments:

1. There are a number of statements that would suggest that there is some uncertainty regarding the robustness of results, and that the analysis of additional cohorts may be needed to substantiate the overall findings. For example, page 4: "It is also plausible that more numerous cohorts of exposed tumors are required to understand whether the observed differences are indeed robust." Page 5: "Further analysis with larger cohorts are required to clarify this point, which appears especially to clarify whether a significant imbalance in favor of deleted chromosomal fragments does occur across platinum-exposed lung tumors." However, the abstract does not seem to reflect this level of uncertainty in reporting the main findings.
2. Many of the findings made appear to apply not to all tumors but are found within tumors of specific cancer types. However, the abstract does not appear to note this.
3. With regards to additional cohorts, there is a POG570 cohort of WGS data on 570 recurrent or metastatic tumors (Nature Cancer 2020, PMID: 35121966), some 82% of which were from patients receiving systemic therapy before biopsy. Is it possible that some of the patterns identified using the HMF datasets could be validated in the POG570 datasets? If not, what numbers of tumors would be needed for the patterns of interest to be reliably identified?
4. The PCAWG cohort is described as comprising all primary tumors, but in fact there are some metastatic tumors in PCAWG cohort. In particular, most of the TCGA melanoma (SKCM) samples are metastatic (PMID: 30401717). This may have bearing on using comparisons between PCAWG and HMF as a surrogate for primary versus metastases.
5. For each boxplot, the number of tumors represented in each group should be indicated somewhere (e.g., along the bottom).
6. For Figure 1a, is a color legend needed here?
7. For analyses comparing HMF to PCAWG (e.g., Figure 1c), the p-values ought to corrected for cancer type (e.g., using a linear regression model with cancer type as a factor).
8. For Figure 1d, are the numbers of tumors in each category indicated in parentheses?
9. For figures 2d and 2e legend, the numbers of tumors in exposed vs unexposed groups for each category should be indicated. Similar for Figures 3a, 3c, 3d.
10. For figure 2c, what is the statistical test used and multiple testing correction applied? Could this be noted in the figure legend?

Significance

The study makes effective use of public genomic resources to make new observations regarding platinum-based anticancer therapies. The observations identify patterns within specific cancer types. The analysis is exploratory in nature and would benefit from independent observation in an independent cohort, though it is not clear whether such cohorts may exist in sufficient numbers.

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity):

The manuscript reports effects on brood size, lifespan and healthspan upon manipulation of C. elegans genes encoding RagA, TOR and Pol III orthologs, as well as other well-characterized lifespan-affecting genes. The results point to complex relationships among TOR and Pol III that are not fully resolved, suggest a role for rpc-1 Pol III that is additive with well-characterized lifespan pathways, indicate a late-life requirement for rpc-1 Pol III to limit lifespan, and, contrary to a previous publication, suggest a muscle requirement for rpc-1 Pol III for lifespan limitation.

Major comments regarding key conclusions:

The work demonstrates that brood size is reduced upon rpc-1 Pol III RNAi feeding from the L4 stage. However, no further analysis is provided to show how later aspects of reproduction impair brood. Minimally, ruling out effects on spermatogenesis would be important since sperm number limits self-fertile brood size. It is also unclear from the methods whether the brood size results include embryonic lethality (post-reproduction). Internal hatching, if it occurred, could also affect interpretation of the results. A change in the reproductive period should be noted if it occurred.

The reviewer is correct that it is important to address the role of Pol III more thoroughly in relation to reproduction.

• The brood size experiments we present simply record the number of hatched progeny. To develop this analysis further we will present the age-specific fecundity data that we generated whilst doing these assays to demonstrate the impact of Pol III on the reproductive period. In addition, we will quantify and present data on the total brood size (dead eggs and hatched progeny) to address whether Pol III also impact embryonic development.
• At 25oC (the temperature that we did these experiments) very few animals suffered internal hatching and those that did were taken out of the analysis – therefore this is unlikely to skew the results.
• The question as to whether Pol III limits egg or sperm function (or later developmental roles) is also interesting and is not yet addressed. To examine this we will: Quantify brood size (dead eggs and hatched progeny) in elegans +/- Pol III RNAi that have been exposed to males during the reproductive period compared to those that reproduce solely as hermaphrodites.

The authors claim that, similar to the relationship previously concluded from aging studies, rpc-1 acts downstream of TORC1. However, this claim is not well supported. In an effort to circumvent early lethality caused by loss of let-363 ("CeTOR"), they use a mutation in raga-1 RagA and demonstrate a further reduction in brood with rpc-1 RNAi. If raga-1(ok386) were a null this result would demonstrate a relationship that is at least partially parallel, not linear. By contrast, double RNAi with let-363 was "non-additive", suggesting a more linear relationship. However, interpretation of these experiments requires (1) that the raga-1 mutation is null and affects only TORC1 signaling, (2) evidence that the double RNAi worked well (e.g., qPCR; see Ahringer et al. 2006 review regarding issues with multi-RNAi), and (3) failure to consider alternative effects of loss of let-363 (e.g., TORC2). Negative results with RNAi are particularly problematic in the absence of convincing evidence that the RNAi worked well. Moreover, results in Figure 1G are difficult to interpret since the initial values are low. Here and elsewhere the genetics descriptions are unconventional, hampering interpretation. For example, what is meant by a mutation being "incomplete"? That it acts as a hypomorph?

We understand the concerns of the reviewer:

• The raga mutant strain that we use is raga-1 (ok386.) This allele harbours a 1242 bp deletion at the raga-1 locus that removes almost the entire coding region of the gene. Details on this allele can be found on Wormbase, and we will reference this in the methods. https://wormbase.org/species/c_elegans/variation/WBVar00091681#02-456-10

For reference, this has been used in several other studies, e.g. doi.org/10.7554/eLife.49158

• We agree that double RNAi can be challenging. Appropriate controls were used here e.g. each RNAi diluted 50:50 with control RNAi in the single treatments and phenotypes were observed in each case (either brood size or lifespan). However, to address the precise knockdown of rpc-1 and let-363 obtained with RNAi we will perform qPCR in response to single and double RNAi treatment (both in WT and raga-1 mutant elegans).
• In addition, we will attempt to measure S6Kinase phosphorylation, a downstream readout of TORC1 signalling in response to raga-1 mutation or let-363 RNAi treatment with and without rpc-1 A phosphor S6 Kinase antibody is commercially available and has been used successfully in C. elegans - doi.org/10.7554/eLife.31268
• Our apologies that the nomenclature was confusing. The CeTOR RNAi nomenclature was ’borrowed’ from other papers describing this tool e.g. org/10.7554/eLife.31268 and doi: 10.1371/journal.pgen.1000972. Here, to make our work clearer, we will change ceTOR to let-363 TOR RNAi and raga-1 to raga-1 RagA in the manuscript – as suggested by the reviewer (see below). The description of ‘incomplete’ mutations will also be amended, and informed by our proposed qPCR analysis.

Another claim is that rpc-1 Pol III limits adult lifespan downstream of TOR. These results are not convincing. The two treatments (raga-1 mutation as "embryonic" and L4 stage "CeTOR" let-363 RNAi as late) are not directly comparable for reasons noted above, and the double RNAi problem hampers interpretation.

Our lifespan data points out that the longevity increase upon Pol III knockdown is additive with TOR/let-363, suggesting a mechanism independent of TOR. Indeed, due to lack of ideal reagents, we were forced to try the double RNAi knockdown approach for TOR/let-363 and Pol III/ rpc-1. To make the data interpretation easier, and rule out the possibility of confounding background RNAi to the maximum possible extent, we have included appropriate RNAi controls. Wherever double RNAi has been used, the effect on the phenotype by 50% dilution of target RNAi with empty-vector control, has also been shown independently and used for the statistical comparison with combinatorial RNAi. Our results have shown that diluting let-363 RNAi and rpc-1 RNAi both to 50%, is enough to impart lifespan increase when initiated from L4 stage.

The nomenclature might be easier to follow if the authors state the actual C. elegans genes manipulated (e.g., let-363 TOR versus raga-1 RagA) rather than using "CeTOR" as a catch-all since these genes are not identical in action.

Thank you for this suggestion. We will implement this in the manuscript where appropriate.

Based on genetic interactions (rsks-1, ife-2, ppp-1, daf-2 and germline loss) they show that rpc-1 RNAi further extends the long lifespan conferred by each of the mutant alleles tested, as well as germline loss induced by two different mutant conditions. These results, though negative, are important. The statement that rpc-1 does not affect global protein synthesis is somewhat overstated without additional experimental support.

We thank the reviewer for supporting our inclusion of ‘negative data’. We agree that our statement relating to protein synthesis is overstated given the data presented. We will soften this to: “rpc-1 does not seem to affect the lifespan incurred by reducing global protein synthesis, although this does not rule out the possibility that Pol III affect protein synthesis by other mechanisms”.

Extending and challenging their own previous work showing an intestinal focus of activity for rpc-1 in limiting longevity (Filer et al., 2017), and noting that RPC-1::GFP detection can be knocked down by RNAi in several tissues, they use a tissue restricted rde-1 expression approach (or sid-1 for neurons) to test the contribution of intestine, hypodermis, neurons, muscle and germline. This new analysis points to a role for the muscle. This result is intriguing and warrants further experiments. To shore up tissue-specific claims the authors could consider (1) additional drivers for intestine and muscle rde-1 in the RNAi experiments, or, ideally, a different approach such as tissue-specific protein degradation (again with multiple drivers), (2) a sufficiency experiment for muscle (wild-type muscle expression in the mutant to demonstrate reversal of the phenotype, or rescue of RNAi defects with an RNAi-insensitive reagent expressed in muscle).

Thank for you appreciating the work we have done here and suggesting further experiments. To take your points one at a time: (1) We have already used the most robust tissue-specific alleles generated and reported in the C. elegans literature so far. It would be a significant amount of work to generate new rde-1 driven tissue specific alleles to double check the Pol III levels/ rpc-1 knockdown response in certain tissues, and we feel this is beyond the scope of this project. Suggestion (2) is interesting and would require us to generate a muscle specific rpc-1 strain. However, there are issues with this approach. Firstly, it would require that we have a rpc-1 mutant to rescue – which we don’t as it is embryonically lethal and secondly it would not be possible to do this experiment using RNAi as the RNAi would then knock down the muscle construct.

The possible explanation for the differences in rde-1 results from the previous work should not be buried in the legends of Figure 3 and Figure S3. Perhaps this leaky background hypothesis should be directly tested (e.g., using the RPC-1::GFP to examine whether residual expression exists in ne219 but not in ne300)? In any case, legend to Figure S3 needs editing: The ne219 background is not itself "intestine-specific", as implied, and the last sentence of Figure S3 legend should be "Thus, the rde-1(ne219)...", right?

The differences between the different tissue-specific strains is interesting. On reflection we agree with the reviewer that it should be included in the main text. We will describe the differences between the two rde-1 alleles ne219 and ne300 in the appropriate section in the manuscript and state our results.

Finally, they show that late-adult rpc-1 RNAi extends lifespan over control RNAi and that, by several movement assays, healthspan is improved upon L4 rpc-1 RNAi, even when RNAi is active in muscle (based on WM118).<br /> The most significant new results are that rpc-1(RNAi) affects brood size, can extend lifespan (though modestly) after day 5 of adulthood, and that muscle may be involved rather than intestine.

Additional comments:

Text throughout should clarify TOR vs presumed TORC1. Methods are insufficient. Important aspects of the lifespan methods and raw data are missing - e.g. exact numbers of worms censored. Exact information regarding statistical analysis is lacking (e.g., which tests, corrections for multiple testing). References should be given for all strains. For the rde-1 strains, it would be helpful to include, in addition to the transgene alleles, the actual promoters used to claim tissue specificity. Note, worms do not have "skeletal" muscle, as implied in the discussion. Figure 5 was not helpful for this reviewer. Figure legend to S3A is confusing: the intestinal signal appears stronger or at least equal, not weaker, in the rpc-1 RNAi background. Were these images collected using the exact same exposure settings?

To address this we will:

• Standardise genetic notation throughout the manuscript (see specific comments above)
• Provide more detail on the transgenic alleles used e.g. promoters driving rde-1.
• The majority of strains were obtained from the CGC but wherever appropriate we will also supply a reference.
• Expand and revise Material and Methods section to appropriately describe all the statistical analyses performed.
• Revise lifespan methods to include censoring detail and lifespan Tables to include information on censored animals.
• Remove the reference to ‘skeletal muscle’ and replace with ‘body wall muscle’.
• Once we have generated new data on the specific knockdowns and downstream targets achieved with let-363 TOR RNAi and raga-1 RagA mutation, as well as on the brood size/dead eggs effects, we will incorporate this information into Fig. 5A for better clarity and readability.
• We can see on reflection that Figure S3A is confusing, mainly due to the gut autofluorescence in both the control and rpc-1 RNAi conditions. We will amend this figure to make this clear and include a selection of close up images of each tissue to make it easier to see the tissue specific knockdown by RNAi.

Reviewer #1 (Significance):

See above. Study will be of interest to aging community.

Reviewer #2 (Evidence, reproducibility and clarity):

The study by Malik and Silva et al describes results of the study investigating the role of RNA Polymerase III in regulating fecundity and lifespan in C. elegans. The authors show that knockdown of Pol III, similar to mTOR suppression, is detrimental for reproduction. Likewise, suppression of either Pol III or mTOR in adult animals extends lifespan via apparently the same pathway. In contrast, Pol III knockdown has an additive effect on lifespan in combination with other established genetic lifespan-extending approaches suggesting that they are working via different mechanisms. Furthermore, using the tissue-specific knockdown of Pol III the authors found that suppression Pol III expression is the muscle, but not other major worm tissues, is sufficient for its lifespan extending effect. Finally, the lifespan extension is also observed when Pol III knockdown is initiated late in adulthood. The overall conclusion is that suppression of Pol III expression late in animal life, particularly in the muscle, is a potential strategy to extend life- and health-span. Overall, the study is well-designed, the tools and results are robust and analysed appropriately. The data presentation is excellent, and the manuscript is clearly written. Addressing the points below will help to improve the clarity further.

We thank the reviewer for their very positive response to our study and are pleased that they found the data convincing. We are extremely pleased that the reviewer agrees with the design and tools used in this study. We can address all of the review’s comments – as discussed below.

Major:

Significant amount of GFP signal is still present in RNAi treated animals, what is the tissue that maintains particularly high levels of expression (Fig. 3A) and how does it affect the conclusions? What is the level of Pol III reduction in different tissues? Could different efficiency of knockdown explain the tissue-specific effect of Pol III downregulation on lifespan? It would be important to show (and, if possible, to quantify) the knockdown efficiency in different tissues using the available reporter

• This experiment had originally been done to test the efficiency of the RNAi, particularly in tissues where rpc-1 RNAi did not impact lifespan. The reviewer is right though, and this information could be analysed further to enhance our study. Figure 3A shows C. elegans expressing the rpc-1::3xflag::gfp reporter. This was used to a) determine the expression pattern of RPC-1 and b) determine the effect of rpc-1 RNAi on this. We noted that RPC-1::GFP is expressed a wide number of tissues and when the reporter strain is treated with rpc-1 RNAi, it is decreased in all tissues. The ‘green’ observed in the RNAi treatment is unfortunately attributable to autofluorescence generated by lysozymes in the C. elegans intestine and masks some of the effects we saw by eye.
• To establish the tissue-specific efficiency of Pol III knockdown and also address the confounding issue of the autofluorescence we will now use a combination of quantitative and qualitative fluorescent microscopy to measure the percentage RPC-1::GFP knockdown in each tissue relevant to this study.

Minor:<br /> Fig. S3B is not cited in the text and the legend for the figure is somewhat confusing, potentially containing errors, this needs to be clarified.

We thank the reviewers for pointing this out. The legend for this figure will be re-written as a result of the analysis described above and we will cite it in the main text.

Reviewer #2 (Significance):

This is the first thorough study of Pol III knockdown as a lifespan extending strategy in C. elegans. In addition to the different laboratory model (previous study of Pol III in ageing primarily focused on Drosophila), this manuscript also offers several novel insights into consequences of Pol III perturbation at phenotypic, as well as mechanistic level in terms of interaction with other longevity pathways. The study will be of interest to those interested in processes underlying longevity and ageing. Considering that this topic is currently in fashion the publication will probably attract attention of not only specialist but also general public.

We are extremely pleased that the reviewer shares our enthusiasm for this study and that they find the experimental evidence compelling.

Reviewer #3 (Evidence, reproducibility and clarity):

Summary: The paper by Yasir Malik et al investigates the genetic interrelationship between TOR signalling and Pol III expression regarding fecundity and longevity in C. elegans. Based on a previous study that defined a role of Pol III downstream of TOR in longevity across various species, this study looks particularly at the relative timing and tissue requirements for TOR and Pol III inhibition in longevity. Data indicate that Pol III acts downstream of TOR in regulating fecundity while there are additive effects regarding survival. The Pol III effect on longevity is based on its role in the muscle. Finally, health-span parameters mirror the survival data.

Major comments: This is a nice study the relies on genetic interaction to ask how TOR and Pol III interact. I find the observation that Pol III inhibition extends survival when initiated at day 5 of adulthood very exciting. In general, the study would benefit from additional data that back up the genetic observations._We thank the reviewer for appreciating the study and the novel insights it provides about the TOR-Pol III inter-relationship. We can address reviewer’s comments with the a few, limited experiments. Discussed below.

In Fig. 1, experiments are done to inhibit TOR to varying degrees in order to perform epistasis experiment. Of course these are difficult to interpret without the use of full KOs/loss of function. So while this is a good solution, it would be important to quantify the level to which TOR signalling is inhibited, optimally with biochemical experiments. We fully appreciate the reviewer’s point. A similar concern was raised by reviewer 1. We propose to address this in two ways: 1) by quantifying mRNA levels by qPCR of let-363 in response to either let-363 TOR RNAi; and 2) by determining the extend of TORC1 activity by using a biochemical readout of the pathway’s activity – S6 Kinase phosphorylation using Western blotting as described here: doi.org/10.7554/eLife.31268 2. General brood size is very low in the WT worms. Normally, one would expect 250-300 offspring per adult worm. It would be helpful if the authors could address this.

Indeed, as pointed out by the reviewer, the WT worms have a brood size of 250-300 eggs when kept at 20oC. but C. elegans exhibit different brood sizes dependent on temperature and these decline in size with increasing temperature. The experiments shown here were carried out at 25oC, where C. elegans produce less offspring. Our observation is in agreement with other studies of similar nature e.g. doi:10.1371/journal.pone.0112377 and doi.org/10.1371/journal.pone.0145925

1. Why were lifespan assays performed at 25C? The standard temperature for the worm is 20C and here I think this is very relevant as the TOR pathway is responsive to suboptimal conditions. I wonder if the results are also true for lower temperatures.

The reviewer raises an interesting point. This study follows from the previous study of Filer et al., Nature 2017 which demonstrated the role of Pol III in ageing. During this study we found and reported that there was a high proportion of intestinal bursting when lifespans were carried out at 20oC, which was ameliorated by carrying out the experiments at 25oC. This was quantified in the original manuscript. To maintain consistency, we continued carrying out Pol III lifespans at this slightly higher temperature. Due to this limitation it is not possible to test the impact of TOR signalling on Pol III at lower temperatures.

Minor comments: 1. It would help to better delineate the rationale for the experiments in Fig. S1. Experiments here are aimed to find mediators of TOR effects distinct from Pol III. Such distinct mediators would be additive to Pol III (as is the case in the figure) and downstream of TOR.

Interpreting epistasis analysis is challenging. We were looking for interactors of Pol III using this targeted genetic approach and working on the premise that if two genes interacted then their effects would be non-additive. However, the reviewer is correct that if two genes are doing the same thing independently then their effects may be additive. Although our data does not suggest these mediators interacting with Pol III in the same pathway, it does not rule out the other possibility. When we re-work the manuscript we will explain our rational more clearly and outline the two scenarios.

Reviewer #3 (Significance):

Strengths: The study advances our knowledge regarding the timing of the Pol III targeting intervention for survival effects.<br /> Limitations: The study relies only on genetic data and not all of it is conclusive.

This study will be interesting for the geroscience community with an eye on TOR inhibition and is relevant to worm biology. I work with C. elegans as a genetic model and I am interested in protein homeostasis, metabolism, health, and longevity.

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 paper by Yasir Malik et al investigates the genetic interrelationship between TOR signalling and Pol III expression regarding fecundity and longevity in C. elegans. Based on a previous study that defined a role of Pol III downstream of TOR in longevity across various species, this study looks particularly at the relative timing and tissue requirements for TOR and Pol III inhibition in longevity. Data indicate that Pol III acts downstream of TOR in regulating fecundity while there are additive effects regarding survival. The Pol III effect on longevity is based on its role in the muscle. Finally, health-span parameters mirror the survival data.

Major comments:

This is a nice study the relies on genetic interaction to ask how TOR and Pol III interact. I find the observation that Pol III inhibition extends survival when initiated at day 5 of adulthood very exciting. In general, the study would benefit from additional data that back up the genetic observations.

1. In Fig. 1, experiments are done to inhibit TOR to varying degrees in order to perform epistasis experiment. Of course these are difficult to interpret without the use of full KOs/loss of function. So while this is a good solution, it would be important to quantify the level to which TOR signalling is inhibited, optimally with biochemical experiments.
2. General brood size is very low in the WT worms. Normally, one would expect 250-300 offspring per adult worm. It would be helpful if the authors could address this.
3. Why were lifespan assays performed at 25C? The standard temperature for the worm is 20C and here I think this is very relevant as the TOR pathway is responsive to suboptimal conditions. I wonder if the results are also true for lower temperatures.

Minor comments:

It would help to better delineate the rationale for the experiments in Fig. S1. Experiments here are aimed to find mediators of TOR effects distinct from Pol III. Such distinct mediators would be additive to Pol III (as is the case in the figure) and downstream of TOR.

Significance

Strengths: The study advances our knowledge regarding the timing of the Pol III targeting intervention for survival effects.

Limitations: The study relies only on genetic data and not all of it is conclusive.

This study will be interesting for the geroscience community with an eye on TOR inhibition and is relevant to worm biology.

I work with C. elegans as a genetic model and I am interested in protein homeostasis, metabolism, health, and longevity.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

The study by Malik and Silva et al describes results of the study investigating the role of RNA Polymerase III in regulating fecundity and lifespan in C. elegans. The authors show that knockdown of Pol III, similar to mTOR suppression, is detrimental for reproduction. Likewise, suppression of either Pol III or mTOR in adult animals extends lifespan via apparently the same pathway. In contrast, Pol III knockdown has an additive effect on lifespan in combination with other established genetic lifespan-extending approaches suggesting that they are working via different mechanisms. Furthermore, using the tissue-specific knockdown of Pol III the authors found that suppression Pol III expression is the muscle, but not other major worm tissues, is sufficient for its lifespan extending effect. Finally, the lifespan extension is also observed when Pol III knockdown is initiated late in adulthood. The overall conclusion is that suppression of Pol III expression late in animal life, particularly in the muscle, is a potential strategy to extend life- and health-span. Overall, the study is well-designed, the tools and results are robust and analysed appropriately. The data presentation is excellent, and the manuscript is clearly written. Addressing the points below will help to improve the clarity further.

Major:

Significant amount of GFP signal is still present in RNAi treated animals, what is the tissue that maintains particularly high levels of expression (Fig. 3A) and how does it affect the conclusions?

What is the level of Pol III reduction in different tissues? Could different efficiency of knockdown explain the tissue-specific effect of Pol III downregulation on lifespan? It would be important to show (and, if possible, to quantify) the knockdown efficiency in different tissues using the available reporter.

Minor:

Fig. S3B is not cited in the text and the legend for the figure is somewhat confusing, potentially containing errors, this needs to be clarified.

Significance

This is the first thorough study of Pol III knockdown as a lifespan extending strategy in C. elegans. In addition to the different laboratory model (previous study of Pol III in ageing primarily focused on Drosophila), this manuscript also offers several novel insights into consequences of Pol III perturbation at phenotypic, as well as mechanistic level in terms of interaction with other longevity pathways. The study will be of interest to those interested in processes underlying longevity and ageing. Considering that this topic is currently in fashion the publication will probably attract attention of not only specialist but also general public.

My expertise is in cellular proteostasis and its perturbation in age-related diseases.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

The manuscript reports effects on brood size, lifespan and healthspan upon manipulation of C. elegans genes encoding RagA, TOR and Pol III orthologs, as well as other well-characterized lifespan-affecting genes. The results point to complex relationships among TOR and Pol III that are not fully resolved, suggest a role for rpc-1 Pol III that is additive with well-characterized lifespan pathways, indicate a late-life requirement for rpc-1 Pol III to limit lifespan, and, contrary to a previous publication, suggest a muscle requirement for rpc-1 Pol III for lifespan limitation.

Major comments regarding key conclusions:

1. The work demonstrates that brood size is reduced upon rpc-1 Pol III RNAi feeding from the L4 stage. However, no further analysis is provided to show how later aspects of reproduction impair brood. Minimally, ruling out effects on spermatogenesis would be important since sperm number limits self-fertile brood size. It is also unclear from the methods whether the brood size results include embryonic lethality (post-reproduction). Internal hatching, if it occurred, could also affect interpretation of the results. A change in the reproductive period should be noted if it occurred.
2. The authors claim that, similar to the relationship previously concluded from aging studies, rpc-1 acts downstream of TORC1. However, this claim is not well supported. In an effort to circumvent early lethality caused by loss of let-363 ("CeTOR"), they use a mutation in raga-1 RagA and demonstrate a further reduction in brood with rpc-1 RNAi. If raga-1(ok386) were a null this result would demonstrate a relationship that is at least partially parallel, not linear. By contrast, double RNAi with let-363 was "non-additive", suggesting a more linear relationship. However, interpretation of these experiments requires (1) that the raga-1 mutation is null and affects only TORC1 signaling, (2) evidnce that the double RNAi worked well (e.g., qPCR; see Ahringer et al. 2006 review regarding issues with multi-RNAi), and (3) failure to consider alternative effects of loss of let-363 (e.g., TORC2). Negative results with RNAi are particularly problematic in the absence of convincing evidence that the RNAi worked well. Moreover, results in Figure 1G are difficult to interpret since the initial values are low. Here and elsewhere the genetics descriptions are unconventional, hampering interpretation. For example, what is meant by a mutation being "incomplete"? That it acts as a hypomorph?
3. Another claim is that rpc-1 Pol III limits adult lifespan downstream of TOR. These results are not convincing. The two treatments (raga-1 mutation as "embryonic" and L4 stage "CeTOR" let-363 RNAi as late) are not directly comparable for reasons noted above, and the double RNAi problem hampers interpretation. The nomenclature might be easier to follow if the authors state the actual C. elegans genes manipulated (e.g., let-363 TOR versus raga-1 RagA) rather than using "CeTOR" as a catch-all since these genes are not identical in action.
4. Based on genetic interactions (rsks-1, ife-2, ppp-1, daf-2 and germline loss) they show that rpc-1 RNAi further extends the long lifespan conferred by each of the mutant alleles tested, as well as germline loss induced by two different mutant conditions. These results, though negative, are important. The statement that rpc-1 does not affect global protein synthesis is somewhat overstated without additional experimental support.
5. Extending and challenging their own previous work showing an intestinal focus of activity for rpc-1 in limiting longevity (Filer et al., 2017), and noting that RPC-1::GFP detection can be knocked down by RNAi in several tissues, they use a tissue restricted rde-1 expression approach (or sid-1 for neurons) to test the contribution of intestine, hypodermis, neurons, muscle and germline. This new analysis points to a role for the muscle. This result is intriguing and warrants further experiments. To shore up tissue-specific claims the authors could consider (1) additional drivers for intestine and muscle rde-1 in the RNAi experiments, or, ideally, a different approach such as tissue-specific protein degradation (again with multiple drivers), (2) a sufficiency experiment for muscle (wild-type muscle expression in the mutant to demonstrate reversal of the phenotype, or rescue of RNAi defects with an RNAi-insensitive reagent expressed in muscle). The possible explanation for the differences in rde-1 results from the previous work should not be buried in the legends of Figure 3 and Figure S3. Perhaps this leaky background hypothesis should be directly tested (e.g., using the RPC-1::GFP to examine whether residual expression exists in ne219 but not in ne300)? In any case, legend to Figure S3 needs editing: The ne219 background is not itself "intestine-specific", as implied, and the last sentence of Figure S3 legend should be "Thus, the rde-1(ne219)...", right?
6. Finally, they show that late-adult rpc-1 RNAi extends lifespan over control RNAi and that, by several movement assays, healthspan is improved upon L4 rpc-1 RNAi, even when RNAi is active in muscle (based on WM118).
7. The most significant new results are that rpc-1(RNAi) affects brood size, can extend lifespan (though modestly) after day 5 of adulthood, and that muscle may be involved rather than intestine.

Additional comments

Text throughout should clarify TOR vs presumed TORC1. Methods are insufficient. Important aspects of the lifespan methods and raw data are missing - e.g. exact numbers of worms censored. Exact information regarding statistical analysis is lacking (e.g., which tests, corrections for multiple testing). References should be given for all strains. For the rde-1 strains, it would be helpful to include, in addition to the transgene alleles, the actual promoters used to claim tissue specificity. Note, worms do not have "skeletal" muscle, as implied in the discussion. Figure 5 was not helpful for this reviewer. Figure legend to S3A is confusing: the intestinal signal appears stronger or at least equal, not weaker, in the rpc-1 RNAi background. Were these images collected using the exact same exposure settings?

Significance

See above. Study will be of interest to aging community.

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

Learn more at Review Commons

---

Reply to the reviewers

Manuscript number: RC-2022-01480

Corresponding author(s): Ananda, Sarkar

1. General Statements

We are thankful to Review commons platform that helped our manuscript critically reviewed with very constructive and valuable feedback. This gave us the opportunity to do the experiments accordingly and significantly improve the manuscript. We are hopeful that this platform will help our manuscript get published in a journal of repute.

2. Point-by-point description of the revisions

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

The manuscript entitled "LDL1 and LDL2 histone demethylases interact with FVE to regulate flowering in Arabidopsis" characterized that LDL1 regulates flowering by binding on the chromatin of MAF4 and MAF5 to repress their expression. Further the authors proposed LDL1/LDL2-FVE model. Here are some comments for this manuscript.

Major problems: 1. This experiment is still not testing or showing/concluding that the whole complex forms on the MAF4 and MAF5

Response: We understand reviewer’s concern regarding the complex. Previously FVE was shown to be a part of co-repressor complex including HDA6, HDA5 and FLD to regulate the expression of FLC and its clade members during floral transition1-3 We showed that LDL1 binds directly to the chromatin of MAF4 and MAF5 to suppress their expression (Figure 1 and 2). Furthermore, we discovered that LDL1 and LDL2 interact with FVE to influence floral transition (Figure 8 and 9). Hung et al., 2018 reported the interaction of LDL1 and LDL2 with HDA6 to regulate circadian rhythm4 and we found that the expression of MAF4 and MAF5 was upregulated in ldl1ldl2hda6 than ldl1ldl2 (Figure 5C and 5D). Therefore, our experimental data, together with previously reported data makes it evident that LDL1 and LDL2 are a part of co-repressor complex through their interaction with FVE and HDA6, which we concluded here. We agree with the reviewer that an additional experiment, such as complex pull-down, will be helpful, but in our opinion, it will only provide additional confirmatory evidence.

2.It is not shown LDL1/LDL2 repress MAF4 and MAF5 by removing H3K4me2 activity. It would be useful to test whether the methylation level of MAF4 and MAF5 has been altered in ldl1/ldl2 mutant

Response: We found altered methylation level in MAF4 and MAF5 chromatin during floral transition in ldl1 and ldl1ldl2 mutants (Figure 6 and 7). We observed that the absence of LDL1, or both LDL1 and LDL2 disturbs the shift in H3K4 methylation status on MAF4 and MAF5 during floral transition and ends up in a more active (enriched in H3K4me3 marks) chromatin state at 19 days. This result, taken together with the increased MAF4 and MAF5 expression in ldl1 and ldl1ldl2 double mutants (Figure 5C and 5D) indicates that LDL1/LDL2 repress MAF4 and MAF5 by altering H3K4 methylation.

3.I suggest that further research is required to provide conclusive evidence concerning the physiology function of LDL1/LDL2-FVE. Such as the expression pattern of LDL1/LDL2, the methylation level of MAF4 and MAF5 before or after floral transition

Response: Taking this suggestion into account, we performed quantification of rosette leaves and flowering time of fvec, ldlfvec and ldl2fvec along with WT, ldl1 and ldl2 (Figure 9). We also observed decreased expression of floral activator genes, FT and SOC1 (targets of MAF4 and MAF5) in fvec, ldlfvec and ldl2fvec in comparison to the WT (Supplementary Figure 10C), which corresponds to their late flowering phenotype.

To understand the role of LDL1and LDL2 during floral transition, we first analyzed the expression of LDL1 and LDL2 during floral transition (Supplementary Figure 8). We observed that the expression of LDL1 and LDL2 expression peaks at 16 days and gets stabilized till 19 days. Then we checked the enrichment of H3K4me1, H3K4me2 and H3K4me3 on MAF4 and MAF5 chromatin in ldl1 and ldl1ldl2 plants with respect to the WT at 16 days (before floral transition) and 19 days (after floral transition). We found an increase in the conversion of H3K4me1 to H3K4me3, when LDL1 and LDL2 were not present (Figure 6 and 7).

Reviewer #1 (Significance (Required)):

The manuscript provide some evidences how LDL1 involve in flowering through epigenetic regulation.

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

Mahima and colleagues investigated LDL1/LDL2-MAF4/MAF5 in Arabidopsis flowering time control. The manuscript contains some interesting observations. To my point of view, however, the data need to be consolidated to support conclusions drawn in the manuscript.

• Title: it does not correctly reflect the manuscript content. Data in relation with FVE were limited to Fig 6, where the data themselves appear preliminary.

Response: We agree with the reviewers that our title didn’t reflect the manuscript content precisely and are happy to take this criticism into consideration. We have revised the title to, “LDL1 and LDL2 affect the dynamics of H3K4 methylation on the chromatin of MAF4 and MAF5 to allow floral transition in Arabidopsis”. Additionally, have provided the quantification data for fvec, ldlfvec and ldl2fvec with respect to WT, ldl1 and ldl2 plants (Figure 9)

• Abstract: most conclusions are over-stated. The current data shown in the manuscript cannot support such strong conclusions.

Response: We have rigorously revised the abstract and toned down the overstated conclusions

• Introduction: It is necessary to make clear that the role of the LDL1 and LDL2 genes in flowering time control had been well established in previous studies, including their repression of transcription of FLC, MAF4 and MAF5 (Berr et al., 2015, Plant J 81:316).

Response: We have revised the introduction to include the previously known roles of LDL1 and LDL2 in regulating flowering time.

• Results:

Regarding LDL1-overexpression lines, 'Relative expression' in Supplementary Fig 2B referred to normalization to WT? The phenotype of plants needs to be shown.

Response: Yes, the level of upregulation of LDL1 expression in different T1 plants (after selection from Hygromycin) was calculated with respect to the WT.

Regarding flowering time, have the observation and measures been performed in the same experiments for the ldl1, ldl1 flc, ldl1 maf4 and ldl1 maf5 mutants (Fig 3 and Supplementary Fig 1)? The late-flowering phenotype of ldl1 shown in Fig 3D-F is much severe than the same mutant shown in the other Figs, any explanation? What's the interpretation that ldl1 is epistatic to flc, maf4 and maf5?

Response: We agree with the reviewer’s observation which is correct. The following quantifications were taken at various points during the study:

flc, ldl1, and ldlflc (Supplementary Figure 1)

WT, ldl1, and ldl1maf4 (Figure 3A, 3B and 3C)

WT, ldl1, and ldl1maf5 (Figure 3D, 3E and 3F)

The rosette leaf numbers and flowering time of the plants in Figure 3D-3F are more severe than the others because seeds were directly sprinkled onto the soil in this phenotyping, whereas in previous phenotypings, plants were grown on 1/2MS plates before being transferred to soil. However, all the components of a single experiment were grown in the same condition. We appreciate your observation, the present data does suggest ldl1 being epistatic to flc, maf4 and maf5.

The in vitro test of LDL1 for its enzyme activity (Fig 4) appears preliminary and fragmented. The quantification data in Fig 4C-D need repeats. Have other histone methylation types (e.g. H3K4me3, H3K27me3, H3K36me3) been tested? The only two types (H3K4me2 and H3K9me2) shown are both down-regulated by LDL1-GST. Can H3K9 demethylation also play a role in flowering time control? In any case, the current in vitro data only are not sufficient to draw the strong conclusions as those appeared in the manuscripts.

Response: Before concluding that LDL1 has H3Kme2 and H3K9me2 demethylase activity, we confirmed it several times__. __Please refer to the PDF file for “response to reviewers” for supporting data.

We analyzed the western band intensity by calculating the area under the curve with imageJ software, which varies between experiments depending on the band intensities, therefore, rather than plotting absolute values of band intensity, we plotted the ratio of LDL1-GST/GST from three independent experiments in Figure 4B. We did perform a preliminary experiment to see if LDL1 has demethylation activity against different methylation marks, such as H3k4me1, me3, H3K9me1, and me3 (1=GST, 2=LDL1-GST), but there was no significant change in the methylation marks in the presence of LDL1. Please refer to the PDF file for “response to reviewers” for supporting data.

H3K9 is a repressive chromatin mark, and its removal would suggest gene activation. Upregulation of FLC, MAF4, and MAF5 in ldl1 and ldl2 mutant suggests LDL1 and LDL2 removes H3k4me2 methylation marks during flowering. However, JMJ28, Jumonji C (JmjC) domain-containing histone demethylase have been shown to positively regulate flowering by removing repressive H3K9me2 marks from the chromatin marks from the chromatin of CONSTANS (CO)5.

In the manuscript, it is saying that LDL1 binds on the chromatin of MAF4 and MAF5. However, I cannot find any data shown to support this conclusion.

Response: We would like to refer to Figure 2A and B where we have provided this information.

Protein-protein interactions, e.g. LDL1/LDL2-FVE in Fig 6A and LDL1-LDL2 and LDL1-HDA5 in Supplementary Fig 5, are examined in yeast two-hybrid assay. Other independent assays would be required.

Response: We have confirmed the interaction of LDL1 and LDL2 with FVE using co-immunoprecipitation assay (Figure 8B). Since Co-IP is a confirmatory experiment, we have done it for positive interactions found through Y2H only. Moreover, in the current manuscript our focus has not been on HDA5, so we didn’t proceed with further experiments.

The study of genetic interaction between fve and ldl1/ldl2 (Fig 6B-D) looks very preliminary. It is unclear how ldl1 fve and ldl2 fve were obtained: by crosses or by CRISPR-Cas9 using ldl1 and ldl2? The phenotypes need more investigations and some molecular data regarding flowering regulatory genes (e.g. MAF4/5) are necessary. In any case, the current title and the related conclusions drawn in the manuscript are over-stated.

Response: We performed the quantification of the genetic interaction between fve and ldl1/ldl2. The binary vector pHSE401-FVE was transformed in ldl1 and ldl2 to produce ldl1fvec and ldl2fvec, respectively. We previously mentioned it in the material methods, but we have now updated it in the results section to avoid confusion.

Following the suggestions, we have scored the phenotype (Figure 9) and checked the expression of flowering regulatory genes (Supplementary Figure 10C).

Fig 7 showed data about MAF5-FLC, MAF5-SVP and MAF5-MAF5 interactions in yeast two-hybrid and about transcriptional repressor activity assay in tobacco leaves using the LUC-reporter. Again, the data need to be confirmed and reproducibility of experiments need to be shown. In addition to proFT:LUC, it is also necessary to have an internal normalization reference construct. Anyway, currently it is far away to allow a strong conclusion such as drawn in the manuscript that MAF5 interacts with FLC and SVP and repress FT to delay floral transition. Response: We have confirmed the interaction of MAF5-FLC, MAF5-SVP and MAF5-MAF5 using co-immunoprecipitation (Figure 10B). We quantified the firefly luciferase activity under proFT using renilla luciferase under pro35s as an internal control and the ratio of LUC/REN represented the promoter activity of FT promoter (Figure 10C).

Reviewer #2 (Significance (Required)):

Topic is interesting, but data are poor to support the conculsions drawn.

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

LDL1 and LDL2 histone demethylases interact with FVE to regulate flowering in Arabidopsis Summary This work study the role on flowering time of LDL1 and LDL2, two Arabidopsis homologs of the histone demethylase LSD1. Although this phenotype was previously described, the authors explore if LDL1 and LDL2 regulate other genes in addition to the floral repressor FLC. In fact, mRNA expression experiments and genetic analyse suggest that LDL1 modules flowering regulating the expression of MAF4 and MAF5, two FLC-like genes that has been less characterized. The also provide some in vitro biochemical evidence of the demethylase activity of LDL1 protein and yeast-two-hybrid data showing the interaction with FVE, another chromatin regulator involved in flowering time.

Major comments 1. Lines 116-117. Please rephrase these lines and remove panels C, D and E from figure 1 (these could be supplementary material). The flowering time phenotype of MAF4 and MAF5 in Col background is very well documented and was described before, see Gu et al Nat. Comm., 2013 (10.1038/ncomms2947) and Kim et al. Plant Cell, 2013 (10.1105/tpc.112.104760)

Response: As per the suggestion, we have modified the discussion and moved the panels 1C, 1D and 1E to the supplementary.

Lines 128-130 and Fig Sup3. The proLDL1:LDL1-GUS cannot be described as fully functional because its flowering time and LDL1 mRNA expression levels has not been compared to the wild-type plant. The line flowers earlier that the ldl1 mutant but it may only partially complement the flowering phenotype.

Response: We have provided additional experiment that the transgene is functional in proLDL1:LDL1-GUS (ldl1) with respect to the WT plants (Supplementary Figure 5A).

Line 135 and Figure 2. How the Chip data was normalized? What are you comparting in your statistical significance tests? Only two regions of each gene were analysed; to assess the binding of LDL1 to MAF4 and MAF5 loci more regions must be analysed.

Response: Normalization of the ChIP data and significance of enrichment of LDL1 was calculated with respect to the fold enrichment in the empty vector control (EV (ldl1)) plants. We only examined the promoter and exon1 of MAF4 and MAF5 for LDL1 enrichment because Hung et al,2019's6 study demonstrated that LDL1 is enriched on the promoter and exon1 of the downstream protein coding genes. However, to check for methylation marks during flowering, we have employed different primer sets on various positions between the promoter and exon1 on MAF4 and MAF5 chromatin.

Figures 6C and 6D. The genetic analysis of ldl mutant with fve-c line is prelaminar and incomplete. The epistasis cannot be evaluated as no quantitative flowering time data is provided. A questionable picture of one lonely plant cannot sustain the conclusions of lines 207-208.

Response: We have modified the picture and quantified the flowering time data to show genetic interaction of ldl1 and ldl2 with fvec mutant plants (Figure 9).

METODS. Please clarify the used mutant alleles for LDL1 LDL2, MAF4, MAF5 and FLC; if they has been previously described; if they are full knock-outs; and, consequently, use the appropriated allele name across the manuscript.

Response: As per the suggestion, we have clarified the different mutant alleles used in the study.

Minor points: 6. I think the title does not describe the work - the interaction with FVE is very relevant but it is not the central theme of the article.

Response: We have changed the title of the study to “LDL1 and LDL2 affect the dynamics of H3K4 methylation on the chromatin of MAF4 and MAF5 to allow floral transition in Arabidopsis”.

It would be very informative to have short-day flowering tome data of the genetic combinations of ldl mutants with flc, maf4 and maf5 mutations.

Response: We absolutely agree that elaborate SD experiment may open interesting avenue for LDL1 mediated regulation of flowering, which might be good for future studies. However, ldl1ldl2 shows late flowering, while maf4 and maf5 exhibit the early flowering phenotype irrespective of the day length7,8.

I found the Discussion section rather too long.

Response: We have shortened the discussion to make it more focused.

Reviewer #3 (Significance (Required)):

Although it is clear that LDL proteins regulate MAF4 and MAF 5. I found that the manuscript lacks of a general overview of flowering time regulation. At the end, it is not clear how LDL proteins regulate flowering time because they regulate FLC, FWA, MAF4 and MAF5: What is more important? Which is the main role of each protein? Are they reductant or do they have specialized functions? In a nut shell, this study is an interesting piece of work for the flowering time field: However, in my opinion, some of the presented data are redundant with previous works and the manuscript may not be relevant for a general audience.

1. Yu, C.-W. et al. HISTONE DEACETYLASE6 Interacts with FLOWERING LOCUS D and Regulates Flowering in Arabidopsis. Plant Physiology 156, 173-184 (2011).
2. Luo, M. et al. Regulation of flowering time by the histone deacetylase HDA 5 in A rabidopsis. The Plant Journal 82, 925-936 (2015).
3. Yu, C.-W., Chang, K.-Y. & Wu, K. Genome-wide analysis of gene regulatory networks of the FVE-HDA6-FLD complex in Arabidopsis. Frontiers in plant science 7, 555 (2016).
4. Hung, F.-Y. et al. The Arabidopsis LDL1/2-HDA6 histone modification complex is functionally associated with CCA1/LHY in regulation of circadian clock genes. Nucleic acids research 46, 10669-10681 (2018).
5. Hung, F.-Y. et al. The Arabidopsis histone demethylase JMJ28 regulates CONSTANS by interacting with FBH transcription factors. The Plant Cell 33, 1196-1211 (2021).
6. Hung, F.-Y. et al. The expression of long non-coding RNAs is associated with H3Ac and H3K4me2 changes regulated by the HDA6-LDL1/2 histone modification complex in Arabidopsis. NAR Genomics and Bioinformatics 2 (2020). 7 Berr, A. et al. The trx G family histone methyltransferase SET DOMAIN GROUP 26 promotes flowering via a distinctive genetic pathway. The Plant Journal 81, 316-328 (2015).

8 Kim, D.-H. and Sibum, S. Coordination of the vernalization response through a VIN3 and

        FLC gene family regulatory network in Arabidopsis. *The Plant Cell *__25, __454-469 (2013)

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

LDL1 and LDL2 histone demethylases interact with FVE to regulate flowering in Arabidopsis

Summary

This work study the role on flowering time of LDL1 and LDL2, two Arabidopsis homologs of the histone demethylase LSD1. Although this phenotype was previously described, the authors explore if LDL1 and LDL2 regulate other genes in addition to the floral repressor FLC. In fact, mRNA expression experiments and genetic analyse suggest that LDL1 modules flowering regulating the expression of MAF4 and MAF5, two FLC-like genes that has been less characterized. The also provide some in vitro biochemical evidence of the demethylase activity of LDL1 protein and yeast-two-hybrid data showing the interaction with FVE, another chromatin regulator involved in flowering time.

Major comments

1. Lines 116-117. Please rephrase these lines and remove panels C, D and E from figure 1 (these could be supplementary material). The flowering time phenotype of MAF4 and MAF5 in Col background is very well documented and was described before, see Gu et al Nat. Comm., 2013 (10.1038/ncomms2947) and Kim et al. Plant Cell, 2013 (10.1105/tpc.112.104760)
2. Lines 128-130 and Fig Sup3. The proLDL1:LDL1-GUS cannot be described as fully functional because its flowering time and LDL1 mRNA expression levels has not been compared to the wild-type plant. The line flowers earlier that the ldl1 mutant but it may only partially complement the flowering phenotype.
3. Line 135 and Figure 2. How the Chip data was normalized? What are you comparting in your statistical significance tests? Only two regions of each gene were analysed; to assess the binding of LDL1 to MAF4 and MAF5 loci more regions must be analysed.
4. Figures 6C and 6D. The genetic analysis of ldl mutant with fve-c line is prelaminar and incomplete. The epistasis cannot be evaluated as no quantitative flowering time data is provided. A questionable picture of one lonely plant cannot sustain the conclusions of lines 207-208.
5. METODS. Please clarify the used mutant alleles for LDL1 LDL2, MAF4, MAF5 and FLC; if they has been previously described; if they are full knock-outs; and, consequently, use the appropriated allele name across the manuscript.

Minor points:

1. I think the tittle does not describe the work - the interaction with FVE is very relevant but it is not the central theme of the article.
2. It would be very informative to have short-day flowering tome data of the genetic combinations of ldl mutants with flc, maf4 and maf5 mutations.
3. I found the Discussion section rather too long.

Significance

Although it is clear that LDL proteins regulate MAF4 and MAF 5. I found that the manuscript lacks of a general overview of flowering time regulation. At the end, it is not clear how LDL proteins regulate flowering time because they regulate FLC, FWA, MAF4 and MAF5: What is more important? Which is the main role of each protein? Are they reductant or do they have specialized functions?

In a nut shell, this study is an interesting piece of work for the flowering time field: However, in my opinion, some of the presented data are redundant with previous works and the manuscript may not be relevant for a general audience.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Mahima and colleagues investigated LDL1/LDL2-MAF4/MAF5 in Arabidopsis flowering time control. The manuscript contains some interesting observations. To my point of view, however, the data need to be consolidated to support conclusions drawn in the manuscript.

• Title: it does not correctly reflect the manuscript content. Data in relation with FVE were limited to Fig 6, where the data themselves appear preliminary.
• Abstract: most conclusions are over-stated. The current data shown in the manuscript cannot support such strong conclusions.
• Introduction: It is necessary to make clear that the role of the LDL1 and LDL2 genes in flowering time control had been well established in previous studies, including their repression of transcription of FLC, MAF4 and MAF5 (Berr et al., 2015, Plant J 81:316).
• Results:

Regarding LDL1-overexpression lines, 'Relative expression' in Supplementary Fig 2B referred to normalization to WT? The phenotype of plants needs to be shown.

Regarding flowering time, have the observation and measures been performed in the same experiments for the ldl1, ldl1 flc, ldl1 maf4 and ldl1 maf5 mutants (Fig 3 and Supplementary Fig 1)? The late-flowering phenotype of ldl1 shown in Fig 3D-F is much severe than the same mutant shown in the other Figs, any explanation? What's the interpretation that ldl1 is epistatic to flc, maf4 and maf5?

The in vitro test of LDL1 for its enzyme activity (Fig 4) appears preliminary and fragmented. The quantification data in Fig 4C-D need repeats. Have other histone methylation types (e.g. H3K4me3, H3K27me3, H3K36me3) been tested? The only two types (H3K4me2 and H3K9me2) shown are both down-regulated by LDL1-GST. Can H3K9 demethylation also play a role in flowering time control? In any case, the current in vitro data only are not sufficient to draw the strong conclusions as those appeared in the manuscripts.

In the manuscript, it is saying that LDL1 binds on the chromatin of MAF4 and MAF5. However, I cannot find any data shown to support this conclusion.

Protein-protein interactions, e.g. LDL1/LDL2-FVE in Fig 6A and LDL1-LDL2 and LDL1-HDA5 in Supplementary Fig 5, are examined in yeast two-hybrid assay. Other independent assays would be required.

The study of genetic interaction between fve and ldl1/ldl2 (Fig 6B-D) looks very preliminary. It is unclear how ldl1 fve and ldl2 fve were obtained: by crosses or by CRISPR-Cas9 using ldl1 and ldl2? The phenotypes need more investigations and some molecular data regarding flowering regulatory genes (e.g. MAF4/5) are necessary. In any case, the current title and the related conclusions drawn in the manuscript are over-stated.

Fig 7 showed data about MAF5-FLC, MAF5-SVP and MAF5-MAF5 interactions in yeast two-hybrid and about transcriptional repressor activity assay in tobacco leaves using the LUC-reporter. Again, the data need to be confirmed and reproducibility of experiments need to be shown. In addition to proFT:LUC, it is also necessary to have an internal normalization reference construct. Anyway, currently it is far away to allow a strong conclusion such as drawn in the manuscript that MAF5 interacts with FLC and SVP and repress FT to delay floral transition.

Significance

Topic is interesting, but data are poor to support the conculsions drawn.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

The manuscript entitled "LDL1 and LDL2 histone demethylases interact with FVE to regulate flowering in Arabidopsis" characterized that LDL1 regulates flowering by binding on the chromatin of MAF4 and MAF5 to repress their expression. Further the authors proposed LDL1/LDL2-FVE model. Here are some comments for this manuscript.

Major problems:

1. This experiment is still not testing or showing/concluding that the whole complex forms on the MAF4 and MAF5
2. It is not shown LDL1/LDL2 repress MAF4 and MAF5 by removing H3K4me2 activity. It would be useful to test whether the methylation level of MAF4 and MAF5 has been altered in ldl1/ldl2 mutant
3. I suggest that further research is required to provide conclusive evidence concerning the physiology function of LDL1/LDL2-FVE. Such as the expression pattern of LDL1/LDL2, the methylation level of MAF4 and MAF5 before or after floral transition

Significance

The manuscript provide some evidences how LDL1 involve in flowering through epigenetic regulation.

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

Learn more at Review Commons

---

Reply to the reviewers

We are grateful to the reviewers for their efforts in critically reading our work. Their meaningful input led us to make the revisions detailed below in our “point-by-point” answers to the reviewer’s comments. The insightful comments have helped us significantly improve the manuscript, allowing us to more accurately quantify and convey our data – we are thankful for that.

Reviewer #1

1. The Figure 1 legend indicates that the BirA tagged strains are mated with ~6000 AviTag yeast strains but results in Figure 2 pie chart account for 4812 total readouts. Presumably 1000 or more strains could not mate or did not produce viable diploids with the BirA tagged strains? It would be helpful to explain this differential. We thank Reviewer #1 for pointing out this gap which occurred exactly as they have interpreted. We have now corrected the figure legend to say exactly how many strains were in the library (5330) and have clearly stated the attrition of strains.

If possible, suggest including more of the raw data (in supplementary) that supports the pie chart in Figure 2. Table S1 shows the 111 proteins that display preference for Ssh1 (out of 586 total interactors?) and the fold change (in rank order) for interaction preference. At a minimum, similar data on Sec61 preference and the list of positive interactors should be included. There may also be useful information in the relative biotinylation signal for each BirA and AviTag combination when significantly above background. This is presumably a readout of AviTag protein abundance, dwell time and orientation to BirA activity. The data could be useful to other investigators.

This is a very good suggestion. We have now added a supplementary table (Supplementary Table S2) with the interaction results for proteins that preferred Sec61 and proteins that did not show any preference.

The authors might want to be more cautious in interpreting impact of the UPR on ssh1 phenotypes in the results and discussion. The Wilkinson et al 2002 paper referenced used very different conditions to detect UPR in ssh1 deletions strains. Jonikas et al 2009 does not detect a chronic UPR in ssh1 deletion cells and the conditions used in the current study seem more similar to the 2009 report. It seems more likely that deficits in translocating/localizing specific proteins causes the observed phenotypes instead of chronic UPR due to reduced ER levels of PDI.

*We agree that as result of the different conditions it is difficult to compare our data to the Wilkinson et al 2002 paper. We have therefore adjusted the text to remove this interpretation. *

Reviewer #2

1. Why was BirA used to study transient interactions? Biotinylation through BirA is slow (that is why it takes several hours to label proximity proteins) and thus it may not be suitable for capturing transient interactions. Instead, TurboID would be more suitable as the biotinylation reaction is faster than BirA. A reasonable explanation using BirA is required. We thank the reviewer for this comment which indeed also reflects our “process” of thinking. Originally, we did try to use TurboID to identify potential cargo proteins. We now have a very robust methodology to look at protein substrates by TurboID (see: https://www.biorxiv.org/content/10.1101/2022.04.27.489741v1) and so this would have obviously been the easier and faster method. However using this approach we mainly observed ribosome subunits and cytosolic proteins for Sec61 and very few, mostly cytosolic, proteins for Ssh1. Our interpretation of this is that since all biotinylation of TurboID strains occurs in parallel there is “competition” from the highly abundant and strong interactors and this does not leave a possibility to detect the low-abundance and even more transient interactions that we would like to measure. It seems that although birA/AviTag are much slower, the specificity and singular ligation site that should be exposed also in co-translational-translocation events, are more suitable for this specific experimental setup. We have now explained this also in the text.

One key question is whether biotinylated proteins identified by this method are substrates or proteins just proximal to Sec61 or Ssh1 due to close cellular localization (e.g. ER membrane) or same protein complex members. An experiment or analysis would be required to confirm that the proteins they identified are indeed potential substrates.

*This is indeed an extremely important point and we have now carefully addressed it in the text. We are certain that the reviewer is right and that many of the biotinylated proteins are same complex members and cytosolic components that happen to be in proximity (maybe regulators?) just as the reviewer suggested. We now clearly write this in the results section. This is why we focused on signal peptide containing proteins. These proteins CAN NOT be complex members nor biotinylated simply due to proximal location on the ER membrane. This is since they reside inside the lumen of the ER if they are soluble or are inserted (if they contain also a transmembrane domain) with their tagged N’ facing the lumen of the ER (So called Type I proteins). The only way such proteins could be biotinylated by the slow BirA on the cytosolic surface is if they passed through the pore of the translocon. *

Along the same line, if proteins identified by this approach are bona fide substrates of Sec61 and Ssh1, proteins having signal peptides should be enriched in the candidate list of substrates. However, it does not look like that according to Figure 2A where the secretome proteins/total proteins ratio appears to be similar among the 4 categories (e.g., Ssh1 preferring, No preference, and Not interacting or excluded). The authors should comment on this.

*We thank Reviewer #2 for highlighting this point that was not clear from our text and figures. There is definitely an enrichment of Signal Peptide (SP) containing proteins amongst the proteins that we think are bona fide substrates however this was not visualized clearly. To highlight this point we have modified Figure 2 and added a bar graph showing the distribution of SP and TMD proteins within the potential secretome. This graph now highlights the enrichment of SP containing proteins in the groups of proteins that preferred Sec61 or Ssh1 in comparison to the group that did not show a preference. *

*We also now add a citation from a new manuscript from the Hegde lab that suggests that indeed soluble SP containing proteins are the key clients for the translocon pore (https://pubmed.ncbi.nlm.nih.gov/36261528/). We have also added a section to the discussion as to why we do not see differential enrichment of SRP or its receptor on either pore although in the past this was suggested to be the key difference between the two translocons. *

Figures 1-2: They should comment on the reproducibility of the method. How many independent experiments were performed? If performed, how was reproducibility of results?

Thank you for highlighting that this was not clarified enough – we have now extended the materials and methods section to make all of the above issues clear and apparent. In short, we performed 3 biological repetitions for each experiment and for each biological repeat we performed 3 technical repeats making our results altogether rely on 9 repeats. We then excluded proteins in two cases

1. If strains were missing in either of the collections (so there was no complete set to compare them) – this caused us to drop 661 strains.
2. In cases where the proteins were expressed at very low levels of extracted poorly in our assay – we defined this as the signal being ten standard deviations (or more) lower than the rest of the signals on the same membrane – this caused us to lose an additional 93 strains. Importantly, the SD between all 9 repeats never rose above 3 (see graph below showing al strains arranged by order in library and the SD between all 9 repeats) and also now stated clearly in the text) hence we think that our method is highly reproducible

Figure 3: It is important to know the overlap of proteins commonly identified in both the interaction screening and protein localization assay. A Venn diagram that compares results between the two high-throughput assays would be useful.

*We have indeed considered making this Venn diagram (See below). However, since the connection between the screens is not direct due to the fact that the protein localization is downstream to translocation events or maybe completely independent of it, we found that the number of specific proteins that are in both screens is low. However, there is a much larger overlap in joining processes and functions, therefore we decide to make the grouping showed in Figure 4B. We would prefer not to show this figure in the manuscript however we leave this to editorial decision. *

Figure 4A (GO term): The authors mentioned that " the most consistent and repeating GO term group was those related to budding and polarity process. These include: "Establishment or maintenance of cell polarity"; "Development process involved in reproduction"; "Bipolar cellular bud site selection"; "Cell budding" and "Structural constituent of cell wall". Are protein sets in these functional categories similar or different? I am asking because GO enrichment analysis often provides apparently different functional categories but similar protein sets are included.

Indeed, this reviewer is totally correct and this is also the case here to some extent. There is some level of overlap between the GO terms. However, in our case this overlap is quite small: Out of the 77 genes that have one of those GO terms assigned only 2 have all 4, 9 have 3 and 4 have 2 of the GO terms therefore we believe that in this case this issue with GO terms hierarchy and assignment is not redundant. We are happy to highlight this in the figure or text if this is deemed to be important.

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:

Cohen et al. presented a high-throughput approach to analyze protein-(putative) substrate interactions in yeast using BirA biotin ligase and its acceptor peptide AVI tag. Using this system, the authors identified the common and unique substrates of translocation pores, Sec61 and Ssh1. Interestingly, the differential substrates between Sec61 and Ssh1 seem to be explained by the degree of hydrophobicity in signal peptide sequences, which was also nicely demonstrated by an experiment showing that swapping the first three amino acids of substrate proteins is sufficient to convert the substrate specificity. While I appreciate that the approach is high-throughput and simple (does not require mass spectrometers), there are some technical comments to be addressed.

1. Why was BirA used to study transient interactions? Biotinylation through BirA is slow (that is why it takes several hours to label proximity proteins) and thus it may not be suitable for capturing transient interactions. Instead, TurboID would be more suitable as the biotinylation reaction is faster than BirA. A reasonable explanation using BirA is required.
2. One key question is whether biotinylated proteins identified by this method are substrates or proteins just proximal to Sec61 or Ssh1 due to close cellular localization (e.g. ER membrane) or same protein complex members. An experiment or analysis would be required to confirm that the proteins they identified are indeed potential substrates.
3. Along the same line, if proteins identified by this approach are bona fide substrates of Sec61 and Ssh1, proteins having signal peptides should be enriched in the candidate list of substrates. However, it does not look like that according to Figure 2A where the secretome proteins/total proteins ratio appears to be similar among the 4 categories (e.g., Ssh1 preferring, No preference, and Not interacting or excluded). The authors should comment on this.
4. Figures 1-2: They should comment on the reproducibility of the method. How many independent experiments were performed? If performed, how was reproducibility of results?
5. Figure 3: It is important to know the overlap of proteins commonly identified in both the interaction screening and protein localization assay. A Venn diagram that compares results between the two high-throughput assays would be useful.
6. Figure 4A (GO term): The authors mentioned that " the most consistent and repeating GO term group was those related to budding and polarity process. These include: "Establishment or maintenance of cell polarity"; "Development process involved in reproduction"; "Bipolar cellular bud site selection"; "Cell budding" and "Structural constituent of cell wall". Are protein sets in these functional categories similar or different? I am asking because GO enrichment analysis often provides apparently different functional categories but similar protein sets are included. 

Referees cross-commenting

The comments from reviewer #1 are reasonable and would further strengthen the quality of the paper.

Significance

The approach is high-throughput and simple (does not require mass spectrometers).

The differential substrates between Sec61 and Ssh1 seem to be explained by the degree of hydrophobicity in signal peptide sequences, which was also nicely demonstrated by an experiment showing that swapping the first three amino acids of substrate proteins is sufficient to convert the substrate specificity.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

In this manuscript the authors develop an unbiased method to measure protein-protein interactions in cells through application of BirA biotin ligase technology. The novel feature of the approach is to append BirA to homologous ER translocon proteins (Sec61 and Ssh1) and then measure biotinylation of the AviTag when fused to the N-terminus of each protein in the yeast proteome. This is accomplished by mating the specific BirA tagged strain with a collection of 6000 yeast strains each with a distinct AviTagged protein. The level of AviTag biotinylation in diploid strains is measured by probing cell lysates applied to filters with fluorescent streptavidin. 2070 proteins displayed interactions with BirA-Sec61 and/or BirA-Ssh1 with a subset of secretory proteins showing preference for one of the translocons. The influence of ssh1 deletion on localization of the GFP-tagged protein library (4127 strains) was also determined and compared with protein interaction data. Moreover, analyses of signal peptides revealed that hydrophobicity of the N-terminal three residues can be sufficient to impart specificity for Sec61 or Ssh1 interaction. The reported findings support their primary conclusions. I have only a few suggestions to strengthen this study.

1. The Figure 1 legend indicates that the BirA tagged strains are mated with ~6000 AviTag yeast strains but results in Figure 2 pie chart account for 4812 total readouts. Presumably 1000 or more strains could not mate or did not produce viable diploids with the BirA tagged strains? It would be helpful to explain this differential.
2. If possible, suggest including more of the raw data (in supplementary) that supports the pie chart in Figure 2. Table S1 shows the 111 proteins that display preference for Ssh1 (out of 586 total interactors?) and the fold change (in rank order) for interaction preference. At a minimum, similar data on Sec61 preference and the list of positive interactors should be included. There may also be useful information in the relative biotinylation signal for each BirA and AviTag combination when significantly above background. This is presumably a readout of AviTag protein abundance, dwell time and orientation to BirA activity. The data could be useful to other investigators.
3. The authors might want to be more cautious in interpreting impact of the UPR on ssh1 phenotypes in the results and discussion. The Wilkinson et al 2002 paper referenced used very different conditions to detect UPR in ssh1 deletions strains. Jonikas et al 2009 does not detect a chronic UPR in ssh1 deletion cells and the conditions used in the current study seem more similar to the 2009 report. It seems more likely that deficits in translocating/localizing specific proteins causes the observed phenotypes instead of chronic UPR due to reduced ER levels of PDI.

Significance

The reported findings support their primary conclusions. The technology development and results are significant, highly relevant and will be of interest to a broad readership in cell and membrane biology.

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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements

        We greatly appreciate the valuable comments from the referees, which have generally been very positive and constructive. The three referees have emphasized the significance of our study that opens a new direction of research regarding the role of RNA modification in viral defense. In addition, the reviewers confirm our view that the audience of our work would be broad.

        The major concerns of the reviewers are limited to four main points:

1. i) to be clearer in our description on the effect of the m6A-YTHDF axis on the viral infectivity and avoid making assumptions on effects on replication (ref. #1 and #3);

2. ii) reviewer 1 finds that the title and conclusion of this manuscript defining YTHDF proteins (ECTs) as "direct effectors of antiviral immunity" is misleading. Nonetheless, as detailed below, Reviewer 1 confuses mere knowledge of effects of m6A with those conferred by YTHDF proteins binding to m6A, and indeed overlooks nearly all evidence presented in the paper for how m6A in AMV confers antiviral resistance (i.e. mechanistic insight); iii) the discussion on the relative importance of antiviral RNA silencing and m6A-YTHDF against AMV;

3. iv) to establish more clearly whether the phase separating capability of IDRs in the reading proteins correlates with the antiviral activity (reviewer 2). We have already completed substantial experimental work to address several of these points. Nonetheless, we find it prudent to ask for an extension of the revision time beyond four weeks to allow for repeats of a few of the infection experiments in question. In the following section, we specify a plan of action for the revisions.

2. Description of the planned revisions

• *Regarding the four major concerns raised by the reviewers, we will experimentally address the last two, whereas we think the first two do not need any further experimental work, as explained in section 4. Thus, the working plan for points #3 and #4 will be as follows:

iii) the discussion on the relative importance of antiviral RNA silencing and m6A-YTHDF against AMV and related viruses

As we mention in the manuscript (discussion, first chapter), AMV *“is one of only very few studied plant RNA viruses for which no anti-RNAi effector has been identified. In addition, prunus necrotic ringspot virus (PNRSV), a virus genetically and functionally closely related to AMV (Pallas et al, 2013), does not induce easily detectable siRNAs, unlike nearly all other studied plant RNA viruses (Herranz et al, 2015)”. *

Thus, we do not come up with a strong judgment on whether RNAi is more or less important than m6A-YTHDFs for AMV resistance.

In any case, although these indirect observations seem to be quite solid, we agree with the reviewer that conclusive evidence to discard RNAi as a defense layer against AMV, at least at the time where ECTs are acting, is lacking. Thus, we plan to evaluate how the absence of the main components of the RNAi machinery affects AMV infection and if this ‘universal’ defense layer interferes/overlaps with the ECTs antiviral defense observed here. Realistically, this will take us 8-10 weeks. The experiments within this topic are based on established and published methods and thus, on solid experience. We do not expect any fallback solution and the results will be conclusive in this sense. We also note that the very time-consuming part of constructing mutants defective in both RNAi and m6A-ECT components (in this case, ect2/ect3/rdr6), as well as a first round of infection assays has already been completed at this point

iv) To establish more clearly whether the phase separating capability of IDRs in the reading proteins correlates with the antiviral activity (Reviewer 2).

We agree with Reviewer 2 that this is an interesting and important question. Hence, we have teamed up with the group of Prof. Kresten Lindorff-Larsen, expert in molecular simulations of protein folding and interaction. The Lindorff-Larsen group has recently published a powerful computational approach to simulate phase separation behavior of intrinsically disordered proteins (IDPs) or regions of proteins (IDRs) (Tesei et al., 2021, Accurate model of liquid-liquid phase behavior of intrinsically disordered proteins from optimization of single-chain properties, PNAS 118, (44) e2111696118). Applying this simulation method to the Arabidopsis ECT proteins establishes two facts that we will incorporate into a revised version:

• The IDR of ECT2 shows marked phase separation propensity, in agreement with the experimental evidence published in Arribas-Hernández et al., 2018, Plant Cell.
• The deletion mutant of ECT2 (ΔN5) with defective antiviral activity, yet unaffected ability to accelerate growth of leaf primordia shows markedly reduced phase separation propensity driven, in the main, by the many tyrosine residues in the region deleted in the mutant. These results suggest that phase separation capability indeed correlates with antiviral activity.

Since not only ECT2, but also ECT3, ECT5 and, to some extent, ECT4, participate in AMV resistance, we plan further simulation work on these proteins during the first two weeks of January 2023 before submission of a revised version of the manuscript.

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

• • All the minor concerns raised by the three reviewers have been addressed and we have incorporated all of their suggestions in this intermediate version.

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

• *As previously mentioned, we believe that points 1 and 2 do not require an experimental approach to be addressed for the following reasons:

i) to be clearer in our description on the effect of the m6A-YTHDF axis on the viral infectivity and avoid making assumptions on effects on replication (ref. #1 and #3)

        We agree with the reviewer that the term 'inhibition of viral replication' was not very appropriate because the idea that was intended to be conveyed was that of viral accumulation.        Hence, we will change this use of language, and we thank the reviewer for pointing out this inaccurate description.

When it comes to differences between effects on infection in inoculated and non-inoculated leaves, there may be a slight misunderstanding, perhaps because we were not clear enough in our originally submitted version. In reality, there are some differences even in inoculated leaves between wild type and ect mutants, especially in the triple mutant, but the slightly higher accumulation in ect mutants is not clearly observed in every experiment and hence, does not always rise to the level of significance. Although it is possible that, at local level, ALKBH9B-mediated m6A would have other ECTs-independent effects, similar to what has been described for some animal viruses (Baquero-Pérez et al., 2021. Viruses), we think that the most likely explanation for this phenomenon is a combination of infection titers and ECT redundancy.

The suggestion to use protoplasts is very accurate, but it would not resolve any doubt in this scenario, because ECTs are mainly expressed in mitotically active cells (Arribas-Hernández et al, 2020, 2018) and, since mature tissues make up the better part of the leaves used to isolate protoplasts, only few of the isolated cells would be useful. In addition, we previously showed that AMV accumulation is reduced in alkbh9b protoplasts compared to WT (Martínez-Pérez et al., 2021. Front. Microbiol.), which suggests that m6A levels of vRNAs are critical for the first stages of the infection, but in that case no problems with the expression pattern of the demethylase were expected.

ii) The title and conclusion of this manuscript defined YTHDF proteins (ECTs) as "direct effectors of antiviral immunity", which is misleading. Effector molecules of an antiviral immunity cannot be identified when the effector mechanism is unknown;

In this regard, we have a very different vision from the one the reviewer proposes. We believe that it is not correct to say that the effector molecules of an antiviral immunity cannot be identified until its mechanism is demonstrated. In fact, RNA silencing effectors were discovered long before their mechanism was elucidated in detail. One molecular interpretation of the Flor’s seminal gene-for-gene model, in terms of receptor/effector recognition, is that specific interaction between the receptor and its recognized (cognate) effector protein triggers resistance.

Furthermore, we strongly believe that we provide enough arguments to propose a model, although, as we comment in the end of the discussion, “we view this model as a conceptual framework of value in the design of future experiments to test its validity”. The reasoning that we show here is the following:

1. The m6A binding proteins are necessary for the antiviral response.
2. At least ECT2 recognizes AMV RNAs in vivo and that its m6A-binding capacity is necessary to play a role in AMV infection.
3. Simply losing methylase activity – with the same developmental defects as ect2/3/ – does not lead to the same degree of loss of resistance, and you can affect AMV resistance without affecting developmental functions of ECT2. Altogether, these observations justify the proposal that m6A exerts antiviral effects by acting as binding sites of ECT proteins in viral RNA, which we consider a clear mechanistic advance.

Bearing in mind that m6A-modified vRNAs might concentrate in replication complexes and that MeRIP-seq methodology to map m6A revealed site multiplicity in the genome of some RNA viruses (Gokhale et al., 2016. Cell Host&Microb; Martínez-Pérez et al., 2017; Lichinchi et al., 2016. Nat Microbiol; Lichinchi et al., 2016. Cell Host&Microb; Marquez-Molins et al, 2022), our results recalled the previously proposed model in which m6A sites multiplicity causes the phase separation of these RNAs through the interaction of the IDRs of the YTH proteins (Ries et al, 2019; Fu & Zhuang, 2020; Gao et al, 2019). Now, with the new simulations of phase separation behavior, although still a model that requires further experimental tests, we have better evidence to support the model that it is related to LLPS of ECT-bound viral RNA. Therefore, we firmly believe that our title conceptually reflects the basic concepts of resistance induction in virus-plant interactions.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Previously, the authors showed evidence that m6A modifications of AMV RNAs erased by the host ALKBH9b enhances AMV spread in Arabidopsis. In this paper, the authors show by transcriptome analysis and RT-qPCR that the accumulation of m6A "reader" proteins, ECT2, 3 and 5 are increased during AMV infection in Arabidopsis. Combined mutations of ect2,3 and 5 led to increased AMV accumulation, suggesting that ECT2,3,5 are critical in inhibition of AMV accumulation in systemic leaves of Arabidopsis. Mutagenesis of ECT2 putative m6A-binding pocket did not restore AMV resistance in double-mutant de23 plant, arguing that m6A reader function of this protein is needed to provide resistance against AMV. Then, proximity-labeling was used to show that ECT2 binds to AMV RNA2 and likely RNA1 in planta. Finally, debilitating both the m6A eraser (ALKBH9b) and the reader (ect2,3,5) restored susceptibility to AMV infection in Arabidopsis, thus providing evidence that the m6A reader proteins are critical for resistance against AMV and that AMV exploits ALKBH9b to fight against ECT2,3,5 in plants. Altogether, these are novel and important findings. The paper is well-written.

I do not have main concerns.

Minor points:

• Abstract: "AMV replication" should be replaced by "AMV infection" or "AMV spread", because the locally infected leaves show similar AMV replication/accumulation in de23 and wt Arabidopsis.

• Alkbh9b mutation causes inhibition of local and systemic movement of AMV, whereas de23 mutant increases AMV accumulation only in systemic leaves (Fig. 1). What is the explanation that de23 does not affect local movement of AMV?

• The last chapter (p19-20) is too speculative and too long, it does not make the paper more interesting: I recommend shortening it and to minimize speculation.

Significance

This paper shows evidence for a new antiviral strategy present in plants. Overall, this is a significant new finding.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summary

In this manuscript, the authors asked the important question of how RNA modification is associated with viral defense in plants. Based on the previous findings that the infectivity of AMV relies on demethylation of viral RNA by recruitment of the cellular m6A demethylase ALKBH9B, in this study, they showed that inactivation of the m6A reader proteins, ECT2, ECT3, and ECT5, is sufficient to restore AMV infectivity in partially resistant alkbh9b mutants. Considering the potential roles of m6A modification in viral defense but the limited knowledge on this topic, the current study opens a new direction of research regarding the role of RNA modification in viral defense.

Major comments

• It is interesting to see that the IDR of ECT2 harbors two separable activities employed to achieve different goals: one that stimulates cellular proliferation by binding to endogenous m6A-containing mRNA, and one that effects basal antiviral resistance when ECT2 binds to hypermethylated viral RNA. Considering that the IDR of a protein contributes the chaperone activity of the protein and then can increase the binding capacity of the protein to different substrates, it would be more informative if the authors discuss whether the RNA chaperone activity of IDR of ECT2 is possibly involved in the different processes.

• It is interesting to propose a working model that m6A site multiplicity in AMV RNA may be a key factor distinguishing it from endogenous mRNA. In that sense, it would be clearer if the authors describe how many m6A sites are present in AMV RNA. Are these m6A sites clustered in certain regions of the viral RNA important for viral replication?

• It is not clear what the correlation between the phase separation capability of ECT proteins and viral infection is.

Minor comments

• In page 6, what is the alfalfa mosaic virus (AMV) RNA 3? Are there AMV RNA 1 and 2? What are their differences? Is the m6A-YTH module specific to AMV RNA 3?

• Name of the mutants; it would be better if same name was used for the mutant throughout the manuscript and in figures; for instance, ect2-1/ect5-2 and de25, as well as ect2-1/ect3-1/ect5-4 and te235 were used, which is not easy to follow.

• Figure 4G legend is missing.

Significance

Considering the potential roles of m6A modification in viral defense but the limited knowledge on this topic, the current study opens a new direction of research regarding the role of RNA modification in viral defense.

The audience will be broad, including any persons who are working on epitranscriptomics, plant sciences, viral infection, and clinical application. I am working on epitranscriptomic RNA modification in plant development and abiotic stress responses.

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 have shown previously that local and systemic infection of alfalfa mosaic virus (AMV) is inhibited and the relative abundance of m6A in the viral RNAs is increased in mutant Arabidopsis plants defective in the m6A demethylase gene ALKBH9B. Here the authors show that genetic inactivation of 2 or 3 m6A-binding proteins (ECTs or m6A readers) enhances the systemic, but not the local, infection of AMV. Notably, the systemic virus resistance of the demethylase mutant plants is largely eliminated by inactivating the same set of ECTs. Moreover, the authors detected in vivo association of ECT2 with AMV RNAs and identified a N-terminal motif of ECT2 that is necessary for AMV resistance, but dispensable for its role in endogenous developmental functions. These findings together provide further evidence to support their earlier conclusion for an antiviral role of m6A methylation of viral RNAs. Unfortunately, several key questions remain unknown and should be addressed to justify publication in EMBO J.

Major comments

1. The authors reported in 2017 that both the local and systemic infection of AMV is inhibited in the m6A demethylase mutant plants (PNAS 114:10755-60). In this work, they show that inactivation of ECTs enhances only the systemic AMV infection, but has no effect on the local infection (Fig. 2). Studies presented in neither Fig. 5 nor Fig. 6 examined possible effects on the local virus infection. Moreover, line 1 of page 13 mentioned a model that ECT binding "causes inhibition of viral replication". To resolve these contradictory descriptions on the step of the virus infection cycle targeted by m6A RNA methylation, it is essential to perform protoplast replication assays to determine whether the mutation of either the demethylase gene or ECTs affects viral RNA replication at single-cell level and to examine local infection for the studies presented in Figs 5 & 6.

2. The title and conclusion of this manuscript defined YTHDF proteins (ECTs) as "direct effectors of antiviral immunity", which is misleading. It remains completely unknown why m6A methylation of viral RNAs is inhibitory to virus infection. The available data suggest that it may act by blocking virus replication and/or movement or by enhancing any of the several known antiviral responses. Effector molecules of an antiviral immunity cannot be identified when the effector mechanism is unknown.

3. At the end of page 13 and elsewhere in this manuscript, the authors conclude that "the m6A-ECT axis constitutes a first, basal layer of antiviral defense", a conclusion that is not supported by the evidence presented. This conclusion will be incorrect if m6A methylation of viral RNAs does not inhibit virus accumulation levels in the protoplast virus replication assays as requested above.

4. In the first paragraph of page 17 and elsewhere in this manuscript, the authors question the relative importance of antiviral RNA silencing against AMV and related viruses as compared to m6A RNA methylation. It is important to determine if AMV becomes more virulent in RNAi-defective mutant plants such as dcl2/4 double mutant and if m6A RNA methylation also confers AMV resistance in RNAi-defective mutant plants.

Minor comments:

1. State the time post-inoculation when the samples were taken for the RNA-seq analysis in Fig.1.

2. State whether all of the northern blotting experiments used RNA extracted from single plant or pooled plants and had been repeated.

3. Verify that the statistical analysis method used in virus titer quantitative analysis is student t-test or one-way ANOVA.

4. The legend to Fig. 4F should be Fig. 4F and 4G.

5. Explain what ∆2 means in Fig. 6B.

6. If both the left and right bars correspond to 1 cm, 9-DAG-old plants would be much bigger than 16-DAG-old plants, which cannot be true.

Significance

Advance:

The authors have shown previously that local and systemic infection of alfalfa mosaic virus (AMV) is inhibited and the relative abundance of m6A in the viral RNAs is increased in mutant Arabidopsis plants defective in the m6A demethylase gene ALKBH9B. Here the authors show that genetic inactivation of 2 or 3 m6A-binding proteins (ECTs or m6A readers) enhances the systemic, but not the local, infection of AMV. Notably, the systemic virus resistance of the demethylase mutant plants is largely eliminated by inactivating the same set of ECTs. Moreover, the authors detected in vivo association of ECT2 with AMV RNAs and identified a N-terminal motif of ECT2 that is necessary for AMV resistance, but dispensable for its role in endogenous developmental functions. These findings together provide further evidence to support their earlier conclusion for an antiviral role of m6A methylation of viral RNAs.

Audience:

Broad, including those interested in RNA modifications, antiviral immunity, and plant biology.

Your expertise:

Antiviral immunity, plant biology

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer 1

Although this is an interesting, and generally well-performed study, it is primarily observational and there are few mechanistic insights provided into how MUC13 modulates barrier function. The authors propose a presumably direct interaction between MUC13 and PKC, which apparently sequesters PKC, preventing this kinase from triggering PKC-dependent increases in TJ barrier function; however, there is no evidence that a MUC13-PKC interaction occurs, that MUC13 is phosphorylated by PKC, or that phosphorylation of MUC13 has any impact on its function or overall barrier function. Thus, the hypothesis is not directly tested and all observations in this manuscript are generally correlative in nature.

While the MUC13 cytoplasmic tail contains a putative PKC-binding motif, we indeed do not show a direct interaction between MUC13 and a member of the PKC family in this manuscript. Unfortunately, we have so far not been able to successfully perform (co-)immunoprecipitation of MUC13 with our current anti-MUC13 antibodies.

To provide more insights into the possible MUC13-PKC interaction, we plan to perform several experiments.

• First, we will determine the expression levels of the different PKC isotypes (PKC alpha, beta, gamma, delta, epsilon, and zeta) in the HRT18 cell lines by western blot.
• Next, we will determine the localization of the relevant PKC isoforms and MUC13 by immunofluorescence microscopy. We are curious to see if we can find a colocalization between MUC13 and a PKC member on the lateral or apical membrane. If we can demonstrate a colocalization, we could follow up with a proximity ligation assay, but this would require the MUC13 antibody directed against the cytoplasmic tail (which only detects the lateral population) and might therefore be challenging.
• Furthermore, since PKC delta protein levels were upregulated in the total lysate of ∆MUC13 cells, we will test a PKC delta-specific inhibitor in the TEER assay.

  Consider quantifying all blots (Fig. 5C, Fig. 6B).

As suggested, we will quantify both blots.

Consider using dot-plots for all quantified data.

The graphs will be altered to include individual measurement points.

Reviewer 2

Fig2E showed two bands with different size in the two MUC13 WT control cell lines. They hypothesized that this could be the consequences of glycosylation different patterns. A sample with untransfected HRT18 might be included in the western blot panel. Additionally, what is the 100kDa band?

Mucin blots are notoriously difficult and these MUC13 blots are the result of a lot of trial and error. We repeated the Western Blot with original HRT18 cells, HRT18 original cell line, as well as the two CRISPR control cells used in the study (WT 1 and WT 2) and one of the full-length MUC13 knockout cells. The higher band was absent from the MUC13 knockout cells, but a small shift in the MUC13 band size can be noted in the WT 1 cells compared to the original and the WT 2 cell lines, possibly indicating a change in the glycosylation pattern. The 100 kDa band remains detectable in all cell lines including the ∆MUC13 cell line, therefore we consider this to be an aspecific background band of the MUC13 antibody. We will add a more extensive Western Blot analysis to the manuscript.

Did the transfection of the inducible GFP-MUC13 plasmid induce any decrease of Claudin1/3/4 in HRT18 or Caco2 cells? Same question regarding PKCdelta.

These are indeed interesting questions. We will perform these experiments with our MUC13-overexpression HRT18 cells.

Reviewer 3

Moreover, the authors should determine if MUC13∆CT localize to TJs, as suggested by the working model in Figure 7C. The subcellular localization of MUC3∆CT could give critical clues for its function, but Figure 2G fails to provide any information and the authors do not present any additional data concerning the localization of MUC13∆CT. Detection of MUC13 in membrane fractions of WT, MUC13∆CT and cells lacking the mucin domain could be a feasible strategy forward.

We will perform additional immunofluorescence experiments to determine the subcellular localization of MUC13-∆CT more accurately. However, detection of the extracellular domain by western blot, as suggested, is not possible due to the incompatibility of the extracellular MUC13-directed hybridoma antibody with the western blot technique. We currently do not have a suitable antibody that recognizes the ED and can be used for western blot.

The authors introduce an inducible MUC13-GFP fusion protein into WT and ∆MUC13 cells and show that it reverses the enhanced TEER upon MUC13 deletion. Unfortunately, the "Materials and Methods" section lacks adequate information on how this fusion protein was designed. Critical questions are the position of the GFP tag within MUC13, whether the fusion protein is correctly processed in HRT18 cells, and if it localizes to the apical or apico-lateral membrane domains? Figure 2H is of low magnification and fails to provide information on the subcellular localization of the MUC13-GFP fusion protein.

The materials and methods section will be adjusted to describe all the design details of the fusion protein. The GFP tag was added to the MUC13 C-terminus with a GGGS linker sequence in between. Processing of the fusion protein seems correct as we observed MUC13-GFP localization to both lateral and apical membranes and no access intracellular build up. As suggested by the reviewer, we will add more detailed immunofluorescence pictures to the manuscript.

Figures 6B-C suggest that PKCdelta levels increase in ∆MUC13 cells, which correlates with higher enrichment of Claudins in membrane fractions. The authors then inhibited PKCdelta and observed reduced recruitment of Claudins to membrane fractions. Since the family of Claudins are differentially regulated by phosphorylation (PMID: 29186552), the authors should investigate the TEER phenotype of WT, ∆MUC13 and MUC13∆CT upon PKC inhibition.

We must clarify that figures 6C-D are done using the PKC inhibitor targeting all conventional PKCs (alpha, beta, gamma) as well as delta (https://www.tocris.com/products/gf-109203x_0741). We recently obtained a PKCdelta-specific inhibitor which we will test in the TEER build-up experiments.

Moreover, the authors predict phosphorylation sites in MUC13CT and suggest a link between PKC and MUC13 (Figure. 6A), however no evidence is presented to support this hypothesis. The authors should either determine if PKC phosphorylates MUC13 and if this modification has implication on MUC13 localization and TJ function, or remove statements regarding MUC13 phosphorylation. The data provided suggest that PKC regulates TJ proteins independent of MUC13.

We will adjust the manuscript to put less emphasis on the putative PKC motifs in the MUC13 cytoplasmic tail. For further details on how we will proceed regarding the possible MUC13-PKC interaction see question 1 from reviewer #1.

Figure 5C. Quantification of at least 3 independent experiments is required.

These data will be added to the manuscript.

Figure 6B. Quantification of at least 3 independent experiments is required.

These data will be added to the manuscript.

Reviewer 4

OPTIONAL: MUC13 is expressed both, in the basolateral membranes and in the apical membrane of intestinal epithelial cells (IECs). Does the authors check the relevance of MUC13 in the formation of microvilli in IECs? Are microvilli different (microvilli staining, number of positive cells to microvilli, length, width or distribution of microvilli) in ΔMUC13 and in MUC13-ΔCT? How the glycocalyx looks like in these cells genetically modified for MUC13?

HRT18 cells do not seem to develop microvilli. However, we plan to stain these cells with a microvilli-specific antibody (ACTUB). The HRT18 cells express mostly MUC13 and relatively low levels of the larger TM mucin MUC1. To study changes in the glycocalyx, we will stain using a MAL-II antibody which targets α-2,3 sialic acids, which are abundantly present in mucins. In this way, we will determine any big changes in the total glycocalyx that may occur in response to the removal of MUC13.

In the figure 1D would be nice to represent the co-localization of MUC13 together with occluding in a graph in each Z-stack so you can visualize in which part of the cell is maximum colocalization of these both components.

These data will be provided.

In the figure 1E, would be great to compare between the two different MUC13 antibodies the apical fraction stained in HRT18 and Caco-2. Specially in the HRT18 cell line since the first antibody did not label apical MUC13 expression meanwhile the second antibody detects the apical expression in these cells. How much lateral lateral stain the C terminal antibody compare with the extracellular antibody for MUC13 and how much stain apically the C terminal antibody compare with the extracellular antibody? Would be nice to see some comparative results using the intensity by Z-stack and plotting in a graph.

This is a good suggestion as it is quite intriguing that both MUC13 antibodies seem to target (partially) different MUC13 populations. We will perform co-staining with both MUC13 antibodies to provide information on which MUC13 populations are detected by each antibody (apical vs lateral membrane).

Manuscript would be improved if in the figure 2H to compare within the same cell line the number of MUC13 positive cells in the WT, number of MUC13 positive cells in WT+pMUC13 and the number of MUC13 positive cells in the ΔMUC13+pMUC13

We will quantify the percentage of MUC13-GFP positive cells in both the WT and ΔMUC13 backgrounds by either microscopy or flow cytometry.

In figure 5C would be helpful to plot in a graph the normalized expression of each TJ protein and compare between the different cells used (WT, ΔMUC13 and MUC13-ΔCT) as you did in figure 5A

We will provide the quantification data of three independent experiments.

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

Reviewer 1

In addition, this model does not explain why all kinase inhibitors tested reverse the increase in TER observed in deltaMUC13 cell lines. Does this reflect the lack of inhibitor specificity or the likelihood that many kinases are involved?

As stated in the manuscript, we think that MLCK, ROCK, and PKC are all essential for TER buildup in the ∆MUC13 cells. Because the roles of MLCK and ROCK are well established, we choose to follow up on the PKC results. We adjusted the text to clarify this point.

The authors do observe that there is an increase in expression of several tight junction-associated proteins, including the claudins, in deltaMUC13 cells. Affected CLDNs include 1, 2, 3, 4, 7, 12. (1) While it appears the authors are arguing that this increased claudin expression results in increased barrier function, they do not sufficiently highlight the well-known role that CLDN2 has in cation transport, and both CLDN-4 and -7 have also been implicated in paracellular ion flux (although this is apparently cell-type specific). These observations would seem to argue against a simple correlation between claudin expression and tight junction barrier function.

The reviewer is right about the different functions of claudins. Claudin-2, -4 and -7 have (potentially) pore-forming properties, while the other claudins restrict paracellular passage. It has been previously demonstrated that the magnitude of paracellular ion and water flux is reflected by the specific repertoire of claudin family members (Shashikanth et al., 2022). In this paper, overexpression of claudin-4 was shown to mobilize and affect polymeric strands of claudin-2, thus blocking its channel activity. Our mass spectrometry data demonstrated a striking increase in claudin-1, -2, -3, -4, -7, and -12 in the MUC13 knockout membranes compared to WT. We hypothesize that the claudin repertoire in the MUC13 knockout cells leads to a more restricted paracellular route (as observed in the TEER and tracer experiments). The pore-forming claudins may be subject to “interclaudin interference” therefore leading to restriction of the total paracellular ion and water flux. We have adjusted the text of the manuscript to clarify this point.

We attempted to investigate claudin-2 expression levels in isolated membranes by Western Blot but were unsuccessful as the antibody did not detect any protein while claudins-1 and -4 could be detected with the same method.

Furthermore, the authors should note the disconnect between paracellular ion flux mediated by claudins and the flux of markers such as dextrans and lucifer yellow, which can be dissociated from claudin function.

We acknowledge that the flux of larger particles (the leak pathway) is not regulated by claudins (which regulates the pore pathway). We aimed to assess both the pore and the leak paracellular pathways, by using different techniques including TEER, small solutes (Lucifer Yellow CH), and larger molecules (4 and 70 kDa FITC-Dextrans). HRT18 wild type cells are already very restrictive to the pass of larger molecules (FITC-Dextrans) but are more permeable to smaller solutes such as Lucifer Yellow (400 Da). We observed that removal of the MUC13 cytoplasmic tail did not affect the TEER, but reduced the paracellular passage of Lucifer Yellow, demonstrating that manipulation of MUC13 can affect both the pore and leak pathways. We adjusted to text to include this point.

The increased expression of claudins in the nominally tail-minus MUC13 without a corresponding change in TER would again seem to argue against a simple correlation;

MUC13-dCT cells showed consistently increased levels of claudins-1 and -2, but not the other claudins. This claudin repertoire (with high claudins-1 and -2, but lower claudin-3, -4, -7, and -12) is apparently not enough to increase TEER. We think that this again reflects the importance of the total claudin composition for the control of the paracellular pathway.

Watch the use of decimal points instead of commas (lines 253 and 256).

Corrected.

Line 543: MilliQ is not a washing agent (or is it?). (Line 535) We use MilliQ as a final step before mounting the glass slides to remove any possible salt deposition that would affect the visualization by microscopy.

We have specified this in the text.

Line 553: TER is the product of total resistance times the area. The units are ohms times area.

Indeed, we have changed this mistake (line 545).

Line 630: Please provide the transfer conditions (voltage, amp, watts?) and transfer buffer when describing the Western blot protocol.

For immunoblotting of MUC13, protein lysates were transferred to 0.2 µm PVDF membranes using the Trans-Blot Turbo Transfer system (Biorad). The transfer was run using the protocol (High MW) which consisted in running for 10 min at 25 volts (V) and 1,3 amperes (A). These experimental data were added to the manuscript.

Reviewer 2

My main concern about this manuscript is that the authors analyzed MUC13 role in intestinal homeostasis and function using colorectal cancer cells. As helpful as cancer cells are, we should always be cautious about extrapolating roles in normal intestinal epithelium or IBD pathology. Obviously, these finding are also interesting in a cancer context. Using GEPIA (http://gepia.cancer-pku.cn/), I observed that MUC13 is overexpressed in colorectal cancer COAD-TCGA dataset (compared to normal colon from GTEX). Similar results were obtained previously by Gupta et al. (ref #10). I am aware that this would be difficult to confirm the main findings in a non-cancerous intestinal cell line but this limit (normal intestine using cancer cells) should be at least discussed in the manuscript.

We appreciate the reviewers’ comments and are aware of the downsides of using cancer-derived cell lines. We have performed the GEPIA analysis ourselves and have an ongoing project about the possible role of MUC13 in colorectal cancer progression. In a separate project, we are collaborating with the Gaultier Laboratory at the University of Virginia which has generated a MUC13 knockout mouse. This model will allow us to study the role of MUC13 in non-cancerous tissue. We recently received intestinal biopsies from these mice which will be stained with MUC13 and claudin antibodies to determine localization in healthy tissue. These experiments will reveal if MUC13 colocalizes with claudin on the lateral membrane in the healthy mouse intestinal tract. In future experiments, we will also address MUC13 localization and function in human intestinal organoids. We have adjusted the discussion to refer to the limitations of using cancer cell lines.

Massey et al (Micro 2021, PMC7014956) previously showed that MUC13 overexpression increased rigidity in PDAC cells and discussed involvement MUC13 link with EMT. MUC13-Her2 interaction was also associated with decrease of E-cadherin suggesting an EMT phenotype. This should be included in the discussion section.

The discussion has been adjusted to include the link with EMT.

The authors performed mass spectrometry analysis. Results are deposited on ProteomeXchange but are not yet publicly released. Among the 1189 membrane protein identified. Did the authors observed alteration of EMT proteins? (decrease of vimentin for example). In the discussion section (lane 347), the authors mentioned the relationship between other membrane bound mucins such as MUC1, MUC4, MUC16 or MUC17 and AJ/TJproteins. Did the authors observed any alteration of these mucin in the mass spectrometry data?

The mass spec analysis was performed on membrane fractions, therefore our dataset will not contain true cytosolic proteins. One of the key EMT proteins, Vimentin, is a cytosolic protein, and indeed it was not found in our dataset. Other EMT-related proteins are shown in the following table. TGF beta 1 was slightly decreased, while E-cadherin and Integrin beta 6 were slightly increased in the ∆MUC13 cells compared to WT cells.

Gene Name

Mean WT

Mean ∆MUC13

Mean MUC13-∆CT

TGFBI (TGB beta 1)

20,54

16,48

18,83

CDH1 (E-cadherin)

22,69

24,57

24,24

ITGB6 (Integrin beta 6)

18,86

21,74

19,19

Vimentin - Cytosolic

-

-

-

CDH2 (Cadherin-2, N-cadherin)

-

-

-

Mucins are large proteins comprised of densely O-glycosylated mucin domains, which makes them extremely challenging to study by mass spectrometry (MS) (Rangel-Angarita et al., 2021). We did not specifically employ mucin-directed technologies in this dataset, thus making the detection of mucins hard. No mucins other than MUC13 were detected. For MUC13, two peptides corresponding to the EGF-like domains in the extracellular domain, a region that is less densely glycosylated. We added a sentence to the description of the mass spec results to include the EMT proteins and other mucins.

Minor points:

Lane 126: HRT18 and Caco2 colon cancer cells instead of intestinal epithelial cells

Corrected.

Lane 181 and lane 514: add "full length" MUC13 DNA sequence

Corrected.

Lane 234: TEER was measured every 12h. How the authors did observed the largest increase at 42h? Was it 48h? Please clarify.

We aimed at measuring every 12 h, however the exact measurements were done at 18h, 24h, and 42 h post-infection. We have corrected this in the manuscript.

Reviewer 3

Line 43 and 46. "Enterocytes" should be replaced with "intestinal epithelial cells", since enterocytes are themselves a distinct subpopulation of IECs.

We have changed it in the manuscript.

Lines 58-60. References in support of the statements should be added.

We added a reference to this sentence.

Lines 188-190. Authors comment on "roundness" of different cell lines. If the parameter is critical for the manuscript, the authors should quantify this phenotype.

The parameter is not critical for the manuscript. We removed the sentence.

Figure 3A. Staining of cell lines should include panels showing localization of MUC13.

Co-staining of MUC13 with occludin in HRT18 cell lines can be found in figure 1D, and MUC13 with E-cadherin in supplementary figure 1.

Lines 323-327 and 390-392. Sentences on these lines contradict each other. The sentences should describe/discuss quantified data presented in Figure 6D.

The reviewer is right that we should be discussing the quantified data in 6D. We adjusted the sentence in line 323-327.

Proteomic data sets should be made publicly available on data depositories.

All proteomics raw data were deposited to the ProteomeXchange Consortium with the dataset identifier PXD029606.

Reviewer 4

OPTIONAL: In the figure 2E, is the extracellular antibody still detecting the MUC13-ΔCT?

No, unfortunately the antibody directed against the MUC13 ED is not compatible with western blot.

In the figure 2G, would be nice to comment possible reasons why the deletion in the first cell line of the MUC13-CT you can still detect with the extracellular antibody some lateral expression of MUC13 meanwhile in the second cell line, the same deletion (MUC13-CT) you cannot see any lateral MUC13 staining with the extracellular antibody.

Yes, this is indeed a puzzling finding, especially because the CRISPR deletion is the same in both cell lines. We will add a sentence about possible reduced stability of the MUC13 without CT domain that leads to a different outcome in both cell lines.

It would be nice that the results from Figure 3H are better explained since it is difficult to follow.

We adjusted the text to explain the experiment in more detail.

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

Reviewer 1

The authors may be overly reliant on TER measurements. Epithelial cells have two parallel resistive pathways: transcellular and paracellular. TER measure the contribution of both. Thus, an increase in TER could result from a decrease in transcellular ion transport. The authors need to measure transcellular ion flow or selectively measure the junctional resistance in a select set of experiments to rule this possibility out.

The reviewer is right that TEER is a sum of the resistance of the transcellular and paracellular pathways. However, due to the high resistance of cell membranes, the current predominantly travels via the paracellular route (Elbrecht et al., 2016). For this reason, TEER measurements are widely accepted techniques for the assessment of ions passage through the paracellular pathway (Shen et al., 2011).

Reviewer 3

Figure 1C. Caco2 and HRT18 cells exhibit distinct MUC13 expression patterns when probed with an antibody against the MUC13 CT; MUC13 localizes almost exclusively to lateral cell junction in HRT18 cells, while a higher portion of MUC13 is present on the apical surface of Caco2 cells. This observation has two possible explanations: 1) the two cell lines express distinct forms of MUC13, or 2) the two cell lines carry distinct machineries for anchoring MUC13 to apical versus apico-lateral membranes. Thus, The authors should take the opportunity to determine the impact of MUC13 deletion on TEER and TJ function in Caco2 cells. Proteomic analysis and functional assays in Caco2 cells may provide more a general mechanism for how MUC13 regulates TJ proteins.

Yes, this would be a great line of investigation. However, we aimed to knockout MUC13 in Caco-2 cell lines (with the same CRISPR/Cas9 protocol as the HRT18 cells) but were unable to obtain Caco-2 knockout clones. We think this might be a consequence of the poor capability of Caco-2 cells to grow as single colonies (a required step in the protocol). Another option is Caco-2 MUC13 knockout cells have reduced viability.

The authors generate cell lines that either lack MUC13 or express MUC13 lacking the cytoplasmic domain. Loss of MUC13 cells resulted in enhanced TEER and increased recruitment of TJ proteins to membrane fractions. MUC13∆CT cells show moderate recruitment of TJ proteins to membranes and no increase in TEER but inhibit paracellular diffusion of Luciferase Yellow across monolayers. Figure 3A suggests that Occludin redistributes to tricellular junctions in ∆MUC13 cells, whereas it is found more laterally in WT and MUC13∆CT cells. These finding suggest that full-length MUC13 interferes with TJ protein complexes. However the impact of the extracellular and intracellular (CT) domains is not fully elucidated. Does the O-glycosylated mucin domain interfere with the extracellular domains Occludin and Claudins? The authors should clarify the contribution of the mucin domain to the observed phenotype, for example by performing the described experiments in a cell line expressing MUC13 lacking the mucin domain.

Mucins are type I membrane proteins with the N-terminal part of the protein on the extracellular site. Therefore, a CRISPR method to specifically remove the glycosylated domain but leave the remainder of the protein in frame is challenging. An additional difficulty is that the ED contains a lot of repeats, complicating the design of specific guide RNAs. To specifically address the contribution of the glycosylated domain, we could complement the MUC13 knockout cell with a construct lacking the ED. However, this would not be comparable to the endogenous MUC13∆CT cell line presented in this manuscript. In future studies, we will strive to address the functions of the different MUC13 domains in more detail.

Figure 5A. Turnover of TJ proteins in membrane fractions occurs faster than over a period of 1-3 days (PMID: 18474622). The authors should determine TJ protein turnover over a period of minutes and hours.

We acknowledge the findings in this interesting paper concerning the continuous remodeling of tight junctions. However, the readout of our biotinylation assay is degradation and the timeframe of degradation turns out to be days and not hours. Within this timeframe remodeling is taking place but it cannot be captured in the total lysate.

Reviewer 4

OPTIONAL: The authors show that the probiotic Lactobacillus plantarum increase epithelial barrier independently of MUC13. Have the authors considered to use other probiotics as Lactobacillus paracasei (10.3389/fcimb.2015.00026), Akkermansia muciniphila (10.1038/emm.2017.282) or some metabolic products from intestinal microbiota as short-chain fatty acids (SCFAs) (10.3389/fphys.2021.650313) to check what is the role of MUC13 and if it is related with other microbe or microbiota metabolite?

Thank you for the suggestion. We have an ongoing project in which we investigate the impact of different probiotic bacteria and plan to investigate whether they have an impact on the epithelial barrier function in a MUC13-dependent manner. This study will lead to a separate publication.

OPTIONAL: The authors successfully delete MUC13 in IECs, both, full length and the cytosolic tail. Have the authors considered targeting the deletion of the PTS domain in MUC13? Could affect that something different from paracellular trafficking as the extracellular detection of microbes and microbial products?

Removal of a domain in the extracellular domain of MUC13 with CRISPR is challenging because mucins are type I membrane proteins, the repeats and possible frameshift, as 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 #4

Evidence, reproducibility and clarity

This work highlight the importance of the transmembrane mucin MUC13 in the control of the intestinal epithelial integrity by the negative regulation of tight junction (TJ) proteins mediated by Protein Kinase C (PKC). To achieve this conclusion, the authors designed CRISPR/Cas9 strategie to generate two types of HRT18 and Caco-2 MUC13 knockout cell lines: one to delete the full MUC13 length (ΔMUC13) and other to target the deletion of the MUC13 cytoplasmic tail (MUC13-ΔCT). More, they design MUC13-GFP plasmid to overexpress MUC13 in WT cells and to rescue the ΔMUC13 cells.

The key conclusions of this project are:

• Transepithelial electrical resistance (TEER) was upregulated in ΔMUC13 with no changes in MUC13-ΔCT compared with the control. The rescue of ΔMUC13 by MUC13-GFP plasmid reduce the TEER to values similar to the WT cells. More, ΔMUC13 and MUC13-ΔCT were more restrictive in the paracellular translocation of small tracers than WT cells suggesting the negative correlation of MUC13 and paracellular conductance of small ions together with higher TEER.
• Absence of MUC13 leads to an upregulation of TJ proteins, which explain the decrease of the paracelullar ion traffic
• Upregulation of TEER in cells lacking MUC13 is dependent on MLCK, ROCK and PKC kinases meanwhile the upregulation of TJ proteins is mediated by PKC proteins. Finally, the authors intend to address in the future the molecular link between MUC13 and PKC during TJ regulation.

The conclusions and the claims from the authors of this work are supported by the data where they extensively test the close relation between the transmembrane mucin MUC13 and the intestinal epithelial integrity.

Major comments:

• OPTIONAL: The authors show that the probiotic Lactobacillus plantarum increase epithelial barrier independently of MUC13. Have the authors considered to use other probiotics as Lactobacillus paracasei (10.3389/fcimb.2015.00026), Akkermansia muciniphila (10.1038/emm.2017.282) or some metabolic products from intestinal microbiota as short-chain fatty acids (SCFAs) (10.3389/fphys.2021.650313) to check what is the role of MUC13 and if it is related with other microbe or microbiota metabolite?
• OPTIONAL: MUC13 is expressed both, in the basolateral membranes and in the apical membrane of intestinal epithelial cells (IECs). Does the authors check the relevance of MUC13 in the formation of microvilli in IECs? Are microvilli different (microvilli staining, number of positive cells to microvilli, length, width or distribution of microvilli) in ΔMUC13 and in MUC13-ΔCT? How the glycocalyx looks like in these cells genetically modified for MUC13?
• OPTIONAL: The authors successfully delete MUC13 in IECs, both, full length and the cytosolic tail. Have the authors considered targeting the deletion of the PTS domain in MUC13? Could affect that something different from paracellular trafficking as the extracellular detection of microbes and microbial products?
• OPTIONAL: In the figure 2E, is the extracellular antibody still detecting the MUC13- ΔCT?

Minor comments:

• In the figure 1D would be nice to represent the co-localization of MUC13 together with occluding in a graph in each Z-stack so you can visualize in which part of the cell is maximum colocalization of these both components.
• In the figure 1E, would be great to compare between the two different MUC13 antibodies the apical fraction stained in HRT18 and Caco-2. Specially in the HRT18 cell line since the first antibody did not label apical MUC13 expression meanwhile the second antibody detects the apical expression in these cells. How much lateral lateral stain the C terminal antibody compare with the extracellular antibody for MUC13 and how much stain apically the C terminal antibody compare with the extracellular antibody? Would be nice to see some comparative results using the intensity by Z-stack and plotting in a graph.
• In the figure 2G, would be nice to comment possible reasons why the deletion in the first cell line of the MUC13-CT you can still detect with the extracellular antibody some lateral expression of MUC13 meanwhile in the second cell line, the same deletion (MUC13-CT) you cannot see any lateral MUC13 staining with the extracellular antibody.
• Manuscript would be improved if in the figure 2H to compare within the same cell line the number of MUC13 positive cells in the WT, number of MUC13 positive cells in WT+pMUC13 and the number of MUC13 positive cells in the ΔMUC13+pMUC13
• It would be nice that the results from Figure 3H are better explained since it is difficult to follow.
• In figure 5C would be helpful to plot in a graph the normalized expression of each TJ protein and compare between the different cells used (WT, ΔMUC13 and MUC13-ΔCT) as you did in figure 5A

Significance

This is a novel study where the authors directly correlate the lack of MUC13 expression with paracellular transport and tight junction proteins. This study describe the high correlation between the transmembrane mucin MUC13 and the integrity of the intestinal epithelium. Therefore, this project is highly valuable not only for the scientific research nowadays, but for future investigations of the intestinal epithelial physiology and biochemistry.

The strengths parts of this study are the different cell constructs including the full deletion of MUC13 and the targeting deletion of MUC13 cytosolic tail. Then, they have been able to directly correlate lack of MUC13 with paracellular traffic where PKC intracellular signal is involved. A limitation of the study is the lack of a cell line lacking the extracellular domain of MUC13, which could give some clues about the direct relation of this membrane mucin with the outer world of the cell (i.e. bacteria).

My field of expertise is intestinal epithelial defense including, but not limited, to mucins.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary:

The authors describe a novel function for transmembrane mucin MUC13 in regulation of tight junctions (TJs) that create an impermeable cell monolayer that allows. paracellular diffusion of very small molecules. The authors use cultured intestinal epithelial cell monolayers. to demonstrate that MUC13 localizes to the apical aspects of IECs as well laterally to tight junctions. CRSIPR/Cas-mediated deletion of MUC13 increased transepithelial resistance (TEER) and reduced the extent of paracellular diffusion of <0.5 kDa molecules across the monolayer. Proteomic analysis revealed that specific TJ proteins are enriched in cell membrane fractions upon deletion of MUC13, while pharmacological inhibition of PKC involved in actomyosin contractility, resulted in loss of TJ proteins from cell membranes and TEER reduction. See major comments for a detailed discussion concerning the findings.

Major comments:

1. Figure 1C. Caco2 and HRT18 cells exhibit distinct MUC13 expression patterns when probed with an antibody against the MUC13 CT; MUC13 localizes almost exclusively to lateral cell junction in HRT18 cells, while a higher portion of MUC13 is present on the apical surface of Caco2 cells. This observation has two possible explanations: 1) the two cell lines express distinct forms of MUC13, or 2) the two cell lines carry distinct machineries for anchoring MUC13 to apical versus apico-lateral membranes. Thus, The authors should take the opportunity to determine the impact of MUC13 deletion on TEER and TJ function in Caco2 cells. Proteomic analysis and functional assays in Caco2 cells may provide more a general mechanism for how MUC13 regulates TJ proteins.
2. The authors generate cell lines that either lack MUC13 or express MUC13 lacking the cytoplasmic domain. Loss of MUC13 cells resulted in enhanced TEER and increased recruitment of TJ proteins to membrane fractions. MUC13∆CT cells show moderate recruitment of TJ proteins to membranes and no increase in TEER but inhibit paracellular diffusion of Luciferase Yellow across monolayers. Figure 3A suggests that Occludin redistributes to tricellular junctions in ∆MUC13 cells, whereas it is found more laterally in WT and MUC13∆CT cells. These finding suggest that full-length MUC13 interferes with TJ protein complexes. However the impact of the extracellular and intracellular (CT) domains is not fully elucidated. Does the O-glycosylated mucin domain interfere with the extracellular domains Occludin and Claudins? The authors should clarify the contribution of the mucin domain to the observed phenotype, for example by performing the described experiments in a cell line expressing MUC13 lacking the mucin domain. Moreover, the authors should determine if MUC13∆CT localize to TJs, as suggested by the working model in Figure 7C. The subcellular localization of MUC3∆CT could give critical clues for its function, but Figure 2G fails to provide any information and the authors do not present any additional data concerning the localization of MUC13∆CT. Detection of MUC13 in membrane fractions of WT, MUC13∆CT and cells lacking the mucin domain could be a feasible strategy forward.
3. The authors introduce an inducible MUC13-GFP fusion protein into WT and ∆MUC13 cells and show that it reverses the enhanced TEER upon MUC13 deletion. Unfortunately, the "Materials and Methods" section lacks adequate information on how this fusion protein was designed. Critical questions are the position of the GFP tag within MUC13, whether the fusion protein is correctly processed in HRT18 cells, and if it localizes to the apical or apico-lateral membrane domains? Figure 2H is of low magnification and fails to provide information on the subcellular localization of the MUC13-GFP fusion protein.
4. Figures 6B-C suggest that PKCdelta levels increase in ∆MUC13 cells, which correlates with higher enrichment of Claudins in membrane fractions. The authors then inhibited PKCdelta and observed reduced recruitment of Claudins to membrane fractions. Since the family of Claudins are differentially regulated by phosphorylation (PMID: 29186552), the authors should investigate the TEER phenotype of WT, ∆MUC13 and MUC13∆CT upon PKC inhibition. Moreover, the authors predict phosphorylation sites in MUC13CT and suggest a link between PKC and MUC13 (Figure. 6A), however no evidence is presented to support this hypothesis. The authors should either determine if PKC phosphorylates MUC13 and if this modification has implication on MUC13 localization and TJ function, or remove statements regarding MUC13 phosphorylation. The data provided suggest that PKC regulates TJ proteins independent of MUC13.

Minor comments:

1. Line 43 and 46. "Enterocytes" should be replaced with "intestinal epithelial cells", since enterocytes are themselves a distinct subpopulation of IECs.
2. Line 59. The authors should note that MUC13 does not have a canonical SEA domain that generates a cleaved heterodimer (PMID: 16369486).
3. Lines 58-60. References in support of the statements should be added.
4. Lines 188-190. Authors comment on "roundness" of different cell lines. If the parameter is critical for the manuscript, the authors should quantify this phenotype.
5. Figure 3A. Staining of cell lines should include panels showing localization of MUC13.
6. Figure 5A. Turnover of TJ proteins in membrane fractions occurs faster than over a period of 1-3 days (PMID: 18474622). The authors should determine TJ protein turnover over a period of minutes and hours.
7. Figure 5C. Quantification of at least 3 independent experiments is required.
8. Figure 6B. Quantification of at least 3 independent experiments is required.
9. Lines 323-327 and 390-392. Sentences on these lines contradict each other. The sentences should describe/discuss quantified data presented in Figure 6D.
10. Proteomic data sets should be made publicly available on data depositories.

Significance

Mucins participate in critical functions in the human intestine. Gel-forming mucins form the mucus layers that separate the gut microbiota from the underlying intestinal epithelial cells (IECs) (PMID: 18806221). Transmembrane mucins are instead anchored to the plasma membrane of various populations of IECs (PMID: 32169835; PMID: 28052300). Despite its discovery over 20 years ago, the functional role of MUC13 in the intestinal epithelium is still debated. MUC13 is expressed in human small intestine and colon under baseline conditions and is dysregulated during inflammation and tumorigenesis, as described by the authors. Thus, understanding how MUC13 expression and localization impact cell function is of great importance for elucidating its function in health and disease. Studies so far have identified transmembrane mucins as biophysical barriers against bacteria (PMID: 33596425) or facilitators of bacterial invasion (PMID: 33824202). The current manuscript can potentially offer novel conceptual insights into how transmembrane mucins govern the integrity of the epithelial monolayer that serves as a firewall between the multitude of microbes in the gut lumen and the immune system. Such insights have implication for both basic and clinical research on inflammatory bowel disease (IBD) and colorectal cancer (CRC). However, while the authors present convincing data that deletion of MUC13 enhances TEER and recruitment of TJ proteins, the study in its current form fail to provide mechanistic proof of how MUC13 impacts individual TJ proteins. Moreover, it is not clear if findings in a specific cultured cell line (HRT18) can be extrapolated to other frequently used intestinal cell lines (e.g. Caco2) and IECs in an in vivo setting. The latter is particularly important since the authors argue that their findings have important implication in intestinal inflammation and cancer.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

Summar: In this manuscript, Segui-Perez and colleagues investigated the role of MUC13 in tight junctions and trans-epithelial barrier function in colon cancer cells. The authors showed that MUC13 is highly expressed throughout the intestine at the apical and lateral membrane. They established Crispr/Cas9 HRT18 cells in which MUC13 (deltaMUC13) or the cytoplasmic tail of MUC13 were deleted. They also performed rescue experiments using GFP-MUC13 constructs. The authors observed that deletion of MUC13 promoted TEER for bigger particles and strengthen tight junctions. Analysis of membrane composition by mass spectrometry showed an upregulation of TJ proteins (Claudins) that is dependent of PKCdelta.

Major points:

1. My main concern about this manuscript is that the authors analyzed MUC13 role in intestinal homeostasis and function using colorectal cancer cells. As helpful as cancer cells are, we should always be cautious about extrapolating roles in normal intestinal epithelium or IBD pathology. Obviously, these finding are also interesting in a cancer context. Using GEPIA (http://gepia.cancer-pku.cn/), I observed that MUC13 is overexpressed in colorectal cancer COAD-TCGA dataset (compared to normal colon from GTEX). Similar results were obtained previously by Gupta et al. (ref #10). I am aware that this would be difficult to confirm the main findings in a non-cancerous intestinal cell line but this limit (normal intestine using cancer cells) should be at least discussed in the manuscript.
2. Massey et al (Micro 2021, PMC7014956) previously showed that MUC13 overexpression increased rigidity in PDAC cells and discussed involvement MUC13 link with EMT. MUC13-Her2 interaction was also associated with decrease of E-cadherin suggesting an EMT phenotype. This should be included in the discussion section.
3. Fig2E showed two bands with different size in the two MUC13 WT control cell lines. They hypothesized that this could be the consequences of glycosylation different patterns. A sample with untransfected HRT18 might be included in the western blot panel. Additionally, what is the 100kDa band?
4. The authors performed mass spectrometry analysis. Results are deposited on ProteomeXchange but are not yet publicly released. Among the 1189 membrane protein identified. Did the authors observed alteration of EMT proteins? (decrease of vimentin for example). In the discussion section (lane 347), the authors mentioned the relationship between other membrane bound mucins such as MUC1, MUC4, MUC16 or MUC17 and AJ/TJproteins. Did the authors observed any alteration of these mucin in the mass spectrometry data?
5. Did the transfection of the inducible GFP-MUC13 plasmid induce any decrease of Claudin1/3/4 in HRT18 or Caco2 cells? Same question regarding PKCdelta.

Minor points:

1. Lane 126: HRT18 and Caco2 colon cancer cells instead of intestinal epithelial cells
2. Lane 181 and lane 514: add "full length" MUC13 DNA sequence
3. Lane 234: TEER was measured every 12h. How the authors did observed the largest increase at 42h? Was it 48h? Please clarify.

Significance

This manuscript is relevant as basic research for both the mucin field and for the intestinal epithelium field. It brings conceptual hypothesis about the role of MUC13 that is less characterized than MUC1 or MUC4.

I have been working on mucins for over 20 years. I found this work well done and very interesting.

I feel that the conclusions are mostly supported by the results. The one semantic limit is that this work is based on cancer cell lines and it is a little bit speculative to extrapolate the finding on normal intestinal epithelium.

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

Learn more at Review Commons

---

Referee #1

Evidence, reproducibility and clarity

The manuscript by Sequi-Perez explores a somewhat novel role for the transmembrane mucin MUC13 in regulating the tight junction barrier. They report that MUC13 is localized, in part, to the apical junctional complex; that depletion of MUC13 increases TER and expression of claudins and OCLN in a membrane fraction; and that partial removal of MUC13 cytoplasmic domain also results in increased expression of OCLN and claudins, but without a corresponding change in TER. Finally, the authors hypothesize that PKCdelta may act in conjunction with MUC13 to regulate paracellular flux in intestinal epithelial cell lines.

Major points:

1. Although this is an interesting, and generally well-performed study, it is primarily observational and there are few mechanistic insights provided into how MUC13 modulates barrier function. The authors propose a presumably direct interaction between MUC13 and PKC, which apparently sequesters PKC, preventing this kinase from triggering PKC-dependent increases in TJ barrier function; however, there is no evidence that a MUC13-PKC interaction occurs, that MUC13 is phosphorylated by PKC, or that phosphorylation of MUC13 has any impact on its function or overall barrier function. Thus, the hypothesis is not directly tested and all observations in this manuscript are generally correlative in nature. In addition, this model does not explain why all kinase inhibitors tested reverse the increase in TER observed in deltaMUC13 cell lines. Does this reflect the lack of inhibitor specificity or the likelihood that many kinases are involved?
2. The authors do observe that there is an increase in expression of several tight junction-associated proteins, including the claudins, in deltaMUC13 cells. Affected CLDNs include 1, 2, 3, 4, 7, 12. (1) While it appears the authors are arguing that this increased claudin expression results in increased barrier function, they do not sufficiently highlight the well-known role that CLDN2 has in cation transport, and both CLDN-4 and -7 have also been implicated in paracellular ion flux (although this is apparently cell-type specific). These observations would seem to argue against a simple correlation between claudin expression and tight junction barrier function. (2) The increased expression of claudins in the nominally tail-minus MUC13 without a corresponding change in TER would again seem to argue against a simple correlation; (3) Furthermore, the authors should note the disconnect between paracellular ion flux mediated by claudins and the flux of markers such as dextrans and lucifer yellow, which can be dissociated from claudin function.
3. The authors may be overly reliant on TER measurements. Epithelial cells have two parallel resistive pathways: transcellular and paracellular. TER measure the contribution of both. Thus, an increase in TER could result from a decrease in transcellular ion transport. The authors need to measure transcellular ion flow or selectively measure the junctional resistance in a select set of experiments to rule this possibility out.

Minor points:

1. Watch the use of decimal points instead of commas (lines 253 and 256).
2. Consider quantifying all blots (Fig. 5C, Fig. 6B).
3. Line 543: MilliQ is not a washing agent (or is it?).
4. Line 553: TER is the product of total resistance times the area. The units are ohms times area.
5. Line 630: Please provide the transfer conditions (voltage, amp, watts?) and transfer buffer when describing the Western blot protocol.
6. Consider using dot-plots for all quantified data.

Significance

This study advances the fields of mucin biology and tight junction barrier function in an incremental manner. The study is well done, but there are few mechanistic insights into how MUC13 modulates paracellular flux in cultured gut epithelial cells.

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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements

We thank all the reviewers for their positive and constructive comments on the manuscript. Changes to the main text have been marked in red text in the uploaded file. We address each of the reviewers’ comments point-by-point below. The major revisions include:

Improved statistical details and attention to subjective language throughout. New TEM data included in the new Figure 1—figure supplement 1 to illustrate the drastic ultrastructural differences between MCs and neighboring epidermal cells. Inclusion of an estimate of the “recombination efficiency” of our keratinocyte lineage trace in Figure 4. Additional quantification of MC density in the different body regions (Figure 6) and prior to squamation (Figure 7F). Imaging of the zebrafish oral mucosa (new Response Figure 1). More nuanced interpretations of the eda and fgf8a mutant phenotypes.

2. Point-by-point description of the revisions

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

The authors describe and characterize the touch system in zebrafish as a new model to study MC development and maintenance. The manuscript is very well written, and the experiments are carefully executed and beautifully illustrated. This study addresses the origin of zebrafish MCs, shows that they are innervated by somatosensory neurons and that they share molecular properties with mammalian MCs. In addition, the authors developed transgenic lines that allow them to study MCs in vivo.

Genetic lineage tracing shows that zf MCs are derived from the epidermis as in mouse and not from neural crest cells, as described for avian MCs. In addition, longevity and turnover of murine MCs was controversial. Here, the authors show that zf MCs constantly turnover and that the distribution and turnover rate in the trunk depends on underlying scales. They show that the loss of scales in eda mutants leads to a decrease in MC production and an increase in MC death showing that scales are required for MC production and maintenance. Using a specific fgf8 mutant allele that causes an increase in Fgf signaling and an increase in scale size they demonstrate that scales are sufficient to induce MCs.

In summary, this manuscript is a rigorous and beautiful characterization of MCs development and maintenance. The authors demonstrate that zebrafish MCs share many characteristics with mammalian MCs. The generation of MCs specific transgenic lines, coupled with existing transgenic lines that label somatosensory cells and cells in the scales sets the stage for detailed further analyses. For example, using these tools one can now study how the size of the MC progenitor domain is controlled, if progenitors migrate and what the identities of the molecular signals from the nerves and scales to progenitors and differentiated MCs are.

Minor comments:

Line 71: Why is the heterogeneity a limitation? Couldn't it also exist in zebrafish?

Thank you for raising this question. The limitations are meant to refer to current limitations of the rodent system and demonstrate an opportunity for a new model system to complement the rodent system. We have rephrased this section to better articulate this point.

Introduction:

“While this system has been useful for understanding many aspects of MC development and function, the rodent system also has several significant limitations that warrant additional models to improve the understanding of MCs.”

We also added the following to the discussion: *“As the majority of the analyses completed here focus on MCs found in the trunk epidermis, it will be intriguing to determine whether all MCs in different skin compartments in the juvenile and adult zebrafish share similar molecular, cellular, and functional properties.” *

Line 295: The authors write: 'Thus, our observations indicate that the decrease in MC cell density in eda mutants is likely due to both reduced MC production and increased MC turnover'.

It should say: '.. increased MC loss'. In mutants the MCs show poor turnover. I believe the term 'turnover' implies that the cells are being replaced, which is only partially happening here.

Thank you for the clarification. We agree with the reviewer and have changed the wording from “turnover” to “loss” in lines 295 and 301.

Line 301: 'The authors state: 'these data suggest that Eda signaling is required for MC development, maintenance, and distribution along the trunk.'

The authors do not show any data that Eda signaling is involved in MC development but only that scales are needed. The MC inducing signals from scales to the epidermis could be independent from Eda signaling. Please rephrase.

Please discuss that not all MC specification/development depends on scales. Even in the scale-less eda mutants some MCs form (as in the inter scale regions in wt?) and even turnover. Do scales secrete a signal that increases proliferation of existing MC progenitors but scales do not affect specification?

We respectfully disagree with the reviewer on the interpretation of these results. Our experimental manipulation (examination of eda-/- vs. sibling controls) only allows us to conclude that Eda signaling - either directly or indirectly - is required for these processes along the trunk. To determine whether signaling from scales is required would require identification of the signal(s) and/or loss/ablation of scales independent of Eda. We have rephrased the results to more clearly state our interpretation. The corresponding portion of the discussion now reads “Further investigations are required to determine whether Eda signaling directly regulates the differentiation of MC progenitors. Alternatively, since eda mutants lack scales (Harris et al., 2008) and have decreased epidermal innervation (Rasmussen et al., 2018), MC development may require scale-derived and/or somatosensory neuron-derived signals.”

Line 320: The authors describe that the fgf8 allele leads to a redistribution of MCs. Is it really a redistribution, or is it ectopic induction or expansion of existing progenitors? Redistribution implies that the expansion is due to a loss of MCs in another region, which I do not see in the data.

Thank you for raising this point about the potentially poor wording choice relating to “redistribution”. We do not yet know whether the distribution of MCs in fgf8a mutants reflects a redistribution, ectopic induction, or expansion of existing progenitors (these are excellent ideas for future studies). Thus, in response to the reviewers comment, we have changed the heading for this results section to “The MC pattern is not predetermined along the trunk” and concluded the section as follows: “... the distribution of MCs tracked with the altered scale size and shape in the mutants, suggesting the MC pattern is not predetermined within the trunk skin compartment (Figure 9E-H).”

Figures:

• Figure 1, panels B-C': EM images are very dark and difficult to see. Letter 'a' is on top of the axon, maybe move to the side and pseudo-color different structures.

In response to these suggestions, we have adjusted the brightness and contrast to lighten the TEM images in Figure 1B-C’ as much as possible. We also moved the ‘a’ off to the side in Figure 1B’ to make the axon more visible. In response to Reviewer #3’s comments (see below), we also added an additional TEM image in the new Figure 1—figure supplement 1 that has presumptive keratinocytes and a MC differentially pseudo-colored. We hesitate to pseudo-color the cells/structures in Figure 1B-C’ for fear of obscuring the underlying TEM images.

• Figure 1, panel D: very difficult to see the magenta axons in the cartoon. Please enlarge and make brighter.

We agree that this needed improvement. In the revised Figure 1D, we made the axons clearer and illustrated the different types of MC-axon associations we observe in Figure 2. We also refer the reader back to this figure in the corresponding axon innervation results section.

• Figure 2, panels A and D: keeping the same antibody stainings in the same color would help with visualization. Matching the bar plots in panel C would be even nicer.

Thank you for the suggestion. The revised Figure 2 now has a consistent color scheme.

• Figure 2, panel C: please identify in the legend if the error bars are SD, SEM or other.

These error bars represent 95% confidence intervals. This information has been added to the figure legend.

• Figure 2, panels G and H: MCs are in cyan in the image, but green in the legend.

This has been corrected.

• Figure 3: include percentages and total number in the image instead of the legend.

The numbers and percentages have now been added to the Figure 3 panels. We have left them in the figure legend for clarity on what was scored.

• Figure 6, panel B: which part of the eye is being depicted?

Thank you for the question. We imaged the corneal epithelium above the lens. This has been clarified in the appropriate parts of Figures 6 and 8 and the corresponding figure legends.

• Figure 6, panel F: please provide error bars and statistics to show that the operculum has a higher density of MCs.

Thank you for the suggestion. In response to the comment, we revised Figure 6F by: 1) increasing the sample size; 2) replotting the data as boxplots rather than bar graphs; and 3) including the results of a one-way ANOVA.

• Figure 7, panels F-H: for simple linear regression, please also provide F and p values.

Thank you for the suggestion. This information has been added to the figure legend.

• Figure 8, panel D: colors for SL do not follow a scale, very hard to understand which is which.

In response to the reviewer’s suggestion, we tried numerous different color palettes. However, we were unable to find a color palette that allowed us to distinguish individual points as well as the rainbow palette used in Figure 8D. Thus, after careful consideration, we have elected to keep the original palette here. For consistency, we have used the same palette in the revised Figure 8–figure supplement 1D and Figure 9–figure supplement 1D.

Methods:

• Line 472: the word "sex" should be used instead of "gender".

Thank you for the correction. This is fixed in the revision.

• Image analysis, line 593. Please provide a more detailed explanation or describe the ImageJ macro used for the analysis.

Our ImageJ macro has been fully annotated and is provided as Figure 2—source code 1 in the revision. The corresponding methods section has also been updated to clarify the methodology.

Reviewer #1 (Significance (Required)):

Soft touch is perceived by Merkel cells (MCs). How MCs develop and are maintained is not well understood because MC development is difficult to study in mammals due to their in utero development. The authors describe and characterize the touch system in zebrafish as a new model to study MC development and maintenance. The study demonstrates that the zebrafish touch system shares many characteristics with its mammalian counterpart, namely its developmental origin, innervation and molecular characteristics. In contrast to mammals, zebrafish transgenic lines that the authors generated, allow the in vivo analysis of Merkel Cell specification, development and maintenance. Therefore this study is the foundation for future detailed cellular and molecular analyses of the touch sensory system and will be of interest to developmental biologists studying stem cells, regeneration and aging, as well as neuroscientists.

We thank the reviewer for their positive assessment of the manuscript.

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

This is a very nice and straightforward paper characterizing mechanosensory Merkel cells in the zebrafish skin. The paper uses a number of criteria, based on our knowledge of Merkel cells in mammals, to identify a population atoh1a expressing cells, with neurosecretory granules and actin rich microvilli as Merkel cells in the zebrafish skin. The authors have used existing transgenic lines and and developed some of their own, described in this paper, to follow the development of Merkel cell in zebrafish. They show that Merkel cells are derived from basal keratinocytes not neural crest cells. They have region specific densities that influenced by underlying structures like scales and fin rays. They go to show that Ectodysplasin signaling promotes Merkel cell development in the trunk skin but not above the eye or operculum. Reduction of Merkel cells in eda mutants suggest that Eda signaling is required for their development and maintenance. Finally they show that alteration of zebfrafish scale pattern using a mutant with exaggerated fgf8a expression also alters merkel cell distribution.

The data presented is clear and the conclusions are supported by their observations.

I have no significant issue with the paper as is.

Reviewer #2 (Significance (Required)):

This study will serve as an excellent basis for future work looking at studies of Merkel cell development and function in fish. Though Merkel cells have been studied in mammals, establishing a zebrafish model for their study will help overcome many barriers that make their analysis difficult in mammals.

We thank the reviewer for their positive assessment of the manuscript.

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

In this manuscript, Brown et al. (2022) seek to characterize and address fundamental questions regarding the development and dynamics of Merkel cells (MCs) in zebrafish (Danio rerio). The authors utilize a diverse and complementary suite of methods to characterize presumptive MCs in the epidermis of adult zebrafish, including electron microscopy, novel transgenic lines, confocal imaging, and various immuno- and non-immunohistological staining techniques. These studies demonstrate that zebrafish MCs share many features with vertebrate (including mammalian) MCs, particularly regarding morphology/structure, putative functions, genetic markers, and bodily distribution.

After establishing the identity of zebrafish MCs, the authors employ lineage tracing and cell tracking analyses to determine that trunk MCs derive from basal keratinocytes and exhibit regular cell turnover. Finally, the authors examine how trunk scales may affect MC development by using established scale mutants. These results show that the presence/absence of scales influences trunk MC development, while scale characteristics (e.g. shape, size) change the distribution of MCs.

MAJOR COMMENTS:

The key conclusions of the manuscript are convincing, however, several points should be addressed by the authors.

Throughout the manuscript, the authors make general claims about zebrafish MCs (zMCs) based on the evidence collected. Yet, most of this evidence (particularly claims about MC turnover, development, structure) comes from examination and experimentation of a specific MC population: trunk MCs located in the scale epidermis. The authors remark upon mammalian MC diversity (lines 73-74) and go on to highlight the diversity of MCs throughout the adult zebrafish (Figure 6), which have differing densities and distribution patterns. Any statements that suggest all zebrafish MCs share certain qualities/features should be carefully considered given the evidence presented.

Thank you for raising this important point. We have added wording in the results and discussion to clearly articulate that the majority of our analyses and conclusions are based on trunk MCs:

Results:

“Anticipating the conclusion of our analysis below, we shall hereafter refer to the epidermal atoh1a+ cells as MCs, with the majority of the analyses completed on trunk MCs unless stated otherwise.”

Discussion:

“As the majority of the analyses completed here focus on MCs found in the trunk epidermis, it will be intriguing to determine whether all MCs in different skin compartments in the juvenile and adult zebrafish share similar molecular, cellular, and functional properties.”

In the manuscript, the authors validate several markers for the identification of zMCs based upon known mammalian markers (e.g. atoh1a, sox2, piezo2, SV2, 5-HT, and AM1-43; Figures 1-3). Yet, another well-known marker for MCs (CK8) is not addressed (Moll, 1995; Moll, 2005). One zebrafish ortholog for CK8 is krt4, a transgene successfully employed in this study to label keratinocytes. Do zMCs express krt4 or other mammalian MC keratins? Answering this question or addressing this discrepancy would further strengthen the authors claims that these cells are bona fide zMCs.

We agree with the reviewer that 1) identification of a keratin(s) that distinguishes MCs from other epidermal cell types in zebrafish would be an excellent reagent; and 2) readers familiar with the mammalian MC literature may similarly wonder why this was not addressed in the manuscript. Indeed, we had considered whether we could identify homologs of CK8, CK20 or other mammalian MC keratins that would label zebrafish MCs. However, despite the confusing nomenclature that would indicate otherwise, the zebrafish keratins share more homology with each other than the corresponding mammalian proteins (Ho et al., 2022; PMID: 34991727). Our revised results section now includes the following to clarify this point: *“For example, keratins, most notably keratin 8 and keratin 20, have been used extensively as markers of mammalian MCs (Moll et al., 1995, 1984). However, zebrafish keratins have undergone extensive gene loss and duplication and are not orthologous to mammalian keratin genes (Ho et al., 2022). Thus, we considered alternative molecular markers to label zebrafish MCs.” *

The authors utilize a previously validated eda mutant line to see if ectodysplasin signaling affects zMC development. While the results of these experiments are convincing, the authors need to make clear whether they are claiming that scales, scale-derived Eda signaling, or Eda signaling alone dictate trunk MC development. It appears that there is some conflation of these ideas, particularly with line 306 ("blocking dermal appendage formation inhibited MC development" is a different claim from 'blocking Eda signaling inhibited MC development'). One way to make this differentiation would be to perform a similar experiment as detailed in Xiao et al. 2016: using a Shh agonist in eda mutants. If scale-specific signals are required in addition to Eda, we would expect to see similar MC densities and patterns in both Shh agonist-treated and non-treated eda mutants.

We agree that our interpretation of these results could have been more clearly articulated in our initial submission. As discussed above in response to Reviewer #1, we do not yet know whether Eda signaling directly or indirectly influences MC development. We have revised the results section to clarify our interpretation of the results as follows: “Together, these data suggest that either Eda signaling, or a scale-derived signal, is required for MC development, maintenance, and distribution along the trunk. Further studies are required to determine the specific scale-derived signal that regulates MC development in the trunk."

The suggestion of using a Shh pathway agonist in eda mutants to attempt to rescue MC differentiation similar to Xiao et al. 2016 is an interesting one. To our knowledge, experiments validating the Smo agonist used by Xiao and colleagues (Hh-Ag1.5) in zebrafish have not been published. We also note that activation of Shh signaling by heat-shock induction of shha expression during squamation led to kyphosis and epidermal migration off of the trunk (Aman et al., 2018; PMID: 30014845). Thus, we respectfully suggest that distinguishing between the various possibilities downstream of Eda is beyond the scope of the current manuscript. We have added a discussion point along these lines: “Further investigations are required to determine whether Eda signaling directly regulates the differentiation of MC progenitors. Alternatively, since eda mutants lack scales (Harris et al., 2008) and have decreased epidermal innervation (Rasmussen et al., 2018), MC development may require scale- and/or somatosensory neuron-derived signals. Finally, we note that trunk MCs are not completely absent in eda mutants, suggesting that a subset of MCs develop independent of Eda signaling.”

Throughout the manuscript, the authors use subjective language (e.g. line 106). While this reviewer does not wish to suppress or alter the authors' voices, careful consideration should be used when employing these types of descriptors. Furthermore, the authors use suggestively quantitative language inappropriately or unjustifiably. For example, in line 221, the authors use "extensive" when describing the co-labeling between atoh1a+ MCs and lineage-traced basal keratinocytes; the percentage of co-labeled cells ranged from 29-32%. Other quantitative descriptors such as "frequently" (line 171) or "uniform" (line 249) describe various features or phenomena without quantification in figures or supplements.

Thank you for this comment. We have paid careful attention to our subjective/statistical language in the revision. Regarding the usage of “uniform” - we have added the wording “relatively uniformly” to descriptions and a statement that our term “uniform” was not specifically quantified. Although the uniform appearance was not specifically quantified, we believe this provides an accurate description of the MC localization pattern in certain skin compartments.

Example word change in Results:

“For example, MCs were distributed relatively uniformly across the eye, although this spatial pattern was not specifically quantified”

In the lineage tracing experiments (Figure 4), the authors note that "recombination is not complete" (lines 1016-1017) to explain why not all zMCs express the basal keratinocyte lineage marker. While this idea could be supported by Figure 4-figure supplement 1, one could postulate that zMCs are derived from multiple progenitor lineages. Using the basal keratinocyte lineage tracing validation, the authors could in theory calculate a "recombination efficiency" of this transgenic line and determine approximately the percent of zMCs they 'lose' as a result. Otherwise, the authors could perform other experiments to support the claim that zMCs derive from basal keratinocytes. For example, could the authors photoconvert basal keratinocytes at 1 dpf and see how many derived MCs are still photoconverted later? Could they do this photoconversion experiment with neural crest cells? Could they ablate neural crest cells and determine if MC number is affected? These additional experiments are not necessarily required for publication, but some explanation of the unexpectedly low percentage of basal keratinocyte lineage marker-labeled MCs would suffice.

We thank the reviewer for raising this important point and the suggestion of calculating a “recombination efficiency”. We note that Cre responsive transgenes are far from a perfect technology in zebrafish as recently characterized by Lalonde et al. (2022; PMID:35582941). In response to the reviewer’s comment, we added an estimate of the recombination efficiency to Figure 4 (panels E, G, H). Importantly, a comparison between the recombination efficiency and percentage of MCs labeled by the basal keratinocyte Cre tracing was not significantly different. Our revised results section reads as follows: “After raising 4-OHT-treated animals to adulthood, we observed variable (2-81%) co-labeling between the basal keratinocyte lineage trace and a MC reporter (Figure 4D’,F). We note that our lineage tracing strategy did not label all basal keratinocytes (Figure 4D; Figure 4—figure supplement 1), suggestive of incomplete Cre-ERT2 induction and/or transgene recombination. Consistent with the latter possibility, a recent analysis demonstrated Tg(actb2:LOXP-BFP-LOXP-DsRed) has a low recombination efficiency compared to other Cre reporter transgenes (Lalonde et al., 2022). To estimate the local recombination efficiency in imaged regions, we thresholded the DsRed channel and calculated the fraction of skin cells labeled (Figure 4E). Importantly, the proportion of MCs labeled by the basal keratinocyte lineage trace was not significantly different from the local recombination efficiency (Figure 4G-H). These observations support a basal keratinocyte origin of most or all zebrafish MCs.”

The authors use appropriate statistics and have sufficient replicates when this information is presented. Yet, the presence or absence of these data is not consistent within figure captions. The authors must ensure that they provide the N of adults and scales (when appropriate), the SL range of adults, and transgenic lines used. Statistics are missing in some figures (for example: Figures 4E, 5D, 5E, 6F, 8S-1E, 9E-H) where it would be appropriate to include them. In some figures, the N changes over time (example: 5D, 5E); an explanation in the 'Methods' section would suffice.

Thank you for noting the need for additional statistics. We have added statistics to the above figures. For Figure 9E-H, we have not added additional statistics. Figure 9E-H serve to graphically visualize differences. We show statistical differences in Figure 9—figure supplement 1 for scale area, aspect ratio, and Feret’s diameter. We have added an explanation related to Figure 5D,E in the methods section: “Animals that died over the course of the experiment were excluded from further analysis.”

MINOR COMMNETS

While the authors present an extensive argument for their claims, addressing these additional comments would further strengthen their story.

Are zMC nuclei lobulated? This ultrastructure characteristic seems to be common in MCs (Chew & Leung, 1994; Tachibana & Nawa, 2002; Moll, 2005; Boulais, 2009).

We have not observed any lobulation of the MC nuclei by TEM, nor was this commented on in the TEM studies of Whitear and colleagues in other teleosts (Lane and Whitear, 1977; PMID: 198137; Whitear, 1989; PMID: 2510796). Nevertheless, we cannot rule out the possibility that serial sectioning or other high resolution analysis of the nuclear shape may reveal such features. In response to the reviewer’s comment, we have added the following paragraph to the discussion: “While our characterization revealed substantial similarities between mammalian and zebrafish MCs, we did observe anatomical differences in line with previous ultrastructural characterizations of teleost MCs (Lane and Whitear, 1977; Whitear, 1989). For example, the nuclei of mammalian MC are commonly lobulated (Boulais et al., 2009; Cheng Chew and Leung, 1994; Moll et al., 2005; Tachibana and Nawa, 2002). While we did not observe lobulation of zebrafish MC nuclei by TEM, we cannot rule out that serial sectioning or high-resolution reconstruction of nuclear shape would reveal lobulation. Mammalian MCs typically localize adjacent to basal keratinocytes (Boot et al., 1992; Cheng Chew and Leung, 1994; Fradette et al., 1995; Mihara et al., 1979; Moll et al., 1996; Smith, Jr, 1977), whereas zebrafish MCs appear in upper strata, typically beneath the periderm (Figure 1D,G’’). As the majority of the analyses completed here focus on MCs found in the trunk epidermis, it will be intriguing to determine whether all MCs in different skin compartments in the juvenile and adult zebrafish share similar properties.”

In Figure 3C and 3", the authors show that AM1-43 labels zMCs. Yet, this technique should also stain sensory axons that associate with MCs (Meyers, 2003). Are axons also stained? Other positive controls for the stains could be useful as a supplement.

The reviewer is correct that Meyers et al., (2003; PMID: 12764092) report AM1-43 staining of neurites that innervate MCs in the whisker follicle. However, they did not report similar staining of neurites innervating touch dome MCs. In murine hairy skin, the related styryl dye FM1-43 appears to most prominently stain MCs and hair follicle-associated lanceolate endings (Banks et al., 2013 PMID: 23440964; Villarino et al., 2022 preprint DOI: 10.1101/2022.05.26.493600). Our revised legend for Figure 3 now includes the following: “AM1-43 has been reported to stain neurites innervating MCs in murine whisker vibrissae (Meyers et al., 2003). However, our AM1-43 staining regiment did not strongly label cutaneous axons, although we cannot exclude low levels of staining.”

All of the stains used in our original Figure 3 have been previously validated in zebrafish, which we have more clearly stated and cited in the corresponding results section of the revision. Because these reagents have all been previously validated and our staining patterns are consistent with the literature, we respectfully suggest that positive controls would add little value to the current manuscript. Nevertheless, in response to the reviewer’s comment, we confirmed our piezo2 FISH staining using an independent method (a piezo2 HCR probe). We have included these HCR results as the updated Figure 3D and moved the original Figure 3D to Figure 3—figure supplement 1.

In Figure 7, the authors argue that as scales develop, MC density increases with scale area. Did the authors compare MC densities of differently-sized scales at the same age? Is fish SL/age a potential confound in the interpretation of these data?

Thank you for the suggestion. In response to the reviewer’s comment, we have replotted the data in Figure 7G,H for animals in the range 8-10 mm SL in Figure 7—figure supplement 1. We have revised the corresponding results section as follows: “The density and number of MCs positively correlated with scale area (Figure 7G,H), although this trend was less pronounced at stages less than 10 mm (Figure 7—figure supplement 1)”. As discussed above in response to reviewer #1’s suggestion, we also now report F-statistics and P-values for the linear regressions in the figure legends.

The authors claim that squamation begins at ~9 mm SL (line 268), prior to which MCs were "rare" in the epidermis (supported by data in Figure 7F). However, Figures 8A and 8G suggest that MCs are not rare prior to squamation/9 mm SL. Are these data in conflict?

Thank you for raising this observation. We do not believe these data are in conflict. Figure 8A and B show images of fish 8.8-8.9 mm SL, immediately prior to squamation. MCs appear about the same time as scales develop but the exact timing varies between animals. To further strengthen this section of the manuscript, we now include quantification of the density of trunk MCs at various stages prior to 9 mm SL (new data added to the developmental timeline in Figure 7F). These data are consistent with our initial interpretation. In the revised results section we clarify this as follows: “Using reporters that label MCs and scale-forming osteoblasts, we rarely observed MCs in the epidermis prior to 8 mm SL (Figure 7B, F). Between 8-10 mm SL, MCs appeared at a low density along the trunk (Figure 7F). MC density rapidly increased from 10-15 mm SL, a period of active scale growth (Figure 7C-F).”

In Figure 6B-E, the panels are incorrectly labeled as "atoh1a:nls-Eos" (figure caption and fluorescence localization show they are atoh1a:Lifeact-EGFP).

The low magnification panels were correctly labeled as atoh1a:nls-Eos. The insets showed atoh1a:Lifeact-EGFP as described in the figure legend. We apologize for the confusion and poor data presentation. We have revised Figure 6 to eliminate the problematic labeling/display.

Figure 9 panels E-H are not referenced in the main body of the text.

Thank you for pointing this out. Fixed in the revised manuscript.

In Figure 6, the authors examine MC densities in the tail, but do not quantify changes here with eda mutants as they did for other regions (eye, operculum) in Figure 8. Why was this region not examined?

We have clarified this point in the revised results section as follows: “eda mutants lack fins at the stages analyzed (Harris et al., 2008) precluding analysis of these regions in the homozygous mutants.”

The authors do a good job in detailing the current literature regarding MCs, however, two missing areas are noticeable: 1) there is no mention of mammalian MCs that reside in the oral mucosa (Hashimoto, 1972) or whether they exist in zebrafish, and 2) no mention of Merkel-like cells (Halata, 2003) and why the cells in this paper are or are likely not Merkel-like cells.

Thank you for the suggestions. Regarding the first point, we revised the introduction to reference (Hashimoto, 1972) as follows: “...vertebrates have diverse types of skin and MCs are found in both hairy and glabrous (non-hairy) skin, as well as mucocutaneous regions such as the gingiva and palate (Hashimoto, 1972; Lacour et al., 1991; Moayedi et al., 2021).” We also imaged the mucosal tissue along the roof palate of the adult mouth and identified atoh1a+ cells (see Response Figure 1 below). Close examination of the atoh1a:Lifeact-EGFP signal revealed these cells have a spherical morphology and extend short processes similar to the MCs described across the body regions examined in Figure 6. However, as the microvillar morphology of the palatal atoh1a+ cells is not identical to those identified in other skin regions, we hesitate to call these MCs without performing additional in-depth analyses. We feel that inclusion of these data in the manuscript could distract the reader from the main focus of our study, therefore we have included them here:

__Response Figure 1. atoh1a+ cells in the adult oral epithelium. (A,B) __Low- (A) and high-magnification (B) confocal micrographs of oral roof palate epithelium in an adult expressing reporters for keratinocytes (Tg(krt4:DsRed)) and atoh1a-expressing cells (Tg(atoh1a:Lifeact-EGFP)). (B’) Reconstructed cross section along the yellow line in B showing two atoh1a+ cells in the upper strata of the oral epithelium. Scale bars: 50 µm (A) and 10 µm (B,B’).

Regarding the second point, we have added the following sentence to the first paragraph of the discussion: “Second, zebrafish MCs extend numerous short, actin-rich microvilli and complex with somatosensory axons, classic morphological hallmarks of MCs (Mihara et al., 1979; Smith, Jr, 1977; Toyoshima et al., 1998). Our morphological observations support the interpretation that these cells are MCs rather than Merkel-like cells, which lack axon association and microvillar processes (reviewed by Halata et al., 2003).”

It may help readers understand MC morphology in context if the authors include a larger picture of the TEM data that highlights the drastic difference in ultrastructure between MCs and neighboring keratinocytes.

Thank you for the suggestion. We added a new figure (Figure 1—figure supplement 1) to the revised manuscript that contains an additional TEM image that we believe illustrates the different morphologies of keratinocytes and MCs. We hope this will help the reader contextualize the morphology and position of MCs within the zebrafish epidermis. This is now referenced in the first results section as follows: “The cells appeared relatively small and spherical with a low cytoplasmic-to-nuclear ratio compared to neighboring keratinocytes (Figure 1B,C; Figure 1—figure supplement 1) …”

Reviewer #3 (Significance (Required)):

The current manuscript provides significant advancements in various biological fields and research communities. For researchers that utilize zebrafish as a model organism, these findings present a new cell type along with novel and essential genetic tools for study. These developments open the possibilities to further understand MCs, their roles in somatosensory function, mechanisms of cell type diversification, and to engage in translational research. For those already researching MCs, this manuscript shows that fundamental questions regarding MC functioning can be rigorously addressed with a new model that can fill the methodological limitations imposed by mammalian biology. Indeed, the authors do a thorough job of introducing and contextualizing our knowledge of MCs and any outstanding gaps. The authors then sit their findings comfortably alongside previous works, largely supporting those findings, and take the extra step to address MC controversies/matters of debate. This technique of supporting the current literature and then uplifting it with new findings makes this work even more impressive. Various audiences will find value in this manuscript, including but not limited to those that study epidermal cell types, the development and influence of skin appendages, somatosensation and sensory disorders, developmental biology, and Merkel cell carcinoma.

We thank the reviewer for their positive assessment of the manuscript.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

In this manuscript, Brown et al. (2022) seek to characterize and address fundamental questions regarding the development and dynamics of Merkel cells (MCs) in zebrafish (Danio rerio). The authors utilize a diverse and complementary suite of methods to characterize presumptive MCs in the epidermis of adult zebrafish, including electron microscopy, novel transgenic lines, confocal imaging, and various immuno- and non-immunohistological staining techniques. These studies demonstrate that zebrafish MCs share many features with vertebrate (including mammalian) MCs, particularly regarding morphology/structure, putative functions, genetic markers, and bodily distribution.

After establishing the identity of zebrafish MCs, the authors employ lineage tracing and cell tracking analyses to determine that trunk MCs derive from basal keratinocytes and exhibit regular cell turnover. Finally, the authors examine how trunk scales may affect MC development by using established scale mutants. These results show that the presence/absence of scales influences trunk MC development, while scale characteristics (e.g. shape, size) change the distribution of MCs.

MAJOR COMMENTS:

The key conclusions of the manuscript are convincing, however, several points should be addressed by the authors.

• Throughout the manuscript, the authors make general claims about zebrafish MCs (zMCs) based on the evidence collected. Yet, most of this evidence (particularly claims about MC turnover, development, structure) comes from examination and experimentation of a specific MC population: trunk MCs located in the scale epidermis. The authors remark upon mammalian MC diversity (lines 73-74) and go on to highlight the diversity of MCs throughout the adult zebrafish (Figure 6), which have differing densities and distribution patterns. Any statements that suggest all zebrafish MCs share certain qualities/features should be carefully considered given the evidence presented.

• In the manuscript, the authors validate several markers for the identification of zMCs based upon known mammalian markers (e.g. atoh1a, sox2, piezo2, SV2, 5-HT, and AM1-43; Figures 1-3). Yet, another well-known marker for MCs (CK8) is not addressed (Moll, 1995; Moll, 2005). One zebrafish ortholog for CK8 is krt4, a transgene successfully employed in this study to label keratinocytes. Do zMCs express krt4 or other mammalian MC keratins? Answering this question or addressing this discrepancy would further strengthen the authors claims that these cells are bona fide zMCs.

• The authors utilize a previously validated eda mutant line to see if ectodysplasin signaling affects zMC development. While the results of these experiments are convincing, the authors need to make clear whether they are claiming that scales, scale-derived Eda signaling, or Eda signaling alone dictate trunk MC development. It appears that there is some conflation of these ideas, particularly with line 306 ("blocking dermal appendage formation inhibited MC development" is a different claim from 'blocking Eda signaling inhibited MC development'). One way to make this differentiation would be to perform a similar experiment as detailed in Xiao et al. 2016: using a Shh agonist in eda mutants. If scale-specific signals are required in addition to Eda, we would expect to see similar MC densities and patterns in both Shh agonist-treated and non-treated eda mutants.

• Throughout the manuscript, the authors use subjective language (e.g. line 106). While this reviewer does not wish to suppress or alter the authors' voices, careful consideration should be used when employing these types of descriptors. Furthermore, the authors use suggestively quantitative language inappropriately or unjustifiably. For example, in line 221, the authors use "extensive" when describing the co-labeling between atoh1a+ MCs and lineage-traced basal keratinocytes; the percentage of co-labeled cells ranged from 29-32%. Other quantitative descriptors such as "frequently" (line 171) or "uniform" (line 249) describe various features or phenomena without quantification in figures or supplements.

• In the lineage tracing experiments (Figure 4), the authors note that "recombination is not complete" (lines 1016-1017) to explain why not all zMCs express the basal keratinocyte lineage marker. While this idea could be supported by Figure 4-figure supplement 1, one could postulate that zMCs are derived from multiple progenitor lineages. Using the basal keratinocyte lineage tracing validation, the authors could in theory calculate a "recombination efficiency" of this transgenic line and determine approximately the percent of zMCs they 'lose' as a result. Otherwise, the authors could perform other experiments to support the claim that zMCs derive from basal keratinocytes. For example, could the authors photoconvert basal keratinocytes at 1 dpf and see how many derived MCs are still photoconverted later? Could they do this photoconversion experiment with neural crest cells? Could they ablate neural crest cells and determine if MC number is affected? These additional experiments are not necessarily required for publication, but some explanation of the unexpectedly low percentage of basal keratinocyte lineage marker-labeled MCs would suffice.

• The authors use appropriate statistics and have sufficient replicates when this information is presented. Yet, the presence or absence of these data is not consistent within figure captions. The authors must ensure that they provide the N of adults and scales (when appropriate), the SL range of adults, and transgenic lines used. Statistics are missing in some figures (for example: Figures 4E, 5D, 5E, 6F, 8S-1E, 9E-H) where it would be appropriate to include them. In some figures, the N changes over time (example: 5D, 5E); an explanation in the 'Methods' section would suffice.

MINOR COMMENTS:

While the authors present an extensive argument for their claims, addressing these additional comments would further strengthen their story.

• Are zMC nuclei lobulated? This ultrastructure characteristic seems to be common in MCs (Chew & Leung, 1994; Tachibana & Nawa, 2002; Moll, 2005; Boulais, 2009).

• In Figure 3C and 3", the authors show that AM1-43 labels zMCs. Yet, this technique should also stain sensory axons that associate with MCs (Meyers, 2003). Are axons also stained? Other positive controls for the stains could be useful as a supplement.

• In Figure 7, the authors argue that as scales develop, MC density increases with scale area. Did the authors compare MC densities of differently-sized scales at the same age? Is fish SL/age a potential confound in the interpretation of these data?

• The authors claim that squamation begins at ~9 mm SL (line 268), prior to which MCs were "rare" in the epidermis (supported by data in Figure 7F). However, Figures 8A and 8G suggest that MCs are not rare prior to squamation/9 mm SL. Are these data in conflict?

• In Figure 6B-E, the panels are incorrectly labeled as "atoh1a:nls-Eos" (figure caption and fluorescence localization show they are atoh1a:Lifeact-EGFP).

• Figure 9 panels E-H are not referenced in the main body of the text.

• In Figure 6, the authors examine MC densities in the tail, but do not quantify changes here with eda mutants as they did for other regions (eye, operculum) in Figure 8. Why was this region not examined?

• The authors do a good job in detailing the current literature regarding MCs, however, two missing areas are noticeable:

1) there is no mention of mammalian MCs that reside in the oral mucosa (Hashimoto, 1972) or whether they exist in zebrafish, and

2) no mention of Merkel-like cells (Halata, 2003) and why the cells in this paper are or are likely not Merkel-like cells.

• It may help readers understand MC morphology in context if the authors include a larger picture of the TEM data that highlights the drastic difference in ultrastructure between MCs and neighboring keratinocytes.

Significance

The current manuscript provides significant advancements in various biological fields and research communities. For researchers that utilize zebrafish as a model organism, these findings present a new cell type along with novel and essential genetic tools for study. These developments open the possibilities to further understand MCs, their roles in somatosensory function, mechanisms of cell type diversification, and to engage in translational research. For those already researching MCs, this manuscript shows that fundamental questions regarding MC functioning can be rigorously addressed with a new model that can fill the methodological limitations imposed by mammalian biology. Indeed, the authors do a thorough job of introducing and contextualizing our knowledge of MCs and any outstanding gaps. The authors then sit their findings comfortably alongside previous works, largely supporting those findings, and take the extra step to address MC controversies/matters of debate. This technique of supporting the current literature and then uplifting it with new findings makes this work even more impressive. Various audiences will find value in this manuscript, including but not limited to those that study epidermal cell types, the development and influence of skin appendages, somatosensation and sensory disorders, developmental biology, and Merkel cell carcinoma.

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

Learn more at Review Commons

---

Referee #2

Evidence, reproducibility and clarity

This is a very nice and straightforward paper characterizing mechanosensory Merkel cells in the zebrafish skin. The paper uses a number of criteria, based on our knowledge of Merkel cells in mammals, to identify a population atoh1a expressing cells, with neurosecretory granules and actin rich microvilli as Merkel cells in the zebrafish skin. The authors have used existing transgenic lines and and developed some of their own, described in this paper, to follow the development of Merkel cell in zebrafish. They show that Merkel cells are derived from basal keratinocytes not neural crest cells. They have region specific densities that influenced by underlying structures like scales and fin rays. They go to show that Ectodysplasin signaling promotes Merkel cell development in the trunk skin but not above the eye or operculum. Reduction of Merkel cells in eda mutants suggest that Eda signaling is required for their development and maintenance. Finally they show that alteration of zebfrafish scale pattern using a mutant with exaggerated fgf8a expression also alters merkel cell distribution.

The data presented is clear and the conclusions are supported by their observations.

I have no significant issue with the paper as is.

Significance

This study will serve as an excellent basis for future work looking at studies of Merkel cell development and function in fish. Though Merkel cells have been studied in mammals, establishing a zebrafish model for their study will help overcome many barriers that make their analysis difficult in mammals.

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 describe and characterize the touch system in zebrafish as a new model to study MC development and maintenance. The manuscript is very well written, and the experiments are carefully executed and beautifully illustrated. This study addresses the origin of zebrafish MCs, shows that they are innervated by somatosensory neurons and that they share molecular properties with mammalian MCs. In addition, the authors developed transgenic lines that allow them to study MCs in vivo.

Genetic lineage tracing shows that zf MCs are derived from the epidermis as in mouse and not from neural crest cells, as described for avian MCs. In addition, longevity and turnover of murine MCs was controversial. Here, the authors show that zf MCs constantly turnover and that the distribution and turnover rate in the trunk depends on underlying scales. They show that the loss of scales in eda mutants leads to a decrease in MC production and an increase in MC death showing that scales are required for MC production and maintenance. Using a specific fgf8 mutant allele that causes an increase in Fgf signaling and an increase in scale size they demonstrate that scales are sufficient to induce MCs.

In summary, this manuscript is a rigorous and beautiful characterization of MCs development and maintenance. The authors demonstrate that zebrafish MCs share many characteristics with mammalian MCs. The generation of MCs specific transgenic lines, coupled with existing transgenic lines that label somatosensory cells and cells in the scales sets the stage for detailed further analyses. For example, using these tools one can now study how the size of the MC progenitor domain is controlled, if progenitors migrate and what the identities of the molecular signals from the nerves and scales to progenitors and differentiated MCs are.

Minor comments:

Line 71: Why is the heterogeneity a limitation? Couldn't it also exist in zebrafish? Line 295: The authors write: 'Thus, our observations indicate that the decrease in MC cell density in eda mutants is likely due to both reduced MC production and increased MC turnover'. It should say: '.. increased MC loss'. In mutants the MCs show poor turnover. I believe the term 'turnover' implies that the cells are being replaced, which is only partially happening here. Line 301: 'The authors state: 'these data suggest that Eda signaling is required for MC development, maintenance, and distribution along the trunk.' The authors do not show any data that Eda signaling is involved in MC development but only that scales are needed. The MC inducing signals from scales to the epidermis could be independent from Eda signaling. Please rephrase. Please discuss that not all MC specification/development depends on scales. Even in the scale-less eda mutants some MCs form (as in the inter scale regions in wt?) and even turnover. Do scales secrete a signal that increases proliferation of existing MC progenitors but scales do not affect specification? Line 320: The authors describe that the fgf8 allele leads to a redistribution of MCs. Is it really a redistribution, or is it ectopic induction or expansion of existing progenitors? Redistribution implies that the expansion is due to a loss of MCs in another region, which I do not see in the data.

Figures:

• Figure 1, panels B-C': EM images are very dark and difficult to see. Letter 'a' is on top of the axon, maybe move to the side and pseudo-color different structures.

• Figure 1, panel D: very difficult to see the magenta axons in the cartoon. Please enlarge and make brighter.

• Figure 2, panels A and D: keeping the same antibody stainings in the same color would help with visualization. Matching the bar plots in panel C would be even nicer.

• Figure 2, panel C: please identify in the legend if the error bars are SD, SEM or other.

• Figure 2, panels G and H: MCs are in cyan in the image, but green in the legend.

• Figure 3: include percentages and total number in the image instead of the legend.

• Figure 6, panel B: which part of the eye is being depicted?

• Figure 6, panel F: please provide error bars and statistics to show that the operculum has a higher density of MCs.

• Figure 7, panels F-H: for simple linear regression, please also provide F and p values.

• Figure 8, panel D: colors for SL do not follow a scale, very hard to understand which is which.

Methods:

• Line 472: the word "sex" should be used instead of "gender".

• Image analysis, line 593. Please provide a more detailed explanation or describe the ImageJ macro used for the analysis.

Significance

Soft touch is perceived by Merkel cells (MCs). How MCs develop and are maintained is not well understood because MC development is difficult to study in mammals due to their in utero development. The authors describe and characterize the touch system in zebrafish as a new model to study MC development and maintenance. The study demonstrates that the zebrafish touch system shares many characteristics with its mammalian counterpart, namely its developmental origin, innervation and molecular characteristics. In contrast to mammals, zebrafish transgenic lines that the authors generated, allow the in vivo analysis of Merkel Cell specification, development and maintenance. Therefore this study is the foundation for future detailed cellular and molecular analyses of the touch sensory system and will be of interest to developmental biologists studying stem cells, regeneration and aging, as well as neuroscientists.

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

Learn more at Review Commons

---

Reply to the reviewers

Official Revision Plan Document:

Manuscript number: #RC-2022-01681

Corresponding author(s): Nicholas, Leigh

1. General Statements

We sincerely appreciate these positive and helpful reviews. We are grateful for the constructive comments and we outline our responses below. Addressing these comments will further broaden the impact of the work and increase the power, reliability, and application of single cell approaches while decreasing the cost and labor intensive collection steps.

As single cell sequencing approaches have entered the mainstream, we are still finding flaws and artifacts from these methods. A major limitation of widely used collection approaches is a difficulty in obtaining biological replicates, which are required to generate robust sequencing datasets. In general, a lack of biological replicates has been a major oversight in the vast majority of single cell studies, and any technique that can facilitate biological replicate collection should be widely applied. The elegance of SNP-based demultiplexing lies in the fact that it can be applied regardless of any external label, applied to previously collected data, and the data are already collected for every sample sequenced. We were pleased to have the reviewers agree and identify the many conceptual advances in this manuscript, with one major critique being noted by one reviewer as a lack of novelty.

Regarding the lack of novelty, we appreciate that SNP-based demultiplexing was not developed as a method within this manuscript, but disagree that a broad benchmarking and validation study that opens the doors to the use of SNP-based demuxing in any species with sufficient between animal genetic heterogeneity lacks novelty. To address this concern, we will now further emphasize the drawbacks and artifacts that can arise in the currently common practice of pooling samples and choosing not to demultiplex, while improving our explanation of our discoveries in this manuscript. The lack of biological replicates in single cell sequencing studies is rampant and needs to be addressed with approaches such as those demonstrated here. We also want to emphasize the importance of validating and benchmarking bioinformatic approaches with orthogonal, priorly established approaches (eg. wet-lab based methods), which had previously not been conducted for SNP-based demultiplexing, outside of human samples. The inbred nature of common lab animals and broad range in quality and availability of genomic resources make this a major step forward in bringing SNP-demultiplexing to all labs. We believe that our paper broadly extends, benchmarks and most importantly validates the advantages and limitations of SNP-based demuxing across various species.

2. Description of the planned revisions

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

“Cardiello et al tested if souporcell (https://pubmed.ncbi.nlm.nih.gov/32366989/) can be used to demultiplex samples for some model organisms, based on identified SNPs. For this, they used synthetic multiplexed data, publicly available datasets and some new datasets, spanning samples from five model organisms. Their analysis indicates that souporcell could be used to

demultiplex scRNA-seq experiments for multiple species, which offers a cost-beneficent approach.

The manuscript reads well and shows this approach can work for different model organisms. However, unfortunately, I am confused about the amount of novelty in this manuscript. The method, souporcell, is already published. The authors indicate souporcell is not validated in non-human samples, but the original paper states that their method works with malaria parasite data (Fig 3b, FigS4). Adapting and using an available tool for different model organisms is good and groups working on different model organisms may find this manuscript useful, but the same could be said for the original article. Due to these reasons, I am not sure whether this manuscript has novelty sufficient for publication.”

__Our response: __We appreciate this constructive criticism that helped us realize that our novelty was not clearly stated in the first version of the manuscript. We need to improve our Introduction and our verbiage as to what has been previously performed and how this current manuscript provides novel insight into multiple previously unanswered questions which broadly extend the utility of SNP-based demultiplexing. To address this comment, we will revamp our Introduction, Results, and Discussion to more clearly highlight the novelty of this work.

Planned revisions:

Defining “validation”. We define validation as establishing the accuracy or validity of a method. Therefore, validation of SNP-based demultiplexing for use in non-human species requires comparison to an already proven, orthogonal method, such as a wet-lab based demultiplexing approach. The souporcell paper does not validate (i.e., confirm with an orthogonal wet-lab method) the results from souporcell in any species but humans. This lack of validation for SNP-based demultiplexing in samples from non-human species made it unclear how and if these approaches would work in other species. Human samples are expected to perform exceptionally well in this approach due to their extremely high genetic diversity and wealth of available genomic resources. Thus, while it was exciting that the original souporcell authors chose to try applying their algorithm to a non-human (e.g., malarial parasite) dataset, the paper left many unanswered questions about potential uses and accuracy. In addition to validating the accuracy of souporcell results in many species, we demonstrated that souporcell shows a relatively poor ability to call doublets in many non-human vertebrates. In addition to highlighting a novel drawback of the method, this demonstrates the need to validate the accuracy of different aspects of tools like souporcell when applied to new systems rather than use souporcell or other SNP-demuxers prior to validation. Highlighting other novel findings in this work: For instance, our assessment of which genomic resources are required for using SNP-based demultiplexing in different species, whether this could be applied to lab animals likely to be inbred to various degrees (and to address other reviewers comments, the inbred level permitted), assessment of the accuracy of SNP-demultiplexing in species with alignment references of varying qualities (i.e., only de novo transcriptome) and genomes of varying sizes (up to 30Gb, 10 times larger that of human, which can be extremely computational intensive), and the exploration of pooling and demultiplexing of multiple species in a single library. Making clear how we made the necessary adjustments to the original souporcell pipeline to successfully apply it to datasets with various resources available in these species.

(Reviewer #1): I also wrote down two minor points below:

“1- Doublets assigned by souporcell compared to the fluor-based assignment look random. In Fig 2 doublet recovery rate looks smaller, and in fig 3 doublet rate prediction looks more random. This is a bit confusing. Is there any explanation for this?”

__Our response: __We agree and thus noted in the manuscript that the detection of doublets in these datasets by Souporcell are not very reliable.

Planned revisions:

We will expand our Discussion to include brief hypotheses for factors that likely contributed to poor doublet detection by souporcell in these analyses. In the Discussion we will clearly suggest complementary approaches for improving the detection/removal of doublets in pooled scRNA-seq experiments through applying external gene expression-based doublet detection programs. We will also attempt to use these programs on at least one of our datasets to see how well independant doublet detection methods complement souporcell on pooled datasets. A full benchmarking of these doublet detection methods already exists and will be referenced in our Discussion.

Reviewer #1: “2- The authors discussed the immune system cells might show some variability in their discussion (referring to fig 3), but this is not clearly shown in the figures as data. Having a percentage bar graph could make it clearer for the readers.”

__Our response: __This is a valid point that we plan to address with the addition of a new figure as well as some clarifications in the text.

Planned revisions:

We will make a supplemental figure for Figure 3 in which we clearly demonstrate animal to animal variability. (bar plot of absolute cell numbers present from each individual animal present in each cell cluster as requested). In the new supplemental figure we will also include a new UMAP plot of fluorescently assigned cell identities belonging only to one of the three animals, which makes it easier to visualize the difference in numbers of cells from each animal present in each individual cell cluster. We will also cite papers that have already demonstrated the phenomena of animal to animal variability in scRNA-seq datasets. We will further emphasize that even in the absence of animal-to-animal variability in co-clustering, that demultiplexing pooled datasets is important because differential expression analysis is greatly enhanced with biological replicates.

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

Major comments:

“1. SNP-based demultiplexing performed well on some species, such as zebrafish and Africa green monkey, from which over 90% of the cells analyzed were correctly identified. However, this accuracy decreases in Pleurodeles samples when a common SNPs VCF file is absent (Fig.3). It showed that cell identity can be more precisely defined with the increase of average read depth (Fig 3B). So, I am wondering whether the mis-defined cells shown in Fig. 3E, actually are cells with lower reads. It is better if the authors can test such a correlation between the cell identity and the depth of reads using the data from Fig. 3E.”

__Our response: __We are thankful to reviewer #2 for raising such a great point. We do see the accuracy of the benchmarking results for this experiment increase with increasing sequence depth/cell quality. However, the reasons for this are potentially more complex than just higher accuracy of souporcell in higher quality cells: The fluorescent-based demultiplexing that is being used for “ground truth” in benchmarking souporcell for this figure is more accurate in cells with higher read depth because more fluorescent gene reads are likely to be captured. Therefore analyzing the accuracy of souporcell relative to fluorescent-based demultiplexing over varying read depths can be confusing because it is possible that both methods improve in accuracy with higher read depth. Figure 3B attempts to illustrate this concept, and to demonstrate why we chose to benchmark only the cells with sufficient read depth (read depth between 5K, and 40K, and >1 fluorescent gene read per cell). We plan to complement our manuscript with additional figures and text that will make this clearer.

Planned revisions:

We will produce a plot similar to Figure 3B, but with a Y axis that is the percent agreement between the two methods. For Figure 4 we will also make a plot showing percent agreement between demux methods versus read depth. This plot will be a useful comparison to investigate whether scRNA read depth is directly affecting the quality of souporcell’s SNP-based demux results. Plotting this comparison for a dataset in which Cellplex/Cell hashing is the benchmarking demux method is a more fair test of the effect of sequencing depth on the souporcell demux results because cellplex results rely on reads from the cellplex library, which are an independent sequencing library from the scRNA reads. We will investigate whether the use of a common VCF file or lack thereof affects souporcell accuracy. To test this, we will try repeating souporcell demux of one dataset with and without a common VCF file input to see if the VCF file inclusion affects the accuracy of souporcell results.

Reviewer #2:

“2. Please discuss limitations of this approach in the manuscript. (1) To which extent, when SNPs are roughly present in the individuals of same species, SNP-based demultiplexing can be applied, e.g., individuals from an inbred strain (c57bl6 mice) would not work.(2) The authors experimentally tested two newt species using SNP-based demultiplexing. When multiple species are experimentally applied, may the cell/nuclei size variation cause problem?”

__Our response: __We agree with Reviewer 2 that this paper brings up many technical questions about the limits to which SNP-based demultiplexing will succeed. These limitations should be addressed more thoroughly in our Discussion section.

Planned revisions:

We will expand our Discussion to more fully discuss the predicted limits for SNP-based demuxing for separating pooled cells from genetically similar individuals. We referenced the single paper previously published which reported that Freemuxlet, a similar approach to souporcell, did not succeed when applied to cells pooled from multiple animals within an inbred mouse strain, but did succeed across mouse strains (though without any validation of results). We will expand this Discussion to address the expected effects of genetic diversity on the success of SNP-based demultiplexing methods. We will also note in this expanded Discussion that SNP-based demuxing worked in this paper on siblings (some of the xenopus, some of the zebrafish), and other SNP-based demuxers have been used successfully for demuxing cells from closely related individuals including human siblings (scSplit) and human maternal/fetal pairs (souporcell). We will expand our Discussion to address the potential drawbacks of pooling cells from different species or tissue types including the possibility of a bias in scRNA-seq sample preparation methods. We expect that variations in cell or nuclei sizes between species could cause biases in cell capture depending on the scRNA-seq library preparation method, especially with microfluidic based scRNA-seq preparation methods. We will search for a dataset that would allow for synthetic pooling of inbred mouse data and, if available, put this through our synthetic pooling and demuxing pipeline. While other papers have reported this does not work with other SNP demux tools, and on comments on the souporcell github (https://github.com/wheaton5/souporcell/issues/154) it does not seem to be working, we feel this would be a nice test/reference for showing the limitations for SNP-based demuxing in highly genetically similar individuals.

(Reviewer #2)* *

“3. What is the upper limit number of samples when using this model. Please make some estimation or discussion about it.”

__Our response: __We think this is a pressing question for the future of SNP-based demuxing and deserves further discussion in this manuscript. This is directly addressed by the authors of souporcell in a github thread with regard to human samples (worked on 21 human samples, may work in up to 40). At this point, we have no reason to believe that the limit on sample numbers should be different in other species.

Planned revisions:

We will include discussion about potential limits for the maximum number of samples that can be pooled and demuxed using this approach. As discussed below in response to reviewer 3, we will quantify the genetic differences in pooled datasets in this manuscript in order to give readers an improved prediction of how well SNP-based demuxers are likely to work on their animals of interest. We will look for previously published pooled dataset from zebrafish that includes multiple dozens of samples and attempt to SNP-demultiplex this pool. While we will be unable to validate the accuracy, given how well SNP-based demuxing has performed we can at least determine if cell origins are assigned.

Reviewer #2: Minor comments:

“1. Please add an algorithm principle of this model.”

__Our response: __Thanks for the suggestion, we will do so.

Planned revision:

We will direct readers to the algorithm principle of souporcell in the original paper and include a flowchart of our workflow for running souporcell piece by piece as we have done in the manuscript. As mentioned above, we will make clear how we made the necessary adjustments to the original souporcell pipeline to successfully apply it to datasets with various resources available in these species.

Reviewer #2:

“2. Give a clear definition of doublets including the ground truth and Souporcell result.”

__Our response: __We appreciate this recommendation. For the purposes of this paper our definition of a ‘doublet’ is a dataset represented by a single cell barcode that actually contains more than one cell. However, true doublets can be identified with absolute certainty only in our synthetically pooled datasets, because no demultiplexing approach used for benchmarking is 100% accurate. Therefore, ‘true doublet’ will refer to known doublets based on synthetically pooled dataset ground truths. Further, for our experimental datasets we will also use ‘confirmed doublet’ to refer to cells that were called doublets by both the ground truth and souporcell. And we will use ‘contested doublet’ to refer to cells in which the experimentally derived ground truth and souporcell result disagree about a potential doublet.

Planned revision:

We will insert a clear definition of doublets used in this paper as described above, including the complexity in identifying which doublets are real given the relationship between ground truth and the souporcell results for each experiment.

Reviewer #2:

“3. Authors should indicate the time cost of running one round of such analysis, the minimal computational requirements?”

__Our response: __This is an important point and will be helpful to readers.

Planned revision:

We will add to the manuscript information on the required time, RAM consumption, and computational requirements for running various setups for souporcell.

Reviewer #3: Major comments:

“The manuscript makes a convincing case for the ability of a preexisting SNP-based demultiplexing tool, called souporcell, to demultiplex pooled samples. The study uses three methods for validation: 1. In silico data pooling; 2. Pooling of transgenic lines; 3. Pooling of cells tagged with CMOs (10x genomics). The results are consistent across experiments.

The authors propose that souporcell is a solution for demultiplexing pooled samples whenever sample tagging methods are not feasible. Although the authors test this approach in several species and conditions, the validation does not cover all possible cases and situations, obviously. Indeed, the authors recommend potential users to run pilot validation experiments with a secondary demultiplexing methods.

However, the manuscript would become more useful if the following points are addressed:

First, what is the genetic relatedness of the individuals pooled in the experiments? What is the SNP frequency in the samples analyzed, and how does that compare to SNP frequency in mouse strains? (The number of SNPs in the VCF is reported in a supplementary table but not discussed in the main text). This point is extremely important: as the authors mention, it is not possible to demultiplex samples from the same mouse strain. Inbreeding is relatively common in laboratory species, even unconventional ones; therefore, information on genetic relatedness and SNP rate would help readers assess whether SNP-based demultiplexing has a good chance to work in their systems. Addressing this point does not require any additional experiments, and computing from the single-cell reads how many SNPs distinguish the individuals pooled here should be straightforward.”

__Our response: __We appreciate the comments raised by reviewer #3.These are valuable critiques and will greatly improve the manuscript.

Planned revisions:

We will expand our Discussion with a paragraph on the limits for genetic differences required for SNP-based demuxing to work, as mentioned in response to Reviewer 2. This will include references to Table 1 values on SNP numbers utilized in each analysis, and hypotheses on the absolute limits for genetic relatedness. We will expand Table 1B to include green monkey. As mentioned in response to Reviewer 2, if previously published data we will also try applying souporcell to data from an inbred mouse line to test run an extreme case of applying SNP-based demuxing to data from very inbred animals. We will more clearly annotate the known relationship between individuals in our experiments, and will discuss this within our Discussion. We will contact the zebrafish and axolotl authors and ask if these animals were siblings. We will identify and apply a method for quantifying the genetic relationship between individuals in each scRNA-seq experiment in this study, to enable us to provide readers with a quantitative measure of genetic diversity present in each experiment. This analysis should shed some light on the requirements for genetic variability in order for SNP-based demultiplexers to succeed.

Reviewer #3:____

“Moreover, the relatively limited number of samples pooled does not validate the use of souporcell with a larger number of samples. For example: in developmental studies, often dozens of embryos are collected and pooled. What are the potential caveats of using souporcell for demultiplexing larger number of samples? The Discussion would be a good place to warn potential users of the limitations of the approach.”

__Our response: __We agree this could still be a limitation, and for developmental studies with multiple dozens of samples, further exploration of optimal demultiplexing methods or the combination of computational and wet-lab based demux methods may be required.

Planned revision:

We will expand our Discussion on predicted limits for SNP-based demuxing of high sample pools, as discussed in response to Reviewer 2. We agree that developmental projects often involve pooling large numbers of samples, so it is worth clearly outlining the benefits and risks of planning to use SNP-based demultiplexing on such high sample pools, and to outline the limits as discussed by the developer of souporcell. As stated above, we will work to identify a previously published pooled zebrafish dataset with multiple dozens of samples and run souporcell on it. While this will not provide any validation it will at the least determine if we are able to assign cell origins, which have thus far been very reliable when assignments have been made.

Reviewer #3: Minor comments:

“- is the accuracy of doublet detection rate a function of number of samples? This can be tested by repeating the monkey in silico experiment with three individuals.”

__Our response: __This is a good question. We do not thing that the number of samples substantially affects the accuracy of doublet detection by souporcell, but we will test this.

Planned revision:

As suggested, we will repeat the monkey analysis with 3 samples to see how this changes doublet detection. Overall, due to the low quality of doublet detection by souporcell found in this manuscript, we will expand our Discussion of doublet detection to propose some potentially useful recommendations for making conservative doublet calls with souporcell external programs (addressed above in response to Reviewer 2. We expect that the more substantial filtering of the monkey datasets relative to the zebrafish dataset prior to pooling contributed to this question. To make these differences more obvious we will more deliberately emphasize the differences in dataset filtering for each experiment.

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

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

From Reviewer 1:

“More generally, showing more direct evidence for the variability of different cell types (not just the immune system) could be informative for scRNA-seq users.”

__Our response: __We do not plan to conduct extensive analyses of other published single cell datasets to provide a further reason for why it is important to have biological replicates for single cell experiments. When building this manuscript, we chose not to pursue the option of publishing an analysis of published single cell datasets in which we could identify artifactual results and animal to animal variability, because we worried that this would be harmful to future open science efforts, and therefore, counterproductive. Further, past papers have already demonstrated the issue of batch effects and animal to animal variability in scRNA-seq datasets, and the requirement for biological replicates to facilitate differential expression analysis. As mentioned above, we will do a better job citing the papers that address these points.

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

Learn more at Review Commons

---

Referee #3

Evidence, reproducibility and clarity

Summary:

In this manuscript, Cardiello and colleagues address the problem of demultiplexing pooled samples in single-cell RNA sequencing (scRNAseq) experiments. The manuscript benchmarks the use of a preexisting SNP-based demultiplexing tool, called souporcell, in pooled samples from non-conventional laboratory species. The validation includes computational pooling of published data from different individuals (zebrafish, green monkey), and generation of new pooled data with independent ground-truth information available (with frogs and three salamander species). The authors conclude that souporcell is suitable for demultiplexing scRNAseq data collected as pools from different individuals. The authors propose that SNP-based demultiplexing can be used to monitor and correct for batch effects, whenever data need to be collected as pools (for example: small sample sizes, developmental datasets etc).

Major comments:

The manuscript makes a convincing case for the ability of a preexisting SNP-based demultiplexing tool, called souporcell, to demultiplex pooled samples. The study uses three methods for validation: 1. In silico data pooling; 2. Pooling of transgenic lines; 3. Pooling of cells tagged with CMOs (10x genomics). The results are consistent across experiments.

The authors propose that souporcell is a solution for demultiplexing pooled samples whenever sample tagging methods are not feasible. Although the authors test this approach in several species and conditions, the validation does not cover all possible cases and situations, obviously. Indeed, the authors recommend potential users to run pilot validation experiments with a secondary demultiplexing methods.

However, the manuscript would become more useful if the following points are addressed:

First, what is the genetic relatedness of the individuals pooled in the experiments? What is the SNP frequency in the samples analyzed, and how does that compare to SNP frequency in mouse strains? (The number of SNPs in the VCF is reported in a supplementary table but not discussed in the main text). This point is extremely important: as the authors mention, it is not possible to demultiplex samples from the same mouse strain. Inbreeding is relatively common in laboratory species, even unconventional ones; therefore, information on genetic relatedness and SNP rate would help readers assess whether SNP-based demultiplexing has a good chance to work in their systems. Addressing this point does not require any additional experiments, and computing from the single-cell reads how many SNPs distinguish the individuals pooled here should be straightforward.

Moreover, the relatively limited number of samples pooled does not validate the use of souporcell with a larger number of samples. For example: in developmental studies, often dozens of embryos are collected and pooled. What are the potential caveats of using souporcell for demultiplexing larger number of samples? The Discussion would be a good place to warn potential users of the limitations of the approach.

Minor comments:

• is the accuracy of doublet detection rate a function of number of samples? This can be tested by repeating the monkey in silico experiment with three individuals.

Significance

The manuscript presents a technical advance, by validating the use of souporcell for demultiplexing scRNAseq data collected from non-conventional animal species.

The audience potentially interested in this paper is relatively broad. Potential readers include biologists that collect and analyze scRNAseq data from pooled samples, for instance scientists working in the fields of embryonic development and evolutionary developmental biology, but also clinical researchers. The manuscript will be particularly interesting for scientists working on amphibians, because souporcell is validated experimentally in three amphibian species.

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:

Provide a short summary of the findings and key conclusions (including methodology and model system(s) where appropriate). Please place your comments about significance in section 2.

This study provided a SNP-based demuxers to facilitate effective experimental design of scRNA-seq. This model used discrepancies in SNPs across species or individuals to trace back the source of cells in scRNA-seq experiments. Benchmarking the performance of demultiplexing, this study analyzed in silico or experimentally pooled scRNA-seq data from species including zebrafish, African green monkeys, Xenopus laevis, axolotl, Pleurodeles waltl, and Notophthalmus viridescens. It demonstrated that high accurately demultiplex can be achieved regardless of existence of genome and a common SNP set. Overall, this study provided an economical, powerful, and less-biased pooled scRNA-seq data analysis method, depending minimally on the availability of genomic resources.

Major comments:

1. SNP-based demultiplexing performed well on some species, such as zebrafish and Africa green monkey, from which over 90% of the cells analyzed were correctly identified. However, this accuracy decreases in Pleurodeles samples when a common SNPs VCF file is absent (Fig.3). It showed that cell identity can be more precisely defined with the increase of average read depth (Fig 3B). So, I am wondering whether the mis-defined cells shown in Fig. 3E, actually are cells with lower reads. It is better if the authors can test such a correlation between the cell identity and the depth of reads using the data from Fig. 3E.
2. Please discuss limitations of this approach in the manuscript. (1) To which extent, when SNPs are roughly present in the individuals of same species, SNP-based demultiplexing can be applied, e.g., individuals from an inbred strain (c57bl6 mice) would not work.(2) The authors experimentally tested two newt species using SNP-based demultiplexing. When multiple species are experimentally applied, may the cell/nuclei size variation cause problem?
3. What is the upper limit number of samples when using this model. Please make some estimation or discussion about it.

Minor comments:

1. Please add an algorithm principle of this model.
2. Give a clear definition of doublets including the ground truth and Souporcell result.
3. Authors should indicate the time cost of running one round of such analysis, the minimal computational requirements?

Significance

1. Accurate demultiplexing of pooled data can reduce the batch effect between data and experimental costs.
2. This model will achieve good results in analyzing cell evolution between different species, or individuals of same species carrying sufficient SNPs.
3. It is sufficient to run this analysis only with a de novo transcriptome, opened the possibility of using pooled sc-RNA analysis on less-investigated species.

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

Cardiello et al tested if souporcell (https://pubmed.ncbi.nlm.nih.gov/32366989/) can be used to demultiplex samples for some model organisms, based on identified SNPs. For this, they used synthetic multiplexed data, publicly available datasets and some new datasets, spanning samples from five model organisms. Their analysis indicates that souporcell could be used to demultiplex scRNA-seq experiments for multiple species, which offers a cost-beneficent approach.

The manuscript reads well and shows this approach can work for different model organisms. However, unfortunately, I am confused about the amount of novelty in this manuscript. The method, souporcell, is already published. The authors indicate souporcell is not validated in non-human samples, but the original paper states that their method works with malaria parasite data (Fig 3b, FigS4). Adapting and using an available tool for different model organisms is good and groups working on different model organisms may find this manuscript useful, but the same could be said for the original article. Due to these reasons, I am not sure whether this manuscript has novelty sufficient for publication. I also wrote down two minor points below:

1. Doublets assigned by souporcell compared to the fluor-based assignment look random. In Fig 2 doublet recovery rate looks smaller, and in fig 3 doublet rate prediction looks more random. This is a bit confusing. Is there any explanation for this?
2. The authors discussed the immune system cells might show some variability in their discussion (referring to fig 3), but this is not clearly shown in the figures as data. Having a percentage bar graph could make it clearer for the readers. More generally, showing more direct evidence for the variability of different cell types (not just the immune system) could be informative for scRNA-seq users.

Significance

scRNA-Seq is becoming a routine approach to assay gene expression profiling. However, it remains costly. There are new approaches to multiplex and demultiplex samples to decrease the cost. Thus, it is good to see that one available tool works for five different model organisms.

Although it is good to see an available tool works for 5 different species, I am not sure about the novelty presented in this manuscript. Technical advances are not clear to this reviewer, as the method is already published. Moreover, this is a technical report manuscript and there is no biological conceptual advance. As a developmental biologist using single-cell mRNA sequencing, someone more directly from the single-cell field may have further comments on novelty, recommendations for references, and could comment on computational aspects in more detail.