Author response:

The following is the authors’ response to the previous reviews.

Public Reviews: 

Reviewer #1 (Public Review): 

In this manuscript by Wu et al., the authors present the high resolution cryoEM structures of the WT Kv1.2 voltagegated potassium channel. Along with this structure the authors have solved several structures of mutants or experimental conditions relevant to the slow inactivation process that these channels undergo and which is not yet completely understood. 

One of the main findings is the determination of the structure of a mutant (W366F) that is thought to correspond to the slow inactivated state. These experiments confirm results in similar mutants in different channels from Kv1.2 that indicate that inactivation is associated with an enlarged selectivity filter. 

Another interesting structure is the complex of Kv1.2 with the pore blocking toxin Dendrotoxin 1. The results shown in the revised version indicate that the mechanism of block is similar to that of related blocking-toxins, in which a lysine residue penetrates in the pore. Surprisingly, in these new structures, the bound toxin results in a pore with empty external potassium binding sites. 

The quality of the structural data presented in this revised manuscript is very high and allows for unambiguous assignment of side chains. The conclusions are supported by the data. This is an important contribution that should further our understanding of voltage-dependent potassium channel gating. In the revised version, the authors have addressed my previous specific comments, which are appended below. 

(1) In the main text's reference to Figure 2d residues W18' and S22' are mentioned but are not labeled in the insets. 

This has been fixed: line 229, p. 9.

(2) On page 8 there is a discussion of how the two remaining K+ ions in binding sites S3 and S4 prevent permeation K+ in molecular dynamics. However, in Shaker, inactivated W434F channels can sporadically allow K+ permeation with normal single-channel conductance but very reduced open times and open probability at not very high voltages. 

This is noted in the discussion Lines 497-500, p. 18

(3) The structures of WT in the absence of K+ shows a narrower selectivity filter, however Figure 4 does not convey this finding. In fact, the structure in Figure 4B is constructed in such an angle that it looks as if the carbonyl distances are increased, perhaps this should be fixed. Also, it is not clear how the distances between carbonyls given in the text on page 12 are measured. Is it between adjacent or kitty-corner subunits? 

We have changed Fig. 4B to show the same view as in Fig. 4A. In the legend we explain that opposing subunits are shown. We no longer give distances, in view of the lack of detectable carbonyl densities.

(4) It would be really interesting to know the authors opinion on the driving forces behind slow inactivation. For example, potassium flux seems to be necessary for channels to inactivate, which might indicate a local conformational change is the trigger for the main twisting events proposed here. 

We address this in the Discussion, line 506-523, pp. 18-19.

Reviewer #2 (Public Review): 

Cryo_EM structures of the Kv1.2 channel in the open, inactivated, toxin complex and in Na+ are reported. The structures of the open and inactivated channels are merely confirmatory of previous reports. The structures of the dendrotoxin bound Kv1.2 and the channel in Na+ are new findings that will of interest to the general channel community. 

Review of the resubmission: 

I thank the authors for making the changes in their manuscript as suggested in the previous review. The changes in the figures and the additions to the text do improve the manuscript. The new findings from a further analysis of the toxin channel complex are welcome information on the mode of the binding of dendrotoxin. 

A few minor concerns: 

(1) Line 93-96, 352: I am not sure as to what is it the authors are referring to when they say NaK2P. It is either NaK or NaK2K. I don't think that it has been shown in the reference suggested that either of these channels change conformation based on the K+ concentration. Please check if there is a mistake and that the Nichols et. al. reference is what is being referred to. 

Thank you for noticing the error. We meant NaK2K and we have changed this throughout.

(2) Line 365: In the study by Cabral et. al., Rb+ ions were observed by crystallography in the S1, S3 and S4 site, not the S2 site. Please correct. 

Thank you. We have re-written this section, lines 364-381, pp. 13-14.

Reviewer #3 (Public Review): 

Wu et al. present cryo-EM structures of the potassium channel Kv1.2 in open, C-type inactivated, toxin-blocked and presumably sodium-bound states at 3.2 Å, 2.5 Å, 2.8 Å, and 2.9 Å. The work builds on a large body of structural work on Kv1.2 and related voltage-gated potassium channels. The manuscript presents a plethora of structural work, and the authors are commended on the breadth of the studies. The structural studies are well-executed. Although the findings are mostly confirmatory, they do add to the body of work on this and related channels. Notably, the authors present structures of DTx-bound Kv1.2 and of Kv1.2 in a low concentration of potassium (which may contain sodium ions bound within the selectivity filter). These two structures add considerable new information. The DTx structure has been markedly improved in the revised version and the authors arrive at well-founded conclusions regarding its mechanism of block. Regarding the Na+ structure, the authors claim that the structure with sodium has "zero" potassium - I caution them to make this claim. It is likely that some K+ persists in their sample and that some of the density in the "zero potassium" structure may be due to K+ rather than Na+. This can be clarified by revisions to the text and discussion. I do not think that any additional experiments are needed. Overall, the manuscript is well-written, a nice addition to the field, and a crowning achievement for the Sigworth lab. 

Most of this reviewer's initial comments have been addressed in the revised manuscript. Some comments remain that could be addressed by revisions of the text. 

Specific comments on the revised version: 

Quotations indicate text in the manuscript. 

(1) "While the VSD helices in Kv1.2s and the inactivated Kv1.2s-W17'F superimpose very well at the top (including the S4-S5 interface described above), there is a general twist of the helix bundle that yields an overall rotation of about 3o at the bottom of the VSD." 

Comment: This seemed a bit confusing. I assume the authors aligned the complete structures - the differences they indicate seem to be slight VSD repositioning relative to the pore rather than differences between the VSD conformations themselves. The authors may wish to clarify. As they point out in the subsequent paragraph, the VSDs are known to be loosely associated with the pore. 

We aligned the VSDs alone, and it is a twist of the VSD helix bundle.

This is now clarified in lines 269-273, p. 10.

(2) Comment: The modeling of DTx into the density is a major improvement in the revision. Figure 3 displays some interactions between the toxin and Kv1.2 - additional side views of the toxin and the channel might allow the reader to appreciate the interactions more fully. The overall fit of the toxin structure into the density is somewhat difficult to assess from the figure. (The authors might consider using ChimeraX to display density and model in this figure.) 

We have added new panels, and stereo pairs, to Figure 3.

(3) "We obtained the structure of Kv1.2s in a zero K+ solution, with all potassium replaced with sodium, and were surprised to find that it is little changed from the K+ bound structure, with an essentially identical selectivity filter conformation (Figure 4B and Figure 4-figure supplement 1)." 

Comment: It should be noted in the manuscript that K+ and Na+ ions cannot be distinguished by the cryo-EM studies - the densities are indistinguishable. The authors are inferring that the observed density corresponds to Na+ because the protein was exchanged from K+ into Na+ on a gel filtration (SEC) column. It is likely that a small amount of K+ remains in the protein sample following SEC. I caution the authors to claim that there is zero K+ in solution without measuring the K+ content of the protein sample. Additionally, it should be considered that K+ may be present in the blotting paper used for cryo-EM grid preparation (our laboratory has noted, for example, a substantial amount of Ca2+ in blotting paper). The affinity of Kv1.2 for K+ has not been determined, to my knowledge - the authors note in the Discussion that the Shaker channel has "tight" binding for K+. It seems possible that some portion of the density in the selectivity filter could be due to residual K+. This caveat should be clearly stated in the main text and discussion. More extensive exchange into Na+, such as performing the entire protein purification in NaCl, or by dialysis (as performed for obtaining the structure of KcsA in low K+ by Y. Zhou et al. & Mackinnon 2001), would provide more convincing removal of K+, but I suspect that the Kv1.2 protein would not have sufficient biochemical stability without K+ to endure this treatment. One might argue that reduced biochemical stability in NaCl could be an indication that there was a meaningful amount of K+ in the final sample used for cryo-EM (or in the particles that were selected to yield the final high-resolution structure).

We now explain in the Methods section, in more detail the steps taken to avoid any residual Na+ contamination during purification, lines 683-687, pp. 24-25. We have changed the text to point out that the ion species cannot be distinguished in the maps, and note results in NaK2K and KcsA (lines 368-381, pp. 13-14).

We note that the same procedures to remove K+ were used for the Kv1.2sW17’F structure (line 385, p. 14). We qualify the ion replacement to say that Na+ replaces “essentially” all K+ (line 607, p. 21).

(4) Referring to the structure obtained in NaCl: "The ion occupancy is also similar, and we presume that Kv1.2 is a conducting channel in sodium solution." 

Comment: Stating that "Kv1.2 is a conducting channel in sodium solution" and implying that conduction of Na+ is achieved by an analogous distribution of ion binding sites as observed for K+ are strong statements to make - and not justified by the experiments provided. Electrophysiology would be required to demonstrate that the channel conducts sodium in the absence of K+. More complete ionic exchange, better control of the ionic conditions (Na+ vs K+), and affinity measurements for K+ would be needed to determine the distribution of Na+ in the filter (as mentioned above). At minimum, the authors should revise and clarify what the intended meaning of the statement "we presume that Kv1.2 is a conducting channel in sodium solution". As mentioned above, it seems possible/likely that a portion of the density in the filter may be due to K+. 

We now present a more detailed argument (lines 376 to 381, pp. 13-14.)

Recommendations for the authors: 

Reviewing Editor: 

After consultation, the reviewers agree that, although the authors have answered most of the comments raised in the previous review, there remains a concern about the structure obtained in the presence if Na. Given that Kv1.2 is more reluctant to slow inactivation, the conducting structure in Na+ could be due to this fact or that it really has higher affinity for K+ than Na+. In the presence of even a small contamination by K+, this ion could thus occupy the selectivity filter, resulting in an open conformation. The authors should clearly state the steps taken to ensure no contamination by K+. It is also possible that indeed the open structure occurs even in the presence of Na+ in the selectivity filter. This should be also discussed, given that this has been observed in other potassium channel structures. 

Reviewer #1 (Recommendations For The Authors): 

In this revised version of the manuscript, the authors have adequately addressed my previous points and improved the clarity and readability of the manuscript. This is a compelling work that shows inactivated structures if the Kv1.2 potassium channel, especially interesting is a structure in the absence of extracellular potassium ions, that can help understand how a reduction in the availability of these ions speed up entrance into the inactivated state in these ion channels. 

I would just recommend that in the absence of functional data (current recordings) when potassium is removed, the authors just use caution in ascribing this structure to an inactivated state. Also, it should be mentioned that the observed ion densities observed in the pore cannot unambiguously be identified as sodium ions. 

Reviewer #3 (Recommendations For The Authors): 

(1)  "The nearby Leu9 is also important as its substitution by alanine also decreases affinity 1000-fold, but we observe no contacts between this residue and residues of the Kv1.2s channel." 

Comment: It seems early in the text to mention the potential interaction of Leu9 to the channel structure. The authors may wish to discuss Leu9 later in the manuscript - a figure showing the location of Leu9 would strengthen the statement. Any hypothesis on why mutation of it has such a profound effect? 

Add a figure panel showing Leu9 position.

We have rewritten the text as suggested, and have identified Leu9 in several panels of Fig. 3.

(2)  "The X-ray structure of a-DTX (Figure 3A)" 

Comment: The authors may wish to cite a reference to this X-ray structure. 

We now cite Skarzynski (1992) on line 321, p. 12.

Loved how it sounded, they did say the same things over and over again, and it was so cool that they had an old English part.

When reading so quick, the brain has little time to process all of the information that was there. By doing this it will affect the way you may think on a specific topic as you may not be able to cover everything in that text.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

My main point of concern is the precision of dissection. The authors distinguish cells isolated from the tailbud and different areas in the PSM. They suggest that the cell-autonomous timer is initiated, as cells exit the tailbud.

This is also relevant for the comparison of single cells isolated from the embryo and cells within the embryo. The dissection will always be less precise and cells within the PSM4 region could contain tailbud cells (as also indicated in Figure 1A), while in the analysis of live imaging data cells can be selected more precisely based on their location. This could therefore contribute to the difference in noise between isolated single cells and cells in the embryo. This could also explain why there are "on average more peaks" in isolated cells (p. 6, l. 7).

This aspect should be considered in the interpretation of the data and mentioned at least in the discussion. (It does not contradict their finding that more anterior cells oscillate less often and differentiate earlier than more posterior ones.)

Reviewer #1 rightly points out that selecting cells in a timelapse is more precise than manual dissection. Manual dissection is inherently variable but we believe in general it is not a major source of noise in our experiments. To control for this, we compared the results of 11 manual dissections of the posterior quarter of the PSM (PSM4) with those of the pooled PSM4 data. In general, we did not see large differences in the distributions of peak number or arrest timing that would markedly increase the variability of the pooled data above that of the individual dissections (Figure 1 – supplement figure 7). We have edited the text in the Results to highlight this control experiment (page 6, lines 13-17).

It is of course possible that we picked up adjacent TB cells when dissecting PSM4, however the reviewer’s assertion that inclusion of TB cells “could also explain why there are "on average more peaks" in isolated cells” is incorrect. Later in the paper we show that cells from the TB have almost identical distributions to PSM4 (mean ± SD, PSM4 4.36 ± 1.44; TB 4.26 ± 1.35; Figure 4 _ supplement 1). Thus, inadvertent inclusion of TB cells while dissecting would in fact not increase the number of peaks.

Here, the authors focus on the question of how cells differentiate. The reverse question is not addressed at all. How do cells maintain their oscillatory state in the tailbud? One possibility is that cells need external signals to maintain that as indicated in Hubaud et al. 2014. In this regard, the definition of tailbud is also very vague. What is the role of neuromesodermal progenitors? The proposal that the timer is started when cells exit the tailbud is at this point a correlation and there is no functional proof, as long as we do not understand how cells maintain the tailbud state. These are points that should be considered in the discussion.

The reviewer asks “How do cells maintain their oscillatory state in the tailbud?”. This is a very interesting question, but as recognized by the reviewer, beyond the scope of our current paper.

We now further emphasize the point “One possibility is that cells need external signals to maintain … as indicated in Hubaud et al. 2014” in the Discussion and added a reference to the review Hubaud and Pourquié 2014 (Signalling dynamics in vertebrate segmentation. Nat Rev Mol Cell Biol 15, 709–721 (2014). https://doi.org/10.1038/nrm3891) (page 18, lines 19-22).

To clarify the definition of the TB, we have stated more clearly in the results (page 15, lines 8-12) that we defined TB cells as all cells posterior to the notochord (minus skin) and analyzed those that survived

>5 hours post-dissociation, did not divide, and showed transient Her1-YFP dynamics.

The reviewer asks: What is the role of neuromesodermal progenitors? In responding to this, we refer to Attardi et al., 2018 (Neuromesodermal progenitors are a conserved source of spinal cord with divergent growth dynamics. Development. 2018 Nov 9;145(21):dev166728. doi: 10.1242/dev.166728).

Around the stage of dissection in zebrafish in our work, there is a small remaining group of cells characterized as NMPs (Sox2 +, Tbxta+ expression) in the dorsal-posterior wall of the TB. These NMPs rarely divide and are not thought to act as a bipotential pool of progenitors for the elongating axis, as is the case in amniotes, rather contributing to the developing spinal cord. How this particular group of cells behaves in culture is unclear as we did not subdivide the TB tissue before culturing. It would be possible to specifically investigate these NMPs regarding a role in TB oscillations, but given the results of Attardi et al., 2018 (small number of cells, low bipotentiality), we argue that it would not be significant for the conclusions of the current work. To indicate this, we included a sentence and a citation of this paper in the results towards the beginning of the section on the tail bud (page 15, lines 8-12).

The authors observe that the number of oscillations in single cells ex vivo is more variable than in the embryo. This is presumably due to synchronization between neighbouring cells via Notch signalling in the embryo. Would it be possible to add low doses of Notch inhibitor to interfere with efficient synchronization, while at the same time keeping single cell oscillations high enough to be able to quantify them?

It is a formal possibility that Delta-Notch signaling may have some impact on the variability in the number of oscillations. However, we argue that the significant amount of cell tracking work required to carry out the suggested experiments would not be justified, considering what has been previously shown in the literature. If Delta-Notch signaling was a major factor controlling the variability of the intrinsic program that we describe, then we would expect that in Delta-Notch mutants the anterior- posterior limits of cyclic gene expression in the PSM would extend beyond those seen in wildtype embryos. Specifically, we might expect to see her1 expression extending more anteriorly in mutants to account for the dramatic increase in the number of cells that have 5, 6, 7 and 8 cycles in culture (Fig. 1E versus Fig. 1I). However, as shown in Holley et al., 2002 (Fig. 5A, B; her1 and the notch pathway function within the oscillator mechanism that regulates zebrafish somitogenesis. Development. 2002 Mar;129(5):1175-83. doi: 10.1242/dev.129.5.1175), the anterior limit of her1 expression in the PSM in DeltaD mutants (aei) is not different to WT. Thus, Delta-Notch signaling may exert a limited control over the number of oscillations, but likely not in excess of one cycle difference.

In the same direction, it would be interesting to test if variation is decreased, when the number of isolated cells is increased, i.e. if cells are cultured in groups of 2, 3 or 4 cells, for instance.

This is a great proposal – however the culture setup used here is a wide-field system that doesn’t allow us to accurately follow more than one cell at a time. Cells that adhere to each other tend to crawl over each other, blurring their identity in Z. This is also why we excluded dividing cells in culture from the analysis. Experiments carried out with a customized optical setup will be needed to investigate this in the future.

It seems that the initiation of Mesp2 expression is rather reproducible and less noisy (+/- 2 oscillation cycles), while the number of oscillations varies considerably (and the number of cells continuing to oscillate after Mesp2 expression is too low to account for that). How can the authors explain this apparent discrepancy?

The observed tight linkage of the Mesp onset and Her1 arrest argue for a single timing mechanism that is upstream of both gene expression events; indeed, this is one of the key implications of the paper. However, the infrequent dissociation of these events observed in FGF-treated cells suggests that more than one timing pathway could be involved, although there are other interpretations. We’ve added more discussion in the text on one vs multi-timers (page 17, lines 19-23 – page 18, line 1 - 8)., see next point.

The observation that some cells continue oscillating despite the upregulation of Mesp2 should be discussed further and potential mechanism described, such as incomplete differentiation.

This is an infrequent (5 out of 54 cells) and interesting feature of PSM4 cells in the presence of FGF. We imagine that this disassociation of clock arrest from mesp on-set timing could be the result of alterations in the thresholds in the sensing mechanisms controlling these two processes. Alternatively - as reviewer 2 argues - it might reflect multiple timing mechanisms at work. We have added a discussion of these alternative interpretations (page 17, lines 19-23 – page 18, line 1 - 8).

Fig. 3 supplement 3 B missing

It’s there in the BioRxiv downloadable PDF and full text – but seems to not be included when previewing the PDF. Thanks for the heads up.

Reviewer #2 (Public Review):

The authors demonstrate convincingly the potential of single mesodermal cells, removed from zebrafish embryos, to show cell-autonomous oscillatory signaling dynamics and differentiation. Their main conclusion is that a cell-autonomous timer operates in these cells and that additional external signals are integrated to tune cellular dynamics. Combined, this is underlying the precision required for proper embryonic segmentation, in vivo. I think this work stands out for its very thorough, quantitative, single-cell real-time imaging approach, both in vitro and also in vivo. A very significant progress and investment in method development, at the level of the imaging setup and also image analysis, was required to achieve this highly demanding task. This work provides new insight into the biology underlying embryo axis segmentation.

The work is very well presented and accessible. I think most of the conclusions are well supported. Here a my comments and suggestions:

The authors state that "We compare their cell-autonomous oscillatory and arrest dynamics to those we observe in the embryo at cellular resolution, finding remarkable agreement."

I think this statement needs to be better placed in context. In absolute terms, the period of oscillations and the timing of differentiation are actually very different in vitro, compared to in vitro. While oscillations have a period of ~30 minutes in vivo, oscillations take twice as long in vitro. Likewise, while the last oscillation is seen after 143 minutes in vivo, the timing of differentiation is very significantly prolonged, i.e.more than doubled, to 373min in vitro (Supplementary Figure 1-9). I understand what the authors mean with 'remarkable agreement', but this statement is at the risk of being misleading. I think the in vitro to in vivo differences (in absolute time scales) needs to be stated more explicitly. In fact, the drastic change in absolute timescales, while preserving the relative ones, i.e. the number of oscillations a cell is showing before onset of differentiation remains relatively invariant, is a remarkable finding that I think merits more consideration (see below).

We have changed the text in the abstract (page 1, line 28) to clarify that the agreement is in the relative slowing, intensity increases and peak numbers.

One timer vs. many timers

The authors show that the oscillation clock slowing down and the timing of differentiation, i.e. the time it takes to activate the gene mesp, are in principle dissociable processes. In physiological conditions, these are however linked. We are hence dealing with several processes, each controlled in time (and thereby space). Rather than suggesting the presence of ‘a timer’, I think the presence of multiple timing mechanisms would reflect the phenomenology better. I would hence suggest separating the questions more consistently, for instance into the following three:

a.  what underlies the slowing down of oscillations?

b.  what controls the timing of onset of differentiation?

c.  and finally, how are these processes linked?

Currently, these are discussed somewhat interchangeably, for instance here: “Other models posit that the slowing of Her oscillations arise due to an increase of time-delays in the negative feedback loop of the core clock circuit (Yabe, Uriu, and Takada 2023; Ay et al. 2014), suggesting that factors influencing the duration of pre-mRNA splicing, translation, or nuclear transport may be relevant. Whatever the identity, our results indicate the timer ought to exert control over differentiation independent of the clock.”(page 14). In the first part, the slowing down of oscillations is discussed and then the authors conclude on 'the timer', which however is the one timing differentiation, not the slowing down. I think this could be somewhat misleading.

To help distinguish the clock’s slowing & arrest from differentiation, we have clarified the text in how we describe our experiments using her1-/-; her7-/- cells (page 10, lines 9-20).

From this and previous studies, we learn/know that without clock oscillations, the onset of differentiation still occurs. For instance in clock mutant embryos (mouse, zebrafish), mesp onset is still occurring, albeit slightly delayed and not in a periodic but smooth progression. This timing of differentiation can occur without a clock and it is this timer the authors refer to "Whatever the identity, our results indicate the timer ought to exert control over differentiation independent of the clock." (page 14). This 'timer' is related to what has been previously termed 'the wavefront' in the classic Clock and Wavefront model from 1976, i.e. a "timing gradient' and smooth progression of cellular change. The experimental evidence showing it is cell-autonomous by the time it has been laid down,, using single cell measurements, is an important finding, and I would suggest to connect it more clearly to the concept of a wavefront, as per model from 1976.

We have been explicit about the connection to the clock & wavefront in the discussion (page 17, line 12-17).

Regarding question a., clearly, the timer for the slowing down of oscillations is operating in single cells, an important finding of this study. It is remarkable to note in this context that while the overall, absolute timescale of slowing down is entirely changed by going from in vivo to in vitro, the relative slowing down of oscillations, per cycle, is very much comparable, both in vivo and in vivo.

We have now pointed out the relative nature of this phenomenon in the abstract, page 1, line 28.

To me, while this study does not address the nature of this timer directly, the findings imply that the cell-autonomous timer that controls slowing down is, in fact, linked to the oscillations themselves. We have previously discussed such a timer, i.e. a 'self-referential oscillator' mechanism (in mouse embryos, see Lauschke et al., 2013) and it seems the new exciting findings shown here in zebrafish provide important additional evidence in this direction. I would suggest commenting on this potential conceptual link, especially for those readers interested to see general patterns.

While we posit that the timer provides positional info to the clock to slow oscillations and instruct its arrest – we do not believe that “the findings imply that the cell-autonomous timer that controls slowing down is, in fact, linked to [i.e., governed by] the oscillations themselves.”. As we show, in her1-/-; her7-/- embryos lacking oscillations, the timing / positional information across the PSM still exists as read-out by Mesp expression. Is this different positional information than that used by the clock? – possibly – but given the tight linkage between Mesp onset and the timing/positioning of clock arrest, both cell-autonomously and in the embryo, we argue that the simplest explanation is that the timing/positional information used by the clock and differentiation are the same. Please see page 10, lines 9-20, as well as the discussion (page 17, lines 19-23; page 18. Lines 1-8 ).

We agree that the timer must communicate to the clock– but this does not mean it is dependent on the clock for positional information.

Regarding question c., i.e. how the two timing mechanisms are functionally linked, I think concluding that "Whatever the identity, our results indicate the timer ought to exert control over differentiation independent of the clock." (page 14), might be a bit of an oversimplification. It is correct that the timer of differentiation is operating without a clock, however, physiologically, the link to the clock (and hence the dependence of the timescale of clock slowing down), is also evident. As the author states, without clock input, the precision of when and where differentiation occurs is impacted. I would hence emphasize the need to answer question c., more clearly, not to give the impression that the timing of differentiation does not integrate the clock, which above statement could be interpreted to say.

As far as we can tell, we do not state that “without clock input, the precision of when and where differentiation occurs is impacted”, and we certainly do not want to give this impression. In contrast, as mentioned above, the her1-/-; her7-/- mutant embryo studies indicate that the lack of a clock signal does not change the differentiation timing, i.e. it does not integrate the clock. Of course, in the formation of a real somite in the embryo, the clock’s input might be expected to cause a given cell to differentiate a little earlier or later so as to be coordinated with its neighbors, for example, along a boundary. But this magnitude of timing difference is within one clock cycle at most, and does not match the large variation seen in the cultured cells that spans over many clock cycles.

A very interesting finding presented here is that in some rare examples, the arrest of oscillations and onset of differentiation (i.e. mesp) can become dissociated. Again, this shows we deal here with interacting, but independent modules. Just as a comment, there is an interesting medaka mutant, called doppelkorn (Elmasri et al. 2004), which shows a reminiscent phenotype "the Medaka dpk mutant shows an expansion of the her7 expression domain, with apparently normal mesp expression levels in the anterior PSM.". The authors might want to refer to this potential in vivo analogue to their single cell phenotype.

Thank you, we had forgotten this result. Although we do not agree that this result necessarily means there are two interacting modules, we have included a citation to the paper, along with a discussion of alternative explanations for the dissociation (page 18, lines 2-14).

One strength of the presented in vitro system is that it enables precise control and experimental perturbations. A very informative set of experiments would be to test the dependence of the cell-autonomous timing mechanisms (plural) seen in isolated cells on ongoing signalling cues, for instance via Fgf and Wnt signaling. The inhibition of these pathways with well-characterised inhibitors, in single cells, would provide important additional insight into the nature of the timing mechanisms, their dependence on signaling and potentially even into how these timers are functionally interdependent.

We agree and in future experiments we are taking advantage of this in vitro system to directly investigate the effect of signaling cues on the intrinsic timing mechanism.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Joint Public Review: 

The molecular mechanisms that mediate the regulated exocytosis of neuropeptides and neurotrophins from neurons via large dense-core vesicles (LDCVs) are still incompletely understood. Motivated by their earlier discovery that the Rab3-RIM1 pathway is essential for neuronal LDCV exocytosis, the authors now examined the role of the Rab3 effector Rabphilin-3A in neuronal LDCV secretion. Based on multiple live and confocal imaging approaches, the authors provide evidence for a synaptic enrichment of Rabphilin-3A and for independent trafficking of Rabphilin-3A and LDCVs. Using an elegant NPY-pHluorin imaging approach, they show that genetic deletion of Rabphilin-3A causes an increase in electrically triggered LDCV fusion events and increased neurite length. Finally, knock-out-replacement studies, involving Rabphilin-3A mutants deficient in either Rab3- or SNAP25-binding, indicate that the synaptic enrichment of Rabphilin-3A depends on its Rab3 binding ability, while its ability to bind to SNAP25 is required for its effects on LDCV secretion and neurite development. The authors conclude that Rabphilin-3A negatively regulates LDCV exocytosis and propose that this mechanism also affects neurite growth, e.g. by limiting neurotrophin secretion. These are important findings that advance our mechanistic understanding of neuronal large dense-core vesicle (LDCV) secretion. 

The major strengths of the present paper are: 

(i) The use of a powerful Rabphilin-3A KO mouse model. 

(ii) Stringent lentiviral expression and rescue approaches as a strong genetic foundation of the study. 

(iii) An elegant FRAP imaging approach. 

(iv) A cutting-edge NPY-pHluorin-based imaging approach to detect LDCV fusion events. 

We thank the reviewers for their positive evaluation of our manuscript.

Weaknesses that somewhat limit the convincingness of the evidence provided and the corresponding conclusions include the following: 

(i) The limited resolution of the various imaging approaches introduces ambiguity to several parameters (e.g. LDCV counts, definition of synaptic localization, Rabphilin-3A-LDCV colocalization, subcellular and subsynaptic localization of expressed proteins, AZ proximity of Rabphilin-3A and LDCVs) and thereby limits the reliability of corresponding conclusions. Super-resolution approaches may be required here. 

We thank the reviewer for their constructive suggestion. We fully agree that super-resolution imaging would produce a more precise localization of RPH3A and co-localization with DCVs. We have now repeated our (co)-localization experiments with STED microscopy. We find that RPH3A colocalized with the pre-synaptic marker Synapsin1 and, to a lesser extent, with the post synaptic marker Homer and DCV marker chromogranin B (new Figure 1). This indicates that RPH3A is highly enriched in synapses, mostly the pre-synapse, and that RPH3A partly co-localizes with DCVs.  

(ii) The description of the experimental approaches lacks detail in several places, thus complicating a stringent assessment. 

We apologize for the lack of detail in explaining the experimental approaches. We have included a more detailed description in the revised manuscript. 

(iii) Further analyses of the LDCV secretion data (e.g. latency, release time course) would be important in order to help pinpoint the secretory step affected by Rabphilin-3A. 

We agree. To address this comment, we have now included the duration of the fusion events (new Figure S2D-F). The start time of the fusion events are shown in the cumulative plots in now Figure 3F and I. The kinetics are normal in the RPH3A KO neurons.

(iv) It remains unclear why a process that affects a general synaptic SNARE fusion protein - SNAP25 - would specifically affect LDCV but not synaptic vesicle fusion. 

We agree that we have not addressed this issue systematically enough in the original manuscript. We have now added a short discussion on this topic in the Discussion of the revised manuscript (p 15, line 380-386). In brief, we do not claim full selectivity for the DCV pathway. Some effects of RPH3A deficiency on the synaptic vesicle cycle have been observed. Furthermore, because DCVs typically do not mix in the synaptic vesicle cluster and fuse outside the active zone (and outside the synapse), DCVs might be more accessible to RPH3A regulation.

(v) The mechanistic links between Rabphilin-3A function, LDCV density in neurites, neurite outgrowth, and the proposed underlying mechanisms involving trophic factor release remain unclear. 

We agree that we have not addressed all these links systematically enough in the original manuscript, although we feel that we have at least postulated the best possible working model to link RPH3A function to DCV exocytosis/neurotrophic factor release and neurite outgrowth (p 15-16, line 396-400). Of course, a single study cannot support all these links with sufficient experimental evidence. We have now added a short text on what we can conclude exactly based on our experiments and how we see the links between RPH3A function, DCV exocytosis/neurotrophic factor release, neurite outgrowth and DCV density in neurites (p 13-14, line 317-325).

Reviewer #1 (Public Review): 

Summary:

The manuscript by Hoogstraaten et al. investigates the effect of constitutive Rabphilin 3A (RPH3A) ko on the exocytosis of dense core vesicles (DCV) in cultured mouse hippocampal neurons. Using mCherry- or pHluorin-tagged NPY expression and EGFP- or mCherry tagged RPHA3, the authors first analyse the colocalization of DCVs and RPH3A. Using FRAP, the authors next analyse the mobility of DCVs and RAB3A in neurites. The authors go on to determine the number of exocytotic events of DCVs in response to high-frequency electrical stimulation and find that RPH3A ko increases the number of exocytotic events by a factor 2-3, but not the fraction of released DCVs in a given cell (8x 50Hz stim). In contrast, the release fraction is also increased in RBP3A KOs when doubling the stimulation number (16x 50Hz). They further observe that RPH3A ko increases dendrite and axon length and the overall number of ChgrB-positive DCVs. However, the overall number of DCVs and dendritic length in ko cells directly correlate, indicating that the number of vesicles per dendritic length remains unaffected in the RPH3A KOs. Lentiviral co-expression of tetanus toxin (TeNT) showed a non-significant trend to reduce axon and dendrite length in RPH3a KOs. Finally, the authors use co-expression of RAB3A and SNAP25 constructs to show that RAB3A but not SNAP25 interaction is required to allow the exocytosis-enhancing effect in RPH3A KOs. 

While the authors' methodology is sound, the microscopy results are performed well and analyzed appropriately, but their results in larger parts do not sufficiently support their conclusions. Moreover, the experiments are not always described in sufficient detail (e.g. FRAP; DCV counts vs. neurite length) to fully understand their claims. 

Overall, I thus feel that the manuscript does not provide a sufficient advance in knowledge. 

Strengths: 

- The authors' methodology is sound, and the microscopy results are performed well and analyzed appropriately. 

- Figure 2: The exocytosis imaging is elegant and potentially very insightful. The effect in the RPH3A KOs is convincing. 

- Figure 4: the logic of this experiment is elegant. It shows that the increased number of DCV fusion events in RPH3A KOs is related to the interaction of RPH3A with RAB3A but not with SNAP25. 

We thank the reviewer for their positive evaluation of our manuscript.

Weaknesses: 

- The results in larger parts do not sufficiently support the conclusions. 

- The experiments are not always described in sufficient detail (e.g. FRAP; DCV counts vs. neurite length) to fully understand their claims. 

- Not of sufficient advance in knowledge for this journal 

- The significance of differences in control experiments WT vs. KO) varies between experiments shown in different figures. 

- Axons and dendrites were not analyzed separately in Figures 1 and 2. 

- The colocalization study in Figure 1 would require super-resolution microscopy. 

To address the reviewers’ comments, we have provided a more detailed explanation of our analysis (p 19-20, line 521-542). In addition, we have repeated our colocalization experiments using STED microscopy, see Joint Public Review item (i).  

Reviewer #2 (Public Review): 

Summary: 

Hoogstraaten et al investigated the involvement of rabphilin-3A RPH3A in DCV fusion in neurons during calcium-triggered exocytosis at the synapse and during neurite elongation. They suggest that RPH3A acts as an inhibitory factor for LDV fusion and this is mediated partially via its interaction with SNAP25 and not Rab3A/Rab27. It is a very elegant study although several questions remain to be clarified. 

Strengths: 

The authors use state-of-the-art techniques like tracking NPY-PHluorin exocytosis and FRAP experiments to quantify these processes providing novel insight into LDCs exocytosis and the involvement of RPH3A. 

We thank the reviewer for their positive evaluation of our manuscript.

Weaknesses: 

At the current state of the manuscript, further supportive experiments are necessary to fully support the authors' conclusions. 

We thank the reviewer for their comments and suggestions. We have performed additional experiments to support our conclusions, see Joint Public Review items (i) – (iv)

Reviewer #3 (Public Review): 

Summary: 

The molecular mechanism of regulated exocytosis has been extensively studied in the context of synaptic transmission. However, in addition to neurotransmitters, neurons also secrete neuropeptides and neurotrophins, which are stored in dense core vesicles (DCVs). These factors play a crucial role in cell survival, growth, and shaping the excitability of neurons. The mechanism of release for DCVs is similar, but not identical, to that used for SV exocytosis. This results in slow kinetic and low release probabilities for DCV compared to SV exocytosis. There is a limited understanding of the molecular mechanisms that underlie these differences. By investigating the role of rabphilin-3A (RPH3A), Hoogstraaten et al. uncovered for the first time a protein that inhibits DCV exocytosis in neurons. 

Strengths: 

In the current work, Hoogstraaten et al. investigate the function of rabphilin-3A (RPH3A) in DVC exocytosis. This RAB3 effector protein has been shown to possess a Ca2+ binding site and an independent SNAP25 binding site. Using colocalization analysis of confocal imaging the authors show that in hippocampal neurons RPH3A is enriched at pre- and post-synaptic sites and associates specifically with immobile DCVs. Using site-specific RPH3A mutants they found that the synaptic location was due to its RAB3 interaction site. They further could show that RPH3A inhibits DCV exocytosis due to its interaction with SNAP25. They came to that conclusion by comparing NPY-pHluorin release in WT and RPH3A KO cells and by performing rescue experiments with RPH3A mutants. Finally, the authors showed that by inhibiting stimulated DCV release, RPH3A controlled the axon and dendrite length possibly through the reduced release of neurotrophins. Thereby, they pinpoint how the proper regulation of DCV exocytosis affects neuron physiology. 

We thank the reviewer for their positive evaluation of our manuscript.

Weaknesses: 

Data context 

One of the findings is that RPH3A accumulates at synapses and is mainly associated with immobile DCVs.

However, Farina et al. (2015) showed that 66% of all DCVs are secreted at synapses and that these DCVs are immobile prior to secretion. To provide additional context to the data, it would be valuable to determine if RPH3A KO specifically enhances secretion at synapses. Additionally, the authors propose that RPH3A decreases DCV exocytosis by sequestering SNAP25 availability. At first glance, this hypothesis appears suitable. However, due to RPH3A synaptic localization, it should also limit SV exocytosis, which it does not. In this context, the only explanation for RPH3A's specific inhibition of DCV exocytosis is that RPH3A is located at a synapse site remote from the active zone, thus protecting the pool of SNAP25 involved in SV exocytosis from binding to RPH3A. This hypothesis could be tested using super-resolution microscopy. 

We thank the reviewer for their suggestion. We have now performed super resolution microscopy, see Joint Public Review item (i). However, these new data do not necessarily explain the stronger effect of RP3A deficiency on DCV exocytosis, relative to SV exocytosis. We have added a short discussion on this topic to the revised manuscript, see Joint Public Review item (iv).

Technical weakness 

One technical weakness of this work consists in the proper counting of labeled DCVs. This is significant since most findings in this manuscript rely on this analysis. Since the data was acquired with epi-fluorescence or confocal microscopy, it doesn't provide the resolution to visualize individual DCVs when they are clumped. The authors use a proxy to count the number of DCVs by measuring the total fluorescence of individual large spots and dividing it by the fluorescence intensity of discrete spots assuming that these correspond to individual DCVs. This is an appropriate method but it heavily depends on the assumption that all DCVs are loaded with the same amount of NPY-pHluorin or chromogranin B (ChgB). Due to the importance of this analysis for this manuscript, I suggest that the authors show that the number of DCVs per µm2 is indeed affected by RPH3A KO using super-resolution techniques such as dSTORM, STED, SIM, or SRRF. 

The reviewer is correct that this is a crucial issue, that we have not addressed optimally until now. We have previously devoted a large part of a previous manuscript to this issue, but have not referred to this previous work clearly enough. We have now clarified this (p 7, line 187-190). In brief, we have previously quantified the ratio between fluorescent intensity of ChgB and NPY-pHluorin in confocal microscopy over the number of dSTORM puncta in sparse areas of WT mouse hippocampal neurons (Persoon et al., 2018). This quantification yielded a unitary fluorescence intensity per vesicle that was very stable of different neurons. Although there might be some underestimation of the total number of DCVs when using confocal microscopy, the study of Persoon et al. (2018) has demonstrated that these parameters correlate well and that the estimations are accurate. Considering that the rF/F0 is similar in RPH3A WT and KO neurons (now Figure S2I), meaning that the intensity of NPY-pHluorin of one fusion event is comparable, we can presume that this correlation also applies for the RPH3A KO neurons.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors): 

Major points: 

(1) The authors perform an extensive analysis regarding the colocalization of RPH3A and DCVs (Figure 1 upper part). This analysis is hampered by the fact that the recorded data has in relation to vesicle size limited resolution (> 1 µm) to allow making strong claims here. In my view, super-resolution microscopy would be required for the co-localization studies shown in Figure 1. 

We fully agree and have now performed super-resolution microscopy, see Joint Public Review item (i)

(2) The FRAP experiments (Figure 1 lower part) cannot be sufficiently understood from what is presented. The methods say that both laser channels were activated during bleaching but NPY-pHluorin is not bleached in Fig.1E. Explanation of the bleaching is not very circumspect. In 1D, it is rather EGFP-RPH3A that is entering the bleached area than the NPY vesicles. These experiments require a more careful explanation of methodology, observed results, and their interpretation. Overall, the observed effects in the original kymograph traces require a better explanation. 

We acknowledge that NPY-pHluorin in Figure 1E (now Figure 2C) is not completely bleached. NPY-pHluorin appeared to be more difficult to bleach than NPY-mCherry. However, it is important to clarify that we merely bleached the neurites to remove the stationary puncta and facilitate our analysis of DCV/RPH3A dynamics. This bleaching step does not affect the interpretation of our results. We apologize that this was not clearly stated in the text and have made the necessary adjustments in legend, results- and methods section, (p 6-7, line 162-163; p 5, line 140-142 and p 19, line 508-513). Additionally, we apologize for the accidental switch of the kymographs for NPY-mCherry and EGFP-RPH3A in Figure 1D (now Figure 2B, C). We greatly appreciate identifying this error.  

(3) Figure 1: The authors need to mention whether axons, dendrites, or both were analyzed throughout the different panels and how they were identified. Is it possible that axons were wrapping around dendrites in their cultures (compare e.g. Shimojo et al., 2015)? Given the limited spatial resolution and because of this wrapping, interpretation of results could be affected. 

We completely agree with the reviewer’s assessment and conclusion. We are unable to distinguish axons from dendrites using this experimental design. We have made sure to specify in the text that our observation that RPH3A does not co-travel with DCVs is true for both dendrites and axons, (p 5, line 150).

(4) Figure 2: The exocytosis imaging is elegant and potentially very insightful. The effect in the RPH3A KOs is convincing. However, the authors determine the efficacy of exocytosis from NPY-pHluorin unquenching of DCVs only. This is only one of several possible parameters to read out the efficiency of exocytosis. Kinetics like e.g. delay between stimulation and start of exocytosis events or release time course of NPY after DCV fusion were not determined. Such analysis could give a better insight into what process before or after the fusion of DCVs is affected by RPH3A ko. 

We fully agree with the reviewer. We have now included the duration of the fusion events (new Figure S2D-F). The start time of the fusion events are shown in the cumulative plots in now Figure 3F and I. The kinetics are normal in the RPH3A KO neurons.

Moreover, it needs to be mentioned whether 2C and D are from WT or ko cultures. It would be best to show representative examples from both genotypes. 

We have now adjusted this in the new figure (now Figure 3C, D).

The number of fusion events is much increased but the release fraction is not significantly changed. While this is consistent with results in Figure 4C it is at variance with 4F. This raises questions about the reliability of the effects in RPH3A KOs. 

The release fraction indicates the number of fusion events normalized to the total DCV pool. In Figure 4D, we observed a slightly bigger pool size, which explains the lack of significance when analyzing the released fraction. In Figure 4G, however, DCV pool sizes are similar between KO and WT, leading to a statistically significant effect on release fraction in KO neurons. Furthermore, Figures 4B and E distinctly show a substantial increase in fusion events in RPH3A KO neurons. This variability in pool size observed could potentially be attributed to variation in culture or inherent biological variability.

Given the increased number of ChgrB-positive DCVs in RPH3A KOs (shown in Figure 2) and that only the cumulative number of exocytosis events were analysed, how can the authors exclude that the RPH3A ko only affects vesicle number but not release, if the % change in released vesicles is not different to WT? Kinetics of release don't seem to be affected. Importantly, what was the density of NPY-pHluorin vesicles in WT vs. ko? 

In Figure 2 (now Figure 5) we show that RPH3A KO neurons are larger and contain more endogenous ChgB+ puncta than WT neurons. This increased number of ChgrB+ puncta scales with their size as puncta density is not increased. A previous study (Persoon et al., 2018) has demonstrated a strong correlation between DCV number and neuron size. Our data show that RPH3A deficiency increased DCV exocytosis, but the released fraction of vesicles depends on the total number of DCVs, which we determined during live recording by dequenching NPY-pHluorin using NH4+. Considering that this is an overexpression of a heterologous DCV-fusion reporter, and not endogenous staining of DCVs, as in the case of ChgrB+ puncta, some variability is not unexpected.

Also in these experiments, the question arises of whether the authors analyse axons, dendrites, or both throughout the different panels and how they were identified. 

In our experimental design we record all fusion events per cell, including both axons and dendrites but excluding the cell soma. We have clarified this in the method section, (p 19, line 508 and p 19, line 521-522).

(5) Figure 3: in D the authors show that ChgrB-pos. DCV density is slightly increased in KOs. How does this relate to the density of NPY-pHluorin DCVS in Figure 2? 

We do not observe a difference in NPY-pHluorin density (see Author response image 1). However, it is important to note that we relied on tracing neurites in live recording images to determine the neuronal size. In contrast, the ChgB density was based on dendritic length using MAP2 (post-hoc) staining was limited. In addition, Chgr+ puncta represent an endogenous DCV staining, NPY-pHluorin quantification is based on overexpression of a heterologous DCV-fusion reporter. These two factors likely contribute some variability.

Author response image 1.

The authors show a non-significant trend of TeNT coexpression to reduce axon and dendrite lengths in RPH3A KOs. While this trend is visible, I think one cannot draw conclusions from that when not reaching significance. The argument of the authors that the increased axon and dendrite lengths are created by growth factor peptide release from DCV during culture time is interesting. However, the fact that TeNT expression shows a trend toward reducing this effect on axons/dendrites is not sufficient to prove the release of such growth factors. 

We agree. We have toned down this speculation in the revised manuscript, (p 15-16, line 395-400).

Lastly, the authors don't provide insight into the mechanisms, of how RPH3A ko increases the number of DCVs per µm dendritic length in the neurons. In my view, there are too many loose ends in this story of how RPH3A ko first increases spontaneous release of DCVs and then enhances neurite growth and DCV density. Did the authors e.g. measure the spontaneous release of DCVs in their cultures? 

We measured spontaneous release of DCVs during the 30s baseline recording prior to stimulation. We observed no difference in spontaneous release between WT and KO neurons (now Figure S2H). However, baseline recording lasted only 30 seconds. It is possible that this was too short to detect subtle effects.

Other points: 

(1) Figure 4: the logic of this experiment is elegant. It shows that the increased number of DCV fusion events in RPH3A KOs is related to the interaction of RPH3A with RAB3A but not with SNAP25. As mentioned above, it is irritating that the reduction of fusion events in KOs and on the release fraction is sometimes reaching significance, but sometimes it does not. Likewise, the absence of significant effects on DCV numbers is not consistent with the results shown in Figures 3C and D. 

DCV numbers in Figure 3 (now Figure 5) are determined by staining for endogenous ChgB, whereas in Figure 4D and G DCV numbers are determined by overexpressing NPY-pHluorin and counting the dequenched puncta following a NH4+ puff.

(2) Figure 1B: truncation of the y-axis needs to be clearly indicated. 

We have replaced this figure with new Figure 1 and have indicated truncations of the y-axis when needed (new Figure 1E). 

(3) Page 10: "Given that neuropeptides are key modulators of adult neurogenesis (Mu et al., 2010), and that RPH3A depletion leads to increased DCV exocytosis, it is coherent that we observed longer neurites in RPH3A KO neurons." I cannot follow the argument of the authors here: what has neurogenesis to do with neurite length? 

We apologize for the confusion. We have clarified this in the revised text, (p 16, line 398-400).

Minor point: 

There are some typos in the manuscript. e.g., page 8: "... may partially dependent on regulated secretion...); page 6: "...to dequence all...". 

Thank you for noticing, we have corrected the typos.

Reviewer #2 (Recommendations For The Authors): 

(1) Supplementary Figure S1A, in my opinion, should be in Figure 1A as it illustrates all the constructs used in this study and helps the reader to follow it up. 

We thank the reviewer for their suggestion. However, we feel that with the adjustments we have made in Figure 1, the illustrations of the constructs fit better in Figure S1, since new Figure 1 shows the localization of endogenous RPH3A and not that of the constructs.  

(2) One of the conclusions of the manuscript is the synaptic localization of the different RPH3A mutants. The threshold for defining synaptic localization is not clear either from the images nor from the analysis: for example, the Menders coefficient for VGut1-Syn1 which is used as a positive control, ranges from 0.65-0.95 and that of RPH3A and Syn1 ranges from 0.5-0.95. These values should be compared to all mutants and the conclusions should be based on such comparison. 

We agree. We have now repeated our initial co-localization experiment with all the RPH3A mutants (now Figure S1D-F).  

(3) Strengthening this figure with STED/SIM/dSTORM microscopy can verify and add a new understanding of the subtle changes of RPH3A localization. 

We fully agree and have now added super-resolution microscopy data, see Joint Public Review item (i).

(4) As RAB3A/RAB27A (ΔRAB3A/RAB27A) loses the punctate distribution, please clarify how can it function at the synapse and not act as a KO. Is it sorted to the synapse and how does it is sorted to the synapse? 

We used lentiviral delivery to introduce our constructs, resulting in the overexpression of ΔRAB3A/RAB27A mutant RPH3A. This overexpression likely compensates for the loss of the punctate distribution of RPH3A, thereby maintaining its limiting effect on DCV exocytosis. It is plausible that under physiological conditions, the mislocalization of RPH3A would lead to increased exocytosis, similar to what we observed in the KO. 

(5) Is RPH3A expressed in both excitatory and inhibitory neurons? 

We agree this is an important question. Single cell RNA-seq already suggests the protein is expressed in both, but we nevertheless decided to test expression of RPH3A protein in excitatory and inhibitory neurons, using immunocytochemistry with VGAT and VGLUT as markers in hippocampal and striatal WT neurons. We found that RPH3A is expressed in both VGLUT+ hippocampal neurons and VGAT+ striatal neurons (new Figure S1A, B).  

(6) The differential use of ChgB and NPY as markers for DCVs should be clarified and compared as these are used at different stages of the manuscript. 

We have previously addressed the comparison between ChgB and NPY-pHluorin (Persoon et al., 2018). We made sure to indicate this more clearly throughout the manuscript to clarify the use of the two markers. 

(7) FRAP experiments- A graph describing NPY recovery should be added as a reference to 2H and discussed. 

We agree. We have made the necessary adjustments (new Figure 2G).

(8) Figure 2E shows some degree of "facilitation" between the 2 8x50 pulses RPH3A KO neurons. Can the author comment on that? What was the reason for using this dual stimulation protocol? 

There is indeed some facilitation between the two 8 x 50 pulses in KO neurons and to a lesser extent also in the WT neurons, which we have observed before in WT neurons (Baginska et al., 2023). Baginska et al. (2023) showed recently that different stimulation protocols can influence certain fusion dynamics, like the ratio of persistent and transient events and event duration. We used two different stimulation protocols to thoroughly investigate the effect of RPH3A on exocytosis, and assess the robustness of our findings regarding the number of fusion events. Fusion kinetics was similar in WT an KO neurons for both stimulation protocols (new Figure 2D-F).

(9) Figure 3 quantifies dendrites length and then moves to quantify both axon and dendrites for the Tetanus toxin experiment. What are the effects of KO on axon length? In the main figures, it is not mentioned but in S3 it seems not to be affected. How does it reconcile with the main conclusion on neurite length? 

Figure 3H (now Figure 6C) shows the effect of the KO on axon length: the axon length is increased in RPH3A KO neurons compared to WT, similar to dendrite length. Re-expressing RPH3A in KO neurons rescues axonal length to WT levels. In Figure S3, we observe a similar trend as in main Figure 3 (new Figure 6), yet this effect did not reach significance. Based on this, we concluded that neurite length is increased upon RPH3A depletion.

(10) For lay readers, please explain the total pool and how you measured it. However, see the next comment. 

We agree. We have now defined this better in the revised manuscript, (p 19, line 524-527 and p 20, line 535-539).

(11) It is a bit hard to understand if the total number of DCV was increased in the KO and if the pool size was increased and in which figure it is quantified. Some sentences like: "A trend towards a larger intracellular DCV pool in KO compared to WT neurons was observed" do not fit with "No difference in DCV pool size was observed between WT and KO neurons (Figure S2D)" or with "During stronger stimulation (16 bursts of 50 APs at 50 Hz), the total fusion and released fraction of DCVs were increased in KO neurons compared to WT". They are not directly supported, or not related to specific figures. Please indicate if the total DCVs pool, as measured by NH4, was increased and based on that, the fraction of the releasable DCVs following the long stimulation. From Figure 2H, the conclusion is an increase in fusion events. In general, NH4 is not quantified clearly- is it quantified in Figure S2C? And if it is a trend, how can it become significant in Figure 3? 

We agree there has been some inconsistency in the way we describe the data on the total number of DCVs. We have addressed this in the revised text to ensure better clarity. The total DCV pool measured by NPY-pHluorin was not significantly increased in KO neurons, we see a trend towards a bigger DCV pool in the 2x8 50 Hz stimulation paradigm (now Figure S2C), therefore the released fraction of vesicles is not increased in Figure 1G (now Figure 3G). The number of DCV in Figure 3 (now Figure 5) is based on endogenous ChgB staining and not overexpression like the DCV pool measured by NPY-pHluorin. In Figure 3 (now Figure 5) we show that RPH3A KO neurons have slightly more ChgB+ puncta compared to WT.

(12) In Figure 3, the quantification is not clear, discrete puncta are not visible but rather a smear of chromogranin staining. How was it quantified? An independent method to count DCV number, size, and distribution like EM is necessary to support and add further understanding. 

We acknowledge that discrete ChgB puncta are not completely visible in Figure 3 (now Figure 5). Besides the inherent limitation in resolution with confocal imaging, we believe that this is due to ChgB accumulation in the KO neurons, as shown in now Figure 5D. Nonetheless, to address this concern of the reviewer, we have selected other images that represent our dataset (now Figure 5A). Furthermore, the number of ChgB+ DCVs was calculated using SynD software (Schmitz et al., 2011; van de Bospoort et al., 2012) (see previous reply). EM would offer valuable independent confirmation on the total DCV number, size and distribution. However, with the current method we already know that vesicle numbers are at least similar. Does that justify the (major) investment in a quantitative EM study? Moreover, this issue does not affect the central message of the current study.

(13) Can the author discuss if the source of DCVs that are released at the synapse is similar or different from the source of DCVs fused while neurites elongate? 

With our current experimental design, we are unable to draw conclusions regarding this aspect. We are not sure how experiments to identify this source (probably the Golgi?) would be crucial to sustain the central message of our study.

(14) An interesting and related question: what are the expression levels of RPH3A during development and neuronal growth during the nervous system development? 

While we have not specifically examined the expression levels of RPH3A over development, public databases show that RPH3A expression increases over time in mice, consistent with other synaptic proteins (Blake et al., 2021; Baldarelli et al., 2021; Krupke et al., 2017). We have now added this to the revised manuscript (p 2, line 55-56).

(15) The conclusion from Figure 4 about the contribution of SNAP25 interaction to RPH3A inhibitory effect is not convincing. The data are scattered and in many neurons, high levels of fusion events were detected. Further or independent experiments are needed to support this conclusion. For example, is the interaction with SNAP25 important for its inhibitory activity in other DCV-releasing systems like adrenal medulla chromaffin cells? 

We agree that further studies in other DCV-releasing systems like chromaffin cells would provide valuable insight into the role of SNAP25 interaction in RPH3A’s inhibitory effect on exocytosis. However, we believe that starting new series of experiments in another model system is outside of the scope of our current study.

(16) Furthermore, the number of DCVs in the KO is similar in this experiment, raising some more questions about the quantification of the number of vesicles, that differ, in different sections of the manuscript (points # 10,11). 

The total DCV pool in the fusion experiments is measured by overexpression NPY-pHluorin, this cannot be directly compared to the number of endogenous ChgB+ DCV in Figure 3 (now Figure 5), see also item (11)

(17) The statement - "RPH3A is the only negative regulator of DCV" is not completely accurate as other DCV inhibitors like tomosyn were described before. 

We agree. By this statement, we intend to convey that RPH3A is the only negative regulator of DCVs without substantial impact on synaptic vesicle exocytosis, unlike Tomosyns. We have clarified this in the revised text, (p 15, line 366-367).

(18) The support for the effect of KO on the "clustering of DCVs" is not convincing. 

The intensity of endogenous ChgB puncta was decreased in RPH3A KO neurons (now Figure 5E). However, the peak intensity induced by single NPY-pHluorin labeled DCV fusion events (quanta) was unchanged (now Figure S2I). This indicates that the decrease in ChgB puncta intensity must be due to a reduced number of DCVs (quanta) in this specific location. We have interpreted that as ‘clustering’, or maybe ‘accumulation’. However, we only put forward this possibility. We are now more careful in our speculations within the text, (p 11 line 271-277).

(19) Final sentence: "where RPH3A binds available SNAP25, consequently restricting the assembly of SNARE complexes" should be either demonstrated or rephrased as no effect of trans or general SNARE complex formation is shown. 

We agree. We have made the necessary adjustments in the text, (p 15, line 387-389).   

(20) A scheme summarizing RPH3A's interaction with synaptic proteins and its effects on DCVs release, maybe even versus its effects on SVs release, should be considered as a figure or graphic abstract. 

We have included a working model in Figure 7.  

(21) Figure 4 logically should come after Figure 2 to summarize the fusion-related chapter before moving to neurite elongation. 

We have placed Figure 4 after Figure 2 (now Figure 3).

Reviewer #3 (Recommendations For The Authors): 

One important finding of this study is that RPH3A downregulates neuron size, possibly by inhibiting DCV release. Additionally, the authors demonstrated that the number of DCVs is directly proportional to the number of DCVs per µm2, and that RPH3A KO reduces DCV clustering. This conclusion was drawn by comparing ChgB with NPY-pHluorin loading of the DCVs. However, this comparison is not valid as ChgB is expressed at an endogenous level and NPY-pHluorin is over-expressed. In the KO situation where DCV exocytosis is enhanced, the available endogenous ChgB may be depleted faster than the overexpressed NPY-pHluorin. Hoogstraaten et al. should either perform a study in which ChgB is overexpressed to test whether the difference in DCV remains or at least provides an alternative interpretation of their data. 

We thank the reviewer for this comment. The reviewer challenges one or two conclusions in our original manuscript (It is not entirely clear to what exactly “This conclusion” refers): (a) “the number of DCVs is directly proportional to the number of DCVs per µm2”, and (b) “that RPH3A KO reduces DCV clustering”. The reviewer probably means that the number of DCVs per neuron is directly proportional to size of the neuron (a) and states this (these) conclusion(s) are “not valid as ChgB is expressed at an endogenous level and NPY-pHluorin is over-expressed” because “endogenous ChgB may be depleted faster than the overexpressed NPY-pHluorin”. We have three arguments to conclude that faster depletion of ChgB cannot affect these two conclusions: (1) DCVs bud off from the Golgi with newly synthesized (fresh) ChgB. Whether or not a larger fraction of DCVs is released does not influence this initial ChgB loading into DCVs (together with over-expressed NPY-pHluorin); (2) in hippocampal neurons merely 1-6% of the total DCV pool undergoes exocytosis (the current study and also extensively demonstrated in Persoon et al., 2018). RPH3A KO neurons release few percent more of the total DCV pool. Hence, “depletion of ChgB” is only marginally different between experimental groups; and (c) the proposed experiment overexpressing ChgB will not help scrutinize our current conclusions as ChgB overexpression is known to affect DCV biogenesis and the total DCV pool, most likely much more than a few percent more release by RPH3A deficiency.

Hoogstraaten et al. conducted a thorough analysis of the impact of RPH3A KO and its rescue using various mutants on dendrite and axon length (see Supplementary Figure 3). However, they did not test the effect of the ΔSNAP25 mutant. The authors demonstrated that this mutant is the least efficient in rescuing DCV exocytosis (Figure 4E). Hence the neurons expressing this mutant should have a similar size to the KO neurons. This finding would strongly support the argument that DCV exocytosis regulates neuron size. Otherwise, it would suggest that RPH3A may have a function in regulating exocytosis at the growth cones that is independent of SNAP25. Since the authors most probably have the data that allows them to measure the neuron size (acquired for Supplementary Figure 2), I suggest that they perform the required analysis. 

We agree this is important and performed new experiments to determine the dendrite length of RPH3A WT, KO and KO neurons expressing the ΔSNAP25 mutant. We observed that the dendrite length of RPH3A KO neurons expressing ΔSNAP25 mutant is indeed similar to KO neurons (new Figure S3C). Although not significant we observe a clear trend towards bigger neurons compared to WT.  This strengthens our conclusion that increased DCV exocytosis contributes to the observed increased neuronal size.

The authors displayed the result of DCV exocytosis in two ways. One is by showing the number of exocytosis events the other is to display the proportion of DCVs that were secreted. They do the latter by dividing the secreted DCV by the total number of DCVs. These are visualized at the end of the experiment through NH4+ application. While this method works well for synaptic secretion as the marker of SV is localized to the SV membrane and remains at the synapse upon SV exocytosis, it cannot be applied in the same manner when it is the DCV content that is labeled as it is released upon secretion. Hence, the total pool of vesicles should be the number of DCV counted upon NH4+ application in addition to those that are secreted. This way of analyzing the total pool of DCV might also explain the difference in this pool size between KO neurons stimulated two times with 8 stimuli instead of one time with 16 stimuli (Sup Fig 2 C and D). This is an important point as it affects the conclusions drawn from Figure 2. 

We thank the reviewed for this comment. We agree, and we have made the necessary adjustments throughout the manuscript. 

The kymogram of DCV exocytic events displayed in Figure 2D shows a majority of persistent (>20s long) events. This is strange as NPY-pHluori corresponds to the released cargo. Previous work using the same labeling and stimulation technique showed that content release occurs in less than 10s (Baginska et al. 2023). The authors should comment on that difference. 

In Baginska et al. (2023), the authors distinguished between persistent and transient events. The transient events are shorter than 10s for the 2x8 and 16x stimulation paradigms, whereas persistent events can last for more than 10s. In our study we did not make this distinction. However, in response to this reviewer, we have now quantified the fusion duration per cell. These new data show that the mean duration is similar between genotypes for both stimulation paradigms. We have added these new data (new Figure S2D-F).

In Figures 1D and E, some puncta in the kymogram appeared to persist after bleaching. This raises questions about the effectiveness of the bleaching procedure for the FRAP experiment. 

The reviewer is correct that NPY-pHluorin in Figure 1E (now Figure 2C) is not fully bleached. NPY-pHluorin was more resistant to bleaching than NPY-mCherry. However, we merely bleached the neurites to facilitate our analysis by reducing fluorescence of the stationary puncta without causing phototoxicity. Some remaining fluorescence after bleaching does not affect our conclusions in any way.

In the discussion, the paragraph titled "RPH3A does not travel with DCVs in hippocampal neurons" is quite confusing and would benefit from a streamlined explanation. 

We thank the reviewed for this comment. We made the necessary adjustments to make this paragraph clearer, (p 14, line 339-351).

First paragraph of page 8 "TeNT expression in KO neurons restored neurite length to WT levels. When compared to KO neurons without TeNT, neurite length was not significantly decreased but displayed a trend towards WT levels (Figure 3G, H)." These two sentences are confusing as they seem contradictory. 

We agree that this conclusion has been too strong. However, we do not see a contradiction. The significant effect between KO and control neurons on both axon and dendrite length is lost upon TeNT expression (which forms the basis for our conclusions cited by the reviewer, now Figure 6B, C). While the difference between KO neurons +/- TeNT did not reach statistical significance. The (strong) trend is clearly in the same direction. We have refined our original conclusion in the revised manuscript, (p 12, line 304-306).

The data availability statement is missing. 

We have added the data availability statement, (p 21, line 571-572).

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

O'Leary and colleagues present data identifying several procedures that alter discrimination between novel and familiar objects, including time, environmental enrichment, Rac-1, context reexposure, and brief reminders of the familiar object. This is complimented with an engram approach to quantify cells that are active during learning to examine how their activation is impacted following each of the above procedures at test. With this behavioral data, authors apply a modeling approach to understand the factors that contribute to good and poor object memory recall.

We thank the Reviewer for summarizing the scope and depth of our manuscript, and indeed for recognizing our efforts. We engage below with the Reviewer’s specific criticisms.

Strengths:

Authors systematically test several factors that contribute to poor discrimination between novel and familiar objects. These results are extremely interesting and outline essential boundaries of incidental, nonaversive memory.<br /> These results are further supported by engram-focused approaches to examine engram cells that are reactivated in states with poor and good object recognition recall.

We thank the Reviewer for these positive comments.

Weaknesses:

For the environmental enrichment, authors seem to suggest objects in the homecage are similar to (or reminiscent of) the familiar object. Thus, the effect of improved memory may not be related to enrichment per se as much as it may be related to the preservation of an object's memory through multiple retrievals, not the enriching experiences of the environment itself. This would be consistent with the brief retrieval figure. Authors should include a more thorough discussion of this.

This is one of the main issues highlighted by the Editor and the Reviewers. We agree that these results dove-tail with the reminder experiments. We have included additional discussion, see line 510-546.

Authors should justify the marginally increased number of engram cells in the non-enrichment group that did not show object discrimination at test, especially relative to other figures. More specific cell counting criteria may be helpful for this. For example, was the DG region counted for engram and cfos cells or only a subsection?

There was a marginal, but non-significant increase in the number of labelled cells within the standard housed mice in Figure 3f. The cell counting criteria was the same across experimental groups and conditions, where the entire dorsal and ventral blade of the dorsal DG was counted for each animal. This non-statistically significant variance may be due to surgical and viral spread difference between mice. We have clarified this in the manuscript, see line 229-232.

It is unclear why the authors chose a reactivation time point of 1hr prior to testing. While this may be outside of the effective time window for pharmacological interference with reconsolidation for most compounds, it is not necessarily outside of the structural and functional neuronal changes accompanied by reconsolidation-related manipulations.

A control experiment was performed to demonstrated that a brief reminder exposure of 5 mins on its own was insufficient to induce new learning that formed a lasting memory (Supplementary Figure S4a). Mice given only a brief acquisition period of 5 mins, exhibited no preference for the novel object when tested 1 hour after training, suggesting the absence of a lasting object memory (Supplementary Figure S4b & c). We therefore used the 1-hour time point for the brief reminder experiment in Figure 4a. We have clarified this within the manuscript and supplementary data see line 258-264.

Figure 5: Levels of exploration at test are inconsistent between manipulations. This is problematic, as context-only reexposures seem to increase exploration for objects overall in a manner that I'm unsure resembles 'forgetting'. Instead, cross-group comparisons would likely reveal increased exploration time for familiar and novel objects. While I understand 'forgetting' may be accompanied by greater exploration towards objects, this is inconsistent across and within the same figure. Further, this effect is within the period of time that rodents should show intact recognition. Instead, context-only exposures may form a competing (empty context) memory for the familiar object in that particular context.

The Reviewer raises an important question, and we agree with the Reviewer that there should be caution and qualification around interpreting these results as “forgetting”. Indeed, for the context-only rexposures, cross-group comparisons show increased exploration time for familiar and novel objects. As the mice exhibit relatively high exploration of both the novel and familiar objects. An alternative explanation would be that the mice have not truly forgotten the familiar object, but rather as the mouse has not seen the familiar object in the last 6 context only sessions, its reappearance makes it somewhat novel again. Therefore, this change in the object’s reappearance triggers the animal’s curiosity, and in turn drives exploration by the animal. In addition, the context-only exposures may form a competing memory for the familiar object in that particular context. We have highlighted this in the results and also included greater discussion. See lines 306-315.

I am concerned at the interpretation that a memory is 'forgotten' across figures, especially considering the brief reminder experiments. Typically, if a reminder session can trigger the original memory or there is rapid reacquisition, then this implies there is some savings for the original content of the memory. For instance, multiple context retrievals in the absence of an object reminder may be more consistent with procedures that create a distinct memory and subsequently recruit a distinct engram.

These findings raise an important question regarding the interpretation of ‘forgetting’. If a reminder trial or experience can trigger the original memory, or there is rapid reacquisition, then this would suggest there is a degree of savings for the original memory content (85, 86). Previous work has emphasized retrieval deficits as a key characteristic of memory impairment, supporting the idea that memory recall or accessibility may be driven by learning feedback from the environment (7, 8, 14–18). Within our behavioral paradigm, a lack of memory expression would still constitute forgetting due to the loss of learned behavioral response in the presence of natural retrieval cues. The changes in memory expression may therefore underlie the adaptive nature of forgetting. This is consistent with the idea that the engram is intact and available, but not accessible. Here we studied natural forgetting, and our data showing memory retrieval following optogenetic reactivation demonstrates that the original engram persists at a cellular level, otherwise activation of those cells would no longer trigger memory recall. We also agree with the reviewer that multiple context retrievals may indeed lead to the formation of a second distinct engram that competes with the original. Recent work suggests that retroactive interference emerges from the interplay of multiple engrams competing for accessibility (18). We have added clarification and included extra discission of this interpretation. See lines 589-598.

Authors state that spine density decreases over time. While that may be generally true, there is no evidence that mature mushroom spines are altered or that this is consistent across figures. Additionally, it's unclear if spine volume is consistently reduced in reactivated and non-reactivated engram cells across groups. This would provide evidence that there is a functionally distinct aspect of engram cells that is altered consistently in procedures resulting in poor recognition memory (e.g. increased spine density relative to spine density of non-reactivated engram cells and non-engram cells)

We thank the Reviewer for their helpful comments on explaining our engram dendritic spine data. We agree with the Reviewer that an analysis of the changes in spine type, as well as the difference between engram and non-engram spines as well and reactivation and non-reactivated engram spines would be interesting and may help to further illuminate the morphological changes of forgetting and memory retrieval. Indeed, future analysis could determine if spine density is reduced in reactivated and non-reactivated engram cells or indeed across engram non-engram cells within different learning conditions. This avenue of investigation could determine if there is a functionally distinct aspect of engram cells that are altered following forgetting (67). However, such analysis is beyond the scope of this study. We have highlighted this limitation and included its discussion. See lines 493-499.

Authors should discuss how the enrichment-neurogenesis results here are compatible with other neurogenesis work that supports forgetting.

We validated the effectiveness of the enrichment paradigm to enhance neural plasticity by measuring adult hippocampal neurogenesis. The hippocampus has been identified as one of the only regions where postnatal neurogenesis continues throughout life (75). Levels of adult hippocampal neurogenesis do not remain constant throughout life and can be altered by experience (41–43, 57).  In addition, adult born neurons have been shown to contribute to the process of forgetting (74, 78, 79). Although the contribution of adult born neurons to cognition and the memory engram is not fully understood (80, 81). Mishra et al, showed that immature neurons were actively recruited into the engram following a hippocampal-dependent task (67). Moreover, increasing the level of neurogenesis rescued memory deficits by restoring engram activity (67). Augmenting neurogenesis further rescued the deficits in spine density in both immature and mature engram neurons in a mouse model of Alzheimer’s disease (67). Whether neurogenesis alters spine density on differentially for reactivated or non-reactivation engrams cells remains to be investigated (67, 68). This avenue of research would help to illuminate the morphological changes following forgetting and provide evidence if there is a functionally distinct aspect of engram cells that is altered in forgetting (67, 68). Our engram labelling strategy which utilized c-fos-tTA transgenic mice combined with an AAV9-TRE-ChR2-eYFP virus does not necessarily label sufficient immature neurons. Future work could utilize a different engram preparation, such as a genetic labelling strategy (TRAP2) or using a different immediate early gene promoter such as Arc to investigate the contribution of new-born neurons to the engram ensemble. We have added additional discussion of how our work fits with previous literature investigating neurogenesis and forgetting. See lines 547-565.

Reviewer #2 (Public Review):

Summary:

The manuscript examines an important question about how an inaccessible, natural forgotten memory can be retrieved through engram ensemble reactivation. It uses a variety of strategies including optogenetics, behavioral and pharmacological interventions to modulate engram accessibility. The data characterize the time course of natural forgetting using an object recognition task, in which animals can retrieve 1 day and 1 week after learning, but not 2 weeks later. Forgetting is correlated with lower levels of cell reactivation (c-fos expression during learning compared to retrieval) and reduction in spine density and volume in the engram cells. Artificial activation of the original engram was sufficient to induce recall of the forgotten object memory while artificial inhibition of the engram cells precluded memory retrieval. Mice housed in an enriched environment had a slower rate of forgetting, and a brief reminder before the retrieval session promoted retrieval of a forgotten memory. Repeated reintroduction to the training context in the absence of objects accelerated forgetting. Additionally, activation of Rac1-mediated plasticity mechanisms enhanced forgetting, while its inhibition prolonged memory retrieval. The authors also reproduce the behavioral findings using a computational model inspired by Rescorla-Wagner model. In essence, the model proposes that forgetting is a form of adaptive learning that can be updated based on prediction error rules in which engram relevancy is altered in response to environmental feedback.

We thank the Reviewer for summarizing the scope and depth of our manuscript, and for recognizing our efforts. We engage below the Reviewer’s specific criticisms of our interpretations.

Strengths:

(1) The data presented in the current paper are consistent with the authors claim that seemingly forgotten engrams sometimes remain accessible. This suggests that retrieval deficits can lead to memory impairments rather than a loss of the original engram (at least in some cases).

We thank the Reviewer for their positive summary.

(2) The experimental procedures and statistics are appropriate, and the behavioral effects appear to be very robust. Several key effects are replicated multiple times in the manuscript.

We thank the Reviewer for their positive comments.

Weaknesses:

(1) My major issue with the paper is the forgetting model proposed in Figure 7. Prior work has shown that neutral stimuli become associated in a manner similar to conditioned and unconditioned stimuli. As a result, the Rescorla-Wagner model can be used to describe this learning (Todd & Homes, 2022). In the current experiments, the neutral context will become associated with the unpredicted objects during training (due to a positive prediction error). Consequently, the context will activate a memory for the objects during the test, which should facilitate performance. Conversely, any manipulation that degrades the association between the context and object should disrupt performance. An example of this can be found in Figure 5A. Exposing the mice to the context in the absence of the objects should violate their expectations and create a negative prediction error. According to the Rescorla-Wagner model, this error will create an inhibitory association between the context and the objects, which should make it harder for the former to activate a memory of the latter (Rescorla & Wagner, 1972). As a result, performance should be impaired, and this is what the authors find. However, if the cells encoding the context and objects were inhibited during the context-alone sessions (Figure 5D) then no prediction error should occur, and inhibitory associations would not be formed. As a result, performance should be intact, which is what the authors observe.

What about forgetting of the objects that occurs over time? Bouton and others have demonstrated that retrieval failure is often due to contextual changes that occur with the passage of time (Bouton, 1993; Rosas & Bouton, 1997, Bouton, Nelson & Rosas, 1999). That is, both internal (e.g. state of the animal) and external (e.g. testing room, chambers, experimenter) contextual cues change over time. This shift makes it difficult for the context to activate memories with which it was once associated (in the current paper, objects). To overcome this deficit, one can simply re-expose animals to the original context, which facilitates memory retrieval (Bouton, 1993). In Figure 2D, the authors do something similar. They activate the engram cells encoding the original context and objects, which enhances retrieval.

Therefore, the forgetting effects presented in the current paper can be explained by changes in the context and the associations it has formed with the objects (excitatory or inhibitory). The results are perfectly predicted by the Rescorla-Wagner model and the context-change findings of Bouton and others. As a result, the authors do not need to propose the existence of a new "forgetting" variable that is driven by negative prediction errors. This does not add anything novel to the paper as it is not necessary to explain the data (Figures 7 and 8).

We thank the reviewer for clearly explaining their concern about our model. We are very sorry that we did not sufficiently explain that our model is, in fact, based on the classic Rescorla-Wagner model. The key equation of the model that updates “engram strength”  is equivalent to the canonical Rescorla-Wagner model that is commonly used in research on reinforcement learning and decision-making (105). One potential minor difference is that we crucially assume different learning rates for positive and negative prediction errors. However, this variant of the Rescorla-Wagner model is common in the computational literature and is generally not regarded as a qualitatively different kind of model. In fact, it allows us to capture that establishing an object-context association (after a positive prediction error) is faster than the forgetting process (through negative errors).

The other equations that are explained in detail in the Methods are necessary to simulate exploration behavior and render the model suitable for model fitting. Concerning exploration behavior, we use the softmax function, which is commonly used in combination with the Rescorla-Wager model, in order to translate the learned quantity (in our case, engram strength) into behavior (here exploration). The other equations are necessary to fit the model to the data (learning rate α and behavioral variability in exploration behavior).

Therefore, we fully agree with the reviewer that the Rescorla-Wagner can explain our empirical results, in particular by assuming that the different manipulations affect the strength of object-context associations, which, in turn, governs forgetting as behaviorally observed. 

In our previous version of the manuscript, we only referred to the Rescorla-Wagner model directly in the Methods. But to make this important point clearer, we now refer to the origin of the model multiple times in the Results section as well. See lines 81, 386-393.

We also agree with the reviewer that the learning/forgetting process can be described in terms of changes in object-context associations (e.g., inhibitory associations after a negative prediction error). Therefore, we now explicitly refer to the relationship between updated object-context associations and forgetting and highlight that we believe that stronger associations signal higher engram “relevancy”. See lines 386-393.

We have extended Figure 7 (new panels a and b), where we illustrate the idea that (a) object-context associations govern forgetting and (b) show the key Rescorla-Wagner equation, including a simple explanation of the main terms (engram strength, prediction error, and learning rate). Finally, we have also extended our discussion of the model, where we now directly state that the Rescorla-Wagner model captures the key results of our experiments. See lines 573-580.

In order to further support a link between our empirical data and computational modeling, we also added extra experiments that showed the modulation of engram cells within the dentate gyrus can regulate these object-context associations. See Supplementary Figure 12a-f and lines 401-404.

To summarize our reply, we agree with the reviewer’s comment and hope that we have clarified the direct relationship to the Rescorla-Wagner model.

(2) I also have an issue with the conclusions drawn from the enriched environment experiment (Figure 3). The authors hypothesize that this manipulation alleviates forgetting because "Experiencing extra toys and objects during environmental enrichment that are reminiscent of the previously learned familiar object might help maintain or nudge mice to infer a higher engram relevancy that is more robust against forgetting.". This statement is completely speculative. A much simpler explanation (based on the existing literature) is that enrichment enhances synaptic plasticity, spine growth, etc., which in turn reduces forgetting. If the authors want to make their claim, then they need to test it experimentally. For example, the enriched environment could be filled with objects that are similar or dissimilar to those used in the memory experiments. If their hypothesis is correct, only the similar condition should prevent forgetting.

We thank the Reviewer for this alternative perspective on our findings. First of all, we agree that this statement is speculative. The effects of enrichment on neural plasticity are well established and it likely contributes to the enhanced memory recall. It is important to emphasize that this process of updating is not necessarily separate from enrichment-induced plasticity at an implementational level, but part of the learning experience within an environment containing multiple objects. The enrichment or, more generally, experience, may therefore enhance memory through the modification of activity of specific engram ensembles. The idea of enrichment facilitating memory updating is consistent with the results obtained by the reminder experiments and further supported by our analysis with the Rescorla-Wagner computational model, where experience updates the accessibility of existing memories, possibly through reactivation of the original engram ensemble.

We would like to further clarify that our explanation concerns the algorithmic level, in contrast to the neural level. Based on the computational analyses using the Rescorla-Wagner model and in line with the reviewer’s previous comment on the model, we believe that forgetting is governed by the strength of object-context associations (or engram relevancy). Our interpretation is that stronger associations signal that the memory or engram representation is important ("relevant") and should not be forgotten. Accordingly, due to a vast majority of experiences with extra cage objects in the enriched environment, mice might generally learn that such objects are common in their environment and potentially relevant in the future (i.e., the object-context association is strong, preventing forgetting). Our speculation of these results is to help unify our empirical data with the computational model.

We believe that the Reviewer's alternative explanation in terms of synaptic plasticity, spine growth is not mutually exclusive with the modelling work. It is possible that the computational mechanisms that we explore based on the Rescorla-Wagner model are neuronally related to the biological mechanisms that the reviewer suggests at the implementational level. Therefore, ultimately, the two perspectives might even complement each other. We have included additional discussion to clarify this point. See lines 510-546.

(3) It is well-known that updating can both weaken or strengthen memory. The authors suggest that memory is updated when animals are exposed to the context in the absence of the objects. If the engram is artificially inhibited (opto) during context-only re-exposures, memory cannot be updated. To further support this updating idea, it would be good to run experiments that investigate whether multiple short re-exposures to the training context (in the presence of the objects or during optogenetic activation of the engram) could prevent forgetting. It would also be good to know the levels of neuronal reactivation during multiple re-exposures to the context in the absence versus context in the presence of the objects.

We thank the Reviewer for their comments. We agree that additional experiments would be helpful to further support the idea of updating. We have performed additional experiments to test the idea that multiple short re-exposures to the training context, in the presence of objects prevents forgetting. In this paradigm, mice were repeatedly exposed to the original object pair (Supplementary Figure S5a). The results indicate that repeated reminder trials facilitate object memory recall (Supplementary Figure 5b&c). These data indicated that subsequent object reminders over time facilitates the transition of a forgotten memory to an accessible memory. See Supplementary Figure S5 and Lines 279-287.

(4) There are a number of studies that show boundary conditions for memory destabilization/reconsolidation. Is there any evidence that similar boundary conditions exist to make an inaccessible engram accessible?

The Reviewer asks an interesting question about boundary conditions and engram accessibility. Boundary conditions could indeed affect the degree of destabilization and reconsolidation, the salience or strength of the memory, as well as the timing of retrieval cues. Future models could focus on understanding the specific boundary conditions in which a memory becomes retrievable and the degree to which it is sufficiently destabilized and liable for updating and forgetting. We have included additional discussion on the potential role of boundary conditions for engram accessibility. See lines 661-666.

(5) More details about how the quantification of immunohistochemistry (c-fos, BrdU, DAPI) was performed should be provided (which software and parameters were used to consider a fos positive neurons, for example).

We have added additional information for the parameters of quantification of immunohistochemistry. See lines 796-809.

(6) Duration of the enrichment environment was not detailed.

We have highlighted the details for the environmental enrichment duration. See lines 756.

Reviewer #3 (Public Review):

Summary:

The manuscript by Ryan and colleagues uses a well-established object recognition task to examine memory retrieval and forgetting. They show that memory retrieval requires activation of the acquisition engram in the dentate gyrus and failure to do so leads to forgetting. Using a variety of clever behavioural methods, the authors show that memories can be maintained and retrieval slowed when animals are reared in environmental enrichment and that normally retrieved memories can be forgotten if exposed to the environment in which the expected objects are no longer presented. Using a series of neural methods, the authors also show that activation or inhibition of the acquisition engram is key to memory expression and that forgetting is due to Rac1.

We thank the Reviewer for summarizing the scope and depth of our manuscript, and indeed for recognizing our efforts. We engage below the Reviewer’s specific criticisms of our interpretations.

Strengths:

This is an exemplary examination of different conditions that affect successful retrieval vs forgetting of object memory. Furthermore, the computational modelling that captures in a formal way how certain parameters may influence memory provides an important and testable approach to understanding forgetting.

The use of the Rescorla-Wagner model in the context of object recognition and the idea of relevance being captured in negative prediction error are novel (but see below).

The use of gain and loss of function approaches are a considerable strength and the dissociable effects on behaviour eliminate the possibility of extraneous variables such as light artifacts as potential explanations for the effects.

We thank the Reviewer for their positive comments.

Weaknesses:

Knowing what process (object retrieval vs familiarity) governed the behavioural effect in the present investigation would have been of even greater significance.

The Reviewer touches on an important issue of the object recognition task. Understanding how experience alters object familiarity versus object retrieval and its impact on learning would help to develop better models of object memory and forgetting. We have added additional discussion. See lines 666-669.

The impact of the paper is somewhat limited by the use of only one sex.

We agree that using only male mice limits the impact of the paper. Indeed, the field of behavioural neuroscience is moving to include sex as a variable. Future experiments should include both male and female mice.

While relevance is an interesting concept that has been operationalized in the paper, it is unclear how distinct it is from extinction. Specifically, in the case where the animals are exposed to the context in the absence of the object, the paper currently expresses this as a process of relevance - the objects are no longer relevant in that context. Another way to think about this is in terms of extinction - the association between the context and the objects is reduced results in a disrupted ability of the context to activate the object engram.

We thank the reviewer for their insightful comment on the connection between engram relevance and memory extinction. Lacagnina et al., demonstrated that extinction training suppressed the reactivation of a fear engram, while activating a second putative extinction ensemble (59). In another study, these extinction engram cells and reward cells were shown to be functionally interchangeable (92). Moreover, in a study conducted by Lay et al., the balance between extinction and acquisition was disrupted by inhibiting the extinction recruited neurons in the BLA and CN (93). These results suggested that decision making after extinction can be governed by a balance between acquisition and extinction specific ensembles (93). Together, this may suggest that in the present study, when mice are repeatedly exposed to the training context, the association between the context and the objects is reduced, resulting in a disrupted ability of the context to activate the object engram. Therefore, memory relevance and extinction may operate similarly to effect engram accessibility, and in essence ‘forgetting’ of object memories may be due to neurobiological mechanisms similar to that of extinction learning (4). We have included additional discussion on the link between our results and the extinction literature. See lines 642-654.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Additional measures that may help interpretation of and clarify data are:

A minute-by-minute analysis for training and testing may provide insight about the learning rate and testing temporal dynamics that may shed light substantially on differential levels of exploration. This should be applied across figures and would support conclusions from models in Figures 7-8 as well.

Locomotion/distance travelled measures.

We have included additional analysis for a minute-by-minute analysis of training and testing of the object memory test at 24 hr, 2 weeks as well as under the standard housing and enrichment conditions. The results further support the initial finding that novel object recognition is increased in mice that recall the object at 24 hr. Similarly, mice housed in the enriched housing initially explore the novel object more compared to the familiar object. See Supplementary Figure 1 and 2, as well as lines 103-105 and 211-213.

The appropriate control for the context exposure figure would be to expose to a novel context in one group and the acquisition/testing context for the other.

We agree with the reviewer that an additional control of a novel context would further support our findings. Indeed, this line of investigate may dove-tail with the other reviewer comments on the role of competing engrams and interference. Future work could investigate the degree to which novel contexts and multiple memories can affect the rate of forgetting through engram updating. We have included additional discussion. See lines 643 and 655. However, in our experience it is necessary to pre-expose mice to different contexts before object exposure (e.g. Autore et al ’23), in order to form discriminate object/context associations. Establishing such a paradigm for this study would be at odds with the established paradigms and schedules in this current study. Moreover, the possibility that the effect of object displacement on forgetting requires the familiar context, or not, does not impact the main conclusions of this study. However, we agree that it is a point for expansion in the future.

A control virus+light group vs simply a no-light condition.

For optogenetic experiments. Control mice underwent the same surgery procedure with virus and optic fibre implantation. However, no light was delivered to excite or inhibit the respective opsin. Previous papers have shown laser light delivered to tissue expressing an AAV-TRE-EYFP lacking an light-opsin does cause cellular excitation. We have clarified this in the text. See lines 726-729.

Reviewer #2 (Recommendations For The Authors):

Minor details:

(1) In the pharmacological modification of Rac 1, please specify what percentage of DMSO was used to dissolve Rac1 inhibitor and correct the typo 'DSMO'

Rac1 inhibitor (Ehop016) was reconstituted and prepared in PBS with 1% Tween-80, 1% DMSO and 30% PEG. We have clarified this in the text and corrected the typo, thank you. See lines 767.

(2) In the penultimate paragraph there is a typo 'predication error'

This is now corrected. Thankyou.

Reviewer #3 (Recommendations For The Authors):

I was unable to find information on what the No Light group consisted of. Was there a control virus infused, were the animals implanted with optical fibres (in the presence or absence of a virus), were they surgical controls, etc?

For optogenetic experiments. No Light Control mice underwent the same surgery procedure with virus and optic fibre implantation. However, no light was delivered to excite or inhibit the respective opsin. We have clarified this in the text. See lines 726-729.

The discussion lacked specificity in places. For example, the idea of eluding to 'other variables' is somewhat vague (p. 21, middle paragraph). Some examples of what other variables could be relevant would be helpful in capturing what direction or relevance the model may have going forward.

We have expanded the discussion of other variables which might impact engram relevance and how the model might be developed moving forward. These may include, boundary conditions of destabilization and reconsolidation, the salience or strength of the memory as well as the timing of retrieval cues or updating experience. Future models could focus on understanding the specific boundary conditions in which a memory becomes retrievable and the degree to which it is sufficiently destabilized and liable for updating and forgetting. The role of perceptual learning on memory retrieval and forgetting may also be an avenue of future investigation. Understanding how experience alters object familiarity versus object retrieval and its impact on learning would also help to develop better models of object memory and forgetting. In the current study, only male mice were utilized. Therefore, future work could also include sex as a variable to fully elucidate the impact of experience on the processes of forgetting. See lines 642-669.

In the same paragraph (p. 21, middle paragraph) there is mention of multiple engrams and how they can compete. The authors reference Autore et al (2023), but I thought Lacagina did this really beautifully also in an experimental setting. This idea is also expressed in Lay et al. (2022). So additional references would further strengthen the authors argument here.

We thank the reviewer for the additional references for discussing engram competition. We have included these papers in the discission. See lines 642-654.

Relatedly, environmental enrichment was considered in terms of object relevance. I wonder if the authors may want to consider thinking about their results in terms of effects on perceptual learning.

Indeed, perceptual learning maybe playing a role in environmental enrichment. We have included additional discussion. See lines 666-669.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

The authors present a model for multisensory correlation detection that is based on the neurobiologically plausible Hassenstein Reichardt detector. It modifies their previously reported model (Parise & Ernst, 2016) in two ways: a bandpass (rather than lowpass) filter is initially applied and the filtered signals are then squared. The study shows that this model can account for synchrony judgement, temporal order judgement, etc in two new data sets (acquired in this study) and a range of previous data sets.

Strengths:

(1) The model goes beyond descriptive models such as cumulative Gaussians for TOJ and differences in cumulative Gaussians for SJ tasks by providing a mechanism that builds on the neurobiologically plausible Hassenstein-Reichardt detector.

(2) This modified model can account for results from two new experiments that focus on the detection of correlated transients and frequency doubling. The model also accounts for several behavioural results from experiments including stochastic sequences of A/V events and sine wave modulations.

Additional thoughts:

(1) The model introduces two changes: bandpass filtering and squaring of the inputs. The authors emphasize that these changes allow the model to focus selectively on transient rather than sustained channels. But shouldn't the two changes be introduced separately? Transients may also be detected for signed signals.

We updated the original model because our new psychophysical evidence demonstrates the fundamental role of unsigned transient for multisensory perception. While the original model received input from sustained unimodal channels (low-pass filters), the new version receives input from unsigned unimodal transient channels. Transient channels are normally modelled through bandpass filters (to remove the DC and high-frequency signal components) and squaring (to remove the sign). While these may appear as two separate changes in the model, they are, in fact, a single one: the substitution of sustained with unsigned transient channels (for a similar approach, see Stigliani et al. 2017, PNAS). Either change alone would not be sufficient to implement a transient channel that accounts for the present results.

That said, we were also concerned with introducing too many changes in the model at once. Indeed, we simply modelled the unimodal transient channels as a single band-pass filter followed by squaring. This is already a stripped-down version of the unsigned transient detectors proposed by Adelson and Bergen in their classic Motion Energy model. The original model consisted of two biphasic temporal filters 90 degrees out of phase (i.e., quadrature filters), whose output is later combined. While a simpler implementation of the transient channels was sufficient in the present study, the full model may be necessary for other classes of stimuli (including speech, Parise, 2024, BiorXiv). Therefore, for completeness, we now include in the Supplementary Information a formal description of the full model, and validate it by simulating our two novel psychophysical studies. See Supplementary Information “The quadrature MCD model” section and Supplementary Figure S8.

(2) Because the model is applied only to rather simple artificial signals, it remains unclear to what extent it can account for AV correlation detection for naturalistic signals. In particular, speech appears to rely on correlation detection of signed signals. Can this modified model account for SJ or TOJ judgments for naturalistic signals?

It can. In a recent series of studies we have demonstrated that a population of spatially-tuned MCD units can account for audiovisual correlation detection for naturalistic stimuli, including speech (e.g. the McGurk Illusion). Once again, unsigned transients were sufficient to replicate a variety of previous findings. We have now extended the discussion to cover this recent research: Parise, C. V. (2024). Spatiotemporal models for multisensory integration. bioRxiv, 2023-12.

Even Nidiffer et al. (2018) which is explicitly modelled by the authors report a significant difference in performance for correlated and anti-correlated signals. This seems to disagree with the results of study 1 reported in the current paper and the model's predictions. How can these contradicting results be explained? If the brain detects correlation on signed and unsigned signals, is a more complex mechanism needed to arbitrate between those two?

We believe the reviewer here refers to our Experiment 2 (where, like Nidiffer at al. (2018) we used periodic stimuli, not Experiment 1, which consists of step stimuli). We were also puzzled by the difference between our Experiment 2 and Nidiffer et al. (2018): we induced frequency doubling, Nidiffer did not. Based on quantitative simulations, we concluded that this difference could be attributed to the fact that while Nidiffer included on each trial an intensity ramp in their periodic audiovisual stimuli, we did not. As a result, when considering the ramp (unlike in Nidiffer’s analyses), all audiovisual signals used by Nidiffer were positively correlated (irrespective of frequency and phase offset), while our signals in Experiment 2 were sometimes correlated and other times not (depending on the phase offset). This important simulation is included in Supplementary Figure S7; we also have now updated the text to better highlight the role of the pedestal in determining the direction of the correlation.

(3) The number of parameters seems quite comparable for the authors' model and descriptive models (e.g. PSF models). This is because time constants require refitting (at least for some experimental data sets) and the correlation values need to be passed through a response mode (i.e. probit function) to account for behavioural data. It remains unclear how the brain adjusts the time constants to different sensory signals.

This is a deep question. For simplicity, here the temporal constants were fitted to the empirical psychometric functions. To avoid overfitting, whenever possible we fitted such parameters over some training datasets, while trying to predict others. However, in some cases, it was necessary to fit the temporal constants to specific datasets. This may suggest that the temporal tuning of those units is not crystalised to some pre-defined values, but is adjusted based on recent perceptual history (e.g., the sequence of trials and stimuli participants are exposed to during the various experiments).

For transparency, here we show how varying the tuning of the temporal constants of the filters affects the goodness of fit of our new psychophysical experiments (Supplementary Figure S8). As it can be readily appreciated, the relative temporal tuning of the unimodal transient detector was critical, though their absolute values could vary over a range of about 15 to over 100ms. The tuning of the low-pass filters of the correlation detector (not shown here) displayed much lower temporal sensitivity over a range between 0.1s to over 1s.

This simulation shows the impact of temporal tuning in our simulations, however, the question remains as to how such a tuning gets selected in the first place. An appealing explanation relies on natural scene statistics: units are temporally tuned to the most common audiovisual stimuli. Although our current empirical evidence does not allow us to quantitatively address this question, in previous simulations (see Parise & Ernst, 2016, Supplementary Figure 8), by analogy with visual motion adaptation, we show how the temporal constants of our model can dynamically adjust and adapt to recent perceptual history. We hope these new and previous simulations address the question about the nature of the temporal tuning of the MCD units.

(4) Fujisaki and Nishida (2005, 2006) proposed mechanisms for AV correlation detection based on the Hassenstein-Reichardt motion detector (though not formalized as a computational model).

This is correct, Fujisaki and Nishida (2005, 2007) also hypothesized that AV synchrony could be detected using a mechanism analogous to motion detection. Interestingly, however, they ruled out such a hypothesis, as their “data do not support the existence of specialized low-level audio-visual synchrony detectors”. Yet, along with our previous work (Parise & Ernst, 2016, where we explicitly modelled the experiments of Fujisaki and Nishida), the present simulations quantitatively demonstrate that a low-level AV synchrony detector is instead sufficient to account for audiovisual synchrony perception and correlation detection. We now credit Fujusaki and Nishida in the modelling section for proposing that AV synchrony can be detected by a cross-correlator.

Finally, we believe the reviewer is referring to the 2005 and 2007 studies of Fujisaki and Nishida (not 2006); here are the full references of the two articles we are referring to:

Fujisaki, W., & Nishida, S. Y. (2005). Temporal frequency characteristics of synchrony–asynchrony discrimination of audio-visual signals. Experimental Brain Research, 166, 455-464.

Fujisaki, W., & Nishida, S. Y. (2007). Feature-based processing of audio-visual synchrony perception revealed by random pulse trains. Vision Research, 47(8), 1075-1093.

Reviewer #2 (Public Review):

Summary:

This is an interesting and well-written manuscript that seeks to detail the performance of two human psychophysical experiments designed to look at the relative contributions of transient and sustained components of a multisensory (i.e., audiovisual) stimulus to their integration. The work is framed within the context of a model previously developed by the authors and is now somewhat revised to better incorporate the experimental findings. The major takeaway from the paper is that transient signals carry the vast majority of the information related to the integration of auditory and visual cues, and that the Multisensory Correlation Detector (MCD) model not only captures the results of the current study but is also highly effective in capturing the results of prior studies focused on temporal and causal judgments.

Strengths:

Overall the experimental design is sound and the analyses are well performed. The extension of the MCD model to better capture transients makes a great deal of sense in the current context, and it is very nice to see the model applied to a variety of previous studies.

Weaknesses:

My one major issue with the paper revolves around its significance. In the context of a temporal task(s), is it in any way surprising that the important information is carried by stimulus transients? Stated a bit differently, isn't all of the important information needed to solve the task embedded in the temporal dimension? I think the authors need to better address this issue to punch up the significance of their work.

In hindsight, it may appear unsurprising that transient signals carry most information for audiovisual integration. Yet, so somewhat unexpectedly, this has never been investigated using perhaps the most diagnostic psychophysical tools for perceived crossmodal timing; namely temporal order and simultaneity judgments–along with carefully designed experiments with quantitative predictions for the effect of either channel. The fact that the results conform to intuitive expectations further supports the value of the present work: grounding empirically with what is intuitively expected. This offers solid psychophysical evidence that one can build on for future advancements. Importantly, developing a model that builds on our new results and uses the same parameters to predict a variety of classic experiments in the field, further supports the current approach.

If “significance” is intended as shaking previous intuitions or theories, then no: this is not a significant contribution. If instead, by significance we intend to build a solid empirical and theoretical ground for future work, then we believe this study is not significant, it is foundational. We hope that this work's significance is better captured in our discussion.

On a side note, there is an intriguing factor around transient vs. sustained channels: what matters is the amount of change, not the absolute stimulus intensity. Previous studies, for example, have suggested a positive cross modal mapping between auditory loudness and visual lightness or brightness [Odegaard et al., 2004]. This study, conversely, challenges this view and demonstrates that what matters for multisensory integration in time is not the intensity of a stimulus, but changes thereof.

In a more minor comment, I think there also needs to be a bit more effort into articulating the biological plausibility/potential instantiations of this sustained versus transient dichotomy. As written, the paper suggests that these are different "channels" in sensory systems, when in reality many neurons (and neural circuits) carry both on the same lines.

The reviewer is right, in our original manuscript we glossed over this aspect. We have now expanded the introduction to discuss their anatomical basis. However, we are not assuming any strict dichotomy between transient and sustained channels; rather, our results and simulations demonstrate that transient information is sufficient to account for audiovisual temporal integration.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

(1) Related to point 2 of the public review, can the authors provide additional results showing that the model can also account for naturalistic signals and more complex stochastic signals?

While working on this manuscript, we were also working in parallel on a project related to audiovisual integration of naturalistic signals. A pre-print is available online [Parise, 2024, BiorXiv], and the related study is now discussed in the conclusions.

(2) As noted in the public review, Fujisaki and Nishida (2005, 2006) already proposed mechanisms for AV correlation detection based on the Hassenstein-Reichardt motion detector. Their work should be referenced and discussed.

We have now acknowledged the contribution of Fujisaki and Nishida in the modelling section, when we first introduce the link between our model and the Hassenstein-Reichardt detectors.

(3) Experimental parameters: Was the phase shift manipulated in blocks? If yes, what about temporal recalibration?

To minimise the effect of temporal recalibration, the order of trials in our experiments was randomised. Nonetheless, we can directly assess potential short-term recalibration effects by plotting our psychophysical responses against both the current SOA, and that of the previous trials. The resulting (raw) psychometric surfaces below are averaged across observers (and conditions for Experiment 1). In all our experiments, responses are obviously dependent on the current SOA (x-axis). However, the SOA of the previous trials (y-axis) does not seem to meaningfully affect simultaneity and temporal order judgments. The psychometric curves above the heatmaps represent the average psychometric functions (marginalized over the SOA of the previous trial).

All in all, the present analyses demonstrate negligible temporal recalibration across trials, likely induced by a random sequence of lags or phase shifts. Therefore, when estimating the temporal constants of the model, it seems reasonable to ignore the potential effects of temporal recalibration. To avoid increasing the complexity of the present manuscript, we would prefer not to include the present analyses in the revised version.

Author response image 1.

Effect of previous trial. Psychometric surfaces for Experiments 1 and 2 plotted against the lag in the current vs. the previous trial. While psychophysical responses are strongly modulated by the lag in the last trial (horizontal axis), they are relatively unaffected by the lag in the previous trial (vertical axis).

(4) The model predicts no differences for experiment 1 and this is what is empirically observed. Can the authors support these null results with Bayes factors?

This is a good suggestion: we have now included a Bayesian repeated measures ANOVA to the analyses of Experiment 1. As expected, these analyses provide further, though mild evidence in support for the null hypothesis (See Table S2). For completeness, the new Bayesian analyses are presented alongside the previous frequentist ones in the revised manuscript.

In the story, the robots are essentially slaves to the humans. They are built in made purely to make human's lives easier. I'm beginning to think this could possibly be a cautionary tale about the way we treat "technology" and "robots" and the possible consequences of expecting them to behave like a human but not treating them like one.

<https://purl.org/cite/theatlantic.com/Bush, Vannevar. 1945. As we may think. Atlantic Monthly, (July). Pp. 101-108.>

I referred (indirectly) to this in an annotation on https://www.theatlantic.com/magazine/archive/1945/07/as-we-may-think/303881/ as "the PDF". As the first page indicates this is rather a PDF—specifically someone's PDF of the ACM's reprint from 1996 (which can be found hanging off this DOI: https://dl.acm.org/doi/10.1145/227181.227186).

The Atlantic's PDF can be found here https://cdn.theatlantic.com/media/archives/1945/07/176-1/132407932.pdf (at least for now).

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

Summary:

This important work advances our understanding of sperm motility regulation during fertilization by uncovering the midpiece/mitochondria contraction associated with motility cessation and structural changes in the midpiece actin network as its mode of action. The evidence supporting the conclusion is solid, with rigorous live cell imaging using state-of-the-art microscopy, although more functional analysis of the midpiece/mitochondria contraction would have further strengthened the study. The work will be of broad interest to cell biologists working on the cytoskeleton, mitochondria, cell fusion, and fertilization. Strengths: The authors demonstrate that structural changes in the flagellar midpiece F-actin network are concomitant to midpiece/mitochondrial contraction and motility arrest during sperm-egg fusion by rigorous live cell imaging using state-of-art microscopy.

Response P1.1: We thank the reviewer for her/his positive assessment of our manuscript.

Weaknesses:

Many interesting observations are listed as correlated or in time series but do not necessarily demonstrate the causality and it remains to be further tested whether the sperm undergoing midpiece contraction are those that fertilize or those that are not selected. Further elaboration of the function of the midpiece contraction associated with motility cessation (a major key discovery of the manuscript) would benefit from a more mechanistic study.

Response P1.2: We thank the reviewer for this point. We have toned down some of our statements since some of the observations are indeed temporal correlations. We will explore some of these possible connections in future experiments. In addition, we have now incorporated additional experiments and possible explanations about the function of the midpiece contraction.

Reviewer #2 (Public Review): 

(1) The authors used various microscopy techniques, including super-resolution microscopy, to observe the changes that occur in the midpiece of mouse sperm flagella. Previously, it was shown that actin filaments form a double helix in the midpiece. This study reveals that the structure of these actin filaments changes after the acrosome reaction and before sperm-egg fusion, resulting in a thinner midpiece. Furthermore, by combining midpiece structure observation with calcium imaging, the authors show that changes in intracellular calcium concentrations precede structural changes in the midpiece. The cessation of sperm motility by these changes may be important for fusion with the egg. Elucidation of the structural changes in the midpiece could lead to a better understanding of fertilization and the etiology of male infertility. The conclusions of this manuscript are largely supported by the data, but there are several areas for improvement in data analysis and interpretation. Please see the major points below.

Response P2.1: We thank the reviewer for the positive comments.

(2) It is unclear whether an increased FM4-64 signal in the midpiece precedes the arrest of sperm motility. in or This needs to be clarified to argue that structural changes in the midpiece cause sperm motility arrest. The authors should analyze changes in both motility and FM4-64 signal over time for individual sperm.

Response P2.2 : We have conducted single cell experiments tracking both FM4-64 and motility as the reviewer suggested (Supplementary Fig S1). We have observed that in all cases, cells gradually diminished the beating frequency and increased FM4-64 fluorescence in the midpiece until a complete motility arrest is observed. A representative example is shown in this Figure but we will reinforce this concept in the results section.

(3) It is possible that sperm stop moving because they die. Figure 1G shows that the FM464 signal is increased in the midpiece of immotile sperm, but it is necessary to show that the FM4-64 signal is increased in sperm that are not dead and retain plasma membrane integrity by checking sperm viability with propidium iodide or other means.

Response P2.3: This is a very good point. In our experiments, we always considered sperm that were motile to hypothesize about the relevance of this observation. We have two types of experiments: 

(1) Sperm-egg Fusion: In experiments where sperm and eggs were imaged to observe their fusion, sperm were initially moving and after fusion, the midpiece contraction (increase in FM4-64 fluorescence was observed) indicating that the change in the midpiece (that was observed consistently in all fusing cells analyzed), is part of the process. 

(2) Sperm that underwent acrosomal exocytosis (AE): we have observed two behaviours as shown in Figure 1: 

a) Sperm that underwent AE and they remain motile without midpiece contraction (they are alive for sure); 

b) Sperm that underwent AE and stopped moving with an increase in FM464 fluorescence. We propose that this contraction during AE is not desired because it will impede sperm from moving forward to the fertilization site when they are in the female reproductive tract. In this case, we acknowledge that the cessation of sperm motility may be attributed to cellular death, potentially correlating with the increased FM4-64 signal observed in the midpiece of immotile sperm that have undergone AE. To address this hypothesis, we conducted image-based flow cytometry experiments, which are well-suited for assessing cellular heterogeneity within large populations.

Author response image 1 illustrates the relationship between cell death and spontaneous AE in noncapacitated mouse sperm, where intact acrosomes are marked by EGFP. Cell death was evaluated using Sytox Blue staining, a dye that is impermeable to live cells and shows affinity for DNA. AE was assessed by the absence of EGFP in the acrosome. 

Author response image 1a indicates a lack of correlation between Sytox and EGFP fluorescence. Two populations of sperm with EGFP signals were found (EGFP+ and EGFP-), each showing a broad distribution of Sytox signal, enabling the distinction between cells that retain plasma membrane integrity (live sperm: Sytox-) and those with compromised membranes (dead cells: Sytox+). The observed bimodal distribution of EGFP signal, regardless of live versus dead cell populations, indicates that the fenestration of the plasma membrane known to occur during AE is a regulated process that does not necessarily compromise the overall plasma membrane integrity. 

These observations are reinforced by the single-cell examples in Author response image 1b, where we were able to identify sperm in four categories: live sperm with intact acrosome (EGFP+/Sytox-), live sperm with acrosomal exocytosis (EGFP-/Sytox-), dead sperm with intact acrosome (EGFP+/Sytox+), and dead sperm with AE (EGFP-/Sytox+). Note the case of AE (lacking EGFP signal) which bears an intact plasma membrane (lacking Sytox Blue signal). Author response image 2 shows single-cell examples of the four categories observed with confocal microscopy to reinforce the observations from Author response image 1a.

Author response image 1.

Fi. Image based flow cytometry analysis (ImageStream Merk II), of non-capacitated mouse sperm, showing the distribution of EGFP signal (acrosome integrity) against Sytox Blue staining (cell viability).  (A) The quadrants show: Sytox Blue + / EGFP low (17.6%), Sytox Blue + / EGFP high (40.1%), Sytox Blue - / EGFP high (20.2%), and Sytox Blue - / EGFP low (21.7%). Each quadrant indicates the percentage of the total sperm population exhibiting the corresponding staining pattern. Axes are presented in a log10 scale of arbitrary units of fluorescence.  (B) Representative single-cell images corresponding to the four categorized sperm populations from the flow cytometry analysis in panel (A). The top row displays sperm with compromised plasma membrane integrity (Sytox Blue +), showing low (left) and high (right) EGFP signals. The bottom row shows sperm with intact plasma membrane (Sytox Blue -), displaying high (left) and low (right) EGFP signal. It is worth noting that when analyzing the percentages in (A), we observed that the data also encompass a population of headless flagella, which was present in all observed categories. Therefore, the percentages should be interpreted with caution.

Author response image 2.

Confocal Microscopy Examples of AE and cell viability. The top row features sperm with compromised plasma membrane integrity (Sytox Blue +) and high EGFP expression; the second row displays sperm with compromised membrane and low EGFP expression; the third row illustrates sperm with intact membrane (Sytox Blue -) and high EGFP expression; the bottom row shows sperm with intact membrane and low EGFP expression. 

Author response images 3-5 provide insight into the relationship between FM4-64 and Sytox Blue fluorescence intensities in non-capacitated sperm (CTRL, Author response image 3), capacitated sperm and acrosome exocytosis events stimulated with 100 µM progesterone (PG, Author response image 4), and capacitated sperm stimulated with 20 µM ionomycin (IONO, Author response image 5). Two populations of sperm with Sytox Blue signals were clearly distinguished (Sytox+ and Sytox-), enabling the discernment between live and dead sperm. Interestingly, the upper right panels of Author response images 3A, 4A, and 5A (Sytox Blue+ / FM4-64 high) consistently show a positive correlation between FM4-64 and Sytox Blue. This observation aligns with the concern raised by Reviewer 2, suggesting that compromised membranes due to cell death provide more binding sites for FM4-64. 

Nonetheless, the lower panels of Author response images 3A, 4A and 5A (Sytox Blue-) show no correlation with FM4-64 fluorescence, indicating that this population can exhibit either low or high FM4-64 fluorescence. As expected, in stark contrast with the CTRL case, the stimulation of AE with PG or IONO in capacitated sperm increased the population of live sperm with high FM4-64 fluorescence (Sytox Blue+ / FM4-64 high: CTRL: 7.85%, PG: 8.73%, IONO: 13.5%). 

Single-cell examples are shown in Author response images 3B, 4B, and 5B, where the four categories are represented: dead sperm with low FM4-64 fluorescence (Sytox Blue+ / FM4-64 low), dead sperm with high FM4-64 fluorescence (Sytox Blue+ / FM4-64 high), live sperm with low FM4-64 fluorescence (Sytox Blue- / FM4-64 low), and live sperm with high FM4-64 fluorescence (Sytox Blue- / FM4-64 high). 

Author response image 3.

Relationship between cell death and FM4-64 fluorescence in capacitated sperm without inductor of RA. Image-based flow cytometry analysis of non-capacitated mouse sperm loaded with FM464 and Sytox Blue dyes, with one and two minutes of incubation time, respectively. (A) The quadrants show: Sytox Blue+ / FM4-64 low (13.3%), Sytox Blue+ / FM4-64 high (49.8%), Sytox Blue- / FM4-64 low (28.1%), and Sytox Blue- / FM4-64 high (7.85%). Each quadrant indicates the percentage of the total sperm population exhibiting the corresponding staining pattern. Axes are presented on a log10 scale of arbitrary units of fluorescence. (B) Representative single-cell images corresponding to the four categorized sperm populations from the flow cytometry analysis in panel (A).

Author response image 4.

Relationship between cell death and FM4-64 fluorescence capacitated sperm stimulated with progesterone. Image-based flow cytometry analysis of non-capacitated mouse sperm loaded with FM4-64 and Sytox Blue dyes, with one and two minutes of incubation time, respectively. (A) The quadrants show: Sytox Blue+ / FM4-64 low (9.04%), Sytox Blue+ / FM4-64 high (61.6%), Sytox Blue- / FM4-64 low (19.7%), and Sytox Blue- / FM4-64 high (8.73%). Each quadrant indicates the percentage of the total sperm population exhibiting the corresponding staining pattern. Axes are presented on a log10 scale of arbitrary units of fluorescence. (B) Representative single-cell images corresponding to the four categorized sperm populations from the flow cytometry analysis in panel (A)

Author response image 5.

Relationship between cell death and FM4-64 fluorescence capacitated sperm stimulated with ionomycin. Image-based flow cytometry analysis of non-capacitated mouse sperm loaded with FM464 and Sytox Blue dyes, with one and two minutes of incubation time, respectively. (A) The quadrants show: Sytox Blue+ / FM4-64 low (4.52%), Sytox Blue+ / FM4-64 high (60.6%), Sytox Blue- / FM4-64 low (20.5%), and Sytox Blue- / FM4-64 high (13.5%). Each quadrant indicates the percentage of the total sperm population exhibiting the corresponding staining pattern. Axes are presented on a log10 scale of arbitrary units of fluorescence. (B) Representative single-cell images corresponding to the four categorized sperm populations from the flow cytometry analysis in panel (A).

Based on the data presented in Author response images 1 to 6, we derive the following conclusions summarized below:

(1) There is no direct relationship between cell death (Sytox Blue-) and AE (EGFP) (Author response images 1 and 2).

(2) There is bistability in the FM4-64 fluorescent intensity. Before reaching a certain threshold, there is no correlation between FM4-64 and Sytox Blue signals, indicating no cell death. However, after crossing this threshold, the FM4-64 signal becomes correlated with Sytox Blue+ cells, indicating cell death (Author response images 4-6).

(3) The Sytox Blue- population of capacitated sperm is sensitive to AE stimulation with progesterone, leading to the expected increase in FM4-64 fluorescence.

Therefore, while the FM4-64 signal alone is not a definitive marker for either AE or cell death, it is crucial to use additional viability assessments, such as Sytox Blue, to accurately differentiate between live and dead sperm in studies of acrosome exocytosis and sperm motility. In the present work, we did not use a cell viability marker due to the complex multicolor, multidimensional fluorescence experiments. However, cell viability was always considered, as any imaged sperm was chosen based on motility, indicated by a beating flagellum. The determination of whether selected sperm die during or after AE remains to be elucidated. The results presented in Figure 2 and Supplementary S1 show examples of motile sperm that experience an increase in FM4-64 fluorescence.

All this information is added to the manuscript (Supplementary Figure 1D).

(4) It is unclear how the structural change in the midpiece causes the entire sperm flagellum, including the principal piece, to stop moving. It will be easier for readers to understand if the authors discuss possible mechanisms.

Response P2.4: As requested, we have incorporated a possible explanation in the discussion section (see line 644-656). We propose three possible hypotheses for the cessation of sperm motility, which can be attributed to the simultaneous occurrence of various events:

(1) Rapid increase in [Ca2+]i levels: A rapid increase in [Ca2+]i levels may trigger the activation of Ca2+ pumps within the flagellum. This process consumes local ATP levels, disrupting glycolysis and thereby depleting the energy required for motility.

(2) Reorganization of the actin cytoskeleton: Alterations in the actin cytoskeleton can lead to changes in the mechanical properties of the flagellum, impacting its ability to move effectively.

(3) Midpiece contraction: Contraction in the midpiece region can potentially interfere with mitochondrial function, impeding the energy production necessary for sustained motility.

(5) The mitochondrial sheath and cell membrane are very close together when observed by transmission electron microscopy. The image in Figure 9A with the large space between the plasma membrane and mitochondria is misleading and should be corrected. The authors state that the distance between the plasma membrane and mitochondria approaches about 100 nm after the acrosome reaction (Line 330 - Line 333), but this is a very long distance and large structural changes may occur in the midpiece. Was there any change in the mitochondria themselves when they were observed with the DsRed2 signal?

Response P2.5: The authors appreciate the reviewer’s observation regarding the need to correct the image in Figure 9A, as the original depiction conveys a misleading representation of the spatial relationship between the mitochondrial sheath and the plasma membrane. This figure has been corrected to accurately reflect a more realistic proximity, while keeping in mind that it is a cartoonish representation.

Regarding the comments about the distances mentioned between former lines 330 and 333, the measurement was not intended to describe the gap between the plasma membrane and the mitochondria but rather the distance between F-actin and the plasma membrane. 

Author response image 6 shows high-resolution scanning electron microscopy (SEM) of two sperm fixed with a protocol tailored to preserve plasma membranes (ref), where the insets clearly show the flagellate architecture in the midpiece with an intact plasma membrane covering the mitochondrial network. A non-capacitated sperm with an intact acrosome is shown in panel A, and a capacitated sperm that has experienced AE is shown in panel B.

Notably, the results depicted in Author response image 6 demonstrate that, irrespective of the AE status, the distance between the plasma membrane and mitochondria consistently remains less than 20 nm, thus confirming the close proximity of these structures in both physiological states. As Reviewer 2 pointed out, if there is no significant difference in the distance between the plasma membrane and mitochondria, then the observed structural changes in the actin network within the midpiece should somehow alter the actual deposition of mitochondria within the midpiece. Figure 5D-F shows that midpiece contraction is associated with a decrease in the helical pitch of the actin network; the distance between turns of the actin helix decreases from  l = 248  nm to  l = 159  nm. This implies a net change in the number of turns the helix makes per 1 µm, from 4 to 6 µm-1.

Author response image 6.

SEM image showing the proximity between plasma membrane and mitochondria. Scale bar 100 nm.

Additionally, a structural contraction can be observed in Figure 5D-F, where the radius of the helix decreases by about 50 nm. To clarify this point, we sought to measure the deposition of individual DsRed2 mitochondria using computational superresolution microscopy—FF-SRM (SRRF and MSSR), Structured Illumination Microscopy (SIM), or a combination of both (SIM + MSSR), in 2D. Author response image 7 shows that these three approaches allow the observation of individual DsRed mitochondria; however, the complexity of their 3D arrangement, combined with the limited space between mitochondria (as seen in Author response image 6), precludes a reliable estimation of mitochondrial organization within the midpiece. To overcome these challenges, we decided to study the midpiece architecture via SEM experiments on non-capacitated versus capacitated sperm stimulated with ionomycin to undergo the AE.

Author response image 7.

Organization of mitochondria observed via FF-SRM and SIM. Scale bar 2 µm. F.N: Fluorescence normalized. F: Frequency

Author response image 8 presents a single-cell comparison of the midpiece architecture in noncapacitated (NC) and acrosome-intact (AI) versus acrosome-reacted (AR) sperm, along with measurements of the midpiece diameter throughout its length. Notably, the diameter of the midpiece increases from the base of the head to more distal regions, ranging from 0.45 nm to 1.10 µm (as shown in Author response images 7 and 8). A significant correlation between the diameter of the flagellum and its curvature was observed (Author response image 9), suggesting a reorganization of the midpiece due to shearing forces. This is further exemplified in Author response images 8 and 9, which provide individual examples of this phenomenon.

Author response image 8.

Comparison of the midpiece architecture in acrosome-intact and acrosome-reacted sperm using scanning electron microscopy (SEM).

As expected, the overall diameter of the midpiece in AI sperm was larger than in AR sperm, with measurements of 0.731 ± 0.008 µm for AI and 0.694 ± 0.007 µm for AR (p = 0.013, Kruskal-Wallis test n > 100, N = 2), as shown in Author response image 10. Additionally, this Author response image 7 indicates that the reorganization of the midpiece architecture involves a change in the periodicity of the mitochondrial network, with frequencies shifting from fNC to fEA mitochondria per micron.  

Author response image 9.

Comparison of the midpiece architecture in acrosome-intact (A) and acrosome-reacted (B) sperm using scanning electron microscopy (SEM).

Collectively, the structural results presented in Figure 5 and Author response images 6 to 10 demonstrate that the AE involves a comprehensive reorganization of the midpiece, affecting its diameter, pitch, and the organization of both the actin and mitochondrial networks. All this information is now incorporated in the new version of the paper (Figure. 2F)

Author response image 10.

Quantification of the midpiece diameter of the sperm flagellum in acrosome-intact and acrosome-reacted sperm analyzed by scanning electron microscopy (SEM). Data is presented as mean ± SEM. Kruskal-Wallis test was employed,  p = 0.013 (AI n=85 , AR n=72).

(6) In the TG sperm used, the green fluorescence of the acrosome disappears when sperm die. Figure 1C should be analyzed only with live sperm by checking viability with propidium iodide or other means.

Response P2.6: We concur with Reviewer 2 that ideally, any experiment conducted for this study should include an intrinsic cell viability test. However, the current research employs a wide array of multidimensional imaging techniques that are not always compatible with, or might be suboptimal for, simultaneous viability assessments. In agreement with the reviewer's concerns, it is recognized that the data presented in Figure 1C may inherently be biased due to cell death. Nonetheless, Author response image 1 demonstrates that the relationship between AE and cell death is more complex than a straightforward all-or-nothing scenario. Specifically, Author response image 1C illustrates a case where the plasma membrane is compromised (Sytox Blue+) yet maintains acrosomal integrity (EGFP+). This observation contradicts Reviewer 1's assertion that "the green fluorescence of the acrosome disappears when sperm die," as discussed more comprehensively in response P2.3.

In light of these observations, we have meticulously revisited the entire manuscript to address and clarify potential biases in our results due to cell death. Consequently, Author response image 5 and its detailed description have been incorporated into the supplementary material of the manuscript to contribute to the transparency and reliability of our findings.

Reviewer #3 (Public Review):

(1) While progressive and also hyperactivated motility are required for sperm to reach the site of fertilization and to penetrate the oocyte's outer vestments, during fusion with the oocyte's plasma membrane it has been observed that sperm motility ceases. Identifying the underlying molecular mechanisms would provide novel insights into a crucial but mostly overlooked physiological change during the sperm's life cycle. In this publication, the authors aim to provide evidence that the helical actin structure surrounding the sperm mitochondria in the midpiece plays a role in regulating sperm motility, specifically the motility arrest during sperm fusion but also during earlier cessation of motility in a subpopulation of sperm post acrosomal exocytosis. The main observation the authors make is that in a subpopulation of sperm undergoing acrosomal exocytosis and sperm that fuse with the plasma membrane of the oocyte display a decrease in midpiece parameter due to a 200 nm shift of the plasma membrane towards the actin helix. The authors show the decrease in midpiece diameter via various microscopy techniques all based on membrane dyes, bright-field images and other orthogonal approaches like electron microscopy would confirm those observations if true but are missing. The lack of additional experimental evidence and the fact that the authors simultaneously observe an increase in membrane dye fluorescence suggests that the membrane dyes instead might be internalized and are now staining intracellular membranes, creating a false-positive result. The authors also propose that the midpiece diameter decrease is driven by changes in sperm intracellular Ca2+ and structural changes of the actin helix network. Important controls and additional experiments are needed to prove that the events observed by the authors are causally dependent and not simply a result of sperm cells dying.

Response P3.1: We appreciate the reviewer's observations and critiques. In response, we have expanded our experimental approach to include alternative methodologies such as mathematical modeling and electron microscopy, alongside further fluorescence microscopy studies. This diversified approach aims to mitigate potential interpretation artifacts and substantiate the validity of our observations regarding the contraction of the sperm midpiece. Additionally, we have implemented further control experiments to fortify the credibility and robustness of our findings, ensuring a more comprehensive and reliable set of results.

First, we acknowledge the concerns raised by Reviewer 2 regarding the interpretation of the magnitude of the observed contraction of the sperm flagellum's midpiece (see response P2.5). Specifically, we believe that the assertion that "... there is a decrease in midpiece parameter due to a 200 nm shift of the plasma membrane towards the actin helix" stated by reviewer 3 needs careful examination. We recognize that the fluorescence microscopy data provided might not conclusively support such a substantial shift. Our live cell imaging and superresolution microscopy experiments indicate that there is a significant decrease in the diameter of the sperm flagellum associated with AE. This is supported by colocalization experiments where FM4-64-stained structures (fluorescing upon binding to membranes) are observed moving closer to Sir-Actinlabeled structures (binding to F-actin). Quantitatively, Figure S5 describes the spatial shift between FM4-64 and Sir-Actin signals, narrowing from a range of 140-210 nm to 50-110 nm (considering the 2nd and 3rd quartiles of the distributions). The mean separation distance between both signals changes from 180 nm in AI cells to 70 nm in AR cells, a net shift of 110 nm. This observation suggests caution regarding the claim of a "200 nm shift of the plasma membrane towards the actin cortex." 

Moreover, the concerns raised by Reviewer #3 about the potential internalization of membrane dyes, which might create a false-positive result by staining intracellular membranes, offer an alternative mechanism to explain a shift of up to 100 nm. This perspective is also supported by the critique from Reviewer #2 regarding the substantial distance (about 100 nm) between the plasma membrane and mitochondria post-acrosome reaction:  “The authors state that the distance between the plasma membrane and mitochondria approaches about 100 nm after the acrosome reaction (…), but this is a very long distance and large structural changes may occur in the midpiece”. These insights have prompted us to refine our methodology and interpretation of the data to ensure a more accurate representation of the underlying biological processes.

Author response image 11 shows a first principles approach in two spatial dimensions to explore three scenarios where a membrane dye, such as FM4-64, stains structures at and within the midpiece of a sperm flagellum, but yet does not result in a net change of diameter. Author response image 11A-C illustrates three theoretical arrangements of fluorescent dyes: Model 1 features two rigid, parallel structures that mimic the plasma membrane surrounding the midpiece of the flagellum. Model 2 builds on Model 1 by incorporating the possibility of dye internalization into structures located near the membrane, suggesting a slightly more complex interaction with nearby membranous intracellular structures. Model 3 represents an extreme scenario where the fluorescent dyes stain both the plasma membrane and internal structures, such as mitochondrial membranes, indicating extensive dye penetration and binding. Author response image 11D-F displays the convolution of the theoretical fluorescent signals from Models 1 to 3 with the theoretical point spread function (PSF) of a fluorescent microscope, represented by a Gaussian-like PSF with a sigma of 19 pixels (approximately 300 nm). This process simulates how each model's fluorescence would manifest under microscopic observation, showing subtle differences in the spatial distribution of fluorescence among the models. Author response image 11G-I reveals the superresolution images obtained through Mean Shift Super Resolution (MSSR) processing of the models depicted in Author response image 11D-F.

By analyzing the three scenarios, it becomes clear that the signals from Models 2 and 3 shift towards the center compared to Model 1, as depicted in Author response image 11J. This shift in fluorescence suggests that the internalization of the dye and its interaction with internal structures might significantly influence the perceived spatial distribution and intensity of fluorescence, thereby impacting the interpretation of structural changes within the midpiece. Consequently, the experimentally observed contraction of up to 100 nm in  could represent an actual contraction of the sperm flagellum's midpiece, a relocalization of the FM4-64 membrane dyes to internal structures, or a combination of both scenarios.

To discern between these possibilities, we implemented a scanning electron microscopy (SEM) approach. The findings presented in Figure 5 and Author response images 7 to 9 conclusively demonstrate that the AE involves a comprehensive reorganization of the midpiece. This reorganization affects its diameter, which changes by approximately 50 nm, as well as the pitch and the organization of both the actin and mitochondrial networks. This data corroborates the structural alterations observed and supports the validity of our interpretations regarding midpiece dynamics during the AE.

Author response image 11.

Modeling three scenarios of midpiece staining with membrane fluorescent dyes.

Secondly, we wish to clarify that in some of our experiments, we have utilized changes in the intensity of FM4-64 fluorescence as an indirect measure of midpiece contraction. This approach is supported by a linear inverse correlation between these variables, as illustrated in Figure S2D. It is important to note that this observation is correlative and indirect; therefore, our data does not directly substantiate the claim that "in a subpopulation of sperm undergoing AE and sperm that fuse with the plasma membrane of the oocyte, there is a decrease in midpiece parameter due to a 200 nm shift of the plasma membrane towards the actin helix". Specifically, we have not directly measured the distance between the plasma membrane and actin cortex in experiments involving gamete fusion.

All the concerns highlighted in this Response P1.1 have been addressed and incorporated into the manuscript. This addition aims to provide comprehensive insight into the experimental observations and methodologies used, ensuring that the data is transparent and accessible for thorough review and replication.

Editor Comment:

As the authors can see from the reviews, the reviewers had quite different degrees of enthusiasm, thus discussed extensively. The major points in consensus are summarized below and it is highly recommended that the authors consider their revisions.

(1) Causality of midpiece contraction with motility arrest is not conclusively supported by the current evidence. Time-resolved imaging of FM4-64 and motility is needed and the working model needs to be revised with two scenarios - whether the sperm contracting indicates a fertilizing sperm or sperm to be degenerated.

(2) The rationale for using FM4-64 as a plasma membrane marker is not clear as it is typically used as an endo-membrane marker, which is also related to the discrepancy of Fluo-4 signal diameter vs. FM4-64 (Figure 4E). The viability of sperm with increased FM4-64 needs to be demonstrated.

(3) The mechanism of midpiece contraction in motility cessation along the whole flagellum is not discussed.

(4) The use of an independent method to support the changes in midpiece diameter/structural changes such as DsRed (transgenic) or TEM.

(5) The claim of Ca2+ change needs to be toned down.

Response Editor: We thank the editor and the reviewers for their thorough and positive assessment of our work and the constructive feedback to further improve our manuscript. Please find below our responses to the reviewers’ comments. We have addressed all these points in the current version. Briefly,

(1) Time resolved images to show the correlation between FM4-64 fluorescence increase and the motility was incorporated

(2) The rationale for using FM4-64 was added.

(3) The mechanism of midpiece contraction was discussed in the paper

(4) An independent method was included to support our conclusions (SEM and other markers not based on membrane dyes)

(5) The results related to the calcium increase were toned down.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

(1) To claim midpiece actin polymerization/re-organization is required for AE, demonstrating that AE does not occur in the presence of actin depolymerizing drugs (e.g., Latrunculin A, Cytochalasin D) would be necessary since the current data only shows the association/correlation. Was the block of AE by actin depolymerization observed?

Response R1.1: We agree with the reviewer but unfortunately, since actin polymerization and or depolymerization in the head are important for exocytosis, we cannot use this experimental approach to dissect both events. Addition of these inhibitors block the occurrence of AE (PMID: 12604633).

(2) Please provide the rationale for using FM4-64 to visualize the plasma membrane since it has been reported to selectively stain membranes of vacuolar organelles. What is the principle of increase of FM4-64 dye intensity, other than the correlation with midpiece contraction? For example, in lines 400-402: the authors mentioned that 'some acrosomereacted moving sperm within the perivitelline space had low FM4-64 fluorescence in the midpiece (Figure 6C). After 20 minutes, these sperm stopped moving and exhibited increased FM4-64 fluorescence, indicating midpiece contraction (Figure 6D).' While recognizing the increase of FM4-64 dye intensity can be an indicator of midpiece contraction, without knowing how and when the intensity of FM4-64 dye changes, it is hard to understand this observation. Please discuss.

Response R1.2: FM4-64 is an amphiphilic styryl fluorescent dye that preferentially binds to the phospholipid components of cell membranes, embedding itself in the lipid bilayer where it interacts with phospholipid head groups. Due to its amphiphilic nature, FM dyes primarily anchor to the outer leaflet of the bilayer, which restricts their internalization. It has been demonstrated that FM4-64 enters cells through endocytic pathways, making these dyes valuable tools for studying endocytosis.

Upon binding, FM4-64's fluorescence intensifies in a more hydrophobic environment that restricts molecular rotation, thus reducing non-radiative energy loss and enhancing fluorescence. These photophysical properties render FM dyes useful for observing membrane fusion events. When present in the extracellular medium, FM dyes rapidly reach a chemical equilibrium and label the plasma membrane in proportion to the availability of binding sites.

In wound healing studies, for instance, the fluorescence of FM4-64 is known to increase at the wound site. This increase is attributed to the repair mechanisms that promote the fusion of intracellular membranes at the site of the wound, leading to a rise in FM4-64 fluorescence. Similarly, an increase in FM4-64 fluorescence has been reported in the heads of both human and mouse sperm, coinciding with AE. In this scenario, the fusion between the plasma membrane and the acrosomal vesicle provides additional binding sites for FM4-64, thus increasing the total fluorescence observed in the head. This dynamic response of FM4-64 makes it an excellent marker for studying these cellular processes in real-time.

This study is the first to report an increase in FM4-64 fluorescence in the midpiece of the sperm flagellum. Figures 5 and Author response images 6 to 9 demonstrate that during the contraction of the sperm flagellum, structural rearrangements occur, including the compaction of the mitochondrial sheath and other membranous structures. Such contraction likely increases the local density of membrane lipids, thereby elevating the local concentration of FM4-64 and enhancing the probability of fluorescence emission. Additionally, changes in the microenvironment such as pH or ionic strength during contraction might further influence FM4-64’s fluorescence properties, as detailed by Smith et al. in the Journal of Membrane Biology (2010). The photophysical behavior of FM4-64, including changes in quantum yield due to tighter membrane packing or alterations in curvature or tension, may also contribute to the increased fluorescence observed. Notably, Figure S2 indicates that other fluorescent dyes like Memglow 700, Bodipy-GM, and FM1-43 also show a dramatic increase in their fluorescence during the midpiece contraction. Investigating whether the compaction of the plasma membrane or other mesoscale processes occur in the midpiece of the sperm flagellum could be a valuable area for future research. The use of fluorescent dyes such as LAURDAN or Nile Red might provide further insights into these membrane dynamics, offering a more comprehensive understanding of the biochemical and structural changes during sperm motility and gamete fusion events.

(3) As the volume of the whole midpiece stays the same while the diameter decreases along the whole midpiece (midpiece contraction), the authors need to describe what changes in the midpiece length they observe during the contraction. Was the length of the midpiece during the contraction measured and compared before and after contraction?

Response R1.3: As requested, we have measured the length of the midpiece in AI and AR sperm. As shown in Author response image 12 (For review purposes only), no statistically significant differences were observed. 

Author response image 12.

Midpiece length measured by the length of mitochondrial DsRed2 fluorescence in EGFP-DsRed2 sperm. Measurements were done before (acrosome-intact) and after (acrosome-reacted) acrosome exocytosis and midpiece contraction. Data is presented as the mean ± sem of 14 cells induced by 10 µM ionomycin. Paired t-test was performed, resulting in no statistical significance. 

(4) Most of all, it is not clear what the midpiece, thus mitochondria, contraction means in terms of sperm bioenergetics and motility cessation. Would the contraction induce mitochondrial depolarization or hyperpolarization, increase or decrease of ATP production/consumption? It will be great if this point is discussed. For example, an increase in mitochondrial Ca2+ is a good indicator of mitochondrial activity (ATP production).

Response R1.4: That is an excellent point. We have discussed this idea in the discussion (line 620-624). We are currently exploring this idea using different approaches because we also think that these changes in the midpiece may have an impact in the function of the mitochondria and perhaps, in their fate once they are incorporated in the egg after fertilization. 

(5) The authors claimed that Ca2+ signal propagates from head to tail, which is the opposite of the previous study (PMID: 17554080). Please clarify if it is a speculation. Otherwise, please support this claim with direct experimental evidence (e.g., high-speed calcium imaging of single cells).

Response R1.5: In that study, it was claimed that a [Ca2+]i  increase that propagates from the tail to the head occurs when CatSper is stimulated. They did not evaluate the occurrence of AE when monitoring calcium.

Our data is in agreement with our previous results (PMID: 26819478) that consistently indicated that only the[Ca2+]i  rise originating in the sperm head is able to promote AE. 

(6) Figure 4E: Please explain how come Fluo4 signal diameter can be smaller than FM4-64 dye if it stains plasma membrane (at 4' and 7').

Response R1.6: When colocalizing a diffraction-limited image (Fluo4) with a super-resolution image (FM4-64), discrepancies in signal sizes and locations can become apparent due to differences in resolution. The Fluo4 signal, being diffraction-limited, adheres to a resolution limit of approximately 200-300 nanometers under conventional light microscopy. This limitation causes the fluorescence signal to appear broader and less defined. Conversely, super-resolution microscopy techniques, such as SRRF (Super-Resolution Radial Fluctuations), achieve resolutions down to tens of nanometers, allowing FM4-64 to reveal finer details at the plasma membrane and display potentially smaller apparent sizes of stained structures. Although both dyes might localize to the same cellular regions, the higher resolution of the FM4-64 image allows it to show a more precise and smaller diameter of the midpiece of the flagellum compared to the broader, less defined signal of Fluo4. To address this, the legend of Figure 4E has been slightly modified to clarify that the FM4-64 image possesses greater resolution. 

(7) Figure 5D-G: the midpiece diameter of AR intact cells was shown ~ 0.8 um or more in Figure 2, while now the radius in Figure 5 is only 300 nm. Since the diameter of the whole midpiece is nearly uniform when the acrosome is intact, clarify how and what brings this difference and where the diameter/radius measurement is done in each figure.

Response R1.7: The difference resides in what is being measured. In Figure 2, the total diameter of the cell is measured, through the maximum peaks of FM4-64 fluorescence which is a probe against plasma membrane. As for Figure 5, the radius shown makes reference to the radius of the actin double helix within the midpiece. To that end, cells were fixed and stained with phalloidin, a F-actin probe.

Minor points

(8) Figure S1 title needs to be changed. The "Midpiece contraction" concept is not introduced when Figure S1 is referred to.

Response R1.8: This was corrected in the new version.

(9) Reference #19: the authors are duplicated.

Response R1.9: This was corrected in the new version.

(10) Line 315-318: sperm undergoing contraction -> sperm undergoing AR/AE?

Response R1.10: This was corrected in the new version.

(11) Line 3632 -> punctuation missing.

Response R1.11: Modified as requested.

(12) Movie S7: please add an arrow to indicate the spermatozoon of interest.

Response R1.12:  The arrow was added as suggested.

(13) Line 515: One result of this study was that the sperm flagellum folds back during fusion coincident with the decrease in the midpiece diameter. The authors did not provide an explanation for this observation. Please speculate the function of this folding for the fertilization process.

Response R1.13: As requested, this is now incorporated in the discussion. We speculate that the folding of the flagellum during fusion further facilitates sperm immobilization because it makes it more difficult for the flagellum to beat. Such processes can enhance stability and increase the probability of fusion success. Mechanistically, the folding may occur as a consequence of the deformation-induced stress that develops during the decrease of midpiece diameter. 

Reviewer #2 (Recommendations For The Authors):

(1) Figure 2C, D, E. Does "-1" on the X-axis mean one minute before induction? If so, the diameter is already smaller and FM4-64 fluorescence intensity is higher before the induction in the spontaneous group. Does the acrosome reaction already occur at "-1" in this group?

Response R2.1: Yes, “-1” means that the measurements of the diameter/FM4-64 fluorescence was done one minute before the induction. And it is correct that the diameter is smaller and FM464 fluorescence higher in the spontaneous group because these sperm underwent acrosome exocytosis before the induction, that is, spontaneously.

(2) Figure 3D. Purple dots are not shown in the graph on the right side.

Response R2.2: Modified as requested.

(3) Lines 404-406. "These results suggest that midpiece contraction and motility cessation occur only after acrosome-reacted sperm penetrate the zona pellucida". Since midpiece contraction and motility cessation also occur before the passage through the zona pellucida (Figure 9B), "only" should be deleted.

Response R2.3: Modified as requested.

Reviewer #3 (Recommendations For The Authors):

(1) Do the authors have a hypothesis as to why the observed decrease in midpiece parameter results in cessation of sperm motility? It would be beneficial for the manuscript to include a paragraph about potential mechanisms in the discussion.

Response R3.1: As requested, a potential mechanism has been proposed in the discussion section (line 644-656).

(2) Since the authors propose in Gervasi et al. 2018 that the actin helix might be responsible for the integrity of the mitochondrial sheath and the localization of the mitochondria, is it possible that the proposed change in plasma membrane diameter and actin helix remodeling for example alters the localization of the mitochondria? TEM should be able to reveal any associated structural changes. In its current state, the manuscript lacks experimental evidence supporting the author's claim that the "helical actin structure plays a role in the final stages of motility regulation". The authors should either include additional evidence supporting their hypothesis or tone down their conclusions in the introduction and discussion.

Response  R3.3: We agree with the reviewer. This is an excellent point. As suggested by this reviewer as well as the other reviewers, we have performed SEM to observe the changes in the midpiece observed after its contraction for two main reasons. First, to confirm this observation using a different approach that does not involve the use of membrane dyes. As shown in Author response image 6-10, we have observed that in addition to the midpiece diameter, there is a reorganization of the mitochondria sheet that is also suggested by the SIM experiments. These observations will be explored with more experiments to confirm the structural and functional changes that mitochondria undergo during the contraction. We are currently investigating this phenomenon, These results are now included in the new Figure  2F.

(3) In line 134: The authors write: 'Some of the acrosome reacted sperm moved normally, whereas the majority remained immotile". Do the authors mean that a proportion of the sperm was motile prior to acrosomal exocytosis and became immotile after, or were the sperm immotile to begin with? Please clarify.

Response R3.4: This statement is based on the quantification of the motile sperm after induction of AE within the AR population (Fig. 1C). 

(4) The authors do not provide any experimental evidence supporting the first scenario. In video 1 a lot of sperm do not seem to be moving to begin with, only a few sperm show clear beating in and out of the focal plane. The highlighted sperm that acrosome-reacted upon exposure to progesterone don't seem to be moving prior to the addition of progesterone. In contrast, the sperm that spontaneously acrosome react move the whole time. In video 1 this reviewer was not able to identify one sperm that stopped moving upon acrosomal exocytosis. Similarly in video 3, although the resolution of the video makes it difficult to distinguish motile from non-motile sperm. In video 2 the authors only show sperm that are already acrosome reacted. Please explain and provide additional evidence and statistical analysis supporting that sperm stop moving upon acrosomal exocytosis.

Response R3.5: In videos 1 and 3, the cells are attached to the glass with concanavalin-A, this lectin makes sperm immotile (if well attached) because both the head and tail stick to the glass. The observed motility of sperm in these videos is likely due to them not being properly attached to the glass, which is completely normal. On the contrary, in videos 2 and 4, sperm are attached to the glass with laminin. This is a glycoprotein that only binds the sperm to the glass through its head, that is why they move freely.

(5) Could the authors provide additional information about the FM4-64 fluorescent dye?

What is the mechanism, and how does it visualize structural changes at the flagellum? Since the whole head lights up, does that mean that the dye is internalized and now stains additional membranes, similar to during wound healing assays (PMID 20442251, 33667528). Or is that an imaging artifact? How do the authors explain the correlation between FM4-64 fluorescence increase in the midpiece and the observed change in diameter? Does FM4-64 have solvatochromatic properties?

Response R3.6: We appreciate the insightful queries posed by Reviewer 3, which echo the concerns initially brought forward by Reviewer 1. For a detailed explanation of the mechanism of FM4-64 dye, how we interpret  it, visualizes structural changes in the flagellum, and its behavior during cellular processes, please refer to our detailed response in Response R1.2. In brief, FM464 is a lipophilic styryl dye that preferentially binds to the outer leaflets of cellular membranes due to its amphiphilic nature. Upon binding, the dye becomes fluorescent, allowing for the visualization of membrane dynamics. The increase in fluorescence in the sperm head or midpiece likely results from the dye’s accumulation in areas where membrane restructuring occurs, such as during AE or in response to changes in the flagellum structure.

Regarding the specific questions about internalization and whether FM4-64 stains additional membranes similarly to what is observed in wound healing assays, it's important to note that FM4-64 can indeed be internalized through endocytosis and subsequently label internal vesicular structures. Additionally, FM4-64 may experience changes in its fluorescence as a result of fusion events that increase the lipid content of the plasma membrane, as observed in studies cited (PMID 20442251, 33667528). This characteristic makes FM4-64 valuable not only for outlining cell membranes but also for tracking the dynamics of both internal and external membrane systems, particularly during cellular events that involve significant membrane remodeling, such as wound healing or AE.

Concerning whether the increased fluorescence and observed changes in diameter are artifacts or reflect real biological processes, the correlation observed likely indicates actual changes in the midpiece architecture through molecular mechanisms that remain to be further elucidated. The data presented in Figures 5 and Author response images 6-10 support that this increase in fluorescence is not merely an artifact but a feature of how FM4-64 interacts with its environment. 

Finally, regarding the solvatochromatic properties of FM4-64, while the dye does show changes in its fluorescence intensity in different environments, its solvatochromatic properties are generally less pronounced than those of dyes specifically designed to be solvatochromatic. FM464's fluorescence changes are more a result of membrane interaction dynamics and dye concentration than of solvatochromatic shifts. 

(6) For the experiment summarized in Figure S1, did the authors detect sperm that acrosome-reacted upon exposure to progesterone and kept moving? This reviewer is wondering how the authors reliably measure FM4-64 fluorescence if the flagellum moves in and out of the focal plane. If the authors observe sperm that keep moving, what was the percentage within a sperm population and how did FM4-64 fluorescence change?

Response R3.6: We did identify sperm that underwent acrosome reaction upon exposure to progesterone and continued to exhibit movement. However, due to the issue raised by the reviewer regarding the flagellum going out of focus, we opted to quantify the percentage of sperm that were adhered to the slide (using laminin). This approach allows for the observation of flagellar position over time, facilitating an easy assessment of fluorescence changes. The percentage of sperm that maintained movement after AE is depicted in Figure 1C.

(7) In Figure S1B it doesn't look like the same sperm is shown in all channels or time points, the hook shown in the EGFP channel is not always pointing in the same direction. If FM4-64 is staining the plasma membrane, how do the authors explain that the flagellum seems to be more narrow in the FM4-64 channel than in the brightfield and DsRed2 channel?

Response 3.7: It is the same sperm, but due to technical limitations images were sequentially acquired. For example, for time 5 minutes after progesterone, all images in DIC were taken, then all images in the EGFP channel, then DsRed2* and finally FM4-64. The reason for this was to acquire images as fast as possible, particularly in DIC images which were then processed to get the beat frequency.

Regarding the flagellum that seems to be more narrow in the FM4-64 channel compared to the BF or DsRed2 channel, the explanation is related to the fact that intensity of the DsRed2 signal is stronger than the other two. This higher signal may have increased the amount of photons captured by the detector.

(8) Overall, it would be beneficial to include statistics on how many sperm within a population did change FM4-64 fluorescence during AE and how many did not, in addition to information about motility changes and viability. Did the authors exclude that the addition of FM4-64 causes cell death which could result in immotile sperm or that only dying sperm show an increase in FM4-64 fluorescence?

Response 3.8: The relationship between cell death and the increase in FM4-64 fluorescence is widely discussed in Response P2.3. In our experiments, we always considered sperm that were motile to hypothesize about the relevance of this observation. We have two types of experiments: 

(1) Sperm-egg Fusion: In experiments where sperm and eggs were imaged to observe their fusion, sperm were initially moving and after fusion, the midpiece contraction (increase in FM4-64 fluorescence was observed) indicating that the change in the midpiece (that was observed consistently in all fusing cells analyzed), is part of the process. 

(2) Sperm that underwent AE: we have observed two behaviours as shown in Figure 1: 

a) Sperm that underwent AE and they remain motile without midpiece contraction (they are alive for sure); 

b) Sperm that underwent AE and stopped moving with an increase in FM464 fluorescence. We propose that this contraction during AE is not desired because it will impede sperm from moving forward to the fertilization site when they are in the female reproductive tract. In this case, we acknowledge that the cessation of sperm motility may be attributed to cellular death, potentially correlating with the increased FM4-64 signal observed in the midpiece of immotile sperm that have undergone AE. To address this hypothesis, we conducted image-based flow cytometry experiments, which are well-suited for assessing cellular heterogeneity within large populations.

Regarding the relationship between the increase in FM4-64 and AE, we have always observed that AE is followed by an increase in FM4-64 in the head in mice (PMID: 26819478) as well as in human (PMID: 25100708) sperm. This was originally corroborated with the EGFP sperm. However, not all the cells that undergo AE increase the FM4-64 fluorescence in the midpiece.

(9) The authors report that a fraction of sperm undergoes AE without a change in FM4-64 fluorescence (Figure 1F). How does the [Ca2+]i change in those cells? Again statistics on the distribution of a certain pattern within a population in addition to showing individual examples would be very helpful.

Response 3.9: A recent work shows that an initial increase in [Ca2+]i  is required to induce changes in flagellar beating necessary for hyperactivation (Sánchez-Cárdenas et al., 2018). However, when [Ca2+]i  increases beyond a certain threshold, flagellar motility ceases. These conclusions are based on single-cell experiments in murine sperm with different concentrations of the Ca2+ ionophore, A23187. The authors reported that complete loss of motility was observed when using ionophore concentrations higher than 1 μM. In contrast, spermatozoa incubated with 0.5 μM A23187 remained motile throughout the experiment. Once the Ca2+ ionophore is removed, the sperm would reduce the concentration of this ion to levels compatible with motility and hyperactivation (Navarrete et al., 2016). However, some of the washed cells did not recover mobility in the recorded time window (Sánchez-Cárdenas et al., 2018). These results would indicate that due to the increase in [Ca2+]i  induced by the ionophore, irreversible changes occurred in the sperm flagellum that prevented recovery of mobility, even when the ionophore was not present in the recording medium. 

Taking into account our results, one possible scenario to explain this irreversible change would be the contraction of the midpiece. Our results demonstrate that the increase in [Ca2+]i observed in the midpiece (whether by induction with progesterone, ionomycin or occurring spontaneously) causes the contraction of this section of the flagellum and its subsequent immobilization. 

(10) While the authors results show that changes in [Ca2+]i correlate with the observed reduction of the midpiece diameter, they do not provide evidence that the structural changes are triggered by Ca2+i influx. It could just be a coincidence that both events spatially overlap and that they temporarily follow each other. The authors should either provide additional evidence or tone down their conclusion.

Response 3.10: We agree with the reviewer. As suggested, we have toned down our conclusion.

(11) Are the authors able to detect the changes in the midpiece diameter independent from FM4-64 or other plasma membrane dyes? An alternative explanation could be that the dyes are internalized due to cell death and instead of staining the plasma membrane they are now staining intracellular membranes, resulting in increased fluorescence and giving the illusion that the midpiece diameter decreased. How do the authors explain that the Bodipy-GM1 Signal directly overlaps with DsRed2 and SIR-actin, shouldn't there be some gap? Since the rest of the manuscript is based on that proposed decrease in midpiece diameter the authors should perform orthogonal experiments to confirm their observation.

Response 3.11: As requested by the reviewer, we have not used new methods to visualize the change in sperm diameter in the midpiece. In neither of them, a membrane dye was used. First, we have performed immunofluorescence to detect a membrane protein (GLUT3). Second, we have used scanning electron microscopy. The results are now incorporated in the new Figure 2FG. In both experiments, a change in the midpiece diameter was observed. Please, also visit responses P2.5 and Author response images 8 to 10.  

Regarding the overlap between the signal of Bodipy GM1 (membrane) and the fluorescence of DsRed2 (mitochondria) and Sir-Actin (F-actin), it is only observed in acrosomereacted sperm, not in acrosome-intact sperm (Figure S4). In our view, these structures become closed after midpiece contraction, and the resolution of the images is insufficient to distinguish them clearly. This issue is also evident in Figure 5B. Therefore, we conducted additional experiments using more powerful super-resolution techniques such as STORM (Figures 5D-F).

(12) The proposed gap of 200 nM between the actin helix and the plasma membrane, has been observed by TEM? Considering that the diameter of the mouse sperm midpiece is about 1 um, that is a lot of empty space which leaves only about 600 nm for the rest of the flagellum. The axoneme is 300 nm and there needs to be room for the ODFs and the mitochondria. Please explain.

Response 3.12: Unfortunately, the filament of polymerized actin cannot be observed by TEM. Furthermore, we were discouraged from trying other approaches, such as utilizing phalloidin gold, because for some reason, it does not work properly.

In our view, the 200 nm gap between the actin cytoskeleton and the plasma membrane is occupied by the mitochondria (that is the size that it is frequently reported based on TEM; see https://doi.org/10.1172/jci.insight.166869).

(13) The results provided by the authors do not convince this reviewer that the actin helix moves, either closer to the plasma membrane or toward the mitochondria, the observed differences are minor and not confirmed by statistical analysis.

Response 3.13: As requested, the title of that section was changed. Moreover, our conclusion is exactly as the reviewer is suggesting: “Since the results of the analysis of SiR-actin slopes were not conclusive, we studied the actin cytoskeleton structure in more detail”. This conclusion is based on the statistical analysis shown in Figure S5D-E.

(14) The fluorescence intensity of all plasma membrane dyes increases in all cells chosen by the authors for further analysis. Could the increase in SiR-Actin fluorescence be explained by a microscopy artifact instead of actin helix remodeling? Alternatively, can the authors exclude that the observed increase in SIR-Actin might be an artifact caused by the increase in FM4-64 fluorescence? Since the brightness in the head similarly increases to the fluorescence in the flagellum the staining pattern looks suspiciously similar. Did the authors perform single-stain controls?

Response 3.14: We had similar concerns when we were doing the experiments using SiR-actin. Although we have performed single stain controls to make sure that the actin helix remodelling occurs during the midpiece contraction, we have performed experiments using higher resolution techniques such as STORM using a different probe to stain actin (Phalloidin).

(15) Should actin cytoskeleton remodeling indeed result in a decrease of actin helix diameter, what do the authors propose is the underlying mechanism? Shouldn't that result in changes in mitochondrial structure or location and be visible by TEM? This reviewer is also wondering why the authors focus so much on the actin helix, while the plasma membrane based on the author's results is moving way more dramatically.

Response 3.15: This raises an intriguing point. Currently, we lack an understanding of the underlying mechanism driving actin remodeling, and we are eager to conduct further experiments to explore this aspect. For instance, we are investigating the potential role of Cofilin in remodeling the F-actin network. Initial experiments utilizing STORM imaging have revealed the localization of Cofilin in the midpiece region, where the actin helix is situated.

Regarding mitochondria, thus far, we have not uncovered any evidence suggesting that acrosome reaction or fusion with the egg induces a rearrangement of these organelles within the structure. The rationale for investigating polymerized actin in depth stems from the fact that, alongside the axoneme and other flagellar structures such as the outer dense fibers and fibrous sheet, these are the sole cytoskeletal components present in that particular tail region.

(14) The fact that the authors observe that most sperm passing through the zona pellucida, which requires motility, display high FM4-64 fluorescence, doesn't that contradict the authors' hypothesis that midpiece contraction and motility cessation are connected? Videos confirming sperm motility and information about pattern distribution within the observed sperm population in the perivitelline space should be provided.

Response 3.14: We believe it is a matter of time, as depicted in Figure 1D, our model shows that first the cells lose the acrosome, present motility and low FM4-64 fluorescence in the midpiece (pattern II) and after that, they lose motility and increase FM4-64 fluorescence in the midpiece (pattern III). That is why, we think that when sperm pass the zona pellucida they present pattern II and after some time they evolve into pattern III. 

(15) In the experiments summarized in Figure 8, did all sperm stop moving? Considering that 74 % of the observed sperm did not display midpiece contraction upon fusion, again doesn't that contradict the authors' hypothesis that the two events are interdependent? Similarly, in earlier experiments, not all acrosome-reacted sperm display a decrease in midpiece diameter or stop moving, questioning the significance of the event. If some sperm display a decrease in midpiece diameter and some don't, or undergo that change earlier or later, what is the underlying mechanism of regulation? The observed events could similarly be explained by sperm death: Sperm are dying × plasma membrane integrity changes and plasma membrane dyes get internalized × [Ca2+]i simultaneously increases due to cell death × sperm stop moving.

Response 3.15: The percentage of sperm that did not exhibit midpiece contraction in Fig.8B is 26%, not 74%, indicating that it does not contradict our hypothesis. However, this still represents a significant portion of sperm that remain unchanged in the midpiece, leaving room for various explanations. For instance, it's possible that: i) the change in fluorescence was not detected due to the event occurring after the recording concluded, or ii) in some instances, this alteration simply does not occur. Nevertheless, we did not track subsequent events in the oocyte, such as egg activation, to definitively ascertain the success of fusion. Incorporation of the dye only manifests the initiation of the process.

(16) The authors propose changes in Ca2+ as one potential mechanism to regulate midpiece contraction, however, the Ca2+ measurements during fusion are flawed, as the authors write in the discussion, by potential Ca2+ fluorophore dilution. Considering that the authors observe high Ca2+ in all sperm prior to fusion, could that be a measuring artifact? Were acrosome-intact sperm imaged with the same settings to confirm that sperm with low and high Ca2+ can be distinguished? Should [Ca2+]i changes indeed be involved in the regulation of motility cessation during fusion, could the authors speculate on how [Ca2+]i changes can simultaneously be involved in the regulation of sperm hyperactivation?

Response 3.16: We agree with the reviewer that our experiments using calcium probes are not conclusive for many technical problems. We have toned down our conclusions in the new version of the manuscript.

(17) 74: AE takes place for most cells in the upper segment of the oviduct, not all of them.

Please correct.

Response 3.17: Corrected in the new version.

(18) 88: Achieved through, or achieved by, please correct.

Response 3.18: Corrected in the new version.

(19) 243: Acrosomal exocytosis initiation by progesterone, please specify.

Response 3.19: Modified in the new version.

(20) 277: "The actin cytoskeleton approaches the plasma membrane during the contraction of the midpiece" is misleading. The author's results show the opposite.

Response 3.20: As suggested, this statement was modified.

(21) 298: Why do the authors find it surprising that the F-actin network was unchanged in acrosome-intact sperm that do not present a change in midpiece diameter?

Response 3.21: The reviewer is right. The sentence was modified.

(22) Figures 5D,F: The provided images do not support a shift in the actin helix diameter.

Response 3.22: The shift in the actin helix diameter is provided in Figure 5E and 5G.

(23) Figure S5C: The authors should show representative histograms of spontaneously-, progesterone induced-, and ionomycin-induced AE. Based on the quantification the SiRactin peaks don't seem to move when the AR is induced by progesterone.

Response 3.23: As requested, an ionomycin induced sperm is incorporated.

(24) 392: Which experimental evidence supports that statement?

Response 3.24: A reference was incorporated. 

Reference 13 is published, please update. Response 3.25: updated as requested.

Author response:

The following is the authors’ response to the previous reviews.

Reviewer #1 (Public Review):

Summary:

The study of human intelligence has been the focus of cognitive neuroscience research, and finding some objective behavioral or neural indicators of intelligence has been an ongoing problem for scientists for many years. Melnick et al, 2013 found for the first time that the phenomenon of spatial suppression in motion perception predicts an individual's IQ score. This is because IQ is likely associated with the ability to suppress irrelevant information. In this study, a high-resolution MRS approach was used to test this theory. In this paper, the phenomenon of spatial suppression in motion perception was found to be correlated with the visuo-spatial subtest of gF, while both variables were also correlated with the GABA concentration of MT+ in the human brain. In addition, there was no significant relationship with the excitatory transmitter Glu. At the same time, SI was also associated with MT+ and several frontal cortex FCs.

Strengths:

(1) 7T high-resolution MRS is used.

(2) This study combines the behavioral tests, MRS, and fMRI.

Weaknesses:

Major:

In Melnick (2013) IQ scores were measured by the full set of WAIS-III, including all subtests. However, this study only used visual spatial domain of gF. I wonder why only the visuo-spatial subtest was used not the full WAIS-III? I am wondering whether other subtests were conducted and, if so, please include the results as well to have comprehensive comparisons with Melnick (2013).

We thank the reviewer for pointing this out. The decision was informed by Melnick’s findings which indicated high correlations between Surround suppression (SI) and the Verbal Comprehension, Perceptual Reasoning, Working Memory, and Processing Speed Indexes, with correlation coefficients of 0.69, 0.47, 0.49, and 0.50, respectively. It is well-established that the hMT+ region of the brain is a sensory cortex involved in visual perception processing (3D perception). Furthermore, motion surround suppression (SI), a specific function of hMT+, aligns closely with this region's activities. Given this context, the Perception Reasoning sub-ability was deemed to have the clearest mechanism for further exploration. Consequently, we selected the most representative subtest of Perception Reasoning—the Block Design Test—which primarily assesses 3D visual intelligence.” For further clarification, due to these reasons, we conducted only the visuo-spatial subtest.

Minor:

Comments:

In the first revised version, we addressed the following recommendations in the 'Author response' file titled 'Recommendation for the authors.' It seems our response may not have reached you successfully. We would like to share and expand upon our response here:

(1) Table 1 and Table supplementary 1-3 contain many correlation results. But what are the main points of these values? Which values do the authors want to highlight? Why are only p-values shown with significance symbols in Table supplementary 2??

(1.1) What are the main points of these values?

Thank reviewer for pointing this out. These correlations represent the relationship between behavior task (SI/BDT) and resting-state functional connectivity. It indicates that left hMT+ is involved in the efficient information integration network when it comes to BDT task. In addition, left hMT+’s surround suppression is involved in several hMT+ - frontal connectivity. Furthermore, the overlap regions between two task indicates the underlying mechanism.

(1.2) Which values do the authors want to highlight?

Table 1 and Table Supplementary 1-3 present the preliminary analysis results for Table 2 and Table Supplementary 4-6. So, we generally report all value. Conversely, in the Table 2 and Table Supplementary 4-6, we highlight the value which support our main conclusion.

(1.3) Why are only p-values shown with significance symbols in Table Supplementary 2?

Thank you for pointing this out, it is a mistake. We have revised it and delete the significance symbols.

(2) Line 27, it is unclear to me what is "the canonical theory".

We thank reviewer for pointing this out. We have revised “the canonical theory" to “the prevailing opinion” (line 27)

(3) Throughout the paper, the authors use "MT+", I would suggest using "hMT+" to indicate the human MT complex, and to be consistent with the human fMRI literature.

We thank reviewer for pointing this out. We have revised them.

(4) At the beginning of the results section, I suggest including the total number of subjects. It is confusing what "31/36 in MT+, and 28/36 in V1" means.

We thank reviewer for pointing this out. We have included the total number of subjects in the beginning of result section. (line 110, line 128)

(5) Line 138, "This finding supports the hypothesis that motion perception is associated with neural activity in MT+ area". This sentence is strange because it is a well-established finding in numerous human fMRI papers. I think the authors should be more specific about what this finding implies.

We thank reviewer for pointing this out. We have revised it to:” This finding is in line with prior results, which indicates that motion perception is associated with neural activity in hMT+ area, but not in EVC (primarily in V1)” (lines 156-158)

(6) There are no unit labels for all x- and y-axies in Figure 1. I only see the unit for Conc is mmol per kg wet weight.

We thank reviewer for pointing this out. Figure 1 is a schematic and workflow chart, so labels for x- and y-axes are not needed. I believe this confusion might pertain to Figure 3. In Figures 3a and 3b, the MRS spectrum does not have a standard y-axis unit as it varies based on the individual physical conditions of the scanner; it is widely accepted that no y-axis unit is used. While the x-axis unit is ppm, which indicate the chemical shift of different metabolites. In Figure 3c, the BDT represents IQ scores, which do not have a standard unit. Similarly, in Figures 3d and 3e, the Suppression Index does not have a standard unit.

(7) Although the correlations are not significant in Figure Supplement 2&3, please also include the correlation line, 95% confidence interval, and report the r values and p values (i.e., similar format as in Figure 1C).

We thank reviewer for pointing this out. We have revised them and include the correlation line, 95% confidence interval, r values and p values.

(8) There is no need to separate different correlation figures into Figure Supplementary 1-4. They can be combined into the same figure.

We thank reviewer for the suggestion. However, each correlation figure in the supplementary figures has its own specific topic and conclusion. Please notes that in the revised version, we have added a figure showing the EVC (primarily in V1) MRS scanning ROI as Supplementary Figure 1. Therefore, the figures the reviewer is concerned about are Supplementary Figure 2-5. The correlation figures in Supplementary Figure 2 indicate that GABA in EVC (primarily in V1) does not show any correlation with BDT and SI, illustrating that inhibition in EVC (primarily in V1) is unrelated to both 3D visuo-spatial intelligence and motion suppression processing. The correlations in Supplementary Figure 3 indicate that the excitation mechanism, represented by Glutamate concentration, does not contribute to 3D visuo-spatial intelligence in either hMT+ or EVC (primarily in V1). Supplementary Figure 4 validates our MRS measurements. Supplementary Figure 5 addresses potential concerns regarding the impact of outliers on correlation significance. Even after excluding two “outliers” from Figures 3d and 3e, the correlation results remain stable.

(9) Line 213, as far as I know, the study (Melnick et al., 2013) is a psychophysical study and did not provide evidence that the spatial suppression effect is associated with MT+.

We thank reviewer for pointing this out. It was a mistake to use this reference, and we have revised it accordingly. (line 242)

(10) At the beginning of the results, I suggest providing more details about the motion discrimination tasks and the measurement of the BDT.

We thank reviewer for pointing this out. We have included some brief description of task in the beginning of result section. (lines 116-120)

(11) Please include the absolute duration thresholds of the small and large sizes of all subjects in Figure 1.

We thank reviewer for the suggestion. We have included these results in Figure 3.

(12) Figure 5 is too small. The items in plot a and b can be barely visible.

We thank reviewer for pointing this out. We increase the size and resolution of the Figure.

Reviewer #3 (Public Review):

(1) Throughout the manuscript, hMT+ connectivity with the frontal cortex has been treated as an a priori hypothesis/space. However, there is no such motivation or background literature mentioned in the Introduction. Can the authors clarify the necessity of functional connectivity? In other words, can BOLD activity of hMT+ in the localizer task substitute for functional connectivity between hMT+ and the frontal cortex?

(1.1) Throughout the manuscript, hMT+ connectivity with the frontal cortex has been treated as an a priori hypothesis/space. However, there is no such motivation or background literature mentioned in the Introduction. Can the authors clarify the necessity of functional connectivity?

We thank reviewer for pointing this out. We offered additional motivation and background literature in the introduction: “Frontal cortex is usually recognized as the cognitive core region (Duncan et al., 2000; Gray et al., 2003). Strong connectivity between the cognitive regions suggests a mechanism for large-scale information exchange and integration in the brain (Barbey, 2018; Cole et al., 2012).  Therefore, the potential conjunctive coding may overlap with the inhibition and/or excitation mechanism of hMT+. Taken together, we hypothesized that 3D visuo-spatial intelligence (as measured by BDT) might be predicted by the inhibitory and/or excitation mechanisms in hMT+ and the integrative functions connecting hMT+ with frontal cortex (Figure 1a).” (lines 67-74). Additionally, we have included a whole-brain analysis for validation. Functional connectivity reveals the information exchange relationships across regions, enhancing our understanding of how hMT+ and the frontal cortex collaborate when solving visual-spatial intelligence tasks.

(1.2) In other words, can BOLD activity of hMT+ in the localizer task substitute for functional connectivity between hMT+ and the frontal cortex?

We thank the reviewer for this question. The localizer task was used solely for defining the hMT+ MRS scanning area. Functional connectivity was measured using resting-state fMRI. Research has shown that resting-state functional connectivity between the frontal cortex and other ROIs can further reveal the neural mechanisms underlying intelligence tasks (Song et al., 2008).

(2) There is an obvious mismatch between the in-text description and the content of the figure:<br /> "In contrast, there was no correlation between BDT and GABA levels in V1 voxels (figure supplement 1a). Further, we show that SI significantly correlates with GABA levels in hMT+ voxels (r = 0.44, P = 0.01, n = 31, Figure 3d). In contrast, no significant correlation between SI and GABA concentrations in V1 voxels was observed (figure supplement 1b)."

We thank reviewer for pointing this out. We have revised it. The revised version is :” In contrast, there was no correlation between BDT and GABA levels in V1 voxels (figure supplement 2a). Further, we show that SI significantly correlates with GABA levels in hMT+ voxels (r = 0.44, P = 0.01, n = 31, Figure 3d). In contrast, no significant correlation between SI and GABA concentrations in V1 voxels was observed (figure supplement 2b).” (lines 151-156)

(3) The authors' response to my previous round of review indicated that the "V1 ROIs" covered a substantial amount of V3 (32%). Therefore, it would no longer be appropriate to call these "V1 ROIs". I'd suggest renaming them as "Early Visual Cortex (EVC) ROIs" to be more accurate. Can the authors justify why choosing the left hemisphere for visual intelligence task, which is typically believed to be right lateralized?

(3.1) The authors' response to my previous round of review indicated that the "V1 ROIs" covered a substantial amount of V3 (32%). Therefore, it would no longer be appropriate to call these "V1 ROIs". I'd suggest renaming them as "Early Visual Cortex (EVC) ROIs" to be more accurate.

We thank the reviewer for pointing this out. We have revised our description of the MRS scanning ROIs to Early Visual Cortex (EVC). Since the majority of our EVC ROIs are in V1 (around 70%) and almost no V2 was included, we decided to mark the EVC ROIs with the explanation "primarily in V1" for better clarification. This terminology has been widely used to better emphasize the V1-based experimental design.

(3.2) Can the authors justify why choosing the left hemisphere for visual intelligence task, which is typically believed to be right lateralized?

We thank the reviewer for pointing this out. The use of the left MT/V5 as a target was motivated by studies demonstrating that left MT+/V5 TMS is more effective at causing perceptual effects (Tadin et al., 2011). Therefore, we chose to use the left hMT+ as our MRS ROI and maintain consistency across different models' ROIs. Additionally, our results support the notion that the visual intelligence task is right lateralized in the frontal cortex. At the resting-fMRI level, we found that significant ROIs, where functional connectivity is highly correlated with BDT scores, are in the right frontal cortex (Figure 5a, b).

(4) "Small threshold" and "large threshold" are neither standard descriptions, and it is unclear what "small threshold" refers to in the following figure caption. Additionally, the unit (ms) is confusing. Does it refer to timing?<br /> "(f) Peason's correlation showing significant negative correlations between BDT and small threshold."

Thank you for pointing this out; we agree with your suggestion. We have revised the terms “small threshold” and “large threshold” to “duration threshold of small grating” and “duration threshold of large grating”, respectively. The unit (ms) refers to timing. The details are described in the methods section: “The duration was adaptively adjusted in each trial, and duration thresholds were estimated using a staircase procedure. Thresholds for large and small gratings were obtained from a 160-trial block that contained four interleaved 3-down/1-up staircases. For each participant, we computed the correct rate for different stimulus durations separately for each stimulus size. These values were then fitted to a cumulative Gaussian function, and the duration threshold corresponding to the 75% correct point on the psychometric function was estimated for each stimulus size”.

(5) In the response letter, the authors mentioned incorporating the neural efficiency hypothesis in the Introduction, but the revised Introduction does not contain such information.

We thank the reviewer for pointing this out. In our revised version, the second paragraph of the introduction addresses the neural efficiency hypothesis: “The “neuro-efficiency” hypothesis is one explanation for individual differences in gF (Haier et al., 1988). This hypothesis puts forward that the human brain’s ability to suppress irrelevant information leads to more efficient cognitive processing. Correspondingly, using a well-known visual motion paradigm (center-surround antagonism) (Liu et al., 2016; Tadin et al., 2003), Melnick et al found a strong link between suppression index (SI) of motion perception and the scores of the block design test (BDT, a subtest of the Wechsler Adult Intelligence Scale (WAIS), which measures the visuo-spatial component (3D domain) of gF (Melnick et al., 2013). Motion surround suppression (SI), a specific function of human extrastriate cortical region, middle temporal complex (hMT+), aligns closely with this region's activities (Gautama & Van Hulle, 2001). Furthermore, hMT+ is a sensory cortex involved in visual perception processing (3D domain) (Cumming & DeAngelis, 2001). These findings suggest that hMT+ potentially plays a significant role in 3D visuo-spatial intelligence by facilitating the efficient processing of 3D visual information and suppressing irrelevant information. However, more evidence is needed to uncover how the hMT+ functions as a core region for 3D visuo-spatial intelligence.” (lines 51-66)

Recommendations for the authors:

Reviewer #1 (Recommendations for The Authors):

In the Code availability, it states that "this paper does not report original code". It seems weird because at least the code to reproduce the figures from the data should be provided.

Thank you for pointing this out. Almost all figures were created using software such as DPABI, BrainNet, and GraphPad Prism 9.5, which are manually operated and do not require code adjustments. However, for the MRS fitting curve, we can provide our MATLAB code for redrawing the MRS fitting. The code has been uploaded to GitHub.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

In this work, Qiu and colleagues examined the effects of preovulatory (i.e., proestrous or late follicular phase) levels of circulating estradiol on multiple calcium and potassium channel conductances in arcuate nucleus kisspeptin neurons. Although these cells are strongly linked to a role as the "GnRH pulse generator," the goal here was to examine the physiological properties of these cells in a hormonal milieu mimicking late proestrus, the time of the preovulatory GnRH-LH surge. Computational modeling is used to manipulate multiple conductances simultaneously and support a role for certain calcium channels in facilitating a switch in firing mode from tonic to bursting. CRISPR knockdown of the TRPC5 channel reduced overall excitability, but this was only examined in cells from ovariectomized mice without estradiol treatment. The patch clamp experiments are comprehensive and overall solid but a direct demonstration of the role of these conductances in being necessary for surge generation (or at least having a direct physiological consequence on surge properties) is lacking, substantially reducing the impact of the findings.

Strengths:

(1) Examination of multiple types of calcium and potassium currents, both through electrophysiology and molecular biology.

(2) Focus on arcuate kisspeptin neurons during the surge is relatively conceptually novel as the anteroventral periventricular nucleus (AVPV) kisspeptin neurons have received much more attention as the "surge generator" population.

(3) The modeling studies allow for direct examination of manipulation of single and multiple conductances, whereas the electrophysiology studies necessarily require examination of each current in isolation. The construction of an arcuate kisspeptin neuron model promises to be of value to the reproductive neuroendocrinology field.

We thank the reviewer for recognizing our comprehensive examination of Kiss-ARH neurons through electrophysiological, molecular and computational modeling of their activity during the preovulatory surge, which as the reviewer pointed out is “conceptually novel.”  We  have bolstered our argument that Kiss1-ARH neurons transition from synchronized firing to burst firing with the E2-mediated regulation of channel expression with the addition of new experiments. We have addressed the recommendations as follows:

Weaknesses:

(1) The novelty of some of the experiments needs to be clarified. This reviewer's understanding is that prior experiments largely used a different OVX+E2 treatment paradigm mimicking periods of low estradiol levels, whereas the present work used a "high E2" treatment model. However, Figures 10C and D are repeated from a previous publication by the same group, according to the figure legend. Findings from "high" vs. "low" E2 treatment regimens should be labeled and clearly separated in the text. It would also help to have direct comparisons between results from low E2 and high E2 treatment conditions.

We have revised Figures 10C and 10D to include new findings (only) on Tac2 and Vglut2 expression in OVX and E2-treated Kiss1ARH.  Most importantly, our E2 treatment regime is clearly stated in the Methods and is exactly the same that was used previously (Qiu, eLife 2016 and Qiu, eLife 2018) for the induction of the LH surge in OVX mice (Bosch, Molecular and Cellular Endocrinology 2013) .

(2) In multiple places, links are made between the changes in conductances and the transition from peptidergic to glutamatergic neurotransmission. However, this relationship is never directly assessed. The data that come closest are the qPCR results showing reduced Tac2 and increased Vglut2 mRNA, but in the figure legend, it appears that these results are from a prior publication using a different E2 treatment regimen.

In the revised Figure 1, we have now included a clear depiction of the transition from synchronized firing driven by NKB signaling in OVX females to burst firing driven by glutamate in E2-treated females. All of the qPCR results in the revised manuscript are new.  We have used the same E2 treatment paradigm as previously published (Qiu, eLife 2018).

(3) Similarly, no recordings of arcuate-AVPV glutamatergic transmission are made so the statements that Kiss1ARH neurons facilitate the GnRH surge via this connection are still only conjecture and not supported by the present experiments.

Using a horizontal hypothalamic slice preparation, we have shown that Kiss1-ARH neurons excite GnRH neurons via Kiss1ARH glutaminergic input to Kiss1AvPV/Pen neurons (summarized in Fig. 12, Qiu, eLife 2016). We did not think that it was necessary to repeat these experiments for the current manuscript.

(4) Figure 1 is not described in the Results section and is only tenuously connected to the statement in the introduction in which it is cited. The relevance of panels C and D is not clear. In this regard, much is made of the burst firing pattern that arises after E2 treatment in the model, but this burst firing pattern is not demonstrated directly in the slice electrophysiology examples.

We have extensively revised Figure 1 to include new whole-cell, current clamp recordings that document burst firing  in  E2-treated, OVX females, which is now cited in the Results.

(5) In Figure 3, it would be preferable to see the raw values for R1 and R2 in each cell, to confirm that all cells were starting from a similar baseline. In addition, it is unclear why the data for TTA-P2 is not shown, or how many cells were recorded to provide this finding.

Before initiating photo-stimulation for each Kiss1-ARH neuron, we adjust the resting membrane potential to -70 mV, as noted  in each panel in Figure 3, through current injections. We have now included new findings on the effects of the T-channel blocker TTA-P2 on slow EPSP in the revised Figure 3. The number of cells tested with each calcium channel blocker is depicted in each of the bar graphs summarizing the effects of the blockers (Figure 3E).

(6) In Figure 5, panel C lists 11 cells in the E2 condition but panel E lists data from 37 cells. The reason for this discrepancy is not clear.

In Figure 5D, we measured the L-, N-, P/Q and R channel currents after pretreatment with TTA-P2 to block the T-type current, whereas in Figure 5C, we measured the total current without TTA-P2.

(7) In all histogram figures, it would be preferable to have the data for individual cells superimposed on the mean and SEM.

In the revised Figures we have included the individual data points for the individual neurons and animals (qPCR). 

(8) The CRISPR experiments were only performed in OVX mice, substantially limiting interpretation with respect to potential roles for TRPC5 in shaping arcuate kisspeptin neuron function during the preovulatory surge.

The TRPC5 channels are most  important for generating slow EPSPs when expression of NKB is high in the OVX state. Conversely, the glutamatergic response becomes more significant when the expression of NKB and TRPC5 channel are muted in the E2-treated state. Therefore, the CRISPR experiments were specifically conducted in OVX mice to maximize the effects.

(9) Furthermore, there are no demonstrations that the CRISPR manipulations impair or alter the LH surge.

In this manuscript, our focus is on the cellular electrophysiological activity of the Kiss1ARH neurons in OVX and E2-treated OVX females. Exploration of CRISPR manipulations related to the LH surge is certainly slated for future  experiments, but these in vivo experiments are  beyond the scope of these comprehensive cellular electrophysiological and molecular studies.

(10) The time of day of slice preparation and recording needs to be specified in the Methods.

We have provided the times of slice preparation and recordings in the revised Methods and Materials.

Reviewer #2 (Public Review):

Summary:

Kisspeptin neurons of the arcuate nucleus (ARC) are thought to be responsible for the pulsatile GnRH secretory pattern and to mediate feedback regulation of GnRH secretion by estradiol (E2). Evidence in the literature, including the work of the authors, indicates that ARC kisspeptin coordinate their activity through reciprocal synaptic interactions and the release of glutamate and of neuropeptide neurokinin B (NKB), which they co-express. The authors show here that E2 regulates the expression of genes encoding different voltage-dependent calcium channels, calcium-dependent potassium channels, and canonical transient receptor potential (TRPC5) channels and of the corresponding ionic currents in ARC kisspeptin neurons. Using computer simulations of the electrical activity of ARC kisspeptin neurons, the authors also provide evidence of what these changes translate into in terms of these cells' firing patterns. The experiments reveal that E2 upregulates various voltage-gated calcium currents as well as 2 subtypes of calcium-dependent potassium currents while decreasing TRPC5 expression (an ion channel downstream of NKB receptor activation), the slow excitatory synaptic potentials (slow EPSP) elicited in ARC kisspeptin neurons by NKB release and expression of the G protein-associated inward-rectifying potassium channel (GIRK). Based on these results, and on those of computer simulations, the authors propose that E2 promotes a functional transition of ARC kisspeptin neurons from neuropeptide-mediated sustained firing that supports coordinated activity for pulsatile GnRH secretion to a less intense firing in glutamatergic burst-like firing pattern that could favor glutamate release from ARC kisspeptin. The authors suggest that the latter might be important for the generation of the preovulatory surge in females.

Strengths:

The authors combined multiple approaches in vitro and in silico to gain insights into the impact of E2 on the electrical activity of ARC kisspeptin neurons. These include patch-clamp electrophysiology combined with selective optogenetic stimulation of ARC kisspeptin neurons, reverse transcriptase quantitative PCR, pharmacology, and CRIPR-Cas9-mediated knockdown of the Trpc5 gene. The addition of computer simulations for understanding the impact of E2 on the electrical activity of ARC kisspeptin cells is also a strength.

The authors add interesting information on the complement of ionic currents in ARC kisspeptin neurons and on their regulation by E2 to what was already known in the literature. Pharmacological and electrophysiological experiments appear of the highest standards. Robust statistical analyses are provided throughout, although some experiments (illustrated in Figures 7 and 8) do have rather low sample numbers.

The impact of E2 on calcium and potassium currents is compelling. Likewise, the results of Trpc5 gene knockdown do provide good evidence that the TRPC5 channel plays a key role in mediating the NKB-mediated slow EPSP. Surprisingly, this also revealed an unsuspected role for this channel in regulating the membrane potential and excitability of ARC kisspeptin neurons.

We thank the reviewer for recognizing that the “pharmacological and electrophysiological experiments appear of the highest standards” and “the addition of the computer modeling for understanding the impact of E2 on the electrical activity of ARC kisspeptin cells is also a strength.  However, we agree with the reviewer that we needed to provide a direct demonstration of “burst-like” firing of Kiss1-ARH neurons, which we have provided in Figure 1. We have addressed the other recommendations as follows:

Weaknesses:

The manuscript also has weaknesses that obscure some of the conclusions drawn by the authors.

One has to do with the fact that "burst-like" firing that the authors postulate ARC kisspeptin neurons transition to after E2 replacement is only seen in computer simulations, and not in slice patch-clamp recordings. A more direct demonstration of the existence of this firing pattern, and of its prominence over neuropeptide-dependent sustained firing under conditions of high E2 would make a more convincing case for the authors' hypothesis.

We have provided  a more direct demonstration of the existence of this firing pattern in the whole-cell current clamp experiments in the revised Figure 1.

In addition, and quite importantly, the authors compare here two conditions, OVX versus OVX replaced with high E2, that may not reflect the physiological conditions (the diestrous [low E2] and proestrous [high E2] stages of the estrous cycle) under which the proposed transition between neuropeptide-dependent sustained firing and less intense burst firing might take place. This is an important caveat to keep in mind when interpreting the authors' findings. Indeed, that E2 alters certain ionic currents when added back to OVX females, does not mean that the magnitude of these ionic currents will vary during the estrous cycle.

We have published that the magnitude of the slow EPSP, which is TRPC5 channel mediated, varies throughout the estrous cycle with the slow EPSP reaching a maximal amplitude during diestrus, which was significantly reduced during proestrus,  similar to that found in OVX compared to E2-treated, OVX females (Figure 2, Qiu, eLife 2016).  Moreover, TRPC5 channel mRNA expression,  similar to the peptides, is downregulated by an E2 treatment (Figure 10 this manuscript) that mimics proestrus levels of the steroid (Bosch et al., Mol Cell Endocrinology 2013). Furthermore, the magnitude of ionic currents is directly proportional to the number of ion channels expressed in the plasma membrane, which we have found correlates with mRNA expression. Therefore, it is likely that the magnitude of these ionic currents will vary during the estrous cycle.

Lastly, the results of some of the pharmacological and genetic experiments may be difficult to interpret as presented. For example, in Figure 3, although it is possible that blockade of individual calcium channel subtypes suppresses the slow EPSP through decreased calcium entry at the somato-dendritic compartment to sustain TRPC5 activation and the slow depolarization (as the authors imply), a reasonable alternative interpretation would be that at least some of the effects on the amplitude of the slow EPSP result from suppression of presynaptic calcium influx and, thus, decreased neurotransmitter and neuropeptide secretion. Along the same lines, in Figure 12, one possible interpretation of the observed smaller slow EPSPs seen in mice with mutant TRPC5 could be that at least some of the effect is due to decreased neurotransmitter and neuropeptide release due to the decreased excitability associated with TRPC5 knockdown.

The reviewer raises a good point, but our previous findings clearly demonstrated that chelating intracellular calcium with BAPTA in whole-cell current clamp recordings abolishes the slow EPSP and persistent firing (Qiu et al., J. Neurosci 2021), which we have noted is the  rationale for dissecting out the contribution of T, R, N, L and P/Q calcium channels to the slow EPSP in our current studies.  The revised Figure 3 also includes the effects of T-channel blocker.

However, to further bolster the argument for the post-synaptic contribution of the calcium channels to the slow EPSP  and eliminate the potential presynaptic effects of the calcium channel blockers on the postsynaptic slow EPSP amplitude, which may result from reduced presynaptic calcium influx and subsequently decreased neurotransmitter release, we have utilized an additional strategy. Specifically, we have measured the response to the externally administered TACR3 agonist senktide under conditions in which the extracellular calcium influx, as well as neurotransmitter and neuropeptide release, are blocked (revised Figure 3).

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

(1) The use of optogenetics in Figure 3 to trigger the slow EPSP could be better clarified in the text.

We have clarified in the Methods the optogenetic protocol for generating the slow EPSP, which we have published previously (Qiu et al., eLife 2016; eLife 2018, J. Neurosci 2021).

(2) The citation for Figure 4C in the text does not match what is shown in the figure.

Figure 4C has been removed in the revised manuscript.

(3) Figure 5 - it would be clearer to have panel D labeled as "model results" or similar to distinguish it from the slice recording data.

Panel D has been labeled as "Model results”.

(4) The text in lines 191-197 in the Results may be better suited to the Discussion.

We have modified the text in order to present the new findings without the discussion points.

(5) It is somewhat confusing to have figure panels cited out of order in the main text (e.g., 7H before 7G and 8H before 8G).

We have edited the text to report the findings in the proper order of the panels in Figures 7 and 8.

Reviewer #2 (Recommendations For The Authors):

- The observations that E2 treatment of OVX mice has an effect on the magnitude of a number of ionic currents does not necessarily mean that these changes will be seen during the estrous cycle, in response to fluctuations in circulating E2 concentrations. Experiments comparing either different estrous cycle stages or OVX mice treated with low or high E2 would be required to gain insight into this question. As such, the relevance of the authors' findings (however interesting these are as they stand) to any potential physiological endocrine/reproductive state transition is questionable, in the reviewer's opinion. The authors should acknowledge this important caveat and moderate the interpretations of their findings and the conclusions of their manuscript accordingly.

We have published that the magnitude of the slow EPSP, which is TRPC5 channel mediated, varies throughout the estrous cycle with the slow EPSP being large during diestrus and significantly reduced during proestrus,  similar to that found in OVX compared to E2-treated, OVX females (Figure 2, Qiu, eLife 2016).  Moreover, TRPC5 channel mRNA expression,  similar to the peptides, is downregulated by an E2 treatment (Figure 10 this manuscript) that mimics proestrus levels of the steroid (Bosch et al., Mol Cell Endocrinology 2013). Furthermore, the magnitude of ionic currents is directly proportional to the number of ion channels expressed in the plasma membrane, which we have found correlates with mRNA expression. Therefore, it is likely that the magnitude of these ionic currents will vary during the estrous cycle.

- The bursting firing pattern that the authors refer to and postulate will favor glutamate release under high E2 conditions is only seen in the computer simulations, not in patch-clamp recordings in brain slices (see also comment below). This substantially weakens some of the conclusions of the manuscript. Unless the authors can convincingly demonstrate a change in ARC kisspeptin firing pattern in response to increasing E2 using electrophysiology, these conclusions should be moderated.

We now include examples of burst firing activity under E2-treatment conditions in Figure 1 and have included summary figure (pie chart) documenting that a significant percentage of cells exhibit this activity with E2 treatment.  

Other comments:

- Title: "E2 elicits distinct firing patterns" is not shown in this work. As such, the title needs to be revised.

We now show these distinct firing patterns in Figure 1, so we think the wording in the title is an accurate reflection of our findings. 

- Abstract: some of the interpretations are overstated, in the reviewer's opinion.

Line 23, "... elevating the whole-cell calcium current and contributing to high-frequency firing" should be moderated, as what is shown by the authors is that blockade of calcium channel subtypes suppresses the slow EPSP and associated firing, the frequency of which is not reported (see also a later comment).

We now include examples of burst firing activity under E2-treatment conditions in Figure 1 and have modified the abstract to state “high frequency burst firing.”

Lines 26-28, that "mathematical modeling confirmed the importance of TRPC5 channels for initiating and sustaining synchronous firing, while GIRK channels, activated by Dyn binding to kappa opioid receptors, were responsible for repolarization" is simply not what the simulations show, in the reviewer's opinion. Indeed, there is no consideration of synchronous activity in the model, which simulates the firing of a single ARC kisspeptin neuron. Further, the model shows that TRPC5 can contribute to overall excitability (firing in response to current injection, Figure 12G) and that increasing TRPC5 conductance increases firing in response to NKB while this is decreased by adding GIRK conductance to the model (Figure 13A). Therefore, considerations of the importance of TRPC5 channels in initiating synchronous firing and the role of Dyn A-induced GIRK activity should not be included in the interpretations of the mathematical simulations.

The significance of synchronization lies in the fact that when neuronal networks synchronize, the behavior of each neuron within the network becomes identical. In such scenarios, the firing of a single neuron mirrors the activity of the entire neuronal network. Consequently, our model simulations, based on a single-cell neuronal model, can be utilized to make reliable inferences about synchronized neuronal activity.

Lines 31-33 (also lines 92-95), that "the transition to burst firing with high, preovulatory levels of E2 facilitates the GnRH surge through its glutamatergic synaptic connection to preoptic Kiss1 neurons" is not supported by the experiments (physiologic or computational) described in the manuscript, and is, therefore, only speculative. These statements should be removed throughout the manuscript.

Previously, we (Qiu et al., (eLife 2016) documented a direct glutamatergic projection from Kiss1-ARH neurons to Kiss1-AVPV/PeN neurons.  Moreover, Lin et al. (Frontiers Endocrinology 2021) demonstrated that low frequency stimulation of Kiss1-ARH:ChR2 neurons, that is known to only release glutamate, boosts the LH surge, and in a follow-up paper the O’Byrne lab blocked this stimulation with ionotropic glutamate antagonists (Shen et al., Frontiers in Endocrinology 2022).  We have included these references in the Introduction and Discussion, but we did not think that it was necessary to cite these papers in the Abstract.  However, we have re-worded this final statement in the Abstract to: “the transition to burst firing with high, preovulatory levels of E2 would facilitate the GnRH surge….” 

- Introduction: the usefulness of Figure 1 is questionable. From reading the figure legend, it is the reviewer's understanding that panels A and B are published elsewhere (there is no description of methods or results in the manuscript). Further, panels C and D are meant to illustrate that ARC kisspeptin neurons display different types of firing in OVX vs E2-treated OVX mice. The legend to C indicates that the trace illustrates "synchronous firing" but shows one cell (how can this be claimed as synchronous?) - the legend to D indicates that the trace "demonstrates" burst firing in ARC kisspeptin neurons. This part of the figure is, in the reviewer's opinion, misleading because these are only two examples (no quantifications or replicates are provided) obtained by stimulating firing in two different endocrine conditions by two different agonists. The "demonstration" of differential firing patterns would require a thorough examination of firing patterns in response to current injections (as in Figure 12 E-F) or in response to the two agonists, under the different hormonal conditions.

Figure 1 has now been completely revised to include new data documenting the different firing patterns.  The methods detailing these experiments can be found in the Material and Methods section.

The introduction presents a rather incomplete picture of what is known regarding how ARC kisspeptin neurons might coordinate their activity to drive episodic GnRH secretion, and it omits published work showing that blockade of glutamate receptors (in particular AMPA receptors) decreases ARC kisspeptin neuron coordinated activity in the brain slices and in vivo and suppresses pulsatile GnRH/LH secretion in mice.

If we are not mistaken, the reviewer is referring to fiber photometry recordings of GCaMP activity, which we cite in the Discussion.  However, for the Introduction we tried to “set the stage” for our studies on measuring the individual channels underlying the different firing patterns and how they are regulated by E2.

The introduction is also quite long with extensive descriptions of previous work by the authors and in other brain areas that would be better suited for the discussion.

Again, we are trying to rationalize why we focused on particular ion channels based on the literature.

- Results: lines 129-132 should be moderated, as whether calcium channels increase excitability or facilitate TRPC5 channel opening has not been directly assessed here.

High frequency optogenetic stimulation of Kiss1-ARH neurons and NKB through its cognate receptor (TACR3) activates TRPC 5 channels (Qiu et al., eLife 2016; J. Neurosci 2021). BAPTA prevents the opening of TRPC5 channels and abrogates the slow EPSP following high frequency stimulation.  Figure 3 documents that inhibition of voltage-activated calcium channels attenuates the slow EPSP, which results in a decrease in excitability.

Lines 145-146, one limitation of this experiment is that blockade of calcium channel subtypes will not only affect calcium entry and subsequent actions of calcium on TRPC5 channels but also impair the release of neurotransmitters and neuropeptides from kisspeptin neurons. The interpretation that "calcium channels contribute to maintaining the sustained depolarization underlying the slow EPSP" needs, therefore, to be moderated as it is not possible to extract the direct contribution of calcium channels to the activation of TRPC5 channels from these experiments.

We cited our previous findings documenting that chelating intracellular calcium with BAPTA abolishes the slow EPSP and persistent firing (Qiu et al., J Neurosci 2021).  However, to eliminate the potential effects of calcium channel blockers on the slow EPSP amplitude, which may result from reduced presynaptic calcium influx and subsequently decreased neurotransmitter and neuropeptide secretion, we adopted a different strategy by comparing responses between Senktide and Cd2+ plus Senktide. Our findings revealed that the non-selective Ca2+ channel blocker Cd2+ significantly inhibited Senk-induced inward current (Figures 3F-H).

Panel C should be removed from Figure 4, as it is published elsewhere.

Figure 4C has been removed.

Lines 168-169, "...E2 treatment led to a significant increase in the peak calcium current density in Kiss1ARH neurons, which was recapitulated as predicted by our computational modeling..." How did the model "predict" this increase in calcium current density? As no information is provided in the methods or supplementary information as to how the effect of E2 was integrated into the model, the authors will need to provide additional narration in the text to explain this statement. The "T-channel inflection" referred to in the figure legend will also need to be explained. Lastly, in Figure 5C, the current density unit should be pA/pF. 

We have added text in the supplementary information to explain how we used the qPCR and electrophysiological data to inform the model regarding the effect that E2 has on the various ionic currents and noted in the Figure 13 legend that the increase/decrease in the conductances is physiologically mediated by E2. We have eliminated the T-channel inflection point (Figure 5D) and corrected the current density label (Figure 5C).

Lines 198-199, please clarify "E2 does not modulate calcium channel kinetics directly but rather alters the mRNA expression to increase the conductance".

We have clarified that “that long-term E2 treatment does not modulate calcium channel kinetics but rather alters the mRNA expression to increase the calcium channel conductance” by referring to the specific figures (i.e., Figures 4, 6) in a previous sentence.

Figures 7 and 8 titles do not accurately reflect the contents: there is nothing about repolarization in the experiments illustrated in Figure 7 or Figure 8. The sample sizes (3 to 4 cells) are also quite small for these experiments.

We have modified the Figure titles per the reviewer’s comments and increased the cell numbers.

The title of Figure 9 also does not fully reflect the figure's contents. Although panel G does suggest that the M current contributes to regulating the membrane potential, the reviewer's reading of this figure panel is that the fractional contribution of the M current does not vary during a short burst of action potentials. The suggestion that "KCNQ channels play a key role in repolarizing Kiss1ARH neurons following burst firing" (line 272) and the statement that "our modeling predicted that M-current contributed to the repolarization following burst firing" (line 273) should be revised accordingly.

The point is that the M-current contributes, albeit a small fraction, to the repolarization during burst firing.

Line 288, please indicate what figure informs this statement.

We have revised the statement since the modeling (Figure 13) comes later in the Results.

Line 311-313, this sentence only superficially describes the simulation, in the reviewer's opinion. Does the model inform on how TRPC5 channels/currents do that? The supplementary information indicates that there is a tone of extracellular neurokinin B embedded in the model. This is important information that should be clearly stated in the manuscript. The authors should also consider discussing the influence of this neurokinin B tone on the contribution of TRPC5 to cell excitability. As a neurokinin B tone in the extracellular space will likely alter the firing of kisspeptin neurons in the model, readers will likely need more information about all this.

In our current ramp simulations of the model (Fig 12 G&H) there is no involvement of neurokinin B (i.e., the NKB parameter  is set to zero), and the effect on the rheobase is solely due to the decrease of the TRPC5 conductance.  In the model, TRPC5 channels are activated by intracellular calcium levels and are therefore contributing to cell excitability even in the absence of extracellular NKB. The NKB tone is used for the simulations presented in Figure 13 where we vary the TRPC5 conductance under saturating levels of extracellular NKB.

Lines 316-318 also read as quite superficial. More explanations of what is illustrated in Figure 13 are needed. In particular, it is unclear from the methods and supplementary information what the different ratios of conductances in OVX+E2 vs in OVX are and how they were varied in the model. Furthermore, it is unclear to the reviewer how the outcome of these simulations matches the authors' postulate that E2 enables a transition to a burst firing pattern that favors glutamate release. Looking at simulated firing in Figure 13B, E2 (by increasing calcium conductances) would tend to enable high-frequency firing within bursts (nearing 50 Hz by eye) and high burst rates (approximately 4 bursts per second), which the reviewer would argue might be expected to cause significant neuropeptide release in addition to that of glutamate.

We have added to the text: “Furthermore, the burst firing of the OVX+E2 parameterized model was supported by elevated h- and Ca 2+-currents (Figure 13B) as well as by the high conductance of Ca2+ channels relative to the conductance of TRPC5 channels (Figure 13C).” We have also provided in the Supplemental Information (Table of Model Parameters) the specific conductances in the OVX and OVX+E2 state and how they are varied to produce the model simulations.

Granted the high frequency firing during a burst could release peptide, but in the E2-treated, OVX females the expression of the peptides are at “rock bottom.”  Therefore, the sustained high frequency firing during the slow EPSP in the OVX state would generate maximum peptide release.

In Figure 13C, the reviewer is unclear on the ranges of TRPC5 conductances shown. The in vitro experiments suggest that E2 suppresses Trpc5 gene expression and might suppress TRPC5 currents. The ratio of gTRPC5(OVX+E2)/gTRPC5(OVX) should, thus, be <1.0. This is not represented in the parameter space provided, making the interpretation of this simulation difficult. Please clarify what the effect of decreasing gTRPC5 will be on firing patterns in the model.

Thank you for pointing this typographical error.  The ratio should be gTRPC5 (OVX)/TRPC5(OVX + E2) for the X-axis.

- Discussion: many statements and conclusions are overreaching and need to be revised; for example lines 320-322, 329-330, 335-338, 369, 371-373, 391-394, 463-464, and 489-494;

We have tempered these statements, so they are not “overreaching.”

Lines 489-494: the authors should integrate published observations that i) ablation of ARC kisspeptin neurons results in increased LH surges in mice and rats and that ii) optogenetic stimulation of ARC kisspeptin fibers in the POA is only effective at increasing LH secretion in a surge-like manner when done at high frequencies (20 Hz), in their discussion of the role of ARC kisspeptin neurons and their firing patterns in the preovulatory surge.

We have included the paper from the O’Byrne lab (Shen et al. Frontiers in Endocrinology 2022) in the Discussion. However, the Mittleman-Smith paper (Endocrinology, 2016) ablating KNDy neurons using NK3-saporin not only targeted KNDy neurons but other arcuate neurons that express NK3 receptors.  Therefore, we have not cited it in the Discussion.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

Major comments: 

My main concern about the manuscript is the extent of both clinical and statistical heterogeneity, which complicates the interpretation of the results. I don't understand some of the antibiotic comparisons that are included in the systematic review. For instance the study by Paul et al (50), where vancomycin (as monotherapy) is compared to co-trimoxazole (as combination therapy). Emergence (or selection) of co-trimoxazole in S. aureus is in itself much more common than vancomycin resistance. It is logical and expected to have more resistance in the co-trimoxazole group compared to the vancomycin group, however, this difference is due to the drug itself and not due to co-trimoxazole being a combination therapy. It is therefore unfair to attribute the difference in resistance to combination therapy. Another example is the study by Walsh (71) where rifampin + novobiocin is compared to rifampin + co-trimoxazole. There is more emergence of resistance in the rifampin + co-trimoxazole group but this could be attributed to novobiocin being a different type of antibiotic than co-trimoxazole instead of the difference being attributed to combination therapy. To improve interpretation and reduce heterogeneity my suggestion would be to limit the primary analyses to regimens where the antibiotics compared are the same but in one group one or more antibiotic(s) are added (i.e. A versus A+B). The other analyses are problematic in their interpretation and should be clearly labeled as secondary and their interpretation discussed. 

Thank you for raising these important points and highlighting the need for clarification. We understand that the reviewer has concerns regarding the following points:

(1) The structure of presenting our analyses, i.e. main analyses and sub-group analyses and their corresponding discussion and interpretation

Our primary interest was whether combining antibiotics has an overarching effect on resistance and to identify factors that explain potential differences of the effect of combining antibiotic across pathogens/drugs. Therefore, pooling all studies, and thereby all combinations of antibiotics, is one of our main analyses. The decision to pool all studies that compare a lower number of antibiotics to a higher number of antibiotics was hence predefined in our previously published study protocol (PROSPERO CRD42020187257).

We indeed, find that heterogeneity is high in our statistical analyses. As planned in our study protocol, we did perform several prespecified sub-group analyses and added additional ones. We now emphasize that several sub-group analyses were performed to investigate heterogeneity (L 119ff): “The overall pooled estimates are based on studies that focus on various clinical conditions/pathogens and compare different antibiotics treatments. To explore the impact of these and other potential sources of heterogeneity on the resistance estimates we performed various sub-group analyses and metaregression.” 

The performed sub-group analyses specifically focused on specific pathogens/clinical conditions (figure 3) or explored heterogeneity due to different antibiotics in comparator arms – as suggested by the reviewer (figure 3B, SI section 6). We find that the heterogeneity remains high even if only resistances to antibiotics common to both arms are considered (SI section 6.1.8). With this analysis we excluded comparisons of different antibiotics (e.g., A vs B+C), such as those between vancomycin and cotrimoxazole named by the reviewer. While we aimed to explore heterogeneity and investigate potential factors affecting the effect of combining antibiotic on resistance, limitations arose due to limited evidence and the nature of data provided by the identified studies. Therefore, interpretability remains also limited for the subgroup analyses, which we highlight in the discussion. (L 186 ff: We accounted for many sources of heterogeneity using stratification and meta-regression, but analyses were limited by missing information and sparse data.) Further, specific subgroup analyses are discussed in more detail in the SI.

(2) Difference in resistance development due to the type of the antibiotics or due to combination therapy?

The reviewer raises an important point, which we also try to make: future studies should be systematically designed to compare antibiotic combination therapy, i.e. identical antibiotics in treatment arms should be used, except for additional antibiotics used in both treatment arms. We already mentioned this point in our discussion but highlight this now by emphasizing how many studies did not have identical antibiotics in their treatment arms. We write in L194ff: “19 (45%) of our included studies compared treatment arms with no antibiotics in common, and 22 studies (52%) had more than one antibiotic not identical in the treatment arms (table 1). To better evaluate the effect of combination therapy, especially more RCTs would be needed where the basic antibiotic treatment is consistent across both treatment arms, i.e. the antibiotics used in both treatment arms should be identical, except for the additional antibiotic added in the comparator arm (table 1).”

Furthermore, we investigated the importance of the type of antibiotics with several subgroup analyses (e.g. SI sections 6.1.8 and 6.1.10). We now further highlight the concern of the type of antibiotics in the result section of the main manuscript, where we discuss the sub-group analysis with no common antibiotics in the treatment arms 131 ff: “Furthermore, a lower number of antibiotics performed better than a higher number if the compared treatment arms had no antibiotics in common (pooled OR 4.73, 95% CI 2.14 – 10.42; I2\=37%, SI table S3), which could be due to different potencies or resistance prevalences of antibiotics as discussed in SI (SI section 6.1.10).” As mentioned above we also perform sub-group analyses, where only resistances of antibiotics common to both arms are considered (SI section 6.1.8). However, as discussed in the corresponding sections, the systematic assessment of antibiotic combination therapy remains challenging as not all resistances against antibiotics used in the arms were systematically measured and reported. Furthermore, the power of these sub-group analyses is naturally a concern, as they include fewer studies. 

Another concern is about the definition of acquisition of resistance, which is unclear to me. If for example meropenem is administered and the follow-up cultures show Enterococcus species (which is intrinsically resistant to meropenem), does this constitute acquisition of resistance? If so, it would be misleading to determine this as an acquisition of resistance, as many people are colonized with Enterococci and selection of Enterococci under therapy is very common. If this is not considered as the acquisition of resistance please include how the acquisition of resistance is defined per included study. Table S1 is not sufficiently clear because it often only contains how susceptibility testing was done but not which antibiotics were tested and how a strain was classified as resistant or susceptible. 

Thank you for pointing out this potential ambiguity. The definition of acquisition of resistance reads now (L 275 ff): “A patient was considered to have acquired resistance if, at the follow-up culture, a resistant bacterium (as defined by the study authors) was detected that was not present in the baseline culture.” We also changed the definition accordingly in the abstract (L 36 ff). We hope that the definition of acquisition is now clearer. Our definition of “acquisition of resistance” is agnostic to bacterial species and hence intrinsically resistant species, as the example raised by the reviewer, can be included if they were only detected during the follow-up culture by the studies. Generally, it was not always clear from the studies, which pathogens were screened for and whether the selection of intrinsically resistant bacteria was reported or not. Therefore, we rely on the studies' specifications of resistant and non-resistant without further distinction from our side, i.e. classifying data into intrinsic and non-intrinsic resistance. Overall, the outcome “acquisition of resistance” can be interpreted as a risk assessment for having any resistant bacterium during or after treatment. In contrast, the outcome “emergence of resistance” is more rigorous, demanding the same species to be detected as more resistant during or after treatment.

The information, which antibiotic susceptibility tests were performed in each individual study can be found in the main text in table 1. However, we agree that this information should be better linked and highlighted again in table S1. We therefore now refer to table 1 in the table description of table S1. L134 ff.: “See table 1 in the main text for which antibiotics the antibiotics tested and reported extractable resistance data”. Furthermore, we added the breakpoints for resistant and susceptible classification if specifically stated in the main text of the study. However, we did not do further research into old guidelines, manufactures manuals or study protocols in case the breakpoints are not specifically stated in the main text as the main goal of this table, in our opinion, is to show a justification, why the studies could be considered for a resistance outcome. We therefore decided against further breakpoint investigations for studies, where the breakpoint is not specifically stated in the main text. 

Line 85: "Even though within-patient antibiotic resistance development is rare, it may contribute to the emergence and spread of resistance." 

Depending on the bug-drug combination, there is great variation in the propensity to develop within-patient antibiotic resistance. For example: within-patient development of ciprofloxacin resistance in Pseudomonas is fairly common while within-patient development of methicillin resistance in S. aureus is rare. Based on these differences, large clinical heterogeneity is expected and it is questionable where these studies should be pooled. 

We agree that our formulation neglects differences in prevalence of within-host resistance emergence depending on bug-drug combinations. We changed our statement in L 86 to: “Within-patient antibiotic resistance development, even if rare, may contribute to the emergence and spread of resistance.”

Line 114: "The overall pooled OR for acquisition of resistance comparing a lower number of antibiotics versus a higher one was 1.23 (95% CI 0.68 - 2.25), with substantial heterogeneity between studies (I2=77.4%)" 

What consequential measures did the authors take after determining this high heterogeneity? Did they explore the source of this large heterogeneity? Considering this large heterogeneity, do the authors consider it appropriate to pool these studies?

Thank you for highlighting this lack of clarity. As mentioned above, we now highlight that we performed several subgroup analyses to investigate heterogeneity. (L 116ff): “The overall pooled estimates are based on studies that focus on various clinical conditions/pathogens and compare different antibiotics treatments. To explore the impact of these and other potential sources of heterogeneity on the resistance estimates we performed various subgroup analyses and meta-regression.” Nevertheless, these analyses faced limitations due to the scarcity of evidence and often still showed a high amount of heterogeneity. Given the lack of appropriate evidence, it is hard to identify the source of heterogeneity. The decision to pool all studies was pre-specified in our previously published study protocol (PROSPERO CRD42020187257) and was motivated by the question whether there is a general effect of combination therapy on resistance development or identify factors that explain potential differences of the effect of combination therapy across bug-drug combinations. Therefore, we think that the presentation of the overall pooled estimate is appropriate, as it was predefined, and potential heterogeneity is furthermore explored in the subgroup analyses. 

Reviewer #1 (Recommendations For The Authors): 

I want to congratulate the investigators for the rigorous approach followed and the - in my opinion - correct interpretation of the data and analysis. The disappointing outcome is independent of the quality of the approach used. Yet, the consequences of that outcome are rather limited, and will not be surprising for - at least - some in the field of antibiotic resistance. 

Thank you for your positive and differentiated feedback.

Reviewer #2 (Recommendations For The Authors): 

Line 93: "The screening of the citations of the 41 studies identified one additional eligible study, for a total of 42 studies". 

Why was this study missed in the search strategy? 

What is the definition of "quasi-RCTs"? Why were these included in the analysis? 

Thank you for pointing out this lack of clarity. The additional study, which was found through screening the references of included studies, was not identified with our search strategy as neither the abstract nor database specific identifiers provided any indications that resistance was measured in this study. We added an explanation in the supplementary materials L 792 ff. and refer to this explanation in the main manuscript (L 95). 

Quasi-randomized trials are trials that use allocation methods, which are not considered truly random. We added this specification in L 95. It now reads: “….two quasi-RCTs, where the allocation method used is not truly random” and in L 252 ff: “Studies were classified as quasi-RCTs if the allocation of participants to study arms was not truly random.” For instance, the study Macnab et al. (1994) assigned patients alternately to the treatment arms. Quasi-randomized controlled trials can lead to biases and especially old studies are more likely to have used quasi-random allocation methods. This can also be seen in our study, where the two quasi-randomized controlled trials were published in 1994 and 1997. The bias is considered in the risk of bias assessment and in our conducted sensitivity analysis regarding the impact of risk of bias on our estimates (supplementary information sections 3.0 and 4.2). Furthermore, one of the two previous conducted meta-analyses comparing beta-lactam monotherapy to beta-lactam and aminoglycoside, which assessed resistance development also included quasi-randomized controlled trials Paul et al 2014. Overall, while designing the study, we decided to include quasi-randomized controlled trials to increase statistical power as we expected that limited statistical power might be a concern and decided to assess potential biases in the risk of bias assessment.  

Line 100: "Consequently, most studies did not have the statistical power to detect a large effect on within-patient resistance development (figure 2 B, SI p 14).". 

Small studies actually have more power to detect large effects while smaller power to detect small effects. Please rephrase. 

Thank you for pointing out this lack of clarity. We rephrased the sentence in order to emphasize our point that the studies are underpowered even if we assume in our power analysis a large effect on resistance development between treatment arms. In this context “the small” studies include too few patients to detect a large difference in resistance development. As resistance development is a rare event, generally studies have to include a larger number of patients to estimate the effect of intervention. We rephrased the sentence in L 101ff to: “Consequently, most studies did not have the statistical power to detect differences in within-patient resistance development even if we assume that the effect on resistance development is large between treatment arms.”

Line 108: "... and prophylaxis for blood cancer patients with four studies (10%) respectively.". 

I would suggest using the medical term hematological malignancy patients. 

Thank you for the suggestion, we changed it as suggested to hematological malignancy patients, also accordingly in the figures, and table 1.

Line 117: "Since the results for the two resistance outcomes are comparable, our focus in the following is on the acquisition of resistance". 

The first OR is 1.23 and the second is 0.74, why do you consider these outcomes as comparable? 

Thank you for pointing out our unprecise formulation. Due to the lack of power the exact estimates need to be interpreted with care. Here, we wanted to make the point that qualitatively the results of both outcomes do not differ in the sense that our analysis shows no substantial difference between a higher and a lower number of antibiotics. We rephrased the sentence to be more precise (L 123ff): “The results for the two resistance outcomes are qualitatively comparable in the sense that individual estimates may differ, but show similar absence of evidence to support either the benefit, harm or equivalence of treating with a higher number of antibiotics. Therefore, our …”. More detailed discussion about differences in estimates can be found in the SI, when the estimates of emergence of resistance are presented (e.g. SI section 2.1).

Line 123: "Furthermore, a lower number of antibiotics performed better than a higher number if the compared treatment arms had no antibiotics in common (pooled OR 4.73, 95% CI 2.14 - 10.42; I 2 =37%, SI p 7).". 

How do you explain this? What does this mean? 

We now added a more detailed explanation in the supplement (L 376ff.): “The result that if the treatment arms had no antibiotics in common a lower number of antibiotics performed better than a higher number of antibiotics could be due to different potencies of antibiotics or resistance prevalences. Further, there could be a bias to combine less potent antibiotics or antibiotics with higher resistance prevalence to ensure treatment efficacy, which couldlead to higher chances to detect resistances in the treatment arm with higher number of antibiotics, e.g. by selecting pre-existing resistance due to antibiotic treatment (see also section 6.1.9).” We furthermore already specifically mention this point in the main manuscript and refer then to the detailed explanation in the SI (L134 ff, “which could be due to different potencies or resistance prevalences of antibiotics as discussed in SI (SI section 6.1.10)”)

Overall, we want to point out that these results need to be interpreted with caution as overall the statistical power is limited to confidently estimate the difference in effect of a higher and lower number of antibiotics.

Line 125: ". In contrast, when restricting the analysis to studies with at least one common antibiotic in the treatment arms are pooled there was little evidence of a difference (pooled OR 0.55, 95% CI 0.28 - 1.07". 

The difference was not statistically significant but there does seem to be an indication of a difference, please rephrase. 

We rephrased the sentence to (L135 ff.): “In contrast, when restricting the analysis to studies with at least one common antibiotic in the treatment arms we found no evidence of a difference, only a weak indication that a higher number of antibiotics performs better (pooled OR 0.55, 95% CI 0.28 – 1.07; I2 \=74%, figure 3B).” 

Line 190: "Similarly, today, relevant cohort studies could be analysed collaboratively using various modern statistical methods to address confounding by indication and other biases (66, 67)". 

However, residual confounding by indication is likely. Please also mention the disadvantages of observational studies compared to RCTs. 

We now highlight that causal inference with observational data comes with its own challenges and stress that randomized controlled trials are still considered the gold standard. L 204ff now reads: “However, even with appropriate causal inference methods, residual confounding cannot be excluded when using observational data (67). Therefore, will remain the gold standard to estimate causal relationships.”

Line 230: "Gram-negative bacteria have an outer membrane, which is absent in grampositive bacteria for instance, therefore intrinsic resistance against antibiotics can be observed in gram-negative bacteria (11)". 

Intrinsic resistance is not unique for Gram-negative bacteria but also exists for Grampositive bacteria. 

We agree with the reviewer that intrinsic resistance is not unique to gram-negative bacteria and refined our writing. We additionally added that differences between gram-negative and gram-positive bacteria are not only to be expected due to differing intrinsic resistances but also due to potential differences in the mechanistic interactions of antibiotics, i.e., synergy or antagonism. The paragraph reads now (SI L289): “The gram status of a bacterium may potentially determine how effective an antibiotic, or an antibiotic combination is. Differences between gram-negative and gram-positive bacteria such as distinct bacterial surface organisation can lead to specific intrinsic resistances of gram-negative and grampositive bacteria against antibiotics (55). These structural differences can lead to varying effects of antibiotic combinations between gram-negative and gram-positive bacteria (56).”

Author response:

The following is the authors’ response to the original reviews.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Line 127. Provide a few more words describing the voltage protocol. To the uninitiated, panels A and B will be difficult to understand. "The large negative step is used to first close all channels, then probe the activation function with a series of depolarizing steps to re-open them and obtain the max conductance from the peak tail current at -36 mV. "

We have revised the text as suggested (revision lines 127 to Line 131): “From a holding potential within the gK,L activation range (here –74 mV), the cell is hyperpolarized to –124 mV, negative to EK and the activation range, producing a large inward current through open gK,L channels that rapidly decays as the channels deactivate. We use the large transient inward current as a hallmark of gK,L. The hyperpolarization closes all channels, and then the activation function is probed with a series of depolarizing steps, obtaining the max conductance from the peak tail current at –44 mV (Fig. 1A).”

Incidentally, why does the peak tail current decay? 

We added this text to the figure legend to explain this: “For steps positive to the midpoint voltage, tail currents are very large. As a result, K+ accumulation in the calyceal cleft reduces driving force on K+, causing currents to decay rapidly, as seen in A (Lim et al., 2011).”

The decay of the peak tail current is a feature of gK,L (large K+ conductance) and the large enclosed synaptic cleft (which concentrates K+ that effluxes from the HC). See Govindaraju et al. (2023) and Lim et al. (2011) for modeling and experiments around this phenomenon.

Line 217-218. For some reason, I stumbled over this wording. Perhaps rearrange as "In type II HCs absence of Kv1.8 significantly increased Rin and tauRC. There was no effect on Vrest because the conductances to which Kv1.8 contributes, gA and gDR activate positive to the resting potential. (so which K conductances establish Vrest???). 

We kept our original wording because we wanted to discuss the baseline (Vrest) before describing responses to current injection.

Vrest is presumably maintained by ATP-dependent Na/K exchangers (ATP1a1), HCN, Kir, and mechanotransduction currents. Repolarization is achieved by delayed rectifier and A-type K+ conductances in type II HCs.

Figure 4, panel C - provides absolute membrane potential for voltage responses. Presumably, these were the most 'ringy' responses. Were they obtained at similar Vm in all cells (i.e., comparisons of Q values in lines 229-230). 

We added the absolute membrane potential scale. Type II HC protocols all started with 0 pA current injection at baseline, so they were at their natural Vrest, which did not differ by genotype or zone. Consistent with Q depending on expression of conductances that activate positive to Vrest, Q did not co-vary with Vrest (Pearson’s correlation coefficient = 0.08, p = 0.47, n= 85).

Lines 254. Staining is non-specific? Rather than non-selective? 

Yes, thanks - Corrected (Line 264).

Figure 6. Do you have a negative control image for Kv1.4 immuno? Is it surprising that this label is all over the cell, but Kv1.8 is restricted to the synaptic pole? 

We don’t have a null-animal control because this immunoreactivity was done in rat. While the cuticular plate staining was most likely nonspecific because we see that with many different antibodies, it’s harder to judge the background staining in the hair cell body layer. After feedback from the reviewers, we decided to pull the KV1.4 immunostaining from the paper because of the lack of null control, high background, and inability to reproduce these results in mouse tissue. In our hands, in mouse tissue, both mouse and rabbit anti-KV1.4 antibodies failed to localize to the hair cell membrane. Further optimization or another method could improve that, but for now the single-cell expression data (McInturff et al., 2018) remain the strongest evidence for KV1.4 expression in murine type II hair cells.

Lines 400-404. Whew, this is pretty cryptic. Expand a bit? 

We simplified this paragraph (revision lines 411-413): “We speculate that gA and gDR(KV1.8) have different subunit composition: gA may include heteromers of KV1.8 with other subunits that confer rapid inactivation, while gDR(KV1.8) may comprise homomeric KV1.8 channels, given that they do not have N-type inactivation .”

Line 428. 'importantly different ion channels'. I think I understand what is meant but perhaps say a bit more. 

Revised (Line 438): “biophysically distinct and functionally different ion channels”.

Random thought. In addition to impacting Rin and TauRC, do you think the more negative Vrest might also provide a selective advantage by increasing the driving force on K entry from endolymph? 

When the calyx is perfectly intact, gK,L is predicted to make Vrest less negative than the values we report in our paper, where we have disturbed the calyx to access the hair cell (–80, Govindaraju et al., 2023, vs. –87 mV, here). By enhancing K+ accumulation in the calyceal cleft, the intact calyx shifts EK—and Vrest—positively (Lim et al., 2011), so the effect on driving force may not be as drastic as what you are thinking.

Reviewer #2 (Recommendations For The Authors):

(1) Introduction: wouldn't the small initial paragraph stating the main conclusion of the study fit better at the end of the background section, instead of at the beginning? 

Thank you for this idea, we have tried that and settled on this direct approach to let people know in advance what the goals of the paper are.

(2) Pg.4: The following sentence is rather confusing "Between P5 and P10, we detected no evidence of a non-gK,L KV1.8-dependent.....". Also, Suppl. Fig 1A seems to show that between P5 and P10 hair cells can display a potassium current having either a hyperpolarised or depolarised Vhalf. Thus, I am not sure I understand the above statement. 

Thank you for pointing out unclear wording. We used the more common “delayed rectifier” term in our revision (Lines 144-147): “Between P5 and P10, some type I HCs have not yet acquired the physiologically defined conductance, gK,L.. N effects of KV1.8 deletion were seen in the delayed rectifier currents of immature type I HCs (Suppl. Fig. 1B), showing that they are not immature forms of the Kv1.8-dependent gK,L channels. ”

(3) For the reduced Cm of hair cells from Kv1.8 knockout mice, could another reason be simply the immature state of the hair cells (i.e. lack of normal growth), rather than less channels in the membrane? 

There were no other signs to suggest immaturity or abnormal growth in KV1.8–/– hair cells or mice. Importantly, type II HCs did not show the same Cm effect.

We further discussed the capacitance effect in lines 160-167: “Cm scales with surface area, but soma sizes were unchanged by deletion of KV1.8 (Suppl. Table 2). Instead, Cm may be higher in KV1.8+/+ cells because of gK,L for two reasons. First, highly expressed trans-membrane proteins (see discussion of gK,L channel density in Chen and Eatock, 2000) can affect membrane thickness (Mitra et al., 2004), which is inversely proportional to specific Cm. Second, gK,L could contaminate estimations of capacitive current, which is calculated from the decay time constant of transient current evoked by small voltage steps outside the operating range of any ion channels. gK,L has such a negative operating range that, even for Vm negative to –90 mV, some gK,L channels are voltage-sensitive and could add to capacitive current.”

(4) Methods: The electrophysiological part states that "For most recordings, we used .....". However, it is not clear what has been used for the other recordings.

Thanks for catching this error, a holdover from an earlier ms. version.  We have deleted “For most recordings” (revision line 466).

Also, please provide the sign for the calculated 4 mV liquid junction potential. 

Done (revision line 476).

Reviewer #3 (Recommendations For The Authors): 

(1) Some of the data in panels in Fig. 1 are hard to match up. The voltage protocols shown in A and B show steps from hyperpolarized values to -71mV (A) and -32 mV (B). However, the value from A doesn't seem to correspond with the activation curve in C.

Thank you for catching this.  We accidentally showed the control I-X curve from a different cell than that in A. We now show the G-V relation for the cell in A.

Also the Vhalf in D for -/- animals is ~-38 mV, which is similar to the most positive step shown in the protocol.

The most positive step in Figure 1B is actually –25 mV. The uneven tick labels might have been confusing, so we re-labeled them to be more conventional.

Were type I cells stepped to more positive potentials to test for the presence of voltage-activated currents at greater depolarizations? This is needed to support the statement on lines 147-148. 

We added “no additional K+ conductance activated up to +40 mV” (revision line 149-150).  Our standard voltage-clamp protocol iterates up to ~+40 mV in KV1.8–/– hair cells, but in Figure 1 we only showed steps up to –25 mV because K+ accumulation in the synaptic cleft with the calyx distorts the current waveform even for the small residual conductances of the knockouts. KV1.8–/– hair cells have a main KV conductance with a Vhalf of ~–38 mV, as shown in Figure 1, and we did not see an additional KV conductance that activated with a more positive Vhalf up to +40 mV.

(2) Line 151 states "While the cells of Kv1.8-/- appeared healthy..." how were epithelia assessed for health? Hair cells arise from support cells and it would be interesting to know if Kv1.8 absence influences supporting cells or neurons. 

We added our criteria for cell health to lines 477-479: “KV1.8–/– hair cells appeared healthy in that cells had resting potentials negative to –50 mV, cells lasted a long time (20-30 minutes) in ruptured patch recordings, membranes were not fragile, and extensive blebbing was not seen.”

Supporting cells were not routinely investigated. We characterized calyx electrical activity (passive membrane properties, voltage-gated currents, firing pattern) and didn’t detect differences between +/+, +/–, and –/– recordings (data not shown). KV1.8 was not detected in neural tissue (Lee et al., 2013). 

(3) Several different K+ channel subtypes were found to contribute to inner hair cell K+ conductances (Dierich et al. 2020) but few additional K+ channel subtypes are considered here in vestibular hair cells. Further comments on calcium-activated conductances (lines 310-317) would be helpful since apamin-sensitive SK conductances are reported in type II hair cells (Poppi et al. 2018) and large iberiotoxin-sensitive BK conductances in type I hair cells (Contini et al. 2020). Were iberiotoxin effects studied at a range of voltages and might calcium-dependent conductances contribute to the enhanced resonance responses shown in Fig. 4? 

We refer you to lines 310-317 in the original ms (lines 322-329 in the revised ms), where we explain possible reasons for not observing IK(Ca) in this study.

(4) Similar to GK,L erg (Kv11) channels show significant Cs+-permeability. Were experiments using Cs+ and/or Kv11 antagonists performed to test for Kv11? 

No. Hurley et al. (2006) used Kv11 antagonists to reveal Kv11 currents in rat utricular type I hair cells with perforated patch, which were also detected in rats with single-cell RT-PCR (Hurley et al. 2006) and in mice with single-cell RNAseq (McInturff et al., 2018).  They likely contribute to hair cell currents, alongside Kv7, Kv1.8, HCN1, and Kir. 

(5) Mechanosensitive ("MET") channels in hair cells are mentioned on lines 234 and 472 (towards the end of the Discussion), but a sentence or two describing the sensory function of hair cells in terms of MET channels and K+ fluxes would help in the Introduction too. 

Following this suggestion we have expanded the introduction with the following lines  (78-87): “Hair cells are known for their large outwardly rectifying K+ conductances, which repolarize membrane voltage following a mechanically evoked perturbation and in some cases contribute to sharp electrical tuning of the hair cell membrane.  Because gK,L is unusually large and unusually negatively activated, it strongly attenuates and speeds up the receptor potentials of type I HCs (Correia et al., 1996; Rüsch and Eatock, 1996b). In addition, gK,L augments a novel non-quantal transmission from type I hair cell to afferent calyx by providing open channels for K+ flow into the synaptic cleft (Contini et al., 2012, 2017, 2020; Govindaraju et al., 2023), increasing the speed and linearity of the transmitted signal (Songer and Eatock, 2013).”

(6) Lines 258-260 state that GKL does not inactivate, but previous literature has documented a slow type of inactivation in mouse crista and utricle type I hair cells (Lim et al. 2011, Rusch and Eatock 1996) which should be considered. 

Lim et al. (2011) concluded that K+ accumulation in the synaptic cleft can explain much of the apparent inactivation of gK,L. In our paper, we were referring to fast, N-type inactivation. We changed that line to be more specific; new revision lines 269-271: “KV1.8, like most KV1 subunits, does not show fast inactivation as a heterologously expressed homomer (Lang et al., 2000; Ranjan et al., 2019; Dierich et al., 2020), nor do the KV1.8-dependent channels in type I HCs, as we show, and in cochlear inner hair cells (Dierich et al., 2020).”

(7) Lines 320-321 Zonal differences in inward rectifier conductances were reported previously in bird hair cells (Masetto and Correia 1997) and should be referenced here.

Zonal differences were reported by Masetto and Correia for type II but not type I avian hair cells, which is why we emphasize that we found a zonal difference in I-H in type I hair cells. We added two citations to direct readers to type II hair cell results (lines 333-334): “The gK,L knockout allowed identification of zonal differences in IH and IKir in type I HCs, previously examined in type II HCs (Masetto and Correia, 1997; Levin and Holt, 2012).”

Also, Horwitz et al. (2011) showed HCN channels in utricles are needed for normal balance function, so please include this reference (see line 171). 

Done (line 184).

(8) Fig 6A. Shows Kv1.4 staining in rat utricle but procedures for rat experiments are not described. These should be added. Also, indicate striola or extrastriola regions (if known). 

We removed KV1.4 immunostaining from the paper, see above.

(9) Table 6, ZD7288 is listed -was this reagent used in experiments to block Gh? If not please omit. 

ZD7288 was used to block gH to produce a clean h-infinity curve in Figure 6, which is described in the legend.

(10) In supplementary Fig. 5A make clear if the currents are from XE991 subtraction. Also, is the G-V data for single cell or multiple cells in B? It appears to be from 1 cell but ages P11-505 are given in legend. 

The G-V curve in B is from XE991 subtraction, and average parameters in the figure caption are for all the KV1.8–/–  striolar type I hair cells where we observed this double Boltzmann tail G-V curve. I added detail to the figure caption to explain this better.

(11) Supplementary Fig. 6A claims a fast activation of inward rectifier K+ channels in type II but not type I cells-not clear what exactly is measured here.

We use “fast inward rectifier” to indicate the inward current that increases within the first 20 ms after hyperpolarization from rest (IKir, characterized in Levin & Holt, 2012) in contrast to HCN channels, which open over ~100 ms. We added panel C to show that the activation of IKir is visible in type II hair cells but not in the knockout type I hair cells that lack gK,L. IKir was a reliable cue to distinguish type I and type II hair cells in the knockout.

For our actual measurements in Fig 6B, we quantified the current flowing after 250 ms at –124 mV because we did not pharmacologically separate IKir and IH.

Could the XE991-sensitive current be activated and contributing?

The XE991-sensitive current could decay (rapidly) at the onset of the hyperpolarizing step, but was not contributing to our measurement of IKir­ and IH, made after 250 ms at –124 mV, at which point any low-voltage-activated (LVA) outward rectifiers have deactivated. Additionally, the LVA XE991-sensitive currents were rare (only detected in some striolar type I hair cells) and when present did not compete with fast IKir, which is only found in type II hair cells.

Also, did the inward rectifier conductances sustain any outward conductance at more depolarized voltage steps? 

For the KV1.8-null mice specifically, we cannot answer the question because we did not use specific blocking agents for inward rectifiers.  However, we expect that there would only be sustained outward IR currents at voltages between EK and ~-60 mV: the foot of IKir’s I-V relation according to published data from mouse utricular hair cells – e.g., Holt and Eatock 1995, Rusch and Eatock 1996, Rusch et al. 1998, Horwitz et al., 2011, etc.  Thus, any such current would be unlikely to contaminate the residual outward rectifiers in Kv1.8-null animals, which activate positive to ~-60 mV. 

(I-HCN is also not a problem, because it could only be outward positive to its reversal potential at ~-40 mV, which is significantly positive to its voltage activation range.)

Author response:

Reviewer #1 (Public Review):

Greter et al. provide an interesting and creative use of lactulose as a "microbial metabolism" inducer, combined with tracking of H2 and other fermentation end products. The topic is timely and will likely be of broad interest to researchers studying nutrition, circadian rhythm, and gut microbiota. However, a couple of moderate to major concerns were noted that may impact the interpretation of the current data:

(1)  Much of the data relies on housing gnotobiotic mice in metabolic cages, but I couldn't find any details of methods to assess contamination during multiple days of housing outside of gnotobiotic isolators/cages. Given the complexity of the metabolic cage system used, sterility would likely be incredibly challenging to achieve. More details needed to be included about how potential contamination of the mice was assessed, ideally with 16S rRNA gene sequencing data of the endpoint samples and/or qPCR for total colonization levels relative to the more targeted data shown.

We thank the reviewer for pointing out that we have not made the experimental setup clear in the text. One of the unique features of our metabolic cage setup is that the mice do not need to be housed outside gnotobiotic isolators, but that the whole system is placed inside an isolator. We have developed and published this system recently (Hoces et al, PLOS Biol 2022), including extensive testing for sterility/gnotobiosis. We will improve clarity in a revised version.

Given that 16S sequencing of germ-free mice will typically produce false positive reads, we used Blautia pseudococcoides as an indicator strain for contaminations. This strain is present in our SPF mouse colony, forms spores that are highly resilient to decontamination measures, and has been the most likely contaminant in our gnotobiotic system. We have checked for presence of this strain in the cecum content of all our animals at the end of each experiment, and only included experiments which had a B. pseudococcoides signal below threshold level.

(2)  The language could be softened to provide a more nuanced discussion of the results. While lactulose does seem to induce microbial metabolism it also could have direct effects on the host due to its osmotic activity or other off-target effects. Thus, it seems more precise to just refer to lactulose specifically in the figure titles and relevant text. Additionally, the degree to which lactulose "disrupts the diurnal rhythm" isn't clear from the data shown, especially given that the markers of circadian rhythm rapidly recover from the perturbation. It is probably more precise to instead state that lactulose transiently induces fermentation during the light phase or something to that effect. The discussion could also be expanded to address what methods are available or could be developed to build upon the concepts here; for example, the use of genetic inducers of metabolism which may avoid the more complex responses to lactulose.

The point about language is well taken. We tried to make the argument that what we call disruption of the diurnal rhythm is acute, meaning that it is not disrupting the rhythm "chronically" (i.e., for longer), but that it recovers rapidly from this transient disruption. Given the confusion this wording is causing we are rephrasing this in a new version of the manuscript.

We also appreciate the mention of concepts from our study that can be built on in future studies, and we will add a paragraph on potential further research.

Despite these concerns, this was still an intriguing and valuable addition to the growing literature on the interface of the microbiome and circadian fields.

We thank the reviewer for all their encouraging and constructive remarks!

Reviewer #2 (Public Review):

Summary:

The authors aimed to investigate how microbial metabolites, such as hydrogen and short-chain fatty acids (SCFAs), influence feeding behavior and circadian gene expression in mice.

Specifically, they sought to understand these effects in different microbial environments, including a reduced community model (EAM), germ-free mice, and SPF mice. The study was designed to explore the broader relationship between the gut microbiome and host circadian rhythms, an area that is not well understood. Through their experiments, the authors hoped to elucidate how microbial metabolism could impact circadian clock genes and feeding patterns, potentially revealing new mechanisms of gut microbiome-host interactions.

Strengths:

The manuscript presents a well-executed investigation into the complex relationship between microbial metabolites and circadian rhythms, with a particular focus on feeding behavior and gene expression in different mouse models. One of the major strengths of the work lies in its innovative use of a reduced community model (EAM) to isolate and examine the effects of specific microbial metabolites, which provides valuable insights into how these metabolites might influence host behavior and circadian regulation. The study also contributes to the broader understanding of the gut microbiome's role in circadian biology, an area that remains poorly understood. The experiments are thoughtfully designed, with a clear rationale that ties together the gut microbiome, metabolic products, and host physiological responses. The authors successfully highlight an intriguing paradox: the significant influence of microbial metabolites in the EAM model versus the lack of effect in germ-free and SPF mice, which adds depth to the ongoing exploration of microbial-host interactions. Despite some methodological concerns, the manuscript offers compelling data and opens up new avenues for research in the field of microbiome and circadian biology.

We thank the reviewer for their encouraging remarks, specifically on the surprising findings that microbial metabolism seems to affect circadian clock gene expression and behavior differently in EAM and SPF mice.

Weaknesses:

The manuscript, while providing valuable insights, has several methodological weaknesses that impact the overall strength of the findings. First, the process for stool collection lacks clarity, raising concerns about potential biases, such as the risk of coprophagia, which could affect the dry-to-wet weight ratio analysis and compromise the validity of these measurements.

We thank the reviewer for pointing out that our description of the specific methods used for collecting feces were presented in a somewhat confusing manner. In short, dry and wet fecal weights were determined based on fecal pellets that were freshly produced and directly collected from restrained mice. To determine total fecal output over time, we collected all fecal pellets produced in a 5 hour window in a cage, determined their dry weight, and then used the water content determined for fresh feces to calculate wet weight. Using this method, we cannot account for potential differences in coprophagia between the groups. However, this is not likely to affect the dry-to-wet ratio of fecal output in our results.

Additionally, the use of the term "circadian" in some contexts appears inaccurate, as "diurnal" might be more appropriate, especially given the uncertainty regarding whether the observed microbiome fluctuations are truly circadian.

Similarly to our answer to reviewer 1 above, we appreciate this remark about imprecise language and have addressed this issue in the text. Indeed, we do not think the microbiota fluctuations are truly circadian, but likely a result of the entrainment through the host's food intake.

Another significant issue is the unexpected absence of an osmotic effect of lactulose in EAM mice, which contradicts the known properties of lactulose as an osmotic laxative. This finding requires further verification, including the use of a positive control, to ensure it is not artifactual.

This is a good point. We have used this lactulose dosage specifically to induce microbial metabolism without causing osmotic diarrhea, and went to some lengths do demonstrate this. In response to this comment (and one by reviewer 3 below about transit time), we are planning an experiment that will use a higher lactulose dose as a positive control.

The presentation of qRT-PCR data as log2-fold changes, with a mean denominator, could introduce bias by artificially reducing variability, potentially leading to spurious findings or increased risk of Type I error. This approach may explain the unexpected activation of both the positive and negative limbs of the circadian clock.

While we agree that our description of the qpcr method used for measuring circadian clock gene expression was lacking detail, we do not see how log2-fold changes (as opposed to, e.g., fold change) would lead to an increased risk of Type 1 error. We did not use a mean denominator for analyzing the data but used the house-keeping data for the same sample as denominator for the respective circadian clock genes. This will be described more clearly in a revised methods section.

Moreover, the lack of detailed information on the primers and housekeeping genes used in the experiments is concerning, particularly given the importance of using non-circadian housekeeping genes for accurate normalization.

We apologize for this omission, it seems like the resource table got lost in the submission, leading to missing information. It will be included in the revised manuscript.

The methods for measuring metabolic hormones, such as GLP-1 and GIP, are also not adequately described. If DPP-IV/protease inhibitor tubes were not used, the data could be unreliable due to the rapid degradation of these hormones by circulating proteases.

We thank the reviewer for spotting this mistake. We will add details of how GLP-1 and GIP were measured to the methods section. While we did not use DPP-IV/protease inhibitor tubes, we added the inhibitors to the syringes when sampling blood, leading to the same effect.

Finally, the manuscript does not address the collection of hormone levels during both fasting and fed phases, a critical aspect for interpreting the metabolic impact of microbial metabolites.

We agree that it will be interesting to measure hormone levels also in the fed phase, and we will include this data in a revised version of the manuscript. Even with that data, a more thorough examination of hormone levels over the diurnal cycle, as suggested by reviewer 3, might be relevant for a full-scale follow-up. Given our data, we of course cannot exclude that there may be time-point-specific differences and therefore have softened the language around this conclusion to state that hormone levels are not acutely changed after a lactulose intervention “at the time-points examined”.

These methodological concerns collectively weaken the robustness of the study's results and warrant careful reconsideration and clarification by the authors.

Because of these weaknesses, the authors have partially achieved their aims by providing novel insights into the relationship between microbial metabolites and host circadian rhythms. The data do suggest that microbial metabolites can significantly influence feeding behavior and circadian gene expression in specific contexts. However, the unexpected absence of an osmotic effect of lactulose, the potential biases introduced by the log2-fold change normalization in qRT- PCR data, and the lack of clarity in critical methodological details weaken the overall conclusions. While the study provides valuable contributions to understanding the gut microbiome's role in circadian biology, the methodological weaknesses prevent a full endorsement of the authors' conclusions. Addressing these issues would be necessary to strengthen the support for their findings and fully achieve the study's aims.

We thank the reviewer again for their careful and critical reading of our work, and for their constructive input. We hope that many of the concerns will be addressed by providing more methodological detail and additional experimental data in the revised version of our manuscript.

Despite the methodological concerns raised, this work has the potential to make a significant impact on the field of circadian biology and microbiome research. The study's exploration of the interaction between microbial metabolites and host circadian rhythms in different microbial environments opens new avenues for understanding the complex interplay between the gut microbiome and host physiology. This research contributes to the growing body of evidence that microbial metabolites play a crucial role in regulating host behaviors and physiological processes, including feeding and circadian gene expression.

We thank the reviewer for their encouraging remarks!

Reviewer #3 (Public Review):

Summary:

In the manuscript by Greter, et al., entitled "Acute targeted induction of gut-microbial metabolism affects host clock genes and nocturnal feeding" the authors are attempting to demonstrate that an acute exposure to a non-nutritive disaccharide (lactulose) promotes microbial metabolism that feeds back onto the host to impact circadian networks. The premise of the study is interesting and the authors have performed several thoughtful experiments to dissect these relationships, providing valuable insights for the field. However, the work presented does not necessarily support some of the conclusions that are drawn. For instance, lactulose is administered during the fasting period to mimic the impact of a feeding bout on the gut microbiota, but it would be important to perform this treatment during the fed state as well to show that the effects on food intake, etc. do not occur.

This is a good point, and we will include an experiment addressing this in a revised version of the manuscript.

To truly draw the conclusion that the current outcomes are directly connected to and mediated via an impact on the host circadian clock, it would be ideal to perform these studies in a circadian gene knock-out animal (i.e., Cry1 or Cry2 KO mice, or perhaps Bmal-VilCre tissue- specific KO mice). If the effects are lost in these animals, this would more concretely connect the current findings to the circadian clock gene network.

We agree that these would be interesting experiments to follow up on the question how the observed effects are actuated by host functions. However, they would require a large amount of preparatory work (including rederiving the KO mice to get them germ-free in our gnotobiotic facility), we argue that they are beyond the scope of this study.

Despite these reservations, the work is promising.

We thank the reviewer for their encouraging assessment.

Strengths:

Attempting to disentangle nutrient acquisition from microbial fermentation and its impact on diurnal dynamics of gut microbes on host circadian rhythms is an important step for providing insights into these host-microbe interactions.

The authors utilize a novel approach in leveraging lactulose coupled with germ-free animals and metabolic cages fitted with detectors that can measure microbial byproducts of fermentation, particularly hydrogen, in real-time.

The authors consider several interesting aspects of lactulose delivery, including how it shifts osmotic balance as well as provides calculations that attempt to explain the caloric contribution of fermentation to the animal in the context of reduced food intake. This provides interesting fundamental insights into the role of microbial outputs on host metabolism.

Thank you!

Weaknesses:

While the authors have done a large amount of work to examine the osmotic vs. metabolic influence of lactulose delivery, the authors have not accounted for the enlarged cecum and increased cecal surface area in germ-free mice. The authors could consider an additional control of cecectomy in germ-free mice.

We thank the reviewer for pointing out the potential effect of the anatomical differences of germ- free and conventionally colonized mice. We agree that when comparing germ-free mice to SPF mice, the enlarged cecum area in germ-free animals could lead to differences in water release or uptake. However, this is not the case in the gnotobiotic mice colonized with our minimal microbiota, which have comparable cecum sizes to germ-free mice, and thus comparing water transport over the cecum wall between those groups can be done without correcting for cecal surface areas. We will add information on cecum sizes in the different experimental groups to a revised version of the manuscript.

The authors have examined GI hormones as one possible mechanism for how food intake is altered by microbial fermentation of lactulose. However, the authors measure PYY and GLP-1 only at a single time point, stating that there are no differences between groups. Given the goal of the studies is to tie these findings back into circadian rhythms, it would be important to show if the diurnal patterns of these GI hormones are altered.

We fully agree that a deeper investigation of the diurnal fluctuations of hormone levels would be an interesting next step in studying whether perturbations in food intake can disturb these rhythms. Doing this for the whole rhythm would really require a full second study. For a revised version of this manuscript, we will add a second time-point of hormone measurements (during the fed phase) to this study. In addition, we will soften the statements made around these data to point out just that hormone level fluctuations could not be detected during specific time points after lactulose treatment, and therefore do not seem to explain the imminent behavioral changes.

Considerations of other factors, such as conjugated vs. deconjugated bile acids, microbial bile salt hydrolase activity, and bile acid resorption, might be an important consideration for how lactulose elicits more influence on ileal circadian clock genes relative to cecum and colon.

We absolutely agree that investigation of microbial bile acid modification and their metabolism by the host would be an interesting topic for a follow-up study.

Measurements of GI transit time (both whole gut and regional) would be an important for consideration for how lactulose might be impacting the ileum vs. cecum vs. colon.

This is also an interesting point, and we will add an assessment of transit time to a revised version of the manuscript.

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

Learn more at Review Commons

Reply to the reviewers

We thank all reviewers for their constructive criticism and suggestions. We have addressed all the points as detailed below. We also added an experiment that strengthens the connection between replication stress and GSF2 and suggests a role of GSF2 in recovery from the DNA replication checkpoint arrest (Fig. 4g).

Reviewer #1 (Evidence, reproducibility and clarity)

Summary

The manuscript by the Khmelinskii group reports that they have successfully constructed two conditional degron libraries of budding yeast for almost all proteins. For this purpose, the authors employed an improved auxin-inducible degron (AID2). Initially, they constructed yeast libraries by fusing HaloTag to the N- or C-terminus of proteins and found that C-terminal tagging is less likely to affect the location and function of proteins (Fig. 1). Based on this finding, the authors fused mNG-AID*-3Myc or AID*-3Myc (AID-v1 or AID-v2 library, respectively) to more than 5600 proteins and found that 4079 proteins were significantly depleted when cells were treated with 5-Ph-IAA (Fig. 2). A fitness defect was observed for over 60% of essential proteins, indicating the target depletion showed the expected phenotype in many cases (Fig. 3). Finally, the authors screened proteins required for maintaining viability in the presence of MMS, CPT and HU, and identified common proteins involved in DNA repair (such as RAD52 epistasis proteins) and other proteins specific for MMS, CPT or HU resistance (Fig. 4). Furthermore, the authors revealed that an ER membrane protein, Gsf2, is required for HU resistance, which was not found in previous studies with the YKO library because gsf2∆ cells in the YKP library had acquired a suppressor mutation (Fig. 4e).

Major comments

1 - In Figure S2a, the authors initially checked the growth of yeast cells expressing OsTIR1(F74G) under the GAL1 promoter, saying that "expression of OsTir1(F74G) from the strong galactose-inducible GAL1 promoter had a negligible impact on yeast fitness (page 3)". To me, the OsTIR1(G74G) expressing cells showed slightly slower growth compared to the control cells. Moreover, the cells expressing it under the very strong GPD promoter showed apparent slow growth, suggesting that OsTIR1(F74G) overexpression caused a side effect. The authors should carefully evaluate the cells with GAL1-OsTIR1(F74G).

Indeed high levels of OsTir1(F74G) impaired growth, at least in the strain background used in our experiments. Expression from the strongest promoter we tested (GPD) resulted in an obvious fitness defect, whereas conditional expression from the strong GAL1 promoter had a small impact on fitness and expression from the weaker CYC1 and ADH1 promoters did not affect fitness (Fig. S2a). Despite the small fitness impairment, we decided to use the GAL1-OsTIR1(F74G) construct for the AID libraries for two reasons: the conditional nature of this promoter is likely to limit adaptation to expression of OsTir1(F74G) and the high expression levels of OsTir1(F74G) are less likely to limit degradation of AID-tagged proteins. We added this explanation to the Results section.

As suggested by the reviewer, we quantitatively evaluated the fitness impact of the GAL1-OsTIR1(F74G) construct. Using the colony size data of the AID-v1 library (grown on galactose medium with 1 µM 5-Ph-IAA, Fig. 2c), we compared colony sizes of OsTIR1– and OsTIR1+ strains for non-essential ORFs. As degradation of non-essential proteins is not expected to affect fitness, the difference in colony size between OsTIR1– and OsTIR1+ strains can be attributed to OsTir1 expression. On average, the presence of the OsTIR1 construct reduced colony size by 7% (median fitness of OsTIR1+ strains relative to OsTIR1– strains of 0.93 ± 0.06, n = 4698 non-essential ORFs). We performed the same comparison for strains that did not exhibited OsTIR1-dependent protein degradation. In this set of strains, the presence of the OsTIR1 construct also reduced colony size by 7% (median fitness of OsTIR1+ strains relative to OsTIR1– strains of 0.93 ± 0.05, n = 624 ORFs in the “not affected” group in Fig. 2d). We added this information to Fig. S3a.

2 - Given the possibility that OsTIR1(F74G) overexpression might cause a growth problem, it is not appropriate to compare OsTIR1+ and OsTIR1- conditions for evaluating growth fitness (Fig. 2). As shown in Fig. S4b, it is more appropriate to compare the +/- 5-Ph-IAA conditions. Additionally, the 5-Ph-IAA concentration used in this study was not clearly mentioned in the method section and figure legends.

The two approaches, comparison of OsTIR1– and OsTIR1+ strains grown on galactose with 5-Ph-IAA (as was done for the AID-v1 library) and comparison of galactose ± 5-Ph-IAA conditions (as was done for the AID-v2 library), have advantages and disadvantages but should yield similar results. The technical noise (due to spatial effects on the screen plates) is lower for the comparison of OsTIR1– and OsTIR1+ strains, as the two strains for each ORF can be grown next to each other on the same plate (Fig. 2c). Furthermore, corrections of spatial effects are more precise with this layout as the frequency of fitness defects per plate is lower. On the other hand, comparison of galactose ± 5-Ph-IAA conditions implicitly corrects for the fitness impact of the GAL1-OsTIR1(F74G) construct, as the fitness distribution of each condition is normalized to the median of that condition, but this fitness impact of OsTir1 cannot be determine from the screen results.

We now explicitly corrected the colony size data of the AID-v1 library for the fitness impact of OsTir1 expression (quantified in the previous point) and updated all the analyses and results shown in Fig. 3, Fig. S3b-e and Fig. S4a. The correction was performed using the multiplicative model, whereby the fitness impacts of OsTir1 expression and degradation of the AID-tagged protein are independent. Overall, our observations and conclusions stand unchanged with the corrected data.

Finally, the 5-Ph-IAA concentration (1 µM) used in all experiments is now indicated in the figure legends and the Methods section.

3 - The authors found that fitness defects were observed for over 60% of essential proteins (Fig. 3). In other words, depletion of the remaining 40% was not enough to induce growth defects. The authors should discuss how the current AID library can be improved to achieve better target depletion. Previous literature reported various possibilities, such as using a tandem degron tag and combining AID with the Tet promoter system (PMID 25181302, 26081484). Although optional, it would be wonderful if the authors would generate an improved library.

Following the reviewer’s suggestion, we added the following statement to the discussion:

“In the future, the libraries could be potentially improved with N-terminal tagging of ORFs that currently exhibit incomplete or no degradation of AID-tagged proteins or using multiple copies of the AID* tag to enhance protein degradation (Kubota et al, 2013; Nishimura & Kanemaki, 2014).”

Minor comments

4 - 5-Ph-IAA is not auxin because it does not induce the auxin responses in plants (PMID 29355850). Therefore, the authors should be careful when they refer to 5-Ph-IAA and should not call it auxin.

We corrected this and now refer to 5-Ph-IAA explicitly throughout the manuscript.

5 - The availability of the HaloTag and AID libraries should be indicated.

We added the following statement to the Methods section: “All strains, plasmids and libraries are available upon request.”

6 - Page 3: "Finally, the extent of AID-dependent degradation varied with protein abundance, in that highly expressed proteins were more likely to be only partially degraded compared to lowly expressed ones (Fig. 2e, Fig. S2e)". Fig. S2e should be Fig. S2d, shouldn't it?

We corrected this mistake.

Reviewer #1 (Significance):

This paper is technically robust and well-conducted. It presents a comprehensive study showcasing the effectiveness of the conditional degron library. The HaloTag libraries will also be useful. The yeast libraries presented in this study will be invaluable for future screenings and studies across all aspects of yeast biology.

Reviewer #2 (Evidence, reproducibility and clarity):

In this study, the authors reported the development and characterization of two AID-tagged strain libraries for the model organism S. cerevisiae. The libraries are based on the latest AID technology, AID2. One library contains a fluorescent protein fused to the AID, whereas the other library does not have the fluorescent protein, thus offering better compatibility with imaging-based screens. The authors show that AID-dependent protein degradation can be achieved for most of the library strains, and a growth phenotype was induced for a high fraction of essential genes. Genetic screens for DNA damage-sensitive mutants showcased the applicability of the libraries.

I only have the following minor comments and suggestions for the authors to consider.

Point 1, Page 3

"Optimized tagging of proteins with these N-terminal localization signals likely also contributes to the lack of correlation between differential fitness defects and occurrence of terminal localization signals (Fig. S1f, Table S2)."

Is this because the genes that cannot tolerate C-terminal addition are already depleted in the C-SWAT library? In the C-SWAT library, a 15-amino-acid linker L3 is added to the C-terminus.

That is certainly a possibility. During construction of SWAT library, tagging with N-SWAT and C-SWAT acceptor modules failed for 251 and 353 ORFs, respectively (Weill et al. 2018, Meurer et al. 2018). However, these ORFs are not enriched in N- or C-terminal localization signals, respectively (4.6% ORFs with C-terminal signals in C-SWAT library vs 3.3% among failed C-SWAT strains; 12.3% ORFs with N-terminal signals in N-SWAT library vs 2.0% among failed N-SWAT strains).

The most significant trend in the data is enrichment of ribosomal subunits in both sets of failed strains: 3.9% and 16.3% of the genes mapped to the GO term “ribosome” in the N-SWAT library and the set of failed N-SWAT strains, respectively; 3.6% and 15.9% of the genes in the C-SWAT library and the set of failed C-SWAT strains, respectively. This is consistent with what was reported by Weill et al. for failed N-SWAT strains.

Point 2, Page 3

"Expression of OsTir1(F74G) from the strong galactose-inducible GAL1 promoter had a negligible impact on yeast fitness (Fig. S2a)."

I wonder why the authors chose to use an inducible promoter to express OsTir1(F74G). In other studies, for example Snyder et al. 2019, OsTir1 has been expressed from a constitutive promoter.

Despite the small fitness impairment, we decided to use the GAL1-OsTIR1(F74G) construct for the AID libraries for two reasons: the conditional nature of this promoter is likely to limit adaptation to expression of OsTir1(F74G) and the high expression levels of OsTir1(F74G) are less likely to limit degradation of AID-tagged proteins. We added this explanation to the Results section.

Please see our response to reviewer 1, points 1 and 2.

Point 3, Page 3

"A similar frequency was previously observed with a set of AID alleles constructed for 758 essential ORFs using the original AID system (Snyder et al, 2019). However, over a third of these alleles exhibited fitness defects even in the absence of auxin, which were further compounded by off-target effects of auxin, highlighting the advantages of the AID2 system."

Snyder et al. 2019 used a TAP-AID-6FLAG tag. The fitness defect in the absence of auxin may not necessarily be due to the AID part of the tag, as TAP tagging is known to compromise the functions of some genes.

We corrected our statement as follows:

“A similar frequency was previously observed with a set of AID alleles constructed for 758 essential ORFs using the original AID system (Snyder et al, 2019). However, over a third of these alleles exhibited fitness defects even in the absence of auxin, which were further compounded by off-target effects of auxin.”

Point 4, Page 3

"Interestingly, complete degradation of 33% of essential proteins did not result in a fitness defect. It is possible that in some cases partial degradation results in low protein levels that are below the detection limit of our assay but are sufficient for viability."

Are these "33% of essential proteins" enriched with genes with low expression levels? I guess genes with low expression levels are more likely to fall below the detection limit even when partially depleted. Are there extreme examples where a highly expressed essential gene does not exhibit a fitness defect when the protein product is no longer detectable?

We performed the analysis suggest by the reviewer, and observed no difference in pre-degradation protein levels between essential & degraded proteins with and without a fitness defect (now shown in Fig. S3b). This also showed that there are indeed several essential proteins with high pre-degradation proteins levels and without a fitness defects upon degradation to below our detection limit: Pgi1, Nhp2, Smt3, Gus1, Dys1, Sis1, Fas2 and Rpo26 (in the abundance bin 4 in Fig. S2f).

In addition, we considered the nature of the essential genes in these two groups. Namely, we compared the frequency of core essential genes, which are always required for viability, and conditional essential genes, which vary in essentiality depending on the genetic background or environment (Bosch-Guiteras & van Leeuwen, 2022). Interestingly, the set of essential and degraded proteins without an accompanying fitness defect was enriched in conditional essential genes defined by two independent measures: essentiality across S. cerevisiae natural isolates (Peter et al, 2018) or with bypass suppression interactions in a laboratory strain (van Leeuwen et al, 2020) (Fig. S3c, odds ratio = 1.6, p-value = 0.04 in a Fisher’s exact test and odds ratio = 1.7, p-value = 0.02, respectively). This suggests that conditional essentiality could explain the observed lack of fitness defects upon degradation of some essential proteins.

We added this analysis to the Results section.

Reviewer #2 (Significance):

This study generated highly valuable resources for functional genomic studies.

Reviewer #3 (Evidence, reproducibility and clarity):

Summary: In this manuscript, the authors construct and analyze a genome-wide collection of AID-tagged S. cerevisiae strains. The manuscript is clearly written and the analysis appears to be thorough. This collection will be quite useful to the yeast community. There are some issues to address, listed below.

1. page 1, abstract - "...with protein abundance and tag accessibility as limiting factors." It's not clear what the authors mean by protein abundance as a limiting factor. Are they referring to the protein level pre-depletion? Please clarify.

That is correct. We clarified this statement as follows:

“Almost 90% of AID-tagged proteins were degraded in the presence of the auxin analog 5-Ph-IAA, with initial protein abundance and tag accessibility as limiting factors.”

1. page 1, second paragraph of the Intro, end of the paragraph - There are publications prior to Van Leeuwen et al. 2016 that describe suppressors lurking in the deletion set. Here are two that should be cited: Hughes et al. 2000, DOI: 10.1038/77116 and Teng et al. 2013, DOI: 10.1016/j.molcel.2013.09.026.

We added the references pointed out by the review.

1. The goal of this work, stated in the first sentence of results, is to construct genome-wide AID libraries. Yet, to test whether N-terminal or C-terminal tagging is better, the authors used a Halo tag. Those results showed that, for the Halo tag, C-terminal tagging was less likely to impair function. Why weren't these tests done with the same AID tag used to build the libraries in the next section? What is the evidence that the results for a Halo tag will be the same as for an AID tag? While hard to find it documented in publications, there is a lot of anecdotal evidence that the type of tag can make a big difference, as well as its location. While this section will be of interest to those using Halo tags, it's not clear how it relates to the rest of the paper, especially given the careful characterization of the AID library in the next section.

We chose the Halo tag due its size (33 kDa), similar to many commonly used fluorescent protein tags and to the mNG-AID*-3myc tag in the AID-v1 library, and lack of evidence for a dominant negative effect on the tagged proteins. This is now stated in the Results section.

We agree that further work is needed to understand how the type of tag, its size and biophysical properties, and the linker between the tag and the protein of interest affect protein localization and function across the proteome. This is now stated in the Results section.

1. Throughout this manuscript, including the Tables, in cases where the protein is no longer detected, please do not describe this as "complete degradation." Instead, please use "not detectable." This is clearly the case for essential proteins that are no longer detected but that still grow, so it is very likely the case for many or all of the others. If the authors have any understanding of the sensitivity of their fluorescence assay, then that would be helpful to know. For example, they could add a control, taking a known amount of a fluorescent protein and analyzing known dilutions to assay the level of detection.

We appreciate the reviewer’s suggestion. We decided against “not detectable” instead of “complete degradation” to avoid confusion with proteins that are not detectable pre-degradation. Nevertheless, we replaced “complete degradation” with “degradation” and added the following explanation to the Results section:

“Out of 5079 proteins detected in OsTIR1– strains, 4455 (~88%) were significantly depleted in OsTIR1+ strains (Fig. 2d, Table S3). 3981 proteins could not be detected specifically in the OsTIR1+ background. Hereafter, we will refer to these proteins as degraded, although it is likely that at least in some cases degradation is not complete but the remainder is below the detection limit of our plate reader assay. Nevertheless, 474 proteins were unequivocally degraded only partially, as they were detectable in the OsTIR1+ background but at reduced levels compared to the OsTIR1– background (Fig. 2d).”

To estimate the detection limit of the colony fluorescence assay, we correlated the background-corrected mNG intensities in OsTIR1– strains with absolute levels (in molecules per cell) of 1167 proteins determined by Lawless et al. (PMID 26750110). Based on a linear fit, the threshold above which proteins are considered “detected” in our analysis, mNG/bkg(OsTIR1–) > 1.2, corresponds to 200 molecules per cell (95% confidence interval 18 to 2187 molecules per cell). We added this information to the Results section and Fig. S2c.

This detection limit is in line with our results, where low abundance proteins such as the centromeric histone Cse4/CENP-A (with two Cse4 molecules per centromere adding to 64 molecules per cell, Aravamudhan et al. PMID: 23623551 and several times that amount elsewhere in the cell, Collins et al. PMID: 15530401) can be detected in the colony assay (Table S3).

1. Fig. S2a and page 3, first full paragraph - The authors wrote that expression of OsTir1(F74G) from the strong GAL1 promoter had a negligible impact on growth. However, the figure shows that there is an obvious effect on growth after 48 hours of incubation, with much smaller colonies. This defect is much less obvious after 72 hours. This difference suggests that the growth effect would have been even more obvious at 24 hours. I think that the text should be modified to indicate this effect.

We now quantified the fitness impact of the GAL1-OsTIR1(F74G) construct and rephrased this part of the manuscript. In addition, we corrected the AID-v1 library screen results for the fitness impact of the GAL1-OsTIR1(F74G) construct and updated all figures and tables. Please see our response to reviewer 1, points 1 and 2.

1. One of the main justifications for the construction of the AID library is to allow assays for essential genes. Yet that was not a feature of the screen for DNA damage response factors. Were any essential genes identified in those screens? It would be of great interest to identify lower levels of 5-Ph-IAA that only mildly affect growth of essential genes and then to repeat the screens.

58 out of the combined 165 potential resistance factors identified in the three screens are essential genes. We added this information to the Results section and essential genes are now indicated in Fig. S5c.

We now show that chemical-genetic interactions for both essential and non-essential genes can be reproduced in spot tests using the MMS screen as an example (Fig. S5d). We also show that additional essential hits can be identified at lower concentrations of 5-Ph-IAA, which allow determining chemical-genetic interactions for strains that otherwise exhibit no growth in 1 μM 5-Ph-IAA (Fig. S5e). As the screens serve as a demonstration of possible uses of the AID libraries, we consider additional exhaustive screening for DNA damage response factors beyond the scope of this manuscript.

1. A big advantage of AID depletion over deletions is the ability to look at strains very shortly after loss of the protein of interest. In many cases in the literature, experiments are done after one or two hours after depletion. Yet in this work, there are no data presented on how effective depletion is in the short term versus after a long period of growth (24 hours). It would be a strong addition to the manuscript to include a time course for at least a subset of the proteins to look at the loss of signal over time, either by fluorescence or by Western blots.

We performed time courses of protein depletion with immunoblotting for 12 strains (4 proteins from the “degraded”, “partially degraded” and “not affected” groups each). The results in Fig. S2e show that “degraded” proteins are depleted to below the detection limit within 60min of 5-Ph-IAA addition, “partially degraded” proteins are depleted less or exhibit a degradation-resistant pool, and the levels of “not affected” proteins remain stable over time, consistent with their classification based on mNG fluorescence in the colony assay. We added this information to the Results section.

Reviewer #3 (Significance):

The library will be of use to the yeast community.

Buildings are more than just structures—they tell stories about the cultures and times they were built in

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

Weakness 1. Enhancing Reproducibility and Robustness: To enhance the reproducibility and robustness of the findings, it would be valuable for the authors to provide specific numbers of animals used in each experiment. Explicitly stating the penetrance of the rod-like neurocranial shape in dact1/2-/- animals would provide a clearer understanding of the consistency of this phenotype. 

In Fig. 3 and Fig. 4 animal numbers were added to the figure and figure legend (line 1111). In Fig. 5 animal numbers were added to the figure. We now state that dact1/2-/- animals exhibit the rod-like neurocranial shape that is completely penetrant (Line 260). 

Weakness 2. Strengthening Single-Cell Data Interpretation: To further validate the single-cell data and strengthen the interpretation of the gene expression patterns, I recommend the following: 

-Provide a more thorough explanation of the rationale for comparing dact1/2 double mutants with gpc4 mutants.

-Employ genotyping techniques after embryo collection to ensure the accuracy of animal selection based on phenotype and address the potential for contamination of wild-type "delayed" animals.

-Supplement the single-cell data with secondary validation using RNA in situ or immunohistochemistry techniques. 

An explanation of our rationale was added to the results section (Lines 391403) and a summary schematic was added to Figure 6 (panel A).

Genotyping of the embryos was not possible but quality control analysis by considering the top 2000 most variable genes across the dataset showed good clustering by genotype, indicating the reproducibility of individuals in each group (See Supplemental Fig. 4).

The gene expression profiles obtained in our single-cell data analysis for gpc4, dact1, and dact2 correlate closely with our in situ hybridization analyses. Further, our data is consistent with published zebrafish single-cell data. We validated our finding of increased capn8 expression in dact1/2 mutants by in situ hybridization. Therefore we are confident in the robustness of our single-cell data.  

Weakness 3. Directly Investigating Non-Cell-Autonomous Effects: To directly assess the proposed non-cell-autonomous role of dact1/2, I suggest conducting transplantation experiments to examine the ability of ectodermal/neural crest cells from dact1/2 double mutants to form wild-type-like neurocranium.  

The reviewer’s suggestion is an excellent experiment and something to consider for future work. Cell transplant experiments between animals of specific genotypes are challenging and require large numbers. It is not possible to determine the genotype of the donor and recipient embryos at the early timepoint of 1,000 cell stage where the transplants would have to be done in the zebrafish. So that each transplant will have to be carried out blind to genotype from a dact1+/-; dact2+/- or dact1-/-; dact2+/- intercross and then both animals have to be genotyped at a subsequent time point, and the phenotype of the transplant recipient be analyzed. While possible, this is a monumental undertaking and beyond the scope of the current study.

Weakness 4. Further Elucidating Calpain 8's Role: To strengthen the evidence supporting the critical role of Calpain 8, I recommend conducting overexpression experiments using a sensitized background to enhance the statistical significance of the findings. 

We thank the reviewer for their suggestion and have now performed capn8 overexpression experiments in embryos generated from dact1/2 double heterozygous breeding. We found a statistically significant effect of capn8 overexpression in the dact1+/-,dact2+/- fish (Lines 462-464 and Fig. 8C,D). 

Minor Comments:  

Comment: Creating the manuscript without numbered pages, lines, or figures makes orientation and referencing harder.  

Revised

Comment: Authors are inconsistent in the use of font and adverbs, which requires extra effort from the reader. ("wntIIf2 vs wnt11f2 vs wnt11f2l"; "dact1/2-/- vs dact1/dact2 -/-"; "whole-mount vs wholemount vs whole mount").  

Revised throughout.

Comment: Multiple sentences in the "Results" belong to the "Materials and Methods" or the "Discussion" section. 

We have worked to ensure that sentences are within the appropriate sections of the manuscript.

Comment: Abstract:

"wnt11f2l" should be "wnt11f2"  

Revised (Line 24).

Comment: Main text:

Page 5 - citation Waxman, Hocking et al. 2004 is used 3x without interruption any other citation. 

Revised (Line 112).

Page 9 - "dsh" mutant is mentioned once in the whole manuscript - is this a mistake?

Revised, Rewritten (Line 196).

Page 10 - Fig 2B does not show ISH.

Revised (Line 229).

Page 11 - "kyn" mutant is mentioned here for the first time but defined on page 15.

Revised (Line 245). Now first described on page 4.

Page 14 - "cranial CNN" should be CNCC.

Revised. (Line 334)

Page 16 - dact1/dact2/gpc4: Fig. 5C is used but it should be Fig 5E.

Revised. (Line 381)

Page 18 - dact1/2-/- or dact1-/-, dact2-/-. 

Revised. (Line 428)

Comment: Methods:

Page 24 - ZIRC () "dot" is missing. ChopChop ")" is missing. "located near the 5' end of the gene" - In the Supplementary Figure 1 looks like in the middle of the gene.

Revised. (Lines 600, 609, 611, respectively).

Page 25 - WISH -not used in the main text.

Revised. (Line 346).

Page 26 - 4% (v/v) formaldehyde; at 4C - 4{degree sign}C; 50% (v/v) ethanol; 3% (w/v) methylcellulose.

Revised. (Lines 659, 660, 662).

Page 27 - 0.1% (w/v) BSA. 

Revised. (Line 668).

Comment: Discussion:

The overall discussion requires more references and additional hypotheses. On page 20, when mentioning 'as single mutants develop normally,' does this refer to the entire animals or solely the craniofacial domain? Are these mutants viable? If they are, it's crucial to discuss this phenomenon in relation to prior morpholino studies and genetic compensation.

Observing how the authors interpret previously documented changes in nodal and shh signaling would be beneficial. While Smad1 is discussed, what about other downstream genes? Is shh signaling altered in the dact1/2 double mutants? 

We have revised the Discussion to include more references (Lines 473, 476, 483, 488, 491, 499, 501, 502, 510, 515, 529, 557, 558) and additional hypotheses (Lines 503-505, 511-519, 522-525). We have added more specific information regarding the single mutants (Lines 270-275, 480-493, Fig. S3). We have added discussion of other downstream genes, including smad1 (Lines 561-572) and shh (Lines 572-580).

Comment: Figures:

Appreciating differences between specimens when eyes were or were not removed is quite hard.

Yes this was an unfortunate oversight, however, the key phenotype is the EP shown in the dissections.

Fig 1. - wntIIf2 vs wnt11f2? C - Thisse 2001 - correct is Thisse et al. 2001.

Revised typo in Fig 1. (And Line 1083).

Fig 1E: These plots are hard to understand without previous and detailed knowledge. Authors should include at least some demarcations for the cephalic mesoderm, neural ectoderm, mesenchyme, and muscle. Missing color code.

We have moved this data to supplementary figure S1 and have added labels of the relevant cell types and have added the color code.

Comment:- Fig 2 - In the legend for C - "wildtype and dact2-/- mutant" and "dact1/2 mutant"; in the picture is dact1-/-, dact2-/-.

Revised (Line 1105).

Fig 2 - B - it is a mistake in 6th condition dact1: 2x +/+, heterozygote (+/-) is missing.

Revised Figure 2B.

Fig 4. - Typo in the legend: dact1/"t"2-/- .

Revised. (Line 1127).

Fig 8C - In my view, when the condition gfp mRNA says "0/197, " none of the animals show this phenotype. I assume the authors wanted to say that all the animals show this phenotype; therefore, "197/197" should be used.

We have removed this data from the figure as there were concerns by the reviewers regarding reproducibility. 

Fig S1 - Missing legend for the 28 + 250, 380 + 387 peaks? RT-qPCR - is not mentioned in the Materials and Methods. In D - ratio of 25% (legend), but 35% (graph).

Revised.(Line 1203, Line 625, Line 1213, respectively).

Fig S2 - The word "identified" - 2x in one sentence. 

Revised. (Line 1230).

Reviewer #2 (Public Review):

Weakness(1) While the qualitative data show altered morphologies in each mutant, quantifications of these phenotypes are lacking in several instances, making it difficult to gauge reproducibility and penetrance, as well as to assess the novel ANC forms described in certain mutants.  

In Fig. 3 and Fig. 4 animal numbers were added to the figure legend. In Fig. 5 animal numbers were added to the figure to demonstrate reproducibility. We now state that dact1/2-/- animals exhibit the rod-like neurocranial shape that is completely penetrant (Line 260). As the altered morphologies that we report are qualitatively significant from wildtype we did not find it necessary to make quantitative measurements. For experiments in which it was necessary to in-cross triple heterozygotes (Fig 3, Fig. 5), we dissected and visually analyzed the ANC of at least 3 compound mutant individuals. At least one individual was dissected for the previously published or described genotypes/phenotypes (i.e. wt, wntllf2-/-, dact1/2-/-, gpc4-/-, wls/-). We realize quantitative measurements may identify subtle differences between genotypes. However, the sheer number of embryos needed to generate these relatively rare combinatorial genotypes and the amount of genotyping required prevented quantitative analyses. 

Weakness 2) Germline mutations limit the authors' ability to study a gene's spatiotemporal functional requirement. They therefore cannot concretely attribute nor separate early-stage phenotypes (during gastrulation) to/from late-stage phenotypes (ANC morphological changes). 

We agree that we cannot concretely attribute nor separate early and latestage phenotypes. Conditional mutants to provide temporal or cell-specific analysis are beyond the scope of this work. Here we speculate based on evidence obtained by comparing and contrasting embryos with grossly similar early phenotypes and divergent late-stage phenotypes. We believe our findings contribute to the existing body of literature on zebrafish mutants with both early convergent extension defects and craniofacial abnormalities.   

Weakness (3) Given that dact1/2 can regulate both canonical and non-canonical wnt signaling, this study did not specifically test which of these pathways is altered in the dact1/2 mutants, and it is currently unclear whether disrupted canonical wnt signaling contributes to the craniofacial phenotypes, even though these phenotypes are typical non-canonical wnt phenotypes. 

Previous literature has attributed canonical wnt, non-canonical wnt, and nonwnt functions to dact, and each of these likely contributes to the dact mutant phenotype (Lines 87-89). We performed cursory analyses of tcf/lef:gfp expression in the dact mutants and did not find evidence to support further analysis of canonical wnt signaling in these fish. Single-cell RNAseq did not identify differential expression of any canonical or non-canonical wnt genes in the dact1/2 mutants.

Further research is needed to parse out the intracellular roles of dact1 and dact2 in response to wnt and tgf-beta signaling. Here we find that dact may also have a role in calcium signaling, and further experiments are needed to elaborate this role.      

Weakness (4) The use of single-cell RNA sequencing unveiled genes and processes that are uniquely altered in the dact1/2 mutants, but not in the gpc4 mutants during gastrulation. However, how these changes lead to the manifested ANC phenotype later during craniofacial development remains unclear. The authors showed that calpain 8 is significantly upregulated in the mutant, but the fact that only 1 out of 142 calpainoverexpressing animals phenocopied dact1/2 mutants indicates the complexity of the system. 

To further test whether capn8 overexpression may contribute to the ANC phenotype we performed overexpression experiments in the resultant embryos of dact1/dact2 double het incross. We found the addition of capn8 caused a small but statistically significant occurrence of the mutant phenotype in dact1/2 double heterozygotes (Fig.8D). We agree with the reviewer that our results indicate a complex system of dysregulation that leads to the mutant phenotype. We hypothesize that a combination of gene dysregulation may be required to recapitulate the mutant ANC phenotype. Further, as capn8 activity is regulated by calcium levels, overexpression of the mRNA alone likely has a small effect on the manifestation of the phenotype. 

Weakness (5) Craniofacial phenotypes observed in this study are attributed to convergent extension defects but convergent extension cell movement itself was not directly examined, leaving open if changes in other cellular processes, such as cell differentiation, proliferation, or oriented division, could cause distinct phenotypes between different mutants. 

Although convergent extension cell movements were not directly examined, our phenotypic analyses of the dact1/2 mutant are consistent with previous literature where axis extension anomalies were attributed to defects in convergent extension (Waxman 2004, Xing 2018, Topczewski 2001). We do not attribute the axis defect to differentiation differences as in situ analyses of established cell type markers show the existence of these cells, only displaced relative to wildtype (Figure 1). We agree that we cannot rule out a role for differences in apoptosis or proliferation however, we did not detect transcriptional differences in dact1/2 mutants that would indicate this in the single-cell RNAseq dataset. Defects in directed division are possible, but alone would not explain that dact1/2 mutant phenotype, particularly the widened dorsal axis (Figure 1).

Major comments:  

Comment (1) The author examined and showed convergent extension phenotype (CE) during body axis elongation in dact1/dact2-/- homozygous mutants. Given that dact2-/- single mutants also displayed shortened axis, the authors should either explain why they didn't analyze CE in dact2-/- (perhaps because that has been looked at in previously published dact2 morphants?) or additionally show whether CE phenotypes are present in dact1 and dact2 single mutants.  

The authors should quantify the CE phenotype in both dact2-/- single mutants and dact1/dact2-/- double mutants, and examine whether the CE phenotypes are exacerbated in the double mutants, which may lend support to the authors' idea that dact1 can contribute to CE. The authors stated in the discussion that they "posit that dact1 expression in the mesoderm is required for dorsal CE during gastrulation through its role in noncanonical Wnt/PCP signaling". However, no evidence was presented in the paper to show that dact1 influences CE during body axis elongation.  

Because any axis shortening in shortening in dact2-/- single mutants was overcome during the course of development and at 5 dpf there was no noticeable phenotype, we did not analyze the single mutants further.  

We have added data to demonstrate the resulting phenotype of each combinatorial genotype to provide a more clear and detailed description of the single and compound mutants (Fig. S3). 

Our hypothesis that dact1 may contribute to convergent extension is based on its apparent ability to compensate (either directly or indirectly) for dact2 loss in the dact2-/- single mutant. 

Comment (2) Except in Fig. 2, I could not find n numbers given in other experiments. It is therefore unclear if these mutant phenotypes were fully or partially penetrant. In general, there is also a lack of quantifications to help support the qualitative results. For example, in Fig. 4, n numbers should be given and cell movements and/or contributions to the ANC should be quantified to statistically demonstrate that the second stream of CNCC failed to contribute to the ANC.  

Similarly, while the fan-shaped and the rod-shaped ANCs are very distinct, the various rod-shaped ANCs need to be quantified (e.g. morphometry or measurements of morphological features) in order for the authors to claim that these are "novel ANC forms", such as in the dact1/2-/-, gpc4/dact1/2-/-, and wls/dact1/2-/- mutants (Fig. 5).  

We have added n numbers for each experiment and stated that the rod-like phenotype of the dact1/2-/- mutant was fully penetrant. 

Regarding CNCC experiments, we repeated the analysis on 3 individual controls and mutants and did not find evidence that CNCC migration was directly affected in the dact1/2 mutant. Rather, differences in ANC development are likely secondary to defects in floor plate and eye field morphometry. Therefore we did not do any further analyses of the CNCCs.

Regarding figure 5, we have added n numbers. We dissected and analyzed a minimum of three triple mutants (dact1/2-/-,gpc4-/- and dact1/2-/-,wls-/-) and numerous dact1/s double mutants and found that the triple mutant ANC phenotype was consistent and recognizably different enough from the dact1/2-/-, or gpc4 or wls single mutant that morphometry measurements were not needed. Further, the triple mutant phenotype (narrow and shortened) appears to be a simple combination of dact1/2 (narrow) and gpc4/wls (shortened) phenotypes. As we did not find evidence of genetic epistasis, we did not analyze the novel ANC forms further.

Comment (3): The authors have attributed the ANC phenotypes in dact1/2-/- to CE defects and altered noncanonical wnt signaling. However, no evidence was presented to support either. The authors can perhaps utilize diI labelling, photoconversionmediated lineage tracing, or live imaging to study cell movement in the ANC and compare that with the cell movement change in the gpc4-/- , and gpc4/dact1/2-/- mutants in order to first establish that dact1/2 affect CE and then examine how dact1/2 mutations can modulate the CE phenotypes in gpc4-/- mutants.  

Concurrently, given that dact1 and dact2 can affect (perhaps differentially) both canonical and non-canonical wnt signaling, the authors are encouraged to also test whether canonical wnt signaling is affected in the ANC or surrounding tissues, or at minimum, discuss the potential role/contribution of canonical wnt signaling in this context.  

Given the substantial body of research on the role of noncanonical wnt signaling and planar cell polarity pathway on convergent extension during axis formation (reviewed by Yang and Mlodzik 2015, Roszko et al., 2009) and the resulting phenotypes of various zebrafish mutants (i.e. Xing 2018, Topczewski 2001), including previous research on dact1 and 2 morphants (Waxman 2004), we did not find it necessary to analyze CE cell movements directly.  

Our finding that CNCC migration was not defective in the dact1/2 mutants and the knowledge that various zebrafish mutants with anterior patterning defects (slb, smo, cyc) have a similar craniofacial abnormality led us to conclude that the rod-like ANC in the dact1/2 mutant was secondary to an early patterning defect (abnormal eye field morphology). Therefore, testing dact1/2 and convergent extension or wnt signaling in the ANC itself was not an aim of this paper.  

Comment (4) The authors also have not ruled out other possibilities that could cause the dact1/2-/- ANC phenotype. For example, increased cell death or reduced proliferation in the ANC may result in the phenotype, and changes in cell fate specification or differentiation in the second CNCC stream may also result in their inability to contribute to the ANC. 

We agree that we cannot rule out whether cell death or proliferation is different in the dact1/2 mutant ANC. However, because we do not find the second CNCC stream within the ANC, this is the most likely explanation for the abnormal ANC shape. Because the first stream of CNCC are able to populate the ANC and differentiate normally, it is most likely that the inability of the second stream to populate the ANC is due to steric hindrance imposed by the abnormal cranial/eye field morphology. These hypotheses would need to be tested, ideally with an inducible dact1/2 mutant, however, this is beyond the scope of this paper.     

Comment (5) The last paragraph of the section "Genetic interaction of dact1/2 with Wnt regulators..." misuses terms and conflates phenotypes observed. For instance, the authors wrote "dact2 haploinsuffciency in the context of dact1-/-; gpc4-/- double mutant produced ANC in the opposite phenotypic spectrum of ANC morphology, appearing similar to the gpc4-/- mutant phenotype". However, if heterozygous dact2 is not modulating phenotypes in this genetic background, its function is not "haploinsuffcient". The authors then said, "These results show that dact1 and dact2 do not have redundant function during craniofacial morphogenesis, and that dact2 function is more indispensable than dact1". However this statement should be confined to the context of modulating gpc4 phenotypes, which is not clearly stated. 

Revised (Lines 380, 382).   

Comment (6) For the scRNA-seq analysis, the authors should show the population distribution in the UMAP for the 3 genotypes, even if there are no obvious changes. The authors are encouraged, although not required, to perform pseudotime or RNA velocity analysis to determine if differentiation trajectories are changed in the NC populations, in light of what they found in Fig. 4. The authors can also check the expression of reporter genes downstream of certain pathways, e.g. axin2 in canonical wnt signaling, to query if these signaling activities are changed (also related to point #3 above). 

We have added population distribution data for the 3 genotypes to Supplemental Figure 4. Although RNA velocity analysis would be an interesting additional analysis, we would hypothesize that the NC population is not driving the differences in phenotype. Rather these are likely changes in the anterior neural plate and mesoderm. 

Comment (7) While the phenotypic difference between gpc4-/- and dact1/2-/- are in the ANC at a later stage, ssRNA-seq was performed using younger embryos. The authors should better explain the rationale and discuss how transcriptomic differences in these younger embryos can explain later phenotypes. Importantly, dact1, dact2, and capn8 expression were not shown in and around the ANC during its development and this information is crucial for interpreting some of the results shown in this paper. For example, if dact1 and dact2 are expressed during ANC development, they may have specific functions during that stage. Alternatively, if dact1 and dact2 are not expressed when the second stream CNCCs are found to be outside the ANC, then the ANC phenotype may be due to dact1/2's functions at an earlier time point. The author's statement in the discussion that "embryonic fields determined during gastrulation effect the CNCC ability to contribute to the craniofacial skeleton" is currently speculative. 

We have reworded our rationale and hypothesis to increase clarity (Lines 391-405). We believe that the ANC phenotype of the dact1/2 mutants is secondary to defective CE and anterior axis lengthening, as has been reported for the slb mutant (Heisenberg 1997, 2000). We utilized the gpc4 mutant as a foil to the dact1/2 mutant, as the gpc4 mutant has defective CE and axis extension without the same craniofacial phenotype.

We have added dact1 and dact2 WISH of 24 and 48 hpf (Fig1. D,E) to show expression during ANC development. 

Comment (8) The functional testing of capn8 did not yield a result that would suggest a strong effect, as only 1 in 142 animals phenocopied dact1/2. Therefore, while the result is interesting, the authors should tone down its importance. Alternatively, the authors can try knocking down capn8 in the dact1/2 mutants to test how that affects the CE phenotype during axis elongation, as well as ANC morphogenesis. 

As overexpression of capn8 in wildtype animals did not result in a significant phenotype, we tested capn8 overexpression in compound dact1/2 mutants as these have a sensitized background. We found a small but statistically significant effect of exogenous capn8 in dact1+/-,dact2+/- animals. While the effect is not what one would expect comparing to Mendelian genetic ratios, the rod-like ANC phenotype is an extreme craniofacial dysmorphology not observed in wildtype or mRNA injected embryos hence significant. The experiment is limited by the available technology of over-expressing mRNA broadly without temporal or cell specificity control. It is possible that if capn8 over-expression was restricted to specific cells (floor plate, notochord or mesoderm) and at the optimal time period during gastrulation/segmentation that the aberrant ANC phenotype would be more robust. We agree with the reviewer that although the finding of a new role for capn8 during development is interesting, its importance in the context of dact should be toned down and we have altered the manuscript accordingly (Lines 455-467).  

Comment (9) A difference between the two images in Fig. 8B is hard to distinguish.

Consider showing flat-mount images. 

We have added flat-mount images to Fig. 8B

Minor comments:

Comment (1) wnt11f2 is spelled incorrectly in a couple of places, e.g. "wnt11f2l" in the abstract and "wntllf2" in the discussion. 

Revised throughout.

Comment (2) For Fig. 1D, the white dact1 and yellow dact2 are hard to distinguish in the merged image. Consider changing one of their colors to a different one and only merge dact1 and dact2 without irf6 to better show their complementarity.  

We agree with the reviewer that the expression patterns of dact1 and dact2 are difficult to distinguish in the merged image. We have added outlines of the cartilage elements to the images to facilitate comparisons of dact1 and dact2 expression (Fig 1F). 

Comment (3) For Fig. 1E, please label the clusters mentioned in the text so readers can better compare expressions in these cell populations.  

We have moved this data to supplementary figure S1 and have added labels.

Comment (4) The citing and labelling of certain figures can be more specific. For example, Fig. S1A, B, and Fig. S1C should be used instead of just Fig. S1 (under the section titled dact1 and dact2 contribute to axis extension...". Similarly, Fig. 4 can be better labeled with alphabets and cited at the relevant places in the text.  

We have modified the labeling of the figures according to the reviewer’s suggestion (Fig S2 (previously S1), Fig4) and have added reference to these labels in the text (Lines 202, 204, 212, 328, 334, 336). 

Comment (5) For Fig. 2B, the (+/+,-/-) on x-axis should be (+/-,-/-).  

Revised in Figure 2B.

Comment (6) Several figures are incorrectly cited. Fig. 2C is not cited, and the "Fig. 2C" and "Fig. 2D" cited in the text should be "Fig. 2D" and "Fig. 2E" respectively. Similarly, Fig. 5C and D are not cited in the text and the cited Fig. 5C should be 5E. The VC images in Fig. 5 are not talked about in the text. Finally, Fig. 7C was also not mentioned in the text.  

We have corrected the labeling and have added descriptions of each panel in the Results (Fig.2 Line 231, 237, 242, Fig 5 Line 373, 381, Fig 7 line 431). 

Comment (7) In the main text, it is indicated that zebrafish at 3ss were used for ssRNAseq, but in the figure legend, it says 4ss. 

Revised (Line 682)

Comment (8) No error bars in Fig. S1B and the difference between the black and grey shades in Fig. S1D is not explained.  

Error bars are not included in the graphs of qPCR results (now Fig S2C) as these are results of a pool of 8 embryos performed one time. We have added a legend to explain the gray vs. black bars (now Fig S2E). 

Reviewer #3 (Public Review):  

Weaknesses: The hypotheses are very poorly defined and misinterpret key previous findings surrounding the roles of wnt11 and gpc4, which results in a very confusing manuscript. Many of the results are not novel and focus on secondary defects. The most novel result of overexpressing calpain8 in dact1/2 mutants is preliminary and not convincing.  

We apologize for not presenting the question more clearly. The Introduction was revised with particular attention to distinguish this work using genetic germline mutants from prior morpholino studies. Please refer to pages 4-5, lines 106-121.

Weakness 1) One major problem throughout the paper is that the authors misrepresent the fact that wnt11f2 and gpc4 act in different cell populations at different times. Gastrulation defects in these mutants are not similar: wnt11 is required for anterior mesoderm CE during gastrulation but not during subsequent craniofacial development while gpc4 is required for posterior mesoderm CE and later craniofacial cartilage morphogenesis (LeClair et al., 2009). Overall, the non-overlapping functions of wnt11 and gpc4, both temporally and spatially, suggest that they are not part of the same pathway.  

We have reworded the text to add clarity. While the loss of wnt11 versus the loss of gpc4 may affect different cell populations, the overall effect is a shortened body axis. We stressed that it is this similar impaired axis elongation phenotype but discrepant ANC morphology phenotypes in the opposite ends of the ANC morphologic spectrum that is very interesting and leads us to investigate dact1/2 in the genetic contexts of wnt11f2 and gpc4.  Pls refer to page 4, lines 73-84. Further, the reviewer’s comment that wnt11 and gpc4 are spatially and temporally distinct is untested. We think the reviewer’s claim of gpc4 acting in the posterior mesoderm refers to its requirement in the tailbud (Marlow 2004). However this does not exclude gpc4 from acting elsewhere as well. Further experiments would be necessary. Both wnt11f2 and gpc4 regulate non-canonical wnt signaling and are coexpressed during some points of gastrulation and CF development (Gupta et al., 2013; Sisson 2015). This data supports the possibility of overlapping roles. 

Weakness 2) There are also serious problems surrounding attempts to relate single-cell data with the other data in the manuscript and many claims that lack validation. For example, in Fig 1 it is entirely unclear how the Daniocell scRNA-seq data have been used to compare dact1/2 with wnt11f2 or gpc4. With no labeling in panel 1E of this figure these comparisons are impossible to follow. Similarly, the comparisons between dact1/2 and gpc4 in scRNA-seq data in Fig. 6 as well as the choices of DEGs in dact1/2 or gpc4 mutants in Fig. 7 seem arbitrary and do not make a convincing case for any specific developmental hypothesis. Are dact1 and gpc4 or dact2 and wnt11 coexpressed in individual cells? Eyeballing similarity is not acceptable.  

We have moved the previously published Daniocell data to Figure S1 and have added labeling. These data are meant to complement and support the WISH results and demonstrate the utility of using available public Daniocell data. Please recommend how we can do this better or recommend how we can remediate this work with specific comment. 

Regarding our own scRNA-seq data, we have added rationale (line 391-403) and details of the results to increase clarity (Lines 419-436). We have added a panel to Figure 6 (panel A) to help illustrate or rationale for comparing dact1/2 to gpc4 mutants to wt. The DEGs displayed in Fig.7A are the top 50 most differentially expressed genes between dact1/2 mutants and WT (Figure 7 legend, line 422-424).   

We have looked at our scRNA-seq gene expression results for our clusters of interest (lateral plate mesoderm, paraxial mesoderm, and ectoderm). We find dact1, dact2, and gpc4 co-expression within these clusters. Knowing whether these genes are coexpressed within the same individual cell would require going back and analyzing the raw expression data. We do not find this to be necessary to support our conclusions. The expression pattern of wnt11f2 is irrelevant here.   

Weakness 3) Many of the results in the paper are not novel and either confirm previous findings, particularly Waxman et al (2004), or even contradict them without good evidence. The authors should make sure that dact2 loss-of-function is not compensated for by an increase in dact1 transcription or vice versa. Testing genetic interactions, including investigating the expression of wnt11f2 in dact1/2 mutants, dact1/2 expression in wnt11f2 mutants, or the ability of dact1/2 to rescue wnt11f2 loss of function would give this work a more novel, mechanistic angle.

We clarified here that the prior work carried out by Waxman using morppholinos, while acceptable at the time in 2004, does not meet the rigor of developmental studies today which is to generate germline mutants. The reviewer’s acceptance of the prior work at face value fails to take the limitation of prior work into account. Further, the prior paper from Waxman et al did not analyze craniofacial morphology other than eyeballing the shape of the head and eyes. Please compare the Waxman paper and this work figure for figure and the additional detail of this study should be clear. Again, this is by no means any criticism of prior work as the prior study suffered from the technological limitations of 2004, just as this study also is the best we can do using the tools we have today. Any discrepancies in results are likely due to differences in morpholino versus genetic disruption and most reviewers would favor the phenotype analysis from the germline genetic context. We have addressed these concerns as objectively as we can in the text (Lines 482-493). The fact that dact1/2 double mutants display a craniofacial phenotype while the single mutants do not, suggests compensation (Lines 503-505), but not necessarily at the mRNA expression level (Fig. S2C). 

This paper tests genetic interaction through phenotyping the wntll/dact1/dact2 mutant.

Our results support the previous literature that dact1/2 act downstream of wnt11 signaling. There is no evidence of cross-regulation of gene expression. We do not expect that changes in wnt11 or dact would result in expression changes in the others.

RNA-seq of the dact1/2 mutants did not show changes in wnt11 gene expression. Unless dact1 and/or dact2 mRNA are under expressed in the wnt11 mutant, we would not expect a rescue experiment to be informative. And as wnt11 is not a focus of this paper, we have not performed the experiment.  

Weakness 4) The identification of calpain 8 overexpression in Dact1/2 mutants is interesting, but getting 1/142 phenotypes from mRNA injections does not meet reproducibility standards.

As the occurrence of the mutant phenotype in wildtype animals with exogenous capn8 expression was below what would meet reproducibility standards, we performed an additional experiment where capn8 was overexpressed in embryos resulting from dact1/dact2 double heterozygotes incross (Fig. 8). We reasoned that an effect of capn8 overexpression may be more robust on a sensitized background. We found a statistically significant effect of capn8 in dact1/2 double heterozygotes, though the occurrence was still relatively rare (6/80). These data suggest dysregulation of capn8 contributes to the mutant ANC phenotype, though there are likely other factors involved. 

Comment: The manuscript title is not representative of the findings of this study.  

We revised the title to strictly describe that we generated and carried out genetic analysis in loss of function compound mutants (Genetic requirement) and that we found capn8 was important which modified this requirement.

Introduction: p.4:

Comment: Anterior neurocranium (ANC) - it has to be stated that this refers to the combined ethmoid plate and trabecular cartilages. 

Thank you, we agree that the ANC and ethmoid plate terminology has been confusing in the literature and we should endeavor to more clearly describe that the phenotypes in question are all in the ethmoid plate and the trabeculae are not affected. ANC has been replaced with ethmoid plate (EP) throughout the manuscript and figures. We also describe that all the observed phenotypes affect the ethmoid plate and not the trabeculae, (pages 13, Lines 265-267).

Comment: Transverse dimension is incorrect terminology - replace with medio-lateral.

Revised (Lines 69, 74).

Comment: Improper way of explaining the relationship between mutant and gene..."Another mutant knypek, later identified as gpc4..." a better  way to explain this would be that the knypek mutation was found to be a non-sense mutation in the gpc4 gene.  

Revised (Line 71)

Comment: "...the gpc4 mutant formed an ANC that is wider in the transverse dimension than the wildtype, in the opposite end of the ANC phenotypic spectrum compared to wnt11f2...These observations beg the question how defects in early patterning and convergent extension of the embryo may be associated with later craniofacial morphogenesis."

This statement is broadly representative of the general failure to distinguish primary from secondary defects in this manuscript. Focusing on secondary defects may be useful to understand the etiology of a human disease, but it is misleading to focus on secondary defects when studying gene function. The rod-like ethmoid of slb mutant results from a CE defect of anterior mesoderm during gastrulation(Heisenberg et al. 1997, 2000), while the wide ethmoid plate of kny mutants results from CE defects of cartilage precursors (Rochard et al., 2016). Based on this evidence, wnt11f2 and gpc4 act in different cell populations at different times.  

It is true that the slb mutant craniofacial phenotype has been stated as secondary to the CE defect during gastrulation and the kny phenotype as primary to chondrocyte CE defects in the ethmoid, however the direct experimental evidence to conclude only primary or only secondary effects does not yet exist. There is no experiment to our knowledge where wnt11f2 was found to not affect ethmoid chondrocytes directly. Likewise, there is no experiment having demonstrated that dysregulated CE in gpc4 mutants does not contribute to a secondary abnormality in the ethmoid. 

Here, we are analyzing the CE and craniofacial phenotypes of the dact1/2 mutants without any assumptions about primary or secondary effects and without drawing any conclusions about wnt11f2 or gpc4 cellular mechanisms.     

Comment: "The observation that wnt11f2 and gpc4 mutants share similar gastrulation and axis extension phenotypes but contrasting ANC morphologies supports a hypothesis that convergent extension mechanisms regulated by these Wnt pathway genes are specific to the temporal and spatial context during embryogenesis."

This sentence is quite vague and potentially misleading. The gastrulation defects of these 2 mutants are not similar - wnt11 is required for anterior mesoderm CE during gastrulation and has not been shown to be active during subsequent craniofacial development while gpc4 is required for posterior mesoderm CE and craniofacial cartilage morphogenesis (LeClair et al., 2009). Here again, the non-spatially overlapping functions of wnt11 and gpc4 suggest that are not part of the same pathway.  

Though the cells displaying defective CE in wnt11f2 and gpc4 mutants are different, the effects on the body axis are similar. The dact1/2 showed a similar axis extension defect (grossly) to these mutants. Our aim with the scRNA-seq experiment was to determine which cells and gene programs are disrupted in dact1/2 mutants. We found that some cell types and programs were disrupted similarly in dact1/2 mutants and gpc4 mutants, while other cells and programs were specific to dact1/2 versus gpc4 mutants. We can speculate that these that were specific to dact1/2 versus gpc4 may be attributed to CE in the anterior mesoderm, as is the case for wnt11. 

p.5

Comment: "We examined the connection between convergent extension governing gastrulation, body axis segmentation, and craniofacial morphogenesis." A statement focused on the mechanistic findings of this paper would be welcome here, instead of a claim for a "connection" that is vague and hard to find in the manuscript.  

We have rewritten this statement (Line 125).

p.7 Results:

Comment: It is unclear why Farrel et al., 2018 and Lange et al., 2023 are appropriate references for WISH. Please justify or edit.  

This was a mistake and has been edited (Page 9).

Comment: " Further, dact gene expression was distinct from wnt11f2." This statement is inaccurate in light of the data shown in Fig1A and the following statements - please edit to reflect the partially overlapping expression patterns.  

We have edited to clarify (Lines 142-143).

p.8

Comment: "...we examined dact1 and 2 expression in the developing orofacial tissues. We found that at 72hpf..." - expression at 72hpf is not relevant to craniofacial morphogenesis, which takes place between 48h-60hpf (Kimmel et al., 1998; Rochard et al., 2016; Le Pabic et al., 2014).  

We have included images and discussion of dact1 and dact2 expression at earlier time points that are important to craniofacial development (Lines 160-171)(Fig 1D,E). 

Comment: "This is in line with our prior finding of decreased dact2 expression in irf6 null embryos". - This statement is too vague. How are th.e two observations "in line".  

We have removed this statement from the manuscript.

Comment: Incomplete sentence (no verb) - "The differences in expression pattern between dact1 and dact2...".  

Revised (Line 172).

Comment: "During embryogenesis..." - Please label the named structures in Fig.1E.

Please be more precise with the described expression time. Also, it would be useful to integrate the scRNAseq data with the WISH data to create an overall picture instead of treating each dataset separately.  

We have moved the previously published Daniocell data to supplementary figure S1 and have labeled the key cell types. 

p.9

Comment: "The specificity of the gene disruption was demonstrated by phenotypic rescue with the injection of dact1 or dact2 mRNA (Fig. S1)." - please describe what is considered a phenotypic rescue.

-The body axis reduction of dact mutants needs to be documented in a figure. Head pictures are not sufficient. Is the head alone affected, or both the head and trunk/tail? Fig.2E suggests that both head and trunk/tail are affected - please include a live embryos picture at a later stage.  

We have added a description of how phenotypic rescue was determined (Line 208). We have added a figure with representative images of the whole body of dact1/2 mutants. Measurements of body length found a shortening in dact1/2 double mutants versus wildtype, however differences were not found to be significantly different by ANOVA (Fig. 3C, Fig. S3, Line 270-275).

p. 11

Comment: "These dact1-/-;dact2-/- CE phenotypes were similar to findings in other Wnt mutants, such as slb and kny (Heisenberg, Tada et al., 2000; Topczewski, Sepich et al., 2001)." The similarity between slb and kny phenotypes should be mentioned with caution as CE defects affect different regions in these 2 mutants. It is misleading to combine them into one phenotype category as wnt11 and gpc4 are most likely not acting in the same pathway based on these spatially distinct phenotypes.  

Here we are referring to the grossly similar axis extension defects in slb and kny mutants. We refer to these mutants to illustrate that dact1 and or 2 deficiency could affect axis extension through diverse mechanisms. We have added text for clarity (Lines 249-252).  

Comment: "No craniofacial phenotype was observed in dact1 or dact2 single mutants. However, in-crossing to generate [...] compound homozygotes resulted in dramatic craniofacial deformity."

This result is intriguing in light of (1) the similar craniofacial phenotype previously reported by Waxman et al (2004) using morpholino- based knock-down of dact2, and the phenomenon of genetic compensation demonstrated by Jakutis and Stainier 2001 (https://doi.org/10.1146/annurev-genet-071719-020342). The authors should make sure that dact2 loss-of-function is not compensated for by an increase in dact1 transcription, as such compensation could lead to inaccurate conclusions if ignored.  

We agree with the reviewer that genetic compensation of dact2 by dact1 likely explains the different result found in the dact2 morphant versus CRISPR mutant. We found increased dact1 mRNA expression in the dact2-/- mutant (Fig S2X) however a more thorough examination is required to draw a conclusion. Interestingly, we found that in wildtype embryos dact1 and dact2 expression patterns are distinct though with some overlap. It would be informative to investigate whether the dact1 expression pattern changes in dact2-/- mutants to account for dact2 loss.   

Comment: "Lineage tracing of NCC movements in dact1/2 mutants reveals ANC composition" - the title is misleading - ANC composition was previously investigated by lineage tracing (Eberhardt et al., 2006; Wada et al., 2005).  

This has been reworded (Line 292)

p.13

Comment: There is no frontonasal prominence in zebrafish.  

This is true, texts have been changed to frontal prominence.  (Lines 293,

299, 320)

Comment: The rationale for investigating NC migration in mutants where there is a gastrula-stage failure of head mesoderm convergent extension is unclear. The whole head is deformed even before neural crest cells migrate as the eye field does not get split in two (Heisenberg et al., 1997; 2000), suggesting that the rod-like ethmoid plate is a secondary defect of this gastrula-stage defect. In addition, neural crest migration and cartilage morphogenesis are different processes, with clear temporal and spatial distinctions.  

We carried out the lineage tracing experiment to determine which NC streams contributed to the aberrantly shaped EP, whether the anteromost NC stream frontal prominence, the second NC stream of maxillary prominence, or both.  We found that the anteromost NCC did contribute to the rod-like EP, which is different from when hedgehod signaling is disrupted,  So while it is possible that the gastrula-effect head mesoderm CE caused a secondary effect on NC migration, how the anterior NC stream and second NC stream are affected differently between dact1/2 and shh pathway is interesting.  We added discussion of this observation to the manuscript (page 23, Lines 514-520). 

p. 14-16

Comment: Based on the heavy suspicion that the rod-like ethmoid plate of the dact1/2 mutant results from a gastrulation defect, not a primary defect in later craniofacial morphogenesis, the prospect of crossing dact1/2 mutants with other wnt-pathway mutants for which craniofacial defects result from craniofacial morphogenetic defects is at the very least unlikely to generate any useful mechanistic information, and at most very likely to generate lots of confusion. Both predictions seem to take form here.  

However, the ethmoid plate phenotype observed in the gpc4-/-; dact1+/-; dact2-/- mutants (Fig. 5E) does suggest that gpc4 may interact with dact1/2 during gastrulation, but that is the case only if dact1+/-; dact2-/- mutants do not have an ethmoid cartilage defect, which I could not find in the manuscript. Please clarify.  

The perspective that the rod-like EP of the dact1/2 is due to gastrulation defect is being examined here. Why would other mutants such as wnt11f2 and gpc4 that have gastrulation CE defects have very different EP morphology, whether primary or secondary NCC effect?  Further dact1 and dact2 were reported as modifiers of Wnt signaling, so it is logical to genetically test the relationship between dact1, dact2, wnt11f2, gpc4 and wls. The experiment had to be done to investigate how these genetic combinations impact EP morphology. This study found that combined loss of dact1, dact2 and wls or gpc4 yielded new EP morphology different than those previously observed in either dact1/2, wls, gpc4, or any other mutant is important, suggesting that there are distinct roles for each of these genes contributing to facial morphology, that is not explained by CE defect alone.   

Comment: I encourage the authors to explore ways to test whether the rod-like ethmoid of dact1/2 mutants is more than a secondary effect of the CE failure of the head mesoderm during gastrulation. Without this evidence, the phenotypes of dact1/2 -gpc4 or - wls are not going to convince us that these factors actually interact.  

Actually, we find our results to support the hypothesis that the ethmoid of the dact1/2 mutants is a secondary effect of defective gastrulation and anterior extension of the body axis. However, our findings suggest (by contrasting to another mutant with impaired CE during gastrulation) that this CE defect alone cannot explain the dysmorphic ethmoid plate. Our single-cell RNA seq results and the discovery of dysregulated capn8 expression and proteolytic processes presents new wnt-regulated mechanisms for axis extension.    

p. 20 Discussion

Comment: "Here we show that dact1 and dact2 are required for axis extension during gastrulation and show a new example of CE defects during gastrulation associated with craniofacial defects."

Waxman et al. (2004) previously showed that dact2 is involved in CE during gastrulation.

Heisenberg et al. (1997, 2000), previously showed with the slb mutant how a CE defect during gastrulation causes a craniofacial defect.  

The Waxman paper using morpholino to disrupt dact2 is produced limited analysis of CE and no analysis of craniofacial morphogenesis. We generated genetic mutants here to validate the earlier morpholino results and to analyze the craniofacial phenotype in detail. We have removed the word “new” to make the statement more clear (Line 475).

Comment: "Our data supports the hypothesis that CE gastrulation defects are not causal to the craniofacial defect of medially displaced eyes and midfacial hypoplasia and that an additional morphological process is disrupted."

It is unclear to me how the authors reached this conclusion. I find the view that medially displaced eyes and midfacial hypoplasia are secondary to the CE gastrulation defects unchallenged by the data presented. 

This statement was removed and the discussion was reworded.

Comment: The discussion should include a detailed comparison of this study's findings with those of zebrafish morpholino studies.  

We have added more discussion to compare ours to the previous morpholino findings (Lines 476-484).

Comment: The discussion should try to reconcile the different expression patterns of dact1 and dact2, and the functional redundancy suggested by the absence of phenotype of single mutants. Genetic compensation should be considered (and perhaps tested).  

The different expression patterns of dact1 and dact2 along with our finding that dact1 and dact2 genetic deficiency differently affect the gpc4 mutant phenotype suggest that dact1 and dact2 are not functionally redundant during normal development. This is in line with the previously published data showing different phenotypes of dact1 or dact2 knockdown. However, our results that genetic ablation of both dact1 and dact2 are required for a mutant phenotype suggests that these genes can compensate upon loss of the other. This would suggest then that the expression pattern of dact1 would be changed in the dact2 mutant and visa versa. We find that this line of investigation would be interesting in future studies. We have addressed this in the Discussion (Lines 485498).

Comment: "Based on the data...Conversely, we propose...ascribed to wnt11f2 "

Functional data always prevail overexpression data for inferring functional requirements.  

This is true.

p.21

Comment: "Our results underscore the crucial roles of dact1 and dact2 in embryonic development, specifically in the connection between CE during gastrulation and ultimate craniofacial development."

How is this novel in light of previous studies, especially by Waxman et al. (2004) and Heisenberg et al. (1997, 2000). In this study, the authors fail to present compelling evidence that craniofacial defects are not secondary to the early gastrulation defects resulting from dact1/2 mutations.  p. 22

We have not claimed that the craniofacial defects are not secondary to the gastrulation defects. In fact, we state that there is a “connection”. Further, we do not claim that this is the first or only such finding. We believe our findings have validated the previous dact morpholino experiments and have contributed to the body of literature concerning wnt signaling during embryogenesis. 

Comment: The section on Smad1 discusses a result not reported in the results section. Any data discussed in the discussion section needs to be reported first in the results section.  

We have added a comment on the differential expression of smad1 to the results section (Lines 446-448).

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

In this manuscript entitled "Hexokinase regulates Mondo-mediated longevity via the PPP and organellar dynamics", Laboy and colleagues investigated upstream regulators of MML-1/Mondo, a key transcription factor that regulates aging and metabolism, using the nematode C. elegans and cultured mammalian cells. By performing a targeted RNAi screen for genes encoding enzymes in glucose metabolism, the authors found that two hexokinases, HXK-1 and HXK-2, regulate nuclear localization of MML-1 in C. elegans. The authors showed that knockdown of hxk-1 and hxk-2 suppressed longevity caused by germline-deficient glp-1 mutations. The authors demonstrated that genetic or pharmacological inhibition of hexokinases decreased nuclear localization of MML-1, via promoting mitochondrial β-oxidation of fatty acids. They found that genetic inhibition of hxk-2 changed the localization of MML-1 from the nucleus to mitochondria and lipid droplets by activating pentose phosphate pathway (PPP). The authors further showed that the inhibition of PPP increased the nuclear localization of mammalian MondoA in cultured human cells under starvation conditions, suggesting the underlying mechanism is evolutionarily conserved. This paper provides compelling evidence for the mechanisms by which novel upstream metabolic pathways regulate MML-1/Mondo, a key transcription factor for longevity and glucose homeostasis, through altering organelle communications, using two different experimental systems, C. elegans and mammalian cells. This paper will be of interest to a broad range of biologists who work on aging, metabolism, and transcriptional regulation. 

Reviewer #2 (Public Review):

Raymond Laboy et.al explored how transcriptional Mondo/Max-like complex (MML-1/MXL-2) is regulated by glucose metabolic signals using germ-line removal longevity model. They believed that MML-1/MXL-2 integrated multiple longevity pathways through nutrient sensing and therefore screened the glucose metabolic enzymes that regulated MML-1 nuclear localization. Hexokinase 1 and 2 were identified as the most vigorous regulators, which function through mitochondrial beta-oxidation and the pentose phosphate pathway (PPP), respectively. MML-1 localized to mitochondria associated with lipid droplets (LD), and MML-1 nuclear localization was correlated with LD size and metabolism. Their findings are interesting and may help us to further explore the mechanisms in multiple longevity models, however, the study is not complete and the working model remains obscure. For example, the exact metabolites that account for the direct regulation of MML-1 were not identified, and more detailed studies of the related cellular processes are needed. 

The identification of responsible metabolites is necessary since multiple pieces of evidence from the study suggests that lipid other than glucose metabolites may be more likely to be the direct regulator of MML-1 and HXK regulate MML-1 indirectly by affecting the lipid metabolism: 1) inhibiting the PPP is sufficient to rescue MML-1 function independent of G6P levels; 2) HXK-1 regulates MML-1 by increasing fatty acid beta-oxidation; 3) LD size correlates with MML-1 nuclear localization and LD metabolism can directly regulate MML-1. The identification of metabolites will be helpful for understanding the mechanism. 

Beta-oxidation and the PPP are involved in the regulation of MML-1 by HXK-1 and HXK-2, respectively. But how these two pathways participate in the regulation is not clear. Is it the beta-oxidation rate or the intermediate metabolites that matters? As for the PPP, it provides substrates for nucleotide synthesis and also its product NADPH is essential for redox balance. Is one of the metabolites or the NADPH levels involved in MML-1 regulation? More studies are needed to provide answers to these concerns. 

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Following are my comments that the authors may want to address to further improve this excellent paper.

Major comments 

(1) Although the authors provided evidence that hexokinases in glucose metabolism are associated with germline-deficient glp-1(-) mutants, they did not mention why they focused on glp-1(-) mutants rather than other longevity mutants. In their previous study (Nakamura et al., 2016), they showed that MML-1 is required for multiple longevity pathways in C. elegans, including reduced mitochondrial respiration and insulin/IGF-1 signaling. Please discuss why the authors focused on glp-1(-) mutants in this paper. It will be even better if the authors test the roles of hexokinases in some other longevity regimens. 

Many thanks for this astute comment. Previously we had shown that mml-1 is required for glp-1, daf-2, and isp-1 longevity, and Johnson et al. had shown a requirement for eat-2, hence the idea that MML-1 is a convergent transcription factor. We first focused on glp-1 because that was the starting point of our screen, and the result was clear and simple: hexokinases regulate MML‑1 nuclear localization and activity in glp-1 and are required for longevity. Naturally, the question arises: do hexokinases behave like MML-1 as convergent longevity regulators across pathways? To address this, we examined the interaction of hxk-1 and hxk-2 with isp-1, daf-2, and raga-1.  Specifically, we now show that:

A. Like glp-1(e2141) mutants, isp-1(qm150) mutants stimulate MML-1 nuclear localization, and the hexokinases are required for isp-1 longevity (Figure 1G-H).

B. daf-2(e1370) mutants do not further stimulate MML-1 nuclear localization beyond basal levels, yet MML-1 is strongly required for daf-2 longevity (Nakamura et al., 2016, Supplementary Figure 1L-M). However, the hexokinases are not required for daf-2 longevity (Supplementary Figure 1M), suggesting that the signaling pathway is wired differently in daf-2, and that other pathways regulate MML-1 activity.

C. raga-1(ok701) mutants stimulate MML-1 nuclear localization and mml-1 is required for raga-1 longevity, suggesting that MML-1 acts downstream of TORC1 signaling (Supplementary Figure 1N-O). However, hexokinases are not required for raga-1 longevity, suggesting that raga-1 acts downstream or parallel to hexokinase signaling (Supplementary Figure 1P).

D. We performed untargeted metabolomics in glp-1, daf-2, and mml-1 single and double mutants and observed that hexose phosphates, which have been shown to regulate MML-1 human homologs MondoA/ChREBP, were differentially regulated between mutants.

Author response image 1.

E. Altogether these experiments reveal that though MML-1 promotes longevity in most pathways, the hexokinases are only required in some (glp-1, isp-1), but not others (raga-1, daf-2). Furthermore, strong MML-1 nuclear localization is often but not always associated with longevity (e.g. daf-2), and the wiring of the signaling pathway is different for various longevity regimens. Consistently, mTOR and Insulin signaling are more functionally linked and therefore may show a more similar genetic profile. Differences in hexose phosphate between glp-1 and daf-2 could explain why MML-1 requires hexokinase function in glp-1 to promote longevity but not in daf-2. However, considerably more work is required to rigorously validate this hypothesis.

(2) In figure 5, the authors investigated whether the association between PPP and MML‑1/MondoA, tested in C. elegans, is conserved in mammals under starvation conditions. The authors should clarify why they tested the MondoA localization upon starvation in cultured human cells. This comment is related to my comment #1 as the authors could determine the roles of hexokinases under dietary restriction (DR)-conditions or in DR-mimetic in eat-2(-) mutants. 

In this case, the actual translatability to a worm longevity pathway was not our goal. Rather, we examined MondoA in cell culture under contrasting conditions of MondoA subcellular localization, where high glucose media had cytosolic/nuclear localization and starvation conditions cytosolic localization. We then showed that similar to our data in worms, PPP inhibition with 6-AN induced MondoA nuclear localization and activity. We now mention this rationale in the results section, lines 352-356.

(3) In figure 2, the authors showed that HXK-2 regulates mitochondrial localization of MML-1, and HXK-1 regulates nuclear localization of MML-1 through mitochondrial β-oxidation in glp‑1(-) mutants. Can the authors test whether mitochondrial β-oxidation affects the effects of hxk RNAi on longevity of glp-1(-) mutants? 

Excellent suggestion. We tried to test this idea and found that acs-2 RNAi alone abolished glp-1 longevity, making epistasis experiments difficult to interpret. This is consistent with published data showing that glp-1 longevity requires NHR-49, a transcription factor that regulates mitochondrial b‑oxidation, that drives acs-2 expression (Ratnappan et al., 2014). It could well be that b‑oxidation inhibition promotes MML-1 nuclear localization but abolishes lifespan extension because of epistatic effects on other transcription factors or processes. Further investigation would be required to elucidate the exact mechanism that goes beyond the scope of the paper.

(4) The authors showed that 2-deoxy-glucose, which decreases the activity of HXK, decreased the nuclear localization of MML-1, and this is consistent with their genetic data. Based on these data, 2-deoxy-glucose is expected to decrease longevity. Interestingly, however, 2-deoxy-glucose has been reported to increase lifespan by restricting glucose, whereas extra glucose intake decreases lifespan in C. elegans, shown by multiple research groups, including M. Ristow, C. Kenyon, and S.J.V. Lee labs. This is seemingly paradoxical and worth discussing with key references, especially because MondoA and Chrebp are known as glucose-responsive transcription factors. 

Thank you for this important comment. 2-DG has been shown to extend lifespan by suppressing glucose metabolism at concentrations ranging from 0.1 to 5 mM, higher concentrations ranging from 20 to 50 mM had the opposite effect decreasing lifespan (Schulz et al., 2007). The concentration we tested was 50 mM 2-DG and observed decreased MML-1 nuclear localization, which is consistent with the previous data showing decreased longevity. We now raise this point in the discussion suggesting that mild inhibition of glucose metabolism has beneficial effects on longevity, while strong suppression causes a shortening of the lifespan (lines 411-414).

Minor comments 

(1) The current Introduction does not include the explicit statement about that MML-1 and MondoA are homologs. Please clarify this as naive readers may be confused.

Thank you for pointing this out. We now say in the intro that MondoA and MML-1 are homologs (lines 59-60).

(2) In figure 1, the effects of hxk-3 on nuclear localization of MML-1 is small compared to those of hxk-1 and hxk-2. Please add speculation about why HXK-3 has different roles in nuclear localization of MML-1 compared to HXK-1 and HXK-2. 

According to GExplore 1.4 (Hutter & Suh, 2016), hxk-3 expression declines during larval development and is low expressed in the adult. Perhaps it has little effect in the young adult, and the other hexokinases suffice to support MML-1 nuclear localization. It also remains possible that hxk-3 is not required in glp-1, but required in other longevity pathways.

(3) The authors tested the effects of genetic inhibition of hxk-1 and hxk-2 on the regulation of MML-1 localization and lifespan of glp-1(-) mutants by using RNAi. I wonder whether the authors can perform the experiments with hxk-1 or hxk-2 loss (or reduction) of function mutants. If they cannot, please discuss the reason and the limitations of RNAi. 

This is an important point raised by the reviewer. We found that RNAi was most effective for phenotypes related to MML-1 nuclear localization and longevity, likely because it results in acute knockdown. We also showed that pharmacological inhibition of hexokinase function with 3BrP and 2‑DG (Supplementary Figure 1B and 1C) and the PPP with 6-AN (Figure 3B) had consistent results with our observation with RNAi.

We generated hexokinase KO mutants by deleting the coding sequence of each hexokinase by CRISPR/Cas9. First, we measured the expression of each hexokinase isozyme in each mutant. Notably, hxk-1(syb1271) null mutant had higher expression of hxk-2 and hxk-3, hxk-2(syb1261) did not significantly affect the expression of hxk-1 and hxk-3, and hxk-3(syb1267) had a mild increase in hxk-2 expression. We followed up on the hxk-1(syb1271) and hxk-2(syb1261) and crossed these mutants with our MML-1::GFP reporter. We observed a modest but significant reduction in MML-1 nuclear localization in both strains. The effect with RNAi is much stronger in comparison to the null mutants, potentially due to a compensatory upregulation of the other hexokinases in the mutants that we do not observe with RNAi (Supplementary Figure 1D-E). Another alternative is that there is a threshold in the effects of hexokinase function on MML-1 nuclear localization. We tried to generate a hxk-1; hxk-2 double mutant but it was lethal and therefore did not pursue this further.

Author response image 2.

(4) Please correct minor typos throughout the manuscript. Following are some examples. <br /> - On page 4, line 111, please correct "Supplementary Figure D-E" to "Supplementary Figure 1D-E". 

- On page 9, line 272, please correct "3A-B" to "4A-B". 

- On page 9, line 275, please correct "S4" to "4". 

- On page 10, line 309, please correct "4A" to "4B" 

Corrected.

(5) In Fig. 3E, please add the information about the scale bars in figure legends.

Corrected.

Reviewer #2 (Recommendations For The Authors):

Here are some detailed suggestions for the authors:

(1) Since MML-1/MXL-2 complex functions in multiple longevity models, e.g. DR, ILS, what are the roles of HXK-1 and HXK-2 in these models? 

We now show that although mml-1 is required in most longevity pathways, hxk-1 and hxk-2 are required in some pathways (glp-1, isp-1) but not others (daf-2, raga-1). See above for more details.

(2) As for the metabolites screening, the lipid metabolic genes can be included. Not only for the above reasons, also previous study had found that the mml-1 mRNA levels and MML-1 GFP nuclear localization were all increased in the glp-1 model, while mml-1 mRNA levels were unaffected by hxk knockdown, suggesting more pathways be involved. 

We agree with the reviewer that understanding what metabolites regulate MML-1 nuclear localization and activity is an important, yet challenging question. Our studies demonstrate a role of glucose metabolism, in particular, hexokinase in this process, consistent with hexose-p being activators of MondoA. Our data also suggest mechanisms beyond hexose-p regulate MML-1, since knockdown of the PPP components stimulates MML-1 even when hxk-2 is depleted and low G6P, and inhibition of the PPP with 6-AN stimulates MondoA nuclear localization under starvation conditions in mammalian cell culture. We tested redox regulation, nucleoside, and lipid metabolism as candidate processes (see below). Notably, our data suggest this other mechanism is tied to lipid metabolism through droplet size since various perturbations that impact LD size and number (atgl-1, dgat-2, tkt-1, Figure 4) affected MML-1 nuclear localization. It remains an open question whether MML-1 is regulated by other metabolites through a ligand-protein interaction or not. We cannot exclude that beyond lipid droplet regulation, specific lipids, other metabolites, or metabolic modules linked to the PPP might regulate MML-1 nuclear localization and activity.

We employed genetic manipulation and pharmacological inhibition to understand the upstream signals that regulate MML-1. These approaches will not be sufficient to determine whether other metabolite(s) are involved in MML-1/MondoA translocation to the nucleus through a direct interaction. Novel technologies that determine protein-metabolite interactions (e.g. MIDAS) will help us answer this question in future work, and go beyond the scope of this paper. As a compromise, we discuss possible metabolites that may orchestrate this based on our observations based on MML‑1 subcellular localization at LD/mitochondria (including PPP and TCA cycle intermediates).

(3) Line 238, it should be "NADPH". 

Corrected.

(4) RNAi targeting enzymes of different branches of PPP can be performed

In our initial screen, we examined the effect of various enzymes of the PPP on MML-1 nuclear localization (Figure 1A, Supplementary Table S1) and found that knockdown of enzymes in both the oxidative phase (PGDH/T25B9.9) and non-oxidative phase (transketolase/TKT-1) affect MML-1 nuclear localization. In line, 6-AN treatment, which affects the oxidative phase, also stimulated MML‑1 nuclear localization (Figure 3B). We also observed that knockdown of enzymes involved in ribose 5P conversion to ribose, ribose 1P, and phosphoribosyl pyrophosphate, an intermediate in nucleotide biosynthesis, decreased MML-1 nuclear localization (rpia-1, F07A11._5, _Y43F4B.5, _R151._2; Supplementary Table S1). Whether MML‑1/MondoA responds to nucleotide pool remains elusive.

(5) As for PPP, these are many possibilities that can be tested. For example, as PPP supplies NADPH for oxidative balance, does MML-1 respond to ROS? Also, it appears the genes in the non-oxidative arm of PPP regulate MML-1, so is nucleotide synthesis involved? 

Thank you for the suggestion. We tested other enzymes involved in NADPH production from the folate cycle and observed a mild but significant reduction of MML-1 nuclear localization upon dao-3i (Supplementary Table S1). Moreover, we tested whether MML-1 nuclear localization is responsive to ROS. While paraquat exposure induced oxidative stress by measuring the transcriptional reporter gst‑4p::GFP (Supplementary Figure 3A), paraquat exposure did not significantly affect MML-1 nuclear localization (Supplementary Figure 3B). Therefore we think it less likely that NADPH production acting through redox regulation is the main effect.

We also tried supplementation with some of the metabolite outputs of PPP including ribose, ribulose, and xylulose, as well as nucleosides (see below), but saw no effect on MML-1 nuclear localization. We agree that further studies are required to pinpoint whether there is another metabolic moiety regulating MML-1 at the protein-ligand level, but this goes beyond the scope of the current investigation.

Author response image 2.

I think that this is a very important point. When conducting assessments, we are acting as scientists conducting an experiment. Like any experiment, there are many different variables at play. Variables include the number of distractions in the room during instruction and during the assessment, the number of distractions in students individual lives, any learning disabilities the student may have, diagnosed or not, and any number of other factors.How can teachers go about assessing the reliability of an assessment and at what point do instructors throw out results all-together?

I love the idea behind this. Rather than assignments being for a grade, I feel as though the majority of the wprk for a class should count for a completion-type score as in the end, the student should not be punished for lack of understanding. Usually when a student does not understand something, it is not necessarily their fault. Sometimes there are underlying conditions the student is dealing with such as ADHD, depression, or autism, which can make learning hard for some students. Why should students be punished for something that is not their fault? Many may argue that it is the students fault as they simply refused to do the work or refused to put effort into the assignment, but I argue that humans, especially children, are naturally curious about the world. If they do not wish to take an opportunity to learn, then usually there is more to it than them simply not wanting to do it. On the other hand, though, many older students have other priorities than completing assignments they may feel are optional, such as a simple completion grade. Instead of trying their best on it, they may deem it worthy to turn in sub-par work instead of actually trying their best on the assignment. So how can we have an assignment that assesses students understanding of a topic without punishing students who do not understand the material? The only possible solution I can think of would be to have the assignment be for a grade that reflects the correctness of the assignment, but allow the students to revise the work until it meets their expectations. This would allow students to prioritize other work over said assignment, if necessary, but also still provide students with urgency to do their best on the assignment.

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

Learn more at Review Commons

Reply to the reviewers

Response to Reviewer #1

Major comments:

1. * The connection between vB12 and MMA is weak, and the attempt to connect these pieces to PPI seems somehow forced. For instance, the authors do not convincingly demonstrate that MMA causes the PPI deficit. Furthermore, vB12 may rescue PPI independently of MMA. The authors should be more transparent about the lack of connection or causality between changes in metabolism and behavior.* We appreciate the reviewer's comment and acknowledge that we have not demonstrated causal relationships between increased MMA, PPI deficits in Tbx1+/- mice and their rescue by vB12. They are associations.

In the revised manuscript, we have clarified this in the Discussion, para.4, by adding the following phrase. “The results of our study do not prove a causal relationship between elevated brain MMA and PPI impairment, nor do they tell us whether rescue of the PPI impairment by vB12 occurs by reducing MMA”.

Regarding the comment of a weak connection between vitamin B12 and MMA, we respectfully disagree.

The biochemistry underlying the connection is outlined clearly in the Introduction, page 4, para. 2.

Patients with vitamin B12 deficiency typically exhibit elevated levels of MMA and administration of vitamin B12 (cobalamin) helps to normalize MMA values (Robert & Brown, 2003). Furthermore, several animal models with genetic alterations in the vitamin B12 pathway exhibit high levels of MMA. For instance, mice lacking the cobalamin transporter have increased MMA (Bernard et al., 2018). Additionally, mice lacking the mutase (Mut), which requires vitamin B12 as a cofactor for the conversion of methylmalonyl CoA to succinyl-CoA for entry into the Krebs cycle, demonstrate elevated levels of MMA and are unresponsive to vitamin B12 treatment (Peters et al. 2006). In the revised manuscript, we have cited these references (Introduction, page 4, para. 2).

Throughout the manuscript, an important control is missing: WT+ vB12 group. Data from this group should be added to Figures 1, 3, 4, 6, and Supplemental Figures 3 and 4 to show the effect of vB12 on WT mice.

All of the experiments reported in the original manuscript included this control group although it was not always included in the data analysis and therefore in the figures, as observed by the reviewer, In the revised manuscript all of the relevant figures and tables now include these data.

In Figures 1C and 3, data from the respective WT controls in the Df1 and Tbx1 cohorts should be shown.

• *

The wild type (WT) animals serve as the control group for both Tbx1+/- and Df/+ mice because they were littermates, obtained from matings between Df1/+ and Tbx1+/- mice. This has been clarified in the Materials and Methods, and in a new cartoon which has been added to Supplementary Figure 1 (1C) showing all of the animal groups used for the various studies (NMR, transcriptomics, behavior).

The Supplemental Table S1 includes 17 WT controls for Tbx1+/- and 6 WT controls for Df1/+, but Figure 1C includes only one group of 11 WT controls. For which group were those 11 WT controls?

Here are several examples of inconsistency in the data: "For this, we first performed a preliminary metabolome analysis using isolated whole brains of male and female Tbx1+/- (n= 18) and WT (n= 10) mice between one and two months of age. A set of metabolites was quantified in brain extracts by liquid chromatography tandem mass spectrometry (LC-MS/MS) (Supplementary Table 1)." Again, the number of mice the authors note in the manuscript does not match that shown in Supplementary Table 1 (Tbx1+/- (n=14) and WT (N=17). "We analyzed whole brain tissue isolated from Df1/+ and WT (control) mice (n = 5 per genotype)." Again, the numbers of mice do not match those in Supplementary Table 1, which notes 6 Df1/+ mice and 6 WT mice.

We apologize for these errors and inconsistencies in the text and tables, all of which have been corrected in the revised manuscript. In addition, we have added the aforementioned cartoon (Supplementary Figure 1C) and we have improved the presentation of the data (genotypes and treatments) in Supplementary Tables 1 and 2. We hope that these changes provide the expected clarity to the data.

MMA is the only metabolite that is similarly changed between Tbx1+/- and Df1/+ brains. This is an interesting observation. However, there is no other overlap in metabolic changes between these two mutants. This is a concern that requires clarification

We appreciate the reviewer's comment. The observation is not altogether surprising considering that the Df1 deletion includes at least 9 genes involved in metabolic pathways (cited refs (Maynard, 2008;Meechan, 2011; Devaraju and Zakharenko, 2017)) any of which might counteract or compensate for changes caused by Tbx1 mutation alone. In addition, heterozygosity for other genes in the deleted region (Df1 encompasses over 20 genes) might affect metabolic processes indirectly. In the revised manuscript we have added the following phrase to the Discussion, para.1 “Thus, even though the two mutants are genetically and metabolically very different, in Df1/+ mice, the MMA phenotype is not affected by heterozygosity of other genes in the deletion.

The authors mention that MMA is not changed in pre-term Tbx1+/- embryos, but no data are provided. What about MMA levels in Df1/+ embryos?

In pre-term Tbx1+/- and WT embryos MMA was undetectable. This is now stated in the final paragraph of the first section of the Results.

We did not measure MMA in Df1/+ embryos. It was not a goal of the study to compare the metabolome of these two genetically very different mutants. The MMA data on Df1/+ mice are presented because they show the potential relevance of this phenotype for the human disease, and they justify the use of the single gene (Tbx1+/-) mutants for studies into metabolism-related disease mechanisms. See also response to point 12

* * In some cases, the differences in metabolites (e.g., glutamine, glutamate, phosphoethanolamine, taurine, leucine, myo-inositol, and niacinamide) between the WT and Tbx1+/- mice is very minimal (Supplemental Figure 2). The y-axis scale should start at 0.

We have changed the y-axis settings where necessary

The vB12 is administered via two different regimens: 1) every 3 days for 28 days, at 4-8 weeks of age for metabolic measurements, and 2) twice a week for 2 months for PPI behavioral testing. Is there any reason the authors chose different protocols?

We apologize for the confusion, which was due to an oversight in the Materials and Methods section that has been corrected. The weekly injection regimen was the same for mice used in the behavioral and metabolic studies, but the treatment time was shorter for the metabolic studies, for practical reasons beyond our control; mice received vB12 injections twice a week, beginning at 4 wks of age and continuing until 8 wks or 12 wks of age for metabolic and behavioral studies respectively.

* * The authors should add the following references to the study: Long et al., Neurogenetics (2006), which shows no change in PPI in Tbx1+/- mice. This discrepancy compared with the current study results and those of Paylor et al., Proc Natl Acad Sci U S A (2006) should be discussed.

We have not cited the study by Long et al. because there are no obvious reasons for the discrepancy (age, mouse strain, sex) that could be discussed. Beyond this of course we cannot comment on data generated by another research group. Nevertheless, the presence of the PPI deficit in Tbx1+/- mice has been confirmed in two different Tbx1 alleles Tbx1 lacZ/+ and Tbx1ΔE5/+, by two different investigators, Dr. Richard Paylor using Tbx1 lacZ/+ mice (Paylor et al. 2001) and Dr. Elvira De Leonibus (co-author of this manuscript) using Tbx1ΔE5/+ mice, in two different countries (USA and Italy) in a rederived colony of mice.

• Figure 6B is a concern. The PPI decrease in the Tbx1+/- group appears to be driven by results from 3-4 mice. First, are those data statistical outliers? *

With all due respect, this is not the case. Eight Tbx1+/- mice, i.e., >50% of those tested had PPI values below the minimum observed in WT mice. The behavioural data were checked for the presence of outliers in each group using the Grubbs test, which yielded negative results. Our finding of PPI deficits in Tbx1+/- mice are in line with previously published data in Tbx1+/- and other animal models of 22q11.2 microdeletion (Paylor et al., 2006; Paylor and Lindsay, 2006; Stark et al., 2008), as well as in humans (Sobin et al., 2005).

Second, experiments in the same mice would be more informative. Do PBS-treated mutants recover PPI if they are treated with vB12 and vice versa? If the authors are concerned about the age difference, they also should include age-dependent effects on PPI.

We decline to perform the proposed experiment for the reasons described in section 4 of the Revision Plan

• *

Because vB12 treatment completely rescued the MMA level in Df1/+ mice (Figure 3), the authors should include a figure showing PPI test results in Df1/+ mice.

Vitamin B12 treatment fully rescued the MMA phenotype in both mutants (Figure 3). Whether it rescues the PPI defect in Df1/+ mice is not important for this study. We used Df1/+ mice as an entry point, in order to give validity to the pursuit of the MMA phenotype in the single gene mutant (Tbx1+/-), in which we expect that it will be easier to find disease mechanisms. For this reason, we focused our attention on identifying metabolic alterations in adult Tbx1+/- mice.

See also response to point 7.

*Figure 1A and B table: Did the authors mean Log2FC instead of FC? The authors also should present the *

*source data by adding supplemental tables that include raw data and normalized conversion, etc., as described in the multivariate statistical data analysis of the LC-MS/MS data. *

• *

The new Figures 1A, Figure 3 and the accompanying tables now state Log2FC. New Supplementary Table 1 presents the raw data that were normalized on the basis of the amount of protein in the samples, described and referenced in the Materials and Methods

"...we identified a new metabolic phenotype that was associated with reduced sensorimotor gating deficits in Tbx1+/- mice". Although the authors showed the PPI rescue by treating Tbx1+/- mice with vB12, that result alone does not prove the association of metabolic phenotype with sensorimotor-gating deficit; other supporting data are needed.

• *

This is perhaps a question of semantics; by associated we mean that the two phenotypes, metabolic alterations and reduced PPI were observed together

The authors stated, "Results showed that there were very few differentially expressed genes in Tbx1+/- vs WT brains, (n=22 out of 14535 expressed genes (Fig. 5 and Supplementary Tab. 2)". However, they described how 3 differentially expressed genes are involved in mitochondrial activity in the Discussion. The authors should describe those 3 genes and their relation to the metabolic change.

The results that the reviewer refers to have changed in the revised manuscript due to the inclusion of the control group WT +vB12 in the data analysis. The transcriptome analysis revealed that vB12 had a stronger impact than genotype, and as a consequence, the statistical analysis of all groups did not highlight minor differences between the two genotypes.

Figure 5B: The authors claimed that they detected similar transcription profiles between WT+vB12 vs. Tbx1+/-+vB12, comparing Tbx1+/-+PBS vs. Tbx1+/-+vB12. This is based on 947 genes being downregulated and 834 being upregulated, which is not appropriate. The authors should normalize those data to the numbers of genes upregulated and downregulated in WT+PBS vs. WT+vB12 respective groups.

We said that we detected similar transcription profiles in PBS-treated WT and Tbx1+/- brains; a WT+vB12 group was not present. The latter group is included in revised manuscript and the data reanalyzed comparing all groups.

See also response to points 2 and 15.

Minor comments

1. Supplementary Table S1 shows the identical MMA concentration "0.2" for 6 controls. Is this correct? This was an error that has been corrected; the value is 0.00 (not detectable).

* Remove the callout for Figure 1C at the end of the second paragraph in Results*.

This figure is no longer present

*There are multiple typos throughout the manuscript. *

Here are several examples:

1. * Fig1B graph- Df/+ => Df1/+* Figure changed in revised manuscript

2. "Together, the hydrophilic and lipophilic results revealed a group of 6 compounds that characterized the brain metabolic differences between Tbx1+/- and WT mice (Figure 2B, 2C)". Figure 2A should be included also. Corrected

3. "In support of this notion, is the finding that...(remove) Removed

4. Remove double periods: "The pathways found are depicted in Figure 2C' which reports the impact of each pathway versus p values.." Corrected

5. Panel labels in all figures are misplaced. Panel labels are aligned correctly

We have performed a spelling and grammar check on the text

"In support of this notion... at least nine orthologs are involved in mitochondrial metabolism". What are those 9 mitochondrial genes? Kolar et al., Schizophr Bull (2023) indicates that there are 8 mitochondrial genes within the 22q11.2 locus. The authors need to list these genes.

This reference, which is a review, has been cited in the Introduction, para.3 along with the genes.

The review presents nine mitochondrial genes which the authors divide into two groups, 1) Genes expressed in mitochondria (SLC25A1, TXNRD2, MRPL40, PRODH, and COMT) and 2) Genes that have been shown to have an impact on mitochondrial function (TANGO2, ZDHHC8, UFD1L, and DGCR8). In the abstract they mention only eight genes, the ninth gene COMT is mentioned in the text.

Reviewer #2 (Significance (Required)):

*The manuscript titled "Tbx1 haploinsufficiency causes brain metabolic and behavioral anomalies in adult mice which are corrected by vitamin B12 treatment" by Caterini et al. presents a comparative metabolomics study in the brains of mouse models carrying a heterozygous mutation in the transcription factor Tbx1. This mutation is contrasted with a chromosomal deficiency encompassing Tbx1, among other gene loci, known as Df1/+, which serves as a mouse model for 22q11 microdeletion syndrome. The primary and most significant finding of the study is that Tbx1 heterozygosity alone induces broad metabolomic alterations in the entire brain parenchyma, despite Tbx1 expression being confined to vascular endothelial cells. The authors leverage this observation to investigate the effects of dietary supplementation with vitamin B12, which alters the metabolome in a manner interpreted by the authors as rescuing or reversing the Tbx1 heterozygosity phenotype. This study holds promise for understanding the individual gene contributions to the penetrant behavioral phenotypes observed in Df1/+ and 22q11 affected subjects. This potential arises from the clear and consequential metabolic phenotypes described, notably the accumulation of methylmalonic acid.

However, despite the intriguing metabolic phenotypes, there are significant issues hindering incontrovertible conclusions. *

Response to Reviewer #2

Major comments

1. Despite the intriguing metabolic phenotypes, there are significant issues hindering incontrovertible conclusions. Chief among these problems is the experimental design's nature, where the effects of genotype and a pharmacological intervention, vitamin B12, are assessed. The current design overlooks the effects of vitamin B12 on wild-type animals in metabolic and behavioral measures, thus precluding the attribution of the effects of vitamin B12 to a rescue. See response to Reviewer 1 (point 2) who made the same criticism. This group is now included in the data analysis of the relevant experiments.

*An alternative explanation, consistent with the measurements, is that vitamin B12 modifies metabolites and transcripts irrespective of genotype. A suggestion of this possibility is the observed effect of B12 lowering glutamate levels in Tbx1 mutant tissue below those in wild-type brain tissue (Fig. 4C). *

This might be true for some metabolites. Indeed, we found 5 metabolites that responded similarly to vB12 in both WT and Tbx1 +/- mice. In contrast, three metabolites responded to vB12 in both WT and Tbx1+/- mice, but the response was more pronounced in Tbx1+/- mice. Finally, a group of eight metabolites was altered exclusively in Tbx1+/- mice after vB12 treatment, including inosine, glutamate and short-chain fatty acids (SCFAs), Figure 4 and Supplementary Figure 6. Thus, overall, our data suggest that with only a few exceptions, the metabolic response to vB12 treatment is genotype-dependent.

• *

This experimental design issue is exacerbated by the multitude of analytes measured by metabolomics, all collectively assumed to change as part of a common genotype-B12 interaction mechanism. This interpretation is feasible only if none of the analytes were to respond to B12 in wild-type animals.

• *

As specified above, the response to vB12 was genotype-dependent. The inclusion of the vB12-treated WT dataset should address this point.

A second major issue arises from the assertion that Tbx1 is exclusively expressed in mouse brain endothelial cells and not in brain parenchyma. A significant unresolved question is how a gene expressed solely in endothelial cells can alter the brain parenchyma metabolome and transcriptome. This issue remains unaddressed and is not sufficiently discussed. If this assertion holds true, then the observations bear great importance in understanding how Df1/+ causes brain phenotypes and, by extension, in human 22q11.

There are quite a lot of published data from the mouse demonstrating the brain endothelial-specific expression of Tbx1 and the lack of expression in other brain cell types. These include studies using reporter genes (Paylor, 2006; Cioffi, 2014), Tbx1Cre based cell fate mapping (Cioffi, 2014, Cioffi, 2022) and single cell whole genome transcriptions (Ximerakis et al., 2019); (https://portals.broadinstitute.org/single_cell/study/aging-mouse-brain). All are cited in the manuscript.

HOW the mutation of Tbx1 alters the brain metabolome and transcriptome will be the object of future studies, Currently, we do not have any data. At the reviewer’s request, we have extended the discussion of this point in the revised manuscript (Discussion, para. 4).

In this vein, the authors should consider that Tbx1 is not expressed in brain endothelial cells in humans and is minimally expressed in fetal astrocytes (see https://brainrnaseq.org/).

https://brainrnaseq.org provides a tool to evaluate the gene expression in the fetal brain. The sequencing was performed on fetal human brain tissue after elective pregnancy termination (4wks-9wks, it is not clear). Our analysis focuses on adult mice, which may contribute to observed differences.

Moreover, Yi et al. (2010) generated a gene expression atlas of human embryogenesis spanning from 4 to 9 weeks of gestational age, revealing downregulation of TBX1 during this timeframe. Conversely, in the normal adult human brain, TBX1 expression is identified in endothelial cells, as indicated in "The Human Brain Cell Atlas v1.0" presented for visualization and data mining through the Chan Zuckerberg Initiative’s CellxGene application, referring to the atlas ontology in Ding et al. (2016). * 3. A third major concern pertains to the general poor quality of the figures. Many figures appear to be directly exported from the software used for data analysis without proper curation. They are inadequately labeled, lack color codes to clarify differences (e.g., volcano plots), feature lettering fonts that are difficult to discern, and have lettering panels placed in awkward positions. Fig. 1 would benefit by the addition of a pathway diagram showing which metabolites are changing. Figure tables/spreadsheets either have sheets labeled in Italian or are empty. Collectively, the manuscript needs more careful data curation and presentation*.

Many of the figures and tables have been modified with respect to the original manuscript and issues of clarity and quality have been improved where necessary.

Other points for consideration are listed below. • The abstract results section does not mention the Df1 mutants at all, and overall the description of the results should be improved

Corrected • The abstract would benefit from defining vB12 before using the abbreviation

Corrected • The section of the Results describing MMA accumulation in the brain would benefit from

• *explaining the choice of 1 month of age for terminal experiments and the choice to use whole brains (are there particularly brain regions suspected to be affected?), * The majority of animals were 2 months of age at sacrifice (age and sex of individual animals are indicated in Supplementary Table 1). Young adult mice were the object of the study for the reason described in the first paragraph of the Results section, namely “Human studies of brain metabolism have mainly been conducted on children and adolescent patients. Therefore, in order to determine whether similar anomalies were present in the mouse models, we performed our studies on young mice between 1 and 2 months of age (Dutta et al., 2016)”.

This is also the age at which the behavioural phenotype has been demonstrated (Paylor et al., 2006), and therefore could, potentially be rescued by vB12 treatment. We do not have regional information pertaining to the adult brain.

2) describing any sex effects for Tbx1 mutants (and clarifying what data points for Tbx1 animals correspond to which sex), and 3) including what sex was used for Df1 experiments.

In preliminary experiment we analyzed males and females’ mice, before electing to use only males. To obtained reliable information about the impact of gender on metabolism and transcription we would have to use much larger numbers of animals. In Supplementary Table 1 pertaining to males and females are now indicated.

• The authors demonstrate that vB12 rescues PPI but use no other behavioral paradigms. It is possible that these mutations and/or vB12 could be impacting anxiety-like behaviors or other behavioral phenotypes. By only including PPI, the authors limit the interpretation of the "rescue" of this phenotype by vB12. * Reduced PPI was the only behavioral anomaly identified in Tbx1+/- mice that were subjected to a standard battery of behavioral tests (Paylor et al., 2006). *

* *Reduced PPI was the only behavioral anomaly identified in Tbx1+/- mice that were subjected to a standard battery of behavioral tests (Paylor et al., 2006).

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

This is a terrific paper looking at influences of Tbx1 heterozygosity on metabolic phenotypes in mice. A weakness is that the locus on effects of B12 is totally unclear--could be neurovascular or even peripheral, but correcting this weakness might include study of Tbx1 conditional mutants, beyond the scope of this study.

• *

Reviewer #3 (Significance (Required)):* good significance

Only two minor suggestions. 'We selected to study primarily Tbx1 single gene mutants because it is the primary candidate disease gene". What is the basis for this statement? Mouse +/- mutants in Mrpl40, Txnrd2, ProhD, and probably others have shown brain phenotypes.*

• *

The basis for TBX1 being considered as the primary candidate disease gene is the finding of TBX1 point mutations in patients who have the full spectrum of clinical phenotypes associated with 22q11.2 deletion syndrome without the chromosomal deletion, namely, congenital heart defects, immune defects, facial dysmorphism, learning defects and developmental delay. Similarly, in the mouse, Tbx1 mutation recapitulates the phenotype observed in multigene deletion mutants, such as Df1/+ mice.

We do not say (or think) that heterozygosity of other genes from del22q11.2 does not contribute to the disease, but mutations of other genes have not been found in individuals with a 22q11.2DS phenotype but without the chromosomal deletion.

In the discussion, the authors could close the loop on low glutamine could result in lower gaba in inhibitory interneurons, and its correction with B12 could restore gaba levels.

Discussion, para. 3. We thank the reviewer for comment. However, the GABA concentration is not altered in Tbx1 haploinsufficient brains; it is only upregulated by Vitamin B12. Therefore, this assumption may be very speculative. Due to differences in the release and reabsorption rates of the three compounds (glutamine, glutamate, and GABA), correctly evaluating the glutamine-glutamate cycle requires separating astrocytes from neurons. We have only discussed the upregulation of glutamate and the GABA response to Vitamin B12, which may counteract the excess of glutamate.

1. __4__Description of analyses that authors prefer not to carry out Please include a point-by-point response explaining why some of the requested data or additional analyses might not be necessary or cannot be provided within the scope of a revision. This can be due to time or resource limitations or in case of disagreement about the necessity of such additional data given the scope of the study. Please leave empty if not applicable.

Reviewer 1, point 11. The PPI decrease in the Tbx1+/- group appears to be driven by results from 3-4 mice. First, are those data statistical outliers?

Second, experiments in the same mice would be more informative. Do PBS-treated mutants recover PPI if they are treated with vB12 and vice versa? If the authors are concerned about the age difference, they also should include age-dependent effects on PPI.

We are unable to perform this experiment because, as stated in the manuscript, the mice were sacrificed at the end of the experiment and the brains preserved for histological analysis (not part of this study). The generation of mice for new experiments would take about one year. With all due respect, we do not believe that the data that would be obtained are sufficiently important to justify, ethically and economically, this work.

Author response:

We want to thank the reviewers for their constructive feedback.

General

The recall values of our method range between 78.6% for all urine cases to 83.3% for feces (and not between 70-80%, as stated by reviewer #2), with a mean precision of 85.6%. This is rather similar to other machine learning-based methods commonly used for the analysis of complicated behavioral readouts. For example, in the paper presenting DeepSqueak for analysis of mouse ultrasonic vocalizations (Coffey et al. DeepSqueak: a deep learning-based system for detection and analysis of ultrasonic vocalizations. Neuropsychopharmacol. 44, 859–868 (2019). https://doi.org/10.1038/s41386-018-0303-6), the recall values reported for both DeepSqueak, Mupet and Ultravox (Fig. 2c, f) are very similar to our method.

We have analyzed and reported all the types of errors made by our methods, which are mostly technical. For example, depositions that overlap the mouse blob for too long till getting cold will be associated with the mouse and therefore will not be detected (“miss” events). These technical errors are not supposed to create a bias for a specific biological condition and, hence, shouldn’t interfere with the use of our method. A video showing all of the mistakes made by our algorithm on the test set was submitted (Figure 2-video 1).

Below we will to relate to specific points and describe our plan to revise the manuscript accordingly.

Detection accuracy

a. It should be noted that when large urine spots are considered, our algorithm got 100% correct classification (Figure 2, supplement 1, panel b). However, small urine deposits are very similar to feces in their appearance in the thermal picture. In fact,  if the feces are not shifted, discrimination can be quite challenging even for human annotators. To demonstrate the accuracy of the proposed method relative to human annotators, we plan to compare its results with the accuracy of a second human annotator.

b. As part of the revision, we plan to test general machine learning-based object detectors such as faster-RCNN or YOLO (as suggested by Reviewer 2) and compare them with our method.

c. To check if our method may introduce bias to the results, we plan to check if the errors are distributed evenly across time, space, and genders.

Design choices

(A) The preliminary detection algorithm has several significant parameters. These are:

a. Minimal temperature rise for detection: 1.1°C rise during 5 sec.

b. Size limits of the detection: 2 - 900 pixels.

c. Minimal cooldown during 40 sec: 1.1°C and at least half the rise.

d. Minimal time between detections in the same location: 30 sec.

We chose to use low thresholds for the preliminary detection to allow detection of very small urinations and to minimize the number of “miss” events, relying on the classifier to robustly reject false alarms. Indeed, we achieved a low rate of miss events: 5 miss events for the entire test set (1 miss event per ~90 minutes of video). We attribute these 5 “miss” events to partial occlusion of the detection by the mouse.

To adjust the preliminary detection parameters to a new environment, one will need to calibrate these parameters in their own setup. Mainly, the size of the detection depends on the resolution of the video, and the cooldown rate might be affected by the material of the floor, as well as the room temperature.

We plan to explore the robustness of these parameters in our setup and report the influence on the accuracy of the preliminary algorithm.

(B) We chose to feed the classifier with 71 seconds of videos (11 seconds before the event and 60 seconds after it) as we wanted the classifier to be able to capture the moment of the deposition, the cooldown process, as well as urine smearing or feces shifting which might give an additional clue for the classification. In the revised paper we plan to report accuracy when using a shorter video for classification.

Generability

a. In the revised version, we plan to report the accuracy of the method used on a different strain of mice (C57), with a different arena color (white arena instead of black).

Statistics

a. In the revised paper, we will explain why we chose each time window for analysis. Also, we will report statistics for different time windows, as suggested by Reviewer 3.

b. Unlike reviewer #2, we don’t think that the small difference in recall rate between urine and feces (78.6% vs. 83.3%, respectively) creates a bias between them. Moreover, we don’t compare the urine rate to the feces rate.

c. In the revised manuscript we will explicitly report the precision scores, although they also appear in our manuscript in Fig. 2- Supplement 1b.

I find this part of this section. Will better help me understand about big data Business analyst Doug Laney suggested that three characteristics make big data different from what came before: volume, velocity, and variety. Volume refers to the sheer amount of data. Velocity focuses on how quickly data arrives as well as how quickly those data become “stale.” Finally, variety reflects the fact that there may be many different kinds of data. Together, these three characteristics are often referred to as the “three Vs” model of big data. Note, however, that even before the dawn of the computer age we’ve had a variety of data, some of which arrives quite quickly, and that can add up to quite a lot of total storage over time.

Author response:

The following is the authors’ response to the previous reviews.

Reviewer 1:

• Although ROC AUC is a widely used metric. Other metrics such as precision, recall, sensitivity, and specificity are not reported in this work. The last two metrics would help readers understand the model’s potential implications in the context of clinical research.

In response to this comment and related ones by Reviewer 2, we have overhauled how we evaluate our models. In the revised version, we have removed Micro ROC-AUC, as this evaluation metric is hard to interpret in the recommender system setting. Instead, the updated version fully focuses on two metrics: ROC-AUC and Precision at 1 of the negative class, both computed per spectrum and then averaged (equivalent to the instance-wise metrics in the previous version of the manuscript). We believe these metrics best reflect the use-case of AMR recommenders. In addition, we have kept (drug-)macro ROC-AUC as a complementary evaluation metric. As the ROC-AUC can be decomposed into sensitivity and specificity (at different prediction probability thresholds), we have added a ROC curve where sensitivity and specificity are indicated in Figure 8 (Appendices).

• The authors did not hypothesize or describe in any way what an acceptable performance of their recommender system should be in order to be adopted by clinicians.

In Section 4.3, we have extended our experiments to include a baseline that represents a “simulated expert”. In short, given a species, an expert can already make some best guesses as to what drugs will be effective or not. To simulate this, we count resistance frequencies per species and per drug in the training set, and use this as predictions of a “simulated expert”.

We now mention in our manuscript that any performance above this level results in a real-world information gain for clinical diagnostic labs.

• Related to the previous comment, this work would strongly benefit from the inclusion of 1-2 real-life applications of their method that could showcase the benefits of their strategy for designing antibiotic treatment in a clinical setting.

While we think this would be valuable to try out, we are an in silico research lab, and the study we propose is an initial proof-of-concept focusing on the methodology. Because of this, we feel a real-life application of the model is out-of-scope for the present study.

• The authors do not offer information about the model features associated with resistance. This information may offer insights about mechanisms of antimicrobial resistance and how conserved they are across species.

In general, MALDI-TOF mass spectra are somewhat hard to interpret. Because of a limited body of work analyzing resistance mechanisms with MALDI-TOF MS, it is hard to link peaks back to specific pathways. For this reason, we have chosen to forego such an analysis. After all, as far as we know, typical MALDI-TOF MS manufacturers’ software for bacterial identification also does not provide interpretability results or insights into peaks, but merely gives an identification and confidence score.

However, we do feel that the whole topic revolving around “the degree of biological insight a data modality might give versus actual performance and usability” merits further discussion. We have ultimately decided not to include a segment in our discussion section as it is hard to discuss this matter concisely.

• Comparison of AUC values across models lacks information regarding statistical significance. Without this information it is hard for a reader to figure out which differences are marginal and which ones are meaningful (for example, it is unclear if a difference in average AUC of 0.02 is significant). This applied to Figure 2, Figure 3, and Table 2 (and the associated supplementary figures).

To make trends a bit more clear and easier to discern, in our revised manuscript, all models are run for 5 replicates (as opposed to 3 in the previous version).

There is an ongoing debate in the ML community whether statistical tests are useful for comparing machine learning models. A simple argument against them is that model runs are typically not independent from each other, as they are all trained on the same data. The assumptions of traditional statistical tests are therefore violated (t-test, Wilcoxon test, etc.). With such tests statistical significance of the smallest differences can simply be achieved by increasing the number of replicates (i.e. training the same models more times).

More complicated but more appropriate statistical tests also exist, such as the 5x2 cross-validated t-test of Dietterich: “Approximate statistical tests for comparing supervised classification learning algorithms”, Neural computation 1998. However, these tests are typically not considered in deep learning, because only 10% of the data can be used for training, which is practically not desirable. The Friedman test of Demšar "On the appropriateness of statistical tests in machine learning." Workshop on Evaluation Methods for Machine Learning in conjunction with ICML. 2008., in combination with posthoc pairwise tests, is still frequently used in machine learning, but that test is only applicable in studies where many datasets are tested.

For those reasons, most deep learning papers that only analyse a few datasets typically do not consider any statistical tests. For the same reasons, we are also not convinced of the added value of statistical tests in our study.

• One key claim of this work was that their single recommender system outperformed specialist (single species-antibiotic) models. However, in its current status, it is not possible to determine that in fact that is the case (see comment above). Moreover, comparisons to species-level models (that combine all data and antibiotic susceptibility profiles for a given species) would help to illustrate the putative advantages of the dual branch neural network model over species-based models. This analysis will also inform the species (and perhaps datasets) for which specialist models would be useful to consider.

We thank the reviewer for this excellent suggestion. In our new manuscript, we have dedicated an entire section of experiments to testing such species-specific recommender models (Section 4.2). We find that species-specific recommender systems generally outperform the models trained globally across all species. As a result, our manuscript has been majorly reworked.

• Taking into account that the clustering of spectra embeddings seemed to be species-driven (Figure 4), one may hypothesize that there is limited transfer of information between species, and therefore the neural network model may be working as an ensemble of species models. Thus, this work would deeply benefit from a comparison between the authors' general model and an ensemble model in which the species is first identified and then the relevant species recommender is applied. If authors had identified cases to illustrate how data from one species positively influence the results for another species, they should include some of those examples.

See the answer to the remark above.

• The authors should check that all abbreviations are properly introduced in the text so readers understand exactly what they mean. For example, the Prec@1 metric is a little confusing.

See the answer to a remark above for how we have overhauled our evaluation metrics in the revised version. In addition, in the revised version, we have bundled our explanations on evaluation metrics together in Section 3.2. We feel that having these explanations in a separate section will improve overall comprehensibility of the manuscript.

• The authors should include information about statistical significance in figures and tables that compare performance across models.

See answer above.

• An extra panel showing species labels would help readers understand Figure 11.

We have tried to play around with including species labels in these plots, but could not make it work without overcrowding the figure. Instead, we have added a reminder in the caption that readers should refer back to an earlier figure for species labels.

• The authors initially stated that molecular structure information is not informative. However, in a second analysis, the authors stated that molecular structures are useful for less common drugs. Please explain in more detail with specific examples what you mean.

In the previous version of our manuscript, we found that one-hot embedding-based models were superior to structure-based drug embedders for general performance. The latter however, delivered better transfer learning performance.

In our new experiments however, we perform early stopping on “spectrum-macro” ROC-AUC (as opposed to micro ROC-AUC in the previous version). As a consequence, our results are different. In the new version of our manuscript, Morgan Fingerprints-based drug embedders generally outperform others both “in general” and for transfer learning. Hence, our previously conflicting statements are not applicable to our new results.

• The authors may want to consider adding a few sentences that summarize the 'Related work' section into the introduction, and converting the 'Related work' section into an appendix.

While we acknowledge that such a section is uncommon in biology, in machine learning research, a “related work” section is very common. As this research lies on the intersection of the two, we have decided to keep the section as such.

Reviewer 2:

• Are the specialist models re-trained on the whole set of spectra? It was shown by Weis et al. that pooling spectra from different species hinders performance. It would then be better to compare directly to the models developed by Weis et al, using their splitting logic since it could be that the decay in performance from specialists comes from the pooling. See the section "Species-stratified learning yields superior predictions" in https://doi.org/10.1038/s41591-021-01619-9.

We train our “specialist” (or now-called “species-drug classifiers”) just as described in Weis et al.: All labels for a drug are taken, and then subsetted for a single species. We have clarified this a bit better in our new manuscript. The text now reads:

“Previous studies have studied AMR prediction in specific species-drug combinations. For this reason, it is useful to compare how the dual-branch setup weighs up against training separate models for separate species and drugs. In Weis et al. (2020b), for example, binary AMR classifiers are trained for the following three combinations: (1) E. coli with Ceftriaxone, (2) K. pneumoniae with Ceftriaxone, and (3) S. aureus with Oxacillin. Here, such "species-drug-specific classifiers" are trained for the 200 most-common combinations of species and drugs in the training dataset.

• Going back to Weis et al. a high variance in performance between species/drug pairs was observed. The metrics in Table 2 do not offer any measurement of variance or statistical testing. Indeed, some values are quite close e.g. Macro AUROC of Specialist MLP-XL vs One-hot M.

See our answer to a remark of Reviewer 1 for our viewpoint on statistical significance testing in machine learning.

• Since this is a recommendation task, why were no recommendation system metrics used, e.g. mAP@K, mRR, and so (apart from precision@1 for the negative class)? Additionally, since there is a high label imbalance in this task (~80% negatives) a simple model would achieve a very high precision@1.

See the answer to a remark above for how we have overhauled our evaluation metrics in the revised version. In addition, in choosing our metrics, we wanted metrics that are both (1) appropriate (i.e. recommender system metrics), but also (2) easy to interpret for clinicians. For this reason, we have not included metrics such as mAP@K or mRR. We feel that “spectrum-macro” ROC-AUC and precision@1 cover a sufficiently broad evaluation set of metrics but are easy enough to interpret.

• A highly similar approach was recently published (https://doi.org/10.1093/bioinformatics/btad717). Since it is quite close to the publication date of this paper, it could be discussed as concurrent work.

We thank the reviewer for bringing our attention to this study. We have added a paragraph in our revised version discussing this paper as concurrent work.

• It is difficult to observe a general trend from Figure 2. A statistical test would be advised here.

See our answer to a remark of Reviewer 1 for our viewpoint on statistical significance testing in machine learning.

• Figure 5. UMAPs generally don't lead to robust quantitative conclusions. However, the analysis of the embedding space is indeed interesting. Here I would recommend some quantitative measures directly using embedding distances to accompany the UMAP visualizations. E.g. clustering coefficients, distribution of pairwise distances, etc.

In accordance with this recommendation, we have computed many statistics on the MALDI-TOF spectra embedding spaces. However, we could not come up with any statistic that illuminated us more than the visualization itself. For this reason, we have kept this section as is, and let the figure speak for itself.

• Weis et al. also perform a transfer learning analysis. How does the transfer learning capacity of the proposed models differ from those in Weis et al?

Weis et al. perform experiments towards “transferability”, not actual transfer learning. In essence, they use a model trained on data from one diagnostic lab towards prediction on data from another. However, they do not conduct experiments to learn how much data such a pre-trained classifier needs to fine-tune it for adequate performance on the new diagnostic lab, as we do. The end of Section 4.4 discusses how our proposed models specifically shine in transfer learning. The paragraph reads:

“Lowering the amount of data required is paramount to expedite the uptake of AMR models in clinical diagnostics. The transfer learning qualities of dual-branch models may be ascribed to multiple properties. First of all, since different hospitals use much of the same drugs, transferred drug embedders allow for expressively representing drugs out of the box. Secondly, owing to multi-task learning, even with a limited number of spectra, a considerable fine-tuning dataset may be obtained, as all available data is "thrown on one pile".”

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1:

(1) One issue that needs to be considered is the nomenclature of the enhancer. The authors have presented data to show this enhancer controls the expression of Ctnnb1 in the stomach, intestine, and colon tissues. However, the name proposed by the authors, ieCtnnb1 (intestinal enhancer of Ctnnb1), doesn't represent its functions. It might be more appropriate to call it a different name, such as gieCtnnb1 (gastrointestinal enhancer of Ctnnb1).

We thank the reviewer for the insightful suggestion and agree that wholemount reporter assays indicated ieCtnnb1 and ieCTNNB1 indeed display activity in the stomach. However, in current study, we focused on the cellular distribution and the function in intestinal epithelia. After careful consideration, we reasoned that the current designation, ieCtnnb1, would be more appropriately represent its expression pattern and functions based on provided evidence. We hope the reviewer could understand our reasoning.  

(2) The writing of this manuscript can be improved in a few places. 

a) The definitions or full names for the abbreviations of some terms, e.g., Ctnnb1, ieCtnnb1, in both abstract and main text, are needed when they first appear. Specifically, Line 108 should be moved to Lines 26 and 95. Lines 125126 are redundant. ieCtnnb1 in Line 130 needs to be defined.

We appreciate the suggestion. In the revision, we have included the definition of Ctnnb1 and the full name of ieCtnnb1 when they first appear in the abstract and the main text. Lines 125-126 were deleted in the revision.

b) Line 192-194, the description of the result needs to be rewritten to reflect

the higher expression of LacZ transcript in eGFP+ cells. 

We would like to emphasize that the key point of this part is that the enhancer activity of ieCtnnb1 is present in both Lgr5-eGFP+ and Lgr5-eGFP- cells. This was validated by single-cell sequencing, which revealed the presence of LacZ transcripts in the Paneth cells. Moreover, we could not confidently conclude that eGFP+ cells have higher expression levels of LacZ, as these measurements were obtained from separate, semi-quantitative RTqPCR experiments.

c)  More details are needed for how the data using human tumor samples were generated and how they were analyzed. 

We thank the suggestion. In the revision, we have provided additional details regarding the data and subsequent analyses of human CRC samples as follows: “We previously conducted paired analyses of chromatin immunoprecipitation sequencing (ChIP-seq) for H3K27ac and H3K4me3, alongside RNA-seq on 68 CRC samples and their adjacent normal (native) tissue (Li et al., 2021).  In the current study, we performed analyses for the enrichment of H3K27ac and H3K4me3 at ieCTNNB1 and CTNNB1 promoter regions, as well as the expression levels of CTNNB1, followed by combined analyses (Figure. 5A, Figure 5 - figure supplement 1).”

d) The genomic structures from multiple species are presented at the bottom of Figure 1a. However, the description and explanation are lacking in both the main text and the figure legend.

We apologize for not presenting clearly. We have added related description in the legend of Figure 1A as “The sequence conservation of the indicated species is shown at the bottom as vertical lines”. We also added an explanation in lines 162-163 of the main text: “Notably, unlike neCtnnb1, the primary sequence of ieCtnnb1 is not conserved among vertebrates (Figure 1A, bottom)”.

Reviewer #2:

(1) One of the main issues emerging during reading concerns the interpretation of the consequence of deleting the ieCtnnb1 enhancer. The authors write on line 235 that the deletion of ieCtnnb1 "undermined" Wnt signaling in the intestinal epithelium. This feels too strong, as the status of the pathway is only mildly affected, testified by the observation that mice with homozygous deletion on ieCtnnb1 are alive and well. The enhancer likely "only" drives higher Ctnnb1 expression, and it does not affect Wnt signaling by other mechanisms. The reduction of Wnt target gene expression upon its deletion is easily interpreted as the consequence of reduced β-catenin. Also the title, in my opinion, allows this ambiguity to stick in readers' minds. In other words, the authors present no evidence that the ieCtnnb1 enhancer controls Wnt signaling dosage via any mechanism other than its upregulation of Ctnnb1 expression in the intestinal epithelium. Reduced Ctnnb1, in turn, could explain the observed reduction of Wnt signaling output and the interesting downstream physiological consequences. Unless the authors think otherwise, I suggest they clarify this throughout the text, including necessary modifications to the title.

We greatly appreciate the reviewer’s important comments and suggestion. We agree that ieCtnnb1’s direct effect on the canonical Wnt signaling is to regulate the transcription of Ctnnb1 in the intestinal epithelia. Therefore, knockout of ieCtnnb1 leads to compromised expression of Ctnnb1 and, consequently, reduced Wnt signaling.  The term “undermined” is indeed too strong and has been revised to “compromised” in the revision (line 237). Similar revisions have been made throughout the manuscript. Particularly, the title was changed into “A Ctnnb1 enhancer transcriptionally regulates Wnt signaling dosage to balance homeostasis and tumorigenesis of intestinal epithelia”. However, as we state in the following point, decreased levels of β-catenin on ieCtnnb1 loss could lead to indirect effect, including the reduced expression of Bambi, which might cause a more significant decrease of nuclear β-catenin.

(2) It is unclear how the reduction of Ctnnb1 mRNA caused by deletion of ieCtnnb1 in mice could lead to a preferential decrease of nuclear more than membranous β-catenin (Fig. 1K and L). This might reflect a general cell autonomous reduction in Wnt signaling activation; yet, it is not clear how this could occur. Do the authors have any explanations for this?

It's a very important question. We observed that in inCtnnb1 knockout epithelia, the expression of Bambi (BMP and activin membrane-bound inhibitor) was significantly downregulated. Since BAMBI has been reported to stabilize β-catenin and facilitate its nuclear translocation, it is likely that the reduced level of BAMBI resulting from the loss of ieCtnnb1 further decreased nuclear βcatenin. In the revision, the expression change of Bambi has been added in Figure 1M. Moreover, the related content was extensively discussed with proper citations: “We noticed that after knocking out ieCtnnb1, the level of βcatenin in the nuclei of small intestinal crypt cells of Ctnnb1Δi.enh mice decreased more significantly compared to that in the cytoplasm (49.5% vs. 29.8%). Although the loss of ieCtnnb1 should not directly lead to reduced nuclear translocation of β-catenin, RNA-seq results showed that the loss of ieCtnnb1 causes a reduction in the expression of Bambi (BMP and activin membranebound inhibitor), a target gene in the canonical Wnt signaling pathway (Figure 1M). BAMBI promotes the binding of Frizzled to Dishevelled, thereby stabilizing β-catenin and facilitating its nuclear translocation (Lin et al., 2008; Liu et al., 2014; Mai et al., 2014; Zhang et al., 2015). Thus, it is likely that the decreased level of BAMBI resulting from the loss of ieCtnnb1 further reduced nuclear βcatenin”. 

(3) In Figure 1 K-L the authors show β-catenin protein level. Why not show its mRNA?

The mRNA levels of Ctnnb1 in small and large intestinal crypts were shown in Figure 1I and 1J, demonstrating reduced expression of Ctnnb1 upon ieCtnnb1 knockout. We hope the reviewer understands that it is unnecessary to measure the nuclear and cytosolic levels of Ctnnb1 transcripts, as the total mRNA level generally reflects the protein level. 

(4) Concerning the GSEA of Figure 1 that includes the Wnt pathway components: a) it would be interesting to see which components and to what extent is their expression affected; b) why should the expression of Wnt components that are not Wnt target genes be affected in the first place? It is odd to see this described uncritically and used to support the idea of downregulated Wnt signaling.

We appreciate the suggestion and apologize for any lack of clarity. The affected components of the Wnt signaling pathway and the extent of their changes are summarized in Figure 1 – figure supplement 3. Additionally, we have provided explanations for their downregulation. For instance, the reduced expression of Wnt3 and Wnt2b ligands in ieCtnnb1-KO crypts may be attributed to the decreased numbers of Paneth cells.  

(5) In lines 251-252 the authors refer to "certain technical issues" in the isolation of cell type from the intestinal epithelium. Why this part should be obscure in the characterization of a tissue for which there are several established protocols of isolation and analysis is not clear. I would rather describe what these issues have been and how they protocol of isolation and analysis is not clear. I would rather describe what these issues have been and how they might have affected the data presented.

We thank the reviewer for pointing this out. The single-cell preparation and sequencing of small intestinal cryptal epithelial cells were carried out largely according to reported protocols with slight modification. The enrichment of live crypt epithelial cells (EpCAM+DAPI-) by flow cytometry and cell filtering after single-cell sequencing were appropriate (Figure 2 – figure supplement 1A1C). We would like to emphasize a few points: 1) Unlike other protocols, we did not exclude immune cells, erythrocytes, or endothelial cells using negative sorting antibodies. 2) When defining cell populations, we focused exclusively on epithelial cell types and did not consider other cell types, such as immune cells. As a result, the so-called “undefined” cells include a mixture of nonepithelial cells. Indeed, markers for erythrocytes (AY036118/Erf1, PMID:12894589) and immune cells (Gm42418 and Lars2, PMID:30940803, PMID: 35659337) were the top three enriched genes in the “undefined” cluster (Figure 2 – figure supplement 1D). 3) Nonetheless, the overall findings remain robust, as key observations such as the loss of Paneth cells and reduced cell proliferation were validated through histological studies. This information has been incorporated into the revised manuscript with related references cited (lines 254-259). 

(6) It is interesting that human SNPs exist that seem to fall within the ieCTNNB1 enhancer and affect the gastrointestinal expression of CTNNB1. Could the author report or investigate whether this SNP is present in human populations that have been considered in large-scale studies for colorectal cancer susceptibility? It seems to me a rather obvious next step of extreme importance to be ignored.

(7) From Figure 5A a reader could conclude that colorectal tumor cells have a higher expression of CTNNB1 mRNA than in normal epithelium. This is the first time I have seen this observation which somewhat undermines our general understanding of Wnt-induced carcinogenesis exclusively initiated by APC mutations whereby it is β-catenin's protein level, not expression of its mRNA, of crucial importance. I find this to be potentially the most interesting observation of the current study, which could be linked to the activity of the enhancer discovered, and I suggest the authors elaborate more on this and perhaps consider it for future experimental follow-ups.

We appreciate the comments and suggestions.  We therefore added related content in the revision (lines 470-475): “Importantly, ieCTNNB1 displayed higher enhancer activity in most CRC samples collected in the study. Moreover, the SNP rs15981379 (C>T) within ieCTNNB1 is associated with the expression of CTNNB1 in the GI tract. Future population studies could investigate how the enhancer activity of ieCTNNB1 and this particular SNP are associated with CRC susceptibility and prognosis”.

(8) I am surprised that the authors, who seem to have dedicated lots of resources to this study, are satisfied by analyzing their ChIP experiments with qPCR rather than sequencing (Figure 6). ChIP-seq would produce a more reliable profile of the HNF4a and CREB1 binding sites on these loci and in other control regions, lending credibility to the whole experiment and binding site identification. Sequencing would also take care of the two following conceptual problems in primer design. 

First: while the strategy to divide enhancer and promoter in 6 regions to improve the resolution of their finding is commendable, I wonder how the difference in signal reflects primers' efficiency rather than HNF4/CREB1 exact positioning. The possibility of distinguishing between regions 2 and 3, for example, in a ChIP-qPCR experiment, also depends on the average DNA fragment length after sonication, a parameter that is not specified here. 

Second: what are the primers designed to detect the ieCtnnb1 enhancer amplifying in the yellow-columns samples of Figure 6G? In this sample, the enhancer is deleted, and no amplification should be possible, yet it seems that a value is obtained and set to 1 as a reference value.

This is indeed a crucial point, and we fully agree with the reviewer that “ChIP-seq would produce a more reliable profile of the HNF4a and CREB1 binding sites on these loci and in other control regions”. However, we believe that our current ChIP-qPCR experiments have adequately addressed the potential concerns raised by the reviewers. (1) We have ensured that the DNA fragment length after sonication falls within the range of 200 bp to 500 bp, with an average length of approximately 300 bp (Author response image 1A). We have stated the point in the revised methods section (line 633). (2) We have randomly inspected 14 out of 26 primer sets used in Figure 6 and its supplemental figure (Author response image 1B-E), confirming that all primer sets demonstrate equal amplification efficiency (ranging from 90% to 110%). This information has also been included in the revised methods section (line 650). (3) Figures 6G and 6H show reduced enrichment of HNF4𝛼 (6G) and p-S133-CREB1 (6H) at the Ctnnb1 promoter in ieCtnnb1 knockout ApcMin/+ tumor tissues. The ChIP-qPCR primers used were positioned at the Ctnnb1 promoter, not at ieCtnnb1, with IgG control enrichment serving as the reference values on the Y-axes. 

Author response image 1.

(A) Agarose gel electrophoresis of sonicated DNA. (B-E) Tests of amplification efficiency for primer sets used in ChIP-qPCR.

(9) The ChIP-qPCR showing preferential binding of pS133-CREB1 in small intestinal crypts and CHT15 cells (line 393) should be shown. 

The ChIP-qPCR results demonstrating preferential binding of p-S133-

CREB1 over CREB1 have been added in revised Figure 6C, 6D and Figure 6 – Supplement 1C.

(10) It is not entirely clear what the blue tracks represent at the bottom of Figures 6C-D and Figure 6 - Figure Supplement 1C-D. The ChIP-seq profiles of both CREB1 and HNF4a shown in Figures 6A and Figure 6 - Figure Supplement 1A do not seem to match. Taking HNF4a, for example from Figure 6 - Figure Supplement 1A it seems to bind on the Ctnnb1 promoter, while in Figure 6 - Figure Supplement 1D the peaks are within the first intron. I realize this might all be a problem with a different scale across figure panels, but I suggest producing a cleared figure.

We apologize for the confusion. We have revised Figure 6C-6D, Figure 6 - figure supplement 1C-D, and the corresponding legends to enhance clarity. (1) The top panels of Figures 6C and 6D respectively highlight shaded regions of ieCTNNB1 (pink) and the CTNNB1 promoter (grey) in Figure 6A, emphasizing the enrichment of p-S133-CREB1.  (2) The top panels of Figure 6 – figure supplement 1C and 1D respectively highlight shaded regions of ieCtnnb1 (pink) and the Ctnnb1 promoter (grey) in Figure 6A – figure supplement 1A, emphasizing the enrichment of HNF4α. (3) Because Figures 6C-6D and Figure 6 - figure supplement 1C-1D respectively correspond to human and mouse genomes, the positions of peaks and scales differ.  

(11) In the intro the authors refer to "TCF-4". I suggest they use the more recent unambiguous nomenclature for this family of transcription factors and call it TCF7L2.

TCF-4 has been changed into TCF7L2 in the revision (line 81)

(12) In lines 121-122, the authors write "Although numerous putative enhancers...only a fraction of them were functionally annotated". To what study/studies are the authors referring? Please provide references.

References were added in the revision (line 124)

(13) In some parts the authors use strong words that should in my opinion be attenuated. Examples are: (i) at line 224, "maintains" would be better substituted with "contribute", as in the absence of ieCtnnb1, Ctnnb1 is still abundantly expressed; (ii) at line 266 "compromised" when the proliferative capacity of CFCs and TACs seems to be only mildly reduced; (iii) at line 286 "disrupts", the genes are simply downregulated.

We thank these great suggestions. 1) On lines 224-225, the sentence was revised to: “These data suggest that ieCtnnb1 plays a specific role in regulating the transcription of Ctnnb1 in intestinal epithelia”. 2) On line 271, “compromised” were replaced with “mildly reduced”. 3) In ieCtnnb1 knockout epithelial cells of small intestine, genes related to secretory functions were decreased, while genes related to absorptive functions were increased. Therefore, the term 'disrupts' is more appropriate than 'downregulates'. 

Reviewer #3:

Line 81, c-Myc should be human MYC (italics) to agree with the other human gene names in this sentence. 

c-Myc has been changed into MYC in the revision (line 82)

Line 215, wildtype should be wild-type. 

“wildtype” has been changed into “wild-type” in the revision (line 215)

Line 224, Elimination of the enhancer did not abolish expression of Ctnnb1; therefore, it would be better to say that it "helps to maintain Ctnnb1 transcription" 

The sentence was changed into “These data suggest that ieCtnnb1 plays a specific role in regulating the transcription of Ctnnb1 in intestinal epithelia” in revision (lines 224-225)

Line 228, perhaps "to activate transcription" is meant. 

“active” has been changed into “activate” in the revision (line 228)

Line 235, consider "reduced" instead of "undermined". 

“undermined” has been replaced with “compromised” in the revision (line 237)

Line 262, "em" dashes should be a both ends of this insertion. 

Line 298, "dysfunctional" would be better.

Line 356, "samples were". 

Line 481, 12-hr (add hyphen). 

All above points have been optimized according to the reviewer’s suggestion.

Line 712, Is "poly-N" meant? 

“Poly-N” indicates undetected bases during sequencing. This explanation was added in the revision (lines 759-760).

Figure 1K, the GAPDH signal is not visible and that panel is unnecessary as there is an H3 control.   

Figure 1K and 1L respectively show levels of nuclear and cytoplasmic βcatenin. GAPDH and H3 were used as internal references for the cytoplasmic and nuclear fractions, respectively, confirming both robust fractionation and equal loading.

Our IEP team is careful to say to our student's parents that it is our recommendation that their student is in the correct placement for the upcoming year. Our unit classes include students that have severe intellectual disabilities. When we have our general ed teachers talk about what standards the general ed students are currently working on, the parent's are sometimes shocked.

At our school, we haven't had any issues with failing to assemble an appropriate IEP team. Our administrators do a great job of ensuring that all IEP team members know what is expected of them. We have sped meetings weekly to ensure that we stay on top of IEP due dates and re-evaluations.

I believe that there is less oversight on the substantive side of IEP implementation accountability. In two years I've never had my progress monitoring checked by anybody. I've also never had parents inquire as to how their child is progressing towards their IEP goals.

Regarding implementation errors, I don't think I've ever seen an administrator verify the services provided by different related service providers such as speech or OT/PT. It may be assumed that these related services are professionals and that they will do their duty to maintain fidelity to the IEP.

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

Learn more at Review Commons

Reply to the reviewers

Reviewer1

I recommend combining Figures 1 and 3 either before or after the results shown in Figure 2, so the reader's expectation for quantification is immediately satisfied.

Response：

Thank you for your suggestion. In the revised manuscript, images of GFP::POP-1 in compound mutants are moved to Figure 3. The schematic diagram of the gonad (previously Fig. 1A) and GFP::POP-1 images in wild type are kept in Figure 1, as they are described in Introduction.

Major comments:

Delving into the figure legend of Fig. 3 and the normalization procedure described in the Methods "Quantification of POP-1 asymmetry in the Z1 and Z4 division" raised concerns. The method therein described is overly complicated but also neglects background subtraction. My first question about this method: what range of distances between daughters is measured in Z? These distances are not discussed in absolute terms, and this is important for our understanding of how much correction for tissue depth might be necessary, as L1s are very thin.

To check my understanding, the authors use as a control a nuclear-localized GFP driven in the somatic gonad precursors in otherwise wild-type worms by the sys-1 promoter. They observe that the regression on a log scale of anterior:posterior (and vice versa) Z1 and Z4 daughter fluorescence over the distance between the daughters in the Z plane is fit by y = −0.034x + 0.0148, which is practically a slope of 0 and an intercept of 0. This means that they observed an ~1:1 ratio (as log(1)=0) of fluorescence in the anterior and posterior daughters of otherwise wild-type worms, at least across the range of very small X values of relevant distances between daughters (again, the relevant range of distances really matters and should be presented), making the normalization seem unnecessary.

Response:

Normalization is essential to compare POP-1 signals between daughter cells since the signal intensities depend on the depth of cells. Depth differences between SGP daughter cells range from 0 to 7.5 micrometers. For example, when we input the maximum difference (7.5) into our correction equation y = −0.034x + 0.0148 (the logarithmically transformed linear regression equation), we get:

y = −0.034 * 7.5 + 0.0148 = -0.2402

To interpret this on the original scale, we apply the inverse logarithmic transformation:

10^(-0.2402) ≈0.575

This result indicates that even if GFP::POP-1 expression is the same in both cells, the depth difference alone can cause approximately a 1.74-fold (1/0. 575) difference in fluorescence intensity.

Similarly, if we use a median value of 3.5 micrometers as the depth difference, we get: y = -0.1042. After the inverse logarithmic transformation, this corresponds to a 0.787 or 1.27 (1/0.787) fold difference in fluorescence intensity.

Without normalization, we risk misinterpreting such differences in expression levels when in reality the expression is the same. Conversely, actual differences in GFP::POP-1 signal could be masked or overestimated due to the depth effect.

In the revised manuscript, examples of depth differences between SGP daughters are shown in Fig. 2S which is added in response to the comment of reviewer 2, asking images of lin-17 mom-5 animals.

In the revised manuscript, we explained the depth effects in the legend of Fig. 3 as follows.

"Since SGP daughter cells are often present at distinct focal planes, we normalized the depth effects on fluorescence intensities (see Materials and Methods for details) for the quantification shown in (B). The images in (A) and (C) are from animals with SGP daughters at similar depths."

Then based on this regression and 95% CI, the authors predict values that reflect true equivalence of fluorescence of POP-1::GFP in the two SGP daughters, compare the observed values to these predictions, and ultimately display in violin plots these differences of observed and expected. Correct?

Response:

Yes, your understanding is correct.

Is this complicated treatment the only way to detect differences in polarity of anterior and posterior daughters of Z1 and Z4? What happens if the authors measure GFP::POP-1 and calculate the following?

Z1p(MGV - background control from same focal plane)

Z1a(MGV - background control from same focal plane)

If this straightforward analysis shows asymmetric signal in the control that is made symmetrical or reversed in the mutants, the hypothesis would seem to be supported with a much more straightforward method. Samples could be analyzed separately in two bins by worm body position, which affects which cell is superficial in the sample. As it is, the Figure 3 Y axis label is hard to interpret without reading the methods at length, diminishing its impact.

Response:

Thank you for the suggestion. Your suggested calculation would be simple if we could assume that control signals (sys-1p::GFP::NLS or sys-1p::GFP::POP-1 in the same wild-type cell) on the same focal plane are the same among animals. However, since there are apparent variations in expression levels among individuals, your suggested method is not appropriate for evaluating differences in sys-1p::GFP::POP-1 intensities between the SGP daughter cells of the same animal.

Missing control: The sys-1 promoter-driven NLS-tagged fluorescent protein as a control to compare to the GFP::POP-1 is analyzed only in the wild-type, and apparently not in the mutants under consideration. Phillips et al. (2007) show that sys-1p transcriptional activity is equivalent between the SGP daughters in wild-type worms, but neither those results nor the method of normalizing to a sys-1p::GFP::NLS signal in this paper address the question of whether sys-1 promoter activity is equivalent in these cells in mutants upstream in the Wnt pathway. If the current method of normalization is to be used, it seems important to normalize to the sys-1p::GFP::NLS regression in each mutant background.

Response:

Thank you for your suggestion. We used sys-1p::GFP::NLS as a control to normalize depth effects, which should be the same across all genotypes because the GFP molecules in SGPs should be equally distributed between SGP daughter cells, not because sys-1 promoter activities are similar among them. Since SGP daughters divide within a short time (about 2 hours), it is likely that the fluorescence of newly synthesized GFP (maturation time of about 1 hour) in SGP daughters is neglectable compared to GFP inherited from the SGP cells. Similarly, sys-1p::GFP::POP-1 signals in SGP daughters reflect the distribution of GFP::POP-1 from SGPs rather than the transcriptional activities of the sys-1 promoter in the daughter cells. sys-1p::GFP::POP-1 or sys-1p::GFP::SYS-1 has been widely used to evaluate polarity of asymmetric divisions in a number of studies, none of which consider transcriptional differences of the sys-1 promoter in the daughter cells.

1. How was lin-17(mn589) generated? if this is the first report of this allele, full information on what the lesion is and how it was derived should to be reported.

Response:

Thank you for your question regarding the lin-17(mn589) allele. We would like to point out that the information about this allele is provided in the Methods section of the original manuscript as follows.

"lin-17(mn589) (gifted by Mike Herman) carries a mutation in the seventh cysteine residue of the CRD domain (C104Y). mn589 exhibits 47% Psa phenotype (indicating T cell polarity defects)."

The methods section lacks a description of how the mes-1 experiments were done, in terms of timing, duration, and temperature; mes-1(bn7) is a temperature sensitive allele.

Response:

Thank you for pointing out the lack of detailed methodology for the mes-1 experiments. The germless phenotype of mes-1 mutants is partial even at high temperatures. We have not performed temperature shifts to observe the phenotype. As per your suggestion, we added the following text to the Strains section:

"mes-1(bn7) is a temperature-sensitive allele with higher penetrance of the germless phenotype at 25{degree sign}C than at 15{degree sign}C, and was grown at 22.5{degree sign}C. The germless phenotype of mes-1(bn7) was observed by the absence of the mex-5::GFP::PH signal through direct observation of epifluorescence."

Minor comments

1. The paper lacks a discussion of precedent in the literature for Wnt-independent Frizzled activity; this is a major finding that is being undersold in the current version of the manuscript.

Response:

Thank you very much for appreciating out finding. We have added the following paragraph to the Discussion section:

"Wnt-independent functions of Frizzled receptors

We have shown that lin-17/Fzd functions in a Wnt-independent manner to control SGP polarity, since the missing DTC phenotype of lin-17; cwn-2 and lin-17 mom-5 was completely rescued by ΔCRD-LIN-17. In addition, SGP polarity is normal in the quintuple Wnt mutant that has mutations in all the Wnt genes (Yamamoto et al., 2011). In seam cells, Wnt receptors including LIN-17/Fzd and MOM-5/Fzd appear to have Wnt-independent functions for cell polarization, as seam cells are still mostly polarized in the quintuple Wnt mutants, while they are strongly unpolarized in the triple receptor mutants (lin-17 mom-5 cam-1/Ror) (Yamamoto et al., 2011). In Drosophila, Fz/Fzd has been primarily considered to function Wnt-independently to coordinate planar cell polarity (PCP) between neighboring cells (Lawrence et al., 2007), though Fz function can still be regulated by Wnt, as PCP orientation can be directed by ectopically expressed Wnt proteins (Wu et al., 2013).

          In Drosophila, Fz regulates PCP by interacting with other PCP components including Van Gogh (Vang). In C. elegans, we found that vang-1/Vang does not appear to function with LIN-17/Fz, since most vang-1 single mutants and cwn-1 cwn-2 vang-1 triple mutants have two gonadal arms (215/216 and 58/58, respectively). As Fz interacts with Disheveled (DSH) in Drosophila PCP regulation, in C. elegans, the Disheveled homologs DSH-2 and MIG-5 regulate SGP polarity (Phillips et al., 2007). Therefore, LIN-17 might regulate the DSH homologs in a Wnt-independent manner. "

Added Reference：

1. Lawrence PA, Struhl G, Casal J. (2007). Planar cell polarity: one or two pathways? Nat Rev Genet. 8, 555-563.
2. Wu, J., Roman, A.C., Carvajal-Gonzalez, J.M., & Mlodzik, M. (2013). Wg and Wnt4 provide long-range directional input to planar cell polarity orientation in Drosophila. Nature Cell Biology, 15(9), 1045-1055.

Important: I think "Fig. 6 Germ cell independent migration of germ cells" title is a typo; should be "Germ cell independent migration of DTCs"

Response:

Thank you for pointing out the typo. We corrected it in the revised manuscript.

This is a very important experiment! I think a greater description of the mes-1 phenotype would be helpful, since loss of germline was not 100% penetrant in mes-1(bn7) hermaphrodites in Strome et al., 1995. The legend says "Germless mes-1 phenotype was confirmed by the absence of the mex-5::GFP::PH signal in the gonad." Consider adding a few sentences to the results (or methods, from which the mes-1 experiments are currently missing) describing that only mes-1 animals that lacked germline fluorescence were analyzed for DTC migration.

Response:

Thank you for providing the context. To address the concerns, we made the following changes to our manuscript:

1. In the Results section, we revised the sentence "We found that 84% of DTCs (n = 90) in germless mes-1 animals..." to "Among mes-1 animals that lack germ cells, we found 84% of DTCs (n = 90)...".
2. We also modified the sentence "We noticed that some germless mes-1 animals..." to "We noticed that some mes-1 animals that lack germ cells...".

Please correct "secreting the Notch ligand LAG-2" this is a membrane-bound, not secreted ligand

Response:

Thank you for your comment. In the revised manuscript, we modified the relevant sentence in the Introduction section as follows:

"Firstly, DTCs function as niche cells for germline stem cells, inhibiting their entry into meiosis by expressing the Notch ligand LAG-2 (Henderson et al., 1994)."

Fig 1. The qualitative loss of polarity would be better depicted with

a grayscale image instead of green-on-black.

Response:

Thank you for your suggestion. The GFP::POP-1 images are raw images of the green channel of the confocal microscopy. We believe that SGP polarity is clearly depicted by them.

Fig. 3 the presentation of these violin plots is confusing. The central text that reads "normal polarity, loss of polarity, reversed polarity" with arrows looks like a second Y axis label attached to the Z4 plot. I recommend rearranging. Consider shading the top, bottom, and central regions and explaining the meaning of the shading in the legend.

Response:

Thank you for your suggestions regarding the presentation of Figure 3. In response to your feedback, we have made the following modifications:

First, we moved the text and arrows from the center to the right side of the figure, creating a clearer layout. As you recommended, we applied shading to the top, bottom, and central regions of the violin plots. Additionally, to explain the meaning of the shading, we added a new explanation to the figure legend. Specifically, we included the following text:

"Values within the 95% CI (between the red lines; light green regions) indicate symmetric localization. Values below the lower red line (light blue regions) indicate reversed localization, while values above the upper red line (light red regions) indicate normal localization."

We applied the same modification to Supplemental Fig. 1.

Reviewer 2

Major comments

1- Are the effects of combining the different Wnts with the lin-17 allele specific to the n3091 allele? It would be important to test another allele, for example the sy277 allele has a similar phenotype and is available at CGC. A null would be even better if it is viable. Alternatively, lin-17(RNAi) could instead be used if efficient enough. This is important since the n3091 allele could differentially alter the binding to the various Wnts, resulting in their distinct phenotypes in that background. However, these distinct phenotypes may not be relevant in a wild-type context.

Response:

Thank you for your insightful comment. The lin-17(n3091) allele contains a nonsense mutation at the 35th codon, located between the second and third cysteine residues in the CRD domain (Wnt binding domain) (Sawa et al 1996). Therefore, it is highly unlikely that the N-terminal protein of 34 amino acids produced in lin-17(n3091) can bind to Wnts. In the revised manuscript, we added the missing-DTC phenotype of lin-17(n671) cwn-2 animals, which show a similar phenotype to lin-17(n3091) cwn-2. n671 is a reference allele in WormBase and has a nonsense mutation. Although sy277 has a deletion in the N-terminal region, its phenotype is weaker than that of n3091 and n671 (Sawa et al 1996).

In the revised manuscript, we described lin-17(n671) cwn-2, in the Table 1, Table S1 and added the following sentence.

"We observed a similar phenotype in lin-17(n671); cwn-2 double mutants, confirming that this genetic interaction is not allele-specific."

2- In the lin-17; mom-5 double mutant which lacks DTCs, are Z1 and Z4 there but they do not express DTC markers, or are they never born? A lineage analysis should be presented. Also, are Z2 and Z3 still there on their own? Please show images.

Response:

Thank you for your comments. We quantified sys-1p::GFP::POP-1 signals in Z1 and Z4 daughter cells of lin-17 mom-5 and have not observed any animals lacking Z1, Z4 or germ cells. In the revised manuscript, as Fig. S2, we added images of sys-1p::GFP::POP-1 localizations in SGP daughters, along with germ cells in lin-17 mom-5 as well as in lin-17 cwn-1 egl-20 cwn-2, both of which were not shown in the original manuscript. In response to Reviewer 1's comment, we also included examples of depth effects on fluorescence intensities in Fig. S2 with images of different focal planes.

Fig. S2 is quoted it at the end of the following sentence.

"Then, we quantified the ratios (on a logarithmic scale) of sys-1p::GFP::POP-1 signal intensities proximal to distal daughter cells in various genotypes (Fig. 3A and Fig. S2)."

The loss of polarity phenotype of lin-17 mom-5 has been described in Phillips et al. We missed to cite this in the original manuscript. We added the citation in the revised manuscript.

"These asymmetries were strongly disrupted and weakly affected in lin-17 mom-5 double and lin-17 single mutants, respectively, as described previously (Phillips et al., 2007; Siegfried et al., 2004)."

Minor comments

1- The term "mirror-symmetry" is redundant. Consider using "symmetry"

or "symmetrical polarity".

Response:

As noted in the cross-comment by Reviewer 1, we believe that "mirror-symmetry" is the appropriate term.

We think that "symmetry" implies the same lineage, whereas the relationship between the Z1 and Z4 lineages is not. "Mirror symmetry" was also used in Herman & Horvitz (1994) to describe the defect in the F lineage in lin-44/Wnt mutants as follows.

"we observed division patterns that were mirror symmetric to those of the wild type (Fig. 2). One plausible explanation is that the polarity of the first asymmetric cell division was reversed, causing the polarities of all subsequent asymmetric cell divisions also to be reversed."

2- "... they are permissively pushed distally by germ cells while proliferating" is confusing as it is unclear what proliferating cell you are referring to - germ cells or the DTC? proliferating? sense. Replace by: "they are pushed distally by proliferating germ cells"

Response:

Thank you for your helpful comment. We agree with your suggestion and modify the sentence as follows:

Original: "... they are permissively pushed distally by germ cells while proliferating" Revised: "... they are pushed distally by proliferating germ cells"

3- Fig. 2 is cited in the text before Fig. 1.

Response:

Thank you for pointing this out. Figure1 is mentioned in the Introduction before Figure 2 is referenced in the Result section in the original manuscript. We think the reviewer might be confused, as the POP-1 localization defect was shown in Figure 1. In response to the reviewer 1's comment, we moved the POP-1 localization images of the compound mutants to Figure 3. In addition, we noticed that in the original manuscript, Figure 1B was mentioned before Figure 1A in the Introduction. Therefore, we have modified the sentences in the Introduction.

The original sentence was:

"In the gonad, at the L1 stage, somatic gonadal precursor cells (SGPs), Z1 and Z4 have LH and HL polarity, respectively (Siegfried et al., 2004) (Fig. 1B). This mirror-symmetric polarity creates their mirror-symmetric lineages producing distal tip cells (DTCs) from the distal daughters (Z1.a and Z4.p) (Fig. 1B)."

The revised sentence now reads:

"In the gonad, at the L1 stage, somatic gonadal precursor cells (SGPs), Z1 and Z4 have LH and HL polarity, respectively, creating their mirror-symmetric lineages producing distal tip cells (DTCs) from the distal daughters (Z1.a and Z4.p) (Siegfried et al., 2004) (Fig. 1A and 1B)."

4- The results also suggest that MOM-5/Frizzled might be the receptor for Wnts regulating DTC production, as lin-17 mom-5 double mutants completely lack DTCs." Table 1 results rather suggest that lin-17 and mom-5 are the two frizzled receptor involved in DTC specification and that they are largely redundant.

Response:

As the reviewer noted, lin-17 and mom-5 function redundantly in DTC specification (SGP polarization). However, their functions are clearly different in terms of genetic interactions with Wnt genes (e.g. lin-17 cwn-2 but not mom-5 cwn-2 show the DTC-missing phenotype). We propose that MOM-5 but not LIN-17 functions as a receptor for Wnts.

I've seen this a few times with people who do not want to identify as feminists. A lot of times their reasoning is that it is no longer necessary. They might say they believe men and women should be equal, but that they already are (at least in the United States), so they do not want to identify as a feminist. I think one reason this might occur is that people might not want to admit to themselves that any injustice is still occurring as this may be distressing. Also, like the text says, I've heard people use progress as proof that feminism should be "done" and men and women are now equal, ignoring the ways that we still aren't. Some of achievements of feminism in the past might be more obvious as they are directly and clearly written into the law (like earning the right to vote), but there are still so many ways that gender inequality persists today.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

In this study, the authors used a stopped-flow method to investigate the kinetics of substrate translocation through the channel in hexameric ClpB, an ATP-dependent bacterial protein disaggregase. They engineered a series of polypeptides with the N-terminal RepA ClpB-targeting sequence followed by a variable number of folded titin domains. The authors detected translocation of the substrate polypeptides by observing the enhancement of fluorescence from a probe located at the substrate's C-terminus. The total time of the substrates' translocation correlated with their lengths, which allowed the authors to determine the number of residues translocated by ClpB per unit time.

Strengths:

This study confirms a previously proposed model of processive translocation of polypeptides through the channel in ClpB. The novelty of this work is in the clever design of a series of kinetic experiments with an engineered substrate that includes stably folded domains. This approach produced a quantitative description of the reaction rates and kinetic step sizes. Another valuable aspect is that the method can be used for other translocases from the AAA+ family to characterize their mechanism of substrate processing.

Weaknesses:

The main limitation of the study is in using a single non-physiological substrate of ClpB, which does not replicate the physical properties of the aggregated cellular proteins and includes a non-physiological ClpB-targeting sequence. Another limitation is in the use of ATPgammaS to stimulate the substrate processing. It is not clear how relevant the results are to the ClpB function in living cells with ATP as the source of energy, a multitude of various aggregated substrates without targeting sequences that need ClpB's assistance, and in the presence of the co-chaperones.

Indeed, we agree that our RepA-Titinx substrates are not aggregates but are model, soluble, substrates used to reveal information about enzyme catalyzed protein unfolding and translocation.  Our substrates are similar to RepA-GFP and GFP-SsrA used by multiple labs including Wickner, Horwich, Sauer, Baker, Shorter, Bukua, to name only a few.  The fact that “this is what everyone does” does not make the substrates physiological or the most ideal. However, this is the technology we currently have until we and others develop something better. In the meantime, we contend that  the results presented here do advance our knowledge on enzyme catalyzed protein unfolding

Part of what this manuscript seeks to accomplish is presenting the development of a single-turnover experiment that reports on processive protein unfolding by AAA+ molecular motors, in this case, ClpB.  Importantly, we are treating translocation on an unfolded polypeptide chain and protein unfolding of stably folded proteins as two distinct reactions catalyzed by ClpB. If these functions are used to disrupt protein aggregates, in vivo, then this remains to be seen.

We contend that processive ClpB catalyzed protein unfolding has not been rigorously demonstrated prior to our results presented here.  Avellaneda et al mechanically unfolded their substrate before loading ClpB (Avellaneda, Franke, Sunderlikova et al. 2020).  Thus, their experiment represents valuable observations reflecting polypeptide translocation on a pre-unfolded protein.  Our previous work using single-turnover stopped-flow experiments employed unstructured synthetic polypeptides and therefore reflects polypeptide translocation and not protein unfolding (Li, Weaver, Lin et al. 2015).  Weibezahn et al used unstructured substrates in their study with ClpB (BAP/ClpP), and thus their results represent translocation of a pre-unfolded polypeptide and not enzyme catalyzed protein unfolding (Weibezahn, Tessarz, Schlieker et al. 2004). 

Many studies have reported the use of  GFP with tags or RepA-GFP and used the loss of GFP fluorescence to conclude protein unfolding.  However, such results do not reveal if ClpB processively and fully translocates the substrate through its axial channel.  One cannot rule out, even when trapping with “GroEL trap”, the possibility that ClpB only needs to disrupt some of the fold in GFP before cooperative unfolding occurs leading to loss of fluorescence.  Once the cooperative collapse of the structure occurs and fluorescence is lost it has not been shown that ClpB will continue to translocate on the newly unfolded chain or dissociate. In fact, the Bukau group showed that folded YFP remained intact after luciferase was unfolded (Haslberger, Zdanowicz, Brand et al. 2008).  Our approach, reported here, yields signal upon arrival of the motor at the c-terminus or within the PIFE distance thus we can be certain that the motor does arrive at the c-terminus after unfolding up to three tandem repeats of the Titin I27 domain.

ATPgS is a non-physiological nucleotide analog.  However, ClpB has been shown to exhibit curious behavior in its presence that we and others, as the reviewer acknowledges, do not fully understand (Doyle, Shorter, Zolkiewski et al. 2007).  Some of the experiments reported here are seeking to better understand that fact.  Here we have shown that ATPgS alone will support processive protein unfolding. With this assay in hand, we are now seeking to go forward and address many of the points raised by this reviewer. 

The authors do not attempt to correlate the kinetic step sizes detected during substrate translocation and unfolding with the substrate's structure, which should be possible, given how extensively the stability and unfolding of the titin I27 domain were studied before. Also, since the substrate contains up to three I27 domains separated with unstructured linkers, it is not clear why all the translocation steps are assumed to occur with the same rate constant.

We assume that all protein unfolding steps occur with the same rate constant, ku.  We conclude that we are not detecting the translocation rate constant, kt, as our results support a model where kt is much faster than ku.  We do think it makes sense that the same slow step occurs between each cycle of protein unfolding.

We have added a discussion relating our observations to mechanical unfolding of tandem repeats of Titin I27 from AFM experiments  (Oberhauser, Hansma, Carrion-Vazquez and Fernandez 2001). Most interestingly, they report unfolding of Titin I27 in 22 nm steps.  Using 0.34 nm per amino acids this yields ~65 amino acids per unfolding step, which is comparable to our kinetic step-size of 57 – 58 amino acids per step.

Some conclusions presented in the manuscript are speculative:

The notion that the emission from Alexa Fluor 555 is enhanced when ClpB approaches the substrate's C-terminus needs to be supported experimentally. Also, evidence that ATPgammaS without ATP can provide sufficient energy for substrate translocation and unfolding is missing in the paper.

In our previous work we have used fluorescently labeled 50 amino acid peptides as substrates to examine ClpB binding (Li, Lin and Lucius 2015, Li, Weaver, Lin et al. 2015).  In that work we have used fluorescein, which exhibits quenching upon ClpB binding.  We have added a control experiment where we have attached alexa fluor 555 to the 50 amino acid substrate so we can be assured the ClpB binds close to the fluorophore.  As seen in supplemental Fig. 1 A  upon titration with ClpB, in the presence of ATPγS, we observe an increase in fluorescence from AF555, consistent with PIFE.  Supplemental Fig. 1 B shows the relative fluorescence enhancement at the peak max increases up to ~ 0.2 or a 20 % increase in fluorescence, due to PIFE, upon ClpB binding.   

Further, peak time is our hypothesized measure of ClpB’s arrival at the dye. Our results indicate that the peak time linearly increases as a function of an increase in the number of folded TitinI27 repeats in the substrates which also supports the PIFE hypothesis. Finally, others have shown that AF555 exhibits PIFE and we have added those references.

The evidence that ATPγS alone can support translocation is shown in Fig. 2 and supplemental Figure 1.  Fig. 2 and supplemental Figure 1 are two different mixing strategies where we use only ATPgS and no ATP at all.  In both cases the time courses are consistent with processive protein unfolding by ClpB with only ATPγS.

Reviewer #2 (Public Review):

Summary:

The current work by Banwait et al. reports a fluorescence-based single turnover method based on protein-induced fluorescence enhancement (PIFE) to show that ClpB is a processive motor. The paper is a crucial finding as there has been ambiguity on whether ClpB is a processive or non-processive motor. Optical tweezers-based single-molecule studies have shown that ClpB is a processive motor, whereas previous studies from the same group hypothesized it to be a non-processive motor. As co-chaperones are needed for the motor activity of the ClpB, to isolate the activity of ClpB, they have used a 1:1 ratio ATP and ATPgS, where the enzyme is active even in the absence of its co-chaperones, as previously observed. A sequential mixing stop-flow protocol was developed, and the unfolding and translocation of RepA-TitinX, X = 1,2,3 repeats was monitored by measuring the fluorescence intensity with the time of Alexa F555 which was labelled at the C-terminal Cysteine. The observations were a lag time, followed by a gradual increase in fluorescence due to PIFE, and then a decrease in fluorescence plausibly due to the dissociation from the substrate allowing it to refold. The authors observed that the peak time depends on the substrate length, indicating the processive nature of ClpB. In addition, the lag and peak times depend on the pre-incubation time with ATPgS, indicating that the enzyme translocates on the substrates even with just ATPgS without the addition of ATP, which is plausible due to the slow hydrolysis of ATPgS. From the plot of substrate length vs peak time, the authors calculated the rate of unfolding and translocation to be ~0.1 aas-1 in the presence of ~1 mM ATPgS and increases to 1 aas-1 in the presence of 1:1 ATP and ATPgS. The authors have further performed experiments at 3:1 ATP and ATPgS concentrations and observed ~5 times increase in the translocation rates as expected due to faster hydrolysis of ATP by ClpB and reconfirming that processivity is majorly ATP driven. Further, the authors model their results to multiple sequential unfolding steps, determining the rate of unfolding and the number of amino acids unfolded during each step. Overall, the study uses a novel method to reconfirm the processive nature of ClpB.

Strengths:

(1) Previous studies on understanding the processivity of ClpB have primarily focused on unfolded or disordered proteins; this study paves new insights into our understanding of the processing of folded proteins by ClpB. They have cleverly used RepA as a recognition sequence to understand the unfolding of titin-I27 folded domains.

(2) The method developed can be applied to many disaggregating enzymes and has broader significance.

(3) The data from various experiments are consistent with each other, indicating the reproducibility of the data. For example, the rate of translocation in the presence of ATPgS, ~0.1 aas-1 from the single mixing experiment and double mixing experiment are very similar.

(4) The study convincingly shows that ClpB is a processive motor, which has long been debated, describing its activity in the presence of only ATPgS and a mixture of ATP and ATPgS.

(5) The discussion part has been written in a way that describes many previous experiments from various groups supporting the processive nature of the enzyme and supports their current study.

Weaknesses:

(1) The authors model that the enzyme unfolds the protein sequentially around 60 aa each time through multiple steps and translocates rapidly. This contradicts our knowledge of protein unfolding, which is generally cooperative, particularly for titinI27, which is reported to unfold cooperatively or utmost through one intermediate during enzymatic unfolding by ClpX and ClpA.

We do not think this represents a contradiction.  In fact, our observations are in good agreement with mechanical unfolding of tandem repeats of Titin I27 using AFM experiments (Oberhauser, Hansma, Carrion-Vazquez and Fernandez 2001).  They showed that tandem repeats of TitinI27 unfolded in steps of ~22 nm.  Dividing 22 nm by 0.34 nm/Amino Acid gives ~65 amino acids per unfolding event.  This implies that, under force, ~65 amino acids of folded structure unfolds in a single step.  This number is in excellent agreement with our kinetic step-size of 65 AA/step. 

Importantly, the experiments cited by the reviewer on ClpA and ClpX are actually with ClpAP and ClpXP.  We assert that this is an important distinction as we have shown that ClpA employs a different mechanism than ClpAP (Rajendar and Lucius 2010, Miller, Lin, Li and Lucius 2013, Miller and Lucius 2014).  Thus, ClpA and ClpAP should be treated as different enzymes but, without question, ClpB and ClpA are different enzymes.

(2) It is also important to note that the unfolding of titinI27 from the N-terminus (as done in this study) has been reported to be very fast and cannot be the rate-limiting step as reported earlier(Olivares et al, PNAS, 2017). This contradicts the current model where unfolding is the rate-limiting step, and the translocation is assumed to be many orders faster than unfolding.

Most importantly, the Olivares paper is examining ClpXP and ClpAP catalyzed protein unfolding and translocation and not ClpB.  These are different enzymes.  Additionally, we have shown that ClpAP and ClpA translocate unfolded polypeptides with different rates, rate constants, and kinetic step-sizes indicating that ClpP allosterically impacts the mechanism employed by ClpA to the extent that even ClpA and ClpAP should be considered different enzymes (Rajendar and Lucius 2010, Miller, Lin, Li and Lucius 2013).  We would further assert that there is no reason to assume ClpAP and ClpXP would catalyze protein unfolding using the same mechanism as ClpB as we do not think it should be assumed ClpA and ClpX use the same mechanism as ClpAP and ClpXP, respectively. 

The Olivares et al paper reports a dwell time preceding protein unfolding of ~0.9 and ~0.8 s for ClpXP and ClpAP, respectively.   The inverse of this can be taken as the rate constant for protein unfolding and would yield a rate constant of ~1.2 s-1, which is in good agreement with our observed rate constant of 0.9 – 4.3 s-1 depending on the ATP:ATPγS mixing ratio.  For ClpB, we propose that the slow unfolding is then followed by rapid translocation on the unfolded chain where translocation by ClpB must be much faster than for ClpAP and ClpXP.  We think this is a reasonable interpretation of our results and not a contradiction of the results in Olivares et al. Moreover, this is completely consistent with the mechanistic differences that we have reported, using the same single-turnover stopped flow approach on the same unfolded polypeptide chains with ClpB, ClpA, and ClpAP (Rajendar and Lucius 2010, Miller, Lin, Li and Lucius 2013, Miller and Lucius 2014, Li, Weaver, Lin et al. 2015).

(3) The model assumes the same time constant for all the unfolding steps irrespective of the secondary structural interactions.

Yes, we contend that this is a good assumption because it represents repetition of protein unfolding catalyzed by ClpB upon encountering the same repeating structural elements, i.e. Beta sheets. 

(4) Unlike other single-molecule optical tweezer-based assays, the study cannot distinguish the unfolding and translocation events and assumes that unfolding is the rate-limiting step.

Although we cannot, directly, distinguish between protein unfolding and translocation we have logically concluded that protein unfolding is likely rate limiting. This is because the large kinetic step-size represents the collapse of ~60 amino acids of structure between two rate-limiting steps, which we interpret to represent cooperative protein unfolding induced by ClpB.  It is not an assumption it is our current best interpretation of the observations that we are now seeking to further test. 

Reviewer #3 (Public Review):

Summary:

The authors have devised an elegant stopped-flow fluorescence approach to probe the mechanism of action of the Hsp100 protein unfoldase ClpB on an unfolded substrate (RepA) coupled to 1-3 repeats of a folded titin domain. They provide useful new insight into the kinetics of ClpB action. The results support their conclusions for the model setup used.

Strengths:

The stopped-flow fluorescence method with a variable delay after mixing the reactants is informative, as is the use of variable numbers of folded domains to probe the unfolding steps.

Weaknesses:

The setup does not reflect the physiological setting for ClpB action. A mixture of ATP and ATPgammaS is used to activate ClpB without the need for its co-chaperones, Hsp70. Hsp40 and an Hsp70 nucleotide exchange factor. This nucleotide strategy was discovered by Doyle et al (2007) but the mechanism of action is not fully understood. Other authors have used different approaches. As mentioned by the authors, Weibezahn et al used a construct coupled to the ClpA protease to demonstrate translocation. Avellaneda et al used a mutant (Y503D) in the coiled-coil regulatory domain to bypass the Hsp70 system. These differences complicate comparisons of rates and step sizes with previous work. It is unclear which results, if any, reflect the in vivo action of ClpB on the disassembly of aggregates.

We agree with the reviewer, there are several strategies that have been employed to bypass the need for Hsp70/40 or KJE to simplify in vitro experiments.  Here we have developed a first of its kind transient state kinetics approach that can be used to examine processive protein unfolding.  We now seek to go forward with examining the mechanisms of hyperactive mutants, like Y503D, and add the co-chaperones so that we can address the limitations articulated by the reviewer.   In fact we already began adding DnaK to the reaction and found that DnaK induced ClpB to release the polypeptide chain (Durie, Duran and Lucius 2018).  However, the sequential mixing strategy developed here was needed to go forward with examining the impact of co-chaperones. 

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Line 1: I recommend changing the title of the paper to remove the terms that are not clearly defined in the text: "robust" and "processive". What are the Authors' criteria for describing a molecular machine as "robust" vs. "not robust"? A definition of processivity is given in equation 2, but its value for ClpB is not reported in the text, and the criteria for classifying a machine as "processive" vs. "non-processive" are not included. Besides, the Authors have previously reported that ClpB is non-processive (Biochem. J., 2015), so it is now clear that a more nuanced terminology should be applied to this protein. Also, Escherichia coli should be fully spelled out in the title.

The title has been changed.  We have removed “robust” as we agree with the reviewer, there is no way to quantify “robust”.  However, we have kept “processive” and have added to the discussion a calculation of processivity since we can quantify processivity.  Importantly, the unstructured substrates used in our previous studies represent translocation and not protein unfolding.  here, on folded substrates, we detect rate-limiting protein unfolding followed by rapid translocation.  Thus, we report a lower bound on protein unfolding processivity of 362 amino acids. 

Line 20: The comment about mitochondrial SKD3 should be removed. SKD3, like ClpB, belongs to the AAA+ family, and it is simply a coincidence that the original study that discovered SKD3 termed it an Hsp100 homolog. The similarity between SKD3 and ClpB is limited to the AAA+ module, so there are many other metazoan ATPases, besides SKD3, that could be called homologs of ClpB, including mitochondrial ClpX, ER-localized torsins, p97, etc.

Removed.

Lines 133-139. Contrary to what the authors state, it is not clear that the "lag-phase" becomes significantly shorter for subsequent mixing experiments (Figure 1E) perhaps except for the last one (2070 s). It is clear, however, that the emission enhancement becomes stronger for later mixes. This effect should be discussed and explained, as it suggests that the pre-equilibrations shorter than ~2000 sec do not produce saturation of ClpB binding to the substrate.

We have added supplemental figure 2, which represents a zoom into the lag region.  This better illustrates what we were seeing but did not clearly show to the reader.  In addition, we address all three changes in the time courses, i.e. extend of lag, change in peak position, and the change in peak height. 

Line 175. The hydrolysis rate of ATPgammaS in the presence of ClpB should be measured and compared to the hydrolysis rate with ATP/ATPgammaS to check if the ratio of those rates agrees with the ratio of the translocation rates. These experiments should be performed with and without the RepA-titin substrate, which could reveal an important linkage between the ATPase engine and substrate translocation. These experiments are essential to support the claim of substrate translocation and unfolding with ATPgammaS as the sole energy source.

The time courses shown in figure 2 and supplemental Figure 1 are collected with only ATPgS and no ATP.  The time courses show a clear increase in lag and appearance of a peak with increasing number of tandem repeats of titin domains.  We do not see an alternate explanation for this observation other than ATPγS supports ClpB catalyzed protein unfolding and translocation.  What is the reviewers alternate explanation for these observations?

We agree with the reviewer that the linkage of ATP hydrolysis to protein unfolding and translocation is essential and we are seeking to acquire this knowledge.  However, a simple comparison of the ratio of rates is not adequate. We contend that a complete mechanistic study of ATP turnover by ClpB is required to properly address this linkage and such a study is too substantial to be included here but is currently underway. 

All that said, the statement on line 175 was removed since we do not report any ATPase measurements in this paper.

Line 199: It is an over-simplification to state that "1:1 mix of ATP to ATPgammaS replaces the need for co-chaperones". This sentence should be corrected or removed. The ClpB co-chaperones (DnaK, DnaJ, GrpE) play a major role in targeting ClpB to its aggregated substrates in cells and in regulating the ClpB activity through interactions with its middle domain. ATPgammaS does not replace the co-chaperones; it is a chemical probe that modifies the mechanism of ClpB in a way that is not entirely understood.

We agree with the reviewer.  The sentence has been modified to point out that the mix of ATP and ATPγS activates ClpB.

Figure 3B, Supplementary Figure 5A. The solid lines from the model fit cannot be distinguished from the data points. Please modify the figures' format to clearly show the fits and the data points.

Done.

Lines 326, 329. It is not clear why the authors mention a lack of covalent modification of substrates by ClpB. AAA+ ATPases do not produce covalent modifications of their substrates.

The issue of covalent modification was presented in the introduction lines 55 – 60 pointing out that much of what we have learned about protein unfolding and translocation catalyzed by ClpA and ClpX is from the observations of proteolytic degradation catalyzed by the associated protease ClpP.  However, this approach is not possible for ClpB/Hsp104 as these motors do not associate with a protease unless they have been artificially engineered to do so. 

Lines 396-399. I am puzzled why the authors try to correlate the size of the detected kinetic step with the length of the ClpB channel instead of the size characteristics of the substrate.

We are attempting to discuss/rationalize the observed large kinetic step-size which, in part, is defined by the structural properties of the enzyme as well as the size characteristics of the substrate.  We have attempted to clarify this and better discuss the properties of the substrate as well as ClpB.

As I mentioned in the Public Review, it is essential to demonstrate that the emission increase used as the only readout of the ClpB position along the substrate is indeed caused by the proximity of ClpB to the fluorophore. One way to accomplish that would be to place the fluorophore upstream from the first I27 domain and determine if the "lag phase" in the emission enhancement disappears.

Alexa Fluor 555 is well established to exhibit PIFE.  However, as in the response to the public review, we have included an appropriate control showing this in supplemental Fig. 1.

Finally, the authors repetitively place their results in opposition to the study of Weibezahn et al. published in 2004 which first demonstrated substrate translocation by engineering a peptidase-associated variant of ClpB. It should be noted that the field of protein disaggregases has moved since the time of that publication from the initial "from-start-to-end" translocation model to a more nuanced picture of partial translocation of polypeptide loops with possible substrate slipping through the ClpB channel and a dynamic assembly of ClpB hexamers with possible subunit exchange, all of which may affect the kinetics in a complex way. However, the present study confirmed the "start-to-end" translocation model, albeit for a non-physiological ClpB substrate, and that is the take-home message, which should be included in the text.

It is not clear to us that the field has “moved on” since Weibezahn et al 2004.  Their engineered construct that they term “BAP” with ClpP is still used in the field despite us reporting that proteolytic degradation is observed in the absence of ATP with that system  (Li, Weaver, Lin et al. 2015) and should, therefore, not be used to conclude processive energy driven translocation. The “partial translocation” by ClpB is also grounded in observations of partial degradation catalyzed by ClpP with BAP from the same group (Haslberger, Zdanowicz, Brand et al. 2008). It is not clear to us that the idea of subunit exchange leading to the possibility of assembly around internal sequences is being considered.  We do agree that this is an important mechanistic possibility that needs further interrogation. We agree with the reviewer, all these factors are confounding and lead to a more nuanced view of the mechanism.

All that said, we have removed some of the opposition in the discussion.

Reviewer #2 (Recommendations For The Authors):

(1) It is assumed that the lag phase will be much longer than the phase in which we see a gradual increase in fluorescence, as the effect of PIFE is significant only when the enzyme is very close to the fluorophore. Particularly for RepA-titin3, the enzyme has to translocate many tens of nm before it is closer to the C-terminus fluorophore. However, in all cases, the lag time is lower or similar to the gradual increase phase (for example, Figure 3B). Could the authors explain this?

The extent of the lag, or time zero until the signal starts to increase, is interpreted to indicate the time the motor moves from it’s initial binding site until it gets close enough to the fluorophore that PIFE starts to occur.  In our analysis we apply signal change to the last intermediate and dissociation or release of unfolded RepA-TitinX.  The increase in PIFE is not “all or nothing”.  Rather, it is starting to increase gradually.  Further, because these are ensemble measurements, and each molecule will exhibit variability in rate there is increased breadth of the peak due to ensemble averaging. 

(2) Although the reason for differences in the peak position (for example, Figure 1E, 2B) is apparent, the reason for variations in the relative intensities has to be given or speculated.

We have addressed the reason for the different peak heights in the revised manuscript.  It is the consequence of the fact that each substrate has slightly different fluorescent labeling efficiencies.  Thus, for each sample there is a mix of labeled and unlabeled substrates both of which will bind to ClpB but the unlabeled ClpB bound substrates do not contribute to the fluorescence signal, but will represent a binding competitor.  Thus, for low labeling efficiency there is a lower concentration of ClpB bound to fluorescent RepA-Titinx and for higher labeling efficiency there is higher concentration of ClpB bound to RepA-Titinx leading to an increased peak height.  RepA-Titin2 has the highest labeling efficiency and thus the largest peak height.

Reviewer #3 (Recommendations For The Authors):

The authors should make it clear that they and previous authors have used different constructs or conditions to bypass the physiological regulation of ClpB action by Hsp70 and its co-factors as mentioned above. In particular, the construct used by Avellaneda et al should be explained when they challenge the findings of those authors.

Minor points:

The lines fitting the experimental points are difficult or impossible to see in Figures 2B, 3B, and s5B.

Fixed

Typo bottom of p6 - "averge"

Fixed

Avellaneda, M. J., K. B. Franke, V. Sunderlikova, B. Bukau, A. Mogk and S. J. Tans (2020). "Processive extrusion of polypeptide loops by a Hsp100 disaggregase." Nature.

Doyle, S. M., J. Shorter, M. Zolkiewski, J. R. Hoskins, S. Lindquist and S. Wickner (2007). "Asymmetric deceleration of ClpB or Hsp104 ATPase activity unleashes protein-remodeling activity." Nature structural & molecular biology 14(2): 114-122.

Durie, C. L., E. C. Duran and A. L. Lucius (2018). "Escherichia coli DnaK Allosterically Modulates ClpB between High- and Low-Peptide Affinity States." Biochemistry 57(26): 3665-3675.

Haslberger, T., A. Zdanowicz, I. Brand, J. Kirstein, K. Turgay, A. Mogk and B. Bukau (2008). "Protein disaggregation by the AAA+ chaperone ClpB involves partial threading of looped polypeptide segments." Nat Struct Mol Biol 15(6): 641-650.

Li, T., J. Lin and A. L. Lucius (2015). "Examination of polypeptide substrate specificity for Escherichia coli ClpB." Proteins 83(1): 117-134.

Li, T., C. L. Weaver, J. Lin, E. C. Duran, J. M. Miller and A. L. Lucius (2015). "Escherichia coli ClpB is a non-processive polypeptide translocase." Biochem J 470(1): 39-52.

Miller, J. M., J. Lin, T. Li and A. L. Lucius (2013). "E. coli ClpA Catalyzed Polypeptide Translocation is Allosterically Controlled by the Protease ClpP." Journal of Molecular Biology 425(15): 2795-2812.

Miller, J. M. and A. L. Lucius (2014). "ATP-gamma-S Competes with ATP for Binding at Domain 1 but not Domain 2 during ClpA Catalyzed Polypeptide Translocation." Biophys Chem 185: 58-69.

Oberhauser, A. F., P. K. Hansma, M. Carrion-Vazquez and J. M. Fernandez (2001). "Stepwise unfolding of titin under force-clamp atomic force microscopy." Proc Natl Acad Sci U S A 98(2): 468-472.

Rajendar, B. and A. L. Lucius (2010). "Molecular mechanism of polypeptide translocation catalyzed by the Escherichia coli ClpA protein translocase." J Mol Biol 399(5): 665-679.

Weibezahn, J., P. Tessarz, C. Schlieker, R. Zahn, Z. Maglica, S. Lee, H. Zentgraf, E. U. Weber-Ban, D. A. Dougan, F. T. Tsai, A. Mogk and B. Bukau (2004). "Thermotolerance requires refolding of aggregated proteins by substrate translocation through the central pore of ClpB." Cell 119(5): 653-665.

Author response:

The following is the authors’ response to the original reviews.

Thank you for your time and consideration on our submission. We also thank the reviewers for their consideration and helpful comments.  We have revised the introduction, results, and discussion sections of the revised manuscript in accordance with the reviewers’ suggestions, which have enhanced the clarity of our work. Specifically, we have clarified that the aim of the study is to report newly discovered sperm behaviours inside the uterus via high resolution deep tissue live imaging, and to stimulate further studies and discussion in the field of postcopulatory sexual selection in mice based on our observations. To the best of our knowledge, many of the specific sperm behaviours described in our manuscript are being reported for the first time, proven through direct observation inside the living reproductive tract.

We have also restructured our manuscript and moved our hypothetical interpretations based on our experimental observations to the discussion section. We hope that these revisions have clarified our claims and that our revised manuscript effectively communicates the importance of our findings and its values in prompting new questions and insight that encourage further studies. We believe that our work clearly demonstrates the importance of sperm/reproductive tract interaction, which cannot be adequately studied in artificial environments, and may become an important guideline for designing future experiments and studies.

Public Reviews: 

Reviewer #1 (Public Review): 

Summary: 

The authors want to determine the role of the sperm hook of the house mouse sperm in movement through the uterus. The authors are trying to distinguish between two hypotheses put forward by others on the role of the sperm hook: (1) the sperm cooperation hypothesis (the sperm hook helps to form sperm trains) vs (2) the migration hypothesis (that the sperm hook is needed for sperm movement through the uterus). They use transgenic lines with fluorescent labels to sperm proteins, and they cross these males to C57BL/6 females in pathogen-free conditions. They use 2-photon microscopy on ex vivo uteri within 3 hours of mating and the appearance of a copulation plug. There are a total of 10 post-mating uteri that were imaged with 3 different males. They provide 10 supplementary movies that form the basis for some of the quantitative analysis in the main body figures. Their data suggest that the role of the sperm hook is to facilitate movement along the uterine wall. 

We thank the reviewer for summarizing our work and the critical review of our paper. As summarized, the sperm hook has been primarily associated with the sperm cooperation (sperm hook) hypothesis and the migration hypothesis. However, we would like to emphasize that the aim of our work is not to cross check between the two hypotheses. Our aim was not to disprove either hypothesis, but rather to develop an experimental platform that enables detailed observation of sperm migration dynamics within the live reproductive tract. 

Through live imaging, we observed both the formation of sperm trains as well as interaction between the sperm and female reproductive tract epithelium. However, in our observations, we could not find advantage in terms of faster movement for the rarely observed sperm trains. While these events were infrequent in our experiments, we are not asserting that the sperm train hypothesis is invalid but rather reporting our observations as is. 

The main findings of our work lie in the newly observed dynamic behaviours of mouse sperm interacting with the female reproductive tract epithelium. Specifically, tapping and associated guided movement along the uterus wall, anchoring and related resistance to internal fluid flow and migration through the utero-tubal junction, and self-organized behaviour while clinging onto the colliculus tubarius. We have extensively revised the manuscript structure to clarify our findings.

Strengths: 

Ex vivo live imaging of fluorescently labeled sperm with 2-photon microscopy is a powerful tool for studying the behavior of sperm. 

Weaknesses: 

The paper is descriptive and the data are correlations. 

The data are not properly described in the figure legends. 

When statistical analyses are performed, the authors do not comment on the trend that sperm from the three males behave differently from each other. This weakens confidence in the results. For example, in Figure 1 the sperm from male 3613 (blue squares) look different from male 838 (red circles), but all of these data are considered together. The authors should comment on why sperm across males are considered together when the individual data points appear to be different across males. 

Thank you for your comments and suggestions. We have revisited all figure legends and made the necessary amendments (shown in the red-lined manuscript). Please note that, for a better flow of the paper, the previous Figure 1 has been changed to Figure 2 in the revised manuscript.

Regarding the analysis using different males, we would like to explain the statistics used. We used generalized linear mixed models to test the effect of the Angle and Distance to the wall on the migration kinetic parameters. The advantage of the generalized linear mixed models is that they consider individual variations in the data as an error term, thereby controlling such individual variations. 

There are two main factors contributing to individual variations. One is, as you pointed out, the difference in sperm from different males. However, we used genetically similar mice, so genetical variations must be minimal. Nonetheless, there must be individual differences that caused variations including age, stress level as well as body conditions. As these factors cannot be controlled, we used the mixed model approach where individual variations are grouped within the individual. This approach enabled us to test the effect of each explanatory variable (Angle and Distance) within an individual. 

The second factor that could cause variations is the female oestrous status. To avoid artifacts that could influence sperm behaviour, we did not use any invasive methods, such as hormone injections, to control or induce female oestrus. We controlled for this possible effect by including the mating date as a random effect. Since each female was used only once, the mating date reflects the variation caused by each female.

To provide further verification that the variation between individual males do not affect our results, we conducted analysis per individual male and mating dates (per each female). As clearly shown, sperm data points from individual males or female also show consistent clear correlations with the distance from the uterus wall. As pointed out, while the mean sperm speed could be different between individuals, they are not the topic we are interested in here. Our interest here is the effect of the distance between sperm and the uterine wall. Additionally, the variation between males is not always larger than those effect of the day (female), which in total suggest that integrating male variation is not essential. We have added this information to Supplementary Figure (Fig. S3) of the revised supplementary materials.

Moving forward, we can also consider the same analysis for the effects of the distance from wall on sperm SWR and LIN (linearity of forward progression) where no statistical significance was found. As see in the following figures, no statistically significant effect of the distance to wall on SWR and LIN are seen in that the regression lines drawn for each male and mating dates.

In summary, the statistical approach we used here has successfully reflected variations in sperm kinetics from different males as well as the variance from different females. We hope that our explanations and additional analysis answer your concerns. 

Movies S8-S10 are single data points and no statistical analyses are performed. Therefore, it is unclear how penetrant the sperm movements are. 

With respect to Movie S8, Figure 4A and B (Figure 5A and B in the current revised manuscript) depict the trajectories of accumulated spermatozoa (sperm trains) in the female uterus, as shown in Movie S8. We have added this information to the revised figure legend (L 293) for clarity. We could not observe sperm trains that moved faster than single sperms during over 100 hours of observation and collection of over 10TB of images. The three sperm trains presented in Fig. 5B were the sperm trains that moved in the head-forward direction. Most other identifiable trains, or clusters, did not move or could not move forward as their heads were entangled randomly. Although we of course agree that a statistical test for Movie S8 (also Fig. 5B) would be great, due to the small number of sperm trains we found, we could not perform meaningful statistical tests. Instead, we provided all data in the box plots in Fig. 5C so that readers can evaluate and understand our points. We believe that this is a more neutral way of presenting our data rather than providing statistical significance.

Regarding Movies S9 and S10, we are not entirely sure whether we understood your comments clearly. It would be very helpful if you could point out more specifically to the manuscript with line numbers as we would like to address your concerns and suggestions, and we believe that your input will improve our manuscript. We did not describe the penetration of sperm in these movies. Movies S9 and S10 are newly found sperm behaviours inside the UTJ and Isthmus. We observed that sperm beating is influenced by the width of luminal space as well as internal flow as see in Movies S9 and S10. As our animal model only expresses red fluorescence in the midpiece, accurate beating frequency measurement cannot be performed. However, we can clearly observe that beating is not continuous and almost results in a halt with respect to reproductive tract variations. We revised our description about the findings about beating speed changes in the revised manuscript (LL 305-335).  

Movies S1B - did the authors also track the movement of sperm located in the middle of the uterus (not close to the wall)? Without this measurement, they can't be certain that sperm close to the uterus wall travels faster. 

We revised the new Movie S1B to include videos that were used for the sperm migration kinetics analysis in Figure 2 (previously Figure 1). As you can see in the movies, the graph, and statistical analysis, there is a clear trend showing spermatozoa migration is slower as a function of distance from the uterus wall. Regarding your comment with respect to the middle of the uterus (not close to the wall), we have added another movie (Movie S1C) that was acquired at different depths from the wall (going towards the centre of the uterus). As clearly seen in Movie S1c, when imaging deeper into the uterus, there are an increasing number of inactive or slow-moving spermatozoa. Since the diameter of the uterus is easily over 2mm, we currently do not have optical access to exactly the centre of the uterus, but for all depths that are observable, spermatozoa near the wall were clearly faster.

Movie S5A - is of lower magnitude (200 um scale bar) while the others have 50 and 20 uM scale bars. Individual sperm movement can be observed in the 20 uM (Movie 5SC). If the authors went to prove that there is no upsucking movement of sperm by the uterine contractions, they need to provide a high magnification image. 

The main focus of video S5A, is the intramural UTJ where spermatozoa are located in rows within narrow luminal space (see Author response image 1). When there is up-suck like sperm passive carriage, there must be sperm movement from the uterus to intramural UTJ as in Author response image 1 left. However, there is no such sperm movement could be seen in our observations, as shown in Movie 5A. Importantly, as you can see in Movie 5A, indicated by an arrow from 5 sec to 6 sec, some spermatozoa are moving downward (see also Author response image 1 right). This is the opposite direction of movement with respect to possible up-suck like sperm carriage. 

Genetical evidence also support up-suck like passive sperm carriage is not the case for sperm migration from the uterus to UTJ. If environmental up-suck like passive transfer plays an important role, it is unlikely that genetically modified spermatozoa cannot pass the entrance of the intramural UTJ (Nakanishi et al., 2004, Biol. Reprod.; Li et al., 2013, J. Mol. Cell Biol.; Larasati et al., 2020, Biol. Reprod.; Qu et al., 2021, Protein Cell). 

Author response image 1.

The left image represents what is expected when up-suck like passive sperm carriage occurs. The right image represents what is actually experimentally observed in the intramural UTJ (see Movie S5A). The direction of the arrowheads indicates the direction of sperm movement.

Movie S8 - if the authors want to make the case that clustered sperm do not move faster than unclustered sperm, then they need to show Movie S8 at higher magnification. They also need to quantify these data. 

We understand your concern. As shown in Figure 5B, we included all sperm kinetics data of each sperm train and unlinked spermatozoon around the trains as individual dots. The only analysis we did not conduct was a statistical test with the data as it could be erroneous due to the large sample size difference (3 trains vs 181 unlinked spermatozoa). As the medians of the four sperm kinetic parameters are similar except SWR, we concluded that they are not necessarily faster than unlinked single spermatozoa. Since there is no known advantage to spermatozoa (including sperm trains) with intermediate moving speeds for sperm competition – for example in IVF, success fertilization rate is high when faster and active spermatozoa with normal shape are selected (Vaughan & Sakkas, 2019, Biol. Reprod.) – it is questionable whether there can be an advantage to the formation of sperm trains whose speed is not faster than unlinked spermatozoa in our data.

However, we do not agree with your comment regarding the need for higher magnification. Measurement of the sperm migration speeds (kinetic parameters) does not require measurement of exact tail movements in this study. Only sperm heads were tracked to measure their trajectory and such tracking was better done at low mag. For example, measuring the speed of a car does not need higher magnifications to visualize the rotation of the wheels. Additionally, including the effect of observation magnification on the sperm kinetic parameters for all 4 GLMM models for Figure 2 (Table S3) does not change the result, which shows that magnification is not a factor that influences our analysis. 

Movie S9C - what is the evidence that these sperm are dead or damaged? 

Thank you for your valid comment. We tracked sperm movements for at least 10 minutes and such entangled spermatozoa in the UTJ never became re-active. As you can see in the new Movie S9b, entangled spermatozoa were also acrosome re-acted (green acrosome head is gone) while active spermatozoa are responding to peristaltic movement by exhibiting movements within the same video. However, as you pointed out, we did not measure their viability with appropriate dyes. Although we also considered about extracting these spermatozoa and performing viability tests, we could not come up with a way to specifically extract the exact spermatozoa that were imaged. Considering your comments, we changed the term damaged or dead to inactive in the revised manuscript (LL 313-316, Legend Figure 6D. LL 380-384).

Movie S10 - both slow- and fast-moving sperm are seen throughout the course of the movie, which does not support the authors' conclusion that sperm tails beat faster over time. 

There must have been a misunderstanding. We did not indicate that sperm beating got faster over time anywhere in the main manuscript, including the figure legend and related movie captions. As correctly pointed out, the sperm beating speed changes over time (not getting faster over time) and shows a correlation with internal fluid flow and width of luminal space (LL 320-332). Please let us know if you meant something else. 

Reviewer #2 (Public Review): 

Summary: 

The specific objective of this study was to determine the role of the large apical hook on the head of mouse sperm (Mus musculus) in sperm migration through the female reproductive tract. The authors used a custom-built two-photon microscope system to obtain digital videos of sperm moving within the female reproductive tract. They used sperm from genetically modified male mice that produce fluorescence in the sperm head and flagellar midpiece to enable visualization of sperm moving within the tract. Based on various observations, the authors concluded that the hook serves to facilitate sperm migration by hooking sperm onto the lining of the female reproductive tract, rather than by hooking sperm together to form a sperm train that would move them more quickly through the tract. The images and videos are excellent and inspirational to researchers in the field of mammalian sperm migration, but interpretations of the behaviors are highly speculative and not supported by controlled experimentation. 

Thank you for your critical review and valuable comments on our manuscript. As pointed out, some of our findings and suggestions were largely observation based. However, to the best of our knowledge, many of our observations are novel, particularly in the context of live imaging inside the female uterus and reproductive tract. We believe these observations open doors to many questions and follow up studies that can be envisioned based on our findings, which is what drives science forward. 

That being said, we entirely agree that many follow up experiments need to be designed and performed, especially to validate the exact molecular mechanisms of the observed dynamics. We acknowledge that it is unfortunate we currently lack the proper molecular experimental toolsets to perform further tests. We have removed much of the hypothetical discussions from the results section and moved them to the discussion section. We hope that our revision more clearly defines the observed experimental data and our interpretations.

Strengths: 

The microscope system developed by the authors could be of interest to others investigating sperm migration. 

The new behaviors shown in the images and videos could be of interest to others in the field, in terms of stimulating the development of new hypotheses to investigate. 

Weaknesses: 

The authors stated several hypotheses about the functions of the sperm behaviors they saw, but the hypotheses were not clearly stated or tested experimentally. 

The hypothesis statements were weakened by the use of hedge words, such as "may". 

We appreciate your helpful comments and have revised our hypotheses and suggestions accordingly. We have removed instances of “may” or revised it to be more direct. We have also moved most of our interpretations and hypotheses from the results to the discussion section. 

It is important to note that experimental approaches to test what we suggested from our findings in the current ex-vivo observation platform are not trivial and require extensive investigation of several unknown factors of the female reproductive tract. For instance, obtaining detailed information on the chemical characteristics and fluid dynamics in the female reproductive tract is essential to build a microfluidic channel that accurately resembles the uterus and oviduct, replicating what we found in an extracted living entire organ. This poses a significant challenge and requires collaborative expertise from many labs, which we hope to build in the near future. 

Furthermore, our biggest concern is that, even if we were to construct the appropriate microfluidic channel to test sperm migration, it is very likely that the sperm behaviours that we observed under natural conditions may not be replicated in artificial environments. This raises questions about whether in-silico or in-vitro findings can truly resemble what we reported here using the ex-vivo observation inside a living organ.

To share our experience related to this difficulty, at the initial stage of our study, we attempted sperm injection combined with fluorescent beads to visualize the fluid flow, as well as dyeing the female reproductive tract and spermatozoa after mating. However, none of these resulted in meaningful results. Another potential approach to perform similar research regarding our claims is using genetical engineering to indirectly confirm the influence of the sperm hook morphology on sperm behaviour. However, such an approach lacks a mechanical demonstration about how the sperm hook interacts with the female reproductive tract. 

It is unfortunate that the sperm behaviours that we found and reported here are considered as highly speculative. The main findings of our work lie in the newly observed dynamic behaviours of mouse sperm interacting with the female reproductive tract epithelium. Specifically, these behaviours include tapping and associated guided movement along the uterus wall, anchoring and related resistance to internal fluid flow and migration through the utero-tubal junction, and self-organized behaviour while clinging onto the colliculus tubarius. 

We have extensively revised the manuscript structure to clarify our findings and integrated our points in the introduction. Although we understand our following hypotheses may be considered speculative and the causative relationship between the sperm hook and its role in sperm migration requires further experimental approaches, we believe that the image-based observation of dynamic behaviours of spermatozoa are solid. We believe our findings will facilitate further studies and discussion in the field of studies on postcopulatory sexual selection in rodents.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors): 

The manuscript is written for an expert in a fairly small field. I recommend that the authors rewrite the manuscript to make it more accessible to people outside of the field. These suggestions include 

(1) Provide a diagram of the female reproductive tract in Figure 1. 

a. Indicate where sperm enter the tract and the location of the oocyte they are trying to reach. 

b. Label all areas of the uterus that are mentioned in this study and be consistent about the label. 

(2) All movies should have a diagram of the location of the uterus that is being imaged. 

Thank you for the great suggestion. We have added a diagram of the female reproductive tract in the revised Figure 1A. In response to your comments 1a and b, we have indicated such information by including eggs in the ampulla and arrows that indicate sperm migration direction. We have also labelled the name of the specific areas that were studied in the manuscript.

We are unsure how to integrate the diagram in all movies without reframing the videos, which could cause serious corruption of the files. More importantly, we think that adding the same diagram to all movies may complicate the visuals and disrupt indications and subject in the movie. Instead, we have referred to the common diagram (Figure 1A) in each movie caption, specifying where the video was taken. Thank you for the suggestion. With this information, we hope readers can now more easily understand where we made the observations. 

(3) The major questions in the field need to be better described in the introduction. 

Thank you for your valuable suggestions and specific comments which have greatly helped improve our manuscript. We have revised our introduction and discussion sections by adding more literature reviews and integrating studies across a wider range of the postcopulatory sexual selection, as per your suggestion (LL 34-57, LL 385-398).

(4) The major question that the authors are trying to address should be described in the introduction. 

Thank you for the helpful suggestion. We have clarified in the introduction that our aim was to contribute to the field of postcopulatory sexual selection in rodents by advancing methodological progress and to stimulate discussion and future research on the function of the sperm hook in murine rodents (LL 76-94) based on our observations.

(5) A discussion of the sperm hook should be provided. How many species have this structure (or similar structure)? 

We have integrated your point into the revised discussion section. Essentially, most murine rodent species have sperm hooks (while their exact shapes differ). However, as there are over 500 species and not all of them have been tested, we do not know exactly how many of them have this structure. Therefore, we included paper references that examined species variations in sperm hook characteristics and their possible correlation with sperm competition (LL 385417) in the discussion. Additionally, we also included papers by Breed (2004) and by Roldan et al (1992) that investigated murine rodents with a sperm hook in the introduction section as well (LL 58-61).  

(6) The figure legends must describe everything in the figure or movie. 

Thank you for the helpful suggestion. We previously thought that our figure legends may be too long. We have included further information in the figure legends and movie captions. We have also revised the movies by adding some clips following our revision (Movie S1).

Reviewer #2 (Recommendations For The Authors): 

Here are some specific concerns I had about the clarity of approach to experiments and interpretations of results. 

In the Introduction, the authors stated that the study was intended to determine the function of the hooks on the mouse sperm heads. However, in the Results section, the authors did not explain the rationale for the first set of experiments with respect to the overall objective of the study. In this experiment, the authors measured the velocities of sperm swimming in the uterus and found that the sperm moved faster when closer to the uterine wall (VCL, VSL). They concluded that migration along the uterine wall "may" be an efficient strategy for reaching the entrance to the uterotubal junction (UTJ) and did not explain how this related to the function of the hooks. 

Thank you for your critical comment and guidance. We have changed the order of Figure 1 and Figure 2 and revised the result section to integrate your points. At the initial stage of the study, we expected to find evidence of the function of sperm trains in aiding sperm migration in the female uterus (which has not been observed in the live uterus; previous works were done invitro with extracted sperm from epididymis or uterus after mating). However, what we found was something unexpected: dynamic sperm hook related movements facilitating sperm migration inside the female uterus by playing a mechanical role in sperm interaction with the uterine wall. These results that were presented in the previous Figure 2 has been reorganized as the new Figure 1.

Based on this observation, our research later moved to clarify whether such sperm-epithelium interaction indeed helps sperm migration. This led us to measure sperm kinetics in relation to their distance and angle to the uterine wall. We have revised our introduction and result parts by integrating these points. We hope that our revision will answer your questions. We have also reduced the use of ‘may’ or ‘can’ in the results section. In the revised manuscript, we have moved such hypotheses to the discussion section and focused on what we observed in the results section.

The authors proposed that the sperm hook "may" play a crucial role in determining the direction of migration. When sperm encountered a uterine wall, significantly more changed migration direction toward the pro-hook direction than toward the anti-hook direction. In Figure 2B, sperm behavior is not visually understandable nor clearly explained. 

Thank you for the helpful comments. We have removed “may” and “might” to make our claim clearer and more concise. We have also revised the previous Figure 2B by combining it with the previous Figure 2C (they have been combined into Figure 1C now). We have also revised Figure 1B by increasing the line thickness of the sperm trajectory of the pro-wall-hook direction and added the anti-wall-hook trajectory. We hope that these revisions make the figure easier to understand.

In Figure 2E, are the authors showing that the tip of the hook is caught between two epithelial cells? Please clarify the meaning of this figure. 

Please clarify the difference between "tapping" and "anchoring". 

Thank you for the detailed comments. As you pointed out, we currently have no evidence whether sperm can be caught in epithelia inter-cellular gaps. We have revised this source of confusion by removing the gap in the revised figure (Figure 1E). We have also included the definition of anchoring (LL 142-143) and tapping (LL 128-130). Anchoring facilitates the attachment of sperm to the uterine epithelia. Such anchoring also involves the catching of the sperm head in the inter-mucosal fold or gap, particularly at the entrance of the intramural UTJ at the end of the uterus. Tapping is the interaction between the head hook and epithelia in which the sperm hook is tapping (or patting) on the surface. Sperm tapping can be a byproduct that results from flagella beating when spermatozoa migrate toward the pro-wall-hook direction along the uterine wall (epithelia) or can play some role in sperm migration. As we currently cannot draw a conclusion, we did not integrate the possible function of the tapping in the manuscript.

The authors proposed that opposite sliding of neighboring mucosal folds lining the UTJ would cause small openings to form, through which only perhaps one sperm at a time could enter and pass through the UTJ into the uterus. This hypothesis was not actually tested. 

Imaging inside deep tissue is challenging due to light scattering as it penetrates through biological tissue. While this is also true for the uterus, the intramural UTJ is especially difficult to image because the UTJ consists of several thick muscle and cell layers (see Movie S5A). Another challenge is that the peristaltic movement of the UTJ results in constant movement, making continuous tracking of single sperms while passing through the entirety of the UTJ impossible in our current experiments. We have moved this hypothesis to the discussion section and restated that this is a pure hypothetical model (LL 399-406). We hope that our model encourages the community in designing or establishing an improved ex-vivo observation system that may be able to test this hypothetical model in the near future.

Next, the authors hypothesized that sperm that encounter the small openings in the UTJ may then be guided onward and the hooks could prevent backward slipping. This was also not tested. 

As you’ve noted, the function of the sperm hook that aids in sliding and preventing backward slipping could not be tested directly in our ex-vivo observation platform that relies on natural movement of the living organ. However, we believe that these limitations also highlight the importance of continued research and the development of more advanced methodologies in this field.

We would also like to note that we provide direct observations of spermatozoa resisting internal flow due to reproductive tract contractions in Movie S3A, B as well as Movie S5B. We referred to these movies and pointed out the role of anchoring (sperm attachment) in preventing sperm from being squeezing out (LL 140-149, LL 224-241). Unfortunately, we cannot conceive of how this behaviour can be tested additionally in any uterus-resembling microfluidic device or ex-vivo systems. In line with your suggestion, we have rewritten the related result section and moved our related discussions in the result part to the discussion section (LL 224-241, LL 399-417). 

The authors observed that large numbers of uterine sperm are attached to the entrance of the UTJ. Some sperm clustered and synchronized their flagellar beating. The authors speculated that this behavior served to push sperm in clusters onward through the UTJ. 

We would like to note that we did not speculate that sperm clustering and their synchronization could serve to push spermatozoa in a cluster to move onward through the UTJ. We only pointed out our observation in recorded videos, that generative flow from the clustered spermatozoa pushed away other spermatozoa as seen in Movie S7 (LL 261-264). Although such sperm cooperation is possible (blocking passage of later sperm), we cannot draw that conclusion from our observation. The possibility you pointed out (pushing sperm onward through the UTJ) was suggested by Qu et al in 2021 [Cooperation-based sperm clusters mediate sperm oviduct entry and fertilization, Protein & Cell] based on their observations on cleared dead reproductive tracts.

The authors found only a few sperm trains in the uterus, UTJ, and oviduct, so they could not measure sufficient numbers of samples to test whether sperm trains swim faster than single sperm. Without sufficient data, they concluded that the "sperm trains did not move faster than unlinked single spermatozoa." 

We would like to take this opportunity to clarify our claims. We do not claim that our current experiments can give the final verdict on whether the sperm train hypothesis for faster swimming is correct or not. The phrase “sperm trains did not move faster” was not intended to mean that the sperm train hypothesis is invalid.  We did not draw a conclusion but dryly described the experimental data that we observed (LL 279-286).  We would once again like to emphasize that the main claim of our manuscript is not to rule out the sperm train hypothesis, but to present the various dynamic interactions of the sperm head with the female reproductive tract. To make the statement more balanced, we revised the sentence as “observed sperm trains did not move faster or slower than unlinked single spermatozoa” (LL 281-282).

The authors hypothesized that the dense sperm clusters at the entrance into the UTJ could prevent the rival's sperm from entering the UTJ (due to plugging entrance and/or creating an outward flow to sweep back the rival's sperm), but they did not test it. 

We agree that we were not able to test such possible function of the sperm cluster at UTJ entrance. Following your concerns, we revised the result part (LL 256-264) by removing most of our discussions related to the observed phenomena. We also integrated some interpretation rather to the discussion section (LL 421-437) and suggested that future works using appropriate microfluidic channel designs or sequential double mating experiments may be performed for additional tests (LL 443-447). However, we would like to point out that Movie S7C clearly shows surrounding sperms that are swept away from the sperm clusters. Since the sperm density is high, this is almost equivalent to a particle image velocimetry experiment, and we can clearly see the effect of the outward flow generated by the sperm clusters.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

This study provides a useful strategy for treating mouse cutaneous squamous cell carcinoma (mCSCC) with serum derived from mCSCC-exposed mice. The exploration of serum-derived antibodies as a potential therapy for curing cancer is particularly promising but the study provides inadequate evidence for specific effects of mCSCC-binding serum antibodies. This study will be of interest to scientists seeking a novel immunotherapic strategy in cancer therapy.

Joint Public Review:

Summary:

This study presents an immunotherapeutic strategy for treating mouse cutaneous squamous cell carcinoma (mCSCC) using serum from mice inoculated with mCSCC. The author hypothesizes that antibodies in the generated serum could aid the immune system in tumor volume reduction. The study results showed a reduction in tumor volume and altered expression of several cancer markers (p53, Bcl-xL, NF-κB, Bax) suggesting the potential effectiveness of this approach.

Strengths:

The approach shows potential effect on preventing tumor progression, from both the tumor size and the cancer biomarker expression levels bringing attention to the potential role of antibodies and B cell responses in cancer therapy.

We greatly appreciate your positive feedback on our study.

Weaknesses:

These are some of the specific things that the author could consider to strengthen the evidence supporting the claims in their study.

(1) The study fails to provide evidence of the specific effect of mCSCC-antibodies on mCSCC. The study utilized serum which also contains many immune response factors like cytokines that could contribute to tumor reduction. There is no information on serum centrifugation conditions, which makes it unclear whether immune components like antigen-specific T cells, activated NK cells, or other immune cells were removed from the serum. The study does not provide evidence of neutralizing antibodies through isolation, analysis of B cell responses, or efficacy testing against specific cancer epitopes. To affirm the specific antibodies' role in the observed immune response, isolating antibodies rather than employing whole serum could provide more conclusive evidence. Purifying the serum to isolate mCSCC-binding antibodies, such as through protein A purification, and ELISA would have been more useful to quantify the immune response. It would be interesting to investigate the types of epitopes targeted following direct tumor cell injection. A more thorough characterization of the antibodies, including B cell isolation and/or hybridoma techniques, would strengthen the claim.

I am deeply appreciative of the reviewer's highly professional comments. Tumor development involves the coexistence of cancer cells at different developmental stages, each harboring a variety of known and unknown mutated proteins. These mutated proteins expose multiple known and unknown epitopes, each capable of stimulating the production of corresponding antibodies in healthy mice. Identifying all these antibodies presents a significant challenge. Current research methodologies, such as ELISA, WB, and ChIP, can only identify known antibodies based on existing antigens. A prerequisite for using these techniques is that both antigens and antibodies are identified. At present, there is no technology available to identify antibodies produced by an unknown mutated protein and epitope. However, I find the reviewer's comments insightful. Perhaps we can initially identify some known mCSCC-antibodies on mCSCC. However, studying the specific effect of these known mCSCC-antibodies on mCSCC is uncertain because we believe that tumor shrinkage results from the combined action of both known and unknown antibodies.

We concur with the reviewer's observations regarding the use of serum, which is rich in immune response factors such as cytokines that could potentially contribute to tumor reduction. In our future research, we plan to systematically analyze the individual roles of these antibodies and cytokines in tumor reduction. In 1973, Nature published a report indicating that serum demonstrated promising results in tumor treatment (Immunotherapy of Cancer with Antibody in Rats. Nature 243, 492 (1973). https://doi.org/10.1038/243492b0). Since then, there have been scarcely any reports on serum therapy for tumors. The primary focus of our study is to evaluate the efficacy of serum therapy in treating tumors. We hypothesize that antibodies and cytokines form a complex interactive network, working in synergy to reduce tumors. Consequently, we believe that studying these antibodies and cytokines in isolation may not yield effective results.

In this study, the methodology section outlines the process of serum preparation. It is important to note that serum is devoid of blood cells. I hypothesized that whole blood might have superior therapeutic effects compared to serum. This is because antibodies could potentially synergize with immune cells (including T cells, B cells, and NK cells), thereby enhancing the effectiveness of the treatment. As previously discussed, these antibodies, cytokines, and immune cells form a complex interactive network aimed at tumor reduction. Consequently, there are numerous factors that could influence the experimental outcomes, which presents a challenge for analyzing the results. Furthermore, the implementation of whole blood transfusion therapy introduces additional considerations, such as potential side effects and reactions associated with blood transfusions.

We thank the reviewers for their suggestion to purify the serum in order to isolate mCSCC-binding antibodies. As we previously mentioned, separating a large number of both known and unknown serum antibodies presents a significant technical challenge. We are eager to discuss and consider suggestions from the reviewers regarding methods to identify a large variety and number of unknown antibodies on cells. Perhaps, as the reviewer suggested, we could begin with known antibodies and employ Protein A purification technology to purify these antibodies and subsequently detect immune responses. We could also categorize the types of epitopes targeted, direct tumor cell injection, to study the epitopes of these types in further studies. The suggestion to study the response of B cells is valuable, and we plan to conduct comprehensive research on the response and status of B cells in our future studies.  

The purification of antibodies to enhance the specificity of their effectiveness against tumors is a critical aspect of our study. However, we would like to address some concerns raised. (1) The separation of all antibodies and cytokines presents a significant technical challenge. Particularly, there is a risk of overlooking antibodies that are present in low concentrations but play crucial roles. (2) What concerns us is that studying the composition separately would lose the overall effectiveness of the study. Our primary concern is that studying these components in isolation could compromise the holistic understanding of the study. This is akin to current research on traditional medicine, where the separation and individual study of compounds often result in a loss of overall therapeutic efficacy. For instance, consider a scenario where 100 antibodies collectively work to shrink a tumor. These antibodies interact with 20 cytokines, forming a complex network that enhances the cytokines' activity against tumor cells. Furthermore, many important antibodies and cytokines are currently unknown. Studying these antibodies in isolation could potentially result in the loss of this therapeutic effect. Therefore, in the discussion section, we have emphasized that our study considers a tumor mass, including tumor cells at various stages of development, as a single entity. As a practicing clinician, my primary focus is on the therapeutic outcomes in tumor treatments, despite the mechanisms of serum therapy remaining largely elusive, liking a black box.

(2) In the study design, the control group does not account for the potential immunostimulatory effects of serum injection itself. A better control would be tumor-bearing mice receiving serum from healthy non-mCSCC-exposed mice. Additionally, employing a completely random process for allocating the treatment groups would be preferable. Also, the study does not explain why intravenous injection of tumor cells would produce superior antibodies compared to those naturally generated in mCSCC-bearing mice.

I concur with the reviewer's perspective that using serum from healthy, non-mCSCC exposed mice as a control could potentially improve our study. Initially, our primary concern was to minimize harm to the mice and avoid excessive blood reactions, which led us to exclude the use of serum from healthy, non-mCSCC exposed mice in our control group. The main objective of our study was to investigate tumor shrinkage through serum treatment, specifically serum-derived antibodies. We anticipated that tumor-bearing mice receiving serum from healthy, non-mCSCC exposed mice would exhibit a response to the injected serum, which would manifest as a blood reaction. However, we did not expect this to result in a tumor treatment effect. If it turns out that normal serum (from healthy, non-mCSCC-exposed mice) possesses tumor-reducing properties, it would indeed be a novel discovery. We appreciate the reviewer's insightful suggestion and will consider incorporating it into our future research.

We concur with the reviewer's observations that the use of a completely random process for assigning treatment groups would be more desirable. Indeed, the complete randomization of the entire process further underscores the efficacy and universality of serum therapy. In this study, we utilized paired mice to mitigate the risk of cross-infection and adverse reactions associated with blood transfusions. We deeply value the reviewer's expert feedback.  

Lastly, the reason why tumor cells, when intravenously injected, produce antibodies superior to those naturally generated in mCSCC-bearing mice, is due to the following reasons. As tumor cells grow, they produce a variety of mutated proteins to adapt to the immune microenvironment and evade the immune system of mCSCC-bearing mice. However, these tumor cells with mutated proteins are exceptionally sensitive and recognizable to healthy mice. This recognition triggers an immune response in healthy mice, leading to the production of specific therapeutic antibodies. This simultaneous production of diverse and abundant antibodies is only achievable by living organisms.

(3) In Figure 2B, it would be more helpful if the author could provide raw data/figures of the tumor than just the bar graph. Similarly in Figure 3, the author should show individual data points in addition to the error bar to visualize the actual distribution.

Raw data (numerical values) have been incorporated into Figures 2B and 3, but the data is placed in the table below the graph. If placed above the error bar, it requires a small font and may not be clear.

(4) The author mentioned that different stages of tumor cells have different surface biomarkers. Therefore, experimenting with injecting tumor cells at various stages could reveal the most immunogenic stage. Such an approach would allow for a comparative analysis of immune responses elicited by tumor cells at different stages of development.

Yes, throughout the course of tumor development, tumor cells at various stages will exhibit distinct markers or possess different mutated proteins. The concept of segregating tumor cells from different stages and independently comparing their immune responses is indeed commendable. Future research could involve isolating cells that express identical biomarkers at each stage for a comparative analysis of the immune responses triggered by the tumor cells. However, this approach diverges from the original intent of this study.

Most tumor cells exist within the same developmental stage. However, this does not imply that all tumor cells within the tumor mass are at the same stage. For instance, a stage III liver cancer tumor may contain both stage I and stage IV tumor cells. Moreover, due to the complexity of tumor development, not all tumor cell surface markers are identical, even for tumors at the same stage. For instance, 20 major proteins and 100 minor proteins are implicated in tumor formation. In fact, random mutations in just 5 of these major proteins and 10 minor proteins can instigate the development of tumors. This implies that the protein pattern (tumor cell surface markers) associated with each individual's tumor is unique. While studying tumor cells at different stages separately allows for the observation of the immune response of tumor cells at each stage, it lacks a comprehensive research and treatment effect. For this reason, the design of this study treats a tumor mass as a whole, encompassing both the primary stage tumor cells and those not in that stage. These tumor cells are then injected to produce corresponding therapeutic antibodies. Furthermore, if tumor cells from only one stage are isolated and specific antibodies are produced against these cells, it could lead to immune escape of tumor cells at other stages, preventing the tumor from shrinking. Therefore, our approach aims to address this issue by considering the tumor mass as a whole.

(5) In the abstract the author mentioned that using mCSCC is a proof-of-concept for this potential cancer treatment strategy. The discussion session should extend to how this strategy might apply to other cancer types beyond carcinoma.

We have incorporated an additional paragraph in the discussion section where we delve into the concepts and experimental principles underpinning this study. This, we believe, addresses the reviewer's query regarding the applicability of our study's methodology to other types of tumors. The process for other tumors also involves isolating cells from the tumor, stimulating therapeutic antibody production in healthy mice using these cells, and ultimately reintroducing these antibodies into mice with tumors to facilitate tumor elimination

Recommendations For The Authors:

The author is encouraged to refine the study's design in future studies considering the weaknesses highlighted above, summarize the results more effectively, and seek opportunities to expand on this promising idea and enhance the research's impact and applicability.

We greatly appreciate the valuable suggestions provided by the editor and reviewers. These insights will certainly be addressed in our future research endeavors.

Suggestions for title modification:

Following the scope of the study, the term 'specific homologous neutralizing-antibodies' may be misleading as neutralizing antibodies typically refer to antibodies preventing viral cell entry. In cancer therapy, 'neutralization' is not a relevant concept, as cancer cells do not infect host cells. Using whole tumor cells as immunogens diverges from the specificity of traditional vaccination approaches that utilize well-defined proteins or antigens. Furthermore, the term "homologous" suggests a precision in targeting that is not demonstrated by reintroducing serum without isolating its specific components. Therapeutic effects should not be attributed to "neutralizing antibodies" without isolating or characterizing the antibody response or verifying their efficacy against specific cancer epitopes. Additionally, it is suggested that you indicate the biological system that your study utilised in the title. More so, this approach is not entirely novel, as seen with the use of adjuvants in some flu vaccines, or in Moderna's cancer vaccine mRNA-4157, which encodes up to 34 patient-specific tumor neoantigens. You can consider the title below or a variant of the same.

Suggested title: Generating serum-based antibodies from tumor-exposed mice: a potential strategy in cutaneous squamous cell carcinoma treatment

I concur with your suggestion and have modified the title to " Generating serum-based antibodies from tumor-exposed mice: a new potential strategy for cutaneous squamous cell carcinoma treatment ". I believe this research remains some new, hence the addition of the word "new". Furthermore, the term "novel" in the paper has been either removed or substituted.

Moreover, I propose that this study shares similarities with Moderna's cancer vaccine mRNA-415, albeit with certain differences. Moderna's cancer vaccine mRNA-415 encodes 34 recognized neoantigens to stimulate an immune response by eliciting specific T cell responses. This is similar to the strategy of some companies developing a protein set for diagnosing lung cancer, liver cancer, among others. Without a doubt, these methods have improved the effectiveness of tumor diagnosis and treatment. However, I think that these methods currently face challenges in completely eradicating tumors because they perceive tumors as a static process and cells that express certain mutated proteins in a fixed manner. I believe that small molecule antibodies, cytokines, and immune cells present in serum that are difficult to detect, have low concentrations, or are unknown are essential for maintaining the expression of important mutant proteins and the escape of tumor cells. This is also the primary reason why tumors are difficult to treat and prone to recurrence at present.

From my perspective, different tumors, as well as different stages of the same tumor, express varying mutated proteins or surface markers. Targeting some may result in others escaping or even creating a more conducive growth environment for those that do escape. Our study adopts a comprehensive view of a tumor block, encompassing tumor cells at different stages and tumor cells at the same stage but expressing different biomarkers. This approach generates a multitude of known and unknown antibodies that work in concert with cytokines and immune cells. While our method may not be capable of generating all mutated proteins and epitope antibodies due to the weakness of some antigens (epitopes of mutated proteins), it can still be effective. As long as the number of tumor cells is reduced below a certain threshold following multiple rounds of treatment with various antibodies produced at different stages, these cancer cells can be eradicated by the body's immune system. This is a process that is real-time and dynamic. Undoubtedly, if it becomes evident that alterations in a set of proteins can bolster the immune system and eradicate tumor cells, then the implications are significant. The immunotherapy proteins, which have demonstrated positive therapeutic effects, developed by certain companies are also predicated on this very principle.

Finally, I greatly appreciate your suggestions, which will be considered and gradually addressed in future research.

Author response:

Public Reviews:

Reviewer #1 (Public Review): 

By mapping H3K4me2 in mouse oocytes and pre-implantation embryos, the authors aim to elucidate how this histone modification is erased and re-established during the parental-to-zygotic transition, as well as how the reprogramming of H3K4me2 regulates gene expression and facilitates zygotic genome activation.

Employing an improved CUT&RUN approach, the authors successfully generated H3K4me2 profiling data from a limited number of embryos. While the profiling experiments are very well executed, several weaknesses, particularly in data analysis, are apparent:

(1) The study emphasizes H3K4me2, which often serves as a precursor to H3K4me3, a well-studied modification during early development. Analyzing the new H3K4me2 dataset alongside published H3K4me3 data is crucial for comprehensively understanding epigenetic reprogramming post-fertilization and the interplay between histone modifications. However, the current analysis is preliminary and lacks depth.

Thank you very much for your valuable suggestions. The data of histone H3K4me3 in humans and mice has been published，and our previous data revealed the unique pattern of H3K4me3 during early human embryos and oocytes (Xia et al., 2019). So, this study mainly focuses on the localization of H3K4me2 in mouse oocytes and preimplantation embryos, how it is erased and re-established during mammalian parental-to-zygote transition, and its function. The combined analysis of H3K4me2 and H3K4me3 is not our main work, but it is not ruled out that there may be new discoveries between these two histones. Previously, our data tended to show that the H3K4me2 not only acts as a precursor of H3K4me3, but also plays its role independently.

(2) Tranylcypromine (TCP) is known as an irreversible inhibitor of monoamine oxidase and LSD1. While the authors suggest TCP inhibits the expression of LSD2, this assertion is questionable. Given TCP's potential non-specific effects in cells, conclusions related to the experiments using TCP should be made with caution.

Thank you for pointing this out, and we thank the reviewer again for the important suggestion. We found that the previous study indicated that TCP was a non-reversible inhibitor of LSD1 and LSD2, but according to our data, the content of LSD1 was very low in the early stages of mouse embryos, which mainly inhibited the function of LSD2. (Binda et al., 2010; Fang et al., 2010 )

(3) Some batches of H3K4me2 antibody are known to cross-react with H3K4me3. Has the H3K4me2 antibody used in CUT&RUN been tested for such cross-reactivity? Heatmaps in the figures indeed show similar distribution for H3K4me2 and H3K4me3, further raising concerns about antibody specificity.

We thank the reviewer for the insightful comments. The H3K4me2 antibody was purchased from Millipore (cat. 07030). Figure 2A shows the specific enrichment area of H3K4me2 in promoter and distal region. Some batches of H3K4me2 antibody are known to cross-react with H3K4me3, but the H3K4me2 antibody we used in our CUT&RUN seems to have Low cross-reactivity.

(4) Certain statements lack supporting references or figures (examples on page 9 can be found on line 245, line 254, and line 258).

Thank you for pointing this out, and we will add references to support the statement in the paper as suggested.

(5) Extensive language editing is recommended to clarify ambiguous sentences. Additionally, caution should be taken to avoid overstatement - most analyses in this study only suggest correlation rather than causality.

Thank you for your kind comments. We will revise the expression in the manuscript later.

Reviewer #2 (Public Review):

Chong Wang et al. investigated the role of H3K4me2 during the reprogramming processes in mouse preimplantation embryos. The authors show that H3K4me2 is erased from GV to MII oocytes and re-established in the late 2-cell stage by performing Cut & Run H3K4me2 and immunofluorescence staining. Erasure and re-establishment of H3K4me2 have not been studied well, and profiling of H3K4me2 in germ cells and preimplantation embryos is valuable to understanding the reprogramming process and epigenetic inheritance.

(1) The authors claim that the Cut & Run worked for MII oocytes, zygotes, and the 2-cell embryos. However, it is unclear if H3K4me2 is erased during the stage or if the Cut & Run did not work for these samples. To support the hypothesis of the erasure of H3K4me2, the authors conducted immunofluorescence staining, and H3k4me2 was undetected in the MII oocyte, PN5, and 2-cell stage. However, the published papers showed strong staining of H3K4me2 at the zygote stage and 2-cell stage ((Ancelin et al., 2016; Shao et al., 2014)). The authors need to cite these papers and discuss the contradictory findings.

The authors used 165 MII oocytes and 190 GV oocytes for the Cut & Run. The amount of DNA in MII oocytes is halved because of the emission of the first polar body. Would it be a reason that H3K4me2 has fewer H3K4me2 peaks in MII oocytes than GV oocytes?

First of all, thank you for your valuable advice. The published papers showed strong staining of H3K4me2 at the zygote stage and 2-cell stage, which is interesting. I think we may have used different parameters in the confocal laser shooting process(Ancelin et al., 2016). We used the same parameter to continuously shoot the blastocyst stage from the GV stage. If we only shot the fertilized egg and the 2-cell stage, I think we may also see weak fluorescence at the 2-cell stage under different parameters. We will refer to this reference and discuss it in the resubmitted version.

Moreover, you mentioned the H3K4me2 has fewer H3K4me2 peaks in MII oocytes than GV oocytes, because the MII expelled the polar body. There is no problem with this logic. However, the first polar body expelled from the MII stage is still in the zona pellucida, and we also collected the polar body in the CUT&RUN experiment; Therefore, compared to GV, the DNA content of MII samples is not halved. After further discussion, we believe that the reduction of H3K4me2 peaks in MII stage compared with GV stage may be closely related to oocyte maturation. It is the specific modification of histones in different forms at different times that affects the chromatin structure change appropriately with the different stages of meiosis. At present, it has been confirmed that H3K4me3 gradually decreases from GV to MII stage during the maturation of human oocytes. H3K27me3 did not change from GV to MII stage.

In Figure 3C, 98% (13,183/13,428) of H3K4me2 marked genes in GV oocytes overlap with those in the 4-cell stage. Furthermore, 92% (14,049/15,112) of H3K4me2 marked genes in sperm overlap with those in the 4-cell stage. Therefore, most regions maintain germ line-derived H3K4me2 in the 4-cell stage. The authors need to clarify which regions of germ line-derived H3K4me2 are maintained or erased in preimplantation embryos. Additionally, it would be interesting to investigate which regions show the parental allele-specific H3K4me2 in preimplantation embryos since the authors used hybrid preimplantation embryos (B6 x DBA).

Thank you very much for your suggestion. Further analysis of which regions show the parental allele-specific H3K4me2 in preimplantation embryos will make the study more interesting. We will discuss this in depth in resubmitted vision.

(2) The authors claim that Kdm1a is rarely expressed during mouse embryonic development (Figure 4A). However, the published paper showed that KDM1a is present in the zygote and 2-cell stage using immunostaining and western blotting ((Ancelin et al., 2016)). Additionally, this paper showed that depletion of maternal KDM1A protein results in developmental arrest at the two-cell stage, and therefore, KDM1a is functionally important in early development. The authors should have cited the paper and described the role of KDM1a in early embryos.

In the analysis of this experiment, we believe that in the early embryonic development of mice, the expression of KDM1A is lower than that of KDM1B, which is relative. Similarly, the transcriptome data we cite also show that KDM1A is expressed at elevated levels during oocyte maturation and fertilization compared to immature oocytes. In addition, the effects of loss of maternal KDM1a on embryonic development were not discussed. We believe that the absence of maternal KDM1b blocks embryonic development, and we will cite and discus the references later.

(3) The authors used the published RNA data set and interpreted that KDM1B (LSD2) was highly expressed at the MII stage (Figure S3A). However, the heat map shows that KDM1B expression is high in growing oocytes but not at 8w_oocytes and MII oocytes. The authors need to interpret the data accurately.

After re-checking the data, we found that there was a problem with the normalization method of our heat map, and we will re-make the heatmap and submit it in the modified version. With reference to Figure 4A, the content of Kdm1b is indeed higher than that of Kdm1a.

(4) All embryos in the TCP group were arrested at the four-cell stage. Embryos generated from KDM1b KO females can survive until E10.5 (Ciccone et al., 2009); therefore, TCP-treated embryos show a more severe phenotype than oocyte-derived KDM1b deleted embryos. Depletion of maternal KDM1A protein results in developmental arrest at the two-cell stage ((Ancelin et al., 2016)). The authors need to examine whether TCP treatment affects KDM1a expression. Western blotting would be recommended to quantify the expression of KDM1A and KDM1B in the TCP-treated embryos.

We will further dig the transcriptome data to confirm the specificity of TCP to KDM1b. In addition, the intervention of TCP on the whole fertilized egg in this study increased the H3K4me2 content, and the embryo development retarding effect was more significant than that obtained by crossing with normal paternal lines after knocking down KDM1B from the mother.

(5) H3K4me2 is increased dramatically in the TCP-treated embryos in Figure 4 (the intensity is 1,000 times more than the control). However, the Cut & Run H3K4me2 shows that the H3K4me2 signal is increased in 251 genes and decreased in 194 genes in the TCP-treated embryos (Fold changes > 2, P < 0.01). The authors need to explain why the gain of H3K4me2 is less evident in the Cut & Run data set than in the immunofluorescence result.

Thanks a lot for your question. In the experimental group, the fluorescence value of H3K4me2 in IF was increased by 1000 times (Figure 4E), and the expression of H3K4Me2-related genes in CR was up-regulated and down-regulated for a total of 445 changes (Figure 6A). In our opinion, as a semi-quantitative analysis, immunofluorescence cannot be compared with the quantitative analysis method of CR because of the different analysis models and threshold Settings.

References

Ancelin, K., ne Syx, L., Borensztein, M., mie Ranisavljevic, N., Vassilev, I., Briseñ o-Roa, L., Liu, T., Metzger, E., Servant, N., Barillot, E., Chen, C.-J., Schü le, R., & Heard, E. (2016). Maternal LSD1/KDM1A is an essential regulator of chromatin and transcription landscapes during zygotic genome activation. https://doi.org/10.7554/eLife.08851.001

Ciccone, D. N., Su, H., Hevi, S., Gay, F., Lei, H., Bajko, J., Xu, G., Li, E., & Chen, T. (2009). KDM1B is a histone H3K4 demethylase required to establish maternal genomic imprints. Nature, 461(7262), 415-418. https://doi.org/10.1038/nature08315

Shao, G. B., Chen, J. C., Zhang, L. P., Huang, P., Lu, H. Y., Jin, J., Gong, A. H., & Sang, J. R. (2014). Dynamic patterns of histone H3 lysine 4 methyltransferases and demethylases during mouse preimplantation development. In Vitro Cellular and Developmental Biology - Animal, 50(7), 603-613. https://doi.org/10.1007/s11626-014-9741-6

References

Xia W, Xu J, Yu G, Yao G, Xu K, Ma X, Zhang N, Liu B, Li T, Lin Z, Chen X, Li L, Wang Q, Shi D, Shi S, Zhang Y, Song W, Jin H, Hu L, Bu Z, Wang Y, Na J, Xie W, Sun YP. Resetting histone modifications during human parental-to-zygotic transition. Science. 2019 Jul 26;365(6451):353-360. doi: 10.1126/science.aaw5118. Epub 2019 Jul 4. PMID: 31273069.

Binda C, Valente S, Romanenghi M, Pilotto S, Cirilli R, Karytinos A, Ciossani G, Botrugno OA, Forneris F, Tardugno M, Edmondson DE, Minucci S, Mattevi A, Mai A. Biochemical, structural, and biological evaluation of tranylcypromine derivatives as inhibitors of histone demethylases LSD1 and LSD2. J Am Chem Soc. 2010 May 19;132(19):6827-33.

Fang R, Barbera AJ, Xu Y, Rutenberg M, Leonor T, Bi Q, Lan F, Mei P, Yuan GC, Lian C, Peng J, Cheng D, Sui G, Kaiser UB, Shi Y, Shi YG. Human LSD2/KDM1b/AOF1 regulates gene transcription by modulating intragenic H3K4me2 methylation. Mol Cell. 2010 Jul 30;39(2):222-33. doi: 10.1016/j.molcel.2010.07.008. PMID: 20670891; PMCID: PMC3518444.

Ancelin K, Syx L, Borensztein M, Ranisavljevic N, Vassilev I, Briseño-Roa L, Liu T, Metzger E, Servant N, Barillot E, Chen CJ, Schüle R, Heard E. Maternal LSD1/KDM1A is an essential regulator of chromatin and transcription landscapes during zygotic genome activation. Elife. 2016 Feb 2;5:e08851. doi: 10.7554/eLife.08851. PMID: 26836306; PMCID: PMC4829419.

Reviewer #3 (Public Review):

Summary:

This study explores the dynamic reprogramming of histone modification H3K4me2 during the early stages of mammalian embryogenesis. Utilizing the advanced CUT&RUN technique coupled with high-throughput sequencing, the authors investigate the erasure and re-establishment of H3K4me2 in mouse germinal vesicle (GV) oocytes, metaphase II (MII) oocytes, and early embryos.

Strengths:

The findings provide valuable insights into the temporal and spatial dynamics of H3K4me2 and its potential role in zygotic genome activation (ZGA).

Weaknesses:

The study primarily remains descriptive at this point. It would be advantageous to conduct further comprehensive functional validation and mechanistic exploration.

Key areas for improvement include enhancing the innovation and novelty of the study, providing robust functional validation, establishing a clear model for H3K4me2's role, and addressing technical and presentation issues. The text would benefit from the introduction of a novel conceptual framework or model that provides a clear explanation of the functional consequences and molecular mechanisms underlying H3K4me2 reprogramming in the transition from parental to early embryonic development.

While the findings are significant, the current manuscript falls short in several critical areas. Addressing major and minor issues will significantly strengthen the study's contribution to the field of epigenetic reprogramming and embryonic development.

Author response:

The following is the authors’ response to the current reviews.

Reviewer #1 (Public Review):

The authors did a great job addressing the weaknesses I raised in the previous round of review, except on the generalizability of the current result in the larger context of multi-attribute decision-making. It is not really a weakness of the manuscript but more of a limitation of the studied topic, so I want to keep this comment for public readers.

The reward magnitude and probability information are displayed using rectangular bars of different colors and orientations. Would that bias subjects to choose an additive rule instead of the multiplicative rule? Also, could the conclusion be extended to other decision contexts such as quality and price, where a multiplicative rule is hard to formulate?

We thank the reviewer for the comment. With regards whether the current type of stimuli may have biased participants to use an additive rule rather, we believe many other forms of stimuli for representing choice attributes would be equally likely to cause a similar bias. This is because the additive strategy is an inherently simplistic and natural way to integrate different pieces of non-interacting information. More importantly, even though it is easy to employ an additive strategy, most participants still demonstrated some levels of employing the multiplicative rule. However, it would indeed be interesting for future studies to explore whether the current composite model remains dominant in situations where the optimal solutions require an additive or subtractive rule, such as those concerning quality and price.

“The same would apply even with a different choice of cues as long as the information is conveyed by two independent visual features.”

“While the additive strategy is a natural and simple approach for integrating non-interacting pieces of information, to some extent, participants also used the multiplicative strategy that was optimal in the current experiment. A general question for such composite models is whether people mix two strategies in a consistent manner on every trial or whether there is some form of probabilistic selection occurring between the two strategies on each trial such that only one strategy is used on any given trial while, on average, one strategy is more probable than the other. It would also be interesting to examine whether a composite model is appropriate in contexts where the optimal solution is additive or subtractive, such as those concerning quality and price.”

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

Summary:

The current study provided a follow-up analysis using published datasets focused on the individual variability of both the distraction effect (size and direction) and the attribute integration style, as well as the association between the two. The authors tried to answer the question of whether the multiplicative attribute integration style concurs with a more pronounced and positively oriented distraction effect.

Strengths:

The analysis extensively examined the impacts of various factors on decision accuracy, with a particular focus on using two-option trials as control trials, following the approach established by Cao & Tsetsos (2022). The statistical significance results were clearly reported.

The authors meticulously conducted supplementary examinations, incorporating the additional term HV+LV into GLM3. Furthermore, they replaced the utility function from the expected value model with values from the composite model.

We thank the reviewer for the positive response and are pleased that the reviewer found our report interesting.

Reviewer #1 Comment 1

Weaknesses:

There are several weaknesses in terms of theoretical arguments and statistical analyses.

First, the manuscript suggests in the abstract and at the beginning of the introduction that the study reconciled the "different claims" about "whether distraction effect operates at the level of options' component attributes rather than at the level of their overall value" (see line 13-14), but the analysis conducted was not for that purpose. Integrating choice attributes in either an additive or multiplicative way only reflects individual differences in combining attributes into the overall value. The authors seemed to assume that the multiplicative way generated the overall value ("Individuals who tended to use a multiplicative approach, and hence focused on overall value", line 20-21), but such implicit assumption is at odds with the statement in line 77-79 that people may use a simpler additive rule to combine attributes, which means overall value can come from the additive rule.

We thank the reviewer for the comment. We have made adjustments to the manuscript to ensure that the message delivered within this manuscript is consistent. Within this manuscript, our primary focus is on the different methods of value integration in which the overall value is computed (i.e., additive, multiplicative, or both), rather than the interaction at the individual level of attributes. However, we do not exclude the possibility that the distractor effect may occur at multiple levels. Nevertheless, in light of the reviewer’s comment, we agree that we should focus the argument on whether distractors facilitate or impair decision making and downplay the separate argument about the level at which distractor effects operate. We have now revised the abstract:

“It is widely agreed that people make irrational decisions in the presence of irrelevant distractor options. However, there is little consensus on whether decision making is facilitated or impaired by the presence of a highly rewarding distractor or whether the distraction effect operates at the level of options’ component attributes rather than at the level of their overall value. To reconcile different claims, we argue that it is important to incorporate consideration of the diversity of people’s ways of decision making. We focus on a recent debate over whether people combine choice attributes in an additive or multiplicative way. Employing a multi-laboratory dataset investigating the same decision making paradigm, we demonstrated that people used a mix of both approaches and the extent to which approach was used varied across individuals. Critically, we identified that this variability was correlated with the effect of the distractor on decision making. Individuals who tended to use a multiplicative approach to compute value, showed a positive distractor effect. In contrast, in individuals who tended to use an additive approach, a negative distractor effect (divisive normalisation) was prominent. These findings suggest that the distractor effect is related to how value is constructed, which in turn may be influenced by task and subject specificities. Our work concurs with recent behavioural and neuroscience findings that multiple distractor effects co-exist.” (Lines 12-26)

Furthermore, we acknowledge that the current description of the additive rule could be interpreted in several ways. The current additive utility model described as:

where  is the options’ utility,  is the reward magnitude,  is the probability, and  is the magnitude/probability weighing ratio . If we perform comparison between values according to this model (i.e., HV against LV), we would arrive at the following comparison:

If we rearrange (1), we will arrive at:

While equations (1) and (2) are mathematically equivalent, equation (1) illustrates the interpretation where the comparison of the utilities occurs after value integration and forming an overall value. On the other hand, equation (2) can be broadly interpreted as the comparison of individual attributes in the absence of an overall value estimate for each option. Nonetheless, while we do not exclude the possibility that the distractor effect may occur at multiple levels, we have made modifications to the main manuscript employ more consistently a terminology referring to different methods of value estimation while recognizing that our empirical results are compatible with both interpretations.

Reviewer #1 Comment 2

The second weakness is sort of related but is more about the lack of coherent conceptual understanding of the "additive rule", or "distractor effect operates at the attribute level". In an assertive tone (lines 77-80), the manuscript suggests that a weighted sum integration procedure of implementing an "additive rule" is equal to assuming that people compare pairs of attributes separately, without integration. But they are mechanistically distinct. The additive rule (implemented using the weighted sum rule to combine probability and magnitude within each option and then applying the softmax function) assumes value exists before comparing options. In contrast, if people compare pairs of attributes separately, preference forms based on the within-attribute comparisons. Mathematically these two might be equivalent only if no extra mechanisms (such as inhibition, fluctuating attention, evidence accumulation, etc) are included in the within-attribute comparison process, which is hardly true in the three-option decision.

We thank the reviewer for the comment. As described in our response to Reviewer #1 Comment 1, we are aware and acknowledge that there may be multiple possible interpretations of the additive rule. We also agree with the reviewer that there may be additional mechanisms that are involved in three- or even two- option decisions, but these would require additional studies to tease apart. Another motivation for the approach used here, which does not explicitly model the extra mechanisms the reviewer refers to was due to the intention of addressing and integrating findings from previous studies using the same dataset [i.e. (Cao & Tsetsos, 2022; Chau et al., 2020)]. Lastly, regardless of the mechanistic interpretation, our results show a systematic difference in the process of value estimation. Modifications to the manuscript text have been made consistent with our motivation (please refer to the reply and the textual changes proposed in response to the reviewer’s previous comment: Reviewer #1 Comment 1).

Reviewer #1 Comment 3

Could the authors comment on the generalizability of the current result? The reward magnitude and probability information are displayed using rectangular bars of different colors and orientations. Would that bias subjects to choose an additive rule instead of the multiplicative rule? Also, could the conclusion be extended to other decision contexts such as quality and price, whether a multiplicative rule is hard to formulate?

We thank the reviewer for the comment. We agree with the observation that the stimulus space, with colour linearly correlated with magnitude, and orientation linearly correlated with probability, may bias subjects towards an additive rule. But that’s indeed the point: in order to maximise reward, subjects should have focused on the outcome space without being driven by the stimulus space. In practice, people are more or less successful in such endeavour. Nevertheless, we argue that the specific choice of visual stimuli we used is no more biased towards additive space than any other. In fact, as long as two or more pieces of information are provided for each option, as opposed to a single cue whose value was previously learned, there will always be a bias towards an additive heuristic (a linear combination), regardless of whether the cues are shapes, colours, graphs, numbers, words.

As the reviewer suggested, the dataset analyzed in the current manuscript suggests that the participants were leaning towards the additive rule. Although there was a general tendency using the additive rule while choosing between the rectangular bars, we can still observe a spread of individuals using either, or both, additive and multiplicative rules, suggesting that there was indeed diversity in participants’ decision making strategies in our data.

In previous studies, it was observed that human and non-human individuals used a mix of multiplicative and additive rules when they were tested on experimental paradigms different from ours (Bongioanni et al., 2021; Farashahi et al., 2019; Scholl et al., 2014). It was also observed that positive and negative distractor effects can be both present in the same data set when human and non-human individuals made decisions about food and social partner (Chang et al., 2019; Louie et al., 2013). It was less clear in the past whether the precise way a distractor affects decision making (i.e., positive/negative distractor effect) is related to the use of decision strategy (i.e., multiplicative/additive rules) and this is exactly what we are trying to address in this manuscript. A follow-up study looking at neural data (such as functional magnetic resonance imaging data) could provide a better understanding of the mechanistic nature of the relationship between distractor effects and decision strategy that we identified here.

We agree with the reviewer that it is true that a multiplicative strategy may not be applicable to some decision contexts. Here it is important to look at the structure of the optimal solution (the one maximizing value in the long run). Factors modulating value (such as probability and temporal delay) require a non-linear (e.g., multiplicative solution), while factors of the cost-benefit form (such as effort and price) require a linear solution (e.g., subtraction). In the latter scenario the additive heuristic would coincide with the optimal solution, and the effect addressed in this study may not be revealed. Nonetheless, the present data supports the notion of distinct neural mechanisms at least for probabilistic decision-making, and is likely applicable to decision-making in general.

Our findings, in conjunction with the literature, also suggest that a positive distractor effect could be a general phenomenon in decision mechanisms that involve the medial prefrontal cortex. For example, it has been shown that the positive distractor effect is related to a decision mechanism linked to medial prefrontal cortex [especially the ventromedial prefrontal cortex (Chau et al., 2014; Noonan et al., 2017)]. It is also known a similar brain region is involved not only when individuals are combining information using a multiplicative strategy (Bongioanni et al., 2021), but also when they are combining information to evaluate new experience or generalize information (Baram et al., 2021; Barron et al., 2013; Park et al., 2021). We have now revised the Discussion to explain this:

“In contrast, the positive distractor effect is mediated by the mPFC (Chau et al., 2014; Fouragnan et al., 2019). Interestingly, the same or adjacent, interconnected mPFC regions have also been linked to the mechanisms by which representational elements are integrated into new representations (Barron et al., 2013; Klein-Flügge et al., 2022; Law et al., 2023; Papageorgiou et al., 2017; Schwartenbeck et al., 2023). In a number of situations, such as multi-attribute decision making, understanding social relations, and abstract knowledge, the mPFC achieves this by using a spatial map representation characterised by a grid-like response (Constantinescu et al., 2016; Bongioanni et al., 2021; Park et al., 2021) and disrupting mPFC leads to the evaluation of composite choice options as linear functions of their components (Bongioanni et al., 2021). These observations suggest a potential link between positive distractor effects and mechanisms for evaluating multiple component options and this is consistent with the across-participant correlation that we observed between the strength of the positive distractor effect and the strength of non-additive (i.e., multiplicative) evaluation of the composite stimuli we used in the current task. Hence, one direction for model development may involve incorporating the ideas that people vary in their ways of combining choice attributes and each way is susceptible to different types of distractor effect.” (Lines 260-274)

Reviewer #1 Comment 4

The authors did careful analyses on quantifying the "distractor effect". While I fully agree that it is important to use the matched two-option trials and examine the interaction terms (DV-HV)T as a control, the interpretation of the results becomes tricky when looking at the effects in each trial type. Figure 2c shows a positive DV-HV effect in two-option trials whereas the DV-HV effect was not significantly stronger in three-option trials. Further in Figure 5b,c, in the Multiplicative group, the effect of DV-HV was absent in the two-option trials and present in the three-option trials. In the Additive group, however, the effect of DV-HV was significantly positive in the two-option trials but was significantly lowered in the three-option trials. Hence, it seems the different distractor effects were driven by the different effects of DV-HV in the two-option trials, rather than the three-option trials?

We thank the reviewer for the comment. While it may be a bit more difficult to interpret, the current method of examining the (DV−HV)T term rather than (DV−HV) term was used because it was the approach used in a previous study (Cao & Tsetsos, 2022).

During the design of the original experiments, trials were generated pseudo-randomly until the DV was sufficiently decorrelated from HV−LV. While this method allows for better group-level examination of behaviour, Cao and Tsetsos were concerned that this approach may have introduced unintended confounding covariations to some trials. In theory, one of the unintended covariations could occur between the DV and specific sets of reward magnitude and probability of the HV and LV. The covariation between parameters can lead to an observable positive distractor effect in the DV−HV as a consequence of the attraction effect or an unintended byproduct of using an additive method of integrating attributes [for further elaboration, please refer to Figure 1 in (Cao & Tsetsos, 2022)]. While it may have some limitations, the approach suggested by Cao and Tsetsos has the advantage of leveraging the DV−HV term to absorb any variance contributed by possible confounding factors such that true distractor effects, if any, can be detected using the (DV−HV)T term.

Reviewer #1 Comment 5

Note that the pattern described above was different in Supplementary Figure 2, where the effect of DV-HV on the two-option trials was negative for both Multiplicative and Additive groups. I would suggest considering using Supplementary Figure 2 as the main result instead of Figure 5, as it does not rely on multiplicative EV to measure the distraction effect, and it shows the same direction of DV-HV effect on two-option trials, providing a better basis to interpret the (DV-HV)T effect.

We thank the reviewer for the comments and suggestion. However, as mentioned in the response to Reviewer #1 Comment 4, the current method of analysis adopted in the manuscript and the interpretation of only (DV−HV)T is aimed to address the possibility that the (DV−HV) term may be capturing some confounding effects due to covariation. Given that the debate that is addressed specifically concerns the (DV−HV)T term, we elected to display Figure 5 within the main text and keep the results of the regression after replacing the utility function with the composite model as Supplementary Figure 5 (previously labelled as Supplementary Figure 2).

Reviewer #2 (Public Review):

This paper addresses the empirical demonstration of "distractor effects" in multi-attribute decision-making. It continues a debate in the literature on the presence (or not) of these effects, which domains they arise in, and their heterogeneity across subjects. The domain of the study is a particular type of multi-attribute decision-making: choices over risky lotteries. The paper reports a re-analysis of lottery data from multiple experiments run previously by the authors and other laboratories involved in the debate.

Methodologically, the analysis assumes a number of simple forms for how attributes are aggregated (adaptively, multiplicatively, or both) and then applies a "reduced form" logistic regression to the choices with a number of interaction terms intended to control for various features of the choice set. One of these interactions, modulated by ternary/binary treatment, is interpreted as a "distractor effect."

The claimed contribution of the re-analysis is to demonstrate a correlation in the strength/sign of this treatment effect with another estimated parameter: the relative mixture of additive/multiplicative preferences.

We thank the reviewer for the positive response and are pleased that the reviewer found our report interesting.

Reviewer #2 Comment 1

Major Issues

(1) How to Interpret GLM 1 and 2

This paper, and others before it, have used a binary logistic regression with a number of interaction terms to attempt to control for various features of the choice set and how they influence choice. It is important to recognize that this modelling approach is not derived from a theoretical claim about the form of the computational model that guides decision-making in this task, nor an explicit test for a distractor effect. This can be seen most clearly in the equations after line 321 and its corresponding log-likelihood after 354, which contain no parameter or test for "distractor effects". Rather the computational model assumes a binary choice probability and then shoehorns the test for distractor effects via a binary/ternary treatment interaction in a separate regression (GLM 1 and 2). This approach has already led to multiple misinterpretations in the literature (see Cao & Tsetsos, 2022; Webb et al., 2020). One of these misinterpretations occurred in the datasets the authors studied, in which the lottery stimuli contained a confound with the interaction that Chau et al., (2014) were interpreting as a distractor effect (GLM 1). Cao & Tsetsos (2022) demonstrated that the interaction was significant in binary choice data from the study, therefore it can not be caused by a third alternative. This paper attempts to address this issue with a further interaction with the binary/ternary treatment (GLM 2). Therefore the difference in the interaction across the two conditions is claimed to now be the distractor effect. The validity of this claim brings us to what exactly is meant by a "distractor effect."

The paper begins by noting that "Rationally, choices ought to be unaffected by distractors" (line 33). This is not true. There are many normative models that allow for the value of alternatives (even low-valued "distractors") to influence choices, including a simple random utility model. Since Luce (1959), it has been known that the axiom of "Independence of Irrelevant Alternatives" (that the probability ratio between any two alternatives does not depend on a third) is an extremely strong axiom, and only a sufficiency axiom for a random utility representation (Block and Marschak, 1959). It is not a necessary condition of a utility representation, and if this is our definition of rational (which is highly debatable), not necessary for it either. Countless empirical studies have demonstrated that IIA is falsified, and a large number of models can address it, including a simple random utility model with independent normal errors (i.e. a multivariate Probit model). In fact, it is only the multinomial Logit model that imposes IIA. It is also why so much attention is paid to the asymmetric dominance effect, which is a violation of a necessary condition for random utility (the Regularity axiom).

So what do the authors even mean by a "distractor effect." It is true that the form of IIA violations (i.e. their path through the probability simplex as the low-option varies) tells us something about the computational model underlying choice (after all, different models will predict different patterns). However we do not know how the interaction terms in the binary logit regression relate to the pattern of the violations because there is no formal theory that relates them. Any test for relative value coding is a joint test of the computational model and the form of the stochastic component (Webb et al, 2020). These interaction terms may simply be picking up substitution patterns that can be easily reconciled with some form of random utility. While we can not check all forms of random utility in these datasets (because the class of such models is large), this paper doesn't even rule any of these models out.

We thank the reviewer for the comment. In this study, one objective is to address an issue raised by Cao and Tsetsos (2022), suggesting that the distractor effect claimed in the Chau et al. (2014) study was potentially confounded by unintended correlation introduced between the distractor and the chooseable options. They suggested that this could be tested by analyzing the control binary trials and the experimental ternary trials in a single model (i.e., GLM2) and introducing an interaction term (DV−HV)T. The interaction term can partial out any unintended confound and test the distractor effect that was present specifically in the experimental ternary trials. We adopted these procedures in our current studies and employed the interaction term to test the distractor effects. The results showed that overall there was no significant distractor effect in the group. We agree with the reviewer’s comment that if we were only analysing the ternary trials, a multinomial probit model would be suitable because it allows noise correlation between the choices. Alternatively, had a multinomial logistic model been applied, a Hausman-McFadden Test could be run to test whether the data violates the assumption of independence of irrelevant alternatives (IIA). However, in our case, a binomial model is preferred over a multinomial model because of: (1) the inclusion of the binary trials, and (2) the small number of trials in which the distractor was chosen (the median was 4% of all ternary trials).

However, another main objective of this study is to consider the possibility that the precise distractor effect may vary across individuals. This is exactly why we employed the composite model to estimate individual’s decision making strategy and investigated how that varied with the precise way the distractor influenced decision making.

In addition, we think that the reviewer here is raising a profound point and one with which we are in sympathy; it is true that random noise utility models can predict deviations from the IIA axiom. Central to these approaches is the notion that the representations of the values of choice options are noisy. Thus, when the representation is accessed, it might have a certain value on average but this value might vary from occasion to occasion as if each sample were being drawn from a distribution. As a consequence, the value of a distractor that is “drawn” during a decision between two other options may be larger than the distractor’s average value and may even have a value that is larger than the value drawn from the less valuable choice option’s distribution on the current trial. On such a trial it may become especially clear that the better of the two options has a higher value than the alternative choice option. Our understanding is that Webb, Louie and colleagues (Louie et al., 2013; Webb et al., 2020) suggest an explanation approximately along these lines when they reported a negative distractor effect during some decisions, i.e., they follow the predictions of divisive normalization suggesting that decisions become more random as the distractor’s value is greater.

An alternative approach, however, assumes that rather than noise in the representation of the option itself, there is noise in the comparison process when the two options are compared. This is exemplified in many influential decision making models including evidence accumulation models such as drift diffusion models (Shadlen & Shohamy, 2016) and recurrent neural network models of decision making (Wang, 2008). It is this latter type of model that we have used in our previous investigations (Chau et al., 2020; Kohl et al., 2023). However, these two approaches are linked both in their theoretical origin and in the predictions that they make in many situations (Shadlen & Shohamy, 2016). We therefore clarify that this is the case in the revised manuscript as follows:

“In the current study and in previous work we have used or made reference to models of decision making that assume that a noisy process of choice comparison occurs such as recurrent neural networks and drift diffusion models (Shadlen & Shohamy, 2016; Wang, 2008). Under this approach, positive distractor effects are predicted when the comparison process becomes more accurate because of an impact on the noisy process of choice comparison (Chau et al., 2020; Kohl et al., 2023). However, it is worth noting that another class of models might assume that a choice representation itself is inherently noisy. According to this approach, on any given decision a sample is drawn from a distribution of value estimates in a noisy representation of the option. Thus, when the representation is accessed, it might have a certain value on average but this value might vary from occasion to occasion. As a consequence, the value of a distractor that is “drawn” during decision between two other options may be larger than the distractor’s average value and may even have a value that is larger than the value drawn from the less valuable choice option’s distribution on the current trial. On such a trial it may become especially clear that the better of the two options has a higher value than the alternative choice option. Louie and colleagues (Louie et al., 2013) suggest an explanation approximately along these lines when they reported a positive distractor effect during some decisions. Such different approaches share theoretical origins (Shadlen & Shohamy, 2016) and make related predictions about the impact of distractors on decision making.” (Lines 297-313)

Reviewer #2 Comment 2

(2) How to Interpret the Composite (Mixture) model?

On the other side of the correlation are the results from the mixture model for how decision-makers aggregate attributes. The authors report that most subjects are best represented by a mixture of additive and multiplicative aggregation models. The authors justify this with the proposal that these values are computed in different brain regions and then aggregated (which is reasonable, though raises the question of "where" if not the mPFC). However, an equally reasonable interpretation is that the improved fit of the mixture model simply reflects a misspecification of two extreme aggregation processes (additive and EV), so the log-likelihood is maximized at some point in between them.

One possibility is a model with utility curvature. How much of this result is just due to curvature in valuation? There are many reasonable theories for why we should expect curvature in utility for human subjects (for example, limited perception: Robson, 2001, Khaw, Li Woodford, 2019; Netzer et al., 2022) and of course many empirical demonstrations of risk aversion for small stakes lotteries. The mixture model, on the other hand, has parametric flexibility.

There is also a large literature on testing expected utility jointly with stochastic choice, and the impact of these assumptions on parameter interpretation (Loomes & Sugden, 1998; Apesteguia & Ballester, 2018; Webb, 2019). This relates back to the point above: the mixture may reflect the joint assumption of how choice departs from deterministic EV.

We thank the reviewer for the comment. They are indeed right to mention the vast literature on curvature in subjective valuation; however it is important to stress that the predictions of the additive model with linear basis functions are quite distinct for the predictions of a multiplicative model with non-linear basis functions. We have tested the possibility that participants’ behaviour was better explained by the latter and we showed that this was not the case. Specifically, we have added and performed model fitting on an additional model with utility curvature based on prospect theory (Kahneman & Tversky, 1979) with the weighted probability function suggested by (Prelec, 1998):

where  and  represent the reward magnitude and probability (both rescaled to the interval between 0 and 1), respectively.  is the weighted magnitude and  is the weighted probability, while  and  are the corresponding distortion parameters. This prospect theory (PT) model is included along with the four previous models (please refer to Figure 3) in a Bayesian model comparison. Results indicate that the composite model remains the best account of participants’ choice behaviour (exceedance probability = 1.000, estimated model frequency = 0.720). We have now included these results in the main text and Supplementary Figure 2:

“Supplementary Figure 2 reports an additional Bayesian model comparison performed while including a model with nonlinear utility functions based on Prospect Theory (Kahneman & Tversky, 1979) with the Prelec formula for probability (Prelec, 1998). Consistent with the above finding, the composite model provides the best account of participants’ choice behaviour (exceedance probability = 1.000, estimated model frequency = 0.720).” (Lines 193-198)

Reviewer #2 Comment 3

3) So then how should we interpret the correlation that the authors report?

On one side we have the impact of the binary/ternary treatment which demonstrates some impact of the low value alternative on a binary choice probability. This may reflect some deep flaws in existing theories of choice, or it may simply reflect some departure from purely deterministic expected value maximization that existing theories can address. We have no theory to connect it to, so we cannot tell. On the other side of the correlation, we have a mixture between additive and multiplicative preferences over risk. This result may reflect two distinct neural processes at work, or it may simply reflect a misspecification of the manner in which humans perceive and aggregate attributes of a lottery (or even just the stimuli in this experiment) by these two extreme candidates (additive vs. EV). Again, this would entail some departure from purely deterministic expected value maximization that existing theories can address.

It is entirely possible that the authors are reporting a result that points to the more exciting of these two possibilities. But it is also possible (and perhaps more likely) that the correlation is more mundane. The paper does not guide us to theories that predict such a correlation, nor reject any existing ones. In my opinion, we should be striving for theoretically-driven analyses of datasets, where the interpretation of results is clearer.

We thank the reviewer for their clear comments. Based on our responses to the previous comments it should be apparent that our results are consistent with several existing theories of choice, so we are not claiming that there are deep flaws in them, but distinct neural processes (additive and multiplicative) are revealed, and this does not reflect a misspecification in the modelling. We have revised our manuscript in the light of the reviewer’s comments in the hope of clarifying the theoretical background which informed both our data analysis and our data interpretation.

First, we note that there are theoretical reasons to expect a third option might impact on choice valuation. There is a large body of work suggesting that a third option may have an impact on the values of two other options (indeed Reviewer #2 refers to some of this work in their Reviewer #2 Comment 1), but the body of theoretical work originates partly in neuroscience and not just in behavioural economics. In many sensory systems, neural activity changes with the intensity of the stimuli that are sensed. Divisive normalization in sensory systems, however, describes the way in which such neural responses are altered also as a function of other adjacent stimuli (Carandini & Heeger, 2012; Glimcher, 2022; Louie et al., 2011, 2013). The phenomenon has been observed at neural and behavioural levels as a function not just of the physical intensity of the other stimuli but as a function of their associated value (Glimcher, 2014, 2022; Louie et al., 2011, 2015; Noonan et al., 2017; Webb et al., 2020).

Analogously there is an emerging body of work on the combinatorial processes that describe how multiple representational elements are integrated into new representations (Barron et al., 2013; Papageorgiou et al., 2017; Schwartenbeck et al., 2023). These studies have originated in neuroscience, just as was the case with divisive normalization, but they may have implications for understanding behaviour. For example, they might be linked to behavioural observations that the values assigned to bundles of goods are not necessarily the sum of the values of the individual goods (Hsee, 1998; List, 2002). One neuroscience fact that we know about such processes is that, at an anatomical level, they are linked to the medial frontal cortex (Barron et al., 2013; Fellows, 2006; Hunt et al., 2012; Papageorgiou et al., 2017; Schwartenbeck et al., 2023). A second neuroscientific fact that we know about medial frontal cortex is that it is linked to any positive effects that distractors might have on decision making (Chau et al., 2014; Noonan et al., 2017). Therefore, we might make use of these neuroscientific facts and theories to predict a correlation between positive distractor effects and non-additive mechanisms for determining the integrated value of multi-component choices. This is precisely what we did; we predicted the correlation on the basis of this body of work and when we tested to see if it was present, we found that indeed it was. It may be the case that other behavioural economics theories offer little explanation of the associations and correlations that we find. However, we emphasize that this association is predicted by neuroscientific theory and in the revised manuscript we have attempted to clarify this in the Introduction and Discussion sections:

“Given the overlap in neuroanatomical bases underlying the different methods of value estimation and the types of distractor effects, we further explored the relationship. Critically, those who employed a more multiplicative style of integrating choice attributes also showed stronger positive distractor effects, whereas those who employed a more additive style showed negative distractor effects. These findings concur with neural data demonstrating that the medial prefrontal cortex (mPFC) computes the overall values of choices in ways that go beyond simply adding their components together, and is the neural site at which positive distractor effects emerge (Barron et al., 2013; Bongioanni et al., 2021; Chau et al., 2014; Fouragnan et al., 2019; Noonan et al., 2017; Papageorgiou et al., 2017), while divisive normalization was previously identified in the posterior parietal cortex (PPC) (Chau et al., 2014; Louie et al., 2011).” (Lines 109-119)

“At the neuroanatomical level, the negative distractor effect is mediated by the PPC, where signal modulation described by divisive normalization has been previously identified (Chau et al., 2014; Louie et al., 2011). The same region is also crucial for perceptual decision making processes (Shadlen & Shohamy, 2016). The additive heuristics for combining choice attributes are closer to a perceptual evaluation because distances in this subjective value space correspond linearly to differences in physical attributes of the stimuli, whereas normative (multiplicative) value has a non-linear relation with them (cf. Figure 1c). It is well understood that many sensory mechanisms, such as in primates’ visual systems or fruit flies’ olfactory systems, are subject to divisive normalization (Carandini & Heeger, 2012). Hence, the additive heuristics that are more closely based on sensory mechanisms could also be subject to divisive normalization, leading to negative distractor effects in decision making.

In contrast, the positive distractor effect is mediated by the mPFC (Chau et al., 2014; Fouragnan et al., 2019). Interestingly, the same or adjacent, interconnected mPFC regions have also been linked to the mechanisms by which representational elements are integrated into new representations (Barron et al., 2013; Klein-Flügge et al., 2022; Law et al., 2023; Papageorgiou et al., 2017; Schwartenbeck et al., 2023). In a number of situations, such as multi-attribute decision making, understanding social relations, and abstract knowledge, the mPFC achieves this by using a spatial map representation characterised by a grid-like response (Constantinescu et al., 2016; Bongioanni et al., 2021; Park et al., 2021) and disrupting mPFC leads to the evaluation of composite choice options as linear functions of their components (Bongioanni et al., 2021). These observations suggest a potential link between positive distractor effects and mechanisms for evaluating multiple component options and this is consistent with the across-participant correlation that we observed between the strength of the positive distractor effect and the strength of non-additive (i.e., multiplicative) evaluation of the composite stimuli we used in the current task. Hence, one direction for model development may involve incorporating the ideas that people vary in their ways of combining choice attributes and each way is susceptible to different types of distractor effect.” (Lines 250-274)

Reviewer #2 Comment 4

(4) Finally, the results from these experiments might not have external validity for two reasons. First, the normative criterion for multi-attribute decision-making differs depending on whether the attributes are lotteries or not (i.e. multiplicative vs additive). Whether it does so for humans is a matter of debate. Therefore if the result is unique to lotteries, it might not be robust for multi-attribute choice more generally. The paper largely glosses over this difference and mixes literature from both domains. Second, the lottery information was presented visually and there is literature suggesting this form of presentation might differ from numerical attributes. Which is more ecologically valid is also a matter of debate.

We thank the reviewer for the comment. Indeed, they are right that the correlation we find between value estimation style and distractor effects may not be detected in all contexts of human behaviour. What the reviewer suggests goes along the same lines as our response to Reviewer #1 Comment 3, multi-attribute value estimation may have different structure: in some cases, the optimal solution may require a non-linear (e.g., multiplicative) response as in probabilistic or delayed decisions, but other cases (e.g., when estimating the value of a snack based on its taste, size, healthiness, price) a linear integration would suffice. In the latter kind of scenarios, both the optimal and the heuristic solutions may be additive and people’s value estimation “style” may not be teased apart. However, if different neural mechanisms associated with difference estimation processes are observed in certain scenarios, it suggests that these mechanisms are always present, even in scenarios where they do not alter the predictions. Probabilistic decision-making is also pervasive in many aspects of daily life and not just limited to the case of lotteries.

While behaviour has been found to differ depending on whether lottery information is presented graphically or numerically, there is insufficient evidence to suggest biases towards additive or multiplicative evaluation, or towards positive or negative distractor effects. As such, we may expect that the correlation that we reveal in this paper, grounded in distinct neural mechanisms, would still hold even under different circumstances.

Taking previous literature as examples, similar patterns of behaviour have been observed in humans when making decisions during trinary choice tasks. In a study conducted by Louie and colleagues (Louie et al., 2013; Webb et al., 2020), human participants performed a snack choice task where their behaviour could be modelled by divisive normalization with biphasic response (i.e., both positive and negative distractor effects). While these two studies only use a single numerical value of price for behavioural modelling, these prices should originate from an internal computation of various attributes related to each snack that are not purely related to lotteries. Expanding towards the social domain, studies of trinary decision making have considered face attractiveness and averageness (Furl, 2016), desirability of hiring (Chang et al., 2019), as well as desirability of candidates during voting (Chang et al., 2019). These choices involve considering various attributes unrelated to lotteries or numbers and yet, still display a combination of positive distractor and negative distractor (i.e. divisive normalization) effects, as in the current study. In particular, the experiments carried out by Chang and colleagues (Chang et al., 2019) involved decisions in a social context that resemble real-world situations. These findings suggests that both types of distractor effects can co-exist in other value based decision making tasks (Li et al., 2018; Louie et al., 2013) as well as decision making tasks in social contexts (Chang et al., 2019; Furl, 2016).

Reviewer #2 Comment 5

Minor Issues:

The definition of EV as a normative choice baseline is problematic. The analysis requires that EV is the normative choice model (this is why the HV-LV gap is analyzed and the distractor effect defined in relation to it). But if the binary/ternary interaction effect can be accounted for by curvature of a value function, this should also change the definition of which lottery is HV or LV for that subject!

We thank the reviewer for the comment. While the initial part of the paper discussed results that were defined by the EV model, the results shown in Supplementary Figure 2 were generated by replacing the utility function based on values obtained by using the composite model. Here, we have also redefined the definition of HV or LV for each subject depending on the updated value generated by the composite model prior to the regression.

References

Apesteguia, J. & Ballester, M. Monotone stochastic choice models: The case of risk and time preferences. Journal of Political Economy (2018).

Block, H. D. & Marschak, J. Random Orderings and Stochastic Theories of Responses. Cowles Foundation Discussion Papers (1959).

Khaw, M. W., Li, Z. & Woodford, M. Cognitive Imprecision and Small-Stakes Risk Aversion. Rev. Econ. Stud. 88, 1979-2013 (2020).

Loomes, G. & Sugden, R. Testing Different Stochastic Specificationsof Risky Choice. Economica 65, 581-598 (1998).

Luce, R. D. Indvidual Choice Behaviour. (John Wiley and Sons, Inc., 1959).

Netzer, N., Robson, A. J., Steiner, J. & Kocourek, P. Endogenous Risk Attitudes. SSRN Electron. J. (2022) doi:10.2139/ssrn.4024773.

Robson, A. J. Why would nature give individuals utility functions? Journal of Political Economy 109, 900-914 (2001).

Webb, R. The (Neural) Dynamics of Stochastic Choice. Manage Sci 65, 230-255 (2019).

Reviewer #3 (Public Review):

Summary:

The way an unavailable (distractor) alternative impacts decision quality is of great theoretical importance. Previous work, led by some of the authors of this study, had converged on a nuanced conclusion wherein the distractor can both improve (positive distractor effect) and reduce (negative distractor effect) decision quality, contingent upon the difficulty of the decision problem. In very recent work, Cao and Tsetsos (2022) reanalyzed all relevant previous datasets and showed that once distractor trials are referenced to binary trials (in which the distractor alternative is not shown to participants), distractor effects are absent. Cao and Tsetsos further showed that human participants heavily relied on additive (and not multiplicative) integration of rewards and probabilities.

The present study by Wong et al. puts forward a novel thesis according to which interindividual differences in the way of combining reward attributes underlie the absence of detectable distractor effect at the group level. They re-analysed the 144 human participants and classified participants into a "multiplicative integration" group and an "additive integration" group based on a model parameter, the "integration coefficient", that interpolates between the multiplicative utility and the additive utility in a mixture model. They report that participants in the "multiplicative" group show a negative distractor effect while participants in the "additive" group show a positive distractor effect. These findings are extensively discussed in relation to the potential underlying neural mechanisms.

Strengths:

- The study is forward-looking, integrating previous findings well, and offering a novel proposal on how different integration strategies can lead to different choice biases.

- The authors did an excellent job of connecting their thesis with previous neural findings. This is a very encompassing perspective that is likely to motivate new studies towards a better understanding of how humans and other animals integrate information in decisions under risk and uncertainty.

- Despite that some aspects of the paper are very technical, methodological details are well explained and the paper is very well written.

We thank the reviewer for the positive response and are pleased that the reviewer found our report interesting.

Reviewer #3 Comment 1

Weaknesses:

The authors quantify the distractor variable as "DV - HV", i.e., the relative distractor variable. Do the conclusions hold when the distractor is quantified in absolute terms (as "DV", see also Cao & Tsetsos, 2023)? Similarly, the authors show in Suppl. Figure 1 that the inclusion of a HV + LV regressor does not alter their conclusions. However, the (HV + LV)*T regressor was not included in this analysis. Does including this interaction term alter the conclusions considering there is a high correlation between (HV + LV)*T and (DV - HV)*T? More generally, it will be valuable if the authors assess and discuss the robustness of their findings across different ways of quantifying the distractor effect.

We thank the reviewer for the comment. In the original manuscript we had already demonstrated that the distractor effect was related to the integration coefficient using a number of complementary analyses. They include Figure 5 based on GLM2, Supplementary Figure 3 based on GLM3 (i.e., adding the HV+LV term to GLM2), and Supplementary Figure 4 based on GLM2 but applying the utility estimate from the composite model instead of expected value (EV). These three sets of analyses produced comparable results. The reason why we elected not to include the (HV+LV)T term in GLM3 (Supplementary Figure 3) was due to the collinearity between the regressors in the GLM. If this term is included in GLM3, the variance inflation factor (VIF) would exceed an acceptable level of 4 for some regressors. In particular, the VIF for the (HV+LV) and (HV+LV)T regressors is 5.420, while the VIF for (DV−HV) and (DV−HV)T is 4.723.

Here, however, we consider the additional analysis suggested by the reviewer and test whether similar results are obtained. We constructed GLM4 including the (HV+LV)T term but replacing the relative distractor value (DV-HV) with the absolute distractor value (DV) in the main term and its interactions, as follows:

GLM4:

A significant negative (DV)T effect was found for the additive group [t(72)=−2.0253, p=0.0465] while the multiplicative group had a positive trend despite not reaching significance. Between the two groups, the (DV)T term was significantly different [t(142)=2.0434, p=0.0429]. While these findings suggest that the current conclusions could be partially replicated, simply replacing the relative distractor value with the absolute value in the previous analyses resulted in non-significant findings. Taking these results together with the main findings, it is possible to conclude that the positive distractor effect is better captured using the relative DV-HV term rather than the absolute DV term. This would be consistent with the way in which option values are envisaged to interact with one another in the mutual inhibition model (Chau et al., 2014, 2020) that generates the positive distractor effect. The model suggests that evidence is accumulated as the difference between the excitatory input from the option (e.g. the HV option) and the pooled inhibition contributed partly by the distractor. We have now included these results in the manuscript:

“Finally, we performed three additional analyses that revealed comparable results to those shown in Figure 5. In the first analysis, reported in Supplementary Figure 3, we added an  term to the GLM, because this term was included in some analyses of a previous study that used the same dataset (Chau et al., 2020). In the second analysis, we added an  term to the GLM. We noticed that this change led to inflation of the collinearity between the regressors and so we also replaced the (DV−HV) term by the DV term to mitigate the collinearity (Supplementary Figure 4). In the third analyses, reported in Supplementary Figure 5, we replaced the utility terms of GLM2. Since the above analyses involved using HV, LV, and DV values defined by the normative Expected Value model, here, we re-defined the values using the composite model prior to applying GLM2. Overall, in the Multiplicative Group a significant positive distractor effect was found in Supplementary Figures 3 and 4. In the Additive Group a significant negative distractor effect was found in Supplementary Figures 3 and 5. Crucially, all three analyses consistently showed that the distractor effects were significantly different between the Multiplicative Group and the Additive Group.” (Lines 225-237)

Reviewer #3 Comment 2

The central finding of this study is that participants who integrate reward attributes multiplicatively show a positive distractor effect while participants who integrate additively show a negative distractor effect. This is a very interesting and intriguing observation. However, there is no explanation as to why the integration strategy covaries with the direction of the distractor effect. It is unlikely that the mixture model generates any distractor effect as it combines two "context-independent" models (additive utility and expected value) and is fit to the binary-choice trials. The authors can verify this point by quantifying the distractor effect in the mixture model. If that is the case, it will be important to highlight that the composite model is not explanatory; and defer a mechanistic explanation of this covariation pattern to future studies.

We thank the reviewer for the comment. Indeed, the main purpose of applying the mixture model was to identify the way each participants combined attributes and, as the reviewer pointed out, the mixture model per se is context independent. While we acknowledge that the mixture model is not a mechanistic explanation, there is a theoretical basis for the observation that these two factors are linked.

Firstly, studies that have examined the processes involved when humans combine and integrate different elements to form new representations (Barron et al., 2013; Papageorgiou et al., 2017; Schwartenbeck et al., 2023) have implicated the medial frontal cortex as a crucial region (Barron et al., 2013; Fellows, 2006; Hunt et al., 2012; Papageorgiou et al., 2017; Schwartenbeck et al., 2023). Meanwhile, previous studies have also identified that positive distractor effects are linked to the medial frontal cortex (Chau et al., 2014; Noonan et al., 2017). Therefore, the current study utilized these two facts to establish the basis for a correlation between positive distractor effects and non-additive mechanisms for determining the integrated value of multi-component choices. Nevertheless, we agree with the reviewer that it will be an important future direction to look at how the covariation pattern emerges in a computational model. We have revised the manuscript in an attempt to address this issue.

“At the neuroanatomical level, the negative distractor effect is mediated by the PPC, where signal modulation described by divisive normalization has been previously identified (Chau et al., 2014; Louie et al., 2011). The same region is also crucial for perceptual decision making processes (Shadlen & Shohamy, 2016). The additive heuristics for combining choice attributes are closer to a perceptual evaluation because distances in this subjective value space correspond linearly to differences in physical attributes of the stimuli, whereas normative (multiplicative) value has a non-linear relation with them (cf. Figure 1c). It is well understood that many sensory mechanisms, such as in primates’ visual systems or fruit flies’ olfactory systems, are subject to divisive normalization (Carandini & Heeger, 2012). Hence, the additive heuristics that are more closely based on sensory mechanisms could also be subject to divisive normalization, leading to negative distractor effects in decision making.

In contrast, the positive distractor effect is mediated by the mPFC (Chau et al., 2014; Fouragnan et al., 2019). Interestingly, the same or adjacent, interconnected mPFC regions have also been linked to the mechanisms by which representational elements are integrated into new representations (Barron et al., 2013; Klein-Flügge et al., 2022; Law et al., 2023; Papageorgiou et al., 2017; Schwartenbeck et al., 2023). In a number of situations, such as multi-attribute decision making, understanding social relations, and abstract knowledge, the mPFC achieves this by using a spatial map representation characterised by a grid-like response (Constantinescu et al., 2016; Bongioanni et al., 2021; Park et al., 2021) and disrupting mPFC leads to the evaluation of composite choice options as linear functions of their components (Bongioanni et al., 2021). These observations suggest a potential link between positive distractor effects and mechanisms for evaluating multiple component options and this is consistent with the across-participant correlation that we observed between the strength of the positive distractor effect and the strength of non-additive (i.e., multiplicative) evaluation of the composite stimuli we used in the current task. Hence, one direction for model development may involve incorporating the ideas that people vary in their ways of combining choice attributes and each way is susceptible to different types of distractor effect.” (Lines 250-274)

Reviewer #3 Comment 3

-  Correction for multiple comparisons (e.g., Bonferroni-Holm) was not applied to the regression results. Is the "negative distractor effect in the Additive Group" (Fig. 5c) still significant after such correction? Although this does not affect the stark difference between the distractor effects in the two groups (Fig. 5a), the classification of the distractor effect in each group is important (i.e., should future modelling work try to capture both a negative and a positive effect in the two integration groups? Or just a null and a positive effect?).

We thank the reviewer for the comment. We have performed Bonferroni-Holm correction and as the reviewer surmised, the negative distractor effect in the additive group becomes non-significant. However, we have to emphasize that our major claim is that there was a covariation between decision strategy (of combining attributes) and distractor effect (as seen in Figure 4). That analysis does not imply multiple comparisons. The analysis in Figure 5 that splits participants into two groups was mainly designed to illustrate the effects for an easier understanding by a more general audience. In many cases, the precise ways in which participants are divided into subgroups can have a major impact on whether each individual group’s effects are significant or not. It may be possible to identify an optimal way of grouping, but we refrained from taking such a trial-and-error approach, especially for the analysis in Figure 5 that simply supplements the point made in Figure 4. The key notion we would like the readers to take away is that there is a spectrum of distractor effects (ranging from negative to positive) that will vary depending on how the choice attributes were integrated.

Reviewer #1 (Recommendations For The Authors):

Reviewer #1 Recommendations 1

Enhancements are necessary for the quality of the scientific writing. Several sentences have been written in a negligent manner and warrant revision to ensure a higher level of rigor. Moreover, a number of sentences lack appropriate citations, including but not restricted to:

- Line 39-41.

- Line 349-350 (also please clarify what it means by parameter estimate" is very accurate: correlation?).

We thank the reviewer for the comment. We have made revisions to various parts of the manuscript to address the reviewer’s concerns.

“Intriguingly, most investigations have considered the interaction between distractors and chooseable options either at the level of their overall utility or at the level of their component attributes, but not both (Chau et al., 2014, 2020; Gluth et al., 2018).” (Lines 40-42)

“Additional simulations have shown that the fitted parameters can be recovered with high accuracy (i.e., with a high correlation between generative and recovered parameters).” (Lines 414-416)

Reviewer #1 Recommendations 2

Some other minor suggestions:

- Correlative vs. Causality: the manuscript exhibits a lack of attentiveness in drawing causal conclusions from correlative evidence (manuscript title, Line 91, Line 153-155).

- When displaying effect size on accuracy, there is no need to show the significance of intercept (Figure 2,5, & supplementary figures).

- Adding some figure titles on Figure 2 so it is clear what each panel stands for.

- In Figure 3, the dots falling on zero values are not easily seen. Maybe increasing the dot size a little?

- Line 298: binomial linking function (instead of binomial distribution).

- Line 100: composite, not compositive.

- Line 138-139: please improve the sentence, if it's consistent with previous findings, what's the point of "surprisingly"?

We thank the reviewer for the suggestions. We have made revisions to the title and various parts of the manuscript to address the reviewer’s concerns.

- Correlative vs. Causality: the manuscript exhibits a lack of attentiveness in drawing causal conclusions from correlative evidence (manuscript title, Line 91, Line 153-155).

We have now revised the manuscript:

“Distractor effects in decision making are related to the individual’s style of integrating choice attributes” (title of the manuscript)

“More particularly, we consider whether individual differences in combination styles could be related to different forms of distractor effect.” (Lines 99-100)

“While these results may seem to suggest that a distractor effect was not present at an overall group level, we argue that the precise way in which a distractor affects decision making is related to how individuals integrate the attributes.” (Lines 164-167)

- When displaying effect size on accuracy, there is no need to show the significance of intercept (Figure 2,5, & supplementary figures).

We have also modified all Figures to remove the intercept.

- Adding some figure titles on Figure 2 so it is clear what each panel stands for.

We have added titles accordingly.

- In Figure 3, the dots falling on zero values are not easily seen. Maybe increasing the dot size a little?

In conjunction with addressing Reviewer #3 Recommendation 6, we have adapted the violin plots into histograms for a better representation of the values.

- Line 298: binomial linking function (instead of binomial distribution).

- Line 100: composite, not compositive.

- Line 138-139: please improve the sentence, if it's consistent with previous findings, what's the point of "surprisingly"?

We have made revisions accordingly.

Reviewer #2 (Recommendations For The Authors):

Reviewer #2 Recommendations 1

Line 294. The definition of DV, HV, LV is not sufficient. Presumably, these are the U from the following sections? Or just EV? But this is not explicitly stated, rather they are vaguely referred to as values." The computational modelling section refers to them as utilities. Are these the same thing?

We thank the reviewer for the suggestion. We have clarified that the exact method for calculating each of the values and updated the section accordingly.

“where HV, LV, and DV refer to the values of the chooseable higher value option, chooseable lower value option, and distractor, respectively. Here, values (except those in Supplementary Figure 5) are defined as Expected Value (EV), calculated by multiplying magnitude and probability of reward.” (Lines 348-350)

Reviewer #2 Recommendations 2

The analysis drops trials in which the distractor was chosen. These trials are informative about the presence (or not) of relative valuation or other factors because they make such choices more (or less) likely. Ignoring them is another example of the analysis being misspecified.

We thank the reviewer for the suggestion and this is related to Major Issue 1 raised by the same reviewer. In brief, we adopted the same methods implemented by Cao and Tsetsos (Cao and Tsetsos, 2022) and that constrained us to applying a binomial model. Please refer to our reply to Major Issue 1 for more details.

Reviewer #2 Recommendations 3

Some questions and suggestions on statistics and computational modeling:

Have the authors looked at potential collinearity between the regressors in each of the GLMs?

We thank the reviewer for the comment. For each of the following GLMs, the average variance inflation factor (VIF) has been calculated as follows:

GLM2 using the Expected Value model:

Author response table 1.

GLM2 after replacing the utility function based on the normative Expected Value model with values obtained by using the composite model:

Author response table 2.

GLM3:

Author response table 3.

As indicated in the average VIF values calculated, none of them exceed 4, suggesting that the estimated coefficients were not inflated due to collinearity between the regressor in each of the GLMs.

Reviewer #2 Recommendations 4

- Correlation results in Figure 4. What is the regression line displayed on this plot? I suspect the regression line came from Pearson's correlation, which would be inconsistent with the Spearman's correlation reported in the text. A reasonable way would be to transform both x and y axes to the ranked data. However, I wonder why it makes sense to use ranked data for testing the correlation in this case. Those are both scalar values. Also, did the authors assess the influence of the zero integration coefficient on the correlation result? Importantly, did the authors redo the correlation plot after defining the utility function by the composite models?

We thank the reviewer for the suggestion. The plotted line in Figure 4 was based on the Pearson’s correlation and we have modified the text to also report the Pearson’s correlation result as well.

If we were to exclude the 32 participants with integration coefficients smaller than 1×10-6 from the analysis, we still observe a significant positive Pearson’s correlation [r(110)=0.202, p=0.0330].

Author response image 1.

Figure 4 after excluding 32 participants with integration coefficients smaller than 1×10-6.

“As such, we proceeded to explore how the distractor effect (i.e., the effect of (DV−HV)T obtained from GLM2; Figure 2c) was related to the integration coefficient (η) of the optimal model via a Pearson’s correlation (Figure 4). As expected, a significant positive correlation was observed [r(142)=0.282, p=0.000631]. We noticed that there were 32 participants with integration coefficients that were close to zero (below 1×10-6). The correlation remained significant even after removing these participants [r(110)=0.202, p=0.0330].” (Lines 207-212)

The last question relates to results already included in Supplementary Figure 5, in which the analyses were conducted using the utility function of the composite model. We notice that although there was a difference in integration coefficient between the multiplicative and additive groups, a correlational analysis did not generate significant results [r(142)=0.124, p=0.138]. It is possible that the relationship became less linear after applying the composite model utility function. However, it is noticeable that in a series of complementary analyses (Figure 5: r(142)=0.282, p=0.000631; Supplementary Figure 3: r(142)=0.278, p=0.000746) comparable results were obtained.

Reviewer #2 Recommendations 5

- From lines 163-165, were the models tested on only the three-option trials or both two and three-opinion trials? It is ambiguous from the description here. It might be worth checking the model comparison based on different trial types, and the current model fitting results do not tell an absolute sense of the goodness of fit. I would suggest including the correctly predicted trial proportions in each trial type from different models.

We thank the reviewer for the suggestion. We have only modeled the two-option trials and the key reason for this is because the two-option trials can arguably provide a better estimate of participants’ style of integrating attributes as they are independent of any distractor effects. This was also the same reason why Cao and Tsetsos applied the same approach when they were re-analyzing our data (Cao and Tsetsos, 2022). We have clarified the statement accordingly.

“We fitted these models exclusively to the Two-Option Trial data and not the Distractor Trial data, such that the fitting (especially that of the integration coefficient) was independent of any distractor effects, and tested which model best describes participants’ choice behaviours.” (Lines 175-178)

Reviewer #2 Recommendations 6

- Along with displaying the marginal distributions of each parameter estimate, a correlation plot of these model parameters might be useful, given that some model parameters are multiplied in the value functions.

We thank the reviewer for the suggestion. We have also generated the correlation plot of the model parameters. The Pearson’s correlation between the magnitude/probability weighting and integration coefficient was significant [r(142)=−0.259, p=0.00170]. The Pearson’s correlation between the inverse temperature and integration coefficient was not significant [r(142)=−0.0301, p=0.721]. The Pearson’s correlation between the inverse temperature and magnitude/probability weighting was not significant [r(142)=−0.0715, p=0.394].

“Our finding that the average integration coefficient  was 0.325 coincides with previous evidence that people were biased towards using an additive, rather than a multiplicative rule. However, it also shows rather than being fully additive ( =0) or multiplicative ( =1), people’s choice behaviour is best described as a mixture of both. Supplementary Figure 1 shows the relationships between all the fitted parameters.” (Lines 189-193)

Reviewer #2 Recommendations 7

Have the authors tried any functional transformations on amounts or probabilities before applying the weighted sum? The two attributes are on entirely different scales and thus may not be directly summed together.

We thank the reviewer for the comment. Amounts and probabilities were indeed both rescaled to the 0-1 interval before being summed, as explained in the methods (Line XXX). Additionally, we have now added and performed model fitting on an additional model with utility curvature based on the prospect theory (Kahneman & Tversky, 1979) and a weighted probability function (Prelec, 1998):

where  and  represent the reward magnitude and probability (both rescaled to the interval between 0 and 1), respectively.  is the weighted magnitude and  is the weighted probability, while  and  are the corresponding distortion parameters. This prospect theory (PT) model was included along with the four previous models (please refer to Figure 3) in a Bayesian model comparison. Results indicate that the composite model remains as the best account of participants’ choice behaviour (exceedance probability = 1.000, estimated model frequency = 0.720).

“Supplementary Figure 2 reports an additional Bayesian model comparison performed while including a model with nonlinear utility functions based on Prospect Theory (Kahneman & Tversky, 1979) with the Prelec formula for probability (Prelec, 1998). Consistent with the above finding, the composite model provides the best account of participants’ choice behaviour (exceedance probability = 1.000, estimated model frequency = 0.720).” (Lines 193-198)

Reviewer #3 (Recommendations For The Authors):

Reviewer #3 Recommendations 1

- In the Introduction (around line 48), the authors make the case that distractor effects can co-exist in different parts of the decision space, citing Chau et al. (2020). However, if the distractor effect is calculated relative to the binary baseline this is no longer the case.

- Relating to the above point, it might be useful for the authors to make a distinction between effects being non-monotonic across the decision space (within individuals) and effects varying across individuals due to different strategies adopted. These two scenarios are conceptually distinct.

We thank the reviewer for the comment. Indeed, the ideas that distractor effects may vary across decision space and across different individuals are slightly different concepts. We have now revised the manuscript to clarify this:

“However, as has been argued in other contexts, just because one type of distractor effect is present does not preclude another type from existing (Chau et al., 2020; Kohl et al., 2023). Each type of distractor effect can dominate depending on the dynamics between the distractor and the chooseable options. Moreover, the fact that people have diverse ways of making decisions is often overlooked. Therefore, not only may the type of distractor effect that predominates vary as a function of the relative position of the options in the decision space, but also as a function of each individual’s style of decision making.” (Lines 48-54)

Reviewer #3 Recommendations 2

- The idea of mixture models/strategies has strong backing from other Cognitive Science domains and will appeal to most readers. It would be very valuable if the authors could further discuss the potential level at which their composite model might operate. Are the additive and EV quantities computed and weighted (as per the integration coefficient) within a trial giving rise to a composite decision variable? Or does the integration coefficient reflect a probabilistic (perhaps competitive) selection of one strategy on a given trial? Perhaps extant neural data can shed light on this question.

We thank the reviewer for the comment. The idea is related to whether the observed mixture in integration models derives from value being actually computed in a mixed way within each trial, or each trial involves a probabilistic selection between the additive and multiplicative strategies. We agree that this is an interesting question and to address it would require the use of some independent continuous measures to estimate the subjective values in quantitative terms (instead of using the categorical choice data). This could be done by collecting pupil size data or functional magnetic resonance imaging data, as the reviewer has pointed out. Although the empirical work is beyond the scope of the current behavioural study, it is worth bringing up this point in the Discussion:

“The current finding involves the use of a composite model that arbitrates between the additive and multiplicative strategies. A general question for such composite models is whether people mix two strategies in a consistent manner on every trial or whether there is some form of probabilistic selection occurring between the two strategies on each trial such that only one strategy is used on any given trial while, on average, one strategy is more probable than the other. To test which is the case requires an independent estimation of subjective values in quantitative terms, such as by pupillometry or functional neuroimaging. Further understanding of this problem will also provide important insight into the precise way in which distractor effects operate at the single-trial level.” (Lines 275-282)

Reviewer #3 Recommendations 3

Line 80 "compare pairs of attributes separately, without integration". This additive rule (or the within-attribute comparison) implies integration, it is just not multiplicative integration.

We thank the reviewer for the comment. We have made adjustments to the manuscript to ensure that the message delivered within this manuscript is consistent.

“For clarity, we stress that the same mathematical formula for additive value can be interpreted as meaning that 1) subjects first estimate the value of each option in an additive way (value integration) and then compare the options, or 2) subjects compare the two magnitudes and separately compare the two probabilities without integrating dimensions into overall values. On the other hand, the mathematical formula for multiplicative value is only compatible with the first interpretation. In this paper we focus on attribute combination styles (multiplicative vs additive) and do not make claims on the order of the operations. More particularly, we consider whether individual differences in combination styles could be related to different forms of distractor effect.” (Lines 92-100)

Reviewer #3 Recommendations 4

- Not clear why the header in line 122 is phrased as a question.

We thank the reviewer for the suggestion. We have modified the header to the following:

“The distractor effect was absent on average” (Line 129)

Reviewer #3 Recommendations 5

- The discussion and integration of key neural findings with the current thesis are outstanding. It might help the readers if certain statements such as "the distractor effect is mediated by the PPC" (line 229) were further unpacked.

We thank the reviewer for the suggestion. We have made modifications to the original passage to further elaborate the statement.

“At the neuroanatomical level, the negative distractor effect is mediated by the PPC, where signal modulation described by divisive normalization has been previously identified (Chau et al., 2014; Louie et al., 2011). The same region is also crucial for perceptual decision making processes (Shadlen & Shohamy, 2016).” (Lines 250-253)

Reviewer #3 Recommendations 6

- In Fig. 3c, there seem to be many participants having the integration coefficient close to 0 but the present violin plot doesn't seem to best reflect this highly skewed distribution. A histogram would be perhaps better here.

We thank the reviewer for the suggestion. We have modified the descriptive plots to use histograms instead of violin plots.

“Figures 3c, d and e show the fitted parameters of the composite model: , the integration coefficient determining the relative weighting of the additive and multiplicative value ( , ); , the magnitude/probability weighing ratio ( , ); and , the inverse temperature ( , ). Our finding that the average integration coefficient  was 0.325 coincides with previous evidence that people were biased towards using an additive, rather than a multiplicative rule.” (Lines 186-191)

Author response:

The following is the authors’ response to the original reviews.

Summary of the changes

Changes in the manuscript were made to clarify some ambiguities raised by the reviewers and to improve the report following their recommendations. A summary of the main changes is listed below:

- The title was changed to better reflect the results of this study - Re-training the model on log transformed FACS scores.

- Testing the specificity of the FEPS to facial expression of pain within this experimental setup by comparing it to the activation maps obtained from the Warm stimulation condition.

- Testing for sensitization/habituation of the behavioral measures (FACS scores and pain ratings).

- Adding a section in the discussion to better address the limitations of this study and provide potential directions for future studies.

Other changes target areas where the original manuscript may have been ambiguous or lacked precision. To address these concerns, additional details have been incorporated, and certain terms have been revised to ensure a more precise and transparent presentation of the information.

Public Reviews:

Reviewer #1 (Public Review):

Picard et al. report a novel neural signature of facial expressions of pain. In other words, they provide evidence that a specific set of brain activations, as measured by means of functional magnetic resonance imaging (fMRI), can tell us when someone is expressing pain via a concerted activation of distinctive facial muscles. They demonstrate that this signature provides a better characterization of this pain behaviour when compared with other signatures of pain reported by past research. The Facial Expression of Pain Signature (FEPS) thus enriches this collection and, if further validated, may allow scientists to identify the neural structures subserving important non-verbal pain behaviour. I have, however, some reservations about the strength of the evidence, relating to insufficient characterization of the underlying processes involved.

We are thankful for the summary of our work. We are hopeful that the modifications made in the latest version effectively address these concerns. The changes are outlined in the summary above, and detailed in the following point-by-point response.

Strengths:

The study relies on a robust machine-learning approach, able to capitalise on the multivariate nature of the fMRI data, an approach pioneered in the field of pain by one of the authors (Dr. Tor Wager). This paper extends Wager's and other colleagues' work attempting to identify specific combinations of brain structures subserving different aspects of the pain experience while examining the extent of similarity/dissimilarity with the other signatures. In doing so, the study provides further methodological insight into fine-grained network characterization that may inspire future work beyond this specific field.

We are thankful for the positive comments.

Weaknesses:

The main weakness concerns the lack of a targeted experimental design aimed to dissect the shared variance explained by activations both specific to facial expressions and to pain reports. In particular, I believe that two elements would have significantly increased the robustness of the findings:

(1) Control conditions for both the facial expressions and the sensory input. An efficient signature should not be predictive of neutral and emotional facial expressions (e.g., disgust) other than pain expressions, as well as it should not be predictive of sensations originating from innocuous warm stimulation or other unpleasant but non-painful stimulation.

We do recognize the lack of specificity testing for the FEPS, especially towards negative emotional facial expressions. This would be relevant to test given the behavioural overlap between the facial expressions of pain and disgust, fear, anger, and sadness (Kunz et al., 2013; Williams, 2003). The experimental design used in this study did not include other negative states. However, we fully support the necessity of collecting data throughout those conditions, and we believe that the present study highlights the importance of such a demonstration. Future research should involve recording facial expressions while exposing participants to stimuli that elicit a range of negative emotions but, to our knowledge, such combination of fMRI and behavioural data is currently unavailable. As raised by the reviewer, this approach would allow us to assess the specificity of the FEPS to the facial expression evoked by pain compared to different affective states. We would like to emphasise that specificity and generalizability testing is a massive amount of work, requiring multiple studies to address comprehensively. A Limitations paragraph addressing this research direction has been added to the Discussion. A conclusion was added to the abstract as follows: “Future studies should explore other pain-relevant manifestations and assess the specificity of the FEPS against other types of aversive or emotional states.”

(2) Graded intensity of the sensory stimulation: different intensities of the thermal stimulation would have caused a graded facial expression (from neutral to pain) and graded verbal reports (from no pain to strong pain), thus offering a sensitive characterisation of the signal associated with this condition (and the warm control condition).

However, these conditions are missing from the current design, and therefore we cannot make a strong conclusion about the generalisability of the signature (regardless of whether it can predict better than other signatures - which may/may not suffer from similar or other methodological issues - another potential interesting scientific question!). The authors seem to work on the assumption that the trials where warm stimulation was delivered are of no use. I beg to disagree. As per my previous comment, warm trials (and associated neutral expressions) could be incorporated into the statistical model to increase the classification sensitivity and precision of the FEPS decoding.

The experience of pain can fluctuate for a fixed intensity or after controlling statistically for the intensity of the stimulation (Woo et al., 2017). Consistent with this, the current study focused on spontaneous facial expression in response to noxious thermal stimuli delivered at a constant intensity that produced moderate to strong pain in every participant. As the reviewer points out, this does not allow us to characterise and compare the stimulus-response function of facial expression and pain ratings. The advantage of the approach adopted is to maximise the number of trials where facial expression is more likely to occur, while ensuring that changes in facial expression and pain ratings are not confounded with changes in stimulus intensity. The manuscript has been revised to clarify that point. However, we do agree that it would be interesting to conduct more studies focusing on facial expression in response to a range of stimulus intensities. This discussion has been added to the Limitations paragraph.

Furthermore, following the reviewer’s suggestion, we performed complementary analyses on the warm trials in the proposed revisions. The dot product (FEPS scores) between the FEPS and the activation maps associated with the warm condition was computed. A linear mixed model was conducted to investigate the association between FEPS scores and the experimental condition (warm vs pain). The trials in the pain condition were divided into two conditions: null FACS scores (painful trials with no facial response; FACS scores = 0) and non-null FACS scores (painful trials with a facial response; FACS > 0). The details of this analysis have been added to the manuscript (see Response of the FEPS to pain and warm section in the Methods; lines 427 to 439) as well as the corresponding results (see Results and Discussion; lines 138 to 158). The FEPS scores were larger in the pain condition where a facial response was expressed, compared to both the pain condition without facial expression and the warm condition. These results confirmed the sensitivity of the FEPS to facial expression of pain.

Reviewer #2 (Public Review):

Summary:

The objective of this study was to further our understanding of the brain mechanisms associated with facial expressions of pain. To achieve this, participants' facial expressions and brain activity were recorded while they received noxious heat stimulation. The authors then used a decoding approach to predict facial expressions from functional magnetic resonance imaging (fMRI) data. They found a distinctive brain signature for pain facial expressions. This signature had minimal overlap with brain signatures reflecting other components of pain phenomenology, such as signatures reflecting subjective pain intensity or negative effects.

We appreciate this concise and accurate summary of our study.

Strength:

The manuscript is clearly written. The authors used a rigorous approach involving multivariate brain decoding to predict the occurrence and intensity of pain facial expressions during noxious heat stimulation. The analyses seem solid and well-conducted. I think that this is an important study of fundamental and clinical relevance.

Weaknesses:

Despite those major strengths, I felt that the authors did not suffciently explain their own interpretation of the significance of the findings. What does it mean, according to them, that the brain signature associated with facial expressions of pain shows a minimal overlap with other pain-related brain signatures?

We express our sincere gratitude for the valuable insights and constructive comments on the strengths and weaknesses of the current study. We thank reviewer 2 for the encouragement to reinforce our interpretation of the significance of the findings, while acknowledging the limitations raised by the three reviewers.

A few questions also arose during my reading.

Question 1: Is the FEPS really specific to pain expressions? Is it possible that the signature includes a facial expression signal that would be shared with facial expressions of other emotions, especially since it involves socio-affective regulation processes? Perhaps this question should be discussed as a limit of the study?

We acknowledge this limitation as outlined in response to Reviewer #1. We have incorporated a Limitations paragraph to provide a more in-depth discussion of this limitation and to explore potential future avenues (lines 225 to 268). Again, please note that the demonstration of specificity is an incremental process that requires a systematic comparison with other conditions where facial expressions are produced without pain. A concluding sentence was added to the abstract to encourage specificity testing in future studies. as indicated above.

Question 2: All AUs are combined together in a composite score for the regression. Given that the authors have other work showing that different AUs may be associated with different components of pain (affective vs. sensory), is it possible that combining all AUs together has decreased the correlation with other pain signatures? Or that the FEPS actually reflects multiple independent signatures?

The question raised is consistent with the work of Kunz, Lautenbacher, LeBlanc and Rainville (2012), and Kunz, Chen and Rainville (2020). In the current study, the pain-relevant action units were combined in order to increase the number of trials where a facial response to pain was expressed, thus enhancing the robustness of our analyses. Given the limited sample size, our current dataset is unfortunately insufficient to perform such analysis as there would not be enough trials to look at the action units separately or in subgroups. While the approach of combining the different AUs has proven to be valid and useful, we recognize the value of investigating potential independent signatures associated with the different AUs within the FEPS, and examining whether those signatures can lead to more similar patterns compared to previously developed pain signatures. This discussion has been included in the Limitations paragraph in the Discussion (lines 225 to 268).

Question 3: Is facial expressivity constant throughout the experiment? Is it possible that the expressivity changes between the beginning and the end of the experiment? For instance, if there is a habituation, or if the participant is less surprised by the pain, or in contrast if they get tired by the end of the experiment and do not inhibit their expression as much as they did at the beginning. If facial expressivity changes, this could perhaps affect the correlation with the pain ratings and/or with the brain signatures; perhaps time (trial number) could be added as one of the variables in the model to address this question.

The concern raised by the reviewer is legitimate. We conducted a mixed-effects model to assess the impact of successive trials and runs on facial expressivity. Results indicate that the FACS scores did not change significantly throughout the experiment, suggesting no notable effect of habituation or sensitization on the facial expressivity in our study. Details about the analysis and the results have been added to the Facial Expression section in the Methods (lines 335 to 346).

Reviewer #3 (Public Review):

In this manuscript, Picard et al. propose a Facial Expression Pain Signature (FEPS) as a distinctive marker of pain processing in the brain. Specifically, they attempt to use functional magnetic resonance imaging (fMRI) data to predict facial expressions associated with painful heat stimulation. The main strengths of the manuscript are that it is built on an extensive foundation of work from the research group, and that experience can be observed in the analysis of fMRI data and the development of the machine learning model. Additionally, it provides a comparative account of the similarities of the FEPS with other proposed pain signatures. The main weaknesses of the manuscript are the absence of a proper control condition to assess the specificity of the facial pain expressions, a few relevant omissions in the methodology regarding the original analysis of the data and its purpose, and a biased interpretation of the results.

I believe that the authors partially succeed in their aims, as described in the introduction, which are to assess the association between pain facial expression and existing pain-relevant brain signatures, and to develop a predictive brain activation model of the facial responses to painful thermal stimulation. However, I believe that there is a clear difference between those aims and the claim of the title, and that the interpretation of the results needs to be more rigorous.

We wish to express our appreciation for the insightful and constructive critique provided. The limitation pertaining to the absence of specificity testing had been addressed in response to Reviewer #1, and it has been incorporated into the manuscript (lines 251 to 258).

The commentary made by Reviewer #3 has drawn our attention to a critical concern, namely the potential misalignment between the study findings and our original title. Consequently, we have changed the title to “A distributed brain response predicting the facial expression of acute nociceptive pain”. We also revised the interpretation of the results in the discussion section and we have added a section on limitations.

Recommendations for the Authors:

Reviewer #1 (Recommendations For The Authors):

I hope the following comments will be useful to improve the manuscript.

Abstract

I felt the abstract could be more clear in terms of experimental or scientific questions, hypotheses/expectations, and findings. I also feel the abstract should briefly support the conclusive claim ("is better than...": how better? Or according to what criterion? This may be more relevant than the final conclusive general sentence that does not specifically address the significance of the findings).

The abstract was revised to reinforce the functional perspective adopted to interpret brain activity produced by noxious stimuli and predicting various pain-relevant manifestations. We also mention explicitly the other pain-relevant signatures against which the FEPS is compared in this report, and we added a concluding sentence highlighting the importance of assessing the specificity of the FEPS in future studies.

Introduction - background and rationale

I would postpone the discussion around pain signature and anticipate the one about the brain mechanisms of facial expressions of pain. This will allow you to reinforce the logical flow of rationale, literature gap/question, why the problem is important, and study aims. Only then go for a review of relevant literature on signatures before providing a more specific final paragraph about the study-specific questions, expectations, and implementation. At the moment this is limited to a single very descriptive short paragraph at the end of the intro.

The introduction was structured to guide the readers through a comprehensive understanding of different pain neurosignatures. The introduction aimed to establish a robust rationale for the subsequent analyses detailed in the results section. Indeed, the presentation of that literature ensured that the discussion around pain signatures is contextualised within a broader continuous framework. We acknowledge the reviewer’s comment on the limited description of the brain mechanisms of facial expression of pain. However, this was addressed in several previous reports of our laboratory (Kunz et al. 2011; Vachon-Presseau et al. 2016; Kunz, Chen, and Rainville 2020). We have added some more details about the brain mechanisms of facial expression, and highlighted those references in the first paragraph of the introduction.

Methods and Results

(1) Was there any indication of power based on the previous work or the other signature papers? If yes, how that would inform the present analysis?

The NPS was trained on 20 participants that experienced 12 trials at each of four different intensities. The assessment of the effect sizes was performed on the Neurological Pain Signature in Han et al. (2022). That study revealed a moderate effect size for predicting between-subject pain reports, and a large one for predicting within-subject pain reports. We trained our model on 34 participants that underwent 16 trials. We expected our results to show a smaller effect size as the current experimental design only allowed us to examine spontaneous changes in the facial expression, as noted in the comments made by Reviewer #1. However, the best way to calculate the unbiased effect size of the results presented in the current study would be to test the unchanged model on new independent datasets (see Reddan, Lindquist, and Wager, 2017). Unfortunately, such datasets do not currently exist.

(2) I would clarify to the reader what is meant by normal range of thermal pain and why is this relevant. Also, I did not find data about this assessment nor about the assessment of facial expressiveness (or reference to where it can be found).

We changed this formulation to “All participants included in this study had normal thermal pain sensitivity” and we added a few references. By targeting a healthy population with normal thermal pain sensitivity, our study sought to identify a predictive brain pattern related to facial expression evoked by typical responses to pain that could eventually be generalised to other individuals from the same population. Details about the assessment of facial expressiveness have been added in the appropriate section in the Methods.

(3) That pain ratings are only weakly associated with facial responses is, in its own right, an interesting finding, as a naïve reader would expect the two to be highly positively correlated. I'd suggest discussing this aspect (in reference to previous research) as it is interesting on both theoretical and empirical grounds.

The likelihood and the strength of pain facial expression generally increase with pain ratings in response to acute noxious stimuli of increasing physical intensities, thereby leading to a positive association between the two responses that is driven by the stimulus. However, the poor correlation or the dissociation between facial pain expression and pain rating is a very well known phenomenon that can be demonstrated easily using experimental methods where the stimulus intensity is held constant and spontaneous fluctuations are observed in both facial expression and pain ratings. This result was not discussed in the current manuscript as it was already addressed in the work of Kunz et al. (2011) and Kunz, Karos and Vervoot (2018). We added the references to these studies in the revised manuscript (lines 330 to 334).

(4) It may be worth having CIs throughout the whole set of analyses.

Thanks for the suggestions, this was an oversight. The confidence intervals have been added in the manuscript where applicable.

(5) I would clarify if there are two measures of the brain signature: dot-product and activation map. Relatedly, I cannot find where the authors explained what "FEPS pattern expression scores". Can the authors please clarify?

The clarification has been added in the manuscript (lines 413 to 414).

(6) There seems to be the assumption that the relationship between pain-relevant brain signatures and facial expressions of pain would be parametric and linear. However, this might not hold true. Did the authors test these assumptions?

We indeed decided to use a linear regression technique (i.e. LASSO regression) to model the association between the brain activity and the facial expression of pain. The algorithm choice was mainly based on the simplicity and the interpretability of that approach, and our limited number of observations. The choice was also coherent with previous studies in the domain (e.g. Wager et al., 2011; Wager et al., 2013; Krishnan et al. 2016; Woo et al., 2017). Using a linear model, we were able to predict above chance level the facial expression evoked by pain using the fMRI activation. However, it is legitimate to think that more complex non linear models can better capture the brain patterns predictive of that behavioural manifestation of pain.

(7) Did the authors assess whether the FACS were better to be transformed/normalised? More generally, I would report any data assessment/transformation that has not been reported.

Thank you for this highly relevant suggestion. FACS scores were indeed not normally distributed and the analyses were conducted again to predict the log transformed FACS scores. This transformation was effective to normalize the distribution (skewness = 0.75, kurtosis = -0.84). The predictive model was confirmed on transformed data.

(8) Page 12: I am not clear on whether all the signatures are included in the same model (like a multiple regression) or if separate regressions are calculated per signature. The authors seem to imply that several regressions have been computed (possibly one per comparison with each signature?).

The correlation between the FACS scores and the pain-related signatures was computed separately for each signature. This information has been clarified.

(9) MVPA: See my main comment about warm trials and experimental/statistical design. For example, the LASSO regression model for the pain trials could be compared with a model using warm trials besides (or instead of) the unfitted model. Otherwise, add the warm trials as another predictor or within the subject level in a dummy fixed factor comprising pain and warm trials.

The inclusion of warm trials in the model training would be inconsistent with the goal of the main analysis to predict the facial expression of pain when a noxious pain stimulus is presented. Secondary analyses were conducted to compare the response of the FEPS to the warm trials compared to noxious pain trials. The dot product between the FEPS and the activation maps (FEPS scores) associated with the warm condition was computed. A linear mixed model was conducted to investigate the association between FEPS scores and the experimental condition (warm vs pain). Additional contrasts compared the warm trials with the pain trials with and without pain facial expression. The details of this analysis have been added to the manuscript (see Response of the FEPS to pain and warm in the Methods) as well as the corresponding results (see Results and Discussion).

(10) I would clarify for the reader why the separate M1 analysis has been run. Although obvious, I feel the reader would benefit from the specific hypothesis about this control analysis being spelled out together with the other statistical hypotheses within the statistical design in a more streamlined manner.

We extended the discussion on the rationale of that analysis and its interpretation taking into account the most recent results using the log transformed FACS scores (lines 125 to 133).

(11) The mixed model aimed to assess the relationship between pain ratings FEPS scores and facial scores is a crucial finding. I believe it speaks to the importance of a more complete design, which I already highlighted. I have a couple of technical questions: did the authors assess random slopes too? And, what was the strategy used to determine the random effects structure?

The linear mixed model considered the participants as a random effect, with random intercepts, considering the grouping structure in our data (i.e., each participant completed multiple trials). The reported results in the original manuscript were considering fixed slopes. However, following the reviewer’s comment, we re-computed the mixed linear models allowing the slopes to vary according to the intensity ratings. The results were changed in the manuscript to represent the output of those models.

(12) The text from lines 63 to 67 could go in the methods.

We decided to include those lines within the Result and Discussion section to give the reader more specification about the FACS scores, as this term is subsequently referenced in the following part of the Results and Discussion section. We are concerned that putting this information only in the Methods section would disrupt the reading.

Reviewer #2 (Recommendations For The Authors):

p. 4-5. When you report the positive weight clusters, you follow up with a sentence specifying which cognitive processes those brain regions are typically associated with. However, when you report the negative weight clusters, you do not specify the cognitive processes typically associated with those brain areas. I think that providing that information would be helpful to the readers.

Thanks for noticing this omission. The information has been added in the most recent version of the manuscript (lines 119 to 121).

p. 9. You specify that the degree of expressiveness of participants was evaluated. How did you evaluate expressiveness? Did you use this variable in your analyses? Were participants excluded based on their degree of expressiveness?

Details about the assessment of facial expressiveness have been added in the appropriate section in the Methods (lines 285 to 289).

p. 10. You explain that two certified FACS-coders evaluated the video recordings to rate the frequency of AUs. Could you please provide more details about the frequency measure? I think that there are different ways in which this could have been done. For instance, were the videos decomposed into frames, and then the frequency measured by summing the number of frames in which the AU occurred? Or was it "expression-based", so one occurrence of an AU (frequency of 1) would correspond to the whole period between its activation onset and offset? Both ways have pros and cons. For example, if the frequency represents the number of frames, then it controls for the total duration of the AU activation within a trial (pro); but if there were multiple activations/deactivations of the AU within one trial, this will not be controlled for (con). And vice-versa with the second way of calculating frequency.

Details about the frequency scores have been added to the manuscript (lines 315 to 319).

p. 11. When you explained how you calculated the association between the facial expression of pain and pain-related brain signatures, I felt that there was some information missing. Did you use the thresholded maps (available in the published articles), or did you somehow have access to the complete, voxel-by-voxel, raw regression coefficient maps?

The unthresholded maps were used. The information has been clarified in the latest version of the manuscript, as well as the details about the availability of the maps (see Data Availability section at the end of the manuscript).

Reviewer #3 (Recommendations For The Authors):

Format

The authors will notice that many observations about the manuscript are related to missing information and a lack of graphical representations. I believe the topic and the content of the manuscript are too complex to condense into a short report.

Title

The claim of the title is simply not substantiated by the content of the manuscript. Demonstrating that the FEPS is a distinctive (i.e., specific) marker of pain processing requires a substantially different experimental design, with more rigorous controls and a broader set of painful stimulations. The manuscript would benefit from a more accurate title.

We agree that the title could better align with our findings. We modified the title accordingly : “A distributed brain response predicting the facial expression of acute nociceptive pain”.

Abstract

I find it puzzling that the authors claim that there is limited knowledge of the neural correlates of facial expression of pain given what they describe in the first paragraph of the introduction. Besides, they propose to reanalyze a dataset that has been extensively described in Kunz et al. (2011), which is unlikely to provide any new significant information.

We respectfully disagree with that comment. We considered that three articles (i.e., Kunz et al., 2011; Vachon-presseau et al., 2016; Kunz, Chen and Rainville, 2020) on the topic do constitute limited knowledge, especially if we compare it to the very large body of literature on the neural correlates associated with pain ratings. Except for these three studies, all the other citations pertain to behavioral studies on facial expression of pain, and do not examine the brain activity related to it. Furthermore, we believe that the complementary nature of the analyses performed in Kunz et al. (2011) and in this manuscript offers new insights into our understanding of facial expression in the context of pain. Indeed, the multivariate approach used in this study addresses some limitations present in Kunz et al. (2011) univariate analyses, mainly that it provides a quantifiable way to compare the similarity between different predictive patterns (Reddan and Wager, 2017). We submit that the assessment of the FEPS against several other pain-relevant signatures provides new and important information.

Furthermore, the abstract does not clearly state the aim, and the first line of the results does not match what the authors claim in the preceding line. The take-home message (last sentence) introduces the concept of a biomarker, which, as stated before, cannot be validated with the current data/experimental design. To put it in plain words, a given facial expression (or a composite score derived from a combination of expressions) cannot be a specific biomarker for pain, because a person can always mimic the same expression without feeling pain. Whether a given facial expression can be predicted from brain activity is a different issue, and whether that prediction can differentiate between painful and non-painful origins of the facial expression is another different issue. Unfortunately, neither of those issues can be tested with the current data/experimental design. The abstract would improve if the authors would circumscribe to what they actually tested, which is accurately described in the last sentence of the Introduction.

The abstract was revised accordingly. The term ‘biomarker’ was used in accordance with preceding studies in the field (see Reddan and Wager, 2017; Lee et al., 2021). Please note that we applied the same reasoning to fluctuations in pain expression as previous studies have applied to pain ratings. Of course, we can not dismiss the possibility of someone mimicking facial expressions. Similar reasoning applies to subjective reports, as individuals can intentionally overestimate their pain experience conveyed through verbal reports. This is another case of specificity testing that cannot be addressed in the present study (see new conclusion of the abstract and discussion of limitations). The challenge of pain assessment is a classical problem within both the scientific and the clinical literature. Here, we suggest that the consideration of multiple manifestations of pain is necessary to address this challenge and will provide a more comprehensive portrait of pain-related brain function.

Introduction

I believe that the Introduction would benefit from a strict definition of what is a marker/biomarker/neuromarkers (all those terms are used in the manuscript) and what are its desirable features (validity, reliability, specificity, etc.). I also believe that the Introduction (and the rest of the text) would benefit from a critical assessment of the term "signature". The Introduction describes four existing "signatures", all of them differing in the experimental condition in which acute nociceptive pain is studied, and proposes a fifth one. Keeping with the analogy, I'm wondering whether they should be called (pain) "signatures" if there is a different one for each experimental acute pain condition, and they are so dissimilar between them when they are tested on the same condition (this dataset).

The last part of that comment raises fundamental methodological potential limitations that should be addressed in more depth in another article. That point goes beyond the scope of a research article. Regarding the stability aspect of the signatures, most of the signatures have not been studied extensively. It is thus difficult to currently assess their reliability. However, Han et al. (2022) showed high within-individual test-retest reliability for the NPS across eight different studies. Given that pain is a multidimensional experience, it is not surprising to find different patterns of activation predictive of different aspects or dimensions of the pain experience (see Čeko et al., 2022 for a similar discussion applied to negative affect).

The authors state that "As an automatic behavioral manifestation, pain facial expression might be an indicator of activity in nociceptive systems, perceptual and evaluative processes, or general negative affect." Doesn't it reflect all three of them? (and instead of or?) Why "might"?

The original sentence has been modified as follows: “As an automatic behavioral manifestation, pain facial expression is considered to be an indicator of activity in nociceptive systems, and to reflect perceptual and affective-evaluative processes” (lines 65 to 67).

Methods

The pain scale should be described. Kunz et al. used a 0-100 scale, where 50 was the pain threshold. This is crucial to interpret the 75-80/100 score for the painful thermal intensity.

The description of the pain scale has been added to the manuscript (lines 299 to 300).

Ratings for warm and painful temperatures should be reported (ideally plotted with individual-trial/subject data). In the same line of reasoning, FACS scores should be reported as well (ideally plotted with individual-trial/subject data). It would be interesting to explore the across-trial variability of pain ratings and FACS scores. That is, do people keep giving the same ratings and making the same facial expression after 16 trials? How much variability is between trials and between subjects?

The point raised in that comment was already addressed in response to a comment made by Reviewer #1 (also see the new Figures S2 and S4; see also lines 335 to 346).

How come only painful trials are analyzed? What if the FEPS signature was the same for warm and painful stimulation, thus reflecting the settings (fMRI experiment, stimulation, etc.) rather than the brain response to the stimuli?

The point raised in that comment was already addressed in response to a comment made by Reviewer #1. There was no pain expression in the warm trials and the FEPS shows no response to warm trials. This is now illustrated in the new Figure S4B (see also lines 138 to 158).

The authors propose to predict the trial-by-trial FACS composite score from the pain ratings using a LMM. However, it is interesting that they aim for an almost constant within- and between-subject pain score (75-80/100) as stated in the Methods. This should theoretically render the linear model invalid since its first (and main) assumption would be that FACS should vary linearly with the pain score. Even if patients were not aware that the temperatures were constant across trials, the variation in pain scores should be explained by random noise for a constant stimulation intensity.

Reviewer #3 raises an important point that we need to clarify. Contrary to the expectation that FACS responses should be strongly correlated to pain ratings, we posited that these response channels depend at least in part on separate brain networks that may be differentially sensitive to a variety of modulatory mechanisms (attention, emotion, expectancy, motor priming, social context, etc.). This implies that part of the variance in FACS is independent from pain ratings. We, therefore, consider what Reviewer #3 refers to as random noise to be relevant and meaningful fluctuations reflecting endogenous processes influencing one’s experience of pain and differentially affecting various output responses.

I noticed that fMRI data was analyzed with SPM5 in the original paper (Kunz et al., 2011) and with SPM8 in this manuscript. Was fMRI data re-processed for this manuscript? Were there any differences between the original analysis and this one that might induce changes in the interpretation of results?

The data were indeed re-processed using SPM8, which was the most recent version available when we started the analyses reported here. We used trial-by-trial activation maps for MVPA, which differs from what was used in the previous study (contrast maps at the level of the conditions, not the trials). We have no reason to believe that the different versions will change the message of this manuscript since those versions do not differ significantly in terms of the fMRI preprocessing pipeline (see SPM8 release notes; https://www.fil.ion.ucl.ac.uk/spm/software/spm8/). Furthermore, the aim of this present study is not to compare the different analysis parameters implemented in SPM5 vs SPM8.

What is the rationale for including PVP in the comparison among signatures? The experimental settings in which it was devised are distant from those described here.

The inclusion of the PVP was aimed at enhancing our comparative analysis with the FEPS, as we sought to investigate the potential functional meaning of the FEPS. The PVP was developed to capture the aversive value of pain, a dimension that is conceptually proximal to the interpretation of the facial expression as a manifestation of the affective response to nociceptive pain.

The LASSO-PCR approach is, in my opinion, not a procedure for (brain) decoding in this context. It is accurately described in the section title as a method for multivariate pattern analysis, or as a variable selection and regularization method for a prediction model. Here, brain activity in specific areas related to pain processing can hardly be described as "encoded", and the method just helps select those activations relevant for explaining a certain outcome (in this case, facial expressions).

We understand the point made by reviewer #3. The term brain decoding was changed for multivariate pattern analysis in the latest version of the manuscript.

Details are missing with regards to the dataset split into training, validation, and testing.

Details about the training and testing procedure were added in the manuscript (lines 383 to 385).

This might just be ignorance from me, so I apologize in advance, but what are "contrast" fMRI images? They are mentioned three times in the text but not really described. Are they the "Pain > Warm" contrasts from the original paper?

We apologize for any confusion caused by the use of the term “contrast images” which suggests a direct comparison between two experimental conditions. We have replaced “contrast images” with “activation maps” to provide a more accurate description of the nature of the data used in the multivariate pattern analysis (lines 388 to 389).

In the "Facial expression" section, the authors run an LMM to test the association between pain ratings (response variable) and facial responses (explanatory variable). If I understand correctly, in the "Multivariate pattern analysis" section they test the association between facial composite scores (response variable) and pain ratings (explanatory variable), but they obtain different results.

The analyses were recomputed on the log transformed data, as mentioned previously in the response to reviewers 1-2. The first model (in the “Facial expression” section) used the log transformed FACS scores as a dependent variable, the pain ratings as the fixed effect, and the participants as the random effect. The results of that analysis suggested that the transformed facial expression scores were not significantly associated with the pain ratings (p = .07). The second model uses both the FEPS pattern expression scores and pain ratings as fixed effects to predict facial responses. This analysis showed the significant contribution of the FEPS to the prediction of FACS scores (p < .001) and no significant effect of the pain ratings. However, a significant interaction was found (p = .03) suggesting that the prediction of the pain facial expression by the FEPS may vary with pain ratings (i.e. moderator effect). Those results have been clarified in the “Multivariate pattern analysis” section in the Methods (lines 416 to 426).

In this same section, what are "FEPS pattern expression scores"? They are used three times in the text, but I could not find their description.

The FEPS pattern expression scores correspond to the dot product between the trial-by-trial activation maps and the unthresholded FEPS signature. This information has been added to the manuscript (lines 413 to 414).

It would not be far-fetched to hypothesize that FACS scores could be predicted using solely activity from the motor cortex. The authors attempted to do this, but only with information from M1. Why did they not use the entire motor cortex, or better, regions of the motor cortex directly linked with the AUs described in the manuscript?

The selection of the primary motor area (M1) was based on the results found in Kunz et al. (2011). In this study, M1 showed the strongest correlation with facial expression of pain. There are numerous possibilities of combinations of multiple brain regions considering a variety of criteria based on distributed networks involved in motor, affective, or pain-related processes. We limited our exploration to the region with the strongest hypothesis due to practical feasibility concerns.

Results and Discussion

As a general recommendation, results should present individual data whenever possible. For example, the association between signatures and facial expression should be plotted using scatterplots.

We have added figures showing individual data when it was applicable (Figure S2; Figure S4).

The authors state that the LASSO-PCR model accounts for the facial responses to pain. I believe this is an overstatement, considering:

- A Pearson's r of 0.49 is usually considered low/weak correlation (moderate at best). In the same line, an R2 of 0.17 means that only 17% of the variance is explained by the model.

More nuanced interpretation of the results has been added to the discussion. A section has been added to highlight the limitations of the study.

- Figure 1 needs to display individual subject data and the ideal regression line.

The model was trained using a k-fold cross-validation procedure. The regression lines thus represent the model’s prediction for each one of the 10 folds (i.e. each fold is trained and tested on a different subset of the data). A scatter plot including the ideal regression line computed across all trials and subjects was added in supplementary material to illustrate the relation between the FACS scores and the FEPS pattern expression scores (Figure S4).

- Looking at Figure 1, it is clear that the model has an intercept different from zero. This means that when the FACS score was zero (i.e., volunteers did not make any distinguishable facial expression), the model predicted a score larger than zero. This is not discussed in the manuscript, and in simple terms, it means that there are brain activation patterns when no discernible facial expression is being made by the volunteers. In the original paper by Kunz et al., two groups of subjects were categorized, and one of them was a facially low- or non-expressive group (n=13). This fact is not even mentioned in the manuscript.

The categorization in the previous report (Kunz et al., 2012) was based on a pre-experimental session. All subjects were included in the current analysis. This is now indicated in the Methods (lines 287 to 289).

- On the other end of the range in Figure 1, differences between the FACS scores near the maximum range (40) are underestimated by 23 to 33 points! I guess that the RMSE is smaller (6-7 points), because many FACS scores are concentrated on the low end of the scale.

This is a very interesting comment. A section discussing the limits of the model to predict the lower and higher FACS scores has been added in the manuscript (lines 232 to 250).

It is of course acceptable to interpret the low similarity between signatures as a sign that each signature describes a different mechanism related to pain processing. However, I believe that a complete discussion should contemplate other competing hypotheses. Considering that all signatures were developed using a similar painful thermal stimulation protocol, it is reasonable to expect larger similarities between signatures. The fact that they are so dissimilar could be a reflection of model overfit, i.e., all these signatures are just fitted to these particular experimental protocols and data, and do not generalize to brain mechanisms of pain processing.

We appreciate the pertinent observation. We have included a limitations section in which we discussed, among other considerations, the possible overfitting of models and the necessity of pursuing generalizability studies (lines 225 to 268).

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

This is an important study on the regulation of chlorophyll biosynthesis in rice embryos. It provides insights into the genetic and molecular interactions that underlie chlorophyll accumulation, highlighting the inhibition of OsGLK1 by OsNF-YB7 and the broader implications for understanding chloroplast development and seed maturation in angiosperms. The results presented, including mutation analysis, gene expression profiles, and protein interaction studies, provide convincing evidence for the function of OsNF-YB7 as a repressor in the chlorophyll biosynthesis pathway.

Thank you very much for your positive assessment of our manuscript. We have carefully revised the manuscript according to the reviewers’ valuable suggestions and comments. For more details, please see the point-to-point response to the reviewers below.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

This manuscript investigates the regulation of chlorophyll biosynthesis in rice embryos, focusing on the role of OsNF-YB7. The rigorous experimental approach, combining genetic, biochemical, and molecular analyses, provides a robust foundation for these findings. The research achieves its objectives, offering new insights into chlorophyll biosynthesis regulation, with the results convincingly supporting the authors' conclusions.

Strengths:

The major strengths include the detailed experimental design and the findings regarding OsNF-YB7's inhibitory role.

Weaknesses:

However, the manuscript's discussion on the practical implications for agriculture and the evolutionary analysis of regulatory mechanisms could be expanded.

Thank you for your insightful comments and suggestions. In the revised manuscript, we discussed the potential application of the chlorophyllous embryo (please see line 270-274). The presence of chlorophyll in the embryo facilitates photosynthesis at early developmental stages, potentially leading to improved seedling growth and vigor (Smolikova and Medvedev, 2016). In crops such as soybean and canola, green embryo is considered as a valuable trait due to its association with enhanced photosynthetic capacity, which consequently promotes fatty acid biosynthesis (Ruuska et al., 2004). However, chlorophyll degradation must be carefully managed during seed maturation to avoid negative effects on seed viability and meal quality (Chung et al., 2006). Interestingly, the green embryo of lotus (Nelumbo nucifera) is widely used as a food ingredient in Asian, Australia, and North America. It is employed in herbal medicine to treat nervous disorders, insomnia, and other conditions (Zhu et al., 2017; Ha et al., 2022), highlighting the significant potential value of the green embryo.

In many chloroembryophytes, such as Arabidopsis, the embryo occupies a large proportion of the seed. From an evolutionary perspective, the presence of chlorophyll in the embryo may promote adaptation in such chloroembryophytes because more reserves can be accumulated in the seed through active photosynthesis, better supporting the embryo development and subsequent seedling growth (Sela et al., 2020). On the other hand, some leucoembryophytes, such as rice, have persistent endosperm rich in storage reserves to nourish embryo development (Liu et al., 2022). Gaining the ability to accumulate chlorophyll in the embryo is unnecessary for such species. In agreement with this hypothesis, cholorophyllous embryos are more prevalent in non-endospermous seeds (Dahlgren, 1980). However, we would like to emphasize that the evolutionary force driving the divergence of chloroembryophytes and leucoembryophytes is currently almost completely unknown and deserves in-depth investigation in the future. We discussed the possible evolution of the ability to accumulate chlorophyll in the embryo, please find the details in Line 276-295.

Reviewer #2 (Public Review):

Summary:

The authors set out to establish the role of the rice LEC1 homolog OsNF-YB7 in embryo development, especially as it pertains to the development of photosynthetic capacity, with chlorophyll production as a primary focus.

Strengths:

The results are well-supported and each approach used complements each other. There are no major questions left unanswered and the central hypothesis is addressed in every figure.

Weaknesses:

There are a handful of sections that could use clarifying for readers, but overall this is a solidly composed manuscript.

The authors clearly achieved their aims; the results compellingly establish a disparity between how this system operates in rice and Arabidopsis. Conclusions are thoroughly supported by the provided data and interpretations. This work will force a reconsideration of the value of Arabidopsis as a model organism for embryo chlorophyll biosynthesis and possibly photosynthesis during embryo maturation more broadly, as rice is a major crop organism and it very clearly does not follow the Arabidopsis model. It will thus be useful to carry out similar tests in other organisms rather than relying on Arabidopsis and attempting to more fully establish the regulatory mechanism in rice.

Thank you very much for your positive comments. We have carefully revised the manuscript according to your and the other reviewers’ comments and suggestions. Particularly, we emphasized the necessary to carry out similar tests in other organisms rather than relying on Arabidopsis to better understand the regulatory mechanism in rice.

Reviewer #3 (Public Review):

Summary:

In this study, the authors set out to understand the mechanisms behind chlorophyll biosynthesis in rice, focusing in particular on the role of OsNF-YB7, an ortholog of Arabidopsis LEC1, which is a positive regulator of chlorophyll (Chl) biosynthesis in Arabidopsis. They showed that OsNF-YB7 loss-of-function mutants in rice have chlorophyll-rich embryos, in contrast to Arabidopsis LEC1 loss-of-function mutants. This contrasting phenotype led the authors to carry out extensive molecular studies on OsNF-YB7, including in vitro and in vivo protein interaction studies, gene expression profiling, and protein-DNA interaction assays. The evidence provided well supported the core arguments of the authors, emphasising that OsNF-YB7 is a negative regulator of Chl biosynthesis in rice embryos by mediating the expression of OsGLK1, a transcription factor that regulates downstream Chl biosynthesis genes. In addition, they showed that OsNF-YB7 interacts with OsGLK1 to negatively regulate the expression of OsGLK1, demonstrating the broad involvement of OsNF-YB7 in rice Chl biosynthetic pathways.

Strengths:

This study clearly demonstrated how OsNF-YB7 regulates its downstream pathways using several in vitro and in vivo approaches. For example, gene expression analysis of OsNF-YB7 loss-of-function and gain-of-function mutants revealed the expression of selected downstream chl biosynthetic genes. This was further validated by EMSA on the gel. The authors also confirmed this using luciferase assays in rice protoplasts. These approaches were used again to show how the interaction of OsNF-YB7 and OsGLK1 regulates downstream genes. The main idea of this study is very well supported by the results and data.

Weaknesses:

From an evolutionary perspective, it is interesting to see how two similar genes have come to play opposite roles in Arabidopsis and rice. It would have been more interesting if the authors had carried out a cross-species analysis of AtLEC1 and OsNF-YB7. For example, overexpressing AtLEC1 in an osnf-yb7 mutant to see if the phenotype is restored or enhanced. Such an approach would help us understand how two similar proteins can play opposite roles in the same mechanism within their respective plant species.

We appreciate your insightful comments and suggestions. It is a very interesting question whether AtLEC1 can fully restore osnf-yb7, given the possible functional divergence between the genes in terms of regulation of chlorophyll biosynthesis in the embryo. We have previously expressed OsNF-YB7 in the lec1-1 background in Arabidopsis, driven by the native promoter of LEC1 (Niu et al., 2021). We found that OsNF-YB7 could almost completely rescue the embryo defects in Arabidopsis, indicating that OsNF-YB7 plays a resemble role in rice as the LEC1 does in Arabidopsis (Niu et al., 2021). We sought to determine whether AtLEC1 can complement the chlorophyll defect in osnf-yb7. However, given the fact that osnf-yb7 shows severe callus induction defect, which is not surprising, because many studies have shown that LEC1 is indispensable for somatic embryo development in various plant species, we are struggling to obtain the genetic materials for analysis. We have to transform OsNF-YB7pro::AtLEC1 into the WT background first, and then cross the transformant with the osnf-yb7 mutant. This is a time-consuming process in rice, but hopefully we will able to isolate a line expressing OsNF-YB7pro::AtLEC1 in the osnf-yb7 background from the resulting segregating population.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

A minor comment regarding the chlorophyll contents quantification in the study. Line 87: "The results showed that WT had an achlorophyllous embryo throughout embryonic development,...." In the TEM result, chloroplast was not observed in the WT embryo sections, indicating a lack of chlorophyll-containing structures, contrary to what was found in the osnf-yb7 embryos where chloroplasts were observed.

The authors stated that the embryo morphologies and Chl autofluorescence data showed that WT had an achlorophyllous embryo throughout embryonic development. However, the quantification of Chl levels in Figure 1D and Figure 4C showed that WT does produce some chlorophylls, albeit at lower levels than osnf-yb7 or OSGLK-OX embryos (WT values in the two figures are slightly different). This discrepancy warrants clarification to ensure consistency and accuracy in the manuscript's findings.

We re-evaluated the Chl content in the embryos of WT and OsGLK1-OX mature seeds. The result confirmed our previous finding that WT embryos produce a small amount of chlorophyll (please see the updated Fig. 4C). Notably, we observed that the dark-grown etiolated plants still have measurable chlorophyll content as reported in many studies (for example, Wang et al., 2017; Yoo et al., 2019), suggesting that there is potential bias in measuring chlorophyll content using an absorbance-based approach. We assume this possibly explains the concern you have raised.

Reviewer #2 (Recommendations For The Authors):

Mild editing for grammar is needed throughout, e.g. line 73, "It is still a mysterious why plant species".

We have carefully edited the grammar.

As a minor point, the placement of figure panels, such as in Figure 1, is not always intuitive.

Thank you for your suggestion. This figure has been revised as suggested. Please see the updated Fig. 1.

What is the significance of the two GFP mutants in Figures 2C and 2D? Is one of those the mislabeled Flag mutant?

The lines showed in Fig. 2C and D were not mislabeled. They were two independent transgenic events, both of which showed that OsNF-YB7 inhibited the expression of OsPORA and OsLHCB4 in rice. The transgenic lines overexpressing OsNF-YB7 tagging with the 3× Flag (NF-YB7-Flag) were also used for this experiment. In agreement, OsPORA and OsLHCB4 were significantly downregulated in the three independent NF-YB7-Flag lines (Fig. S4C), confirming the results showed in Fig. 2C and D.

In Figures 2G and 2H, what is that enormous band at the bottom of the gel?

The bands at the bottom of the gel were free probes. We indicated this in the revised figure.

Not until the Materials and Methods section did I realize that any of this study was being done in tobacco; the Introduction implies it's rice vs. Arabidopsis and it might be a good idea to mention the organism of study somewhere before Figure 6.

We apologize for any confusion caused by our previous writing. While the majority of this study was performed with rice plants or protoplasts, the split complementary LUC assays and BiFC assays were performed with tobacco. We have specified these in the revised manuscript as suggested.

Reviewer #3 (Recommendations For The Authors):

It would be nice if the author could show what the phenotype is in AtLEC1 OX in osnf-yb7 and also OsNF-YB7 OX in atlec1 mutants.

Thank you for your suggestion. We have previously expressed OsNF-YB7 in the lec1-1 background of Arabidopsis, driven by the native promoter of Arabidopsis LEC1 (Niu et al., 2021). Since OsNF-YB7 could rescue the embryo morphogenesis defects in Arabidopsis (Niu et al., 2021), we assumed that OsNF-YB7 plays a similar role in rice as the LEC1 does in Arabidopsis. However, it remains unknown whether expression of LEC1 in osnf-yb7 may restore the chlorophyllous embryo phenotype in rice. As the generation of genetic material is time-consuming, and especially given the fact that osnf-yb7 has a severe callus induction defect, we are struggling to obtain the complementary line for analysis. We have to transform OsNF-YB7pro::AtLEC1 in a WT background first, and then cross the transformant with the osnf-yb7 mutant. Hopefully, we will be able to isolate a line expressing OsNF-YB7pro::AtLEC1 in osnf-yb7 background, from the derived segregating population. We discussed the reviewer’s concern in the revised manuscript, please see Line 369-376.

Line 46, I think it is vague to mention that 'Like most plant species'. Some species might have different copy numbers, for example, a single GLK in liverwort M. polymorpha.

The statement has been revised. Please see Line 46.

Figures 2F and 5B, why was only one promoter region used for OsLHCB4? It would be better to have more regions like OsPORA.

Thank you for your comments. Here, we have examined more promoter regions (P1, P2 and P3) in the revised manuscript as suggested, among which, the previously selected promoter region (P3) contains both the G-box and CCAATC motifs that can be potentially recognized by GLK1. Consistent to our previous report, the results showed that OsNF-YB7 (left) and OsGLK1 (right) were associated with the P3 region, but showed no significant differences in the other probes. Please see the results in Fig. 2F and Fig. 5B of the revised manuscript.

Legend of Figures 2G, H, OsPORA (I), and OsLHCB (J) should be (G) and (H) respectively.

Corrected.

References

Chung, D.W., Pruzinska, A., Hortensteiner, S., and Ort, D.R. (2006). The role of pheophorbide a oxygenase expression and activity in the canola green seed problem. Plant Physiol 142, 88-97.

Ha, T., Kim, M.S., Kang, B., Kim, K., Hong, S.S., Kang, T., Woo, J., Han, K., Oh, U., Choi, C.W., and Hong, G.S. (2022). Lotus Seed Green Embryo Extract and a Purified Glycosyloxyflavone Constituent, Narcissoside, Activate TRPV1 Channels in Dorsal Root Ganglion Sensory Neurons. J Agric Food Chem 70, 3969-3978.

Liu, J., Wu, M.W., and Liu, C.M. (2022). Cereal Endosperms: Development and Storage Product Accumulation. Annu Rev Plant Biol 73, 255-291.

Niu, B., Zhang, Z., Zhang, J., Zhou, Y., and Chen, C. (2021). The rice LEC1-like transcription factor OsNF-YB9 interacts with SPK, an endosperm-specific sucrose synthase protein kinase, and functions in seed development. Plant J 106, 1233-1246.

Ruuska, S.A., Schwender, J., and Ohlrogge, J.B. (2004). The capacity of green oilseeds to utilize photosynthesis to drive biosynthetic processes. Plant Physiol 136, 2700-2709.

Sela, A., Piskurewicz, U., Megies, C., Mene-Saffrane, L., Finazzi, G., and Lopez-Molina, L. (2020). Embryonic Photosynthesis Affects Post-Germination Plant Growth. Plant Physiol 182, 2166-2181.

Smolikova, G.N., and Medvedev, S.S. (2016). Photosynthesis in the seeds of chloroembryophytes. Russ J Plant Physl+ 63, 1-12.

Wang, Z., Hong, X., Hu, K., Wang, Y., Wang, X., Du, S., Li, Y., Hu, D., Cheng, K., An, B., and Li, Y. (2017). Impaired Magnesium Protoporphyrin IX Methyltransferase (ChlM) Impedes Chlorophyll Synthesis and Plant Growth in Rice. Front Plant Sci 8, 1694.

Yoo, C.Y., Pasoreck, E.K., Wang, H., Cao, J., Blaha, G.M., Weigel, D., and Chen, M. (2019). Phytochrome activates the plastid-encoded RNA polymerase for chloroplast biogenesis via nucleus-to-plastid signaling. Nat Commun 10, 2629.

Zhu, M., Liu, T., Zhang, C., and Guo, M. (2017). Flavonoids of Lotus (Nelumbo nucifera) Seed Embryos and Their Antioxidant Potential. J Food Sci 82, 1834-1841.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Reviews):

Summary: 

The authors use a combination of biochemistry and cryo-EM studies to explore a complex between the cap-binding complex and an RNA binding protein, ALYREF, that coordinates mRNA processing and export.

Strengths: 

The biochemistry and structural biology are supported by mutagenesis which tests the model in vitro. The structure provides new insight into how key events in RNA processing and export are likely to be coordinated.

Weaknesses: 

The authors provide biochemical studies to confirm the interactions that they identify; however, they do not perform any studies to test these models in cells or explore the consequences of mRNA export from the nucleus. In fact, several of the amino acids that they identified in ALYREF that are critical for the interaction, as determined by their own biochemical studies, are conserved in budding yeast Yra1 (residues E124/E128 are E/Q in budding yeast and residues Y135/V138/P139 are F/S/P), where the impact on poly(A) RNA export from the nucleus could be readily evaluated. The authors could at least mention this point as part of the implications and the need for future studies. No one seems to have yet targeted any of these conserved residues, so this would be a logical extension of the current work.

We thank the reviewer for the feedback on our work. ALYREF coordinates pre-mRNA processing and export through interactions with a plethora of mRNA biogenesis factors including the DDX39B subunit of the TREX complex, CBC, EJC, and 3’ processing factors. ALYREF mediates the recruitment of the TREX complex on nascent transcripts which depends on its interactions with both CBC and EJC. Our work and studies by others indicate that ALYREF uses overlapping interfaces including both the N-terminal WxHD motif and the RRM domain to bind CBC and EJC. Thus, ALYREF mutants deficient in CBC interaction will also disrupt the ALYREF-EJC interaction and are not ideal for functional studies. In addition, the CBC plays important roles in multiple steps of mRNA metabolism through interactions with a plethora of factors, which often interact competitively with CBC. Identification of separation-of-function mutations on CBC or ALYREF that specifically disrupt their interaction but not other cellular complexes containing CBC or ALYREF would be an important future area to test the model in cells. 

We appreciate the reviewer’s insightful comments regarding yeast Yra1. Thus far, the physical and functional connection between Yra1 and CBC in yeast has not been demonstrated. There are major differences between yeast Yra1 and human ALYREF. Given the lack of an EJC in S. cerevisiae, it is unclear whether Yra1 acts in a similar manner as human ALYREF. In addition, Yra1 does not contain a WxHD motif in its N-terminal unstructured region, which is involved in CBC and EJC interactions in ALYREF. Characterization of the Yra1-CBC interaction will be an interesting future direction. We now include a discussion about yeast Yra1 in the newly added “Conclusion and perspectives” section. 

Specific suggestions:

The authors could put their work in context by speculating how some of the amino acids that they identify as being critical for the interactions they identify could contribute to cancer. For example, they mention mutations of interacting residues in NCBP2 are associated with human cancers, pointing out that NCBP2 R105C amino acid substitution has been reported in colorectal cancer and the NCBP2 I110M mutation has been found in head and neck cancer. Do the authors speculate that these changes would decrease the interaction between NCBP2 and ALYREF and, if so, how would this contribute to cancer? They also mention that a K330N mutation in NCBP1 in human uterine corpus endometrial carcinoma, where Y135 on the α2 helix of mALYREF2 makes a hydrogen bond with K330 of NCBP1. How do they speculate loss of this interaction would contribute to cancer?

In the revised manuscript, we include a discussion about these CBC mutants found in human cancers in the “Conclusion and perspectives” section. We think some of these CBC mutants, such as NCBP-1 K330N, could reduce interaction with ALYREF. Compromised CBC-ALYREF interaction will affect the recruitment of the TREX complex on nascent transcripts and cause dysregulation of mRNA export. In addition, that could also change the partition of CBC and ALYREF in different cellular complexes and cause perturbation of various steps in mRNA biogenesis that are regulated by CBC and ALYREF. Thus far, it is unclear whether and how loss of the CBC-ALYREF interaction directly contributes to cancer. Our work and that of others provide molecular insights to test in future studies. 

Reviewer #2 (Public Reviews):

Summary: 

In this manuscript, Bradley and his colleagues represented the cryo-EM structure of the nuclear cap-binding complex (CBC) in complex with an mRNA export factor, ALYREF, providing a structural basis for understanding CBC regulating gene expression.

Strengths: 

The authors successfully modeled the N-terminal region and the RRM domain of ALYREF (residues 1-183) within the CBC-ALYREF structure, which revealed that both the NCBP1 and NCBP2 subunits of the CBC interact with the RBM domain of ALYREF. Further mutagenesis and pull-down studies provided additional evidence to the observed CBC-ALYREF interface. Additionally, the authors engaged in a comprehensive discussion regarding other cellular complexes containing CBC and/or ALYREF components. They proposed potential models that elucidated coordinated events during mRNA maturation. This study provided good evidence to show how CBC effectively recruits mRNA export factor machinery, enhancing our understanding of CBC regulating gene expression during mRNA transcription, splicing, and export. 

Weaknesses: 

No in vivo or in vitro functional data to validate and support the structural observations and the proposed models in this study. Cryo-EM data processing and structural representation need to be strengthened. 

We appreciate the reviewer’s comments and suggestions. The fact that ALYREF uses highly overlapped binding interfaces for CBC and EJC interactions prevents us from a clear functional dissection of the ALYREF-CBC interaction using in vitro assays or in cells at the current stage. Please also see our response to Reviewer 1. 

In this revised manuscript, we have reprocessed the cryo-EM data using a different strategy which yields significantly improved maps. We have made improvements to the presentation of the structural work based on the reviewer’s specific comments. 

Reviewer #3 (Public Reviews):

Summary: 

The authors carried out structural and biochemical studies to investigate the multiple functions of CBC and ALYREF in RNA metabolism.

Strengths: 

For the structural study part, the authors successfully revealed how NCBP1 and NCBP2 subunits interact with mALYREF (residues 1-155). Their binding interface was then confirmed by biochemical assays (mutagenesis and pull-down assays) presented in this study. 

Weaknesses: 

The authors did not provide functional data to support their proposed models. The authors should include more details regarding the workflow of their cryo-EM data processing in the figure. 

We thank the reviewer for the comments. We completely agree that testing the proposed models in cells would be ideal. However, as we also respond to the other reviewers, functional studies are premature at the current stage because both ALYREF and CBC are components of many cellular complexes that regulate mRNA metabolism. Separation-of-function mutations on CBC or ALYREF first need to be identified in future studies for further investigation. Please also see our response to Reviewer 1. 

As suggested by the reviewer, we have included more details of the cryo-EM workflow in this revised manuscript. We have also included various validation measures including 3DFSC analyses, map vs model FSC curves, and representative density maps at various protein-protein binding interfaces. 

Recommendations for the Authors:

Reviewer #1 (Recommendations for the Authors):

Major points:

The authors should take advantage of Figure 1, which shows the domain structures of NCBP1, NCBP2, and ALYREF to indicate for the reader specifically which protein domains are included in the biochemical and structural analyses. In the current version of the manuscript, there is plenty of space to indicate below each domain structure precisely what regions are included.

In this revised manuscript, we have revised Figure 1A to indicate the protein constructs used in this work. 

Although it is fine to combine the Results and Discussion, the authors should really offer a concluding paragraph to highlight the novel results from this study and put the results in context.

We thank the reviewer for the recommendation. We now include a “Conclusion and perspectives” section in this revised manuscript.  

Minor comments:

Page 5, last sentence (and others) starts a sentence with the word "Since" when likely "As" which does not imply a time element to the phrase, is the correct word.

"Since the ALYREF/mALYREF2 interaction with the CBC is conserved and mALYREF2 exhibits better solubility, we focused on mALYREF2 in the cryo-EM investigations."

Would be more correct as: "As the ALYREF/mALYREF2 interaction with the CBC is conserved and mALYREF2 exhibits better solubility, we focused on mALYREF2 in the cryo-EM investigations."

We thank the reviewer for the comments. We have made the corrections. 

The word 'data' is plural so the sentence at the bottom of p.9 that includes the phrase "...in vivo data shows.." should read "..in vivo data show.." 

Corrected in the revised manuscript.

Reviewer #2 (Recommendations for the Authors):

Major points:

(1) The authors claimed the improved solubility of mouse ALYREF2 (mALYREF2, residues 1-155) compared to the previously employed ALYREF construct. However, human ALYREF has already been purified successfully for pull down assay, indicating soluble human ALYREF obtained, why not use human ALYREF directly? Please clarify. 

Pull-down studies were performed with GST-tagged ALYREF. For cryo-EM studies, untagged ALYREF is preferred to avoid potential issues that may arise from the expression tag. However, untagged ALYREF is less soluble than GST-tagged ALYREF and is not amenable for structural studies. We have revised the text to clarify this point. 

(2) The authors confirmed CBC-ALYREF interfaces through mutagenesis and pull-down assays in vitro. However, it would be more informative and interesting to include functional assays in vitro or/and in vivo with mutagenesis. 

We completely concur with the reviewer that testing the proposed models in vitro and in vivo would be important. However, as we pointed out in our response to public reviews, the highly overlapped binding interfaces on ALYREF for CBC and EJC interactions pose a great challenge for functional studies. Furthermore, both ALYREF and CBC are multifunctional factors and interact with a number of partners. Ideally, separation-of-function mutants that specifically disrupt the CBC-ALYREF interaction but not others need to be identified in future studies in order to perform functional studies. 

(3) About cryo-EM data processing and structural representation:

(1) In the description of the cryo-EM data processing, the authors claimed they did heterogeneous refinement, homogenous refinement, and then local refinement. This reviewer is puzzled by this process because the normal procedure should be non-uniform refinement following homogenous refinement. If the authors did not perform non-uniform refinement, they should do it because it would significantly improve the quality and resolution of cryo-EM maps. In addition, the right local refinement should include mask files and only show the density/map of the local region. 

We thank the reviewer for the suggestions. In response to the reviewer’s comment on the preferred orientation issue (point 5 below), we reprocessed the cryo-EM data and obtained significantly improved cryo-EM maps. In this revised manuscript, the CBC-mALYREF map was refined using homogeneous refinement; the CBC map was refined using homogenous refinement followed by non-uniform refinement. Refinement masks are included in Figure 2-figure supplement1. 

(2) Further local refinements with signal subtraction should be performed to improve the density and resolution of mALYREF2. 

We tested local refinement with or without signal subtraction using masks covering mALYREF2 and various regions of CBC. Unfortunately, this approach did not improve the density of mALYREF2. We suspect that the small size of mALYREF2 (77 residues for the RRM domain) and the intrinsic flexibility of CBC are the limiting factors in these attempts. 

(3) Figures with cryoEM map showing the side chains of the residues on the CBC-mALYREF2 interface should be included to strengthen the claims. Authors could add the map to Figure 3b/c or present it as a supplementary figure.

We include new supplementary figures (Figure 3-figure supplement 1) to show the electron densities corresponding to the views in Figure 3B and 3C. Residues labeled in Figure 3B and 3C are shown in sticks in these supplementary figures.

(4) For cryo-EM date processing, the authors have omitted lots of important details. Could the authors elaborate on the data processing with more details in the corresponding Figure and Methods Sections? Only one abi-initial model from the picked good particles was displayed in the figure. Are there any other different conformations of 3D classes for the dataset? In addition, too few classes have been considered in 3D classification, more classes may give a class with better density and resolution.

We thank the reviewer for the comments. We have reprocessed the cryo-EM data. A major change is to use Topaz for particle picking. We now include more details for data processing in Figure 2-figure supplement 1 and the method section. The cryo-EM sample is relatively uniform. Ab-initio reconstruction and heterogenous refinement yielded only one good class and the other classes are “junk” classes (omitted in the workflow figure). No major conformational changes were observed throughout the multiple rounds of heterogenous refinement for both CBC and CBCmALYREF2. In this revised manuscript, we have been able to obtain significantly improved maps through the new data processing strategy employing Topaz as illustrated in Figure 2-figure supplement 1 to 5.

(5) Angular distribution plots should be included to show if there is a preferred orientation issue. Based on the presented maps in validation reports, there may exist a preferred orientation issue for the reported two cryo-EM maps. Detailed 3D-Histogram and directional FSC plots for all the cryo-EM maps using 3DFSC web server should be presented to show the overall qualities (https://www.nature.com/articles/nmeth.4347 and https://3dfsc.salk.edu/).

We thank the reviewer for the recommendations. In response to the reviewer’s comment on the preferred orientation issue, we reprocessed the cryo-EM data. Topaz was used for particle picking instead of template picking. 3DFSC analyses indicate that the new CBC-mALREF2 map has a sphericity of 0.946, which is a significant improvement from the previous map which has a sphericity of 0.815. Consistently, the maps presented in this revised manuscript show significantly improved densities. We now include angular distribution and 3DFSC analyses of the EM maps (Figure 2-figure supplement 2 and 4). 

(6) Figures of model-to-map FSCs need to be present to demonstrate the quality of the models and the corresponding ones (model resolution when FSC=0.5) should also be included in Table 1. The accuracy of the model is important for structural explanations and description.

The model-to-map FSCs are now included in Figure 2-figure supplement 3A and 5A. The model resolutions of CBC-mALYREF2 and CBC are estimated to be 3.5 Å and 3.6 Å at an FSC of 0.5. These numbers are now included in Table 1. 

(7) In addition, figures of local density maps with different regions of the models, showing side chains, are necessary and important to justify the claimed resolutions. 

We now include density maps overlayed with residue side chains at various regions. For the CBCmALYREF2 map, density maps are shown at the mALYREF2-NCBP1 interfaces (Figure 3-figure supplement 1A and 1B), mALYREF2-NCBP2 interface (Figure 3-figure supplement 1C), NCBP1NCPB2 interface (Figure 2-figure supplement 5B), and the region near m7G (Figure 2-figure supplement 5C). For the CBC map, density maps are shown at the NCBP1-NCPB2 interface (Figure 2-figure supplement 3B) and the region near m7G (Figure 2-figure supplement 3C). 

Minor points:

(1) A figure superimposing the models from the CBC-mALYREF2 amp and mALYREF2 alone map is necessary to present that there are no other CBC binding-induced conformational changes in CBC except the claimed by the authors. In addition, a figure showing the density of m7GpppG should be included as well.  

Overlay of CBC and CBC-mALYREF2 models is now presented in Figure 2-figure supplement 3D. Comparing CBC and CBC-mALYREF2, NCBP1 and NCBP2 have a RMSD of 0.32 Å and 0.30 Å, respectively. The density maps near the M7G cap analog are shown in Figure 2-figure supplement 3C for CBC and Figure 2-figure supplement 5C for CBC-mALYREF2. 

(2) Authors obtained the two maps from one dataset, so "we first determined" and "we next determined" (page 6) should be replaced with something like "One class of 3D cryo-EM map revealed' and "Another class of 3D cryo-EM map defined". 

We have revised the text as suggested by the reviewer.  

(3) In 'Abstract', 'a mRNA export factor' should be 'an mRNA export factor'. 

Corrected in the revised manuscript.

(4) In 'Abstract', the final sentence 'Comparison of CBC- ALYREF to other CBC and ALYREF containing cellular complexes provides insights into the coordinated events during mRNA transcription, splicing, and export' doesn't read smoothly, I would suggest revising it to 'Comparing CBC-ALYREF with other cellular complexes containing CBC and/or ALYREF components provides insight into the coordinated events during mRNA transcription, splicing, and export.' 

We thank the reviewer for the recommendation and have revised accordingly. 

(5) In paragraph 'CBC-ALYREF and viral hijacking of host mRNA export pathway', line 6, the sentences preceding and following the term 'However' indicate a progressive or parallel relationship, rather than a transitional one. To enhance the coherence, I would suggest replacing 'However' with 'Furthermore' or 'In addition'. 

Corrected in the revised manuscript.

(6) In both Figure 5 and Figure 6, the depicted models are proposed and constructed exclusively through the comparison of the CBC-partial ALYREF with other cellular complexes containing components of CBC and/or ALYREF, which need to be confirmed by more studies. To prevent potential confusion and misunderstandings, it is recommended to replace the term 'model' with 'proposed model'. 

Corrected in the revised manuscript.

Reviewer #3 (Recommendations for the Authors):

Major points:

(1) In the Results and Discussion section, the authors mentioned "Recombinant human ALYREF protein was shown to interact with the CBC in RNase-treated nuclear extracts." However, they used mouse ALYREF for cryo-EM investigations. Can the authors include an explanation for this choice during the revision?  

In our work, we used a mixture of glutamic acid and arginine to increase the solubility of GSTALYREF. For cryo-EM studies, we use untagged ALYREF to avoid potential issues that may arise from the expression tag. However, untagged ALYREF is less soluble than GST-tagged ALYREF and is not suitable for structural studies in standard buffers. We have made further clarification on this point in this revised manuscript. 

(2) In the paragraph on "CBC-ALYREF interfaces", the authors stated "For example, E97 forms salt bridges with K330 and K381 of NCBP1. Y135 on the α2 helix of mALYREF2 makes a hydrogen bond with K330 of NCBP1. The importance of this interface between ALYREF and NCBP1 is highlighted by a K330N mutation found in human uterine corpus endometrial carcinoma." I fail to see a strong connection between their structural observations and previous findings regarding the role of a K330N mutation found in human uterine corpus endometrial carcinoma. The authors should add more words to thread these two parts.  

In response to the reviewer’s comment, we now move the discussion of these CBC mutants to the newly added “Conclusion and perspectives” section. 

(3) The authors should include side chains of the residues in their figure of Local resolution estimation and FSC curves, especially when they are presenting the binding interface between two components. 

We have now included density maps that are overlayed with structural models showing side chains of critical residues. These maps include the NCBP1-mALYREF2 interfaces (Figure 3-figure supplement 1A and 1B), NCBP2-mALYREF2 interface (Figure 3-figure supplement 1C), NCBP1NCBP2 interface (Figure 2-figure supplement 3B and 5B), and the m7G cap region (Figure 2figure supplement 3C and 5C). 

Minor points: 

(1) Some grammatical mistakes need to be corrected. For example, it is "an mRNA" instead of "a mRNA".  

Corrected in the revised manuscript.

(2) The authors can provide more information for the audience to know better about ALYREF when it first appears in the 5th line in the Abstract section. For example, "It promotes mRNA export through direct interaction with ALYREF, a key mRNA export factor, ...". 

We have revised the sentence based on the reviewer’s comment.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Recommendations For The Authors):

Some of the data is problematic and does not always support the authors' conclusions:

(1) Fig. 1K and H are identical.

Thank you for pointing out this problem in manuscript. We apologize for this unintentional mistake and have replaced Fig. 1K.

(2) The graph in Figure 2B contradicts the text. It is not obvious how the image was quantified to produce the histological score graph..

We thank the reviewer for pointing out this problem in manuscript, as the reviewer suggested, we have replaced the Figure 2B.

(3) In Figures 2C and D, there is no clear pattern of changes in pro-inflammatory or anti-inflammatory cytokines, despite the authors' claims in the text.

We appreciate the comment, we think the reason is that the level of cytokines in the tissue is low, so the pattern of changes is not obvious.

(4) It is unclear why the anti-dsDNA antibody does not stain the nucleus in Figure 4B. The staining with anti-dsDNA and DAPI does not match well. Figure 5H shows there is still lots of cytosolic DNA in OGT-/- HCF-1-C, measured by DAPI. These data do not support the authors' conclusion that HCFC600 eliminates cytosolic DNA accumulation (line 229). There is no support for the authors' claim that HCF-1 restrains the cGAS-STING pathway (line 330).

We thank these insightful comments, the most critical step in staining cytosolic DNA is to proceed to a low-permeabilization as to allow the antibody to cross the cellular membrane but not the nuclear membrane, that’s why the anti-dsDNA antibody does not stain the nucleus. In Figure 5H, we think we used a high concentrated DAPI to do the staining and nucleus DNA get stained, looks like it’s the cytosolic DNA. 

(5) In Figure 5B, there is no increase in HCF-1 cleavage after OGT over-expression.

We appreciate the reviewer for his/her comment, we think the reason is that we used the cell line to stably overexpress OGT-GFP and we may have missed the time point when the increase in HCF-1 cleavage occurred, so there is no big increase of it. However, there is a significant increase in Figure 5C.

(6) In Figure 7, the TNF-a staining does not inspire confidence.

We thank the reviewer for his/her comment, from both Figure 7K (MC38 tumor model) and Figure 7N (LLC tumor model), we observed a significant increase in TNF-α+ CD8+ T cells in the group treated with the combination of OSMI-1 and anti-PD-L1 compared to the control group, as evidenced by the clear clustering.

The writing needs significant improvement:

(1) There are multiple English grammar mistakes throughout the paper. It is recommended that the authors run the manuscript through an editing service.

We thank the reviewer for his/her suggestion. We apologize for the poor language of our manuscript. We worked on the manuscript for a long time and the repeated addition and removal of sentences and sections obviously led to poor readability. We have now worked on both language and readability and have also involved native English speakers for language corrections. We really hope that the flow and language level have been substantially improved.

(2) Some passages are misleading -- lines 161-162, line 217, lines 241-242, 263-264, 299-300. They need to be changed substantially.

We apologize for these mistakes, we have changed them.

(3) Figure legends should be rewritten. Currently, they are too abbreviated to be understood.

We apologize for that, we have rewritten them.

(4) Discussion should also be thoroughly reworked. Currently, it is merely restating the authors' findings. The authors should put their findings in the broader context of the field.

We apologize for that. For a better understanding of our study, we have reworked the discussion.

Reviewer #2 (Recommendations For The Authors):

(1) Previous studies (DOI: 10.1093/nar/gkw663, 10.1016/j.jgg.2015.07.002, 10.1016/j.dnarep.2022.103394) have suggested that OGT deficiency triggers DNA damage, connecting it to DNA repair and maintenance through various mechanisms. This should be acknowledged in the manuscript. Conversely, the role of HCF1 and its cleaved products in maintaining genomic integrity hasn't been previously shown. The authors investigate HCF1's role solely in the context of OGT inhibition. It is unclear whether this is also true under other stimuli that trigger DNA damage, whether fragments of HCF1 specifically reduce DNA damage, or if HSF1 is involved in the basal machinery that would be defective only in the absence of OGT.

We have acknowledged the manuscript mentioned above. In this paper we focused on the OGT function, which is related to HCF1. The role of HCF1 and its cleaved products in maintaining genomic integrity is an interesting topic, we may focus on it in next project.

(2) In villin-CRE-deficient mice, the authors observe generic inflammation in the intestine unrelated to tumor development. It's unclear if this also occurs in the presence of OGT inhibitors in mice, whether these inhibitors induce a systemic inflammatory (Type I interferon) response, or if certain tissues like the intestine or proliferating tumor cells are more susceptible to such a response.

We thank the comment, yes, investigating whether OGT inhibitors induce an inflammatory response, either systemically or tissue-specifically, is a very interesting project to focus on. However, in our current paper, we use a genetic method to identify the role of OGT deficiency in intestine inflammation-induced tumor development. This approach provides convincing evidence for our hypothesis. We may test the effect of OGT inhibitors on inflammation and tumor development in our next project.

(3) Another critical observation is the magnitude of the interferon response triggered by DNA damage in the OGT-deficient models. While it's known that DNA damage can activate cGAS-STING, the response's extent in the absence of OGT prompts the question of whether additional OGT-specific features could explain this phenomenon. For example, Lamin A, essential for nuclear envelope integrity and shown to be O-glycosylated (DOI: 10.3390/cells7050044), and other components of the nuclear envelope or its repair might be affected by OGT. The impact of OGT inhibition on nuclear envelope integrity compared to other DNA-damaging agents could be explored.

We appreciate the comment, in this project, we find an OGT binding protein, HCF1, though LC–MS/MS assay, it’s a top one candidate in binding profiles, so we focus on it. Like Lamin A and other components of the nuclear envelope still are good targets to check, we may explore these in our next project.

(4) The authors also demonstrate a correlation between OGT expression in tumors compared to healthy tissues. However, the reason is unclear, raising questions about whether this is a consequence of proliferation or metabolic deregulation in the cancer. The authors should address this aspect.

We appreciate the reviewer’s insightful point. It is very good questions and very interesting research. However, in this paper we focused on how OGT influence its downstream molecules to promote tumor, we didn’t check why OGT is increased in tumors, it is not the scope of this current work, we would love to investigate it in the future.

Minor points

Please add the legend to Figures S2, S3 and S5.

We thank the comment, we have added the legend to Figures S2, S3 and S5.

The sentence line 137 should be clarified as OGT deficiency seems more related to increased inflammation in this model.

We thank the comment, we have corrected the sentence line 137.

Line 732 has a ( typo before the number 34.

We thank the comment, we have corrected the sentence line 732.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

In this important study, the authors manually assessed randomly selected images published in eLife between 2012 and 2020 to determine whether they were accessible for readers with deuteranopia, the most common form of color vision deficiency. They then developed an automated tool designed to classify figures and images as either "friendly" or "unfriendly" for people with deuteranopia. While such a tool could be used by publishers, editors or researchers to monitor accessibility in the research literature, the evidence supporting the tools' utility was incomplete. The tool would benefit from training on an expanded dataset that includes different image and figure types from many journals, and using more rigorous approaches when training the tool and assessing performance. The authors also provide code that readers can download and run to test their own images. This may be of most use for testing the tool, as there are already several free, user-friendly recoloring programs that allow users to see how images would look to a person with different forms of color vision deficiency. Automated classifications are of most use for assessing many images, when the user does not have the time or resources to assess each image individually.

Thank you for this assessment. We have responded to the comments and suggestions in detail below. One minor correction to the above statement: the randomly selected images published in eLife were from articles published between 2012 and 2022 (not 2020).

Public Reviews:

Reviewer #1 (Public Review):

Summary:

The authors of this study developed a software application, which aims to identify images as either "friendly" or "unfriendly" for readers with deuteranopia, the most common color-vision deficiency. Using previously published algorithms that recolor images to approximate how they would appear to a deuteranope (someone with deuteranopia), authors first manually assessed a set of images from biology-oriented research articles published in eLife between 2012 and 2022. The researchers identified 636 out of 4964 images as difficult to interpret ("unfriendly") for deuteranopes. They claim that there was a decrease in "unfriendly" images over time and that articles from cell-oriented research fields were most likely to contain "unfriendly" images. The researchers used the manually classified images to develop, train, and validate an automated screening tool. They also created a user-friendly web application of the tool, where users can upload images and be informed about the status of each image as "friendly" or "unfriendly" for deuteranopes.

Strengths:

The authors have identified an important accessibility issue in the scientific literature: the use of color combinations that make figures difficult to interpret for people with color-vision deficiency. The metrics proposed and evaluated in the study are a valuable theoretical contribution. The automated screening tool they provide is well-documented, open source, and relatively easy to install and use. It has the potential to provide a useful service to the scientists who want to make their figures more accessible. The data are open and freely accessible, well documented, and a valuable resource for further research. The manuscript is well written, logically structured, and easy to follow.

We thank the reviewer for these comments.

Weaknesses:

(1) The authors themselves acknowledge the limitations that arise from the way they defined what constitutes an "unfriendly" image. There is a missed chance here to have engaged deuteranopes as stakeholders earlier in the experimental design. This would have allowed [them] to determine to what extent spatial separation and labelling of problematic color combinations responds to their needs and whether setting the bar at a simulated severity of 80% is inclusive enough. A slightly lowered barrier is still a barrier to accessibility.

We agree with this point in principle. However, different people experience deuteranopia in different ways, so it would require a large effort to characterize these differences and provide empirical evidence about many individuals' interpretations of problematic images in the "real world." In this study, we aimed to establish a starting point that would emphasize the need for greater accessibility, and we have provided tools to begin accomplishing that. We erred on the side of simulating relatively high severity (but not complete deuteranopia). Thus, our findings and tools should be relevant to some (but not all) people with deuteranopia. Furthermore, as noted in the paper, an advantage of our approach is that "by using simulations, the reviewers were capable of seeing two versions of each image: the original and a simulated version." We believe this step is important in assessing the extent to which deuteranopia could confound image interpretations. Conceivably, this could be done with deuteranopes after recoloration, but it is difficult to know whether deuteranopes would see the recolored images in the same way that non-deuteranopes see the original images. It is also true that images simulating deuteranopia may not perfectly reflect how deuteranopes see those images. It is a tradeoff either way. We have added comments along these lines to the paper.

(2) The use of images from a single journal strongly limits the generalizability of the empirical findings as well as of the automated screening tool itself. Machine-learning algorithms are highly configurable but also notorious for their lack of transparency and for being easily biased by the training data set. A quick and unsystematic test of the web application shows that the classifier works well for electron microscopy images but fails at recognizing red-green scatter plots and even the classical diagnostic images for color-vision deficiency (Ishihara test images) as "unfriendly". A future iteration of the tool should be trained on a wider variety of images from different journals.

Thank you for these comments. We have reviewed an additional 2,000 images, which were randomly selected from PubMed Central. We used our original model to make predictions for those images. The corresponding results are now included in the paper.

We agree that many of the images identified as being "unfriendly" are microscope images, which often use red and green dyes. However, many other image types were identified as unfriendly, including heat maps, line charts, maps, three-dimensional structural representations of proteins, photographs, network diagrams, etc. We have uploaded these figures to our Open Science Framework repository so it's easier for readers to review these examples. We have added a comment along these lines to the paper.

The reviewer mentioned uploading red/green scatter plots and Ishihara test images to our Web application and that it reported they were friendly. Firstly, it depends on the scatter plot. Even though some such plots include green and red, the image's scientific meaning may be clear. Secondly, although the Ishihara images were created as informal tests for humans, these images (and ones similar to them) are not in eLife journal articles (to our knowledge) and thus are not included in our training set. Thus, it is unsurprising that our machine-learning models would not classify such images correctly as unfriendly.

(3) Focusing the statistical analyses on individual images rather than articles (e.g. in figures 1 and 2) leads to pseudoreplication. Multiple images from the same article should not be treated as statistically independent measures, because they are produced by the same authors. A simple alternative is to instead use articles as the unit of analysis and score an article as "unfriendly" when it contains at least one "unfriendly" image. In addition, collapsing the counts of "unfriendly" images to proportions loses important information about the sample size. For example, the current analysis presented in Fig. 1 gives undue weight to the three images from 2012, two of which came from the same article. If we perform a logistic regression on articles coded as "friendly" and "unfriendly" (rather than the reported linear regression on the proportion of "unfriendly" images), there is still evidence for a decrease in the frequency of "unfriendly" eLife articles over time.

Thank you for taking the time to provide these careful insights. We have adjusted these statistical analyses to focus on articles rather than individual images. For Figure 1, we treat an article as "Definitely problematic" if any image in the article was categorized as "Definitely problematic." Additionally, we no longer collapse the counts to proportions, and we use logistic regression to summarize the trend over time. The overall conclusions remain the same.

Another issue concerns the large number of articles (>40%) that are classified as belonging to two subdisciplines, which further compounds the image pseudoreplication. Two alternatives are to either group articles with two subdisciplines into a "multidisciplinary" group or recode them to include both disciplines in the category name.

Thank you for this insight. We have modified Figure 2 so that it puts all articles that have been assigned two subdisciplines into a "Multidisciplinary" category. The overall conclusions remain the same.

(4) The low frequency of "unfriendly" images in the data (under 15%) calls for a different performance measure than the AUROC used by the authors. In such imbalanced classification cases the recommended performance measure is precision-recall area under the curve (PR AUC: https://doi.org/10.1371%2Fjournal.pone.0118432) that gives more weight to the classification of the rare class ("unfriendly" images).

We now calculate the area under the precision-recall curve and provide these numbers (and figures) alongside the AUROC values (and figures). We agree that these numbers are informative; both metrics lead to the same overall conclusions.

Reviewer #2 (Public Review):

Summary:

An analysis of images in the biology literature that are problematic for people with a color-vision deficiency (CVD) is presented, along with a machine learning-based model to identify such images and a web application that uses the model to flag problematic images. Their analysis reveals that about 13% of the images could be problematic for people with CVD and that the frequency of such images decreased over time. Their model yields 0.89 AUC score. It is proposed that their approach could help making biology literature accessible to diverse audiences.

Strengths:

The manuscript focuses on an important yet mostly overlooked problem, and makes contributions both in expanding our understanding of the extent of the problem and in developing solutions to mitigate the problem. The paper is generally well-written and clearly organized. Their CVD simulation combines five different metrics. The dataset has been assessed by two researchers and is likely to be of high-quality. Machine learning algorithm used (convolutional neural network, CNN) is an appropriate choice for the problem. The evaluation of various hyperparameters for the CNN model is extensive.

We thank the reviewer for these comments.

Weaknesses:

The focus seems to be on one type of CVD (deuteranopia) and it is unclear whether this would generalize to other types.

We agree that it would be interesting to perform similar analyses for protanopia and other color-vision deficiencies. But we leave that work for future studies.

The dataset consists of images from eLife articles. While this is a reasonable starting point, whether this can generalize to other biology/biomedical articles is not assessed.

This is an important point. We have reviewed an additional 2,000 images, which were randomly selected from PubMed Central, and used our original model to make predictions for those images. The corresponding results are now included in the paper.

"Probably problematic" and "probably okay" classes are excluded from the analysis and classification, and the effect of this exclusion is not discussed.

We now address this in the Discussion section.

Machine learning aspects can be explained better, in a more standard way.

Thank you. We address this comment in our responses to your comments below.

The evaluation metrics used for validating the machine learning models seem lacking (e.g., precision, recall, F1 are not reported).

We now provide these metrics (in a supplementary file).

The web application is not discussed in any depth.

The paper includes a paragraph about how the Web application works and which technologies we used to create it. We are unsure which additional aspects should be addressed.

Reviewer #3 (Public Review):

Summary:

This work focuses on accessibility of scientific images for individuals with color vision deficiencies, particularly deuteranopia. The research involved an analysis of images from eLife published in 2012-2022. The authors manually reviewed nearly 5,000 images, comparing them with simulated versions representing the perspective of individuals with deuteranopia, and also evaluated several methods to automatically detect such images including training a machine-learning algorithm to do so, which performed the best. The authors found that nearly 13% of the images could be challenging for people with deuteranopia to interpret. There was a trend toward a decrease in problematic images over time, which is encouraging.

Strengths:

The manuscript is well organized and written. It addresses inclusivity and accessibility in scientific communication, and reinforces that there is a problem and that in part technological solutions have potential to assist with this problem.

The number of manually assessed images for evaluation and training an algorithm is, to my knowledge, much larger than any existing survey. This is a valuable open source dataset beyond the work herein.

The sequential steps used to classify articles follow best practices for evaluation and training sets.

We thank the reviewer for these comments.

Weaknesses:

I do not see any major issues with the methods. The authors were transparent with the limitations (the need to rely on simulations instead of what deuteranopes see), only capturing a subset of issues related to color vision deficiency, and the focus on one journal that may not be representative of images in other journals and disciplines.

We thank the reviewer for these comments. Regarding the last point, we have reviewed an additional 2,000 images, which were randomly selected from PubMed Central, and used our original model to make predictions for those images. The corresponding results are now included in the paper.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

N/A

Thank you.

Reviewer #2 (Recommendations For The Authors):

- The web application link can be provided in the Abstract for more visibility.

We have added the URL to the Abstract.

- They focus on deuteranopia in this paper. It seems that protanopia is not considered. Why? What are the challenges in considered this type of CVD?

We agree that it would be interesting to perform similar analyses for protanopia and other color-vision deficiencies. But we leave that work for future studies. Deuteranopia is the most common color-vision deficiency, so we focused on the needs of these individuals as a starting point.

- The dataset is limited to eLife articles. More discussion of this limitation is needed. Couldn't one also include some papers from PMC open access dataset for comparison?

We have reviewed an additional 2,000 images, which we randomly selected from PubMed Central, and used our original model to make predictions for those images. The corresponding results are now included in the paper.

- An analysis of the effect of selecting a severity value of 0.8 can be included.

We agree that this would be interesting, but we leave it for future work.

- "Probably problematic" and "probably okay" classes are excluded from analysis, which may oversimplify the findings and bias the models. It would have been interesting to study these classes as well.

We agree that this would be interesting, but we leave it for future work. However, we have added a comment to the Discussion on this point.

- Some machine learning aspects are discussed in a non-standard way. Class weighting or transfer learning would not typically be considered hyperparameters."corpus" is not a model. Description of how fine-tuning was performed could be clearer.

We have updated this wording to use more appropriate terminology to describe these different "configurations." Additionally, we expanded and clarified our description of fine tuning.

- Reporting performance on the training set is not very meaningful. Although I understand this is cross-validated, it is unclear what is gained by reporting two results. Maybe there should be more discussion of the difference.

We used cross validation to compare different machine-learning models and configurations. Providing performance metrics helps to illustrate how we arrived at the final configurations that we used. We have updated the manuscript to clarify this point.

- True positives, false positives, etc. are described as evaluation metrics. Typically, one would think of these as numbers that are used to calculate evaluation metrics, like precision (PPV), recall (sensitivity), etc. Furthermore, they say they measure precision, recall, precision-recall curves, but I don't see these reported in the manuscript. They should be (especially precision, recall, F1).

We have clarified this wording in the manuscript.

- There are many figures in the supplementary material, but not much interpretation/insights provided. What should we learn from these figures?

We have revised the captions and now provide more explanations about these figures in the manuscript.

- CVD simulations are mentioned (line 312). It is unclear whether these methods could be used for this work and if so, why they were not used. How do the simulations in this work compare to other simulations?

This part of the manuscript refers to recolorization techniques, which attempt to make images more friendly to people with color vision deficiencies. For our paper, we used a form of recolorization that simulates how a deuteranope would see a figure in its original form. Therefore, unless we misunderstand the reviewer's question, these two types of simulation have distinct purposes and thus are not comparable.

- relu -> ReLU

We have corrected this.

Reviewer #3 (Recommendations For The Authors):

The title can be more specific to denote that the survey was done in eLife papers in the years 2012-2022. Similarly, this should be clear in the abstract instead of only "images published in biology-oriented research articles".

Thank you for this suggestion. Because we have expanded this work to include images from PubMed Central papers, we believe the title is acceptable as it stands. We updated the abstract to say, "images published in biology- and medicine-oriented research articles"

Two mentions of existing work that I did not see are to Jambor and colleagues' assessment on color accessibility in several fields: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8041175/, and whether this work overlaps with the 'JetFighter' tool

(https://elifesciences.org/labs/c2292989/jetfighter-towards-figure-accuracy-and-accessibility).

Thank you for bringing these to our attention. We have added a citation to Jambor, et al.

We also mention JetFighter and describe its uses.

Similarly, on Line 301: Significant prior work has been done to address and improve accessibility for individuals with CVD. This work can be generally categorized into three types of studies: simulation methods, recolorization methods, and estimating the frequency of accessible images.

- One might mention education as prior work as well, which might in part be contributing to a decrease in problematic images (e.g., https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8041175/)

We now suggest that there are four categories and include education as one of these.

Line 361, when discussing resources to make figures suitable, the authors may consider citing this paper about an R package for single-cell data: https://elifesciences.org/articles/82128

Thank you. We now cite this paper.

The web application is a good demonstration of how this can be applied, and all code is open so others can apply the CNN in their own uses cases. Still, by itself, it is tedious to upload individual image files to screen them. Future work can implement this into a workflow more typical to researchers, but I understand that this will take additional resources beyond the scope of this project. The demonstration that these algorithms can be run with minimal resources in the browser with tensorflow.js is novel.

Thank you.

General:

It is encouraging that 'definitely problematic' images have been decreasing over time in eLife. Might this have to do with eLife policies? I could not quickly find if eLife has checks in place for this, but given that JetFighter was developed in association with eLife, I wonder if there is an enhanced awareness of this issue here vs. other journals.

This is possible. We are not aware of a way to test this formally.

Reviewer #1 (Public Review):

Summary:

A nice study trying to identify the relationship between E. coli O157 from cattle and humans in Alberta, Canada.

Strengths:

(1) The combined human and animal sampling is a great foundation for this kind of study.

(2) Phylogenetic analyses seem to have been carried out in a high-quality fashion.

Weaknesses:

I think there may be a problem with the selection of the isolates for the primary analysis. This is what I'm thinking:

(1) Transmission analyses are strongly influenced by the sampling frame.

(2) While the authors have randomly selected from their isolate collections, which is fine, the collections themselves are not random.

(3) The animal isolates are likely to represent a broad swathe of diversity, because of the structured sampling of animal reservoirs undertaken (as I understand it).

(4) The human isolates are all from clinical cases. Clinical cases of the disease are likely to be closely related to other clinical cases, because of outbreaks (either detected, or undetected), and the high ascertainment rate for serious infections.

(5) Therefore, taking an equivalent number of animal and clinical isolates, will underestimate the total diversity in the clinical isolates because the sampling of the clinical isolates is less "independent" (in the statistical sense) than sampling from the animal isolates.

(6) This could lead to over-estimating of transmission from cattle to humans.

(7) "We hypothesize that the large proportion of disease associated with local transmission systems is a principal cause of Alberta's high E. coli O157:H7 incidence" - this seems a bit tautological. There is a lot of O157 because there's a lot of transmission. What part of the fact it is local means that it is a principal cause of high incidence? It seems that they've observed a high rate of local transmission, but the reasons for this are not apparent, and hence the cause of Alberta's incidence is not apparent. Would a better conclusion not be that "X% of STEC in Alberta is the result of transmission of local variants"? And then, this poses a question for future epi studies of what the transmission pathway is.

Author response:

Reviewer #1 (Public Review):

Summary:

A nice study trying to identify the relationship between E. coli O157 from cattle and humans in Alberta, Canada.

Strengths:

(1) The combined human and animal sampling is a great foundation for this kind of study.

(2) Phylogenetic analyses seem to have been carried out in a high-quality fashion.

Weaknesses:

I think there may be a problem with the selection of the isolates for the primary analysis. This is what I'm thinking:

(1) Transmission analyses are strongly influenced by the sampling frame.

(2) While the authors have randomly selected from their isolate collections, which is fine, the collections themselves are not random.

(3) The animal isolates are likely to represent a broad swathe of diversity, because of the structured sampling of animal reservoirs undertaken (as I understand it).

(4) The human isolates are all from clinical cases. Clinical cases of the disease are likely to be closely related to other clinical cases, because of outbreaks (either detected, or undetected), and the high ascertainment rate for serious infections.

(5) Therefore, taking an equivalent number of animal and clinical isolates, will underestimate the total diversity in the clinical isolates because the sampling of the clinical isolates is less "independent" (in the statistical sense) than sampling from the animal isolates.

(6) This could lead to over-estimating of transmission from cattle to humans.

We appreciate the reviewer’s careful thoughts about our sampling strategy. We agree with points (1) and (2), and we will provide additional details on the animal collections as requested.

We agree with point (3) in theory but not in fact. As shown in Figure 3a, the cattle isolates were very closely related, despite the temporal and geographic breadth of sampling within Alberta. The median SNP distance between cattle sequences was 45 (IQR 36-56), compared to 54 (IQR 43-229) SNPs between human sequences from cases in Alberta during the same years. Additionally, as shown in Figure 2, only clade A and B isolates – clades that diverge substantially from the rest of the tree – were dominated by human cases in Alberta. We will better highlight this evidence in the revision.

We agree with the reviewer in point (4) that outbreaks can be an important confounder of phylogenetic inference. This is why we down-sampled outbreaks (based on genetic relatedness, not external designation) in our extended analyses (lines 192-194). We did not do this in the primary analysis, because there were no large clusters of identical isolates. Figure 3b shows a limited number of small clusters; however, clustered cattle isolates outnumbered clustered human isolates, suggesting that any bias would be in the opposite direction the reviewer suggests. Regarding severe cases being oversampled among the clinical isolates, this is absolutely true and a limitation of all studies utilizing public health reporting data. We will make this limitation to generalizability clearer in the discussion. However, as noted above, clinical isolates were more variable than cattle isolates, so it does not appear to have heavily biased the analysis.

We disagree with the reviewer on point (5). While the bias toward severe cases could make the human isolates less independent, the relative sampling proportions are likely to induce greater distance between clinical isolates than cattle isolates, which is exactly what we observe (see response to point (3) above). Cattle are E. coli O157:H7’s primary reservoir, and humans are incidental hosts not able to sustain infection chains long-term. Not only is the bacteria prevalent among cattle, cattle are also highly prevalent in Alberta. Thus, even with 89 sampling points, we are still capturing a small proportion of the E. coli O157:H7 in the province. Being able to sample only a small proportion of cattle’s E. coli O157:H7 increases the likelihood of only sampling from the center of the distribution, making extreme cases such as that shown at the very bottom of the tree in Figure 3b, rare and important. In comparison, sampling from human cases constitutes a higher proportion of human infections relative to cattle, and is therefore more representative of the underlying distribution, including extremes. We will add this point to the limitations. As with the clustering above, if anything, this outcome would have biased the study away from identifying cattle as the primary reservoir. Additionally, the relatively small proportion of cattle sampled makes our finding that 15.7% of clinical isolates were within 5 SNPs of a cattle isolate, the distance most commonly used to indicate transmission for E. coli O157:H7, all the more remarkable.

Because of the aforementioned points, we disagree with the reviewer’s conclusion in point (6). We believe transmission from cattle-to-humans is likely underestimated for the reasons given above. Not only do all prior studies indicate ruminants as the primary reservoirs of E. coli O157:H7, and humans as only incidental hosts, our specific data do not support the reviewer’s individual contentions. That said, we will conduct a sensitivity analysis as recommended to determine the impact of sampling and inclusion of the small clusters on our primary findings.

(7) We hypothesize that the large proportion of disease associated with local transmission systems is a principal cause of Alberta's high E. coli O157:H7 incidence" - this seems a bit tautological. There is a lot of O157 because there's a lot of transmission. What part of the fact it is local means that it is a principal cause of high incidence? It seems that they've observed a high rate of local transmission, but the reasons for this are not apparent, and hence the cause of Alberta's incidence is not apparent. Would a better conclusion not be that "X% of STEC in Alberta is the result of transmission of local variants"? And then, this poses a question for future epi studies of what the transmission pathway is.

The reviewer is correct, and the suggestion for the direction of future studies was our intent with this statement. We will revise it.

Reviewer #2 (Public Review):

This study identified multiple locally evolving lineages transmitted between cattle and humans persistently associated with E. coli O157:H7 illnesses for up to 13 years. Furthermore, this study mentions a dramatic shift in the local persistent lineages toward strains with the more virulent stx2a-only profile. The authors hypothesized that this phenomenon is the large proportion of disease associated with local transmission systems is a principal cause of Alberta's high E. coli O157:H7 incidence. These opinions more effectively explain the role of the cattle reservoir in the dynamics of E. coli O157:H7 human infections.

(1) The authors acknowledge the possibility of intermediate hosts or environmental reservoirs playing a role in transmission. Further discussion on the potential roles of other animal species commonly found in Alberta (e.g., sheep, goats, swine) could enhance the understanding of the transmission dynamics. Were isolates from these species available for analysis? If not, the authors should clearly state this limitation.

We will expand the discussion of other species in Alberta, as suggested, including other livestock, wildlife, and the potential role of birds and flies. Unfortunately, we did not have sequences available from other species, and we will add this to the limitations. Sequences from other species may be available from sequences collected by others, which as we note in the limitations do not have sufficient metadata to assign them to Alberta vs. the rest of Canada. While we have requested this data, we have been unsuccessful in obtaining it. We will continue to pursue it.

(2) The focus on E. coli O157:H7 is understandable given its prominence in Alberta and the availability of historical data. However, a brief discussion on the potential applicability of the findings to non-O157 STEC serogroups, and the limitations therein, would be beneficial. Are there reasons to believe the transmission dynamics would be similar or different for other serogroups?

We appreciate this comment and will expand our discussion of relevance to non-O157 STEC. Other authors have proposed that transmission dynamics differ, and studies of STEC risk factors, including our own, support this. However, there has been very little direct study of non-O157 transmission dynamics and there is even less cross-species genomic and metadata available for non-O157 isolates of concern.

(3) The authors briefly mention the need for elucidating local transmission systems to inform management strategies. A more detailed discussion on specific public health interventions that could be targeted at the identified LPLs and their potential reservoirs would strengthen the paper's impact.

We agree with the reviewer that this would be a good addition to the manuscript. The public health implications for control are several and extend to non-STEC reportable zoonotic enteric infections, such as Campylobacter and Salmonella. We will add a discussion of these.

(4) Understanding the relationship between specific risk factors and E. coli O157:H7 infections is essential for developing effective prevention strategies. Have case-control or cohort studies been conducted to assess the correlation between identified risk factors and the incidence of E. coli O157:H7 infections? What methodologies were employed to control for potential confounders in these studies?

Yes, there have been several case-control studies of reported cases. Many of these are referenced in the discussion in terms of the contribution of different sources to infection. However, we will add a more explicit discussion of risk factors.

(5) The study's findings are noteworthy, particularly in the context of E. coli O157:H7 epidemiology. However, the extent to which these results can be replicated across different temporal and geographical settings remains an open question. It would be constructive for the authors to provide additional data that demonstrate the replication of their sampling and sequencing experiments under varied conditions. This would address concerns regarding the specificity of the observed patterns to the initial study's parameters.

We appreciate the reviewer’s comment, as we are currently building on this analysis with an American dataset with different types of data available than were used in this study. We will add a discussion of this. We will also be adding a sensitivity analysis to the manuscript simulating a different sampling approach, which should also be informative to this question.

Reviewer #2 (Public Review):

Summary:

The goal is to ask if common species when studied across their range tend to have larger ranges in total. To do this the authors examined a very large citizen science database which gives estimates of numbers, and correlated that with the total range size, available from Birdlife. The average correlation is positive but close to zero, and the distribution around zero is also narrow, leading to the conclusion that, even if applicable in some cases, there is no evidence for consistent trends in one or other direction.

Strengths:

The study raises a dormant question, with a large dataset.

Weaknesses:

This study combines information from across the whole world, with many different habitats, taxa, and observations, which surely leads to a quite heterogeneous collection.

First, scale. Many of the earlier analyses were within smaller areas, and for example, ranges are not obviously bounded by a physical barrier. I assume this study is only looking at breeding ranges; that should be stated, as 40% of all bird species migrate, and winter limitation of populations is important. Also are abundances only breeding abundances or are they measured through the year? Are alien distributions removed?

Second, consider various reasons why abundance and range size may be correlated (sometimes positively and sometimes negatively) at large scales. Combining studies across such a large diversity of ecological situations seems to create many possibilities to miss interesting patterns. For example:

(1) Islands are small and often show density release.

(2) North temperate regions have large ranges (Rapoport's rule) and higher population sizes than the tropics.

(3) Body size correlates with global range size (I am unsure if this has recently been tested but is present in older papers) and with density. For example, cosmopolitan species (barn owl, osprey, peregrine) are relatively large and relatively rare.

(4) In the consideration of alien species, it certainly looks to me as if the law is followed, with pigeon, starling, and sparrow both common and widely distributed. I guess one needs to make some sort of statement about anthropogenic influences, given the dramatic changes in both populations and environments over the past 50 years.

(5) Wing shape correlates with ecological niche and range size (e.g. White, American Naturalist). Aerial foraging species with pointed wings are likely to be easily detected, and several have large ranges reflecting dispersal (e.g. barn swallow).

Third, biases. I am not conversant with ebird methodology, but the number appearing on checklists seems a very poor estimate of local abundance. As noted in the paper, common species may be underestimated in their abundance. Flocking species must generate large numbers, skulking species few. The survey is often likely to be in areas favorable to some species and not others. The alternative approach in the paper comes from an earlier study, based on ebird but then creating densities within grids and surely comes with similar issues.<br /> Biases are present in range as well. Notably, tropical mountain-occupying species have range sizes over-estimated because holes in the range are not generally accounted for (Ocampo-Peñuela et al Nature communications). These species are often quite rare too.

Fourth, random error. Random error in ebird assessments is likely to be large, with differences among observers, seasons, days, and weather (e.g. Callaghan et al. 2021, PNAS). Range sizes also come with many errors, which is why occupancy is usually seen as the more appropriate measure.

If we consider both range and abundance measurements to be subject to random error in any one species list, then the removal of all these errors will surely increase the correlation for that list (the covariance shouldn't change but the variances will decrease). I think (but am not sure) that this will affect the mean correlation because more of the positive correlations appear 'real' given the overall mean is positive. It will definitely affect the variance of the correlations; the low variance is one of the main points in the paper. A high variance would point to the operation of multiple mechanisms, some perhaps producing negative correlations (Blackburn et al. 2006).

On P.80 it is stated: "Specifically, we can quantify how AOR will change in relation to increases in species richness and sampling duration, both of which are predicted to reduce the magnitude of AORs" I haven't checked the references that make this statement, but intuitively the opposite is expected? More species and longer durations should both increase the accuracy of the estimate, so removing them introduces more error? Perhaps dividing by an uncertain estimate introduces more error anyway. At any rate, the authors should explain the quoted statement in this paper.

It would be of considerable interest to look at the extreme negative and extreme positive correlations: do they make any biological sense?

Discussion:

I can see how publication bias can affect meta-analyses (addressed in the Gaston et al. 2006 paper) but less easily see how confirmation bias can. It seems to me that some of the points made above must explain the difference between this study and Blackburn et al. 2006's strong result.

Certainly, AOR really does seem to be present in at least some cases (e.g. British breeding birds) and a discussion of individual cases would be valuable. Previous studies have also noted that there are at least some negative and some non-significant associations, and understanding the underlying causes is of great interest (e.g. Kotiaho et al. Biology Letters).

It is not easy to crtitque your own work in a way such as these authors did. However, by stating in their book that "only four of our twenty-five mini-biographies were of women," shows that it is okay to admit your faults. Nobody is perfect and it is foolish to illustrate yourself as such. Another benefit of this particular group making these statements is that it draws the student to look closer at these types of things. To ask questions, such as, "Out of these authors, how many are women? How many are of a different race?" While these questions may cause some backlash for "discrimination", they are valid questions for this instance. As long as you are not using gender or race in a hateful way, it is okay to observe these things. It is common sense that people of different genders and races might have different opinions, life styles, experiences, and so much more.

This is good to know that we are able to receive extensions. I know with papers I tend to reach a writers block, or sometimes need extra time to re-read my paper and make sure it is up to my standards. I think that this is very beneficial to the student, and you as a professor. I think with having that policy allows writers to be more comfortable with turning in their work that is completed, it also doesn't waste your time either, by reading/grading a paper that could've used a little more work.

I am looking forward to writing about topics that are more uncomfortable. I feel like, as students, we focus a lot on writing reports and more analytical projects. I hope that this class allows us to have a more vulnerable perspective on writing.

Author response:

The following is the authors’ response to the current reviews.

The concerns raised during the review have been incorporated into the discussion of the results, and the need for further research is acknowledged in the paper. This is not possible in the present study, as the clinical project has been completed and further patients cannot be enrolled without starting a new project. We are confident that the results are scientifically valid and that the methodology was scientifically sound and up to date. They were obtained on a dataset that was obviously large enough to allow 20% of it to be set aside and a machine-learned classifier to be trained on the remaining 80%, which then assigned samples to neuropathy with an accuracy better than guessing.

Furthermore, our results are at least tentatively replicated in a completely independent data set from another patient cohort. The strengths and limitations of the study design, in particular the latter, are discussed in the necessary depth. In summary, the machine-learned results provided major hits on one side and probably unimportant lipids on the other side of the variable importance scale. Both could be verified in vitro. We are therefore confident that we have contributed to the advancement of knowledge about cancer therapy-associated neuropathy and look forward to further developments in this area.

The following is the authors’ response to the original reviews.

Weaknesses Reviewer 1: 

There are a number of weaknesses in the study. The small sample size is a significant limitation of the study. Out of 31 patients, only 17 patients were reported to develop neuropathy, with significant neuropathy (grade 2/3) in only 5 patients. The authors acknowledge this limitation in the results and discussion sections of the manuscript, but it limits the interpretation of the results. Also acknowledged is the limited method used to assess neuropathy. 

We agree with the reviewer that the cohort size and assessment of neuropathy are limitations of our study as we already described in the corresponding section of the manuscript. However, occurrence and grade of the neuropathy are in line with results reported from previous studies. From these studies, the expected occurrence of neuropathy with our therapeutic regimen is around 50-70% (54.9% in our cohort), and most patients (80-90%) are expected to experience Grade 1 neuropathy after 12 weeks (13). In these studies, neuropathy is assessed by using questionnaires or by grading via NCTCTCAE as in our study. In summary, assessment and occurrence of neuropathy of our reported cohort are in line with previous reports.

Potentially due to this small number of patients with neuropathy, the machine learning algorithms could not distinguish between samples with and without neuropathy. Only selected univariate analyses identified differences in lipid profiles potentially related to neuropathy.  

The data analysis consistently followed a "mixture of experts" approach, as this seems to be the most successful way to deal with omics data. We have elaborated on this in the Methods section, including several supporting references. Regarding the quoted sentence from the results section, after rereading it, we realized that it was somewhat awkwardly worded. What we mean is now better worded in the results section, namely “Although the three algorithms detected neuropathy in new cases, unseen during training, at balanced accuracy of up to 0.75, while only the guess level of 0.5 was achieved when using permuted data for training, the 95% CI of the performance measures was not separated from guess level”. Therefore, multivariate feature selection was not considered a valid approach, since it requires that the algorithms from which the feature importance is read can successfully perform their task of class assignment (4). Therefore, univariate methods (Cohen's d, FPR, FWE) were preferred, as well as a direct hypothesis transfer of the top hits from the abovementioned day1/2 assessments to neuropathy. Classical statistics consisting of direct group comparisons using Kruskal-Wallis tests (5) were performed.” 

It was our approach to investigate the data set in an unbiased manner by different machine learning algorithms and select those lipids that the majority of the algorithms considered important for distinguishing the patient groups (majority voting). This way, the inconsistencies and limitations of a single evaluation method, such as regression analysis, that occur in some datasets, can be mitigated. 

Three sphingolipid mediators including SA1P differed between patients with and without neuropathy at the end of treatment. These sphingolipids were elevated at the end of treatment in the cohort with neuropathy, relative to those without neuropathy. However, across all samples from pre to post-paclitaxel treatment, there was a significant reduction in SA1P levels. It is unclear from the data presented what the underlying mechanism for this result would be. 

We agree with the reviewer that our study does not identify the mechanism by which paclitaxel treatment alters sphingolipid concentrations in the plasma of patients. It has been reported before that paclitaxel may increase expression and activity of serine palmitoyltransferase (SPT) which is the crucial enzyme and rate-limiting step in the denovo synthesis of sphingolipids. This may be associated with a shift towards increased synthesis of 1-deoxysphingolipids and a decrease of “classical” sphingolipids (6) and may explain the general reduction of SA1P and other sphingolipid levels after paclitaxel treatment in our study. 

It is also conceivable that paclitaxel reduces the release of sphingolipids into the plasma. Paclitaxel is a microtubule stabilizing agent (7) that may interfere with intracellular transport processes and release of paracrine mediators. 

The mechanistic details of paclitaxel involvement in sphingolipid metabolism or transport are highly interesting but identifying them is beyond the scope of our manuscript.

If elevated SA1P is associated with neuropathy development, it would be expected to increase in those who develop neuropathy from pre to post-treatment time points. 

There is a general trend of reduced plasma SA1P concentrations following paclitaxel treatment. Nevertheless, patients experiencing neuropathy exhibit significantly elevated SA1P levels post-treatment. 

It has been shown before that paclitaxel-induced neuropathic pain requires activation of the S1P1 receptor in a preclinical study (8). Moreover, a meta-analysis of genome-wide association studies (GWAS) from two clinical cohorts identified multiple regulatory elements and increased activity of S1PR1 associated with paclitaxel-induced neuropathy (9). These data imply that enhanced S1P receptor activity and signaling are key drivers of paclitaxel-induced neuropathy. It seems that both, increased levels of the sphingolipid ligands in combination with enhanced expression and activity of S1P receptors can potentiate paclitaxel-induced neuropathy in patients. This explains why also decreased SA1P concentrations after paclitaxel treatment can still enhance neuropathy via the S1PRTRPV1 axis in sensory neurons.

We added this paragraph to the discussions section of our manuscript.

Primary sensory neuron cultures were used to examine the effects of SA1P application.

SA1P application produced calcium transients in a small proportion of sensory neurons. It is not clear how this experimental model assists in validating the role of SA1P in neuropathy development as there is no assessment of sensory neuron damage or other hallmarks of peripheral neuropathy. These results demonstrate that some sensory neurons respond to SA1P and that this activity is linked to TRPV1 receptors. However, further studies will be required to determine if this is mechanistically related to neuropathy.

As we detected elevated levels of SA1P in the plasma of PIPN patients, we can assume higher concentrations in the vicinity of sensory neurons. These neurons are the main drivers for neuropathy and neuropathic pain and are strongly affected by paclitaxel in their activity (10-15). Also, TRPV1 shows altered activity patterns in response to paclitaxel treatment (16). Because of its relevance for nociception and pathological pain, TRPV1 activity is a suitable and representative readout for pathological pain states in peripheral sensory neurons (17, 18), which is why we investigated them.

We would like to point out the potency of SA1P to increase capsaicin-induced calciumtransients in sensory neurons at submicromolar concentrations. 

We also agree with the reviewer that further studies need to investigate the underlying mechanisms in more detail. We added this sentence to the final paragraph in the discussion section of our manuscript.

Weaknesses Reviewer 2: 

The article is poorly written, hindering a clear understanding of core results. While the study's goals are apparent, the interpretation of sphingolipids, particularly SA1P, as key mediators of paclitaxel-induced neuropathy lacks robust evidence. 

We agree that the relevance of SA1P as key mediator of paclitaxel-induced neuropathy might be overstated and changed the wording throughout the manuscript accordingly. However, we would like to point out the potency of this lipid to increase capsaicin-induced calcium-transients in sensory neurons at submicromolar concentrations. 

Also, the lipid signature in the plasma of PIPN patients shows a unique pattern and sphingolipids are the group that showed the strongest alterations when comparing the patient groups. We also measured eicosanoids, such as prostaglandins, linoleic acid metabolites, endocannabinoids and other lipid groups that have previously been associated with influences on pain perception or nociceptor sensitization. However, none of these lipids showed significant differences in their concentrations in patient plasma. This is why we consider sphingolipids as contributors to or markers of paclitaxel-induced neuropathy in patients.

We also revised the entire article to improve its clarity.

The introduction fails to establish the significance of general neuropathy or peripheral neuropathy in anticancer drug-treated patients, and crucial details, such as the percentage of patients developing general neuropathy or peripheral neuropathy, are omitted. This omission is particularly relevant given that only around 50% of patients developed neuropathy in this study, primarily of mild Grade 1 severity with negligible symptoms, contradicting the study's assertion of CIPN as a significant side effect. 

As we already described in the introduction, CIPN is a serious dose- and therapy-limiting side effect, which affects up to 80% of treated patients. This depends on dose and combination of chemotherapeutic agents. For paclitaxel, therapeutic doses range from 80 – 225 mg/m². As CIPN symptoms are dose-dependent, the number of PIPN patients that receive a high paclitaxel dose is higher than the number of PIPN patient receiving a low dose.

In our study, we mainly used a low dose paclitaxel, because this therapeutic regimen is the most widely used paclitaxel monotherapy. From previous studies, the expected occurrence of neuropathy with this therapeutic regimen is around 50-70%, and most patients (8090%) are expected to experience Grade 1 neuropathy after 12 weeks (1-3).

Our results are within the range reported by these studies (54.9% patients with neuropathy). Also, as we highlight in Table S1, the neuropathy symptoms persist in most cases for several years after chemotherapy, affecting quality of life of these patients which makes it far from being a negligible symptom.

We added some more information concerning PIPN in the introduction section in which we emphasize the clinical problem.

The lack of clarity in distinguishing results obtained by lipidomics using machine learning methods and conventional methods adds to the confusion. The poorly written results section fails to specify SA1P's downregulation or upregulation, and the process of narrowing down to sphingolipids and SA1P is inadequately explained. 

We have tried to keep the machine learning part in the main manuscript short and moved major parts of it to a supplement. However, as this has been claimed to have led to a lack of clarity, we have expanded the description of the data analysis and added extensive explanations and supporting references for the mixed expert approach that was used throughout the analysis. We hope this is now clear.

Integrating a significant portion of the discussion section into the results section could enhance clarity. An explanation of the utility of machine learning in classifying patient groups over conventional methods and the citation of original research articles, rather than relying on review articles, may also add clarity to the usefulness of the study. 

As suggested by the reviewer, we moved the relevant parts from the discussion to the results section in the revised version of our manuscript.

Reviewer #1 (Recommendations For The Authors): 

Figure 2 should be better explained or removed. In its current form, it does not add to the interpretation of the manuscript.  

As mentioned above, we have expanded the description of the ESOM/U-matrix method in the Methods section and rewritten the figure legend. In addition, we have annotated the U-matrix in the figure. The method has been reported extensively in the computer science and biomedical literature, and a more detailed description in the referenced papers would go beyond the current focus on lipidomics. However, we believe that this discussion is sufficiently detailed for the readers of this report: "… a second unsupervised approach was used to verify the agreement between the lipidomics data structure and the prior classification, implemented as self-organizing maps (SOM) of artificial neurons (19). In the special form of an “emergent” SOM (ESOM (20)), the present map consisted of 4,000 neurons arranged on a two-dimensional toroidal grid with 50 rows and 80 columns (21, 22). ESOM was used because it has been repeatedly shown to correctly detect subgroup structures in biomedical data sets comparable to the present one (20, 22, 23). The core principle of SOM learning is to adjust the weights of neurons based on their proximity to input data points. In this process, the best matching unit (BMU) is identified as the neuron closest to a given data point. The adaptation of the weights is determined by a learning rate (η) and a neighborhood function (h), both of which gradually decrease during the learning process. Finally, the groups are projected onto separate regions of the map. On top of the trained ESOM, the distance structure in the high-dimensional feature space was visualized in the form of a so-called U-matrix (24) which is the canonical tool for displaying the distance structures of input data on ESOM (21). 

The visual presentation facilitates data group separation by displaying the distances between BMUs in high-dimensional space in a color-coding that uses a geographical map analogy, where large "heights" represent large distances in feature space, while low "valleys" represent data subsets that are similar. "Mountain ranges" with "snow-covered" heights visually separate the clusters in the data. Further details about ESOM can be found in (24)."

The second patient cohort is only included in the discussion - with cohort details in the supplementary material and figures included in the main text. Perhaps these data should be removed entirely. The findings are described as trends and not statistically significant and multiple issues with this second cohort are mentioned in the discussion. 

We agree with the reviewer that including the second patient cohort in the discussion is inadequate. Of course, there are differences between the patient cohorts that do not allow direct comparison and that are highlighted in the section on limitations of the study. However, we still think it is interesting and relevant to show these data, because we used our algorithms trained on the first patient cohort to analyze the second cohort. And these data support the main results. 

We therefore moved the entire paragraph to the results section of to improve coherence of our manuscript. The passage was introduced with the subheading:  “Support of the main results in an independent second patient cohort”.

The title does not reflect the content of the paper and should be changed to better reflect the content and its significance. 

We change the title to “Machine learning and biological validation identify sphingolipids as potential mediators of paclitaxel-induced neuropathy in cancer patients” to avoid overstating the results as suggested by the Reviewer.

Further, the discussion should be modified to avoid overstating the results. 

As the reviewer suggests, we changed the wording to avoid overstating the results. 

Reviewer #2 (Recommendations For The Authors): 

Please address the absence of clear neuropathy in the majority of patients after treatment with paclitaxel in your discussion. 

As stated above, occurrence and grade of the neuropathy are in line with the results from previous studies. From these studies, the expected occurrence of neuropathy with our therapeutic regimen is around 50-70%, (the variability is due to differences in the assessment methods) and most patients (80-90%) are expected to experience Grade 1 neuropathy after 12 weeks (1-3). 

We added this information in the discussion section of the revised manuscript.

Line 65: Kindly replace review articles with original research articles for proper citation. 

We replaced the review articles with original publications, focusing on clinical observations. We added the following publications: Jensen et al., Front Neurosci 2020; Chen et al., Neurobiol Aging 2018; Igarashi et al., J Alzheimers Dis. 2011; Kim et al., Oncotarget 2017 as references 17-20 in the revised version of our manuscript.

Line 260: The mention of SA1P is introduced here without prior reference (do not use words like "again", or "see above", if it is not previously mentioned). Adjust the text for coherence.

We agree with the reviewer that the introduction of SA1P in this passage in incoherent. We replaced the sentence in line 260 with: 

The small set of lipid mediators emerging from all three methods as informative for neuropathy included the sphingolipid sphinganine-1-phosphate (SA1P), also known as dihydrosphingosine-1-phosphate (DH-S1P)…”

Lines 301-315: Consider relocating several lines from this section to the results section for improved clarity. 

We moved the lines 309-312 explaining the algorithm selection and their validation success in the corresponding results section (Lipid mediators informative for assigning postpaclitaxel therapy samples to neuropathy).

Lines 382-396: Move this content to the results section to enhance the organization and coherence of the manuscript. 

We moved the entire paragraph to the results section of our manuscript to improve coherence. The passage was introduced with the subheading:  “Support of the main results in an independent second patient cohort”.

References

(1) Barginear M, Dueck AC, Allred JB, Bunnell C, Cohen HJ, Freedman RA, et al. Age and the Risk of Paclitaxel-Induced Neuropathy in Women with Early-Stage Breast Cancer (Alliance A151411): Results from 1,881 Patients from Cancer and Leukemia Group B (CALGB) 40101. Oncologist. 2019;24(5):617-23.

(2) Mauri D, Kamposioras K, Tsali L, Bristianou M, Valachis A, Karathanasi I, et al. Overall survival benefit for weekly vs. three-weekly taxanes regimens in advanced breast cancer: A metaanalysis. Cancer Treat Rev. 2010;36(1):69-74.

(3) Budd GT, Barlow WE, Moore HC, Hobday TJ, Stewart JA, Isaacs C, et al. SWOG S0221: a phase III trial comparing chemotherapy schedules in high-risk early-stage breast cancer. J Clin Oncol. 2015;33(1):58-64.

(4) Lötsch J, and Ultsch A. Pitfalls of Using Multinomial Regression Analysis to Identify ClassStructure-Relevant Variables in Biomedical Data Sets: Why a Mixture of Experts (MOE) Approach Is Better. BioMedInformatics. 2023;3(4):869-84.

(5) Kruskal WH, and Wallis WA. Use of Ranks in One-Criterion Variance Analysis. J Am Stat Assoc. 1952;47(260):583-621.

(6) Kramer R, Bielawski J, Kistner-Griffin E, Othman A, Alecu I, Ernst D, et al. Neurotoxic 1deoxysphingolipids and paclitaxel-induced peripheral neuropathy. FASEB J. 2015;29(11):4461-72.

(7) Field JJ, Diaz JF, and Miller JH. The binding sites of microtubule-stabilizing agents. Chem Biol. 2013;20(3):301-15.

(8) Janes K, Little JW, Li C, Bryant L, Chen C, Chen Z, et al. The development and maintenance of paclitaxel-induced neuropathic pain require activation of the sphingosine 1-phosphate receptor subtype 1. J Biol Chem. 2014;289(30):21082-97.

(9) Chua KC, Xiong C, Ho C, Mushiroda T, Jiang C, Mulkey F, et al. Genomewide Meta-Analysis Validates a Role for S1PR1 in Microtubule Targeting Agent-Induced Sensory Peripheral Neuropathy. Clin Pharmacol Ther. 2020;108(3):625-34.

(10) Kawakami K, Chiba T, Katagiri N, Saduka M, Abe K, Utsunomiya I, et al. Paclitaxel increases high voltage-dependent calcium channel current in dorsal root ganglion neurons of the rat. J Pharmacol Sci. 2012;120(3):187-95.

(11) Pittman SK, Gracias NG, Vasko MR, and Fehrenbacher JC. Paclitaxel alters the evoked release of calcitonin gene-related peptide from rat sensory neurons in culture. Exp Neurol. 2013.

(12) Luo H, Liu HZ, Zhang WW, Matsuda M, Lv N, Chen G, et al. Interleukin-17 Regulates NeuronGlial Communications, Synaptic Transmission, and Neuropathic Pain after Chemotherapy.

Cell reports. 2019;29(8):2384-97 e5.

(13) Pease-Raissi SE, Pazyra-Murphy MF, Li Y, Wachter F, Fukuda Y, Fenstermacher SJ, et al. Paclitaxel Reduces Axonal Bclw to Initiate IP3R1-Dependent Axon Degeneration. Neuron. 2017;96(2):373-86 e6.

(14) Duggett NA, Griffiths LA, and Flatters SJL. Paclitaxel-induced painful neuropathy is associated with changes in mitochondrial bioenergetics, glycolysis, and an energy deficit in dorsal root ganglia neurons. Pain. 2017.

(15) Li Y, Adamek P, Zhang H, Tatsui CE, Rhines LD, Mrozkova P, et al. The Cancer Chemotherapeutic Paclitaxel Increases Human and Rodent Sensory Neuron Responses to TRPV1 by Activation of TLR4. J Neurosci. 2015;35(39):13487-500.

(16) Hara T, Chiba T, Abe K, Makabe A, Ikeno S, Kawakami K, et al. Effect of paclitaxel on transient receptor potential vanilloid 1 in rat dorsal root ganglion. Pain. 2013;154(6):882-9.

(17) Jardin I, Lopez JJ, Diez R, Sanchez-Collado J, Cantonero C, Albarran L, et al. TRPs in Pain Sensation. Front Physiol. 2017;8:392.

(18) Julius D. TRP Channels and Pain. Annual review of cell and developmental biology.

2013;29:355-84.

(19) Kohonen T. Self-Organized Formation of Topologically Correct Feature Maps. Biol Cybern. 1982;43(1):59-69.

(20) Lötsch J, Lerch F, Djaldetti R, Tegder I, and Ultsch A. Identification of disease-distinct complex biomarker patterns by means of unsupervised machine-learning using an interactive R toolbox (Umatrix). Big Data Analytics. 2018;3(1):5.

(21) Ultsch A. 2003.

(22) Lotsch J, Geisslinger G, Heinemann S, Lerch F, Oertel BG, and Ultsch A. Quantitative sensory testing response patterns to capsaicin- and ultraviolet-B-induced local skin hypersensitization in healthy subjects: a machine-learned analysis. Pain. 2018;159(1):11-24.

(23) Lötsch J, Thrun M, Lerch F, Brunkhorst R, Schiffmann S, Thomas D, et al. Machine-Learned Data Structures of Lipid Marker Serum Concentrations in Multiple Sclerosis Patients Differ from Those in Healthy Subjects. Int J Mol Sci. 2017;18(6).

(24) Lötsch J, and Ultsch A. Cham: Springer International Publishing; 2014:249-57.

Author response:

Public Reviews:

Reviewer #1 (Public Review):

Summary:

In this manuscript, Wu et al. introduce a novel approach to reactivate the Muller glia cell cycle in the mouse retina by simultaneously reducing p27Kip1 and increasing cyclin D1 using a single AAV vector. The approach effectively promotes Muller glia proliferation and reprograming without disrupting retinal structure or function. Interestingly, reactivation of the Muller glia cell cycle downregulates IFN pathway, which may contribute to the induced retinal regeneration. The results presented in this manuscript may offer a promising approach for developing Müller glia cell-mediated regenerative therapies for retinal diseases.

Strengths:

The data are convincing and supported by appropriate, validated methodology. These results are both technically and scientifically exciting and are likely to appeal to retinal specialists and neuroscientists in general.

Weaknesses:

There are some data gaps that need to be addressed.

(1) Please label the time points of AAV injection, EdU labeling, and harvest in Figure 1B.

We thank the reviewer for highlighting the lack of clarity in our experimental design. We will label all experiment timelines in the figures where appropriate in the revised version.

(2) What fraction of Müller cells were transduced by AAV under the experimental conditions?

We apologize for not clearly conveying the transduction efficiency. The retinal region adjacent to the injection site, typically near the central retina, exhibits a transduction efficiency of nearly 100%. In contrast, the peripheral retina shows a lower transduction efficiency compared to the central region. We will include the quantification of AAV transduction efficiency in the revised manuscript.

The quantification of Edu+ MG or other markers was conducted in the area with the highest efficiency. 

(3) It seems unusually rapid for MG proliferation to begin as early as the third day after CCA injection. Can the authors provide evidence for cyclin D1 overexpression and p27 Kip1 knockdown three days after CCA injection?

In our pilot study, we tested the onset time of GFP expression from AAV-GFAP-GFP following intravitreal injection. We observed GFP expression in MG as early as two days post-infection. These findings will be included in the revised manuscript. Additionally, we plan to perform qPCR or Western blot analysis to confirm cyclin D1 overexpression and p27kip1 knockdown at the onset of Müller glia proliferation, which will also be included in the revised manuscript.

(4) The authors reported that MG proliferation largely ceased two weeks after CCA treatment. While this is an interesting finding, the explanation that it might be due to the dilution of AAV episomal genome copies in the dividing cells seems far-fetched.

We believe that the lack of durability in high Cyclin D1 and low p27kip1 levels in MG contributes to the cessation of their proliferation. A potential reason for the loss of high Cyclin D1 overexpression and p27kip1 knockdown during MG proliferation could be the dilution of the AAV episomal genome. However, testing this hypothesis is challenging. Instead, we plan to provide direct evidence in the revised manuscript by examining the levels of Cyclin D1 and p27kip1 in the retina treated with CCA before and after the peak of MG proliferation.

Reviewer #2 (Public Review):

This manuscript by Wu, Liao et al. reports that simultaneous knockdown of P27Kip1 with overexpression of Cyclin D can stimulate Muller glia to re-enter the cell cycle in the mouse retina. There is intense interest in reprogramming mammalian muller glia into a source for neurogenic progenitors, in the hopes that these cells could be a source for neuronal replacement in neurodegenerative diseases. Previous work in the field has shown ways in which mouse Muller glia can be neurogenically reprogrammed and these studies have shown cell cycle re-entry prior to neurogenesis. In other works, typically, the extent of glial proliferation is limited, and the authors of this study highlight the importance of stimulating large numbers of Muller glia to re-enter the cell cycle with the hopes they will differentiate into neurons. While the evidence for stimulating proliferation in this study is convincing, the evidence for neurogenesis in this study is not convincing or robust, suggesting that stimulating cell cycle-reentry may not be associated with increasing regeneration without another proneural stimulus.

Below are concerns and suggestions.

Intro:

(1) The authors cite past studies showing "direct conversion" of MG into neurons. However, these studies (PMID: 34686336; 36417510) show EdU+ MG-derived neurons suggesting cell cycle re-entry does occur in these strategies of proneural TF overexpression.

We thank the reviewer for pointing this out. We will revise the statement to "MG neurogenesis," which encompasses both direct conversion and Müller glia proliferation followed by neuronal differentiation.

(2) Multiple citations are incorrectly listed, using the authors first name only (i.e. Yumi, et al; Levi, et al;). Studies are also incompletely referenced in the references.

We apologize for the mistake with the reference. We will fix these mistakes in the revised version.

Figure 1:

(3) When are these experiments ending? On Figure 1B it says "analysis" on the end of the paradigm without an actual day associated with this. This is the case for many later figures too. The authors should update the paradigms to accurately reflect experimental end points.

We thank the reviewer for highlighting the lack of clarity in our experimental design. We will label all experiment timelines in the figures where appropriate in the revised version.

(4) Are there better representative pictures between P27kd and CyclinD OE, the EdU+ counts say there is a 3 fold increase between Figure 1D&E, however the pictures do not reflect this. In fact, most of the Edu+ cells in Figure 1E don't seem to be Sox9+ MG but rather horizontally oriented nuclei in the OPL that are likely microglia.

Thanks to the reviewer for pointing this out. We will replace the image of Cyclin D1 which a better representative image.

(5) Is the infection efficacy of these viruses different between different combinations (i.e. CyclinD OE vs. P27kd vs. control vs. CCA combo)? As the counts are shown in Figure 1G only Sox9+/Edu+ cells are shown not divided by virus efficacy. If these are absolute counts blind to where the virus is and how many cells the virus hits, if the virus efficacy varies in efficiency this could drive absolute differences that aren't actually biological.

Because the AAV-GFAP-Cyclin D1 and AAV-GFAP-Cyclin D1-p27kip1 shRNA viruses do not carry a fluorescent reporter gene, we cannot easily measure viral efficacy in the same experiment. We believe that variations in viral efficacy cannot account for the significant differences in MG proliferation for two reasons: 1) We injected the same titer for all three viruses, and 2) Viral infection efficacy is very high, approaching 100% in the central retina. Nonetheless, to rule out the possibility that the differences in MG proliferation among the Cyclin D overexpression, p27kip1 knockdown, and CCA groups are due to variations in viral efficacy, we will include the p27kip1 knockdown and Cyclin D1 overexpression efficiencies for all four groups using qPCR and/or Western blot analysis in the revised manuscript.

(6) According to the Jax laboratories, mice aren't considered aged until they are over 18months old. While it is interesting that CCA treatment does not seem to lose efficacy over maturation I would rephrase the findings as the experiment does not test this virus in aged retinas.

Thank you to the reviewer for bringing this to our attention. We will void using “aged mice” in our revised manuscript.

(7) Supplemental Figure 2c-d. These viruses do not hit 100% of MG, however 100% of the P27Kip staining is gone in the P27sh1 treatment, even the P27+ cell in the GCL that is likely an astrocyte has no staining in the shRNA 1 picture. Why is this?

For Supplementary Figure 2c-d, we focused on the central area where knockdown efficiency was high, approaching 100%. We will replace this image with one that includes both high and low Müller glia transduction efficiency regions, clearly demonstrating the complete loss of p27kip1 staining in the area of high transduction efficiency.

Figure 2

(8) Would you expect cells to go through two rounds of cell cycle in such a short time? The treatment of giving Edu then BrdU 24 hours later would have to catch a cell going through two rounds of division in a very short amount of time. Again the end point should be added graphically to this figure.

We thank the reviewer for raising this important point. While the typical cell cycle time for human cells is approximately 24 hours, we hypothesized that 24 hours would be the most likely timepoint to capture cells continuously progressing through the cell cycle. However, we acknowledge that we cannot exclude the possibility of some cells entering a second cell cycle at much later timepoints.

In the revised manuscript, we will carefully qualify our conclusion to state that the majority of MG do not immediately undergo another cell division, rather than making a definitive statement. This more cautious phrasing will better reflect the limitations of the 24-hour timepoint and allow for the potential of a small subset of cells proceeding through additional rounds of division at later stages.

Figure 3

(9) I am confused by the mixing of ratios of viruses to indicate infection success. I know mixtures of viruses containing CCA or control GFP or a control LacZ was injected. Was the idea to probe for GFP or LacZ in the single cell data to see which cells were infected but not treated? This is not shown anywhere?

The virus infection was not uniform across the entire retina. To mark the infection hotspots, we added 10% GFP virus to the mixture. Regions of the retina with low infection efficiency were removed by dissection and excluded from the scRNA-seq analysis. We apologize for not clearly explaining this methodological detail in the original text, and will update the Methods section accordingly.

(10) The majority of glia sorted from TdTomato are probably not infected with virus. Can you subset cells that were infected only for analysis? Otherwise it makes it very hard to make population judgements like Figure 3E-H if a large portion are basically WT glia.

This question is related to the last one. Since the regions with high virus infection efficiency were selectively dissected and isolated for analysis, the percentage of CCA-infected MG should constitute the majority in the scRNA-seq data.

(11) Figure 3C you can see Rho is expressed everywhere which is common in studies like this because the ambient RNA is so high. This makes it very hard to talk about "Rod-like" MG as this is probably an artifact from the technique. Most all scRNA-seq studies from MG-reprogramming have shown clusters of "rods" with MG hybrid gene expression and these had in the past just been considered an artifact.

We agree that the low levels of Rho in other MG clusters (such as quiescent, reactivated, and proliferating MG) are likely due to RNA contamination. However, the level of Rho in the rod-like MG is significantly higher than in the other clusters, indicating that this is unlikely to be solely due to contamination.

As shown in Supplementary Figure 7A-C, a cluster of MG-rod hybrid cells (cluster C4) was present in all three experimental groups at similar ratios, and this hybrid cluster was excluded from further analysis. In contrast, the rod-like Müller glia (cluster C3) were predominantly found in the CCA and CCANT groups, suggesting a genuine response to CCA treatment.

Furthermore, we will conduct Rho and Gnat1 RNA in situ hybridization on the dissociated retinal cells to further support the conclusion that rod-specific genes are upregulated in a subset of MG in the revised manuscript.

(12) It is mentioned the "glial" signature is downregulated in response to CCA treatment. Where is this shown convincingly? Figure H has a feature plot of Glul , which is not clear it is changed between treatments. Otherwise MG genes are shown as a function of cluster not treatment.

We will add box plots of several MG-specific genes to better illustrate the downregulation of the glial signature in the relevant cell cluster in the revised manuscript.

Figure 4

(13) The authors should be commended for being very careful in their interpretations. They employ the proper controls (Er-Cre lineage tracing/EdU-pulse chasing/scRNA-seq omics) and were very careful to attempt to see MG-derived rods. This makes the conclusion from the FISH perplexing. The few puncta dots of Rho and GNAT in MG are not convincing to this reviewer, Rho and GNAT dots are dense everywhere throughout the ONL and if you drew any random circle in the ONL it would be full of dots. The rigor of these counts also comes into question because some dots are picked up in MG in the INL even in the control case. This is confusing because baseline healthy MG do not express RNA-transcripts of these Rod genes so what is this picking up? Taken together, the conclusion that there are Rod-like MG are based off scRNA-seq data (which is likely ambient contamination) and these FISH images. I don't think this data warrants the conclusion that MG upregulate Rod genes in response to CCA.

We performed RNA in situ hybridization on retinal sections because we aimed to correlate cell localization with rod gene expression. We understand the reviewer’s concern that the punctate signals of Rho and GNAT1 in the ONL MG may actually originate from neighboring rods. In the revised manuscript, we will conduct RNAscope on dissociated retinal cells to avoid this issue.

Figure 5

(14) Similar point to above but this Glul probe seems odd, why is it throughout the ONL but completely dark through the IPL, this should also be in astrocytes can you see it in the GCL? These retinas look cropped at the INL where below is completely black. The whole retinal section should be shown. Antibodies exist to GS that work in mouse along with many other MG genes, IHC or western blots could be done to better serve this point.

Indeed, the GCL was cropped out in Figure 5 A-B. We have other images with all retinal layers, which we will use in the revised manuscript. Additionally, we will perform the GS antibody staining to demonstrate partial MG dedifferentiation following CCA treatment.

Figure 6

(15) Figure 6D is not a co-labeled OTX2+/ TdTomato+ cell, Otx2 will fill out the whole nucleus as can be seen with examples from other MG-reprogramming papers in the field (Hoang, et al. 2020; Todd, et al. 2020; Palazzo, et al. 2022). You can clearly see in the example in Figure 6D the nucleus extending way beyond Otx2 expression as it is probably overlapping in space. Other examples should be shown, however, considering less than 1% of cells were putatively Otx2+, the safer interpretation is that these cells are not differentiating into neurons. At least 99.5% are not.

We have additional examples of Otx2+ Tdt+ Edu+ cells, which suggest that MG neurogenesis to Otx2+ cells does occur, despite the low efficiency. We will include these images in the revised manuscript.

(16) Same as above Figure 6I is not convincingly co-labeled HuC/D is an RNA-binding protein and unfortunately is not always the clearest stain but this looks like background haze in the INL overlapping. Other amacrine markers could be tested, but again due to the very low numbers, I think no neurogenesis is occurring.

We have additional examples of HuC/D+ Tdt+ Edu+ cells, which we will show in the revised manuscript.

(17) In the text the authors are accidently referring to Figure 6 as Figure 7.

We thank the reviewer for pointing out the mistake. We will correct the mistake in the revised manuscript.

Figure 7

(18) I like this figure and the concept that you can have additional MG proliferating without destroying the retina or compromising vision. This is reminiscent of the chick MG reprogramming studies in which MG proliferate in large numbers and often do not differentiate into neurons yet still persist de-laminated for long time points.

General:

(19) The title should be changed, as I don't believe there is any convincing evidence of regeneration of neurons. Understanding the barriers to MG cell-cycle re-entry are important and I believe the authors did a good job in that respect, however it is an oversell to report regeneration of neurons from this data.

We thank the reviewer for the suggestion. We will consider changing the title in the revised manuscript.

(20) This paper uses multiple mouse lines and it is often confusing when the text and figures switch between models. I think it would be helpful to readers if the mouse strain was added to graphical paradigms in each figure when a different mouse line is employed.

We will label the mouse lines used in each experiment in the figures where appropriate.

Reviewer #1 (Public Review):

Summary:

In this paper the researchers aimed to address whether bees causally understand string-pulling through a series of experiments. I first briefly summarize what they did:

- In experiment 1, the researchers trained bees without string and then presented them with flowers in the test phase that either had connected or disconnected strings, to determine what their preference was without any training. Bees did not show any preference.

- In experiment 2, bees were trained to have experience with string and then tested on their choice between connected vs. disconnected string.

- Experiment 3 was similar except that instead of having one option which was an attached string broken in the middle, the string was completely disconnected from the flower.

- In experiment 4, bees were trained on green strings and tested on white strings to determine if they generalize across color.

- In experiment 5, bees were trained on blue strings and tested on white strings.

- In experiment 6, bees were trained where black tape covered the area between the string and the flower (i.e. so they would not be able to see/ learn whether it was connected or disconnected).

- In experiments 2-6, bees chose the connected string in the test phase.

- In experiment 7, bees were trained as in expt 3 and then tested where string was either disconnected or coiled i.e. still being 'functional' but appearing different.

- In experiment 8, bees were trained as before and then tested on string that was in a different coiled orientation, either connected or disconnected.

- In experiments 7 and 8 the bees showed no preference.

Strengths:

I appreciate the amount of work that has gone into these experiments and think they are a nice, thorough set of experiments. I enjoyed reading the paper and felt that it was overall well-written and clear. I think experiment 1 shows that bees do not have an untrained understanding of the function of the string in this context. The rest of the experiments indicate that with training, bees have a preference for unbroken over broken string and likely use visual cues learned during training to make this choice. They also show that as in other contexts, bees readily generalize across different colors.

The 'weaknesses' that I previously listed were dealt with by the authors in the revised version of the manuscript. I think the only point that we disagreed on was relating to the ecological relevance of the task to the bees.

Here is my previous comment:

I think the paper would be made stronger by considering the natural context in which the bee performs this behavior. Bees manipulate flowers in all kinds of contexts, and scrabble with their legs to achieve nectar rewards. Rather than thinking that it is pulling a string, my guess would be that the bee learns that a particular motor pattern within their usual foraging repertoire (scrabbling with legs), leads to a reward. I don't think this makes the behavior any less interesting - in fact, I think considering the behavior through an ecological lens can help make better sense of it.

The authors disagreed, writing the following:

"Here we respectfully disagree. The solving of Rubik s cube by humans could be said to be version of finger movements naturally required to open nuts or remove ticks from fur, but this is somewhat beside the point: it s not the motor<br /> sequences that are of interest, but the cognition involved. A general approach in work on animal intelligence and cognition is to deliberately choose paradigms that are outside the animals daily routines this is what we have done here, in asking whether there is means end comprehension in bee problem solving. Like comparable studies on this question in other animals, the experiments are designed to probe this question, not one of ecological validity."

I think the difference would be that humans know that they are doing a rubik's cube whereas I do not think that the bee knows that it is pulling string- I think the bee thinks that it is foraging on a flower. Therefore, I stand by my statement that I think it's worth considering what the bee is experiencing in this task and how it relates to what it would be doing while foraging. I think that as animal cognition researchers we can design tasks that are distinct from what the animal would naturally encounter to ask specific questions about what they are thinking- but that we can never remove the ecological context since the animal will always be viewing the task through that lens. However, I think this may be a philosophical difference in opinion and I am happy with the manuscript as it stands.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

In this paper, the researchers aimed to address whether bees causally understand string-pulling through a series of experiments. I first briefly summarize what they did:

- In experiment 1, the researchers trained bees without string and then presented them with flowers in the test phase that either had connected or disconnected strings, to determine what their preference was without any training. Bees did not show any preference.

- In experiment 2, bees were trained to have experience with string and then tested on their choice between connected vs. disconnected string.

- experiment 3 was similar except that instead of having one option which was an attached string broken in the middle, the string was completely disconnected from the flower.

- In experiment 4, bees were trained on green strings and tested on white strings to determine if they generalize across color.

- In experiment 5, bees were trained on blue strings and tested on white strings.

- In experiment 6, bees were trained where black tape covered the area between the string and the flower (i.e. so they would not be able to see/ learn whether it was connected or disconnected).

- In experiments 2-6, bees chose the connected string in the test phase.

- In experiment 7, bees were trained as in experiment 3 and then tested where the string was either disconnected or coiled i.e. still being 'functional' but appearing different.

- In experiment 8, bees were trained as before and then tested on a string that was in a different coiled orientation, either connected or disconnected.

- In experiments 7 and 8 the bees showed no preference.

Strengths:

I appreciate the amount of work that has gone into this study and think it contains a nice, thorough set of experiments. I enjoyed reading the paper and felt that overall it was well-written and clear. I think experiment 1 shows that bees do not have an untrained understanding of the function of the string in this context. The rest of the experiments indicate that with training, bees have a preference for unbroken over broken string and likely use visual cues learned during training to make this choice. They also show that as in other contexts, bees readily generalize across different colors.

Weaknesses:

(1) I think there are 2 key pieces of information that can be taken from the test phase - the bees' first choice and then their behavior across the whole test. I think the first choice is critical in terms of what the bee has learned from the training phase - then their behavior from this point is informed by the feedback they obtain during the test phase. I think both pieces of information are worth considering, but their behavior across the entire test phase is giving different information than their first choice, and this distinction could be made more explicit. In addition, while the bees' first choice is reported, no statistics are presented for their preferences.

We agree with the reviewer that the first choice is critical in terms of what the bumblebees have learned from the training phase. We analyzed the bees’ first choice in Table 1, and we added the tested videos. The entire connected and disconnected strings were glued to the floor, the bees were unable to move either the connected or disconnected strings, and avoid learning behavior during the tests. We added the data of bee's each choice in the Supplementary table.

(2) It seemed to me that the bees might not only be using visual feedback but also motor feedback. This would not explain their behavior in the first test choice, but could explain some of their subsequent behavior. For example, bees might learn during training that there is some friction/weight associated with pulling the string, but in cases where the string is separated from the flower, this would presumably feel different to the bee in terms of the physical feedback it is receiving. I'd be interested to see some of these test videos (perhaps these could be shared as supplementary material, in addition to the training videos already uploaded), to see what the bees' behavior looks like after they attempt to pull a disconnected string.

We added supplementary videos of testing phase. As noted in General Methods, both connected and disconnected strings were glued to the floor to prevent the air flow generated by flying bumblebees’ wings from changing the position of the string during the testing phase. The bees were unable to move either the connected or disconnected strings during the tests, and only attempted to pull them. Therefore, the difference in the friction/weight of pulling the both strings cannot be a factor in the test.

(3) I think the statistics section needs to be made clearer (more in private comments).

We changed the statistical analysis section as suggested by the reviewer.

(4) I think the paper would be made stronger by considering the natural context in which the bee performs this behavior. Bees manipulate flowers in all kinds of contexts and scrabble with their legs to achieve nectar rewards. Rather than thinking that it is pulling a string, my guess would be that the bee learns that a particular motor pattern within their usual foraging repertoire (scrabbling with legs), leads to a reward. I don't think this makes the behavior any less interesting - in fact, I think considering the behavior through an ecological lens can help make better sense of it.

Here we respectfully disagree. The solving of Rubik’s cube by humans could be said to be version of finger-movements naturally required to open nuts or remove ticks from fur, but this is somewhat beside the point: it’s not the motor sequences that are of interest, but the cognition involved. A general approach in work on animal intelligence and cognition is to deliberately choose paradigms that are outside the animals’ daily routines-this is what we have done here, in asking whether there is means-end comprehension in bee problem solving. Like comparable studies on this question in other animals, the experiments are designed to probe this question, not one of ecological validity.

Reviewer #2 (Public Review):

Summary:

The authors wanted to see if bumblebees could succeed in the string-pulling paradigm with broken strings. They found that bumblebees can learn to pull strings and that they have a preference to pull on intact strings vs broken ones. The authors conclude that bumblebees use image matching to complete the string-pulling task.

Strengths:

The study has an excellent experimental design and contributes to our understanding of what information bumblebees use to solve a string-pulling task.

Weaknesses:

Overall, I think the manuscript is good, but it is missing some context. Why do bumblebees rely on image matching rather than causal reasoning? Could it have something to do with their ecology? And how is the task relevant for bumblebees in the wild? Does the test translate to any real-life situations? Is pulling a natural behaviour that bees do? Does image matching have adaptive significance?

We appreciate the valuable comment from the reviewer. Our explanation, which we have now added to the manuscript, is as follows:

“Different flower species offer varying profitability in terms of nectar and pollen to bumblebees; they need to make careful choices and learn to use floral cues to predict rewards (Chittka, 2017). Bumblebees can easily learn visual patterns and shapes of flower (Meyer-Rochow, 2019); they can detect stimuli and discriminate between differently coloured stimuli when presented as briefly as 25 ms (Nityananda et al., 2014). In contrast, causal reasoning involves understanding and responding to causal relationships. Bumblebees might favor, or be limited to, a visual approach, likely due to the efficiency and simplicity of processing visual cues to solve the string-pulling task. ”

As above, it worth noting that our work is not designed as an ecological study, but one about the question of whether causal reasoning can explain how bees solve a string-pulling puzzle. We have a cognitive focus, in line with comparable studies on other animals. We deliberately chose a paradigm that is to some extent outside of the daily challenges of the animal.

Reviewer #3 (Public Review):

Summary:

This paper presents bees with varying levels of experience with a choice task where bees have to choose to pull either a connected or unconnected string, each attached to a yellow flower containing sugar water. Bees without experience of string pulling did not choose the connected string above chance (experiment 1), but with experience of horizontal string pulling (as in the right-hand panel of Figure 4) bees did choose the connected string above chance (experiments 2-3), even when the string colour changed between training and test (experiments 4-5). Bees that were not provided with perceptual-motor feedback (i.e they could not observe that each pull of the string moved the flower) during training still learned to string pull and then chose the connected string option above chance (experiment 6). Bees with normal experience of string pulling then failed to discriminate between connected and unconnected strings when the strings were coiled or looped, rather than presented straight (experiments 7-8).

Weaknesses:

The authors have only provided video of some of the conditions where the bees succeeded. In general, I think a video explaining each condition and then showing a clip of a typical performance would make it much easier to follow the study designs for scholars. Videos of the conditions bees failed at would be highly useful in order to compare different hypotheses for how the bees are solving this problem. I also think it is highly important to code the videos for switching behaviours. When solving the connected vs unconnected string tasks, when bees were observed pulling the unconnected string, did they quickly switch to the other string? Or did they continue to pull the wrong string? This would help discriminate the use of perceptual-motor feedback from other hypotheses.

We added the test videos as suggested by the reviewer, and we added the data for each bee's choice. However, both connected and disconnected strings were glued to the floor, and therefore perceptual-motor feedback was equal and irrelevant between the choices during the test.

The experiments are also not described well, for my below comments I have assumed that different groups of bees were tested for experiments 1-8, and that experiment 6 was run as described in line 331, where bees were given string-pulling training without perceptual feedback rather than how it is described in Figure 4B, which describes bees as receiving string pulling training with feedback.

We now added figures of Experiment 6 and 7 in the Figure 1B, and we mentioned that different groups of bees were tested for Experiments 1-9.

The authors suggest the bees' performance is best explained by what they term 'image matching'. However, experiment 6 does not seem to support this without assuming retroactive image matching after the problem is solved. The logic of experiment 6 is described as "This was to ensure that the bees could not see the familiar "lollipop shape" while pulling strings....If the bees prefer to pull the connected strings, this would indicate that bees memorize the arrangement of strings-connected flowers in this task." I disagree with this second sentence, removing perceptual feedback during training would prevent bees memorising the lollipop shape, because, while solving the task, they don't actually see a string connected to a yellow flower, due to the black barrier. At the end of the task, the string is now behind the bee, so unless the bee is turning around and encoding this object retrospectively as the image to match, it seems hard to imagine how the bee learns the lollipop shape.

We agree with the reviewer that while solving the task in the last step during training, the bees don't actually see a string connected to a yellow flower, due to the black barrier. Since the full shape is only visible after the pulling is completed and this requires the bee to “check back” on the entire display after feeding, to basically conclude “ this is the shape that I need to be looking for later”.

Another possibility is that bumblebees might remember the image of the “lollipop shape” while training the bees in the first step, in which the “lollipop shape” was directly presented to the bumblebee in the early step of the training.

We added the experiment suggested by the reviewer, and the result showed that when a green table was placed behind the string to obscure the “lollipop shape” at any point during the training phase, the bees were unable to identify the connected string. The result further supports that bumblebees learn to choose the connected string through image matching.

Despite this, the authors go on to describe image matching as one of their main findings. For this claim, I would suggest the authors run another experiment, identical to experiment 6 but with a black panel behind the bee, such that the string the bee pulls behind itself disappears from view. There is now no image to match at any point from the bee's perspective so it should now fail the connectivity task.

Strengths:

Despite these issues, this is a fascinating dataset. Experiments 1 and 2 show that the bees are not learning to discriminate between connected and unconnected stimuli rapidly in the first trials of the test. Instead, it is clear that experience in string pulling is needed to discriminate between connected and unconnected strings. What aspect of this experience is important? Experiment 6 suggests it is not image matching (when no image is provided during problem-solving, but only afterward, bees still attend to string connectivity) and casts doubt on perceptual-motor feedback (unless from the bee's perspective, they do actually get feedback that pulling the string moves the flower, video is needed here). Experiments 7 and 8 rule out means-end understanding because if the bees are capable of imagining the effect of their actions on the string and then planning out their actions (as hypotheses such as insight, means-end understanding and string connectivity suggest), they should solve these tasks. If the authors can compare the bees' performance in a more detailed way to other species, and run the experiment suggested, this will be a highly exciting paper

We appreciate the valuable comment from the reviewer. We compared the bees' performance to other species, and conducted the experiment as suggested by the reviewer.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Smaller comments:

Line 64: is the word 'simple' needed here? It could also be explained by more complex forms of associative learning, no?

We deleted “simple”.

Methods:

Line 230: was it checked that this was high-contrast for the bees?

We added the relevant reference in the revised manuscript.

Line 240: how much sucrose solution was present in the flowers?

We added 25 microliters sucrose solution in the flowers. We added the information in the revised manuscript.

Line 266: check grammar.

We checked the grammar as follows: “During tests, both strings were glued to the floor of the arena to prevent the air flow generated by flying bumblebees’ wings from changing the position of the string.”

Statistical analysis:

- What does it mean that "Bees identity and colony were analyzed with likelihood ratio tests"?

Bees identity and colony was set as a random variable. We changed the analysis methods in the revised manuscript, and results of the all the experiments did not changed.

- Line 359: do you mean proportion rather than percentage?

We mean the percentage.

- "the number of total choices as weights" - this should be explained further. This is the number of choices that each bee made? What was the variation and mean of this number? If bees varied a lot in this metric, it might make more sense to analyze their first choice (as I see you've done) and their first 10 choices or something like that - for consistency.

This refers to the total number of choices made by each bumblebee. We added the mean and standard error of each bee’s number of choices in Table 1. Some bees pulled the string fewer than 10 times; we chose to include all choices made by each bee.

- More generally I think the first test is more informative than the subsequent choices, since every choice after their first could be affected by feedback they are getting in that test phase. Or rather, they are telling you different things.

All the bees were tested only once, however, you might be referring to the first choice. We used Chi-square test to analyze the bumblebees’ first choices in the test. It is worth noting that both connected and disconnected strings were glued to the floor. The bees were unable to move either the connected or disconnected strings during the tests, and only attempted to pull them. Therefore,the feedback from pulling either the connected or disconnected strings is the same.

- Line 362: I think I know what you mean, but this should be re-phrased because the "number of" sounds more appropriate for a Poisson distribution. I think what you are testing is whether each individual bee chose the connected or the disconnected string - i.e. a 0 or 1 response for each bee?

We agree with the reviewer that each bee chose the connected or the disconnected string - i.e. a 0 or 1 response for each bee, but not the number. We clarify this as: “The total number of the choices made by each bee was set as weights.” 

- Line 364-365: here and elsewhere, every time you mention a model, make it clear what the dependent and independent variables are. i.e. for the mixed model, the 'bee' is the random factor? Or also the colony that the bee came from? Were these nested etc?

We clarify this in the revised manuscript. The bee identity and colony is the random factor in the mixed model.

- Line 368: "Latency to the first choice of each bee was recorded" - why? What were the hypotheses/ predictions here?

The latency to the first choice was intended to see if the bumblebees were familiarizing with the testing pattern. A shorter delay time might indicate that the bumblebees were more familiar with the pattern.

- Line 371: "Multiple comparisons among experiments were.." - do you mean 'within' experiments? It seems that treatments should not be compared between different experiments.

We mean multiple comparisons among different experiments; we clarify this in the revised manuscript.

Results

Experiment 1: From the methods, it sounded like you both analyzed the bees' first choice and their total no. of choices, but in the results section (and Figure 1) I only see the data for all choices combined here.

In table 1 and in the text you report the number of bees that chose each option on their first choice, but there are no statistical results associated with these results. At the very least, a chi square or binomial test could be run.

Line 138: "Interestingly, ten out of fifteen bees pulled the connected string in their first choice" - this is presented like it is a significant majority of bees, but a chi-square test of 10 vs 5 has a p-value = 0.1967

We used the Chi square test to analyzed of the bees’ first choice. We also added the analyzed data in the Table 1.

Line 143: "It makes sense because the bees could see the "lollipop shape" once they pulled it out from the table." - this feels more like interpretation (i.e. Discussion) rather than results.

We moved the sentence to the discussion.

Line 162: again this feels more like interpretation/ conjecture than results.

We removed the sentence in the results.

Line 184: check grammar.

We checked the grammar. We changed “task” to “tasks”.

Figures

I really appreciated the overview in Figure 5 - though I think this should be Figure 1? Even if the methods come later in eLife, I think it would be nice to have that cited earlier on (e.g. at the start of the results) to draw the reader's attention to it quickly, since it's so helpful. It also then makes the images at the bottom of what is currently Figure 1 make more sense. I also think that the authors could make it clearer in Figure 5 which strings are connected vs disconnected in the figure (even if it means exaggerating the distance more than it was in real life). I had to zoom in quite a bit to see which were connected vs. not. Alternatively, you could have an arrow to the string with the words "connected" "disconnected" the first time you draw it - and similar labels for the other string conditions.

We appreciate the valuable comment from the reviewer. We changed Figure 5 to Figure 2, and Figure 4 to Figure 1. We cited the Figures at the start of the results. We also changed the gap distance between the disconnected strings. Additionally, we added arrows to indicate “connected” and “disconnected” strings in the Figure.

Figure 1 - I think you could make it clearer that the bars refer to experiments (e.g. have an x-axis with this as a label). Also, check the grammar of the y-axis.

We added the experiments number in the Figures. Additionally, we checked the grammar of the y-axis. We changed “percentages” to “parentage”. 

I also think it's really helpful to see the supplementary videos but I think it would be nice to see some examples of the test phase, and not just the training examples.

We added Supplementary videos of the testing phase.

Reviewer #2 (Recommendations For The Authors):

Below are also some minor comments:

L40: "approaches".

We changed “approach” to “approaches”.

L42: but likely mainly due to sampling bias of mammals and birds.

We changed the sentence as follows: String pulling is one of the most extensively used approaches in comparative psychology to evaluate the understanding of causal relationships (Jacobs & Osvath, 2015), with most research focused on mammals and birds, where a food item is visible to the animal but accessible only by pulling on a string attached to the reward (Taylor, 2010; Range et al., 2012; Jacobs & Osvath, 2015; Wakonig et al., 2021).

L64: remove "in this study"

We removed “in this study”.

L64: simple associative learning of what? Isn't your image matching associative too?

We removed “ simple”.

L97: remove "a" before "connected".

We removed “a” before “connected”.

L136-138: but maybe they could still feel the weight of the flower when pulling?

Because both strings were glued to the floor in the test phase, the feedback was the same and therefore irrelevant. This information is noted in the General Methods.

L161: what are these numbers?

We removed the latency in the revised manuscript.

L167/ Table 1: I realise that the authors never tried slanted strings to check if bumblebees used proximity as a cue. Why?

This was simply because we wanted to focus on whether bumblebees could recognize the connectivity of the string.

Discussion: Why did you only control for colour of the string? What if you had used strings with different textures or smells? Unclear if the authors controlled for "bumblebee smell" on the strings, i.e., after a bee had used the string, was the string replaced by a new one or was the same one used multiple times?

We used different colors to investigate featural generalization of the visual display of the string connected to the flower in this task. We controlled for color because it is a feature that bumblebees can easily distinguish.

Both the flowers and the strings were used only once, to prevent the use of chemosensory cues. We clarify this in the revised manuscript.

L182: since what?

We deleted “since” in the revised manuscript.

L182-188: might be worth mentioning that some crows and parrots known for complex cognition perform poorly on broken strings (e.g., https://doi.org/10.1098/rspb.2012.1998 ; https://doi.org/10.1163/1568539X-00003511 ; https://doi.org/10.1038/s41598-021-94879-x ) and Australian magpies use trial and error (https://doi.org/10.1007/s00265-023-03326-6).

We added the following sentences as suggested by the reviewer: “It is worth noting that some crows and parrots known for complex cognition perform poorly on the broken string task without perceptual feedback or learning. For example, New Caledonian crows use perceptual feedback strategies to solve the broken string-pulling task, and no individual showed a significant preference for the connected string when perceptual feedback was restricted (Taylor et al., 2012). Some Australian magpies and African grey parrots can solve the broken string task, but they required a high number of trials, indicating that learning plays a crucial role in solving this task (Molina et al., 2019; Johnsson et al., 2023).”

L193: maybe expand on this to put the task into a natural context?

We added the following sentences as suggested by the reviewer:

“Different flower species offer varying profitability in terms of nectar and pollen to bumblebees; they need to make careful choices and learn to use floral cues to predict rewards (Chittka, 2017). Bumblebees can easily learn visual patterns and shapes of flower (Meyer-Rochow, 2019); they can detect stimuli and discriminate between differently coloured stimuli when presented as briefly as 25 ms (Nityananda et al., 2014). In contrast, causal reasoning involves understanding and responding to causal relationships. Bumblebees might favor, or be limited to, a visual approach, likely due to the efficiency and simplicity of processing visual cues to solve the string-pulling task. ”

L204: is causal understanding the same as means-end understanding?

Means-end understanding is expressed as goal-directed behavior, which involves the deliberate and planned execution of a sequence of steps to achieve a goal. Includes some understanding of the causal relationship (Jacobs & Osvath, 2015; Ortiz et al., 2019). .

L235: this is a very big span of time. Why not control for motivation? Cognitive performance can vary significantly across the day (at least in humans).

Bumblebee motivation is understood to be rather consistent, as those that were trained and tested came to the flight arena of their own volition and were foragers looking to fill their crop load each time to return it to the colony.

L232: what is "(w/w)" ? This occurs throughout the manuscript.

“w/w” represents the weight-to-weight percentage of sugar.

L250: this sentence sounds odd. "containing in the central well.." ?? Perhaps rephrase? Unclear what central well refers to? Did the flowers have multiple wells?

We rephrased the sentence as follows: For each experiment, bumblebees were trained to retrieve a flower with an inverted Eppendorf cap at the center, containing 25 microliters of 50% sucrose solution, from underneath a transparent acrylic table

L268: why euthanise?

The reason for euthanizing the bees is that new foragers will typically only become active after the current ones were removed from the hive.

L270: chemosensory cues answer my concern above. Maybe make it clear earlier.

We moved this sentence earlier in the result.

L273: did different individuals use different pulling strategies? Do you have the data to analyse this? This has been done on birds and would offer a nice comparison.

We analyzed the string-pulling strategies among different individuals, and provided Supplementary Table 1 to display the performances of each individual in different string-pulling experiments.

L365: unclear why both models. Would be nice to see a GLM output table.

The duration of pulling different kinds of strings were first tested with the Shapiro-Wilk test to assess data normality. The duration data that conforms to a normal distribution was compared using linear mixed-effects models (LMM), while the data that deviates from normality were examined with a generalized linear-mixed model (GLMM). We added a GLM and GLMM output table in the revised manuscript.

L377: should be a space between the "." and "This".

We added a space between the “.” and “This”.

L383-390: some commas and semicolons are in the wrong places.

We carefully checked the commas and semicolons in this sentence.

Reviewer #3 (Recommendations For The Authors):

Minor comments

Line 32: seems to be missing a word, suggest "the bumblebees' ability to distinguish".

we added “the” in the revised manuscript.

Line 47: it would be good to reference other scholars here, this is the central focus of all work in comparative psychology.

We added the reference in the revised manuscript.

Line 50-61: I think the string-pulling literature could be described in more detail here, with mention of perceptual-motor feedback loops as a competing hypothesis to means-end understanding (see Taylor et al 2010, 2012). It seems a stretch to suggest that "String-pulling studies have directly tested means-end comprehension in various species", when perceptual-motor feedback is a competing hypothesis that we have positive evidence for in several species.

We mentioned the perceptual-motor feedback in the introduction as follow:

“Multiple mechanisms can be involved in the string-pulling task, including the proximity principle, perceptual feedback and means-end understanding (Taylor et al., 2012; Wasserman et al., 2013; Jacobs & Osvath, 2015; Wang et al., 2020). The principle of proximity refers to animals preferring to pull the reward that is closest to them (Jacobs & Osvath, 2015). Taylor et al. (2012) proposed that the success of New Caledonian crows in string-pulling tasks is based on a perceptual-motor feedback loop, where the reward gradually moves closer to the animal as they pull the strings. If the visual signal of the reward approaching is restricted, crows with no prior string-pulling experience are unable to solve the broken string task (Taylor et al., 2012).

However, when a green table was placed behind the string to obscure the “lollipop” structure during the training, the bees could not see the “lollipop” during the initial training stage or after pulling the string from under the table. In this situation, the bees were unable to identify the connected string, further proving that bumblebees chose the connected string based on image matching.

Line 68: suggest remove 'meticulously'.

We removed “meticulously”.

Line 99: This is an exciting finding, can the authors please provide a video of a bee solving this task on its first trial?

We added videos in the supplementary materials.

Line 133: perceptual-motor feedback loops should be introduced in the introduction.

We introduced perceptual-motor feedback loops in the revised manuscript.

Line 136: please clarify the prior experience of these bees, it is not clear from the text.

We clarified the prior experience of these bees as follow: Bumblebees were initially attracted to feed on yellow artificial flowers, and then trained with transparent tables covered by black tape (S7 video) through a four-step process.

Line 138: from the video it is not possible to see the bee's perspective of this occlusion. Do the authors have a video or image showing the feedback the bees received? I think this is highly important if they wish to argue that this condition prevents the use of both image matching and a perceptual-motor feedback loop.

We prevented the use of image matching: the bees were unable to see the flower moving towards them above the table during the training phase in this condition. But the bees may receive visual image both after pulling the string out from the table and in the initial stages of training in this condition.

Line 147: please clarify what experience these bees had before this test.

We added the prior experience of bumblebees before training as follow: We therefore designed further experiments based on Taylor et al. (2012) to test this hypothesis. Bumblebees were first trained to feed on yellow artificial, and then trained with the same procedure as Experiment 2, but the connected strings were coiled in the test.

Line 155: This is a highly similar test to that used in Taylor et al 2012, have the authors seen this study?

We mentioned the reference in the revised manuscript as follows: We therefore designed further experiments based on Taylor et al. (2012) to test this hypothesis.

Line 183: This sentence needs rewriting "Since the vast majority of animals, including dogs 183 (Osthaus et al., 2005), cats (Whitt et al., 2009), western scrub-jays (Hofmann et al.,2016) and azure-winged magpies (Wang et al., 2019) are failing in such tasks spontaneously".

We changed the sentence as suggested by the reviewer as follow:  Some animals, including dogs (Osthaus et al., 2005), cats (Whitt et al., 2009), western scrub-jays (Hofmann et al., 2016) and azure-winged magpies (Wang et al., 2019) fail in such task spontaneously.

Line 186: "complete comprehension of the functionality of strings is rare" I am not sure the evidence in the current literature supports any animal showing full understanding, can the authors explain how they reach this conclusion?

We wished to say that few animal species could distinguish between connected and disconnected strings without trial and error learning. We revised the sentence as follows:

It is worth noting that some crows and parrots known for complex cognition perform poorly on broken string task without perceptual feedback or learning. For example, New Caledonian crows use perceptual feedback strategies to solve broken string-pulling task, and no individual showed a significant preference for the connected string when perceptual feedback is restricted (Taylor et al., 2012). Some Australian magpies and African grey parrots can solve the broken string task, but it required a high number of trials, indicating that learning plays a crucial role in solving this task (Molina et al., 2019; Johnsson et al., 2023).

Line 190: the authors need to clarify which part of their study provides positive evidence for this conclusion.

We added the evidence for this conclusion as follows: Our findings suggest that bumblebees with experience of string pulling prefer the connected strings, but they failed to identify the interrupted strings when the string was coiled in the test.

Line 265: was the far end of the string glued only?

The entire string was glued to the floor, not just the far ends of the string.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews: 

Reviewer #1 (Public Review):

Summary: 

In this paper, the authors used target agnostic MBC sorting and activation methods to identify B cells and antibodies against sexual stages of Plasmodium falciparum. While they isolated some Mabs against PFs48/45 and PFs230, two well-known candidates for "transmission blocking" vaccines, these antibodies' efficacies, as measured by TRA, did not perform as well as other known antibodies. They also isolated one cross-reactive mAb to proteins containing glutamic acid-rich repetitive elements, that express at different stages of the parasite life cycle. They then determined the structure of the Fab with the highest protein binder they could determine through protein microarray, RESA, and observed homotypic interactions. 

Strengths: 

-  Target agnostic B cell isolation (although not a novel methodology). 

-  New cross-reactive antibody with some "efficacy" (TRA) and mechanism (homotypic interactions) as demonstrated by structural data and other biophysical data. 

Weaknesses: 

The paper lacks clarity at times and could benefit from more transparency (showing all the data) and explanations. 

We have added the oocyst count data from the SMFA experiments as Supplementary Table 2, and ELISA binding curves underlying Figure 4B as Supplementary Figure 5.

In particular: 

- define SIFA 

- define TRAbs 

We have carefully gone through the manuscript and have introduced abbreviations at first use, removed unnecessary abbreviations and removed unnecessary jargon to increase readability.

- it is not possible to read the Figure 6B and C panels. 

We regret that the labels in Supplementary Figures 6 and 7 were of poor quality and have now included higher resolution images to solve this issue.

Reviewer #2 (Public Review): 

This manuscript by Amen, Yoo, Fabra-Garcia et al describes a human monoclonal antibody B1E11K, targeting EENV repeats which are present in parasite antigens such as Pfs230, RESAs, and 11.1. The authors isolated B1E11K using an initial target agnostic approach for antibodies that would bind gamete/gametocyte lysate which they made 14 mAbs. Following a suite of highly appropriate characterization methods from Western blotting of recombinant proteins to native parasite material, use of knockout lines to validate specificity, ITC, peptide mapping, SEC-MALS, negative stain EM, and crystallography, the authors have built a compelling case that B1E11K does indeed bind EENV repeats. In addition, using X-ray crystallography they show that two B1E11K Fabs bind to a 16 aa RESA repeat in a head-to-head conformation using homotypic interactions and provide a separate example from CSP, of affinity-matured homotypic interactions. 

There are some minor comments and considerations identified by this reviewer, These include that one of the main conclusions in the paper is the binding of B1E11K to RESAs which are blood stage antigens that are exported to the infected parasite surface. It would have been interesting if immunofluorescence assays with B1E11K mAb were performed with blood-stage parasites to understand its cellular localization in those stages. 

In the current manuscript, we provide multiple lines of evidence that B1E11K binds (with high affinity) to repeats that are present in RESAs, i.e. through micro-array studies, in vitro binding experiments such as Western blot, ELISA and BLI, and through X-ray crystallography studies on B1E11k – repeat peptide complexes. Taken together, we think we provide compelling evidence that B1E11k binds to repeats present in RESA proteins. We do agree that studies on the function of this mAb against other stages of the parasite could be of interest, but as our manuscript focuses on the sexual stage of the parasite, we feel that this is beyond scope of the current work. However, this line of inquiry will be strongly considered in follow up studies.   

Reviewer #3 (Public Review): 

The manuscript from Amen et al reports the isolation and characterization of human antibodies that recognize proteins expressed at different sexual stages of Plasmodium falciparum. The isolation approach was antigen agnostic and based on the sorting, activation, and screening of memory B cells from a donor whose serum displays high transmission-reducing activity. From this effort, 14 antibodies were produced and further characterized. The antibodies displayed a range of transmission-reducing activities and recognized different Pf sexual stage proteins. However, none of these antibodies had substantially lower TRA than previously described antibodies. 

The authors then performed further characterization of antibody B1E11K, which was unique in that it recognized multiple proteins expressed during sexual and asexual stages. Using protein microarrays, B1E11K was shown to recognize glutamate-rich repeats, following an EE-XX-EE pattern. An impressive set of biophysical experiments was performed to extensively characterize the interactions of B1E11K with various repeat motifs and lengths. Ultimately, the authors succeeded in determining a 2.6 A resolution crystal structure of B1E11K bound to a 16AA repeat-containing peptide. Excitingly, the structure revealed that two Fabs bound simultaneously to the peptide and made homotypic antibody-antibody contacts. This had only previously been observed with antibodies directed against CSP repeats. 

Overall I found the manuscript to be very well written, although there are some sections that are heavy on field-specific jargon and abbreviations that make reading unnecessarily difficult. For instance, 'SIFA' is never defined. 

We have carefully gone through the manuscript and have introduced abbreviations at first use, removed unnecessary abbreviations and removed unnecessary jargon to increase readability.

Strengths of the manuscript include the target-agnostic screening approach and the thorough characterization of antibodies. The demonstration that B1E11K is cross-reactive to multiple proteins containing glutamate-rich repeats, and that the antibody recognizes the repeats via homotypic interactions, similar to what has been observed for CSP repeat-directed antibodies, should be of interest to many in the field. 

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors): 

Figure 1 - why only gametes ELISA and Spz or others?  

The volumes of the single B cell supernatants were too small to screen against multiple antigens/parasite stages. As we aimed to isolate antibodies against the sexual stages of the parasite, our assay focused on this stage and supernatants were not tested against other stages. Furthermore, we screened for reactivity against gametes as TRA mAbs likely target gametes rather than other forms of sexual stage parasites.

Figure 2 A 

(a) Wild type (WT) and Pfs48/45 knock-out (KO) gametes.

(b) I am a bit confused about what GMT is vs Pfs48/45 

We have changed the column titles in Figure 2A to “wild-type gametes” and “Pfs48/45 knockout gametes” to improve clarity.  

(c) Binding is high % why is it red? 

We chose to present the results in a heatmap format with a graded color scale, from strong binders in red to weak binders in green. It has now been clarified in the legend of the figure. 

Please state acronyms clearly 

TRA - transmission reducing activity 

SMFA - standard membrane feeding assay 

We have added the full terms to clarify the acronyms.

1123- VRC01 (not O1)

We have corrected this.

Figure 2 C bottom panels, clarify which ones are TRAbs (Assuming the Mabs with over 80% TRA at 500 ug/ml) (right gel) and the ones that are not (left gel)? 

In the Western blot in Figure 2c, we have marked the antibodies with >80% TRA with an asterisk.

Furthermore, we have replaced ‘TRAbs’ by ‘mAbs with >80% TRA at 500 µg/mL’ in the figure legend.

ITC show the same affinity of the Fab to the 2 peptides but not the ELISA, not the BLI/SPR would be more appropriate. Any potential explanation?  

The way binding affinity is determined across various techniques can result in slight differences in determined values. For instance, ELISAs utilize long incubation times with extensive washing steps and involve a spectroscopic signal, isothermal titration calorimetry (ITC) uses calorimetric signal at different concentration equilibriums to extract a KD, and BLI determines kinetic parameters for KD determination. Discrepancies in binding affinities between orthologous techniques have indeed been observed previously in the context of peptide-antibody binding (e.g. PMID: 34788599).

Despite this, regardless of technique, the relative relationships in all three sets of data is the same - higher binding affinity is observed to the longer P2 peptide. This is the main takeaway of the section. As the reviewer suggests, BLI is likely the most appropriate readout here and is the only value explicitly mentioned in the main text. We primarily use ITC to support our proposed binding stoichiometry which is important to substantiate the SEC-MALS and nsEM data in Figure 4H-I. We added the following sentences to help reinforce these points: “The determined binding affinity from our ITC experiments (Table 1) differed from our BLI experiments (Fig. 4D and 4E), which can occur when measuring antibody-peptide interactions. Regardless, our data across techniques all trend toward the same finding in which a stronger binding affinity is observed toward the longer RESA P2 (16AA) peptide.”

Figure 5C - would be helpful to have the peptide sequence above referring to what is E1, E2 etc... 

We added two panels (Figure 5C-D) showcasing the binding interface that shows the peptide numbering in the context of the overall complex. We hope that this will help better orient the reader. 

Figure S4 - maybe highlight in different colors the EENVV, EEIEE, Etc, etc 

Repeats found in the sequence of the various proteins in Figure S4 have now been highlighted with different colors.

Line 163 - why 14 mabs if 11 wells? Isn't it 1 B cell per well? The authors should explain right away that some wells have more than 1 B cell and some have 1 HC, 1LC, and 1 KC. 

We agree that this was somewhat confusing and have modified the text which now reads: “We obtained and cloned heavy and light chain sequences for 11 out of 84 wells. For three wells we obtained a kappa light chain sequence and for five wells a lambda light chain sequence. For three wells we obtained both a lambda and kappa light chain sequence suggesting that either both chains were present in a single B cell or that two B cells were present in the well. For all 14 wells we retrieved a single heavy chain sequence. Following amplification and cloning, 14 mAbs, from 11 wells, were expressed as full human IgG1s (Table S1) (Dataset S1).”

Line 166-167 - were they multiple HC (different ones) as well when Lambda and kappa were present?

This is not clear at first. 

We clarified this point in the text, see also comment above.

Line 177 - expressed Pfs48/45 and Pfs230, is it lacking both or just Pfs48/45 (as stated on line 172)? 

Pfs48/45 binds to the gamete surface via a GPI anchor, while Pfs230 is retained to the surface through binding to Pfs48/45. Hence, the Pfs48/45 knockout parasite will therefore also lack surfacebound Pfs230. We have added a sentence to the Results clarifying this: “The mAbs were also tested for binding to Pfs48/45 knock-out female gametes, which lack surface-bound Pfs48/45 and Pfs230”.

Show the ELISA data used to calculate EC50 in Figure 3. 

ELISA binding curves are now shown as Figure S5.

Line 313-315 - what if you reverse, capture the Fab (peptide too small even if biotinylated?) 

As anticipated by the Reviewer, immobilizing the Fab and dipping into peptide did not yield appreciable signal for kinetic analysis and thus the experiment from this setup is not reported. 

Line 341 - add crystal structure 

This has now been added.

There is a bit too much speculation in the discussion. For e.g. "The B1C5L and B1C5K mAbs were shown to recognize Domain 2 of Pfs48/45 and exhibited moderate potency, as previously described for Abs with such specificity (27). These 2 mAbs were isolated from the same well and shared the same heavy chain; their three similar characteristics thus suggest that their binding is primarily mediated by the heavy chain". Actual data will reinforce this statement. 

As B1C5L and B1C5K recognized domain 2 of Pfs48/45 with similar affinity, this strongly suggests that binding is mediated though the heavy chain. Structural analysis could confirm this statement, but this is out of the scope of this study.  

Reviewer #2 (Recommendations For The Authors): 

Figure 1: This figure provides a description of the workflow. To make it more relevant for the paper, the authors could add relevant numbers as the workflow proceeds. 

(a) For example, how many memory B cells were sorted, how many supernatants were positive, and then how many mAbs were produced? These numbers can be attached to the relevant images in the workflow. 

We modified the figure to include the numbers. 

(b) For the "Supernatant screening via gamete extract ELISA", please change to "Supernatant screening via gamete/gametocyte extract ELISA". 

We modified the statement as suggested. 

Line 155: The manuscript states that 84 wells reacted with gamete/gametocyte lysate. The following sentence states that "Out of the 21 supernatants that were positive...". Can the authors provide the summary of data for all 84 wells or why focus on only 21 supernatants? 

We screened all supernatants against gamete lysate, and only a subset against gametocyte lysate. In total, we found 84 positive supernatants that were reactive to at least one of the two lysates. 21 of those 84 positive were screened against both lysates. We have modified the text to clarify the numbers:

“After activation, single cell culture supernatants potentially containing secreted IgGs were screened in a high-throughput 384-well ELISA for their reactivity against a crude Pf gamete lysate (Fig. S1B). A subset of supernatants was also screened against gametocyte lysate (S1C). In total, supernatants from 84 wells reacted with gamete and/or gametocyte lysate proteins, representing 5.6% of the total memory B cells. Of the 21 supernatants that were screened against both gamete and gametocyte lysates, six recognized both, while nine appeared to recognize exclusively gamete proteins, and six exclusively gametocyte proteins.”

Please note that all 84 positive wells were taken forward for B cell sequencing and cloning. 

Line 171: SIFA is introduced for the first time and should be completely spelled out.

We have corrected this. 

Figure 2: 

(a) In Figure 2A, can you change the column title from "% pos KO GMT" to "% pos Pfs48/45 KO GMT"?

We have changed the column titles.  

(b) In Figure 2B, the SMFA results have been converted to %TRA. Can the authors please provide the raw data for the oocyst counts and number of mosquitoes infected in Supplementary Materials? 

We have added oocyst count data in Table S2, to which we refer in the figure legend. 

(c) For Figure 2F, the authors do have other domains to Pfs230 as described in Inklaar et al, NPJ Vaccines 2023. An ELISA/Western to the other domains could identify the binding site for B2C10L, though we appreciate this is not the central result of this manuscript. 

We thank the reviewer for this suggestion. We are indeed planning to identify the target domain of B2C10L using the previously described fragments, but agree with the reviewer that this not the focus of the current manuscript and decided to therefore not include it in the current report.

Line 116: The word sporozoites appears in subscript and should be corrected to be normal text. 

We have corrected this.

Line 216: Typo "B1E11K" 

We have corrected this.

Materials and Methods: 

(a) PBMC sampling: Please add the ethics approval codes in this section. 

Donor A visited the hospital with a clinical malaria infection and provided informed consent for collection of PBMCs. We have modified the method section to clarify this. 

“Donor A had lived in Central Africa for approximately 30 years and reported multiple malaria infections during that period. At the time of sampling PBMCs, Donor A had recently returned to the Netherlands and visited the hospital with a clinical malaria infection. After providing informed consent, PBMCs were collected, but gametocyte prevalence and density were not recorded.”

(b) Gamete/Gametocyte extract ELISA: Can the authors please provide the concentration of antibodies used for the positive and negative controls (TB31F, 2544, and 399) 

We have added the concentrations for these mAbs in the methods section.

Recombinant Pfs48/45 and Pfs230 ELISA: Please state the concentration or molarity used for the coating of recombinant Pfs48/45 and Pfs230CMB. 

We have added the concentrations, i.e. 0.5 µg/mL, to the methods section.

Western Blotting: The protocol states that DTT was added to gametocyte extracts (Line 594), but Western Blots in Figures 2 and 3 were performed in non-reducing conditions. Please confirm whether DTT was added or not. 

Thank you for noting this. We did not use DTT for the western blots and have removed this line from the methods section.

Reviewer #3 (Recommendations For The Authors): 

Below are a few minor comments to help improve the manuscript. 

(1) In Figure 4E, are the BLI data fit to a 1:1 binding model? The fits seem a bit off, and from ITC and X-ray studies it is known that 2 Fabs bind 1 peptide. The second Fab should presumably have higher affinity than the first Fab since the second Fab will make interactions with both the peptide and the first Fab. It may be better to fit the BLI data to a 2:1 binding model. 

The 2:1 (heterogeneous ligand) model assumes that there are two different independent binding sites. However, the second binding event described is dependent on the first binding event and thus this model also does not accurately reflect the system. Given that there is not an ideal model to fit, we instead are careful about the language used in the main text to describe these results. Additionally, we also include a sentence to the results section to ensure that the proper findings/interpretations are highlighted: “…our data all trend toward the same finding in which a stronger binding affinity is observed toward the longer RESA P2 (16AA) peptide.”

(2) The sidechain interactions shown in Figures 5C and D could probably be improved. The individual residues are just 'floating' in space, causing them to lack context and orientation. 

We added two panels (Fig. 5C-D) showcasing the binding interface that shows the peptide numbering in the context of the overall complex. We hope that this will help orient the reader.  

(3) The percentage of Ramachandran outliers should be listed in Table 2. Presumably, the value is 0.2%, but this is omitted in the current table. 

Table 2 has been modified to include the requested information explicitly.

I think this illustrates why it could be important to, as Nick says, separate our students into a few boxes because it makes it easier to think about, and then we can think of obstacles and solutions that may come up in each group while lesson planning.