GA

GA

GA

GA

ga

ga

ga

ga

Ga

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

GA

Reviewer #3 (Public review):

Here the Wang et al resubmit their manuscript describing the events in the establishment of polarity in MDCK cells cultured in vitro. As with the original version, the description is throughout and is important to the field to report as it establishes a hierarchy of events in polarization, placing Par3 upstream of centrosome positioning and apical membrane component trafficking. Unfortunately, in the revised version, the authors addressed almost none of my points. They did a cursory job of responding in the rebuttal letter but made little attempt to actually address what was being asked or to incorporate any of my suggestions into the manuscript. The particularly egregious examples are cited below:

Comments on revisions:

(1) My original main experimental concern was not addressed: I had originally asked what role microtubules play in the process of polarization (either centrosomal or non-centrosomal). An obvious model is that Gp135, Rab11, etc. are delivered to the AMIS on centrosomal microtubules. Centrosomes might be also be pulled to the AMIS via cortically derived microtubules as is the case in the C. elegans intestine where the centrosome moves apically on apical microtubules via dynein directed transport to the cortically anchored minus ends. The authors do not explore the role of microtubules in the revision, citing that it was not possible to observe the microtubules directly or to perform nocodazole experiments during polarization. Instead, the authors use a relatively new genetic tool to disrupt centrosomal microtubules. They appear to succeed in displacing centrosomal g-tubulin using this tool, but without being able to observe microtubules, a remaining caveat of this experiment is that it is still unclear whether the authors have removed centrosomal microtubules. Compounding this issue is that this tool has never been used in MDCK cells. The authors conclude "we found that cells lacking centrosomal microtubules were still able to polarize and position the centrioles apically.", but they have not shown this, instead the data suggest this conclusion and the authors should acknowledge the caveat that they have no idea whether centrosomal microtubules are abolished. Similarly, the authors also state: "Additionally, although PCNT knockout cells show reduced microtubule nucleation ability, they still recruit a small amount of γ-tubulin". Where are the data that show that microtubule nucleation is reduced in these PCNT knock out cells?

(2) Many of my comments were addressed in the rebuttal, but not in the text.<br /> The non-centrosomal GP135 in Figure 2 is not acknowledged or explained.

That the polarity index does not actually measure polarity, but nuclear-centrosome distance is not acknowledged or explained in the paper.

I still don't believe that the quantification in Figure 3D matches the images I am being shown in Figure 3A. In the centrinone treatment condition, there is certainly an enrichment of GP135 at the AMIS that is not detected in the quantification. The method described in the rebuttal might miss this enrichment if it is offset from line drawn between the centroid of the two nuclei.

Cell height changes in the centrosome depleted cysts are still referenced in the text ("the cell heights of the centrosome-depleted cysts are less uniform"), but no specific data or image is called out. Currently, Figure 3G is referenced, but that is a graph of GP135 intensity

In my original review, I called on the authors to comment on the striking similarity of the mechanisms they documented in MDCK cells to what has been shown in in vivo systems. The authors did not do this, instead restating in the rebuttal some features of what they found. But, the mechanisms shown here are remarkably similar to the polarization of primordia that generate tubular organs in vivo. Perhaps most striking is the similarity to the C> elegans intestine where Par3 localizes to the cortex at the site of an apical MTOC that pulls the centrosome to the apical surface via dynein (Feldman and Priess, 2012). Instead of discussing this similarity, the authors state: "Par3 is likely to regulate centrosome positioning through some intermediate molecules or mechanisms, but its specific mechanism is still unclear and requires further investigation." Given the acetylated tubulin signal emanating from the Par3 positive patch in Figure 5E and F, I suspect similar mechanisms to the C. elegans intestine are at play here. Such a parallel should be noted in the Discussion.

I had originally commented that "I find the results in Figure 6G puzzling. Why is ECM signaling required for Gp135 recruitment to the centrosome. Could the authors discuss what this means?" The authors responded that "The data in Figure 6G do not indicate that ECM signaling is required for the recruitment of Gp135 to the centrosome". In Figure 6G, the localization of GP135 to the centrosome appears significantly delayed compared to its localization to the centrosome in images where cells were cultured in Matrigel. Indeed, the authors argue that the centrosomal localization precedes and contributes to its localization to the AMIS. In the absence of ECM, GP135 localizes to the membrane before it localizes to the centrosome and its localization to the centrosome appears significantly reduced. Thus, my original and current interpretation is that ECM signaling is somehow required for the centrosomal targeting of GP135. One could make a competition argument, i.e. that the cortex in the absence of ECM is somehow a more desirable place to localize than the centrosome, but this experiment also argues that the centrosome does not need to be a source of this material in order for it to end up on the cortex.

(3) There needs to be precision in the language used in many places:

I don't understand this line in the abstract: "When cultured in Matrigel, de novo polarization of a single epithelial cell is often coupled with mitosis." If a cell has divided, it is no longer a single cell.

The authors state in the Introduction "Because of its strong ability to nucleate microtubules, the centrosome functions as the primary microtubule organizing center", but then state ""In polarized epithelial cells, the centrosome is localized at the apical region during interphase, which contributes to the construction of an asymmetric microtubule network conducive to polarized vesicle trafficking". In the latter statement, I assume the authors are describing the well-characterized apical microtubule network in epithelial cells that is non-centrosomal. Thus, the latter sentence is at odds with the former.

The authors continually refer to Par3 as a tight junction protein. "Par3, which controls tight junction assembly to partition the apical surface from the basolateral surface". To my knowledge, PARD3 is an apical protein with similar localization to C. elegans PAR-3 and Drosophila Bazooka. PARD3B is a junctional protein. I assume that the antibody that the authors are using is to PARD3 and not PARD3B? Can the authors please clarify this in the text.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

Wang, Po-Kai, et al., utilized the de novo polarization of MDCK cells cultured in Matrigel to assess the interdependence between polarity protein localization, centrosome positioning, and apical membrane formation. They show that the inhibition of Plk4 with Centrinone does not prevent apical membrane formation, but does result in its delay, a phenotype the authors attribute to the loss of centrosomes due to the inhibition of centriole duplication. However, the targeted mutagenesis of specific centrosome proteins implicated in the positioning of centrosomes in other cell types (CEP164, ODF2, PCNT, and CEP120) did not affect centrosome positioning in 3D cultured MDCK cells. A screen of proteins previously implicated in MDCK polarization revealed that the polarity protein Par-3 was upstream of centrosome positioning, similar to other cell types.

Strengths:

The investigation into the temporal requirement and interdependence of previously proposed regulators of cell polarization and lumen formation is valuable to the community. Wang et al., have provided a detailed analysis of many of these components at defined stages of polarity establishment. Furthermore, the generation of PCNT, p53, ODF2, Cep120, and Cep164 knockout MDCK cell lines is likely valuable to the community.

Weaknesses:

Additional quantifications would highly improve this manuscript, for example it is unclear whether the centrosome perturbation affects gamma tubulin levels and therefore microtubule nucleation, it is also not clear how they affect the localization of the trafficking machinery/polarity proteins. For example, in Figure 4, the authors measure the intensity of Gp134 at the apical membrane initiation site following cytokinesis, but there is no measure of Gp134 at the centrosome prior to this.

We thank the reviewer for this important suggestion. Previous studies have shown that genes encoding appendage proteins and CEP120 do not regulate γ-tubulin recruitment to centrosomes (Betleja, Nanjundappa, Cheng, & Mahjoub, 2018; Vasquez-Limeta & Loncarek, 2021). Although the loss of PCNT reduces γ-tubulin levels, this reduction is partially compensated by AKAP450. Even in the case of PCNT/AKAP450 double knockouts, low levels of γ-tubulin remain at the centrosome (Gavilan et al., 2018), suggesting that it is difficult to completely eliminate γ-tubulin by perturbing centrosomal genes alone.

To directly address this question, in the revised manuscript (Page 8, Paragraph 4; Figure 4—figure supplement 3), we employed a recently reported method to block γ-tubulin recruitment by co-expressing two constructs: the centrosome-targeting carboxy-terminal domain (C-CTD) of CDK5RAP2 and the γ-tubulin-binding domain of NEDD1 (N-gTBD). This approach effectively depleted γ-tubulin and abolished microtubule nucleation at the centrosome (Vinopal et al., 2023). Interestingly, despite the reduced efficiency of apical vesicle trafficking, these cells were still able to establish polarity, with centrioles positioned apically. These results suggest that microtubule nucleation at the centrosomes (centrosomal microtubules) facilitates—but is not essential for—polarity establishment.

Regarding Figure 4, we assume the reviewer was referring to Gp135 rather than Gp134. In the revised manuscript (Page 8, Paragraph 2; Figure 4I), we observed a slight decrease in Gp135 intensity near PCNT-KO centrosomes at the pre-Abs stage. However, its localization at the AMIS following cytokinesis remained unaffected. These results suggest that the loss of PCNT has a limited impact on Gp135 localization. 

Reviewer #2 (Public review):

Summary:

The authors decoupled several players that are thought to contribute to the establishment of epithelial polarity and determined their causal relationship. This provides a new picture of the respective roles of junctional proteins (Par3), the centrosome, and endomembrane compartments (Cdc42, Rab11, Gp135) from upstream to downstream.

Their conclusions are based on live imaging of all players during the early steps of polarity establishment and on the knock-down of their expression in the simplest ever model of epithelial polarity: a cell doublet surrounded by ECM.

The position of the centrosome is often taken as a readout for the orientation of the cell polarity axis. There is a long-standing debate about the actual role of the centrosome in the establishment of this polarity axis. Here, using a minimal model of epithelial polarization, a doublet of daugthers MDCK cultured in Matrigel, the authors made several key observations that bring new light to our understanding of a mechanism that has been studied for many years without being fully explained:

(1) They showed that centriole can reach their polarized position without most of their microtubule-anchoring structures. These observations challenge the standard model according to which centrosomes are moved by the production and transmission of forces along microtubules.

(2) However) they showed that epithelial polarity can be established in the absence of a centriole.

(3) (Somehow more expectedly) they also showed that epithelial polarity can't be established in the absence of Par3.

(4) They found that most other polarity players that are transported through the cytoplasm in lipid vesicles, and finally fused to the basal or apical pole of epithelial cells, are moved along an axis which is defined by the position of centrosome and orientation of microtubules.

(5) Surprisingly, two non-daughter cells that were brought in contact (for 6h) could partially polarize by recruiting a few Par3 molecules but not the other polarity markers.

(6) Even more surprisingly, in the absence of ECM, Par 3 and centrosomes could move to their proper position close to the intercellular junction after cytokinesis but other polarity markers (at least GP135) localized to the opposite, non-adhesive, side. So the polarity of the centrosome-microtubule network could be dissociated from the localisation of GP135 (which was believed to be transported along this network).

Strengths:

(1) The simplicity and reproducibility of the system allow a very quantitative description of cell polarity and protein localisation.

(2) The experiments are quite straightforward, well-executed, and properly analyzed.

(3) The writing is clear and conclusions are convincing.

Weaknesses:

(1) The simplicity of the system may not capture some of the mechanisms involved in the establishment of cell polarity in more physiological conditions (fluid flow, electrical potential, ion gradients,...).

We agree that certain mechanisms may not be captured by this simplified system. However, the model enables us to observe intrinsic cellular responses, minimize external environmental variables, and gain new insights into how epithelial cells position their centrosomes and establish polarity. 

(2) The absence of centriole in centrinone-treated cells might not prevent the coalescence of centrosomal protein in a kind of MTOC which might still orient microtubules and intracellular traffic. How are microtubules organized in the absence of centriole? If they still form a radial array, the absence of a centriole at the center of it somehow does not conflict with classical views in the field.

Previous studies have shown that in the absence of centrioles, centrosomal proteins can relocate to alternative microtubule-organizing centers (MTOCs), such as the Golgi apparatus (Gavilan et al., 2018). Furthermore, centriole loss leads to increased nucleation of non-centrosomal microtubules (Martin, Veloso, Wu, Katrukha, & Akhmanova, 2018). However, these microtubules typically do not form the classical radial array or a distinct star-like organization. 

While this non-centrosomal microtubule network can still support polarity establishment, it does so less efficiently—similar to what is observed in p53-deficient cells undergoing centriole-independent mitosis (Meitinger et al., 2016). Thus, although the absence of centrioles does not completely prevent microtubule-based organization or polarity establishment, it impairs their spatial coordination and reduces overall efficiency compared to a centriole-centered microtubule-organizing center (MTOC). 

(3) The mechanism is still far from clear and this study shines some light on our lack of understanding. Basic and key questions remain:

(a) How is the centrosome moved toward the Par3-rich pole? This is particularly difficult to answer if the mechanism does not imply the anchoring of MTs to the centriole or PCM.

Previous studies have shown that Par3 interacts with dynein, potentially anchoring it at the cell cortex (Schmoranzer et al., 2009). This interaction enables dynein, a minus-enddirected motor, to exert pulling forces on microtubules, thereby promoting centrosome movement toward the Par3-enriched pole.

In our experiments (Figure 4), we attempted to disrupt centrosomal microtubule nucleation by knocking out multiple genes involved in centrosome structure and function, including ODF2 and PCNT. Under these perturbations, γ-tubulin still remained detectable at the centrosome, and we were unable to completely eliminate centrosomal microtubules. 

To address this question more directly, we employed a strategy to deplete γ-tubulin from centrosomes by co-expressing the centrosome-targeting C-terminal domain (C-CTD) of CDK5RAP2 and the γ-tubulin-binding domain of NEDD1 (N-gTBD). As shown in the new data of the revised manuscript (Page 8, Paragraph 4; Figure 4—figure supplement 3), this approach effectively depleted γ-tubulin from centrosomes, thereby abolishing microtubule nucleation at the centrosome. 

Surprisingly, even under these conditions, centrioles remained apically positioned (Page 8, Paragraph 4; Figure 4—figure supplement 3), indicating that centrosomal microtubules are not essential for centrosome movement during polarization.

Given these findings, we agree that the precise mechanism by which the Par3-enriched cortex attracts or guides centrosome movement remains unclear. Although dynein–Par3 interactions may contribute, further studies are needed to elucidate how centrosome repositioning occurs in the absence of microtubule-based pulling forces from the centrosome itself.

(b) What happens during cytokinesis that organises Par3 and intercellular junction in a way that can't be achieved by simply bringing two cells together? In larger epithelia cells have neighbours that are not daughters, still, they can form tight junctions with Par3 which participates in the establishment of cell polarity as much as those that are closer to the cytokinetic bridge (as judged by the overall cell symmetry). Is the protocol of cell aggregation fully capturing the interaction mechanism of non-daughter cells?

We speculate that a key difference between cytokinesis and simple cell-cell contact lies in the presence or absence of actomyosin contractility during the process of cell division. Specifically, contraction of the cytokinetic ring generates mechanical forces between the two daughter cells, which are absent when two non-daughter cells are simply brought together. While adjacent epithelial cells can indeed form tight junctions and recruit Par3, the lack of shared cortical tension and contractile actin networks between non-daughter cells may lead to differences in how polarity is initiated. This mechanical input during cytokinesis may serve as an organizing signal for centrosome positioning. This idea is supported by recent work showing that the actin cytoskeleton can influence centrosome positioning (Jimenez et al., 2021), suggesting that contractile actin structures formed during cytokinesis may contribute to spatial organization in a manner that cannot be replicated by simple aggregation. 

In our experiments, we simply captured two cells that were in contact within Matrigel. We cannot say for sure that it captures all the interaction mechanisms of non-daughter cells, but it does provide a contrast to daughter cells produced by cytokinesis. 

Reviewer #3 (Public review):

Here, Wang et al. aim to clarify the role of the centrosome and conserved polarity regulators in apical membrane formation during the polarization of MDCK cells cultured in 3D. Through well-presented and rigorous studies, the authors focused on the emergence of polarity as a single MDCK cell divided in 3D culture to form a two-cell cyst with a nascent lumen. Focusing on these very initial stages, rather than in later large cyst formation as in most studies, is a real strength of this study. The authors found that conserved polarity regulators Gp135/podocalyxin, Crb3, Cdc42, and the recycling endosome component Rab11a all localize to the centrosome before localizing to the apical membrane initiation site (AMIS) following cytokinesis. This protein relocalization was concomitant with a repositioning of centrosomes towards the AMIS. In contrast, Par3, aPKC, and the junctional components E-cadherin and ZO1 localize directly to the AMIS without first localizing to the centrosome. Based on the timing of the localization of these proteins, these observational studies suggested that Par3 is upstream of centrosome repositioning towards the AMIS and that the centrosome might be required for delivery of apical/luminal proteins to the AMIS.

To test this hypothesis, the authors generated numerous new cell lines and/or employed pharmacological inhibitors to determine the hierarchy of localization among these components. They found that removal of the centrosome via centrinone treatment severely delayed and weakened the delivery of Gp135 to the AMIS and single lumen formation, although normal lumenogenesis was apparently rescued with time. This effect was not due to the presence of CEP164, ODF2, CEP120, or Pericentrin. Par3 depletion perturbed the repositioning of the centrosome towards the AMIS and the relocalization of the Gp135 and Rab11 to the AMIS, causing these proteins to get stuck at the centrosome. Finally, the authors culture the MDCK cells in several ways (forced aggregation and ECM depleted) to try and further uncouple localization of the pertinent components, finding that Par3 can localize to the cell-cell interface in the absence of cell division. Par3 localized to the edge of the cell-cell contacts in the absence of ECM and this localization was not sufficient to orient the centrosomes to this site, indicating the importance of other factors in centrosome recruitment.

Together, these data suggest a model where Par3 positions the centrosome at the AMIS and is required for the efficient transfer of more downstream polarity determinants (Gp135 and Rab11) to the apical membrane from the centrosome. The authors present solid and compelling data and are well-positioned to directly test this model with their existing system and tools. In particular, one obvious mechanism here is that centrosome-based microtubules help to efficiently direct the transport of molecules required to reinforce polarity and/or promote lumenogenesis. This model is not really explored by the authors except by Pericentrin and subdistal appendage depletion and the authors do not test whether these perturbations affect centrosomal microtubules. Exploring the role of microtubules in this process could considerably add to the mechanisms presented here. In its current state, this paper is a careful observation of the events of MCDK polarization and will fill a knowledge gap in this field. However, the mechanism could be significantly bolstered with existing tools, thereby elevating our understanding of how polarity emerges in this system.

We agree that further exploration of microtubule dynamics could strengthen the mechanistic framework of our study. In our initial experiments, we disrupted centrosome function through genetic perturbations (e.g., knockout of PCNT, CEP120, CEP164, and ODF2). However, consistent with previous reports (Gavilan et al., 2018; Tateishi et al., 2013), we found that single-gene deletions did not completely eliminate centrosomal microtubules. Furthermore, imaging microtubule organization in 3D culture presents technical challenges. Due to the increased density of microtubules during cell rounding, we were unable to obtain clear microtubule filament structures—either using α-tubulin staining in fixed cells or SiR-tubulin labeling in live cells. Instead, the signal appeared diffusely distributed throughout the cytosol.

To overcome this, we employed a recently reported approach by co-expressing the centrosome-targeting carboxy-terminal domain (C-CTD) of CDK5RAP2 and the γtubulin-binding domain (gTBD) of NEDD1 to completely deplete γ-tubulin and abolish centrosomal microtubule nucleation (Vinopal et al., 2023). In our new data presented in the revised manuscript (Page 8, Paragraph 4; Figure 4—figure supplement 3), we found that cells lacking centrosomal microtubules were still able to polarize and position the centrioles apically. However, the efficiency of polarized transport of Gp135 vesicles to the apical membrane was reduced. These findings suggest that centrosomal microtubules are not essential for polarity establishment but may contribute to efficient apical transport. 

Reference

Betleja, E., Nanjundappa, R., Cheng, T., & Mahjoub, M. R. (2018). A novel Cep120-dependent mechanism inhibits centriole maturation in quiescent cells. Elife, 7. doi:10.7554/eLife.35439

Gavilan, M. P., Gandolfo, P., Balestra, F. R., Arias, F., Bornens, M., & Rios, R. M. (2018). The dual role of the centrosome in organizing the microtubule network in interphase. EMBO Rep, 19(11). doi:10.15252/embr.201845942

Jimenez, A. J., Schaeffer, A., De Pascalis, C., Letort, G., Vianay, B., Bornens, M., . . . Thery, M. (2021). Acto-myosin network geometry defines centrosome position. Curr Biol, 31(6), 1206-1220 e1205. doi:10.1016/j.cub.2021.01.002

Martin, M., Veloso, A., Wu, J., Katrukha, E. A., & Akhmanova, A. (2018). Control of endothelial cell polarity and sprouting angiogenesis by non-centrosomal microtubules. Elife, 7. doi:10.7554/eLife.33864

Meitinger, F., Anzola, J. V., Kaulich, M., Richardson, A., Stender, J. D., Benner, C., . . . Oegema, K. (2016). 53BP1 and USP28 mediate p53 activation and G1 arrest after centrosome loss or extended mitotic duration. J Cell Biol, 214(2), 155-166. doi:10.1083/jcb.201604081

Schmoranzer, J., Fawcett, J. P., Segura, M., Tan, S., Vallee, R. B., Pawson, T., & Gundersen, G. G. (2009). Par3 and dynein associate to regulate local microtubule dynamics and centrosome orientation during migration. Curr Biol, 19(13), 1065-1074. doi:10.1016/j.cub.2009.05.065

Tateishi, K., Yamazaki, Y., Nishida, T., Watanabe, S., Kunimoto, K., Ishikawa, H., & Tsukita, S. (2013). Two appendages homologous between basal bodies and centrioles are formed using distinct Odf2 domains. J Cell Biol, 203(3), 417-425. doi:10.1083/jcb.201303071

Vasquez-Limeta, A., & Loncarek, J. (2021). Human centrosome organization and function in interphase and mitosis. Semin Cell Dev Biol, 117, 30-41. doi:10.1016/j.semcdb.2021.03.020

Vinopal, S., Dupraz, S., Alfadil, E., Pietralla, T., Bendre, S., Stiess, M., . . . Bradke, F. (2023). Centrosomal microtubule nucleation regulates radial migration of projection neurons independently of polarization in the developing brain. Neuron, 111(8), 1241-1263 e1216. doi:10.1016/j.neuron.2023.01.020.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Figures:

(1) Figure 3 B+C - Although in comparison to Figure 2 it appears the p53 mutation does not affect θN-C, or Lo-c. the figure would benefit from direct comparison to control cells.

We appreciate your suggestion to improve the clarity of the figure. In response, we have revised Figure 3B+C to include control cell data, allowing for clearer side-by-side comparisons in the updated figures. 

(2) Figure 3D - Clarify if both were normalized to time point 0:00 of the p53 KO. The image used appears that Gp135 intensity increases substantially between 0:00 and 0:15 in the figure, but the graph suggests that the intensity is the same if not slightly lower.

Figure 3D – The data were normalized to the respective 0:00 time point for each condition. Because the intensity profile was measured along a line connecting the two nuclei, Gp135 signal could only be detected if it appeared along this line. However, the images shown are maximum-intensity projections, meaning that Gp135 signals from peripheral regions are projected onto the center of the image. This may create the appearance of increased intensity at certain time points (e.g., Figure 3A, p53-KO + CN, 0:00–0:15). 

(3) Figure 4A: The diagram does not accurately represent the effect of the mutations, for example, PCNT mutation likely doesn't completely disrupt PCM (given gamma-tubulin is still visible in the staining), but instead results in its disorganization, Cep164 also wouldn't be expected to completely ablate distal appendages.

Thank you for your comment. We have modified the figure in the revised manuscript (Figure 4A) to more clearly depict the defective DAs. 

(4) Figure 4 + Supplements: A more in-depth characterization of the mutations would help address the previous comment and strengthen the manuscript. Especially as these components have previously been implicated in centrosome transport.

Thank you for your valuable suggestion. As noted in previous studies, CEP164 is essential for distal appendage function and basal body docking, with its loss resulting in blocked ciliogenesis (Tanos et al., 2013); CEP120 is required for centriole elongation and distal appendage formation, and its loss also results in blocked ciliogenesis (Comartin et al., 2013; Lin et al., 2013; Tsai, Hsu, Liu, Chang, & Tang, 2019); ODF2 functions upstream in the formation of subdistal appendages, and its loss eliminates these structures and impairs microtubule anchoring (Tateishi et al., 2013); and PCNT functions as a PCM scaffold, necessary for the recruitment of PCM components and for microtubule nucleation at the centrosome (Fong, Choi, Rattner, & Qi, 2008; Zimmerman, Sillibourne, Rosa, & Doxsey, 2004). 

Given that the phenotypes of these mutants have been well characterized in the literature. Here, we further focus on their roles in centrosome migration and polarized vesicle trafficking within the specific context of our study. 

(5) Figure 4: It would be interesting to measure the Gp135 intensity at the centrosomes, given that the model proposes it is trafficked from the centrosomes to the AMIS.

Thank you for your suggestion. We have included measurements of Gp135 intensity at the centrosomes during the Pre-Abs stage in the revised figure (Figure 4I). Our data show no significant differences in Gp135 intensity between wild-type (WT) and CEP164-, ODF2-, or CEP120-knockout (KO) cell lines. However, a slight decrease in Gp135 intensity was observed in PCNT-KO cells. 

(6) Figure 6F shows that in suspension culture polarity is reversed, however, in Figure 6G gp135 still localizes to the cytokinetic furrow prior to polarity reversal. Given this paper demonstrates Par-3 is upstream of centrosome positioning, it would be important to have temporal data of how Par-3 localizes prior to the ring observed in 6F.

Thank you for your comment. We have included a temporal analysis of Par3 localization using fixed-cell staining in the revised figure (Figure 6—figure supplement 1D). This analysis shows that Par3 also localizes to the cytokinesis site during the Pre-Abs stage, prior to ring formation observed during the Post-CK stage (Figure 6F). Interestingly, during the Pre-Abs stage, the centrosomes also migrate toward the center of the cell doublets in suspension culture, and Gp135 surrounding the centrosomes is also recruited to a region near the center (Figure 6—figure supplement 1E). These data suggest that Par3 also is initially recruited to the cytokinesis site before polarity reversal, potentially promoting centrosome migration. The main difference from Matrigel culture is the peripheral localization of Par3 and Gp135 in suspension, which is likely due to the lack of external ECM signaling. 

Results:

(1) Page 7 Paragraph 1 - consistently use AMIS (Apical membrane initiation site) rather than "the apical site".

Thank you for your helpful comment. We have revised the manuscript (Page 7, Paragraph 1) and will now use "AMIS" (Apical Membrane Initiation Site) instead of "the apical site" throughout the text. 

(2) Page 7 Paragraph 4 - A single sentence explaining why the p53 background had to be used for the Cep120 deletion would be beneficial. Did the cell line have a reduced centrosome number? Does this effect apical membrane initiation similar to centrinone?

We have revised the text (Page 7, Paragraph 4) to clarify that we were unable to generate a CEP120 KO line in p53-WT cells for unknown reasons. CEP120-KO cells have a normal number of centrosome, but their centrioles are shorter. Because this KO line still contains centrioles, the effect is different from centrinone treatment, which results in a complete loss of centrioles. 

(3) Page 10 paragraph 4 - This paragraph is confusing to read. I understand that in the cysts and epithelial sheet the cytokinetic furrow is apical, therefore a movement towards the AMIS could be due to its coincidence with the furrow. However, the phrasing "....we found that centrosomes move towards the apical membrane initiation site direction before bridge abscission. Taken together these findings indicate the position is strongly associated with the site of cytokinesis but not with the apical membrane" is confusing to the reader.

We have revised the manuscript (Page 11, paragraph 4) to change the AMIS as the center of the cell doublet. During de novo epithelial polarization, the apical membrane has not yet formed at the Pre-Abs stage. However, at the Pre-Abs stage, the centrosome has already migrated toward the site of cytokinesis, suggesting that centrosome positioning is correlated with the site of cell division. A similar phenomenon occurs in fully polarized epithelial cysts and sheets, where the centrosomes also migrate before bridge abscission. Thus, we propose that the position of the centrosome is closely associated with the site of cytokinesis and is independent of apical membrane formation. 

Discussion

(1) Page 11, Paragraph 2 - citations needed when discussing previous studies.

Thank you for your suggestion. We have included the necessary references to the discussion of the previous studies in the revised manuscript (Page 12, Paragraph 2). 

(2) Page 12, Paragraph 2 - This section of the discussion would be strengthened by discussing the role of the actomyosin network in defining centrosome position (Jimenez et al., 2021). It seems plausible that the differences observed in the different conditions could be due to altered actomyosin architecture. Especially where the cells haven't undergone cytokinesis.

We appreciate the suggestion of a role for the actomyosin network in determining centrosome positioning. Recent studies have indeed highlighted the role of the actomyosin network in regulating centrosome centering and off-centering (Jimenez et al., 2021). During the pre-abscission stage of cell division, the actomyosin network undergoes significant dynamic changes, with the contractile ring forming at the center and actin levels decreasing at the cell periphery. In contrast, under aggregated cell conditions—meaning cells that have not undergone division—the actomyosin network does not exhibit such dynamic changes. The loss of actomyosin remodeling may therefore influence whether the centrosome moves. Thus, alterations in actomyosin architecture may contribute to the differences observed under various conditions, particularly when cells have not yet completed cytokinesis. We have revised Paragraph 2 on Page 13 to briefly mention the referenced study and to propose that the actomyosin network may influence centrosome positioning, contributing to our observed results. This addition strengthens the discussion and clarifies our findings. 

(3) Page 12 paragraph 3 - Given that centrosome translocation during cytokinesis in MDCK cells (this study) appears to be similar to that observed in HeLa cells and the zebrafish Kupffers vesicle (Krishnan et al., 2022) it would be interesting to discuss why Rab11a and PCNT may not be essential to centrosome positioning in MDCK cells.

Thank you for your insightful comment. We agree that it is interesting that centrosome translocation during cytokinesis in MDCK cells (as observed in our study) is similar to that observed in HeLa cells and zebrafish Kupffer's vesicle (Krishnan et al., 2022). However, there are notable differences between these systems that may help explain why Rab11a and PCNT are not essential for centrosome positioning in MDCK cells.

Our study used 3D culture of MDCK cells, while the reference study examined adherent culture of HeLa cells. In the adherent culture, cells attached to the culture surface form large actin stress fibers on their basal side, which weakens the actin networks in the apical and intercellular regions. In contrast, the 3D culture system used in our study better preserves cell polarity and the integrity of the actin network, which might contribute to centrosome positioning independent of Rab11a and PCNT. Differences in culture conditions and actin network architecture may explain why Rab11a and PCNT are not required for centrosome positioning in MDCK cells.

Furthermore, the referenced study focused on Rab11a and PCNT in zebrafish embryos at 3.3–5 hours post-fertilization (hpf), a time point before the formation of the Kupffer’s vesicle. At this stage, the cells they examined may not yet have become epithelial cells, which may also influence the requirement of Rab11a and PCNT for centrosome positioning. We hypothesize that during the pre-abscission stage, centrosome migration toward the cytokinetic bridge occurs primarily in epithelial cells, and that the polarity and centrosome positioning mechanisms in these cells may differ from those in other cell types, such as zebrafish embryos.

Furthermore, data from Krishnan et al. (2022) suggest that cytokinesis failure in pcnt+/- heterozygous embryos and Rab11a functional-blocked embryos may be due to the presence of supernumerary centrosomes. Consistent with this, our data show that blocking cytokinesis inhibits centrosome movement in MDCK cells. However, in our MDCK cell lines with PCNT or Rab11a knockdown, we did not observe significant cytokinesis failure, and centrosome migration proceeded normally. 

Reviewer #2 (Recommendations for the authors):

Suggestions for experiments:

(1) A description of the organization of microtubules in the absence of centriole, or in the absence of ECM would be interesting to understand how polarity markers end up where you observed them. This easy experiment may significantly improve our understanding of this system.

Previous studies have shown that in the absence of centrioles, microtubule organization undergoes significant changes. Specifically, the number of non-centrosomal microtubules increases, and these microtubules are not radially arranged, leading to the absence of focused microtubule organizing centers in centriolar-deficient cells (Martin, Veloso, Wu, Katrukha, & Akhmanova, 2018). This disorganized microtubule network reduces the efficiency of vesicle transport during de novo epithelial polarization at the mitotic preabscission stage. 

In contrast, the organization of microtubules under ECM-free conditions remains less well characterized. Here, we show that while the ECM plays a critical role in establishing the direction of epithelial polarity, it does not influence the positioning of the centrosome, the microtubule-organizing center (MTOC).  

(2) Would it be possible to knock down ODF2 and pericentrin to completely disconnect the centrosome from microtubules?

ODF2 is the base of subdistal appendages. When ODF2 is knocked out, it affects the recruitment of all downstream proteins to the subdistal appendages (Mazo, Soplop, Wang, Uryu, & Tsou, 2016). One study has shown that ODF2 knockout cells almost completely lost subdistal appendage structures and significantly reduced the microtubule asters surrounding the centrioles (Tateishi et al., 2013). However, although pericentrin (PCNT) is the main scaffold of the pericentriolar matrix (PCM) of centrosomes, the microtubule organization ability of centrosomes can be compensated by AKAP450, a paralog of PCNT, after PCNT knockout. A previous study has even shown that in cells with a double knockout of PCNT and AKAP450, γ-tubulin can still be recruited to the centrosomes, and centrosomes can still nucleate microtubules (Gavilan et al., 2018). This suggests that there are other proteins or pathways that promote microtubule nucleation on centrosomes. We are unsure whether the triple knockout of ODF2, PCNT, and AKAP450 can completely disconnect the centrosome from microtubules. However, a recent study reported a simpler approach involving the expression of dominant-negative fragments of the γ-tubulinbinding protein NEDD1 and the activator CDK5RAP2 at the centrosome (Vinopal et al., 2023). In our revised manuscript (Page 8, Paragraph 4; Figure 4—figure supplement 3), we applied this strategy, which resulted in the depletion of nearly all γ-tubulin from the centrosome. This indicates a strong suppression of centrosomal microtubule nucleation and an effective disconnection of the centrosome from the microtubule network. 

(3) The study does not distinguish the role of cytokinesis from the role of tight junctions, which form only after cytokinesis and not simply by bringing cells into contact. Would it be feasible and interesting to study the polarization after cytokinesis in cells that could not form tight junctions (due to the absence of Ecad or ZO1 for example)?

Studying cell polarization after cytokinesis in cells unable to form tight junctions is a promising area of research.

Recent studies have shown that mouse embryonic stem cells (mESCs) cultured in Matrigel can form ZO-1-labelled tight junctions at the midpoint of cell–cell contact even in the absence of cell division. However, in the absence of E-cadherin, ZO-1 localization is significantly impaired. Interestingly, despite the loss of E-cadherin, the Golgi apparatus and centrosomes remain oriented toward the cell–cell interface (Liang, Weberling, Hii, Zernicka-Goetz, & Buckley, 2022). These findings suggest that cell polarity can be maintained independently of tight junction formation, highlighting the potential value of studying cell polarization that lack tight junctions.

Furthermore, while studies have explored the effects of knocking down tight junction components such as JAM-A and Cingulin on lumen formation in MDCK 3D cultures (Mangan et al., 2016; Tuncay et al., 2015), the role of ZO-1 in this context remains underexplored. Cingulin knockdown has been shown to disrupt endosome targeting and the formation of the AMIS, while both JAM-A and Cingulin knockdown result in actin accumulation at multiple points, leading to the formation of multi-lumen structures rather than a reversal of polarity. However, previous research has not specifically investigated centrosome positioning in JAM-A and Cingulin knockdown cells, an area that could provide valuable insights into how polarity is maintained in the absence of tight junctions. 

Writing details:

(1) The migration of the centrosome in the absence of appendages or PCM is proposed to be ensured by compensatory mechanisms ensuring the robustness of microtubule anchoring to the centrosome. It could also be envisaged that the centrosome motion does not require this anchoring and that other yet unknown moving mechanisms, based on an actin network for example, might exist.

Thank you for your valuable comments. We agree that there may indeed be some unexpected mechanisms that allow centrosomes to move independently of microtubule anchoring to the centrosome, such as mechanisms based on actin filaments or noncentrosomal microtubules; these mechanisms are worth further investigation.

In response to your suggestion, in the Paragraph 5 of the discussion section, we further clarified that while a microtubule anchoring mechanism might be one explanation, other mechanisms could also influence centrosome movement in the absence of appendages or PCM. Additionally, we revised the Paragraph 4 regarding the possibility of actin network-driven centrosome movement and emphasized the importance of future research for a deeper understanding of these processes. 

(2) The actual conclusion of the study of Martin et al (eLife 2018) is not simply that centrosome is not involved in cell polarization but that it hinders cell polarization!

Thank you for your valuable feedback. We agree with the findings of Martin et al. (eLife 2018) that centrosome is not irrelevant to cell polarity, but rather they inhibit cell polarization. Therefore, we have revised the manuscript (Page 2, Paragraph 2) to more accurately reflect this viewpoint. 

(3) This study recalls some conclusions of the study by Burute et al (Dev Cell 2017), in particular the role of Par3 in driving centrosome toward the intercellular junction of daughter cells after cytokinesis. It would be welcome to comment on the results of this study in light of their work.

Thank you for your valuable feedback. The study by Burute et al. (Dev Cell, 2017) showed that in micropattern-cultures of MCF10A cells, the cells exhibit polarity and localize their centrosomes towards the intercellular junction, while downregulation of Par3 gene expression disrupts this centrosome positioning. This result is similar to our findings in 3D cultured MDCK cells and consistent with previous studies in C. elegans intestinal cells and migrating NIH 3T3 cells (Feldman & Priess, 2012; Schmoranzer et al., 2009), indicating that Par3 indeed influences centrosome positioning in different cellular systems. However, Par3 does not directly localize to the centrosome; rather, it localizes to the cell cortex or cell-cell junctions. Therefore, Par3 likely regulates centrosome positioning through other intermediary molecules or mechanisms, but the specific mechanism remains unclear and requires further investigation. 

(4) Could the term apico-basal be used in the absence of a basement membrane to form a basal pole?

We understand that using the term "apico-basal" in the absence of a basement membrane might raise some questions. Traditionally, the apico-basal axis refers to the polarity of epithelial cells, where the apical surface faces the lumen or external environment, and the basal surface is oriented toward the basement membrane. However, in the absence of a basement membrane, such as in certain in vitro systems or under specific experimental conditions, polarity along a similar axis can still be observed. In such cases, the term "apico-basal" can still be used to describe the polarity between the apical domain and the region where it contacts the substrate or adjacent cells. 

(5) The absence of centrosome movement to the intercellular bridge in spread cells in culture is not so surprising considering the work of Lafaurie-Janvore et al (Science 2018) about the role of cell spreading in the regulation of bridge tension and abscission delay.

Thank you for your valuable comment. Indeed, previous studies have shown that in some cell types, the centrosome does move toward the intercellular bridge in spread cells (Krishnan et al., 2022; Piel, Nordberg, Euteneuer, & Bornens, 2001), but other studies have suggested that this movement may not be significant and it may not occur in universally observed across all cell types (Jonsdottir et al., 2010). In our study, we aim to demonstrate that this phenomenon is more pronounced in 3D culture systems compared to 2D spread cell culture systems. Previous studies and our work have observed that centrosome migration occurs during the pre-abscission stage, but whether this migration is directly related to cytokinetic bridge tension or the time of abscission remains an open question. Further research is needed to explore the potential relationship between centrosome positioning, cytokintic bridge tension, and the timing of abscission. 

(6) GP135 (podocalyxin) has been proposed to have anti-adhesive/lubricant properties (hence its pro-invasive effect). Could it be possible that once localized at the cell surface it is systematically moved away from regions that are anchored to either the ECM or adjacent cells? So its localization away from the centrosome in an ECM-free experiment would not be a consequence of defective targeting but relocalization after reaching the plasma membrane?

Thank you for your valuable comment. We agree that GP135 may indeed move directly across the cell surface, away from the region where it interacts with the ECM or adjacent cells. This re-localization could be due to its anti-adhesive or lubricating properties, which may facilitate its displacement from these adhesive sites. To validate this, it is necessary to employ higher-resolution real-time imaging system to observe the dynamic behavior of GP135 on the cell surface.

However, this does not contradict our main conclusion. Under suspension culture conditions without ECM, the centrosome positioning in cell doublets is indeed decoupled from apical membrane orientation. This suggests that the localization of the centrosome and the apical membrane is regulated by different mechanisms. Specifically, the GP135 protein tends to accumulate away from areas of contact with the ECM or adjacent cells, possibly through movement within the cell membrane or by recycling endosome transport. In contrast, centrosome positioning is closely related to the cytokinesis site. Our study clearly elucidates the differences between these two polarity properties. 

Reviewer #3 (Recommendations for the authors):

Major:

(1) To me, a clear implication of these studies is that Gp135, Rab11, etc. are delivered to the AMIS on centrosomal microtubules. The authors do not explore this model except to say that depletion of SD appendage or pericentrin has no effect on the protein relocalization to the AMIS. However, the authors do not observe microtubule association with the centrosome in these KO conditions. This analysis is imperative to interpret existing results since these are new KO conditions in this cell/culture system and parallel pathways (e.g. CDK5RAP2) are known to contribute to microtubule association with the centrosome. An ability to comment on the mechanism by which the centrosome contributes to the efficiency of polarization would greatly enhance the paper.

Microtubule requirement could also be tested in numerous additional ways requiring varying degrees of new experiments:

(a) faster live cell imaging at abscission to see if the deposition of those components appears to traffic on MTs;

(b) live cell imaging with microtubules (e.g. SPY-tubulin) and/or EB1 to determine the origin and polarity of microtubules at the pertinent stages;

For (a) and (b), because the cells were cultured in Matrigel, they tended to be round up, with a dense internal structure that made observation difficult. In contrast, under adherent culture conditions, the cells were flattened with a more dispersed internal structures, making them easier to observe. We had previously used SPY-tubulin to label microtubules for live cell imaging; however, due to the dense microtubule structure in 3D culture, the image contrast was reduced, and we could not clearly observe the microtubule network within the cells. 

(c) acute nocodazole treatment at abscission to determine the effect on protein localization.

Regarding the method of using nocodazole to study microtubule requirements at the abscission stage, we believe that nocodazole treatment may lead to cytokinesis failure. Cell division failure results in the formation of binucleated cells, which are unable to establish cell polarity. Furthermore, nocodazole treatment cannot distinguish between centrosomal and non-centrosomal microtubules, making it unsuitable for studying the specific role of centrosomal microtubules in this process.

In our new data (Figure 4-figure supplementary 3) presented in the revised manuscript, we employed a recently reported method by co-expressing of the centrosome-targeting carboxy-terminal domain (C-CTD) of CDK5RAP2 and the γ-tubulin-binding domain (gTBD) of NEDD1 to completely deplete γ-tubulin and abolish centrosomal microtubule nucleation (Vinopal et al., 2023). We found that cells lacking centrosomal microtubules were still able to polarize and position the centrioles apically. However, the efficiency of polarized transport of Gp135 vesicles to the apical membrane was reduced. These findings suggest that centrosomal microtubules are not essential for polarity establishment but may contribute to facilitate efficient apical transport. 

(2) Similar to the expanded analysis of the role of microtubules in this system, it would be excellent if the author could expand on the role of Par3 and the centrosome, although this reviewer recognizes that the authors have already done substantial work. For example, what are the consequences of Gp135 and/or Rab11 getting stuck at the centrosome? Do the authors have any later images to determine when and if these components ever leave the centrosome? Existing literature focuses on the more downstream consequence of Par3 removal on single-lumen formation. 

Similarly, could the authors expand on the description of polarity disruption following centrinone treatment? It is clear that Gp135 recruitment is disrupted, but how and when do things get fixed and what else is disrupted at the very earliest stages of AMIS formation? The authors have an excellent opportunity to really expand on what is known about the requirements for these conserved components.

Regarding the use of centrinone in treatment, we speculate that Gp135 can still accumulate at the AMIS over time, although the efficiency of its recruitment may be reduced.

Furthermore, under similar conditions, other apical membrane components (such as the Crumbs3 protein) may exhibit similar characteristics to Gp135 protein. 

(3) Perhaps satisfying both of the above asks, could the authors do a faster time-lapse at the relevant time points, i.e. as proteins are being recruited to the AMIS (time points between 1Aiv and v)? This type of imaging again might help shed light on the mechanism.

We believe the above questions are very important and may require further experimental verification in the future. 

Minor:

(1) What is the green patch of Gp135 in Figure 2A that does not colocalize with the centrosome? Is this another source of Gp135 that is being delivered to the AMIS? This type of patch is also visible in Figure 3A 15 and 30-minute panels.

During mitosis, membrane-composed organelles such as the Golgi apparatus are typically dispersed throughout the cytoplasm. However, during the pre-abscission stage, these organelles begin to reassemble and cluster around the centrosome. Furthermore, they also accumulate in the region between the nucleus and the cytokinetic bridge, corresponding to the “patch” mentioned in Figure 2A. 

Live cell imaging results showed that this Gp135 patch initially appears in a region not associated with the centrosome. Subsequently, they were either directly transported to the AMIS or fused with the centrosome-associated Gp135 and transported together. Notably, this patch was only observed when Gp135 was overexpressed in cells. No such distinct protein patches were observed when staining endogenous Gp135 protein (Figure 1A), suggesting that overexpression of Gp135 protein may lead to a localized increase in its concentration in that region. 

(2) I am confused by the "polarity index" quantification as this appears to just be a nucleus centrosome distance measurement and wouldn't, for example, distinguish if the centrosomes separated from the nucleus but were on the basal side of the cell.

The position of the centrosome within the cell (i.e., its distance from the nucleus) can indeed serve as an indicator of cell polarity (Burute et al., 2017). We acknowledge that this quantitative method does not directly capture the specific direction in which the centrosome deviates from the cell center. To address this limitation, we have incorporated information about the angle between the nucleus and the centrosome, which allows for a more accurate description of changes in cell polarity (Rodriguez-Fraticelli, Auzan, Alonso, Bornens, & Martin-Belmonte, 2012). 

(3) How is GP135 "at AMIS" measured? Is an arbitrary line drawn? This is important later when comparing to centrinone treatment in Figure 3D where the quantification does not seem to accurately capture the enrichment of Gp135 that is seen in the images.

To measure the expression level of Gp135 in the "AMIS" region of the cell, we first connected the centers of the two cell nuclei in three-dimensional space to form a straight line. Then, we used the Gp135 expression intensity at the midpoint of this line as the representative value for the AMIS region. This method is based on the assumption that the AMIS region is most likely located between the centers of the two cell nuclei. Therefore, this quantitative method provides a standardized assessment tool for comparing Gp135 expression levels under different conditions. 

(4) The authors reference cell height (p.7) but no data for this measurement are shown

Thank you for the comment. Although we did not perform quantitative measurements, the differences in cell height are clearly visible in Figure 3E (p53-KO + CN), which visually illustrates this phenomenon. 

(5) Can the authors comment on the seeming reduction of Par3 in p53 KO cells?

We did not observe a reduction of Par3 in p53-KO cells in our experiments.

(6) Can the authors make sense of the E-cad localization: Figure 5, Supplement 2.

Our study revealed that E-cadherin begins to accumulate at the cell-cell contact sites during the pre-abscission stage. Its appearance is similar to that of ZO-1, which also appears near the cell division site during this phase. Therefore, the behavior of E-cadherin contrasts sharply with that of Gp135, further highlighting the unique trafficking mechanisms of apical membrane proteins during this process. 

(7) I find the results in Figure 6G puzzling. Why is ECM signaling required for Gp135 recruitment to the centrosome. Could the authors discuss what this means?

We appreciate the reviewer’s valuable comments and thank you for the opportunity to clarify this point. The data in Figure 6G do not indicate that ECM signaling is required for the recruitment of Gp135 to the centrosome. Rather, our findings suggest that even in the absence of ECM, the centrosomes can migrate to a polarized position similar to that in Matrigel culture. This suggests that centrosome migration and the orientation of the nucleus–centrosome axis may be independent of ECM signaling and are primarily driven by cytokinesis alone. 

Regarding the localization of Gp135, previous studies have shown that ECM signaling through integrin promotes endocytosis, which is crucial for the internalization of Gp135 from the cell membrane and its subsequent transport to the AMIS (Buckley & St Johnston, 2022). Our study found that, prior to its accumulation at the AMIS, Gp135 transiently localizes around the centrosome. In the absence of ECM, due to reduced endocytosis, Gp135 primarily remains on the cell membrane and does not undergo intracellular trafficking.  

(8) The authors end the Discussion stating that these studies may have implication for in vivo settings, yet do not discuss the striking similarities to the C. elegans and Drosophila intestine or the findings from any other more observational studies of tubular epithelial systems in vivo (e.g. mouse kidney polarization, zebrafish neuroepithelium, etc.). These models should be discussed.

Thank you for your valuable comment. Indeed, all types of epithelial tissues or tubular epithelial systems in vivo share some common features during cell division, which have been well-documented across various species. 

These features include: during interphase, the centrosome is located at the apical surface of the cells; after the cell enters mitosis, the centrosome moves to the lateral side of the cell to regulate spindle orientation; and during cytokinesis, the cleavage furrow ingresses asymmetrically from the basal to the apical side, with the cytokinetic bridge positioned at the apical surface. Our study using MDCK 3D culture and transwell culture systems successfully mimicked these key features, demonstrating that these in vitro models are of significant value for studying cell polarization dynamics. 

Based on our observations, we speculate that the centrosome may return to the apical surface after anaphase, just before bridge abscission. This is consistent with our findings from studies using MDCK 3D cultures and transwell systems, which showed that the centrosome relocates prior to the final stages of cytokinesis.

Additionally, we propose that de novo polarization of the kidney tubule in vivo may not solely depend on the aggregation and mesenchymal-epithelial transition (MET) of the metanephric mesenchyme. It may also be related to the cell division process, which triggers centrosome migration and polarized vesicle trafficking. These processes likely contribute to enhancing cell polarization, as we observed in our in vitro models.

We hope this will further clarity the potential implications of our findings for in vivo model studies, as well as and their broader impact on the field of tubular epithelial cell polarization research. 

(9) There are several grammatical issues/typos throughout the paper. A careful readthrough is required. For example:

this sentence makes no sense "that the centrosome acts as a hub of apical recycling endosomes and centrosome migration during cytokinetic pre-abscission before apical membrane components are targeted to the AMIS"

We carefully reviewed the paper and made necessary revisions to address the issues raised. In particular, we revised certain sentences to improve clarity and readability (Page 5, Paragraph 3). 

(10) P.8: have been previously reported [to be] involved in MDCK...

We appreciate the reviewer's valuable suggestions. We have revised the sentence accordingly (Page 9, Paragraph 2). 

(11) This sentence seems misplaced: "Cultured conditions influence cellular polarization preferences."

The sentence itself is fine, but to improve the coherence and clarity of the paragraph, we adjusted the paragraph structure and added some transitional phrases (Page 13, Paragraph 1).  

(12) "Play a downstream role in Par3 recruitment" doesn't make sense, this should just be downstream of Par3 recruitment.

Thank you for your suggestion. We have revised the wording accordingly, changing it to "downstream of Par3 recruitment" (Page 10, Paragraph 2).  

Reference

Buckley, C. E., & St Johnston, D. (2022). Apical-basal polarity and the control of epithelial form and function. Nat Rev Mol Cell Biol, 23(8), 559-577. doi:10.1038/s41580-022-00465-y

Burute, M., Prioux, M., Blin, G., Truchet, S., Letort, G., Tseng, Q., . . . Thery, M. (2017). Polarity Reversal by Centrosome Repositioning Primes Cell Scattering during Epithelial-to-Mesenchymal Transition. Dev Cell, 40(2), 168-184. doi:10.1016/j.devcel.2016.12.004

Comartin, D., Gupta, G. D., Fussner, E., Coyaud, E., Hasegan, M., Archinti, M., . . . Pelletier, L. (2013). CEP120 and SPICE1 cooperate with CPAP in centriole elongation. Curr Biol, 23(14), 13601366.

doi:10.1016/j.cub.2013.06.002

Feldman, J. L., & Priess, J. R. (2012). A role for the centrosome and PAR-3 in the hand-off of MTOC function during epithelial polarization. Curr Biol, 22(7), 575-582. doi:10.1016/j.cub.2012.02.044

Fong, K. W., Choi, Y. K., Rattner, J. B., & Qi, R. Z. (2008). CDK5RAP2 is a pericentriolar protein that functions in centrosomal attachment of the gamma-tubulin ring complex. Mol Biol Cell, 19(1), 115-125. doi:10.1091/mbc.e07-04-0371

Gavilan, M. P., Gandolfo, P., Balestra, F. R., Arias, F., Bornens, M., & Rios, R. M. (2018). The dual role of the centrosome in organizing the microtubule network in interphase. EMBO Rep, 19(11). doi:10.15252/embr.201845942

Jimenez, A. J., Schaeffer, A., De Pascalis, C., Letort, G., Vianay, B., Bornens, M., . . . Thery, M. (2021). Acto-myosin network geometry defines centrosome position. Curr Biol, 31(6), 1206-1220 e1205. doi:10.1016/j.cub.2021.01.002

Jonsdottir, A. B., Dirks, R. W., Vrolijk, J., Ogmundsdottir, H. M., Tanke, H. J., Eyfjord, J. E., & Szuhai, K. (2010). Centriole movements in mammalian epithelial cells during cytokinesis. BMC Cell Biol, 11, 34. doi:10.1186/1471-2121-11-34

Krishnan, N., Swoger, M., Rathbun, L. I., Fioramonti, P. J., Freshour, J., Bates, M., . . . Hehnly, H. (2022). Rab11 endosomes and Pericentrin coordinate centrosome movement during preabscission in vivo. Life Sci Alliance, 5(7). doi:10.26508/lsa.202201362

Liang, X., Weberling, A., Hii, C. Y., Zernicka-Goetz, M., & Buckley, C. E. (2022). E-cadherin mediates apical membrane initiation site localisation during de novo polarisation of epithelial cavities. EMBO J, 41(24), e111021. doi:10.15252/embj.2022111021

Lin, Y. N., Wu, C. T., Lin, Y. C., Hsu, W. B., Tang, C. J., Chang, C. W., & Tang, T. K. (2013). CEP120 interacts with CPAP and positively regulates centriole elongation. J Cell Biol, 202(2), 211219. doi:10.1083/jcb.201212060

Mangan, A. J., Sietsema, D. V., Li, D., Moore, J. K., Citi, S., & Prekeris, R. (2016). Cingulin and actin mediate midbody-dependent apical lumen formation during polarization of epithelial cells. Nat Commun, 7, 12426. doi:10.1038/ncomms12426

Martin, M., Veloso, A., Wu, J., Katrukha, E. A., & Akhmanova, A. (2018). Control of endothelial cell polarity and sprouting angiogenesis by non-centrosomal microtubules. Elife, 7. doi:10.7554/eLife.33864

Mazo, G., Soplop, N., Wang, W. J., Uryu, K., & Tsou, M. F. (2016). Spatial Control of Primary Ciliogenesis by Subdistal Appendages Alters Sensation-Associated Properties of Cilia. Dev Cell, 39(4), 424-437. doi:10.1016/j.devcel.2016.10.006

Piel, M., Nordberg, J., Euteneuer, U., & Bornens, M. (2001). Centrosome-dependent exit of cytokinesis in animal cells. Science, 291(5508), 1550-1553. doi:10.1126/science.1057330

Rodriguez-Fraticelli, A. E., Auzan, M., Alonso, M. A., Bornens, M., & Martin-Belmonte, F. (2012). Cell confinement controls centrosome positioning and lumen initiation during epithelial morphogenesis. J Cell Biol, 198(6), 1011-1023. doi:10.1083/jcb.201203075

Schmoranzer, J., Fawcett, J. P., Segura, M., Tan, S., Vallee, R. B., Pawson, T., & Gundersen, G. G. (2009). Par3 and dynein associate to regulate local microtubule dynamics and centrosome orientation during migration. Curr Biol, 19(13), 1065-1074. doi:10.1016/j.cub.2009.05.065

Tanos, B. E., Yang, H. J., Soni, R., Wang, W. J., Macaluso, F. P., Asara, J. M., & Tsou, M. F. (2013). Centriole distal appendages promote membrane docking, leading to cilia initiation. Genes Dev, 27(2), 163-168. doi:10.1101/gad.207043.112

Tateishi, K., Yamazaki, Y., Nishida, T., Watanabe, S., Kunimoto, K., Ishikawa, H., & Tsukita, S. (2013). Two appendages homologous between basal bodies and centrioles are formed using distinct Odf2 domains. J Cell Biol, 203(3), 417-425. doi:10.1083/jcb.201303071

Tsai, J. J., Hsu, W. B., Liu, J. H., Chang, C. W., & Tang, T. K. (2019). CEP120 interacts with C2CD3 and Talpid3 and is required for centriole appendage assembly and ciliogenesis. Sci Rep, 9(1), 6037. doi:10.1038/s41598-019-42577-0

Tuncay, H., Brinkmann, B. F., Steinbacher, T., Schurmann, A., Gerke, V., Iden, S., & Ebnet, K. (2015). JAM-A regulates cortical dynein localization through Cdc42 to control planar spindle orientation during mitosis. Nat Commun, 6, 8128. doi:10.1038/ncomms9128

Vinopal, S., Dupraz, S., Alfadil, E., Pietralla, T., Bendre, S., Stiess, M., . . . Bradke, F. (2023). Centrosomal microtubule nucleation regulates radial migration of projection neurons independently of polarization in the developing brain. Neuron, 111(8), 1241-1263 e1216. doi:10.1016/j.neuron.2023.01.020

Zimmerman, W. C., Sillibourne, J., Rosa, J., & Doxsey, S. J. (2004). Mitosis-specific anchoring of gamma tubulin complexes by pericentrin controls spindle organization and mitotic entry. Mol Biol Cell, 15(8), 3642-3657. doi:10.1091/mbc.e03-11-0796.

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

In this manuscript, Chengjian Zhao et al. focused on the interactions between vascular, biliary, and neural networks in the liver microenvironment, addressing the critical bottleneck that the lack of high-resolution 3D visualization has hindered understanding of these interactions in liver disease.

Strengths:

This study developed a high-resolution multiplex 3D imaging method that integrates multicolor metallic compound nanoparticle (MCNP) perfusion with optimized CUBIC tissue clearing. This method enables the simultaneous 3D visualization of spatial networks of the portal vein, hepatic artery, bile ducts, and central vein in the mouse liver. The authors reported a perivascular structure termed the Periportal Lamellar Complex (PLC), which is identified along the portal vein axis. This study clarifies that the PLC comprises CD34⁺Sca-1⁺ dual-positive endothelial cells with a distinct gene expression profile, and reveals its colocalization with terminal bile duct branches and sympathetic nerve fibers under physiological conditions.<br />

Weaknesses:

This manuscript is well-written, organized, and informative. However, there are some points that need to be clarified.

(1) After MCNP-dye injection, does it remain in the blood vessels, adsorb onto the cell surface, or permeate into the cells? Does the MCNP-dye have cell selectivity?

The experimental results showed that after injection, the MCNP series nanoparticles predominantly remained within the lumens of blood vessels and bile ducts, with their tissue distribution determined by physical perfusion. No diffusion of the dye signal into the surrounding parenchymal tissue was observed, nor was there any evidence of adsorption onto the cell surface or entry into cells. The newly added Supplementary Figure S2A–H further confirmed this feature, demonstrating that the dye signals were strictly confined to the luminal space, clearly delineating the continuous course of blood vessels and the branching morphology of bile ducts. These findings strongly support the conclusion that “MCNP dyes are distributed exclusively within the luminal compartments.”

Therefore, the MCNP dyes primarily serve as intraluminal tracers within the tissue rather than as labels for specific cell types.

(2) All MCNP-dyes were injected after the mice were sacrificed, and the mice's livers were fixed with PFA. After the blood flow had ceased, how did the authors ensure that the MCNP-dyes were fully and uniformly perfused into the microcirculation of the liver?

Thank you for the reviewer’s valuable comments. Indeed, since all MCNP dyes were perfused after the mice were euthanized and blood circulation had ceased, we cannot fully ensure a homogeneous distribution of the dye within the hepatic microcirculation. The vascular labeling technique based on metallic nanoparticle dyes used in this study offers clear imaging, stable fluorescence intensity, and multiplexing advantages; however, it also has certain limitations. The main issue is that the dye distribution within the hepatic parenchyma can be affected by factors such as lobular overlap, local tissue compression, and variations in vascular pathways, resulting in regional inhomogeneity of dye perfusion. This is particularly evident in areas where multiple lobes converge or where anatomical structures are complex, leading to local dye accumulation or over-perfusion.

In our experiments, we attempted to minimize local blockage or over-perfusion by performing PBS pre-flushing and low-pressure, constant-speed perfusion. Nevertheless, localized dye accumulation or uneven distribution may still occur in lobe junctions or structurally complex regions. Such variation represents one of the methodological limitations. Overall, the dye signals in most samples remained confined to the vascular and biliary lumens, and the distribution pattern was highly reproducible.

We have addressed this issue in the Discussion section but would like to emphasize here that, although this system has clear advantages, it remains sensitive to anatomical variability in the liver—such as lobular overlap and vascular heterogeneity. At vascular junctions, local perfusion inhomogeneity or dye accumulation may occur; therefore, injection strategies and perfusion parameters should be adjusted according to liver size and vascular condition to improve reproducibility and imaging quality. It should also be noted that the results obtained using this method primarily aim to visualize the overall and fine anatomical structures of the hepatic vascular system rather than to quantitatively reflect hemodynamic processes. In the future, we plan to combine in vivo perfusion or dynamic fluid modeling to further validate the diffusion characteristics of the dyes within the hepatic microcirculation.

(3) It is advisable to present additional 3D perspective views in the article, as the current images exhibit very weak 3D effects. Furthermore, it would be better to supplement with some videos to demonstrate the 3D effects of the stained blood vessels.

Thank you for the reviewer’s valuable comments. In response to the suggestion, we have added perspective-rendered images generated from the 3D staining datasets to provide a more intuitive visualization of the spatial morphology of the hepatic vasculature. These images have been included in Figure S2A–J. In addition, we have prepared supplementary videos (available upon request) that dynamically display the three-dimensional distribution of the stained vessels, further enhancing the spatial perception and visualization of the results.

(4) In Figure 1-I, the authors used MCNP-Black to stain the central veins; however, in addition to black, there are also yellow and red stains in the image. The authors need to explain what these stains are in the legend.

Thank you for the reviewer’s constructive comment. In Figure 1I, MCNP-Black labels the central vein (black), MCNP-Yellow labels the portal vein (yellow), MCNP-Pink labels the hepatic artery (pink), and MCNP-Green labels the bile duct (green). We have revised the Figure 1 legend to include detailed descriptions of the color signals and their corresponding structures to avoid any potential confusion.

(5) There is a typo in the title of Figure 4F; it should be "stem cell".

Thank you for the reviewer’s careful correction. We have corrected the spelling error in the title of Figure 4F to “stem cell” and updated it in the revised manuscript.

(6) Nuclear staining is necessary in immunofluorescence staining, especially for Figure 5e. This will help readers distinguish whether the green color in the image corresponds to cells or dye deposits.

We thank the reviewer for the valuable suggestion. We understand that nuclear staining can help determine the origin of fluorescence signals. However, in our three-dimensional imaging system, the deep signal acquisition range after tissue clearing often causes nuclear dyes such as DAPI to generate highly dense and widespread fluorescence, especially in regions rich in vascular structures, which can obscure the fine vascular and perivascular details of interest. Therefore, this study primarily focuses on high-resolution visualization of the spatial architecture of the vascular and biliary systems. We have added an explanation regarding this point in Figures S2I–J.

Reviewer #2 (Public review):

Summary:

The present manuscript of Xu et al. reports a novel clearing and imaging method focusing on the liver. The authors simultaneously visualized the portal vein, hepatic artery, central vein, and bile duct systems by injecting metal compound nanoparticles (MCNPs) with different colors into the portal vein, heart left ventricle, inferior vena cava, and the extrahepatic bile duct, respectively. The method involves: trans-cardiac perfusion with 4% PFA, the injection of MCNPs with different colors, clearing with the modified CUBIC method, cutting 200 micrometer thick slices by vibratome, and then microscopic imaging. The authors also perform various immunostaining (DAB or TSA signal amplification methods) on the tissue slices from MCNP-perfused tissue blocks. With the application of this methodical approach, the authors report dense and very fine vascular branches along the portal vein. The authors name them as 'periportal lamellar complex (PLC)' and report that PLC fine branches are directly connected to the sinusoids. The authors also claim that these structures co-localize with terminal bile duct branches and sympathetic nerve fibers, and contain endothelial cells with a distinct gene expression profile. Finally, the authors claim that PLC-s proliferate in liver fibrosis (CCl4 model) and act as a scaffold for proliferating bile ducts in ductular reaction and for ectopic parenchymal sympathetic nerve sprouting.

Strengths:

The simultaneous visualization of different hepatic vascular compartments and their combination with immunostaining is a potentially interesting novel methodological approach.

Weaknesses:

This reviewer has several concerns about the validity of the microscopic/morphological findings as well as the transcriptomics results. In this reviewer's opinion, the introduction contains overstatements regarding the potential of the method, there are severe caveats in the method descriptions, and several parts of the Results are not fully supported by the documentation. Thus, the conclusions of the paper may be critically viewed in their present form and may need reconsideration by the authors.

We sincerely thank the reviewer for the thorough evaluation and constructive comments on our study. We fully understand and appreciate the reviewer’s concerns regarding the methodological validity and interpretation of the results. In response, we have made comprehensive revisions and additions to the manuscript as follows:

First, we have carefully revised the Introduction and Discussion sections to provide a more balanced description of the methodological potential, removing statements that might be considered overstated, and clarifying the applicable scope and limitations of our approach (see the revised Introduction and Discussion).

Second, we have substantially expanded the Methods section with detailed information on model construction, imaging parameters, data processing workflow, and technical aspects of the single-cell transcriptomic reanalysis, to enhance the transparency and reproducibility of the study.

Third, we have added additional references and explanatory notes in the Results section to better support the main conclusions (see Section 6 of the Results).

Finally, we have rechecked and validated all experimental data, and conducted a verification analysis using an independent single-cell RNA-seq dataset (Figure S6). The results confirm that the morphological observations and transcriptomic findings are consistent and reproducible across independent experiments.

We believe these revisions have greatly strengthened the reliability of our conclusions and the overall scientific rigor of the manuscript. Once again, we sincerely appreciate the reviewer’s valuable comments, which have been very helpful in improving the logic and clarity of our work.

Reviewer #3 (Public review):

Summary:

In the reviewed manuscript, researchers aimed to overcome the obstacles of high-resolution imaging of intact liver tissue. They report successful modification of the existing CUBIC protocol into Liver-CUBIC, a high-resolution multiplex 3D imaging method that integrates multicolor metallic compound nanoparticle (MCNP) perfusion with optimized liver tissue clearing, significantly reducing clearing time and enabling simultaneous 3D visualization of the portal vein, hepatic artery, bile ducts, and central vein spatial networks in the mouse liver. Using this novel platform, the researchers describe a previously unrecognized perivascular structure they termed Periportal Lamellar Complex (PLC), regularly distributed along the portal vein axis. The PLC originates from the portal vein and is characterized by a unique population of CD34⁺Sca-1⁺ dual-positive endothelial cells. Using available scRNAseq data, the authors assessed the CD34⁺Sca-1⁺ cells' expression profile, highlighting the mRNA presence of genes linked to neurodevelopment, biliary function, and hematopoietic niche potential. Different aspects of this analysis were then addressed by protein staining of selected marker proteins in the mouse liver tissue. Next, the authors addressed how the PLC and biliary system react to CCL4-induced liver fibrosis, implying PLC dynamically extends, acting as a scaffold that guides the migration and expansion of terminal bile ducts and sympathetic nerve fibers into the hepatic parenchyma upon injury.

The work clearly demonstrates the usefulness of the Liver-CUBIC technique and the improvement of both resolution and complexity of the information, gained by simultaneous visualization of multiple vascular and biliary systems of the liver at the same time. The identification of PLC and the interpretation of its function represent an intriguing set of observations that will surely attract the attention of liver biologists as well as hepatologists; however, some claims need more thorough assessment by functional experimental approaches to decipher the functional molecules and the sequence of events before establishing the PLC as the key hub governing the activity of biliary, arterial, and neuronal liver systems. Similarly, the level of detail of the methods section does not appear to be sufficient to exactly recapitulate the performed experiments, which is of concern, given that the new technique is a cornerstone of the manuscript.

Nevertheless, the work does bring a clear new insight into the liver structure and functional units and greatly improves the methodological toolbox to study it even further, and thus fully deserves the attention of readers.

Strengths:

The authors clearly demonstrate an improved technique tailored to the visualization of the liver vasulo-biliary architecture in unprecedented resolution.

This work proposes a new biological framework between the portal vein, hepatic arteries, biliary tree, and intrahepatic innervation, centered at previously underappreciated protrusions of the portal veins - the Periportal Lamellar Complexes (PLCs).

Weaknesses:

Possible overinterpretation of the CD34+Sca1+ findings was built on re-analysis of one scRNAseq dataset.

Lack of detail in the materials and methods section greatly limits the usefulness of the new technique to other researchers.

We thank the reviewer for this important comment. We agree that when conclusions are mainly based on a single dataset, overinterpretation should be avoided. In response to this concern, we have carefully re-evaluated and clearly limited the scope of our interpretation of the scRNA-seq analysis. In addition, we performed a validation analysis using an independent single-cell RNA-seq dataset (see new Figure S6), which consistently confirmed the presence and characteristic transcriptional profile of the periportal CD34⁺Sca1⁺ endothelial cell population. These supplementary analyses strengthen the robustness of our findings and address the reviewer’s concern regarding potential overinterpretation.

In the revised manuscript, we have also greatly expanded the Materials and Methods section by providing detailed information on sample preparation, imaging parameters, data processing workflow, and single-cell reanalysis procedures. These revisions substantially improve the transparency and reproducibility of our methodology, thereby enhancing the usability and reference value of this technique for other researchers.

Recommendations for the authors:

Reviewer #2 (Recommendations for the authors):

Introduction

(1) In general, the Introduction is very lengthy and repetitive. It needs extensive shortening to a maximum of 2 A4 pages.

We thank the reviewer for the valuable suggestions. We have thoroughly condensed and restructured the Introduction, removing redundant content and merging related paragraphs to make the theme more focused and the logic clearer. The revised Introduction has been shortened to within two A4 pages, emphasizing the scientific question, innovation, and technical approach of the study.

(2) Please correct this erroneous sentence:

'...the liver has evolved the most complex and densely n organized vascular network in the body, consisting primarily of the portal vein system, central vein system, hepatic artery system, biliary system, and intrahepatic autonomic nerve network [6, 7].'

We thank the reviewer for pointing out this spelling error. The revised sentence is as follows:

“…the liver has evolved the most complex and densely organized ductal-vascular network in the body, consisting primarily of the portal vein system, central vein system, hepatic artery system, biliary system, and intrahepatic autonomic nerve network [6, 7].”

(3) '...we achieved a 63.89% improvement in clearing efficiency and a 20.12% increase in tissue transparency'

Please clarify what you exactly mean by 'clearing efficiency' and 'increased tissue transparency'.

We thank the reviewer for the valuable comments and have clarified the relevant terminology in the revised manuscript.

“Clearing efficiency” refers to the improvement in the time required for the liver tissue to become completely transparent when treated with the optimized Liver-CUBIC protocol (40% urea + H₂O₂), compared with the conventional CUBIC method. In this study, the clearing time was reduced from 9 days to 3.25 days, representing a 63.89% increase in time efficiency.

“Tissue transparency” refers to the ability of the cleared tissue to transmit visible light. We quantified the optical transparency by measuring light transmittance across the 400–900 nm wavelength range using a microplate reader. The results showed that the average transmittance increased by 20.12%, indicating that Liver-CUBIC treatment markedly enhanced the optical clarity of the liver tissue.

(4) I am concerned about claiming this imaging method as real '3D imaging'. Namely, while the authors clear full lobes, they actually cut the cleared lobes into 200-micrometer-thick slices and perform further microscopy imaging on these slices. Considering that they focus on ductular structures of the liver (such as vasculature, bile duct system, and innervations), 200 micrometer allows a very limited 3D overview, particularly in comparison with the whole-mount immuno-imaging methods combined with light sheet microscopy (such as Adori 2021, Liu 2021, etc). In this context, I feel several parts of the Introduction to be an overstatement: besides of emphasizing the advantages of the technique (such as simultaneous visualization of different hepatic vascular compartments and the bile duct system by MCNPs, the combination with immunostainings), the authors must honestly discuss the limitations (such as limited tissue overview, potential dye perfusion problems - uneven distribution of the dye etc).

We appreciate the reviewer’s insightful comments. It is true that most of the imaging depth in this study was limited to approximately 200 μm, and thus it could not achieve whole-liver three-dimensional imaging comparable to light-sheet microscopy. However, the primary focus of our study was to resolve the microscopic intrahepatic architecture, particularly the spatial relationships among blood vessels, bile ducts, and nerve fibers. Through high-resolution imaging of thick tissue sections, combined with MCNP-based multichannel labeling and immunofluorescence co-staining, we were able to accurately delineate the three-dimensional distribution of these microstructures within localized regions.

In addition to thick-section imaging, we also obtained whole-lobe dye perfusion data (as shown in Figure S1F), which comprehensively depict the three-dimensional branching patterns and distribution of the vascular systems within the liver lobe. These images were acquired from intact liver lobes perfused with MCNP dyes, revealing a continuous vascular network extending from major trunks to peripheral branches, thereby demonstrating that our approach is also capable of achieving organ-level visualization.

We have added this image and a corresponding description in the revised manuscript to more comprehensively present the coverage of our imaging system, and we have incorporated this clarification into the Discussion section.

Method

(5) More information may be needed about MCNPs:

a) As reported, there are nanoparticles with different colors in brightfield microscopy, but the particles are also excitable in fluorescence microscopy. Would you please provide a summary about excitation/emission wavelengths of the different MCNPs? This is crucial to understand to what extent the method is compatible with fluorescence immunohistochemistry.

We thank the reviewer for the careful attention and professional suggestion. We fully agree that this issue is critical for evaluating the compatibility of our method with fluorescent immunohistochemistry. Different types of metal compound nanoparticles (MCNPs) have clearly distinguishable spectral properties:

- MCNP-Green and MCNP-Yellow: AF488-matched spectra, with excitation/emission wavelengths of 495/519 nm.

- MCNP-Pink: Designed for far-red spectra, with excitation/emission wavelengths of 561/640 nm.

- MCNP-Black: Non-fluorescent, appearing black under bright-field microscopy only.

The above information has been added to the Materials and Methods section.

b) Also, is there more systematic information available concerning the advantage of these particles compared to 'traditional' fluorescence dyes, such as Alexa fluor or Cy-dyes, in fluorescence microscopy and concerning their compatibility with various tissue clearing methods (e.g., with the frequently used organic-solvent-based methods)?

We thank the reviewer for the detailed question. Compared with conventional organic fluorescent dyes, MCNP offers the following advantages:

- Enhanced photostability: Its inorganic core-shell structure resists fading even after hydrogen peroxide bleaching.

- High signal stability: Fluorescence is maintained during aqueous-based clearing (e.g., CUBIC) and multiple rounds of staining without quenching.

We appreciate the reviewer’s suggestion. In our Liver-CUBIC system, MCNP nanoparticles exhibited excellent multi-channel labeling stability and fluorescence signal retention. Regarding compatibility with other clearing methods (e.g., SCAFE, SeeDB, CUBIC), since these methods have limited effectiveness for whole-liver clearing (see Figure 2 of Tainaka, et al. 2014) and cannot meet the requirements for high-resolution microstructural imaging in this study, we consider further testing of their compatibility unnecessary.

In summary, MCNP dye demonstrates superior signal stability and spectral separation compared with conventional organic fluorescent dyes in multi-channel, long-term, high-transparency three-dimensional tissue imaging.

c) When you perfuse these particles, to which structures do they bind inside the ducts (vessels, bile ducts)? Is the 48h post-fixation enough to keep them inside the tubes/bind them to the vessel walls? Is there any 'wash-out' during the complex cutting/staining procedure? E.g., in Figure 2D: the 'classical' hepatic artery in the portal triad is not visible - but the MCNP apparently penetrated to the adjacent sinusoids at the edge of the lobulus. Also, in Figure 3B, there is a significant mismatch between the MNCP-green (bile duct) signal and the CD19 (epithelium marker) immunostaining. Please discuss these.

The experimental results showed that following injection, MCNP nanoparticles primarily remained within the vascular and biliary lumens, and their tissue distribution depended on physical perfusion. No dye signal was observed to diffuse into the surrounding parenchyma, nor did the particles adhere to cell surfaces or enter cells. The newly added Supplementary Figures S2A–H further confirm this feature: the dye signal is strictly confined within the lumens, clearly delineating continuous vascular paths and biliary branching patterns, strongly supporting the conclusion that “MCNP dye is distributed only within luminal spaces.”

Thus, MCNP dye mainly serves as an intraluminal tracer rather than a label for specific cell types.

We provide the following explanations and analyses regarding MCNP distribution in the hepatic vascular and biliary systems and its post-fixation stability:

- Potential signal displacement during sectioning/immunostaining: During slicing and immunostaining, a small number of particles may be washed away due to mechanical cutting or washing steps; however, the overall three-dimensional structure retains high spatial fidelity.

- Observation in Figure 2D: MCNP was seen entering the sinusoidal spaces at the lobule periphery, but hepatic arteries were not visible, likely due to limitations in section thickness. Although arteries were not apparent in this slice, arterial distribution around the portal vein is visible in Figure 2C. It should be noted that Figures 2C, D, and E do not represent whole-liver imaging, so not all regions necessarily contain visible hepatic arteries. For easier identification, the main hepatic artery trunk is highlighted in cyan in Figure 2E.

- Incomplete biliary signal in Figure 3B: This may be because CK19 labeling only covers biliary epithelial cells, whereas MCNP-green distributes throughout the biliary lumen. In Figure 3B, the terminal MCNP-green signal exhibits irregular polygonal structures, which we interpret as the canalicular regions.

(6) Which fixative was used for 48h of postfixation (step 6) after MCNP injections?

After MCNP injection, mouse livers were post-fixed in 4% paraformaldehyde (PFA) for 48 hours. This fixation condition effectively “locks” the MCNP particles within the vascular and biliary lumens, maintaining their spatial positions, while also being compatible with subsequent sectioning and multi-channel immunostaining analyses.

The above information has been added to the Materials and Methods section

(7) What is the 'desired thickness' in step 7? In the case of immunostained tissue, a 200-micrometer slice thickness is mentioned. However, based on the Methods, it is not completely clear what the actual thickness of the tissue was that was examined ultimately in the microscopes, and whether or not the clearing preceded the cutting or vice versa.

We appreciate the reviewer’s question. The “desired thickness” referred to in step 7 of the manuscript corresponds to the thickness of tissue sections used for immunostaining and high-resolution microscopic imaging, which is typically around 200 µm. We selected 200 µm because this thickness is sufficient to observe the PLC structure in its entirety, allows efficient staining, and preserves tissue architecture well. Other researchers may choose different section thicknesses according to their experimental needs.

In this study, the processing order for immunostained tissue samples was sectioning followed by clearing, as detailed below:

Section Thickness

To ensure antibody penetration and preservation of three-dimensional structure, tissue sections were typically cut to ~200 µm. Thicker sections can be used if more complete three-dimensional structures are required, but adjustments may be needed based on antibody penetration and fluorescence detection conditions.

Clearing Sequence

After sectioning, slices were processed using the Liver-CUBIC aqueous-based clearing system.

(8) More information is needed concerning the 'deep-focus microscopy' (Keyence), the applied confocal system, and the THUNDER 'high resolution imaging system': basic technical information, resolutions, objectives (N.A., working distance), lasers/illumination, filters, etc.

In this study, all liver lobes (left, right, caudate, and quadrate lobes) were subjected to Liver-CUBIC aqueous-based clearing to ensure uniform visualization of MCNP fluorescence and immunolabeling throughout the three-dimensional imaging of the entire liver.

The above information has been added to the Materials and Methods section.

Imaging Systems and Settings

VHX-6000 Extended Depth-of-Field Microscope: Objective: VH-Z100R, 100×–1000×; resolution: 1 µm (typical); illumination: coaxial reflected; transmitted illumination on platform: ON.

Zeiss Confocal Microscope (980): Objectives: 20× or 40×; image size: 1024 × 1024. Fluorescence detection was set up in three channels:

- Channel 1: 639 nm laser, excitation 650 nm, emission 673 nm, detection range 673–758 nm, corresponding to Cy5-T1 (red).

- Channel 2: 561 nm laser, excitation 548 nm, emission 561 nm, detection range 547–637 nm, corresponding to Cy3-T2 (orange).

- Channel 3: 488 nm laser, excitation 493 nm, emission 517 nm, detection range 490–529 nm, corresponding to AF488-T3 (green).

Leica THUNDER Imager 3D Tissue: Fluorescence detection in two channels:

- Channel 1: FITC channel (excitation 488 nm, emission ~520 nm).

- Channel 2: Orange-red channel (excitation/emission 561/640 nm).<br /> Equipped with matching filter sets to ensure signal separation.

The above information has been added to the Materials and Methods section.

(9) Liver-CUBIC, step 2: which lobe(s) did you clear (...whole liver lobes...).

In this study, all liver lobes (left, right, caudate, and quadrate lobes) were subjected to Liver-CUBIC aqueous-based clearing to ensure uniform visualization of MCNP fluorescence and immunolabeling throughout the three-dimensional imaging of the entire liver.

The above information has been added to the Materials and Methods section.

(10) For the DAB and TSA IHC stainings, did you use free-floating slices, or did you mount the vibratome sections and do the staining on mounted sections?

In this study, fixed livers were first sectioned into thick slices (~200 µm) using a vibratome. Subsequently, DAB and TSA immunohistochemical (IHC) staining were performed on free-floating sections. During the entire staining process, the slices were kept floating in the solutions, ensuring thorough antibody penetration in the thick sections while preserving the three-dimensional tissue architecture, thereby facilitating multiple rounds of staining and three-dimensional imaging.

(11) Regarding the 'transmission quantification': this was measured on 1 mm thick slices. While it is interesting to make a comparison between different clearing methods in general, one must note that it is relatively easy to clear 1mm thick tissue slices with almost any kind of clearing technique and in any tissues. The 'real' differences come with thicker blocks, such as >5mm in the thinnest dimension. Do you have such experiences (e.g., comparison in whole 'left lateral liver lobes')?

In this study, we performed three-dimensional visualization of entire liver lobes to depict the distribution of MCNPs and the overall spatial architecture of the vascular and biliary systems (Figure S1F). However, due to the limitations of the plate reader and fluorescence imaging systems in terms of spatial resolution and light penetration depth, quantitative analyses were conducted only on tissue sections approximately 1 mm thick.

Regarding the comparative quantification of different clearing methods, as the reviewer noted, nearly all aqueous- or organic solvent–based clearing techniques can achieve relatively uniform transparency in 1 mm-thick tissue sections, so differences at this thickness are limited. We have not yet conducted systematic comparisons on whole-lobe sections thicker than 5 mm and therefore cannot provide “true” difference data for thicker tissues.

(12) There is no method description for the ELMI studies in the Methods.

Transmission Electron Microscopy (TEM) Analysis of MCNPs

Before imaging, the MCNP dye solution was centrifuged at 14,000 × g for 10 minutes at 4 °C to remove aggregates and impurities. The supernatant was collected, diluted 50-fold, and 3–4 μL of the sample was applied onto freshly glow-discharged Quantifoil R1.2/1.3 copper grids (Electron Microscopy Sciences, 300 mesh). The sample was allowed to sit for 30 seconds to enable particle adsorption, after which excess liquid was gently wicked away with filter paper and the grid was air-dried at room temperature. The sample was then negatively stained with 1% uranyl acetate for 30 seconds and air-dried again before imaging.

Negative-stain TEM images were acquired using a JEOL JEM-1400 transmission electron microscope operating at 120 kV and equipped with a CCD camera. Data acquisition followed standard imaging conditions.

The above information has been added to the Materials and Methods section.

(13) Please, provide a method description for the applied CCl4 cirrhosis model. This is completely missing.

(1) Under a fume hood, carbon tetrachloride (CCl₄) was dissolved in corn oil at a 1:3 volume ratio to prepare a working solution, which was filtered through a 0.2 μm filter into a 30 mL glass vial. In our laboratory, to mimic chronic injury, mice in the experimental group were intraperitoneally injected at a dose of 1 mL/kg body weight per administration.

(2) Mice were carefully removed from the cage and placed on a scale to record body weight for calculation of the injection volume.

(3) The needle cap was carefully removed, and the required volume of the pre-prepared CCl₄ solution was drawn into the syringe. The syringe was gently flicked to remove any air bubbles.

(4) Mice were placed on a textured surface (e.g., wire cage) and restrained. When the mouse was properly positioned, ideally with the head lowered about 30°, the left lower or right lower abdominal quadrant was identified.

(5) Holding the syringe at a 45° angle, with the bevel facing up, the needle was inserted approximately 4–5 mm into the abdominal wall, and the calculated volume of CCl₄ was injected.

(6) Mice were returned to their cage and observed for any signs of discomfort.

(7) Needles and syringes were disposed of in a sharps container without recapping. A new syringe or needle was used for each mouse.

(8) To establish a progressive liver fibrosis model, injections were administered twice per week (e.g., Monday and Thursday) for 3 or 6 consecutive weeks (n=3 per group). Control mice were injected with an equal volume of corn oil for 3 or 6 weeks (n=3 per group).

(9) Forty-eight hours after the last injection, mice were euthanized by cervical dislocation, and livers were rapidly harvested. Portions of the liver were processed for paraffin embedding and histological sectioning, while the remaining tissue was either immediately frozen or used for subsequent molecular biology analyses.

The above information has been added to the Materials and Methods section.

(14) Please provide a method description for the quantifications reported in Figures 5D, 5F, and 6E.

ImageJ software was used to analyze 3D stained images (Figs. 5F, 6E), and the ultra-depth-of-field 3D analysis module was used to analyze 3D DAB images (Fig. 5D). The specific steps are as follows:

Figure 5D: DAB-stained 3D images from the control group and the CCl₄ 6-week (CCl₄-6W) group were analyzed. For each group, 20 terminal bile duct branch nodes were randomly selected, and the actual path distance along the branch to the nearest portal vein surface was measured. All measurements were plotted as scatter plots to reflect the spatial extension of bile ducts relative to the portal vein under different conditions.

Figure 5F: TSA 3D multiplex-stained images from the control group, CCl₄ 3-week (CCl₄-3W), and CCl₄ 6-week (CCl₄-6W) groups were analyzed. For each group, 5 terminal bile duct branch nodes were randomly selected, and the actual path distance along the branch to the nearest portal vein surface was measured. Measurements were plotted as scatter plots to illustrate bile duct spatial extension.

Figure 6E: TSA 3D multiplex-stained images from the control, CCl₄-3W, and CCl₄-6W groups were analyzed. For each group, 5 terminal nerve branch nodes were randomly selected, and the actual path distance along the branch to the nearest portal vein surface was measured. Scatter plots were generated to depict the spatial distribution of nerves under different treatment conditions.

(15) Please provide a method description for the human liver samples you used in Figure S6. Patient data, fixation, etc...

The human liver tissue samples shown in Figure S6 were obtained from adjacent non-tumor liver tissues resected during surgical operations at West China Hospital, Sichuan University. All samples used were anonymized archived tissues, which were applied for scientific research in accordance with institutional ethical guidelines and did not involve any identifiable patient information. After being fixed in 10% neutral formalin for 24 hours, the tissues were routinely processed for paraffin embedding (FFPE), and sectioned into 4 μm-thick slices for immunostaining and fluorescence imaging.

Results

(16) While it is stated in the Methods that certain color MCNPs were used for labelling different structures (i.e., yellow: hepatic artery; green: bile duct; portal vein: pink; central veins: black), in some figures, apparently different color MCNPs are used for the respective structures. E.g., in Figure 1J, the artery is pink and the portal vein is green. Please clarify this.

The color assignment of MCNP dyes is not fixed across different experiments or schematic illustrations. MCNP dyes of different colors are fundamentally identical in their physical and chemical properties and do not exhibit specific binding or affinity for particular vascular structures. We select different colors based on experimental design and imaging presentation needs to facilitate distinction and visualization, thereby enhancing recognition in 3D reconstruction and image display. Therefore, the color labeling in Figure 1F is primarily intended to illustrate the distribution of different vascular systems, rather than indicating a fixed correspondence to a specific dye or injection color.

(17) In Figure 1J, the hepatic artery is extremely shrunk, while the portal vein is extremely dilated - compared to the physiological situation. Does it relate to the perfusion conditions?

We appreciate the reviewer’s attention. In fact, under normal physiological conditions, the hepatic arteries labeled by CD31 are naturally narrow. Therefore, the relatively thin hepatic arteries and thicker portal veins shown in Figure 1J are normal and unrelated to the perfusion conditions. See figure 1E of Adori et al., 2021.

(18) Re: MCNP-black labelled 'oval fenestrae': the Results state 50-100 nm, while they are apparently 5-10-micron diameter in Figure 1I. Accordingly, the comparison with the ELMI studies in the subsequent paragraph is inappropriate.

We thank the reviewer for the correction. The previous statement was a typographical error. In fact, the diameter of the “elliptical windows” marked by MCNP-black is 5–10 μm, so the diameter of 5–10 μm shown in Figure 1I is correct.

(19) Please, correct this erroneous sentence: 'Pink marked the hepatic arterial system by injection extrahepatic duct (Figure 2B).'

Original sentence: “The hepatic arterial system was labeled in pink by injection through the extrahepatic duct (Figure 2B).”

Revised sentence: “The hepatic arterial system was labeled in pink by injection through the left ventricle (Figure 2B).”

(20) How do you define the 'primary portal vein tract'?

We thank the reviewer for the question. The term “primary portal vein tract” refers to the first-order branches of the portal vein that enter the liver from the hepatic hilum. These are the major branches arising directly from the main portal vein trunk and are responsible for supplying blood to the respective hepatic lobes. This definition corresponds to the concept of the first-order portal vein in hepatic anatomy.

(21) I am concerned that the 'periportal lamellar complex (PLC)' that the Authors describe really exists as a distinct anatomical or functional unit. I also see these in 3D scans - in my opinion, these are fine, lower-order portal vein branches that connect the portal veins to the adjacent sinusoid. The strong MCNP-labelling of these structures may be caused by the 'sticking' of the perfused MCNP solutions in these 'pockets' during the perfusion process. What do these structures look like with SMA or CD31 immunostaining? Also, one may consider that the anatomical evaluation of these structures may have limitations in tissue slices. Have you ever checked MCNP-perfused, cleared full live lobes in light sheet microscope scans? I think this would be very useful to have a comprehensive morphological overview. Unfortunately, based on the presented documentation, I am also not convinced that PLCs are 'co-localize' with fine terminal bile duct branches (Figure 3E, S3C), or with TH+ 'neuronal bead chain networks' (Fig 6C). More detailed and more convincing documentation is needed here.

We thank the reviewer for the detailed comments. Regarding the existence and function of the periportal lamellar complex (PLC), our observations are based on MCNP-Pink labeling of the portal vein, through which we were able to identify the PLC structure surrounding the portal branches. It should be noted that the PLC represents a very small anatomical structure. Although we have not yet performed light-sheet microscopy scanning, we anticipate that such imaging would primarily visualize larger portal vein branches. Nevertheless, this does not affect our overall conclusions.

We also appreciate the reviewer’s suggestion that the observed structures might result from MCNP adherence during perfusion. To verify the structural characteristics of the PLC, we performed immunostaining for SMA and CD31, which revealed a specific arrangement pattern of smooth muscle and endothelial markers rather than simple perfusion-induced deposition (Figures 4F and S6B).

Regarding the apparent colocalization of the PLC with terminal bile duct branches (Figures 3E and S3C) and TH⁺ neuronal bead-like networks (Figure 6C), we acknowledge that current literature evidence remains limited. Therefore, we have carefully described these observations as possible spatial associations rather than definitive conclusions. Future studies integrating high-resolution three-dimensional imaging with functional analyses will help to further clarify the anatomical and physiological significance of the PLC.

(22) 'Extended depth-of-field three-dimensional bright-field imaging revealed a strict 1:1 anatomical association between the primary portal vein trunk (diameter 280 {plus minus} 32 μm) and the first-order bile duct (diameter 69 {plus minus} 8 μm) (Figures 3A and S3A)'.

How do you define '1:1 anatomical association'? How do you define and identify the 'order' (primary, secondary) of vessel and bile duct branches in 200-micrometer slices?

We thank the reviewer for the question. In this study, the term “1:1 anatomical correlation” refers to the stable paired spatial relationship between the main portal vein trunk and its corresponding primary bile duct within the same portal territory. In other words, each main portal vein branch is accompanied by a primary bile duct of matching branching order and trajectory, together forming a “vascular–biliary bundle.”

The definitions of “primary” and “secondary” branches were based on extended-depth 3D bright-field reconstructions, considering both branching hierarchy and vessel/duct diameters: primary branches arise directly from the main trunk at the hepatic hilum and exhibit the largest diameters (averaging 280 ± 32 μm for the portal vein and 69 ± 8 μm for the bile duct), whereas secondary branches extend from the primary branches toward the lobular interior with smaller calibers.

(23) In my opinion, the applied methodical approach in the single cell transcriptomics part (data mining in the existing liver single cell database and performing Venn diagram intersection analysis in hepatic endothelial subpopulations) is largely inappropriate and thus, all the statements here are purely speculative. In my opinion, to identify the molecular characteristics of such small and spatially highly organized structures like those fine radial portal branches, the only way is to perform high-resolution spatial transcriptomic.

We thank the reviewer for the comment. We fully acknowledge the importance of high-resolution spatial transcriptomics in identifying the fine structural characteristics of portal vein branches. Due to current funding and technical limitations, we were unable to perform such high-resolution spatial transcriptomic analyses. However, we validated the molecular features of the PLC using another publicly available liver single-cell RNA-sequencing dataset, which provided preliminary supporting evidence (Figures S6B and S6C). In the manuscript, we have carefully stated that this analysis is exploratory in nature and have avoided overinterpretation. In future studies, high-resolution spatial omics approaches will be invaluable for more precisely delineating the molecular characteristics of these fine structures.

(24) 'How the autonomic nervous system regulates liver function in mice despite the apparent absence of substantive nerve fiber invasion into the parenchyma remains unclear.'

Please consider the role of gap junctions between hepatocytes (e.g., Miyashita, 1991; Seseke, 1992).

In this study, we analyzed the spatial distribution of hepatic nerves in mice using immunofluorescence staining and found that nerve fibers were almost exclusively confined to the portal vein region (Figure S6A). Notably, this distribution pattern differs markedly from that in humans. Previous studies have shown that, in human livers, nerves are not only located around the portal veins but also present along the central veins, interlobular septa, and within the parenchymal connective tissue (Miller et al., 2021; Yi, la Fleur, Fliers & Kalsbeek, 2010).

Further research has provided a physiological explanation for this interspecies difference: even among species with distinct sympathetic innervation patterns in the parenchyma—i.e., with or without direct sympathetic input—the sympathetic efferent regulatory functions may remain comparable (Beckh, Fuchs, Ballé & Jungermann, 1990). This is because signals released from aminergic and peptidergic nerve terminals can be transmitted to hepatocytes through gap junctions as electrical signals (Hertzberg & Gilula, 1979; Jensen, Alpini & Glaser, 2013; Seseke, Gardemann & Jungermann, 1992; Taher, Farr & Adeli, 2017).

However, the scarcity of nerve fibers within the mouse hepatic parenchyma suggests that the mechanisms by which the autonomic nervous system regulates liver function in mice may differ from those in humans. This observation prompted us to further investigate the potential role of PLC endothelial cells in this process.

(25) Please, correct typos throughout the text.

We thank the reviewer for this comment. We have carefully proofread the entire manuscript and corrected all typographical errors and minor language issues throughout the text.

Reviewer #3 (Recommendations for the authors):

(1) A strong recommendation - the authors ought to challenge their scRNAsq- re-analysis with another scRNAseq dataset, namely a recently published atlas of adult liver endothelial, but also mesenchymal, immune, and parenchymal cell populations https://pubmed.ncbi.nlm.nih.gov/40954217/, performed with Smart-seq2 approach, which is perfectly suitable as it brings higher resolution data, and extensive cluster identity validation with stainings. Pietilä et al. indicate a clear distinction of portal vein endothelial cells into two populations that express Adgrg6, Jag1 (e2c), from Vegfc double-positive populations (e5c and e2c). Moreover, the dataset also includes the arterial endothelial cells that were shown to be part of the PLC, but were not followed up with the scRNAseq analysis. This distinction could help the authors to further validate their results, better controlling for cross-contaminations that may occur during scRNAseq preparation.

We thank the reviewer for the valuable suggestion. As noted, we have further validated the molecular characteristics of the PLC using a recently published atlas of adult liver endothelial cells (Pietilä et al., 2023, PMID: 40954217). This dataset, generated using the Smart-seq2 technique, provides high-resolution transcriptomic profiles. By analyzing this dataset, we identified a CD34⁺LY6A⁺ portal vein endothelial cell population within the e2 cluster, which is localized around the portal vein. We then examined pathways and gene expression patterns related to hematopoiesis, bile duct formation, and neural signaling within these cells. The results revealed gene enrichment patterns consistent with those observed in our primary dataset, further supporting the robustness of our analysis of the PLC’s molecular characteristics.

(2) Improving the methods section is highly recommended, this includes more detailed information for material and protocols used - catalog numbers; protocol details of the usage - rocking platforms, timing, and tubes used for incubations; GitHub or similar page with code used for the scRNA seq re-analysis.

We thank the reviewer for the valuable suggestion. We have added more detailed information regarding the materials and experimental procedures in the Methods section, including catalog numbers, incubation conditions (such as the type of shaker, incubation time, and tube specifications), and other relevant parameters.

(3) In Figure 2A, the authors claim the size of the nanoparticle is 100nm, while based on the image, the size is ~150-180nm. A more thorough quantification of the particle size would help users estimate the usability of their method for further applications.

We thank the reviewer for the comment. In the TEM image shown in Figure 2A, the nanoparticles indeed appear to be approximately 150–200 nm in size. We have re-verified the particle dimensions and will update the corresponding description in the Methods section to allow readers to more accurately assess the applicability of this approach.

(4) In Figure 3E, it is not clear what is labeled by the pink signal. Please consider labeling the structures in the figure.

We thank the reviewer for the valuable comment. The pink signal in Figure 3E was originally intended to label the hepatic artery. However, a slight spatial misalignment occurred during the labeling process, making its position appear closer to the central vein rather than the portal vein in the image. To avoid misunderstanding, we will add clear annotations to the image and clarify this deviation in the figure legend in the revised version. It should also be noted that this figure primarily aims to illustrate the spatial relationship between the bile duct and the portal vein, and this minor deviation does not affect the reliability of our experimental conclusions.

(5) The following statement is not backed by quantification as it ought to be „Dual-channel three-dimensional confocal imaging combined with CK19 immunostaining revealed that the sites of dye leakage did not coincide with the CK19-positive terminal bile duct epithelium, but instead were predominantly localized within regions adjacent to the PLC structures".

We thank the reviewer for the valuable comment. We have added the corresponding quantitative analysis to support this conclusion. Quantitative assessment of the extended-depth imaging data revealed that dye leakage predominantly occurred in regions adjacent to the PLC structure, rather than in the perivenous sinusoidal areas. The corresponding results have been presented in the revised Figure 3G.

(6) Similarly, Figure 4F is central to the Sca1CD34 cell type identification but lacks any quantification, providing it would strengthen the key statement of the article. A possible way to approach this is also by FACS sorting the double-positive cells and bluk/qRT validation.

We thank the reviewer for raising this point. We agree that quantitative validation of the Sca1⁺CD34⁺ population by FACS sorting could further support our conclusions. However, the primary focus of this study is on the spatial localization and transcriptional features of PLC endothelial cells. The identification of the Sca1⁺CD34⁺ subset is robustly supported by multiple complementary approaches, including three-dimensional imaging, co-staining with pan-endothelial markers, and projection mapping analyses. Collectively, these lines of evidence provide a solid basis for characterizing this unique endothelial population.

(7) The images in Figure S4D are not comparable, as the Sca1-stained image shows a longitudinal section of the PV, but the other stainings are cross-sections of PVs.

We thank the reviewer for the careful comment. We agree that the original Sca1-stained image, being a longitudinal section of the portal vein, was not optimal for direct comparison with other cross-sectional images. We have replaced it with a cross-sectional image of the portal vein to ensure comparability across all images. The updated image has been included in the revised Supplementary Figure S4D.

(8) I might be wrong, but Figure 4J is entirely missing, and only a cartoon is provided. Either remove the results part or provide the data.

We appreciate the reviewer’s careful observation. Figure 4J was intentionally designed as a schematic illustration to summarize the structural relationships and spatial organization of the portal vein, hepatic artery, and PLC identified in the previous panels (Figures 4A–4I). It does not represent newly acquired experimental data, but rather serves to provide a conceptual overview of the findings.

To avoid misunderstanding, we have clarified this point in the figure legend and the main text, stating that Figure 4J is a schematic summary rather than an experimental image. Therefore, we respectfully prefer to retain the schematic figure to aid readers’ interpretation of the preceding results.

(9) The methods section lacks information about the CCL4concentration, and it is thus hard to estimate the dosage of CCL4 received (ml/kg). This is important for the interpretation of the severity of the fibrosis and presence of cirrhosis, as different doses may or may not lead to cirrhosis within the short regimen performed by the authors [PMID: 16015684 DOI: 10.3748/wjg.v11.i27.4167]. Validation of the fibrosis/cirrhosis severity is, in this case, crucial for the correct interpretation of the results. If the level of cirrhosis is not confirmed, only progressive fibrosis should be mentioned in the manuscript, as these two terms cannot be used interchangeably.

Thank you for the reviewer’s comment. We indeed omitted the information on the concentration of carbon tetrachloride (CCl₄) in the Methods section. In our experiments, mice received intraperitoneal injections of CCl₄ at a dose of 1 mL/kg body weight, twice per week, for a total of six weeks. We have revised the manuscript accordingly, using the term “progressive fibrosis” to avoid confusion between fibrosis and cirrhosis.

(10) The following statement is not backed by any correlation analysis: "Particularly during liver fibrosis progression, the PLC exhibits dynamic structural extension correlating with fibrosis severity,.. ".

We thank the reviewer for the comment. The original statement that the “PLC correlates with fibrosis severity” lacked support from quantitative analysis. To ensure a precise description, we have revised the sentence as follows: “During liver fibrosis progression, the PLC exhibits dynamic structural extension.”

(11) Similarly, the following statement is not followed by data that would address the impact of innervation on liver function: "How the autonomic nervous system regulates liver function in mice despite the apparent absence of substantive nerve fiber invasion into the parenchyma remains unclear.".

This section has been revised. In this study, we analyzed the spatial distribution of nerves in the mouse liver using immunofluorescence staining. The results showed that nerve fibers were almost entirely confined to the portal vein region (Figure S6A). Notably, this distribution pattern differs significantly from that in humans. Previous studies have demonstrated that in the human liver, nerves are not only distributed around the portal vein but also present in the central vein, interlobular septa, and connective tissue of the hepatic parenchyma (Miller et al., 2021; Yi, la Fleur, Fliers & Kalsbeek, 2010).

Previous studies have further explained the physiological basis for this difference: even among species with differences in parenchymal sympathetic innervation (i.e., species with or without direct sympathetic input), their sympathetic efferent regulatory functions may still be similar (Beckh, Fuchs, Ballé & Jungermann, 1990). This is because signals released by adrenergic and peptidergic nerve terminals can be transmitted to hepatocytes as electrical signals through intercellular gap junctions (Hertzberg & Gilula, 1979; Jensen, Alpini & Glaser, 2013; Seseke, Gardemann & Jungermann, 1992; Taher, Farr & Adeli, 2017). However, the scarcity of nerve fibers in the mouse hepatic parenchyma suggests that the mechanism by which the autonomic nervous system regulates liver function in mice may differ from that in humans. This finding also prompts us to further explore the potential role of PLC endothelial cells in this process.

(12) Could the authors discuss their interpretation of the results in light of the fact that the innervation is lower in cirrhotic patients? https://pmc.ncbi.nlm.nih.gov/articles/PMC2871629/. Also, while ADGRG6 (Gpr126) may play important roles in liver Schwann cells, it is likely not through affecting myelination of the nerves, as the liver nerves are not myelinated https://pubmed.ncbi.nlm.nih.gov/2407769/ and https://www.pnas.org/doi/10.1073/pnas.93.23.13280.

We have revised the text to state that although most hepatic nerves are unmyelinated, GPR126 (ADGRG6) may regulate hepatic nerve distribution via non-myelination-dependent mechanisms. Studies have shown that GPR126 exerts both Schwann cell–dependent and –independent functions during peripheral nerve repair, influencing axon guidance, mechanosensation, and ECM remodeling (Mogha et al., 2016; Monk et al., 2011; Paavola et al., 2014).

(13) The manuscript would benefit from text curation that would:

a) Unify the language describing the PLC, so it is clear that (if) it represents protrusions of the portal veins.

We have standardized the description of the PLC throughout the manuscript, clearly specifying its anatomical relationship with the portal vein. Wherever appropriate, we indicate that the PLC represents protrusions associated with the portal vein, avoiding ambiguous or inconsistent statements.

b) Increase the accuracy of the statements.

Examples: "bile ducts, and the central vein in adult mouse livers."

We have refined all statements for accuracy.

c) Reduce the space given to discussion and results in the introduction, moving them to the respective parts. The same applies to the results section, where discussion occurs at more places than in the Discussion part itself.

We have edited the Introduction, removing detailed results and functional explanations, and retaining only a concise overview.

Examples: "The formation of PLC structures in the adventitial layer may participate in local blood flow regulation, maintenance of microenvironmental homeostasis, and vascular-stem cell interactions."

"This finding suggests that PLC endothelial cells not only regulate the periportal microcirculatory blood flow, but also establish a specialized microenvironment that supports periportal hematopoietic regulation, contributing to stem cell recruitment, vascular homeostasis, and tissue repair. "

"Together, these findings suggest the PLC endothelium may act as a key regulator of bile duct branching and fibrotic microenvironment remodeling in liver cirrhosis. " This one in particular would require further validation with protein stainings and similar, directly in your model.

d) Provide a clear reference for the used scRNA seq so it's clear that the data were re-analyzed.

Example: "single-cell transcriptomic analysis revealed significant upregulation of bile duct-related genes in the CD34⁺Sca-1⁺ endothelium of PLC in cirrhotic liver, with notably high expression of Lgals1 (Galectin-1) and HGF(Figure 5G) "

When describing the transcriptional analysis of PLC endothelial cells, we explicitly cited the original scRNA-seq dataset (Su et al., 2021), clarifying that these data were reanalyzed rather than newly generated.

e) Introducing references for claims that, in places, are crucial for further interpretation of experiments.

Examples: "It not only guides bile duct branching during development but also"; the authors show no data from liver development.

Thank you for pointing this out. We have revised the relevant statement to ensure that the claim is accurate and well-supported.

f) Results sentence "Instead, bile duct epithelial cells at the terminal ducts extended partially along the canalicular network without directly participating in the formation of the bile duct lumen." Lacks a callout to the respective Figure.

We would like to thank the reviewers for pointing out this issue. In the revised manuscript, the relevant image (Figure 3D) has been clearly annotated with white arrows to indicate the phenomenon of terminal cholangiocytes extending along the bile canaliculi network. Additionally, the schematic diagram on the right side clearly shows the bile canaliculi, cholangiocytes, and bile flow direction using arrows and color coding, thus intuitively corresponding to the textual description.

(14) Formal text suggestions: The manuscript text contains a lot of missed or excessive spaces and several typos that ought to be fixed. A few examples follow:

a) "densely n organized vascular network "

b) "analysis, while offering high spatial "

c) "specific differences, In the human liver, "

d) Figure 4F has a typo in the description.

e) "generation of high signal-to-noise ratio, multi-target " SNR abbreviation was introduced earlier.

f) Canals of Hering, CoH abbreviation comes much later than the first mention of the Canals of Hering.

We thank the reviewer for the helpful comment regarding textual consistency. We have carefully reviewed and revised the entire manuscript to improve the accuracy, clarity, and consistency of the text.

Reviewer #1 (Public review):

Domínguez-Rodrigo and colleagues make a largely convincing case for habitual elephant butchery by Early Pleistocene hominins at Olduvai Gorge (Tanzania), ca. 1.8-1.7 million years ago. They present this at a site scale (the EAK locality, which they excavated), as well as across the penecontemporaneous landscape, analyzing a series of findspots that contain stone tools and large-mammal bones. The latter are primarily elephants, but giraffids and bovids were also butchered in a few localities.

The authors claim that this is the earliest well-documented evidence for elephant butchery; doing so requires debunking other purported cases of elephant butchery in the literature, or in one case, reinterpreting elephant bone manipulation as being nutritional (fracturing to obtain marrow) rather than technological (to make bone tools). The authors' critical discussion of these cases may not be consensual, but it surely advances the scientific discourse. The authors conclude by suggesting that an evolutionary threshold was achieved at ca. 1.8 ma, whereby regular elephant consumption rich in fats and perhaps food surplus, more advanced extractive technology (the Acheulian toolkit), and larger human group size had coincided.

The fieldwork and spatial statistics methods are presented in detail and are solid and helpful, especially the excellent description (all too rare in zooarchaeology papers) of bone conservation and preservation procedures. The results are detailed and clearly presented.

The authors achieved their aims, showcasing recurring elephant butchery in 1.8-1.7 million-year-old archaeological contexts. The authors cautiously emphasize the temporal and spatial correlation of 1) elephant butchery, 2) Acheulian toolkits, and 3) larger sites, and discuss how these elements may be causally related.

Overall, this is an interesting manuscript of broad interest that presents original data and interpretations from the Early Pleistocene archaeology of Olduvai Gorge. These observations and the authors' critical review of previously published evidence are an important contribution that will form the basis for building models of Early Pleistocene hominin adaptation.

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

Domínguez-Rodrigo and colleagues make a moderately convincing case for habitual elephant butchery by Early Pleistocene hominins at Olduvai Gorge (Tanzania), ca. 1.8-1.7 million years ago. They present this at the site scale (the EAK locality, which they excavated), as well as across the penecontemporaneous landscape, analyzing a series of findspots that contain stone tools and large-mammal bones. The latter are primarily elephants, but giraffids and bovids were also butchered in a few localities. The authors claim that this is the earliest well-documented evidence for elephant butchery; doing so requires debunking other purported cases of elephant butchery in the literature, or in one case, reinterpreting elephant bone manipulation as being nutritional (fracturing to obtain marrow) rather than technological (to make bone tools). The authors' critical discussion of these cases may not be consensual, but it surely advances the scientific discourse. The authors conclude by suggesting that an evolutionary threshold was achieved at ca. 1.8 ma, whereby regular elephant consumption rich in fats and perhaps food surplus, more advanced extractive technology (the Acheulian toolkit), and larger human group size had coincided.

The fieldwork and spatial statistics methods are presented in detail and are solid and helpful, especially the excellent description (all too rare in zooarchaeology papers) of bone conservation and preservation procedures. However, the methods of the zooarchaeological and taphonomic analysis - the core of the study - are peculiarly missing. Some of these are explained along the manuscript, but not in a standard Methods paragraph with suitable references and an explicit account of how the authors recorded bone-surface modifications and the mode of bone fragmentation. This seems more of a technical omission that can be easily fixed than a true shortcoming of the study. The results are detailed and clearly presented.

By and large, the authors achieved their aims, showcasing recurring elephant butchery in 1.8-1.7 million-year-old archaeological contexts. Nevertheless, some ambiguity surrounds the evolutionary significance part. The authors emphasize the temporal and spatial correlation of (1) elephant butchery, (2) Acheulian toolkits, and (3) larger sites, but do not actually discuss how these elements may be causally related. Is it not possible that larger group size or the adoption of Acheulian technology have nothing to do with megafaunal exploitation? Alternative hypotheses exist, and at least, the authors should try to defend the causation, not just put forward the correlation. The only exception is briefly mentioning food surplus as a "significant advantage", but how exactly, in the absence of food-preservation technologies? Moreover, in a landscape full of aggressive scavengers, such excess carcass parts may become a death trap for hominins, not an advantage. I do think that demonstrating habitual butchery bears very significant implications for human evolution, but more effort should be invested in explaining how this might have worked.

Overall, this is an interesting manuscript of broad interest that presents original data and interpretations from the Early Pleistocene archaeology of Olduvai Gorge. These observations and the authors' critical review of previously published evidence are an important contribution that will form the basis for building models of Early Pleistocene hominin adaptation.

This is a good example of the advantages of the eLife reviewing process. It has become much too common, among traditional peer-reviewing journals, to reject articles when there is no coincident agreement in the reviews, regardless of the heuristics (i.e., empirically-supported weight) of the arguments on both reviewers. Reviewers 1 and 2 provide contrasting evaluations, and the eLife dialogue between authors and reviewers enable us to address their comments differentially. Reviewer 1 (R1), whose evaluation is overall positive, remarks that the methods of the zooarchaeological and taphonomic analysis are missing. We have added them now in the revised version of our manuscript. R1 also remarks that our work highlights correlation of events, but not necessarily causation. We did not establish causation because such interpretations bear a considerable amount of speculation (and they might have fostered further criticism by R2); however, in the revised version, we expanded our discussion of these issues substantially. Establishing causation among the events described is impossible, but we certainly provide arguments to link them.

Reviewer #2 (Public review):

The authors argue that the Emiliano Aguirre Korongo (EAK) assemblage from the base of Bed II at Olduvai Gorge shows systematic exploitation of elephants by hominins about 1.78 million years ago. They describe it as the earliest clear case of proboscidean butchery at Olduvai and link it to a larger behavioral shift from the Oldowan to the Acheulean.

The paper includes detailed faunal and spatial data. The excavation and mapping methods appear to be careful, and the figures and tables effectively document the assemblage. The data presentation is strong, but the behavioral interpretation is not supported by the evidence.

The claim for butchery is based mainly on the presence of green-bone fractures and the proximity of bones and stone artifacts. These observations do not prove human activity. Fractures of this kind can form naturally when bones break while still fresh, and spatial overlap can result from post-depositional processes. The studies cited to support these points, including work by Haynes and colleagues, explain that such traces alone are not diagnostic of butchery, but this paper presents them as if they were.

The spatial analyses are technically correct, but their interpretation extends beyond what they can demonstrate. Clustering indicates proximity, not behavior. The claim that statistical results demonstrate a functional link between bones and artifacts is not justified. Other studies that use these methods combine them with direct modification evidence, which is lacking in this case.

The discussion treats different bodies of evidence unevenly. Well-documented cut-marked specimens from Nyayanga and other sites are described as uncertain, while less direct evidence at EAK is treated as decisive. This selective approach weakens the argument and creates inconsistency in how evidence is judged.

The broader evolutionary conclusions are not supported by the data. The paper presents EAK as marking the start of systematic megafaunal exploitation, but the evidence does not show this. The assemblage is described well, but the behavioral and evolutionary interpretations extend far beyond what can be demonstrated.

We disagree with the arguments provided by Reviewer 2 (R2). The arguments are based on two issues: bone breakage and spatial association. We will treat both separately here.

Bone breakage

R2 argues that:

“The claim for butchery is based mainly on the presence of green-bone fractures and the proximity of bones and stone artifacts. These observations do not prove human activity. Fractures of this kind can form naturally when bones break while still fresh, and spatial overlap can result from post-depositional processes. The studies cited to support these points, including work by Haynes and colleagues, explain that such traces alone are not diagnostic of butchery, but this paper presents them as if they were.”

In our manuscript, we argued that green-breakage provides an equally good (or even  better) taphonomic evidence of butchery if documented following clear taphonomic indicators. Not all green breaks are equal and not all “cut marks” are unambiguously identifiable as such. First, “natural” elephant long limb breaks have been documented only in pre/peri-mortem stages when an elephant breaks a leg. As a matter of fact, they have only been reported in publication on femora, the thinnest long bone (Haynes et al., 2021). Unfortunately, they have been studied many months after the death of the individuals, and the published diagnosis is made under the assumption that no other process intervened in the modification of those bones during this vast time span. Most of the breaks resulting from pre-mortem fractures produce long smooth, oblique/helical outlines. Occasionally, some flake scarring may occur on the cortical surface. This has been documented as uneven, small-sized, spaced, and we are not sure if it resulted from rubbing of broken fragments while the animal was alive and attempting to walk or some may have resulted from dessication of the bone after one year. When looking at them in detail, such breaks contain sometimes step-microfractures and angular (butterfly-like) outlines. Sometimes, they may be accompanied by pseudo-notches, which are distinct and not comparable to the deep notches that hammerstone breaking generates on the same types of bones. Commonly, the edges of the breaks show some polishing, probably from separate break planes rubbing against each other. It should be emphasized that the experimental work on hammerstone breaking documented by Haynes et al. (2021) is based on bone fracture properties of bones that are no longer completely green. The cracking documented in their hammerstone experimentation, with very irregular outlines differs from the cracking that we are documented in butchery of recently dead elephants.

All this contrasts with the overlapping notches and flake scars (mostly occurring on the medullary side of the bone), both of them bigger in size, with clear smooth, spiral and longitudinal trajectories, with a more intensive modification on the medullary surface, and with sharp break edges resulting from hammerstone breaking of the green bone. No “natural” break has been documented replicating the same morphologies displayed in the Supplementary File to our paper. We display specimens with inflection points, hackle marks on the breaks, overlapping scarring on the medullary surface, with several specimens displaying percussion marks and pitting (also most likely percussion marks). Most importantly, we document this patterned modification on elements other than femora, for which no example has been documented of purported morphological equifinality caused by pre-mortem “natural” breaking. In contrast, such morphologies are documented in hammerstone-broken completely green bones (work in progress). We cited the works of Haynes to support this, because they do not show otherwise. As a matter of fact, Haynes himself had the courtesy of making a thorough reading of our manuscript and did not encounter any contradiction with his work. 

Spatial association

R2 argues in this regard:

“The spatial analyses are technically correct, but their interpretation extends beyond what they can demonstrate. Clustering indicates proximity, not behavior. The claim that statistical results demonstrate a functional link between bones and artifacts is not justified. Other studies that use these methods combine them with direct modification evidence, which is lacking in this case.”

We should emphasize that there is some confusion in the use and interpretation of clustering by R2 when applied to EAK. R2 appears to interpret clustering as the typical naked-eye perception of the spatial association of different items. In contrast, we rely on the statistical concept of clustering, more specifically on spatial interdependence or covariance, which is different. Items may appear visually clustered but still be statistically independent. This could, for example, result from two independent depositional episodes that happen to overlap spatially. In such cases, the item-to-item relationship does not necessarily show any spatial interdependence between classes other than simple clustering (i.e., spatial coincidence in intensity).

Spatial statistical interdependence, on the other hand, reflects a spatial relationship or co-dependence between different items. This goes beyond the mere fact that classes appear clustered: items between classes may show specific spatial relationships — they may avoid each other or occupy distinct positions in space (regular co-dependence), or they may interact within the same spatial area (clustering co-dependence). Our tests indicate the latter for EAK.

Such patterns are difficult to explain when depositional events are unrelated, since the probability that two independent events would generate identical spatial patterns in the same loci is very low. They are also difficult to reconcile when post-depositional processes intervene and resediment part of the assemblage (Domínguez-Rodrigo et al. 2018).

Finally, R2 concludes:

“The discussion treats different bodies of evidence unevenly. Well-documented cut-marked specimens from Nyayanga and other sites are described as uncertain, while less direct evidence at EAK is treated as decisive. This selective approach weakens the argument and creates inconsistency in how evidence is judged.”

The Nyayanga hippo remains bearing modifications have not been well-documented cut marks. Neither R2 nor we can differentiate those marks from those inflicted by natural abrasive processes in coarse-grained sedimentary contexts, where the carcasses are found. The fact that the observable microscopic features (through low-quality photographs as appear in the original publication) differ between the cut marks documented on smaller animals and those inferred for the hippo remains makes them even more ambiguous. Nowhere in our manuscript do we treat the EAK evidence (or any other evidence) as decisive, but as the most likely given the methods used and the results reported.

References

Haynes G, Krasinski K, Wojtal P. 2021. A Study of Fractured Proboscidean Bones in Recent and Fossil Assemblages. Journal of Archaeological Method and Theory 28:956–1025.

Domínguez-Rodrigo, M., Cobo-Sánchez, L., yravedra, J., Uribelarrea, D., Arriaza, C., Organista, E., Baquedano, E. 2018. Fluvial spatial taphonomy: a new method for the study of post-depositional processes. Archaeological and Anthropological Sciences 10: 1769-1789.

Recommendations for authors:

Reviewer #1 (Recommendations for the authors):

I have several recommendations that, in my opinion, could enhance the communication of this study to the readers. The first point is the only crucial one.

(1) A detailed zooarchaeological methods section must be added, with explanations (or references to them) of precisely how the authors defined and recorded bone-surface modifications and mode of bone fragmentation.

This appears in the revised version of the manuscript in the form of a new sub-section within the Methods section.

(2) The title could be improved to better represent the contents of the paper. It contains two parts: the earliest evidence for elephant butchery (that's ok), and revealing the evolutionary impact of megafaunal exploitation. The latter point is not actually revealed in the manuscript, just alluded to here and there (see also below).

We have elaborated on this in the revised version, linking megafaunal exploitation and anatomical changes (which appear discussed in much more detail in the references indicated).

(3) The abstract does not make it clear whether the authors think that the megafaunal adaptation strongly correlates with the Acheulian technocomplex. It seems that they do, so please make this point apparent in the abstract.

From a functional point of view, we document the correlation, but do not believe in the causation, since most butchering tools around these megafaunal carcasses are typologically non Acheulian. We have indicated so in the abstract.

(4) Please define what you mean by "megafauna". How large should an animal be to be considered as megafauna in this particular context?

We have added this definition: we identify as “megafauna” those animals heavier than 800 kg.

(5) In the literature survey, consider also this Middle Pleistocene case-study of elephant butchery, including a probable bone tool: Rabinovich, R., Ackermann, O., Aladjem, E., Barkai, R., Biton, R., Milevski, I., Solodenko, N., and Marder, O., 2012. Elephants at the middle Pleistocene Acheulian open-air site of Revadim Quarry, Israel. Quaternary International, 276, pp.183-197.

Added to the revised version

(6) The paragraph in lines 123-160 is unclear. Do the authors argue that the lack of evidence for processing elephant carcasses for marrow and grease is universal? They bring forth a single example of a much later (MIS 5) site in Germany. Then, the authors state the huge importance of fats for foragers (when? Where? Surely not in all latitudes and ecosystems). This left me confused - what exactly are you trying to claim here?

We have explained this a little more in the revised text. What we pointed out was that most prehistoric (and modern) elephant butchery sites leave grease-containing long bones intact. Evidence of anthropogenic breakage of these elements is rather limited. The most probably reason is the overabundance of meat and fat from the rest of the carcass and the time-consuming effort needed to access the medullary cavity of elephant long bones.

(7) The paragraph in lines 174-187 disrupts the flow of the text, contains previously mentioned information, ends with an unclear sentence, and could be cut.

(8) Results: please provide the MNI for the EAK site (presumably 1, but this is never mentioned).

Done in the revised version.

(9) Lines 292 - 295: The authors found no traces of carnivoran activity (carnivoran remains, coprolites, or gnawing marks on the elephant bones), yet they attribute the absence of some non-dense skeletal elements to carnivore ravaging. I cannot understand this rationale, given that other density-mediated processes could have deleted the missing bones and epiphysis.

This interpretation stems from our observations of several elephant carcasses in the Okavango delta in Botswana. Those that were monitored showed deletion of remains (i.e., disappearance of certain bones, like feet) without necessarily imprinting damage on the rest of the carcass. Carnivore intervention in an elephant death site can result in deletion of a few remains without much damage (if any), or if hyena clans access the carcass, much more conspicuous damage can be documented. There is a whole range of carnivore signatures in between. We are currently working on our study of several elephant carcasses subjected to these highly variable degrees of carnivore impact.

(10) Lines 412 - 422: "The clustering of the elephant (and hippopotamus) carcasses in the areas containing the highest densities of landscape surface artifacts is suggestive of a hominin agency in at least part of their consumption and modification." - how so? It could equally suggest that both hominins and elephants were drawn to the same lush environments.

We agree. Both hominins and megafauna must have been drawn to the same ecological loci for interaction to emerge. However, the fact that the highest density clusters of artifacts coincide with the highest density of carcasses “showing evidence of having been broken”, is suggestive of hominin use and consumption.

(11) Discussion: I suggest starting the Discussion with a concise appraisal of the lines of evidence detailed in the Results and their interpretation, and only then, the critical reassessment of other studies. Similarly, a new topic starts in line 508, but without any subheading or an introductory sentence that could assist the readers.

We added the introductory lines of the former Conclusion section to the revised Discussion section, as suggested by R1.

(12) Line 607: Neumark-Nord are Late Pleistocene sites (MIS 5), not Middle Pleistocene.

Corrected.

(13) Regarding the ambiguity in how megafaunal exploitation may be causally related to the other features of the early Acheulian, the authors can develop the discussion. Alternatively, they should explicitly state that correlation is not causation, and that the present study adds the megafaunal exploitation element to be considered in future discussion of the shifts in lifestyles 1.8 million years ago.

We have done so.

Reviewer #2 (Recommendations for the authors):

The following detailed comments are provided to help clarify arguments, ensure accurate representation of cited literature, and strengthen the logical and methodological framing of the paper. Line numbers refer to the version provided for review.

(1) Line 55: Such concurrency (sometimes in conjunction with other variables)

The term "other variables" is very vague. I would suggest expanding on this or taking it out altogether.

(2) Line 146: Megafaunal long bone green breakage (linked to continuous spiral fractures on thick cortical bone) is probably a less ambiguous trace of butchery than "cut marks", since many of the latter could be equifinal and harder to identify, especially in contexts of high abrasion and trampling (Haynes et al., 2021, 2020).

This reasoning is not supported by the evidence or the cited sources. Green-bone spiral fractures only show that a bone broke while it was fresh and do not reveal who or what caused it. Carnivore feeding, trampling, and natural sediment pressure can all create the same patterns, so these fractures are not clearer evidence of butchery than cut marks. Cut marks, when they are preserved and morphologically clear, remain the most reliable indicator of human activity. The Haynes papers actually show the opposite of what is claimed here. They warn that spiral fractures and surface marks can form naturally and that fracture patterns alone cannot be used to infer butchery. This section should be revised to reflect what those studies actually demonstrate.

The reasoning referred to in line 146 is further explained below in the original text as follows:

“Despite the occurrence of green fractures on naturally-broken bones, such as those trampled by elephants (Haynes et al., 2020), those occurring through traumatic fracturing or gnawed by carnivores (Haynes and Hutson, 2020), these fail to reproduce the elongated, extensive, or helicoidal spiral fractures (uninterrupted by stepped sections), accompanied by the overlapping conchoidal scars (both cortical and medullary), the reflected scarring, the inflection points, or the impact hackled break surfaces and flakes typical of dynamic percussive breakage. Evidence of this type of green breakage had not been documented earlier for the Early Pleistocene proboscidean or hippopotamid carcasses, beyond the documentation of flaked bone with the purpose of elaboration of bone tools (Backwell and d’Errico, 2004; Pante et al., 2020; Sano et al., 2020).”

The problem in the way that R2 uses Haynes et al.´s works is that R2 uses features separately. Natural breaks occurring while the bone is green can generate spiral smooth breaks, for example, but it is not the presence of a single feature that invalidates the diagnosis of agency or that is taphonomically relevant, but the concurrence of several of them. The best example of a naturally (pre-mortem) broken bone was published by Haynes et al.

The natural break shows helical fractures, subjugated to linear (angular) fracture outlines. Notice how the crack displays a zig-zag. The break is smooth but most damage occurs on the cortical surface, with flaking adjacent to the break and step micro-fracturing on the edges. The cortical scarring is discontinuous (almost marginal) and very small, almost limited to the very edge of the break. No modification occurs on the medullary surface. No extensive conchoidal fractures are documented, and certainly none inside the medullary surface of the break.

Compare with Figure S8, S10, S17 and S34 (all specimens are shown in their medullary surface):

In these examples, we see clearly modified medullary surfaces with multiple green breaks and large-sized step fractures, accompanied in some examples by hackle marks. Some show large overlapping scars (of substantially bigger size than those documented in the natural break image). Not a single example of naturally-broken bones has been documented displaying these morphologies simultaneously. It is the comprehensive analysis of the co-occurrence of these features and not their marginal and isolated occurrence in naturally-broken bones that make a difference in the attribution of agency. Likewise, no example of naturally-broken bone has been published that could mimic any of the two green-broken bones documented at EAK. In contrast, we do have bones from our on-going experimentation with green elephant carcasses that jointly reproduce these features. See also Figure 6 of the article to find another example without any modern referent in the naturally-broken bones documented.

We should emphasize that R2 is inaccurately portraying what Haynes et al.´s results really document. Contrary to R2´s assertion, trampling does not reproduce any of the examples shown above. Neither do carnivores. It should be stressed that Haynes & Harrod only document similar overlapping scarring on the medullary surface of bones, when using much smaller animals. In all the carnivore damage repertoire that they document for elephants, durophagous spotted hyenas can only inflict furrowing on the ends of the biggest long bones, especially if they are adults. Long bone midshafts remain inaccessible to them. The mid-shaft portions of bones that we document in our Supplementary File and at EAK cannot be the result of hyena (or carnivore damage) for this reason, and also because their intense gnawing on elephant bones leaves tooth marking on most of the elements that they modify, being absent in our sample.

(3) Line 176: other than hominins accessed them in different taphonomically-defined stages- stages - the "Stages" is repeated twice

Defined in the revised version

(4) Line 174: Regardless of the type of butchery evidence - and with the taphonomic caveat that no unambiguous evidence exists to confirm that megafaunal carcasses were hunted or scavenged other than hominins accessed them in different taphonomically-defined stages- stages - the principal reasons for exploring megafaunal consumption in early human evolution is its origin, its episodic or temporally-patterned occurrence, its impact on hominin adaptation to certain landscapes, and its reflection on hominin group size and site functionality.

This sentence is confusing and needs to be rewritten for clarity. It tries to combine too many ideas at once, and the phrasing makes it hard to tell what the main point is. The taphonomic caveat in the middle interrupts the sentence and obscures the argument. It should be broken into separate, clearer statements that distinguish what evidence exists, what remains uncertain, and what the broader goals of the discussion are.

We believe the ideas are displayed clearly

(5) Line 179: landscapes, and its reflection on hominin group size and site functionality. If hominins actively sought the exploitation of megafauna, especially if targeting early stages of carcass consumption, the recovery of an apparent surplus of resources reflects a substantially different behavior from the small-group/small-site pattern documented at several earlier Oldowan anthropogenic sites (Domínguez-Rodrigo et al., 2019) -or some modern foragers, like the Hadza, who only exploit megafaunal carcasses very sporadically, mostly upon opportunistic encounters (Marlowe, 2010; O'Connell et al., 1992; Wood, 2010; Wood and Marlowe, 2013).

This sentence makes a reasonable point, but is written in a confusing way. The idea that early, deliberate access to megafauna would represent a different behavioral pattern from smaller Oldowan or modern foraging contexts is valid, but the sentence is awkward and hard to follow. It should be rephrased to make the logic clearer and more direct.

We believe the ideas are displayed clearly

(6) Line 186: When the process started of becoming megafaunal commensal started has major implications for human evolution.

This sentence is awkward and needs to be rewritten for clarity. The phrasing "when the process started of becoming megafaunal commensal started" is confusing and grammatically incorrect. It could be revised to something like "Determining when hominins first began to interact regularly with megafauna has major implications for human evolution," or another version that clearly identifies the process being discussed.

Modified in the revised version

(7) Line189: The multiple taphonomic biases intervening in the palimpsestic nature of most of these butchery sites often prevent the detection of the causal traces linking megafaunal carcasses and hominins. Functional links have commonly been assumed through the spatial concurrence of tools and carcass remains; however, this perception may be utterly unjustified as we argued above. Functional association of both archaeological elements can more securely be detected through objective spatial statistical methods. This has been argued to be foundational for heuristic interpretations of proboscidean butchery sites (Giusti, 2021). Such an approach removes ambiguity and solidifies spatial functional association, as demonstrated at sites like Marathousa 1 (Konidaris et al., 2018) or TK Sivatherium (Panera et al., 2019). This method will play a major role in the present study.

This section overstates what spatial analysis can demonstrate and misrepresents the cited studies. The works by Giusti (2021), Konidaris et al. (2018), and Panera et al. (2019) do use spatial statistics to examine relationships between artifacts and faunal remains, but they explicitly caution that spatial overlap alone does not prove functional or behavioral association. These studies argue that clustering can support such interpretations only when combined with detailed taphonomic and stratigraphic evidence. None of them claims that spatial analysis "removes ambiguity" or "solidifies" functional links. The text should be revised to reflect the more qualified conclusions of those papers and to avoid implying that spatial statistics can establish behavioral causation on their own.

We disagree. Both works (Giusti and Panera) use spatial statistical tools to create an inferential basis reinforcing a functional association of lithics and bones. In both cases, the anthropogenic agency inferred is based on that. We should stress that this only provides a basis for argumentation, not a definitive causation. Again, those analyses show much more than just apparent visual clustering.

(8) Line 200: Here, we present the discovery of a new elephant butchery site (Emiliano Aguirre Korongo, EAK), dated to 1.78 Ma, from the base of Bed II at Olduvai Gorge. It is the oldest unambiguous proboscidean butchery site at Olduvai.

It is fine to state the main finding in the introduction, but the phrasing here is too strong. Calling EAK "the oldest unambiguous proboscidean butchery site" asserts certainty before the evidence is presented. The claim should be stated more cautiously, for example, "a new site that provides early evidence for proboscidean butchery," so that the language reflects the strength of the data rather than pre-judging it.

We understand the caution by R2, but in this case, EAK is the oldest taphonomically-supported evidence of elephant butchery at Olduvai (see discussion about FLK North in the text). Whether this is declared at the beginning or the end of the text is irrelevant.

(9) Line 224: The drying that characterizes Bed II had not yet taken place during this moment.

This sentence reads like a literal translation. It should be rewritten for clarity.

Modified in the revised version

(10) Line 233: During the recent Holocene, the EAK site was affected by a small landslide which displaced the...

This section contains far more geological detail than is needed for the argument. The reader only needs to know that the site block was displaced by a small Holocene landslide but retains its stratigraphic integrity. The extended discussion of regional faults, seismicity, and slope processes goes well beyond what is necessary for context and distracts from the main focus of the paper.

We disagree. The geological information is what is most commonly missing from most archaeological reports. Here, it is relevant because of the atypical process and because it has been documented only twice with elephant butchery sites. Explaining the dynamic geological process that shaped the site helps to understand its spatial properties.

(11) Line 264: In June 2022, a partial elephant carcass was found at EAK on a fragmented stratigraphic block...

This section reads like field notes rather than a formal site description. Most of the details about the discovery sequence, trench setup, and excavation process are unnecessary for the main text. Only the basic contextual information about the find location, stratigraphic position, and anatomical composition is needed. The rest could be condensed or moved to the methods or supplementary material.

We disagree. See reply above.

(12) Line 291: hominins or other carnivores. Ongoing restoration work will provide an accurate estimate of well-preserved and modified fractions of the assemblage.

This sentence is unclear and needs to specify what kind of restoration work is being done and what is meant by well-preserved and modified fractions. It is not clear whether modified refers to surface marks, diagenetic alteration, or something else. If the bones are still being cleaned or prepared, the analysis is incomplete, and the counts cannot be considered final. If restoration only means conservation or stabilization, that should be stated clearly so the reader understands that it does not affect the results. As written, it is not clear whether the data presented here are preliminary or complete.

We added: For this reason, until restoration is concluded, we cannot produce any asssertion about the presence or absence of bone surface modifications.

(13) Line 294: The tibiae were well preserved, but the epiphyseal portions of the femora were missing, probably removed by carnivores, which would also explain why a large portion of the rib cage and almost all vertebrae are missing.

This explanation is not well supported. The missing elements could be the result of other forms of density-mediated destruction, such as sediment compaction or post-depositional fragmentation, especially since no tooth marks were found. Given the low density of ribs, vertebrae, and femoral epiphyses, these processes are more likely explanations than carnivore removal. The text should acknowledge these alternatives rather than attributing the pattern to carnivore activity without direct evidence.

Sediment compaction and post-depositional can break bones but cannot make them disappear. Our excavation process was careful enough to detect bone if present. Their absence indicates two possibilities: erosion through the years at the front of the excavation or carnivore intervention. Carnivores can take elephant bones without impacting the remaining assemblage (see our reply above to a similar comment).

(14) Line 304: The fact that the carcass was moved while encased in its sedimentary context, along with the close association of stone tools with the elephant bones, is in agreement with the inference that the animal was butchered by hominins. A more objective way to assess this association is through spatial statistical analysis.

The authors state that "the carcass was moved while encased in its sedimentary context, along with the close association of stone tools with the elephant bones, is in agreement with the inference that the animal was butchered by hominins." This does not logically follow. Movement of the block explains why the bones and tools remain together, not how that association was created. The preserved association alone does not demonstrate butchery, especially in the absence of cut marks or other direct evidence of hominin activity.

Again, we are sorry that R2 is completely overlooking the strong signal detected by the spatial statistical analysis. The way that the block moved, it preserved the original association of bones and tools. This statement is meant to clarify that despite the allochthonous nature of the block, the original autochthonous depositional process of both types of archaeological materials has been preserved. The spatial association, as statistically demonstrated, indicates that the functional link is more likely than any other alternative process. The additional fact that nowhere else in that portion of the outcrop do we identify scatters of tools (all appear clustered at a landscape scale with the elephant) adds more support to this interpretation. This would have been further supported by the presence of cut marks, no doubt, but their absence does not indicate lack of functional association, since as Haynes´ works have clearly shown, most bulk defleshing of modern elephant leaves no traces on most bones.

(15) Line 370: This also shows that the functional connection between the elephant bones and the tools has been maintained despite the block post-sedimentary movement.

The spatial analyses appear to have been carried out appropriately, and the interpretations of clustering and segregation are consistent with the reported results. However, the conclusion that the "functional connection" between bones and tools has been maintained goes beyond what spatial correlation alone can demonstrate. These analyses show spatial proximity and scale-dependent clustering but cannot, by themselves, confirm a behavioral or functional link.

R2 is making this comment repeatedly and we have addressed it more than once above. We disagree and we refer to our replies above to sustain it.

(16) Line 412: The clustering of the elephant (and hippopotamus) carcasses in the areas containing the highest densities of landscape surface artifacts is suggestive of a hominin agency in at least part of their consumption and modification. The presence of green broken elephant long bone elements in the area surveyed is only documented within such clusters, both for lower and upper Bed II. This constitutes inverse negative evidence for natural breaks occurring on those carcasses through natural (i.e., non-hominin) pre- and peri-mortem limb breaking (Haynes et al., 2021, 2020; Haynes and Hutson, 2020). In this latter case, it would be expected for green-broken bones to show a more random landscape distribution, and occur in similar frequencies in areas with intense hominin landscape use (as documented in high density artifact deposition) and those with marginal or non-hominin intervention (mostly devoid of anthropogenic lithic remains).

The clustering of green-bone fractures with stone tools is intriguing but should be interpreted cautiously. The Haynes references are misrepresented here. Those studies address both cut marks and green-bone (spiral) fractures, emphasizing that each can arise through non-hominin processes such as trampling, carcass collapse, and sediment loading. They do not treat green fractures as clearer evidence of butchery; in fact, they caution that such breakage patterns can occur naturally and even form clustered distributions in areas of repeated animal activity. The claim that these studies support spiral fractures as unambiguous indicators of hominin activity, or that natural breaks would be randomly distributed, is not accurate.

We would like to emphasize again that the Haynes´references are not misrepresented here. See our extensive reply above. If R2 can provide evidence of natural breakage patterns resulting from pre-mortem limb breaking or post-mortem trampling resulting in all limb bones being affected by these processes and resulting in smooth spiral breaks, accompanied with extensive and overlapping scarring on the medullary surface, in conjunction with the other features described in our replies above, then we would be willing to reconsider. With the evidence reported until now, that does not occur simultaneously on specimens resulting from studies on modern elephant bones.

R2 seems to contradict him(her)self here by saying that Haynes studies show that cut marks are not reliable because they can also be reproduced via trampling. Until this point, R2 had been saying that only cut marks could demonstrate a functional link and support butchery. Haynes´ studies do not deal experimentally with sediment loading.

(17) Line 424: This indicates that from lower Bed II (1.78 Ma) onwards, there is ample documented evidence of anthropogenic agency in the modification of proboscidean bones across the Olduvai paleolandscapes. The discovery of EAK constitutes, in this respect, the oldest evidence thereof at the gorge. The taphonomic evidence of dynamic proboscidean bone breaking across time and space supports, therefore, the inferences made by the spatial statistical analyses of bones and lithics at the site.

This conclusion is overstated. The claim of "ample documented evidence of anthropogenic agency" is too strong, given that the main support comes from indirect indicators like green-bone fractures and spatial clustering rather than clear butchery marks. It would be more accurate to say that the evidence suggests or is consistent with possible hominin involvement. The final sentence also conflates association with causation; spatial and taphonomic data can indicate a relationship, but do not confirm that the carcasses were butchered by hominins.

The evidence is based on spatially clustering (at a landscape scale) of tools and elephant (and other megafaunal taxa) bones, in conjunction with a large amount of green-broken elements. This interpretation, if we compare it against modern referents is supported even stronger. In the past few years, we have been conducting work on modern naturally dead elephant carcasses in Botswana and Zambia, and of the several carcasses that we have seen, we have not identified a single case of long bone shaft breaks like those described by Haynes as natural or like those we describe here as anthropogenic. This probably means that they are highly unlikely or marginal occurrences at a landscape scale. This seems to be supported by Haynes´ work too. Out of the hundreds of elephant carcasses that he has monitored and studied over the years for different works, we have managed to identify only two instances where he described natural pre-mortem breaks. This certainly qualifies as extremely marginal. 

Most of the Results section is clearly descriptive, but beginning with "The clustering of the elephant (and hippopotamus) carcasses..." the text shifts from reporting observations to drawing behavioral conclusions. From this point on, it interprets the data as evidence of hominin activity rather than simply describing the patterns. This part would be more appropriate for the Discussion, or should be rewritten in a neutral, descriptive way if it is meant to stay in the Results.

This appears extensively discussed in the Discussion section, but the data presented in the results is also interpreted in that section, following a clear argumental chain.

(18) Line 433: A recent discovery of a couple of hippopotamus partial carcasses at the 3.0-2.6 Ma site of Nyayanga (Kenya), spatially concurrent with stone artifacts, has been argued to be causally linked by the presence of cut marks on some bones (Plummer et al., 2023). The only evidence published thereof is a series of bone surface modifications on a hippo rib and a tibial crest, which we suggest may be the result of byproduct of abiotic abrasive processes; the marks contrast noticeably with the well-defined cut marks found on smaller mammal bones (Plummer et al. ́s 2023: Figure 3C, D) associated with the hippo remains (Plummer et al., 2023).

The authors suggest that the Nyayanga marks could result from abiotic abrasion, but this claim does not engage with the detailed evidence presented by Plummer et al. (2023). Plummer and colleagues documented well-defined, morphologically consistent cut marks and considered the sedimentary context in their interpretation. Raising abrasion as a general possibility without addressing that analysis gives the impression of selective skepticism rather than an evaluation grounded in the published data.

We disagree again on this matter. R2 does not clarify what he/she means by well-defined or morphologically consistent. We provide an alternative interpretation of those marks that fit their morphology and features and that Plummer at al did not successfully exclude. We also emphasize that the interpretation of the Nyayanga marks was made descriptively, without any analytical approach and with a high degree of subjectivity by the researcher. All of this disqualifies the approach as well defined and keeps casting an old look at modern taphonomy. Descriptive taphonomy is a thing of the 1980´s. Today there are a plethora of analytical methods, from multivariate statistics, to geometric morphometrics to AI computer vision (so far the most reliable) which represent how taphonomy (and more specifically, analysis of bone surface modifications) should be conducted in the XXI century. This approaches would reinforce interpretations as preliminarily published by Plummer et al, provided they reject alternative explanations like those that we have provided.

(19) Line 459: It would have been essential to document that the FLK N6 tools associated with the elephant were either on the same depositional surface as the elephant bones and/or on the same vertical position. The ambiguity about the FLK N6 elephant renders EAK the oldest secure proboscidean butchery evidence at Olduvai, and also probably one of the oldest in the early Pleistocene elsewhere in Africa.

The concern about vertical mixing is fair, but the tone makes it sound like the association is definitely not real. It would be more accurate to say that the evidence is ambiguous, not that it should be dismissed altogether.

We have precisely done so. We do not dismiss it, but we cannot take it for anything solid since we excavated the site and show how easily one could make functional associations if forgetting about the third dimension. It is not a secure butchery site. This is what we said and we stick to this statement.

(20) Line 479: In all cases, these wet environments must have been preferred places for water-dependent megafauna, like elephants and hippos, and their overlapping ecological niches are reflected in the spatial co-occurrence of their carcasses. Both types of megafauna show traces of hominin use through either cutmarked or percussed bones, green-broken bones, or both (Supplementary Information).

The environmental part is good, but the behavioral interpretation is too strong. Saying elephants and hippos "must have been" drawn to these areas is too certain, and claiming that both "show traces of hominin use" makes it sound like every carcass was modified. It should be clearer that only some have possible evidence of this.

The sentence only refers to both types of fauna taxonomically. No inference can be drawn therefor that all carcasses are modified.

(21) Line 496: In most green-broken limb bones, we document the presence of a medullary cavity, despite the continuous presence of trabecular bone tissue on its walls.

This sentence is confusing and doesn't seem to add anything meaningful. All limb bones naturally have a medullary cavity lined with trabecular bone, so it's unclear why this is noted as significant. The authors should clarify what they mean here or remove it if it's simply describing normal bone structure.

No. Modern elephant long bones do not have a hollow medullary cavity. All the medullary volume is composed of trabecular tissue. Some elephants in the past had hollow medullary cavities, which probably contained larger amounts of marrow and fat. 

(22) Line 518: We are not confident that the artefacts reported by de la Torre et al are indeed tools.

While I generally agree with this statement, the paragraph reads as defensive rather than comparative. It would help if they briefly summarized what de la Torre et al. actually argued before explaining why they disagree.

We devote two full pages of the Discussion section to do so precisely.

(23) Lines 518-574: They are similar to the green-broken specimens that we have reported here...

This part is very detailed but inconsistent. They argue that the T69 marks could come from natural processes, but they use similar evidence (green fractures, overlapping scars) to argue for human activity at EAK. If equifinality applies to one, it applies to both.

We are confused by this misinterpretation. Features like green fractures and overlapping scars (among others) can be used to detect anthropogenic agency in elephant bone breaking; that is, any given specimen can be determined to have been an “artifact” (in the sense of human-created item), but going from there to interpreting an artifact as a tool, there is a large distance. Whereas an artifact (something made by a human) can be created indirectly through several processes (for example, demarrowing a bone resulting in long bone fragments), a tool suggest either intentional manufacture and use or both. That is the difference between de la Torre et al.´s interpretation and ours. We believe that they are showing anthropogenically-made items, but they have provided no proof that they were tools.

(24) Line 576: A final argument used by the authors to justify the intentional artifactual nature of their bone implements is that the bone tools were found in situ within a single stratigraphic horizon securely dated to 1.5 million years ago, indicating systematic production rather than episodic use. This is taphonomically unjustified.

The reasoning here feels uneven in how clustering evidence is used. At EAK, clustering of bones and artifacts is taken as meaningful evidence of hominin activity, but here the same pattern at T69 is treated as a natural by-product of butchery or carnivore activity. If clustering alone cannot distinguish between intentional and incidental association, the authors should clarify why it is interpreted as diagnostic in one case but not in the other.

Again, we are confused by this misinterpretation. It applies to two different scenarios/questions:

a) is there a functional link between tools and bones at EAK and T69? We have statistically demonstrated that at EAK and we think de la Torre et al. is trying to do the same for T69, although using a different method. 

b) Are the purported tools at T69 tools? Are those that we report here tools? In this regard there is no evidence for either case and given that several bones from T69 come from animals smaller than elephants, we do not discard that carnivores might have been responsible for those, whereas hominin butchery might have been responsible for the intense long limb breaking at that site. It remains to be seen how many (if any) of those specimens were tools.

(25) Line 600: If such a bone implement was a tool, it would be the oldest bone tool documented to date (>1.7 Ma).

The comparison to prior studies is useful, and the point about missing use-wear traces is well taken. However, the last lines feel speculative. If no clear use evidence has been found, it's premature to suggest that one specimen "would be the oldest bone tool." That claim should be either removed or clearly stated as hypothetical.

It clearly reads as hypothetical.

(26) Line 606: Evidence documents that the oldest systematic anthropogenic exploitation of proboscidean carcasses are documented (at several paleolandscape scales) in the Middle Pleistocene sites of Neumark-Nord (Germany)(Gaudzinski-Windheuser et al., 2023a, 2023b).

This is the first and only mention of Neumark-Nord in the paper, and it appears without any prior discussion or connection to the rest of the study. If this site is being used for comparison or as part of a broader temporal framework, it needs to be introduced and contextualized earlier. As written, it feels out of place and disconnected from the rest of the argument.

This is a Late Pleistocene site and we do not see the need to present it earlier, given that the scope of this work is Early Pleistocene.

(27) Line 608: Evidence of at least episodic access to proboscidean remains goes back in time (see review in Agam and Barkai, 2018; Ben-Dor et al., 2011; Haynes, 2022).

The distinction between "systematic" and "episodic" exploitation is useful, but the authors should clarify what criteria define each. The phrase "episodic access...goes back in time" is vague and could be replaced with a clearer statement summarizing the nature of the earlier evidence.

It is self-explanatory

(28) Line 610: Redundant megafaunal exploitation is well documented at some early Pleistocene sites from Olduvai Gorge (Domínguez-Rodrigo et al., 2014a, 2014b; Organista et al., 2019, 2017, 2016).

The phrase "redundant megafaunal exploitation" needs clarification. "Redundant" is not standard terminology in this context. Does this mean repeated, consistent, or overlapping behaviors? Also, while these same Olduvai sites are mentioned earlier, this phrasing also introduces new interpretive language not used before and implies a broader behavioral generalization than what the data actually show.

Webster: Redundant means repetitive, occurring multiple times.

(29) Line 612: At the very same sites, the stone artifactual assemblages, as well as the site dimensions, are substantially larger than those documented in the Bed I Oldowan sites (Diez-Martín et al., 2024, 2017, 2014, 2009).

The placement and logic of this comparison are unclear. The discussion moves from Middle Pleistocene Neumark-Nord to early Pleistocene Olduvai sites, then to Bed I Oldowan contexts without clearly signaling the temporal or geographic transitions. If the intent is to contrast Acheulean vs. Oldowan site scale or organization, that connection needs to be made explicit. As written, it reads as a disjointed shift rather than a continuation of the argument.

We disagree. Here, we finalize by bringing in some more recent assemblages where hominin agency is not in question.

(30) Line 616: Here, we have reported a significant change in hominin foraging behaviors during Bed I and Bed II times, roughly coinciding with the replacement of Oldowan industries by Acheulian tool kits -although during Bed II, both industries co-existed for a substantial amount of time (Domínguez-Rodrigo et al., 2023; Uribelarrea et al., 2019, 2017).

This section should be restructured for flow. The reference to behavioral change during Bed I-II and the overlap of Oldowan and Acheulean industries is important, but feels buried after a long detour. Consider moving this earlier or rephrasing so the main conclusion (behavioral change across Beds I-II) is clearly stated first, followed by supporting examples.

It is not within the scope of this work and is properly described in the references mentioned.

(31) Line 620: The evidence presented here, together with that documented by de la Torre et al. (2025), represents the most geographically extensive documentation of repeated access to proboscidean and other megafaunal remains at a single fossil locality.

The phrase "most geographically extensive documentation of repeated access" overstates what has been demonstrated. The evidence presented is site-specific and does not justify such a broad superlative. This should be toned down or supported with comparative quantitative data.

We disagree. There is no other example where such an abundant record of green-broken elements from megafauna is documented. Neumark-Nord is more similar because it shows extensive evidence of butchery, but not so much about degreasing.

(32) Line 623: The transition from Oldowan sites, where lithic and archaeofaunal assemblages are typically concentrated within 30-40 m2 clusters, to Acheulean sites that span hundreds or even over 1000 m2 (as in BK), with distinct internal spatial organization and redundancy in space use across multiple archaeological layers spanning meters of stratigraphic sequence (Domínguez-Rodrigo et al., 2014a, 2009b; Organista et al., 2017), reflects significant behavioral and technological shifts.

This sentence about site size and spatial organization repeats earlier claims without adding new insight. If it's meant as a synthesis, it should explicitly say how the spatial expansion relates to changes in behavior or mobility, not just describe the difference.

In the Conclusion section these correlations have been explained in more detail to add some causation.

(33) Line 628: This pattern likely signifies critical innovations in human evolution, coinciding with major anatomical and physiological transformations in early hominins (Dembitzer et al., 2022; Domínguez-Rodrigo et al., 2021, 2012).

The conclusion that this "signifies critical innovations in human evolution" is too sweeping, given the data presented. It introduces physiological and anatomical transformation without connecting it to any evidence in this paper. Either cite the relevant findings or limit the claim to behavioral implications.

The references cited elaboration in extension this. The revised version of the Conclusion section also elaborates on this.

Overall, the conclusions section reads as a loosely connected set of assertions rather than a focused synthesis. It introduces new interpretations and terminology not supported or developed earlier in the paper, and the argument jumps across temporal and geographic scales without clear transitions. The discussion should be restructured to summarize key results, clarify the scope of interpretation, and avoid speculative or overstated claims about evolutionary significance.

We have done so, supported by the references used in addition to extending some of the arguments

(34) Line 639: The systematic excavation of the stratigraphic layers involved a small crew.

This sentence is not necessary.

No comment

(35) Line 643: The orientation and inclination of the artifacts were recorded using a compass and an inclinometer, respectively.

What were these measurements used for (e.g., post-depositional movement analysis, spatial patterning)? A short note on the purpose would make this more meaningful.

Fabric analysis has been added to the revised version.

(36) Line 659: Restoration of the EAK elephant bones

This section could be streamlined and clarified. It includes procedural detail that doesn't contribute to scientific replicability (e.g., the texture of gauze, number of consolidant applications), while omitting some key information (such as how restoration may have affected analytical results). It also contains interpretive comments ("most of the assemblage has been successfully studied") that don't belong in Methods.

No comment

(37) Line 689: In the field laboratory, cleaning of the bone remains was carried out, along with adhesion of fragments and their consolidation when necessary.

Clarify whether cleaning or adhesion treatments might obscure or alter bone surface modifications, as this has analytical implications.

These protocols do not impact bone like that anymore.

(38) Line 711: (b) Percussion Tools - Includes hammerstones or cobbles exhibiting diagnostic battering, pitting, and/or impact scars consistent with percussive activities.

Define how diagnostic features (battering, pitting) were identified - visual inspection, magnification, or quantitative criteria?

Both macro and microscopically

(39) Line 734: We conducted the analysis in three different ways after selecting the spatial window, i.e., the analysed excavated area (52.56 m2).

Clarify why the 52.56 m² spatial window was chosen. Was this the total excavated area or a selected portion?

It was what was left of the elephant accumulation after erosion.

(40) Line 728: The spatial statistical analyses of EAK.

Adding one or two sentences at the start explaining the analytical objective, such as testing spatial association between faunal and lithic materials, would help readers understand how each analysis relates to the broader research questions.

This is well explained in the main text

(41) Line 782: An intensive survey seeking stratigraphically-associated megafaunal bones was carried out in the months of June 2023 and 2024.

It would help to specify whether the same areas were resurveyed in both field seasons or if different zones were covered each year. This information is important for understanding sampling consistency and potential spatial bias.

Both areas were surveyed in both field seasons. We were very consistent.

(42) Line 787: We focused on proboscidean bones and used hippopotamus bones, some of the most abundant in the megafaunal fossils, as a spatial control.

Clarify how the hippopotamus remains functional as a "spatial control." Are they used as a proxy for water-associated taxa to test habitat patterning, or as a baseline for comparing carcass distribution? The meaning of "control" in this context is ambiguous.

As a proxy for megafaunal distribution given their greater abundance over any other megafaunal taxa.

(43) Line 789: Stratigraphic association was carried out by direct observation of the geological context and with the presence of a Quaternary geologist during the whole survey.

This is good methodological practice, but it would be helpful to describe how stratigraphic boundaries were identified in the field (for example, by reference to tuffs or marker beds). That information would make the geological framework more replicable.

This is basic geological work. Of course, both tuffs and marker beds were followed.

(44) Line 791: When fossils found were ambiguously associated with specific strata, these were excluded from the present analysis.

You might specify what proportion of the total finds were excluded due to uncertain stratigraphic association. Reporting this would indicate the strength of the stratigraphic control.

This was not quantified but it was a very small amount compared to those whose stratigraphic provenience was certain.

(45) Line 799: The goals of this survey were: a) collect a spatial sample of proboscidean and megafaunal bones enabling us to understand if carcasses on the Olduvai paleolandscapes were randomly deposited or associated to specific habitats.

You might clarify how randomness or habitat association was tested.

Randomness was tested spatially and comparing density according to ecotone. Same for habitat association.

(46) The Methods section provides detailed information about excavation, restoration, and spatial analyses but omits critical details about the zooarchaeological and taphonomic procedures. There is no explanation of how faunal remains were analyzed once recovered, including how cut marks, percussion marks, or green bone fractures were identified or what magnification or diagnostic criteria were used. The authors also do not specify the analytical unit used for faunal quantification (e.g., NISP, MNI, MNE, or other), making it unclear how specimen counts were generated for spatial or taphonomic analyses. Even if these details are provided in the Supplementary Information, the main text should include at least a concise summary describing the analytical framework, the criteria for identifying surface modifications and fracture morphology, and the quantification system employed. This information is essential for transparency, replicability, and proper evaluation of the behavioral interpretations.

See reply above. There is a new subsection on taphonomic methods now.

Supplementary information:

(47) The Supplementary Information includes a large number of green-broken proboscidean specimens from other Olduvai localities (BK, LAS, SC, FLK West), but it is never explained why these are shown or how they relate to the EAK study. The main analysis focuses entirely on the EAK elephant, including so much unrelated material without any stated purpose, which makes the supplement confusing. If these examples are meant only to illustrate the appearance of green fractures, that should be stated. Otherwise, the extensive inclusion of non-EAK material gives the impression that they were part of the analyzed assemblage when they were not.

This is stated in the opening paragraph to the section.

(48) Line 96: A small collection of green-broken elephant bones was retrieved from the lower and upper Bed II units.

It would help to clarify whether these specimens are part of the EAK assemblage or derive from other Bed II localities. As written, it is not clear whether this description refers to material analyzed in the main text or to comparative examples shown only in the Supplementary Information.

No, EAK only occupies the lower Bed II section. They belong in the Bed II paleolandscape units.

(49) Line 97: One of them, a proximal femoral shaft found within the LAS unit, has all the traces of having been used as a tool (Figure 6).

This says the bone tool in Figure 6 is from LAS, but the main text caption identifies it as from EAK. If I am not mistaken, EAK is a site at the base of Bed II, and LAS is a separate stratigraphic unit higher in the sequence, so the authors should clarify which is correct.

Our mistake. It provenience is from LAS in the vicinity of EAK.

(50) Line 186: Figure S20. Example of other megafaunal long bone shafts showing green breaks.

Not cited in text or SI narrative. No indication where these bones come from or why they are relevant.

It appears justified in the revised version.

(51) Line 474: Figure S28-S30. Hyena-ravaged giraffe bones from Chobe (Botswana).

These figures are not discussed in the text or SI, and their relevance to the study is unclear. The authors should explain why these modern comparative examples were included and how they inform interpretations of the Olduvai assemblages.

It appears justified in the revised version.

(52) Line 498: Figure S31. Bos/Bison bone from Bois Roche (France).

This figure is not mentioned in the text or Supplementary Information. The authors should specify why this specimen is shown and how it contributes to the study's taphonomic or behavioral comparisons.

It appears justified in the revised version.

(53) Line 504: Figure S32. Miocene Gomphotherium femur from Spain.

This figure is never referenced in the paper. The authors should clarify the purpose of including a Miocene specimen from outside Africa and explain what it adds to the interpretation of Bed II material.

It appears justified in the revised version.

(54) Line 508: Figure S33. Elephant femoral shaft from BK (Olduvai).

This figure appears to show comparative material but is not cited or discussed in the text. The authors should explain why the BK material is presented here and how it relates to EAK or the broader analysis.

There are two figures labeled S33.

It appears justified in the revised version.

(55) Line 515: Figure S33. Tibia fragment from a large medium-sized bovid displaying multiple overlapping scars on both breakage planes inflicted by carnivore damage.

Because this figure repeats the S33 label and is not cited or explained in the text, it is unclear why this specimen is included or how it contributes to the study. The authors should correct the duplicate numbering and clarify the purpose of this figure.

It appears justified in the revised version.

(56) Line 522: Same specimen as shown in Figure S30, viewed on its medial side.

This is not the same bone as S30. This figure is not discussed in the text or Supplementary Information. The authors should clarify why it is included and how it relates to the rest of the analysis.

It appears justified in the revised version.

Author response:

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

Public Reviews: 

Reviewer #1 (Public review): 

Summary: 

This paper focuses on understanding how covalent inhibitors of peroxisome proliferator-activated receptor-gamma (PPARg) show improved inverse agonist activities. This work is important because PPARg plays essential roles in metabolic regulation, insulin sensitization, and adipogenesis. Like other nuclear receptors, PPARg, is a ligand-responsive transcriptional regulator. Its important role, coupled with its ligand-sensitive transcriptional activities, makes it an attractive therapeutic target for diabetes, inflammation, fibrosis, and cancer. Traditional non-covalent ligands like thiazolininediones (TZDs) show clinical benefit in metabolic diseases, but utility is limited by off-target effects and transient receptor engagement. In previous studies, the authors characterized and developed covalent PPARg inhibitors with improved inverse agonist activities. They also showed that these molecules engage unique PPARg ligand binding domain (LBD) conformations whereby the c-terminal helix 12 penetrates into the orthosteric binding pocket to stabilize a repressive state. In the nuclear receptor superclass of proteins, helix 12 is an allosteric switch that governs pharmacologic responses, and this new conformation was highly novel. In this study, the authors did a more thorough analysis of how two covalent inhibitors, SR33065 and SR36708 influence the structural dynamics of PPARg LBD. 

Strengths: 

(1) The authors employed a compelling integrated biochemical and biophysical approach.  

(2) The cobinding studies are unique for the field of nuclear receptor structural biology, and I'm not aware of any similar structural mechanism described for this class of proteins.  

(3) Overall, the results support their conclusions.  

(4) The results open up exciting possibilities for the development of new ligands that exploit the potential bidirectional relationship between the covalent versus non-covalent ligands studied here. 

Weaknesses: 

(1) The major weakness in this work is that it is hard to appreciate what these shifting allosteric ensembles actually look like on the protein structure. Additional graphical representations would really help convey the exciting results of this study. 

We thank the review for the comments. In response to the specific recommendations below, we added two new figures—Figure 1 and Figure 8 in this resubmission—that hopefully address the weakness identified by the reviewer.

Reviewer #2 (Public review): 

Summary: 

The authors use ligands (inverse agonists, partial agonists) for PPAR, and coactivators and corepressors, to investigate how ligands and cofactors interact in a complex manner to achieve functional outcomes (repressive vs. activating). 

Strengths: 

The data (mostly biophysical data) are compelling from well-designed experiments. Figures are clearly illustrated. The conclusions are supported by these compelling data. These results contribute to our fundamental understanding of the complex ligand-cofactor-receptor interactions. 

Weaknesses: 

This is not the weakness of this particular paper, but the general limitation in using simplified models to study a complex system. 

We appreciate the reviewer’s comments. Breaking down a complex system into a simpler model system, when possible, provides a unique lens with which to probe systems with mechanistic insight. While simplified models may not always explain the complexity of systems in cells, for example, our recently published work showed that a simplified model system — biochemical assays using reconstituted PPARγ ligand-binding domain (LBD) protein and peptides derived from coregulator proteins (similar to the assays in this current work) and protein NMR structural biology studies using PPARγ LBD — can explain the activity of ligand-induced PPARγ activation and repression to a high degree (pearson/spearman correlation coefficients ~0.7-0.9):

MacTavish BS, Zhu D, Shang J, Shao Q, He Y, Yang ZJ, Kamenecka TM, Kojetin DJ. Ligand efficacy shifts a nuclear receptor conformational ensemble between transcriptionally active and repressive states. Nat Commun. 2025 Feb 28;16(1):2065. doi: 10.1038/s41467-025-57325-4. PMID: 40021712; PMCID: PMC11871303.

Recommendations for the authors

Reviewer #1 (Recommendations for the authors): 

(1) More set-up is needed in the results section. The first paragraph is unclear on what is new to this study versus what was done previously. Likewise, a brief description of the assays used and the meaning behind differences in signals would help the general reader along. 

We modified the last paragraph of the introduction and first results section to hopefully better set the stage for what was done previously vs. what is new/recollected in this study. In our results section, we also include more description about what the assays measure.

(2) Since this paper is building on previous work, additional figures are needed in the introduction and discussion. Graphical depictions of what was found in the first study on how these ligands uniquely influence PPARg LBD conformation. A new model/depiction in the discussion for what was learned and its context with the rest of the field. 

Our revised manuscript includes a new Figure 1 describing the possible allosteric mechanism by which a covalent ligand inhibits binding of other non-covalent ligands that was inferred from our previous study; and a new Figure 8 with a model for what has been learned.

(3) It is stated that the results shown are representative data for at least two biological replicates. However, I do not see the other replicates shown in the supplementary information. 

We appreciate the Reviewer’s emphasis on data reproducibility and rigor. We confirm that the biochemical and cellular assay data presented are indeed representative of consistent findings observed across two or more biological replicates—and we show representative data in our figures but not the extensive replicate data in supplementary information consistent with standard practices.

(4) Figure 1a could benefit from labels of antagonists, inverse agonist, etc., next to each chemical structure. Likewise, if any co-crystal or other models are available it would be helpful to include those for comparison. 

We added the pharmacological labels to Figure 2a (old Figure 1a).

(5) The figure legends don't seem to match up completely with the figures. For example, Figure 2b states that fitted Ki values +/- standard deviation. are stated in the legend, but it's shown as the log Ki. 

We revised the figure legends to ensure they display the appropriate errors as reported from the data fitting.

(6) EC50, IC50, Ki, and Kd values alongside reported errors and R2 values for the fits should be reported in a table. 

Our revised manuscript now includes a Source Data file (Figure 5—source data 1.xlsx) of the data (n=2) plotted in Figure 5 (old Figure 4) so that readers can regenerate the plots and calculate the errors and R2 values if desired. Otherwise, fitted values and errors are reported in figures when fitting in Prism permitted and reported errors; when Prism was unable to fit data or fit the error, n.d. (not determined) is specified.

(7) Statistical analysis is missing in some places, for example, Figure 1b. 

We revised Figure 2b (old Figure 1b) to include statistical testing.

Reviewer #2 (Recommendations for the authors): 

I suggest that the authors discuss the following points to broaden the significance of the results: 

(1) The two partial agonists MRL24 and nTZDpa) are "partial" in the coactivator and corepressor recruitment assays, but are "complete" in the TR-FRET ligand displacement assay (Figure 2). Please explain that a partial agonist is defined based on the functional outcome (cofactor recruitment in this study) but not binding affinity/efficacy. 

We added the following sentence to describe the partial agonist activity of these compounds: “These high affinity ligands are partial agonists as defined on their functional outcome in coregulator recruitment and cellular transcription; i.e., they are less efficacious than full agonists at recruiting peptides derived from coactivator proteins in biochemical assays (Chrisman et al., 2018; Shang et al., 2019; Shang and Kojetin, 2024) and increasing PPARγ-mediated transcription (Acton et al., 2005; Berger et al., 2003).“

(2) Will the discovery reported here be broadly applicable? 

(a) Applicable if other partial agonists and inhibitors are used? 

(b) Applicable if different coactivators/corepressors, or different segments of the same cofactor, are used?

(c) Applicable to other NRs (their AF-2 are similar but with sequence variation)?

(d) The term "allosteric" might mean different things to different people - many readers might think that it means a "distal and unrelated" binding pocket. It might be helpful to point out that in this study, the allosteric site is actually "proximal and related". 

We expanded our introduction and/or discussion sections to expand upon these concepts; specific answers as follows:

(a) Orthosteric partial agonists?—yes, because helix 12 would clash with an orthosteiric ligand; other covalent inhibitors?—it depends on whether the covalent inhibitor stabilizes helix 12 in the orthosteric pocket.

(b) yes with some nuanced exceptions where certain segments of the same coregulator protein bind with high affinity and others apparently do not bind or bind with low affinity

(c) it is not clear yet if other NRs share a similar ligand-induced conformational ensemble to PPARγ

(d) we addressed this point in the 4th paragraph of the introduction “...the non-covalent ligand binding event we previously described at the alternate/allosteric site, which is proximal to the orthosteric ligand-binding pocket, …”

Reviewer #1 (Public review):

Summary:

This study aims to explore how different forms of "fragile nucleosomes" facilitate RNA Polymerase II (Pol II) transcription along gene bodies in human cells. The authors propose that pan-acetylated, pan-phosphorylated, tailless, and combined acetylated/phosphorylated nucleosomes represent distinct fragile states that enable efficient transcription elongation. Using CUT&Tag-seq, RNA-seq, and DRB inhibition assays in HEK293T cells, they report a genome-wide correlation between histone pan-acetylation/phosphorylation and active Pol II occupancy, concluding that these modifications are essential for Pol II elongation.

Strengths:

(1) The manuscript tackles an important and long-standing question about how Pol II overcomes nucleosomal barriers during transcription.

(2) The use of genome-wide CUT&Tag-seq for multiple histone marks (H3K9ac, H4K12ac, H3S10ph, H4S1ph) alongside active Pol II mapping provides a valuable dataset for the community.

(3) The integration of inhibition (DRB) and recovery experiments offers insight into the coupling between Pol II activity and chromatin modifications.

(4) The concept of "fragile nucleosomes" as a unifying framework is potentially appealing and could stimulate further mechanistic studies.

Weaknesses:

(1) Misrepresentation of prior literature

The introduction incorrectly describes findings from Bintu et al., 2012. The cited work demonstrated that pan-acetylated or tailless nucleosomes reduce the nucleosomal barrier for Pol II passage, rather than showing no improvement. This misstatement undermines the rationale for the current study and should be corrected to accurately reflect prior evidence.

(2) Incorrect statement regarding hexasome fragility

The authors claim that hexasome nucleosomes "are not fragile," citing older in vitro work. However, recent studies clearly showed that hexasomes exist in cells (e.g., PMID 35597239) and that they markedly reduce the barrier to Pol II (e.g., PMID 40412388). These studies need to be acknowledged and discussed.

(3) Inaccurate mechanistic interpretation of DRB

The Results section states that DRB causes a "complete shutdown of transcription initiation (Ser5-CTD phosphorylation)." DRB is primarily a CDK9 inhibitor that blocks Pol II release from promoter-proximal pausing. While recent work (PMID: 40315851) suggests that CDK9 can contribute to CTD Ser5/Ser2 di-phosphorylation, the manuscript's claim of initiation shutdown by DRB should be revised to better align with the literature. The data in Figure 4A indicate that 1 µM DRB fully inhibits Pol II activity, yet much higher concentrations (10-100×) are needed to alter H3K9ac and H4K12ac levels. The authors should address this discrepancy by discussing the differential sensitivities of CTD phosphorylation versus histone modification turnover.

(4) Insufficient resolution of genome-wide correlations

Figure 1 presents only low-resolution maps, which are insufficient to determine whether pan-acetylation and pan-phosphorylation correlate with Pol II at promoters or gene bodies. The authors should provide normalized metagene plots (from TSS to TTS) across different subgroups to visualize modification patterns at higher resolution. In addition, the genome-wide distribution of another histone PTM with a different localization pattern should be included as a negative control.

(5) Conceptual framing

The manuscript frequently extrapolates correlative genome-wide data to mechanistic conclusions (e.g., that pan-acetylation/phosphorylation "generate" fragile nucleosomes). Without direct biochemical or structural evidence. Such causality statements should be toned down.

Author response:

Public Reviews: 

Reviewer #1 (Public review): 

Summary: 

This study aims to explore how different forms of "fragile nucleosomes" facilitate RNA Polymerase II (Pol II) transcription along gene bodies in human cells. The authors propose that pan-acetylated, pan-phosphorylated, tailless, and combined acetylated/phosphorylated nucleosomes represent distinct fragile states that enable eFicient transcription elongation. Using CUT&Tagseq, RNA-seq, and DRB inhibition assays in HEK293T cells, they report a genome-wide correlation between histone pan-acetylation/phosphorylation and active Pol II occupancy, concluding that these modifications are essential for Pol II elongation. 

Strengths: 

(1) The manuscript tackles an important and long-standing question about how Pol II overcomes nucleosomal barriers during transcription. 

(2) The use of genome-wide CUT&Tag-seq for multiple histone marks (H3K9ac, H4K12ac, H3S10ph, H4S1ph) alongside active Pol II mapping provides a valuable dataset for the community. 

(3) The integration of inhibition (DRB) and recovery experiments oFers insight into the coupling between Pol II activity and chromatin modifications. 

(4) The concept of "fragile nucleosomes" as a unifying framework is potentially appealing and could stimulate further mechanistic studies. 

Really appreciate the positive or affirmative comments from the reviewer.

Weaknesses: 

(1)  Misrepresentation of prior literature 

The introduction incorrectly describes findings from Bintu et al., 2012. The cited work demonstrated that pan-acetylated or tailless nucleosomes reduce the nucleosomal barrier for Pol II passage, rather than showing no improvement. This misstatement undermines the rationale for the current study and should be corrected to accurately reflect prior evidence. 

What we said is according to the original report in the publication (Bintu et al., Cell, 2012). Here is the citation from the report:

Page 739,(Bintu, L. et al., Cell, 2012)(PMID: 23141536)

“Overall transcription through tailless and acetylated nucleosomes is slightly faster than through unmodified nucleosomes (Figure 1C), with crossing times that are generally under 1 min (39.5 ± 5.7 and 45.3 ± 7.6 s, respectively). Both the removal and acetylation of the tails increase eFiciency of NPS passage:71% for tailless nucleosomes and 63% for acetylated nucleosomes (Figures 1C and S1), in agreement with results obtained using bulk assays of transcription (Ujva´ ri et al., 2008).”

We will cite this original sentence in our revision.

(2) Incorrect statement regarding hexasome fragility

The authors claim that hexasome nucleosomes "are not fragile," citing older in vitro work. However, recent studies clearly showed that hexasomes exist in cells (e.g., PMID 35597239) and that they markedly reduce the barrier to Pol II (e.g., PMID 40412388). These studies need to be acknowledged and discussed. 

“hexasome” was introduced in the transcription field four decades ago. Later, several groups claimed that “hexasome” is fragile and could be generated in transcription elongation of Pol II. However, their original definition was based on the detection of ~100 bps DNA fragments (MNase resistant) in vivo by Micrococcal nuclease sequencing (MNase-seq), which is the right length to wrap up one hexasome histone subunit (two H3/4 and one H2A/2B) to form the sub-nucleosome of a hexasome. As we should all agree that acetylation or phosphorylation of the tails of histone nucleosomes will lead to the compromised interaction between DNA and histone subunits, which could lead to the intact naïve nucleosome being fragile and easy to disassemble, and easy to access by MNase. Fragile nucleosomes lead to better accessibility of MNase to DNA that wraps around the histone octamer, producing shorter DNA fragments (~100 bps instead of ~140 bps). In this regard, we believe that these ~100 bps fragments are the products of fragile nucleosomes (fragile nucleosome --> hexasome), instead of the other way around (hexasome --> fragile). 

Actually, two early reports from Dr. David J.  Clark’s group from NIH raised questions about the existence of hexasomes in vivo (PMID: 28157509) (PMID: 25348398).

From the report of PMID:35597239, depletion of INO80 leads to the reduction of “hexasome” for a group of genes, and the distribution of both “nucleosomes” and “hexasomes” with the gene bodies gets fuzzier (less signal to noise). In a recent theoretical model (PMID: 41425263), the corresponding PI found that chromatin remodelers could act as drivers of histone modification complexes to carry out different modifications along gene bodies. The PI found that INO80 could drive NuA3 (a H3 acetyltransferase) to carry out pan-acetylation of H3 and possibly H2B as well in the later runs of transcription of Pol II for a group of genes (SAGA-dependent). It suggests that the depletion of INO80 will affect (reduce) the pan-acetylation of nucleosomes, which leads to the drop of pan-acetylated fragile nucleosomes, subsequently the drop of “hexasomes”. This explains why depletion of INO80 leads to the fuzzier results of either nucleosomes or “hexasomes” in PMID: 35597239. The result of PMID: 35597239 could be a strong piece of evidence to support the model proposed by the corresponding PI (PMID: 41425263).

From a recent report: PMID:40412388, the authors claimed that FACT could bind to nucleosomes to generate “hexasomes”, which are fragile for Pol II to overcome the resistance of nucleosomes. It was well established that FACT enhances the processivity of Pol II in vivo via its chaperonin property. However, the exact working mechanism of FACT still remains ambiguous. A report from Dr. Cramer’s group showed that FACT enhances the elongation of regular genes but works just opposite for pausing-regulated genes (PMID: 38810649). An excellent review by Drs. Tim Formosa and Fred Winston showed that FACT is not required for the survival of a group of differentiated cells (PMID: 33104782), suggesting that FACT is not always required for transcription. It is quite tricky to generate naïve hexasomes in vitro according to early reports from the late Dr. Widom’s group. Most importantly, the new data (the speed of Pol II, the best one on bare DNA is ~27 bps/s) from the report of PMID: 40412388, which is much slower than the speed of Pol II in vivo: ~2.5 kbs/min or ~40 bps/s. From our recovering experiments (Fig. 4C, as mentioned by reviewer #3), in 20 minutes (the period between 10 minutes and 30 minutes, due to the property of CUT-&TAG-seq, of which Pol II still active after cells are collected, there is a big delay of complete stop of Pol II during the procedure of CUT&TAG experiments, so the first period of time does not actually reflect the speed of Pol II, which is ~5 kb/min), all Pol IIs move at a uniform speed of ~2.5 kbs/min in vivo. Interestingly, a recent report from Dr. Shixin Liu’s group (PMID: 41310264) showed that adding SPT4/5 to the transcription system with bare DNA (in vitro), the speed of Pol II reaches ~2.5kbs/min, exactly the same one as we derived in vivo. Similar to the original report (PMID: 23141536), the current report of PMID:40412388 does not mimic the conditions in vivo exactly.

There is an urgent need for a revisit of the current definition of “hexasome”, which is claimed to be fragile and could be generated during the elongation of Pol II in vivo. MNase is an enzyme that only works when the substrate is accessible. In inactive regions of the genome, due to the tight packing of chromatin, MNase is not accessible to individual nucleosomes within the bodies of a gene or upstream of promoters, which is why we only see phased/spacing or clear distribution of nucleosomes at the transcription start sites, but it becomes fuzzy downstream or upstream of promoters. On the other hand, for fragile nucleosomes, the accessibility to MNase should increase dramatically, which leads to the ~100 bps fragments. Based on the uniform rate (2.5 kbs/min) of Pol II for all genes derived from human 293T cells and the similar rate (2.5 kbs/min) of Pol II on bare DNA in vitro, it is unlikely for Pol II to pause in the middle of nucleosomes to generate “hexasomes” to continue during elongation along gene bodies. Similar to RNAPs in bacterial (no nucleosomes) and Archaea (tailless nucleosomes), there should be no resistance when Pol IIs transcribe along all fragile nucleosomes within gene bodies in all eukaryotes, as we characterized in this manuscript. 

(3)  Inaccurate mechanistic interpretation of DRB 

The Results section states that DRB causes a "complete shutdown of transcription initiation (Ser5-CTD phosphorylation)." DRB is primarily a CDK9 inhibitor that blocks Pol II release from promoter-proximal pausing. While recent work (PMID: 40315851) suggests that CDK9 can contribute to CTD Ser5/Ser2 di-phosphorylation, the manuscript's claim of initiation shutdown by DRB should be revised to better align with the literature. The data in Figure 4A indicate that 1 M DRB fully inhibits Pol II activity, yet much higher concentrations (10-100 ) are needed to alter H3K9ac and H4K12ac levels. The authors should address this discrepancy by discussing the differential sensitivities of CTD phosphorylation versus histone modification turnover. 

Yes, it was reported that DRB is also an inhibitor of CDK9. However, if the reviewer agrees with us and the current view in the field, the phosphorylation of Ser5-CTD of Pol II is the initiation of transcription for all Pol II-regulated genes in eukaryotes. CDK9 is only required to work on the already phosphorylated Ser5-CTD of Pol II to release the paused Pol II, which only happens in metazoans. From a series of works by us and others: CDK9 is unique in metazoans, required only for the pausing-regulated genes but not for regular genes. We found that CDK9 works on initiated Pol II (Ser5-CTD phosphorylated Pol II) and generates a unique phosphorylation pattern on CTD of Pol II (Ser2ph-Ser2ph-Ser5ph-CTD of Pol II), which is required to recruit JMJD5 (via CID domain) to generate a tailless nucleosome at +1 from TSS to release paused Pol II (PMID: 32747552). Interestingly, the report from Dr. Jesper Svejstrup’s group (PMID: 40315851) showed that CDK9 could generate a unique phosphorylation pattern (Ser2ph-Ser5ph-CTD of Pol II), which is not responsive to the popular 3E10 antibody that recognizes the single Ser2phCTD of Pol II.  This interesting result is consistent with our early report showing the unique phosphorylation pattern (Ser2ph-Ser2ph-Ser5ph-CTD of Pol II) is specifically generated by CDK9 in animals, which is not recognized by 3E10 either (PMID: 32747552). Actually, an early report from Dr. Dick Eick’s group (PMID: 26799765) showed the difference in the phosphorylation pattern of the CTD of Pol II between animal cells and yeast cells.  We have characterized how CDK9 is released from 7SK snRNP and recruited onto paused Pol II via the coupling of JMJD6 and BRD4 (PMID: 32048991), which was published on eLIFE. It is well established that CDK9 works after CDK7 or CDK8. From our PRO-seq data (Fig. 3) and CUT&TAG-seq data of active Pol II (Fig. 4), adding DRB completely shuts down all genes via inhibiting the initiation of Pol II (generation of Ser5ph-CTD of Pol II). Due to the uniqueness of CDK9 only in metazoans, it is not required for the activation of CDK12 or CDK13 (they are orthologs of CTK1 in yeast), as we demonstrated recently (PMID: 41377501). Instead, we found that CDK11/10 acts as the ortholog of Bur1 kinase from yeast, is essential for the phosphorylation of Spt5, the link of CTD of Pol II, and CDK12 (PMID: 41377501). 

(4) Insufficient resolution of genome-wide correlations 

Figure 1 presents only low-resolution maps, which are Insufficient o determine whether pan-acetylation and pan-phosphorylation correlate with Pol II at promoters or gene bodies. The authors should provide normalized metagene plots (from TSS to TTS) across different subgroups to visualize modification patterns at higher resolution. In addition, the genome-wide distribution of another histone PTM with a diFerent localization pattern should be included as a negative control. 

A popular view in the field is that the majority of genomes are inactive since they do not contain coding RNAs, which are responsible for ~20,000 protein candidates characterized in animals. However, our genomewide characterization using the four histone modification marks, active Pol II, and RNA-seq, shows a different story. Figure 1 shows that most of the human genome of HEK293T is active in producing not only protein-coding RNAs but also non-coding RNAs (the majority of them). We believe that Figure 1 could change our current view of the activity of the entire genome, and should be of great interest to general readers as well as researchers on genomics. Furthermore, it is a basis for Figure 2, which is a zoom-in of Figure 1.  

(5) Conceptual framing 

The manuscript frequently extrapolates correlative genome-wide data to mechanistic conclusions (e.g., that pan-acetylation/phosphorylation "generate" fragile nucleosomes). Without direct biochemical or structural evidence. Such causality statements should be toned down.  

The reviewer is right, we should tone down the strong sentences. However, we believe that our data is strong enough to derive the general conclusion. The reviewer may agree with us that the entire field of transcription and epigenetics has been stagnant in recent decades, but there is an urgent need for fresh ideas to change the current situation. Our novel discoveries, for sure, additional supporting data are needed, should open up a brand new avenue for people to explore. We believe that a new era of transcription will emerge based on our novel discoveries. We hope that this manuscript will attract more people to these topics. As Reviewer #3 pointed out, this story establishes the connection between transcription and epigenetics in the field. 

Reviewer #2 (Public review): 

Summary: 

In this manuscript, the authors use various genomics approaches to examine nucleosome acetylation, phosphorylation, and PolII-CTD phosphorylation marks. The results are synthesized into a hypothesis that 'fragile' nucleosomes are associated with active regions of PolII transcription. 

Strengths: 

The manuscript contains a lot of genome-wide analyses of histone acetylation, histone phosphorylation, and PolII-CTD phosphorylation. 

Weaknesses: 

This reviewer's main research expertise is in the in vitro study of transcription and its regulation in purified, reconstituted systems. 

Actually, the pioneering work of the establishment of in vitro transcription assays at Dr. Robert Roeder’s group led to numerous groundbreaking discoveries in the transcription field. The contributions of in vitro work in the transcription field are the key for us to explore the complexity of transcription in eukaryotes in the early times and remain important currently.

I am not an expert at the genomics approaches and their interpretation, and overall, I had a very hard time understanding and interpreting the data that are presented in this manuscript.  I believe this is due to a problem with the manuscript, in that the presentation of the data is not explained in a way that's understandable and interpretable to a non-expert.

Thanks for your suggestions. You are right, we have problems expressing our ideas clearly in this manuscript, which could confuse. We will make modifications accordingly per your suggestions.

For example: 

(1) Figure 1 shows genome-wide distributions of H3K9ac, H4K12ac, Ser2phPolII, mRNA, H3S10ph, and H4S1ph, but does not demonstrate correlations/coupling - it is not clear from these data that pan-acetylation and pan-phosphorylation are coupled with Pol II transcription. 

Figure 1 shows the overall distribution of the four major histone modifications, active Pol II, and mRNA genome-wide in human HEK293T cells. It tells general readers that the entire genome is quite active and far more than people predicted that most of the genome is inactive, since just a small portion of the genome expresses coding RNAs (~20,000 in animals). Figure 1 shows that the majority of the genome is active and expresses not only coded mRNA but also non-coding RNAs. After all, it is the basis of Figure 2, which is a zoom-in of Figure 1. However, it is beyond the scope of this manuscript to discuss the non-coding RNAs. 

(2) Figure 2 - It's not clear to me what Figure 2 is supposed to be showing. 

(A) Needs better explanation - what is the meaning of the labels at the top of the gel lanes? 

Figure 2 is a zoom-in for the individual gene, which shows how histone modifications are coupled with Pol II activity on the individual gene. We will give a more detailed explanation of the figure per the reviewer’s suggestions.

(B) This reviewer is not familiar with this technique, its visualization, or its interpretation - more explanation is needed. What is the meaning of the quantitation graphs shown at the top? How were these calculated (what is on the y-axis)? 

Good suggestions, we will do some modifications.

(3) To my knowledge, the initial observation of DRB eFects on RNA synthesis also concluded that DRB inhibited initiation of RNA chains (pmid:982026) - this needs to be acknowledged. 

Thanks for the reference, which is the first report to show the DRB inhibits initiation of Pol II in vivo. We will cite it in the revision.  

(4) Again, Figures 4B, 4C, 5, and 6 are very difficult to understand - what is shown in these heat maps, and what is shown in the quantitation graphs on top? 

Thanks for the suggestions, we will give a more detailed description of the Figures.  

Reviewer #3 (Public review): 

Summary: 

Li et al. investigated the prevalence of acetylated and phosphorylated histones (using H3K9ac, H4K12ac, H3S10ph & H4S1ph as representative examples) across the gene body of human HEK293T cells, as well as mapping elongating Pol II and mRNA. They found that histone acetylation and phosphorylation were dominant in gene bodies of actively transcribing genes. Genes with acetylation/phosphorylation restricted to the promoter region were also observed. Furthermore, they investigated and reported a correlation between histone modifications and Pol II activity, finding that inhibition of Pol II activity reduced acetylation/phosphorylation levels, while resuming Pol II activity restored them. The authors then proposed a model in which panacetylation or pan-phosphorylation of histones generates fragile nucleosomes; the first round of transcription is accompanied by panacetylation, while subsequent rounds are accompanied by panphosphorylation. 

Strengths: 

This study addresses a highly significant problem in gene regulation. The author provided riveting evidence that certain histone acetylation and/or phosphorylation within the gene body is correlated with Pol II transcription. The author furthermore made a compelling case that such transcriptionally correlated histone modification is dynamic and can be regulated by Pol II activity. This work has provided a clearer view of the connection between epigenetics and Pol II transcription. 

Thanks for the insightful comments, which are exactly what we want to present in this manuscript. 

Weaknesses: 

The title of the manuscript, "Fragile nucleosomes are essential for RNA Polymerase II to transcribe in eukaryotes", suggests that fragile nucleosomes lead to transcription. While this study shows a correlation between histone modifications in gene bodies and transcription elongation, a causal relationship between the two has not been demonstrated. 

Thanks for the suggestions. What we want to express is that the generation of fragile nucleosomes precedes transcription, or, more specifically, transcription elongation. The corresponding PI wrote a hypothetical model on how pan-acetylation is generated by the coupling of chromatin remodelers and acetyltransferase complexes along gene bodies, in which chromatin remodelers act as drivers to carry acetyltransferases along gene bodies to generate pan-acetylation of nucleosomes (PMID: 41425263). We have a series of work to show how “tailless nucleosomes” at +1 from transcription start sites are generated to release paused Pol II in metazoans (PMID: 28847961) (PMID: 29459673) (PMID: 32747552) (PMID: 32048991).   We still do not know how pan-phosphorylation along gene bodies is generated. It should be one of the focuses of our future research.

Reviewer #1 (Public review):

This study by Vitar et al. probes the molecular identity and functional specialization of pH-sensing channels in cerebrospinal fluid-contacting neurons (CSFcNs). Combining patch-clamp electrophysiology, laser-based local acidification, immunohistochemistry, and confocal imaging, the authors propose that PKD2L1 channels localized to the apical protrusion (ApPr) function as the predominant dual-mode pH sensor in these cells.

The work establishes a compelling spatial-physiological link between channel localization and chemosensory behavior. The integration of optical and electrical approaches is technically strong, and the separation of phasic and sustained response modes offers a useful conceptual advance for understanding how CSF composition is monitored.

Several aspects of data interpretation, however, require clarification or reanalysis-most notably the single-channel analyses (event counts, Po metrics, and mixed parameters), the statistical treatment, and the interpretation of purported "OFF currents." Additional issues include PKD2L1-TRPP3 nomenclature consistency, kinetic comparison with ASICs, and the physiological relevance of the extreme acidification paradigm. Addressing these points will substantially improve reproducibility and mechanistic depth.

Overall, this is a scientifically important and technically sophisticated study that advances our understanding of CSF sensing, provided that the analytical and interpretative weaknesses are satisfactorily corrected.

(1) The authors should re-analyze electrophysiological data, focusing on macroscopic currents rather than statistically unreliable Po calculations. Remove or revise the Po analysis, which currently conflates current amplitude and open probability.

(2) PKD2L1-TRPP3 nomenclature should be clarified and all figure labels, legends, and text should use consistent terminology throughout.

(3) The authors should reinterpret the so-called OFF currents as pH-dependent recovery or relaxation phenomena, not as distinct current species. Remove the term "OFF response" from the manuscript.

(4) Evidence for physiological relevance should be provided, including data from milder acidification (pH 6.5-6.8) and, where appropriate, comparisons with ASIC-mediated currents to place PKD2L1 activity in context.

(5) Terminology and data presentation should be unified, adopting consistent use of "predominant" (instead of "exclusive") and "sustained" (instead of "tonic"), and all statistical formats and units should be standardized.

(6) The Discussion should be expanded to address potential Ca²⁺-dependent signaling mechanisms downstream of PKD2L1 activation and their possible roles in CSF flow regulation and central chemoreception.

Author response:

Public Reviews:

Reviewer #1 (Public review):

This study by Vitar et al. probes the molecular identity and functional specialization of pH-sensing channels in cerebrospinal fluid-contacting neurons (CSFcNs). Combining patch-clamp electrophysiology, laser-based local acidification, immunohistochemistry, and confocal imaging, the authors propose that PKD2L1 channels localized to the apical protrusion (ApPr) function as the predominant dual-mode pH sensor in these cells.

The work establishes a compelling spatial-physiological link between channel localization and chemosensory behavior. The integration of optical and electrical approaches is technically strong, and the separation of phasic and sustained response modes offers a useful conceptual advance for understanding how CSF composition is monitored.

Several aspects of data interpretation, however, require clarification or reanalysis-most notably the single-channel analyses (event counts, Po metrics, and mixed parameters), the statistical treatment, and the interpretation of purported "OFF currents." Additional issues include PKD2L1-TRPP3 nomenclature consistency, kinetic comparison with ASICs, and the physiological relevance of the extreme acidification paradigm. Addressing these points will substantially improve reproducibility and mechanistic depth.

Overall, this is a scientifically important and technically sophisticated study that advances our understanding of CSF sensing, provided that the analytical and interpretative weaknesses are satisfactorily corrected.

(1) The authors should re-analyze electrophysiological data, focusing on macroscopic currents rather than statistically unreliable Po calculations. Remove or revise the Po analysis, which currently conflates current amplitude and open probability.

We agree with the reviewer that the Po analysis has strong limitations, particularly in experiments where the recording times are short, such as when extracellular pH is changed via photolysis (Figure 4D) or puff application (Figure 3Aa). To circumvent this problem and not rely solely on Po estimations, we used alternative methods, including an analysis of the total membrane charge (extensively used throughout the manuscript, as in Figures 3A and 4D) and an analysis of event latencies (Figure 4G). Nevertheless, single channel recordings contain information that is not included in the macroscopic current analysis. In the revised version, we intend to stress that the elementary current amplitude is conserved during manipulations such as pH changes, leaving the total number of channels (N) and the channel open probability (Po) as possible culprits for the current changes. Since these changes are rapid and reversible, it is likely that N remains constant while Po changes. To address the reviewer’s concern, we propose the following changes/reanalysis: (i) report in each condition the minimum N (based on the maximum number of simultaneously open channels; for example, in Figure 3Aa, the minimum N goes from 4-5 in control conditions to 1 during the puff of the pH 6.4 solution). Although imperfect, this method provides a tentative estimate of Po; (ii) report the fraction of time that the channels remain open; (iii) revise the text and figures to use the expression “apparent Po” instead of “Po”, acknowledging the limitations of the measurement in short recordings. We also acknowledge that some traces (Figure 3Aa, top) may appear confusing, as they seem to show macroscopic currents. We will modify these figures by including the amplitude histograms (as in Figure 1Bb) to clearly demonstrate that recordings from CSFcNs primarily reflect single-channel activity when challenged with pH changes.

(2) PKD2L1-TRPP3 nomenclature should be clarified and all figure labels, legends, and text should use consistent terminology throughout.

We agree with the reviewer that the nomenclature for the polycystin protein family is confusing. In this manuscript, we have followed the nomenclature  proposed in a recent comprehensive review on polycystin channels by Palomero, Larmore and DeCaen (Palomero et al. 2023), which refer to the channels by their gene names. As indicated in that review, the PKD2L1 channel corresponds to TRPP2 (previously known as TRPP3, see their Table 1). However, in another recent review on TRP channels,  the PKD2L1 channel is referred to as TRPP3 (Zhang et al. 2023). To prevent any ambiguity, we will remove references to the TRPP nomenclature from the text and exclusively use the PKD2L1 acronym.

(3) The authors should reinterpret the so-called OFF currents as pH-dependent recovery or relaxation phenomena, not as distinct current species. Remove the term "OFF response" from the manuscript.

Although largely used in the literature, we concur with the reviewer that the term “OFF response” is not very helpful from a biophysical perspective as it may imply the existence of a distinct current. Consequently, we will remove the terms “OFF response” and “OFF current” from the revised manuscript and replace them with the term “photolysis-evoked PKD2L1 current”. Furthermore, to improve the logical flow, we will condense the two sections (“The proton-induced current is an off-current” and “The off-current is mediated by the activation of PKD2L1 channels”) into a single, new section titled “The photolysis-induced current is mediated by PKD2L1 channels”. This consolidation will prevent the artificial separation of the description of this current. Finally, we will revise the discussion to better characterize this photolysis-evoked phenomenon as a recovery current.

(4) Evidence for physiological relevance should be provided, including data from milder acidification (pH 6.5-6.8) and, where appropriate, comparisons with ASIC-mediated currents to place PKD2L1 activity in context.

This point is partly addressed in Figure 3. The data indicate that  PKD2L1 channels are highly sensitive to pH variations within the physiological range. To strengthen this conclusion, we will add the EC50 values derived from the curve fittings to the figure. Regarding ASIC-mediated currents, one of our main conclusions is that ASICs are not present in the apical process (ApPr), as the effects of proton photolysis in the ApPr are not blocked by ASIC antagonists. Our results suggest that PKD2L1 channels are the exclusive pH sensitive channels in the ApPr. ASIC channels likely mediate acid sensitivity in the soma, although we have not investigated the latter in detail. We intend to modify the Discussion in order to provide a physiological framework linking channel activity with physiological and pathophysiological pH changes. 

(5) Terminology and data presentation should be unified, adopting consistent use of "predominant" (instead of "exclusive") and "sustained" (instead of "tonic"), and all statistical formats and units should be standardized.

Folllowing the reviewer’s suggestions, an exhaustive rephrasing will be performed to unify terminology, data presentation and correct the text.

(6) The Discussion should be expanded to address potential Ca²⁺-dependent signaling mechanisms downstream of PKD2L1 activation and their possible roles in CSF flow regulation and central chemoreception.

This is indeed a very interesting and currently unresolved point in the physiology of CSFcNs. Published data indicate that calcium influx through PKD2L1 channels is a key regulator of apical process (ApPr) physiology. These channels are calcium permeable yet are also inhibited by intracellular calcium (DeCaen et al. 2016). Additionally, ultrastructural data show that the ApPr is rich in mitochondria and tubulo-vesicular structures resembling the Golgi apparatus (Bruni et Reddy 1987; Bjugn et al. 1988; Nakamura et al. 2023), intracellular organelles critical for calcium homeostasis. Altogether, this evidence suggests that intra-ApPr calcium concentration must be finely regulated, both in space and time, for the ApPr to fulfill its physiological roles. Based on the existing literature, we can speculate that these calcium signals are decoded by several systems: (i) calcium may act as a second messenger, linking the activation of the multimodal PKD2L1 channels to changes in CSFcN excitability, which in turn regulates spinal neuronal networks controlling locomotor activity; (ii) calcium could initiate the neurosecretion of various molecules from the ApPr into the central canal, as proposed by the Wyart group in the zebrafish in the context of bacterial infections (Prendergast et al. 2023); (iii) calcium could activate the Hedgehog signaling pathway (as has been shown by Delling et al. 2013); iv) calcium could modulate CSF flow by modulating ependymal cells ciliary activity. Resolving these downstream pathways is essential to fully define the role of CSFcNs as integrators of cerebrospinal fluid homeostasis. We will expand on this topic in the Discussion section of the revised ms.

Reviewer #2 (Public review):

Summary:

Cerebrospinal fluid contacting neurons (CSF-cNs) are GABAergic cells surrounding the spinal cord central canal (CC). In mammals, their soma lies sub-ependymally, with a dendritic-like apical extension (AP) terminating as a bulb inside the CC.

How this anatomy-soma and AP in distinct extracellular environments relate to their multimodal CSF-sensing function remains unclear.

The authors confirm that in GATA3:GFP mice, where these cells are labeled, that CSFcNs exhibit prominent spontaneous electrical activity mediated by PKD2L1 (TRPP2) channels, non-selective cation channels with ~200 pS conductance modulated by protons and mechanical forces.

They investigated PKD2L1 pH sensitivity and its effects on CSFcN excitability. They uncovered that PKD2L1 generates both phasic and tonic currents, bidirectionally modulated by pH with high sensitivity near physiological values.

Combining electrophysiology (intact and isolated AP recordings) with elegant laser-photolysis, they show that functional PKD2L1 channels localize specifically to the apical extension (AP).

This spatial segregation, coupled with PKD2L1's biophysical properties (high conductance, pH sensitivity) and the AP's unique features (very high input resistance), renders CSFcN excitability highly sensitive to PKD2L1 modulation. Their findings reveal how the AP's properties are optimised for its sensory role.

Strengths:

This is a very convincing demonstration using elegant and challenging approaches (uncaging, outside out patch of the AP) together to form a complete understanding of how these sensory cells can detect the changes of pH in the CSF so finely.

Weaknesses:

The following do not constitute weaknesses; rather, they are minor requests that this reviewer considers would complete this beautiful study.

(1) It would be nice to quantify further the relation in spontaneous as well as in acidic or basic pH between the effects observed on channel opening and holding current: do they always vary together and in a linear way?

Following the reviewer’s suggestion, we performed a Spearman’s rank correlation test. The analysis revealed a significant correlation between the changes in the apparent open probability and the holding current in paired experiments (control vs pH 6.4 pressure applications; p < 0.05, Spearman r = 0.72 and critical value = 0.67). The Pearson correlation coefficient calculated on the same data set was r = 0.63 (critical value = 0.632), indicating that the correlation is not linear. We thank the reviewer for raising this point and will add this analysis to the manuscript.

(2) Since CSF-cNs also respond to changes in osmolarity (Orts Dell Immagine 2013) & mechanosensory stimulations in a PKD2L1 dependent manner (Sternberg NC 2018), it would be nice to test the same results whether the same results hold true on the role of PKD2L1 in AP for pressure application of changes in osmolarity.

This is a very important point. As the reviewer notes, previous experimental evidence indicates that CSFcNs are also sensitive to osmolarity changes and mechanical stimulation in a PKD2L1-dependent manner. It is therefore reasonable to assume that, similar to pH sensitivity, osmotic and mechanical sensitivity depend on channels localized to the apical process (ApPr). Regarding mechanosensitivity, this spatial segregation could be tested by mechanically stimulating either the ApPr or the soma with a piezo-controlled blunt pipette (see, for example, Hao et al. 2013). Assessing sensitivity to osmotic changes, however, is more challenging, as pressure application lacks the spatial resolution to discriminate between compartments in such a compact cell. In theory, a highly localized osmotic jump could be achieved via photolysis, provided a caged compound that releases many osmotic particles simultaneously is used. In typical photolysis experiments, a localized osmotic change is produced, but its amplitude is very low (on the order of 1 to 2 mOsm).

In mice, like in fish (Sternberg et al, NC 2018), we can observe throughout the figures that a large fraction of the channel activity occurs with partial and very fast openings of the PKD2L1 channel. I recommend the authors analyse the points below:

(a) To what extent do these partial openings of the channel contribute to the changes in holding current and resting potential?

As the reviewer indicates, these partial and rapid openings are characteristic of PKD2L1 single-channel activity and appear to be conserved across species. However, estimating their precise contribution to the sustained current would require a detailed channel model, which is currently lacking. Indeed, the exact mechanism underlying this prominent sustained current in CSFcNs remains unknown and should definitely be addressed in future work.

(b) In the trace from the outside out AP, it looks like the partial transient openings are gone. Can the authors verify whether these partial openings are only present in somatic recordings?

The outside-out recordings from the apical process also show some partial openings (see the upper trace in Figure 4Db). We will specifically mention this important point in the revised version of the ms. 

(3) Previous studies have observed expression of metabotropic Glutamate receptors in CSF-cNs (transcriptome from Prendergast et al CB 2023). The authors only used blockers for ionotropic glutamate receptors in their recordings: could it be that these metabotropic receptors influence the response to uncaging of MNI-Glu when glutamate is co-released with a proton?

We thank the reviewer for pointing out the presence of metabotropic glutamate receptors in CSFcNs. However, our evidence indicates that metabotropic receptors do not contribute to the response when uncaging MNI-glutamate. This conclusion is supported by two observations: (i) the response obtained when uncaging MNI-γLGG, which does not release glutamate (Figure 5Ab), and (ii) the response obtained when uncaging protons from DPNI-GABA (data not shown) (DPNI-GABA is a GABA cage with photochemistry similar to MNI cages that also releases a proton upon photolysis; Trigo et al. 2009), are the same. In both experiments (uncaging MNI-γLGG or DPNI-GABA) a clear photolysis-evoked PKD2L1 current is observed.

(4) In the outside out patch of the AP, PKD2L1 unitary currents appear rare. Could it be that the disruption in the cilium or underlying actin/myosin cytoskeleton drastically alter the open probability of the channel?

The reviewer is correct in noting that the opening frequency of PKD2L1 channels appears lower in outside-out patches than in whole-ApPr recordings, although we have not quantified this. We interpreted this difference as reflecting a lower channel number. However, as the reviewer suggests, a plausible alternative explanation is that the channel's biophysical properties are altered when removed from its native ionic environment or when it loses interactions with regulatory proteins. We will address this point in the Discussion.

(5) Could the authors use drugs against ASIC to specify which ASIC channels contribute to the pH response in the soma?

As described in the manuscript, we performed experiments with ASIC antagonists, although we did not attempt to characterize the specific ASIC subtype mediating the somatic response. Based on the published literature, we used both psalmotoxin-1, which blocks ASIC1 channels, and APETx2, which blocks ASIC3 channels. The presence of ASIC1 in mouse CSFcNs has been demonstrated previously (Orts-Del’immagine et al. 2012; Orts-Del’Immagine et al. 2016), while ASIC3 has been identified in lamprey CSFcNs (Jalalvand et al. 2016). When applying an acidic solution to the soma, we recorded an inward current that was substantially blocked by psalmotoxin-1, although a small residual component persisted, consistent with the earlier findings of Orts-Del’Immagine et al. We did not attempt to block this remaining Psalmotoxin1‑insensitive component.

(6) This is out of the scope of this study, but we did observe in fish a very rarely-opening channel in the PKD2L1KO mutant. I wonder if the authors have similar observations in the conditions where PKD2L1 is mainly in the closed state.

We have never seen such kind of openings in our recordings (when the channel is closed or in the presence of dibucaine).

References

Bjugn, R, H K Haugland, et P R Flood. 1988. “Ultrastructure of the mouse spinal cord ependyma”. Journal of Anatomy 160 (octobre): 117‑25.

Bruni, J. E., et K. Reddy. 1987. “Ependyma of the Central Canal of the Rat Spinal Cord: A Light and Transmission Electron Microscopic Study”. Journal of Anatomy 152 (juin): 55‑70.

Delling, Markus, Paul G. DeCaen, Julia F. Doerner, Sebastien Febvay, et David E. Clapham. 2013. ”Primary cilia are specialized calcium signaling organelles”. Nature 504 (7479): 311‑14 https://doi.org/10.1038/nature12833.

Hao, Jizhe, Jérôme Ruel, Bertrand Coste, Yann Roudaut, Marcel Crest, et Patrick Delmas. 2013. “Piezo-Electrically Driven Mechanical Stimulation of Sensory Neurons”. In Ion Channels, édité par Nikita Gamper, vol. 998. Methods in Molecular Biology. Humana Press. https://doi.org/10.1007/978-1-62703-351-0_12.

Jalalvand, Elham, Brita Robertson, Hervé Tostivint, Peter Wallén, et Sten Grillner. 2016. “The Spinal Cord Has an Intrinsic System for the Control of pH”. Current Biology: CB 26 (10): 1346‑51. https://doi.org/10.1016/j.cub.2016.03.048.

Nakamura, Yuka, Miyuki Kurabe, Mami Matsumoto, et al. 2023. “Cerebrospinal Fluid-Contacting Neuron Tracing Reveals Structural and Functional Connectivity for Locomotion in the Mouse Spinal Cord”. eLife 12 (février): e83108. https://doi.org/10.7554/eLife.83108.

Orts-Del’Immagine, Adeline, Riad Seddik, Fabien Tell, et al. 2016. “A Single Polycystic Kidney Disease 2-like 1 Channel Opening Acts as a Spike Generator in Cerebrospinal Fluid-Contacting Neurons of Adult Mouse Brainstem”. Neuropharmacology 101 (février): 549‑65. https://doi.org/10.1016/j.neuropharm.2015.07.030.

Orts-Del’immagine, Adeline, Nicolas Wanaverbecq, Catherine Tardivel, Vanessa Tillement, Michel Dallaporta, et Jérôme Trouslard. 2012. “Properties of Subependymal Cerebrospinal Fluid Contacting Neurones in the Dorsal Vagal Complex of the Mouse Brainstem”. The Journal of Physiology 590 (16): 3719‑41. https://doi.org/10.1113/jphysiol.2012.227959.

Prendergast, Andrew E., Kin Ki Jim, Hugo Marnas, et al. 2023. “CSF-Contacting Neurons Respond to Streptococcus Pneumoniae and Promote Host Survival during Central Nervous System Infection”. Current Biology 33 (5): 940-956.e10. https://doi.org/10.1016/j.cub.2023.01.039.

Trigo, Federico F., George Papageorgiou, John E. T. Corrie, et David Ogden. 2009. “Laser photolysis of DPNI-GABA, a tool for investigating the properties and distribution of GABA receptors and for silencing neurons in situ”. Journal of Neuroscience Methods 181 (2): 159‑69. https://doi.org/10.1016/j.jneumeth.2009.04.022.

Pour le premier point : dans le console.log, "listeRadio[index].value" fonctionne aussi (dans ce cas, pas besoin de mettre un argument "event" dans la fonction fléchée).

Reviewer #1 (Public review):

Mitochondrial staining difference is convincing, but the status of the mitos, fused vs fragmented, elongated vs spherical, does not seem convincing. Given the density of mito staining in CySC, it is difficult to tell what is an elongated or fused mito vs the overlap of several smaller mitos.

I'm afraid the quantification and conclusions about the gstD1 staining in CySC vs. GSCs is just not convincing-I cannot see how they were able to distinguish the relevant signals to quantify once cell type vs the other.

The overall increase in gstD1 staining with the CySC SOD KD looks nice, but again I can't distinguish different cel types. This experiment would have been more convincing if the SOD KD was mosaic, so that individual samples would show changes in only some of the cells. Still, it seems that KD of SOD in the CySC does have an effect on the germline, which is interesting.

The effect of SOD KD on the number of less differentiated somatic cells seems clear. However, the effect on the germline is less clear and is somewhat confusing. Normally, a tumor of CySC or less differentiated Cyst cells, such as with activated JAK/STAT, also leads to a large increase in undifferentiated germ cells, not a decrease in germline as they conclude they observe here. The images do not appear to show reduced number of GSCs, but if they counted GSCs at the niche, then that is the correct way to do it, but its odd that they chose images that do not show the phenotype. In addition, lower number of GSCs could also be caused by "too many CySCs" which can kick out GSCs from the niche, rather than any affect on GSC redox state. Further, their conclusion of reduced germline overall, e.g. by vasa staining, does not appear to be true in the images they present and their indication that lower vasa equals fewer GSCs is invalid since all the early germline expresses Vasa.

The effect of somatic SOD KD is perhaps most striking in the observation of Eya+ cyst cells closer to the niche. The combination of increased Zfh1+ cells with many also being Eya+ demonstrates a strong effect on cyst cell differentiation, but one that is also confusing because they observe increases in both early cyst cells (Zfh1+) as well as late cyst cells (Eya+) or perhaps just an increase in the Zfh1/Eya double-positive state that is not normally common. The effects on the RTK and Hh pathways may also reflect this disturbed state of the Cyst cells.

However, the effect on germline differentiation is less clear-the images shown do not really demonstrate any change in BAM expression that I can tell, which is even more confusing given the clear effect on cyst cell differentiation.

For the last figure, any effect of SOD OE in the germline on the germline itself is apparently very subtle and is within the range observed between different "wt" genetic backgrounds.

Comments on revisions:

Upon re-re-review, the manuscript is improved but retains many of the flaws outlined in the first reviews.

[[Bert Hubert c]] ziet geen rol (meer) voor Eurostack als initiatief, na de positieve invloed in 2024 en 2025. Zijn omschrijving wijst opnieuw naar noodzaak van actie

Foundation wants to move from talk to walk, but unclear how. This is where the groundswell comes in right, the smaller providers joining forces a la Nextcloud, the NLnet stuff EU funding.

[[Frank Karlitschek p]] has been doing this for over decade already, and that needs mentioning. I talked to him [[Berlin 2014]] at re:publica about this, in the light of the steps E and I were taking in our personal digitisation, and when I moved my company to nextcloud.

Reviewer #2 (Public review):

Summary:

This study presents a detailed single-cell transcriptomic analysis of the postnatal development of mouse anterior chamber tissues. Analysis focused on the development of cells that comprise Schlemm's Canal (SC) and trabecular meshwork (TM).

Strengths:

This developmental atlas represents a valuable resource for the research community. The dataset is robust, consisting of ~130,000 cells collected across seven time points from early post-natal development to adulthood. Analyses reveal developmental dynamics of SC and TM populations and describe the developmental expression patterns of genes associated with glaucoma.

Weaknesses:

(1) Throughout the paper, the authors place significant weight on the spatial relationships of UMAP clusters, which can be misleading (See Chari and Patcher, Plos Comb Bio 2023). This is perhaps most evident in the assessment of vascular progenitors (VP) into BEC and SEC types (Figures 4 and 5). In the text, VPs are described as a common progenitor for these types, however, the trajectory analysis in Figure 5 denotes a path of PEC -> BEC -> VP -> SEC. These two findings are incongruous and should be reconciled. The limitations of inferring relationships based on UMAP spatial positions should be noted.

(2) Figure 2d does not include P60. It is also noted that technical variation resulted in fewer TM3 cells at P21; was this due to challenges in isolation? What is the expected proportion of TM3 cells at this stage?

(3) In Figures 3a and b it is difficult to discern the morphological changes described in the text. Could features of the image be quantified or annotated to highlight morphological features?

(4) Given the limited number of markers available to identify SC and TM populations during development, it would be useful to provide a table describing potential new markers identified in this study.

(5) The paper introduces developmental glaucoma (DG), namely Axenfeld-Rieger syndrome and Peters Anomaly, but the expression analysis (Figure S20) does not annotate which genes are associated with DG.

Author response:

Public Reviews:

Reviewer #1 (Public review): 

Summary: 

This study presents a comprehensive single-cell atlas of mouse anterior segment development, focusing on the trabecular meshwork and Schlemm's canal. The authors profiled ~130,000 cells across seven postnatal stages, providing detailed and solid characterization of cell types, developmental trajectories, and molecular programs. 

Strengths: 

The manuscript is well-written, with a clear structure and thorough introduction of previous literature, providing a strong context for the study. The characterization of cell types is detailed and robust, supported by both established and novel marker genes as well as experimental validation. The developmental model proposed is intriguing and well supported by the evidence. The study will serve as a valuable reference for researchers investigating anterior segment developmental mechanisms. Additionally, the discussion effectively situates the findings within the broader field, emphasizing their significance and potential impact for developmental biologists studying the visual system. 

Weaknesses: 

The weaknesses of the study are minor and addressable. As the study focuses on the mouse anterior segment, a brief discussion of potential human relevance would strengthen the work by relating the findings to human anterior segment cell types, developmental mechanisms, and possible implications for human eye disease. Data availability is currently limited, which restricts immediate use by the community. Similarly, the analysis code is not yet accessible, limiting the ability to reproduce and validate the computational analyses presented in the study. 

In the revised version we will highlight the human relevance of our work in the discussion section. Additionally, data and codes are public on single cell portal and GEO, accession numbers have been updated.

Reviewer #2 (Public review): 

Summary: 

This study presents a detailed single-cell transcriptomic analysis of the postnatal development of mouse anterior chamber tissues. Analysis focused on the development of cells that comprise Schlemm's Canal (SC) and trabecular meshwork (TM). 

Strengths: 

This developmental atlas represents a valuable resource for the research community. The dataset is robust, consisting of ~130,000 cells collected across seven time points from early post-natal development to adulthood. Analyses reveal developmental dynamics of SC and TM populations and describe the developmental expression patterns of genes associated with glaucoma. 

Weaknesses: 

(1) Throughout the paper, the authors place significant weight on the spatial relationships of UMAP clusters, which can be misleading (See Chari and Patcher, Plos Comb Bio 2023). This is perhaps most evident in the assessment of vascular progenitors (VP) into BEC and SEC types (Figures 4 and 5). In the text, VPs are described as a common progenitor for these types, however, the trajectory analysis in Figure 5 denotes a path of PEC -> BEC -> VP -> SEC. These two findings are incongruous and should be reconciled. The limitations of inferring relationships based on UMAP spatial positions should be noted. 

(2) Figure 2d does not include P60. It is also noted that technical variation resulted in fewer TM3 cells at P21; was this due to challenges in isolation? What is the expected proportion of TM3 cells at this stage? 

(3) In Figures 3a and b it is difficult to discern the morphological changes described in the text. Could features of the image be quantified or annotated to highlight morphological features? 

(4) Given the limited number of markers available to identify SC and TM populations during development, it would be useful to provide a table describing potential new markers identified in this study. 

(5) The paper introduces developmental glaucoma (DG), namely Axenfeld-Rieger syndrome and Peters Anomaly, but the expression analysis (Figure S20) does not annotate which genes are associated with DG.

(1) We agree that inferring biological relationships from the spatial arrangement of UMAP clusters has limitations and we will qualify our interpretation accordingly in the text. We will also add clarifying language to the trajectory analysis in Figure 5. The intended developmental trajectory is PEC → VP → BEC and SEC; however, the cluster labels in Figure 5 were applied incorrectly. Specifically, VP-BECs were mislabeled as BECs, which led to the confusion.

(2) We recently published the P60 dataset separately (Tolman, Li, Balasubramanian et al., eLife 2025); these data consist of integrated single-nucleus multiome profiles that were subjected to in-depth analysis. Additionally, we found that integrating the P60 dataset with the developmental datasets obscured sub-clustering of mature cell types. In future manuscripts, we will pursue a more detailed analysis of TM development and perform time point–specific clustering, similar to the approach we used for endothelial cells (Figure 4e).

Comparing proportions of cells at different ages and as the eyes grows needs to be done cautiously. Notwithstanding the limitations, the proportions of TM1, TM2, and TM3 clusters are expected to be similar between P14 and P21 as the proportions at P14 and P60 are similar when comparing to the separately analyzed P60 data.  Importantly, our dissection strategy changed with age: from P2 to P14, we removed approximately one-third of the cornea, whereas at P21 and P60 we removed most of the cornea to help maximize representation of limbal cells as the eyes grew. This change in dissection likely contributed to the reduced number of TM3 cells observed at P21.  TM3 cells are enriched anteriorly (at-least in adult) and so are located closer to the corneal cut during dissection of the P21 eyes (which despite being larger than younger ages are still small and more delicate to accurately dissect than at P60) and are therefore more likely to be lost. Additional details are provided in the Methods section.

(3) For Figure 3a and b, we will work to add clarity by providing additional annotations and an additional illustration.

(4) We will include a table listing potential new markers for developing SC and TM populations.

(5) We will annotate the genes associated with DG in Figure S20.

charge current

Section ini ga perlu ada, karena kesalahan generation

DOI: 10.1016/j.ydbio.2025.11.011

Resource: (Cell Signaling Technology Cat# 9661, RRID:AB_2341188)

Curator: @scibot

SciCrunch record: RRID:AB_2341188

What is this?

il y a deux actualités régionales et pas de nationale ci-dessous. Peut-être relayer ce message de FNE? https://fne.asso.fr/actualites/violences-contre-les-associations-ca-continue-et-c-est-insupportable

Briefing : Éduquer à l'Orientation, de l'École Primaire à Parcoursup

Synthèse Exécutive

Ce document synthétise les perspectives et stratégies clés concernant l'éducation à l'orientation dans le système éducatif français, de l'école primaire à l'enseignement supérieur.

Il ressort que l'orientation est un processus continu et complexe qui doit commencer dès le plus jeune âge pour être efficace.

L'enjeu principal est de lutter contre les déterminismes sociaux, géographiques et de genre qui se construisent très tôt.

Les enseignants jouent un rôle central, non pas seulement informatif, mais comme des acteurs majeurs permettant aux élèves d'élargir leurs horizons et de se projeter dans un avenir concret.

Les points critiques à retenir sont :

• L'orientation est un processus et non un acte ponctuel : Elle ne doit pas se réduire au choix final sur une plateforme comme Parcoursup, mais être un cheminement accompagné tout au long de la scolarité.

• La précocité de l'intervention est essentielle : Les actions menées dès l'école primaire sont fondamentales pour "ouvrir le champ des possibles" avant que les stéréotypes ne se cristallisent.

• La lutte contre les déterminismes est l'objectif suprême : L'école a pour mission de permettre à chaque jeune de s'émanciper de son genre, de son origine sociale ou de son territoire pour construire son propre parcours.

• Une approche collaborative est indispensable : Le succès de l'accompagnement repose sur l'articulation entre les différents acteurs : enseignants, parents, psychologues de l'Éducation nationale (PsyEN), et partenaires externes comme les Régions.

• Le développement de compétences transversales est central : Au-delà de la découverte des métiers, il s'agit de développer chez l'élève la connaissance de soi, l'esprit critique, la capacité à s'informer et à se projeter dans un monde incertain.

1. Redéfinir l'Orientation : Un Processus Complexe et Multifactoriel

L'orientation est un concept polysémique, perçu différemment par chaque acteur impliqué dans le parcours de l'élève. Comprendre ces perspectives est essentiel pour construire un accompagnement cohérent.

• Pour les parents : L'orientation est souvent définie par un lieu d'affectation final. Leur principal intérêt est de voir leur enfant épanoui dans un environnement choisi.

• Pour les enseignants : L'objectif est de garantir la réussite de l'élève dans son parcours. Ils observent les compétences de l'adolescent pour l'orienter vers des formations où il pourra augmenter son potentiel et réussir.

• Pour les spécialistes (PsyEN) : L'accent est mis sur le développement psychocognitif. Ils accompagnent les élèves dans la projection, ce qui implique d'accepter de faire des deuils et de renoncer à certaines options, un processus parfois complexe à l'adolescence.

Ces acteurs, auxquels s'ajoute la Région qui détient la compétence de l'information sur les métiers et les formations, doivent s'articuler pour offrir un soutien complet.

Yoril Baudoin, DRAIO : "Les différents acteurs de l'orientation dans le système éducatif doivent s'articuler pour accompagner le jeûne dans ses projets et dans l'accomplissement de ses rêves."

2. Le "Parcours Avenir" : Une Approche Longitudinale de l'Orientation

Le "Plan Avenir", et plus spécifiquement son "plan pluriannuel d'éducation à l'orientation", formalise la nécessité de voir l'orientation comme une trajectoire continue.

• Une vision de parcours : Le plan vise à créer une articulation fluide entre l'école, le collège, le lycée et l'enseignement supérieur. Il met l'accent sur le "parcours" plutôt que sur le "projet" ponctuel.

• L'importance du primaire : L'implication du premier degré, bien que récente, est jugée cruciale.

Estelle Blanchard, directrice d'école, souligne que les élèves du primaire se posent déjà des questions sur leur avenir mais manquent de références.

L'objectif est donc de leur "ouvrir le champ des possibles" dès la maternelle, en découvrant les métiers des parents puis en élargissant progressivement les horizons.

• Le rôle renforcé des enseignants : Le plan rappelle que les enseignants sont les premiers au contact des élèves et ont un impact significatif sur leurs trajectoires.

Leur mission est d'agir activement contre les déterminismes pour permettre à chaque élève de se dire : "Et pourquoi pas moi ?".

Yoril Baudoin, DRAIO : "[Le plan Avenir, c'est] dire aux enseignants vous avez une place très forte pour travailler sur les déterminismes quel qu'ils soient et pour permettre finalement à des jeunes de se dire 'Bah l'école peut me servir d'ascenseur social ou en tout cas de m'extirper d'une assignation si tenté que ce soit mon choix.'"

3. Lutter contre les Déterminismes : Les Inégalités au Cœur du Sujet

L'un des objectifs centraux de l'éducation à l'orientation est de corriger les inégalités qui entravent les choix des élèves. Celles-ci sont de plusieurs ordres.

Déterminisme Géographique

Particulièrement prégnant dans les territoires ruraux et enclavés, il se manifeste par :

• Des choix par défaut : Des élèves aux bons résultats scolaires s'orientent vers des filières professionnelles par dépit, simplement parce que le lycée général le plus proche est trop éloigné.

• La peur de l'inconnu : Les familles et les élèves appréhendent les "grandes villes", ce qui constitue un frein majeur aux projets ambitieux.

• Solutions concrètes mises en place :

◦ Visites de lycées dès le CM1-CM2 : Pour se projeter et découvrir les formations existantes.  

◦ Implication des familles : Inviter les parents à participer à ces visites pour "dédramatiser le passage au lycée".   

◦ Sorties scolaires stratégiques : Organiser des voyages dans de grandes villes (ex: Lille) en utilisant les transports en commun (train, métro) pour familiariser élèves et parents avec cet environnement.

Déterminisme Social

Ce biais affecte non seulement les élèves mais aussi la perception des éducateurs.

• L'autocensure des accompagnants : Face à un élève de milieu modeste, la tendance peut être de proposer un "projet par petit pas" (ex: Bac Pro puis BTS) en anticipant des difficultés financières, une approche qui ne serait pas adoptée pour un élève de milieu favorisé.

• Solutions proposées :

◦ Informer sur les droits : Communiquer de manière explicite sur les aides financières et les dispositifs de soutien (bourses, services du CROUS) pour lever les freins financiers.   

◦ Intégrer les services sociaux dans les temps forts de l'orientation, comme les forums des métiers.

Déterminisme de Genre

Les stéréotypes de genre sont ancrés dès le plus jeune âge et influencent fortement les projections professionnelles.

• Des projections stéréotypées : En maternelle, les choix sont très genrés ("princesse", "chevalier").

Le travail consiste à déconstruire ces clichés (ex: les princesses peuvent être fortes et se sauver seules).

• Le risque de renforcer les stéréotypes : L'intervention de professionnels peut être contre-productive.

Un chef d'entreprise affirmant que l'industrie a besoin de filles parce qu'elles sont "minutieuses et plus sérieuses" ancre le stéréotype au lieu de le combattre.

• Les limites des "rôles modèles" : Présenter des figures au parcours exceptionnel peut être intimidant et perçu comme inaccessible, au lieu d'inspirer.

L'idéal est de présenter des modèles plus proches, dans une "zone proximale de développement", auxquels les jeunes peuvent s'identifier.

L'Influence Familiale

Les témoignages d'élèves révèlent le poids de l'entourage dans les décisions finales.

• Le dilemme de l'écoute : Une élève de terminale explique son tiraillement entre les conseils de ses professeurs (basés sur ses notes) et ceux de ses parents (basés sur sa personnalité et sa gestion du stress).

• Les choix faits "pour" les parents : Un élève raconte avoir choisi un lycée général "pour sa mère", alors qu'il aspirait à une filière professionnelle manuelle.

Ce cheminement, bien que contraint, lui a permis de mûrir son projet et de mieux l'argumenter par la suite.

4. Développer les Compétences à s'Orienter : Stratégies et Outils

Éduquer à l'orientation ne se limite pas à informer, mais vise à équiper les élèves d'un ensemble de compétences pour naviguer leur parcours.

Les Outils Pédagogiques

| Stratégie | Description | Objectifs | | --- | --- | --- | | Le Référentiel de Compétences | Cadre officiel pour le collège et le lycée qui balise les compétences à développer (ex: s'informer, se projeter). | Intégrer l'orientation dans chaque discipline et non comme une activité annexe. Développer l'esprit critique, notamment face à l'information en ligne. | | L'Introspection et Connaissance de Soi | Travail sur les compétences psychosociales : estime de soi, confiance, savoir-être, apprendre à apprendre. | Permettre à l'élève de connaître ses forces, ses faiblesses et de prendre conscience de ses capacités via des outils comme l'auto-évaluation. | | L'Immersion et l'Expérience | Mise en situation concrète : journées portes ouvertes dans les lycées, stages, "classe en entreprise" (une classe délocalisée une fois par semaine dans une entreprise locale). | Rendre l'apprentissage concret en "passant par le geste". Faciliter la projection et démystifier le monde professionnel. L'interaction entre pairs est un levier puissant. | | La Gestion des "Grands Rêves" | Accompagner les élèves qui aspirent à des carrières très sélectives (footballeur, influenceur) sans briser leurs ambitions. | Le rêve est un moteur essentiel. Le rôle de l'éducateur est d'aider l'élève à développer les compétences nécessaires et la réflexivité pour, si besoin, réajuster son projet. | | L'Intelligence Artificielle (IA) | Utilisation d'outils comme les chatbots (ex: Orian) pour fournir de l'information. | Le risque est de court-circuiter le cheminement humain. L'opportunité est d'utiliser l'IA pour la recherche documentaire afin de libérer du temps pour l'accompagnement humain qualitatif. |

Le témoignage d'une élève sur l'anxiété générée par les choix finaux sur Parcoursup illustre l'échec d'une approche tardive.

Analyse de Yoril Baudoin : "Ce qu'elle nous décrit là c'est à un moment donné l'orientation se réduit à un acte administratif où il faut faire un choix et là d'un seul coup on n'est pas prêt [...] alors que nous souhaitons travailler collégialement pour montrer que c'est un cheminement."

More photodegradation of DA linked to DOM and halides, not other seawater consitutes

Looking at figure 2a, it appears that some sites that are of low confidence using one method are identical across both methods. For these low confidence sites, it is custom to test a secondary sequence that incudes the second most likely residues. I couldn't find supplement table 2, so it is possible you've already done that but i just couldn't look at the sequences to verify, and it wasn't clear from the text either.

Aïe... Ici pour moi, ça coince. La notion de "population" est trop vague... aussi "qui demande"

Le phénomène est bien plus insidieux et subtil. La pente vers l'extrêmedroitisation est déjà là. Elle répond à un affect particulier, très pulsionnel : celui de l'homme du ressentiment nietzschéen. La population est donc l'agglomérat, la collection de ces ressentiments individuels. "population" renvoie à une notion 'collective' où le collectif n'est pas collectif justement !

j'ajouterais la dimension psychique, ce qui donnerait "qu’il est constitutif de nos organes physiologiques5, psychiques, , techniques et de toutes les perspectives d’organisations sociales possibles;"

Author response:

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

Reviewer #1:

In this well-written and timely manuscript, Rieger et al. introduce Squidly, a new deep learning framework for catalytic residue prediction. The novelty of the work lies in the aspect of integrating per-residue embeddings from large protein language models (ESM2) with a biology-informed contrastive learning scheme that leverages enzyme class information to rationally mine hard positive/negative pairs. Importantly, the method avoids reliance on the use of predicted 3D structures, enabling scalability, speed, and broad applicability. The authors show that Squidly outperforms existing ML-based tools and even BLAST in certain settings, while an ensemble with BLAST achieves state-of-the-art performance across multiple benchmarks. Additionally, the introduction of the CataloDB benchmark, designed to test generalization at low sequence and structural identity, represents another important contribution of this work.

We thank the reviewer for their constructive and encouraging assessment of the manuscript. We appreciate the recognition of Squidly’s biology-informed contrastive learning framework with ESM2 embeddings, its scalability through the avoidance of predicted 3D structures, and the contribution of the CataloDB benchmark. We are pleased that the reviewer finds these aspects to be of value, and their comments will help us in further clarifying the strengths and scope of the work.

The manuscript acknowledges biases in EC class representation, particularly the enrichment for hydrolases. While CataloDB addresses some of these issues, the strong imbalance across enzyme classes may still limit conclusions about generalization. Could the authors provide per-class performance metrics, especially for underrepresented EC classes?

We thank the reviewer for raising this point. We agree that per-class performance metrics provide important insight into generalizability across underrepresented EC classes. In response, we have updated Figure 3 to include two additional panels: (i) per-EC F1, precision and recall scores, and (ii) a relative display of true positives against the total number of predictable catalytic residues. These additions allow the class imbalance to be more directly interpretable. We have also revised the text between lines 316-321 to better contextualize our generalizability claims in light of these results.

An ablation analysis would be valuable to demonstrate how specific design choices in the algorithm contribute to capturing catalytic residue patterns in enzymes.

We agree an ablation analysis is beneficial to show the benefits of a specific approach. We consider the main design choice in Squidly to be how we select the training pairs, hence we chose a standard design choice for the contrastive learning model. We tested the effect of different pair schemes on performance and report the results in Figure 2A and lines 244258. These results are a targeted ablation in which we evaluate Squidly against AEGAN using the AEGAN training and test datasets, while systematically varying the ESM2 model size and pair-mining scheme. As a baseline, we included the LSTM trained directly on ESM2 embeddings and random pair selection.  We showed that indeed the choice of pairs has a large impact on performance, which is significantly improved when compared to naïve pairing. This comparison suggests that performance gains are attributable to reactioninformed pair-mining strategies. We recognize that the way these results were originally presented made this ablation less clear. We have revised the wording in the Results section (lines 244-247) and updated the caption to Figure 2A to emphasize the purpose of this section of the paper.

The statement that users can optionally use uncertainty to filter predictions is promising but underdeveloped. How should predictive entropy values be interpreted in practice? Is there an empirical threshold that separates high- from low-confidence predictions? A demonstration of how uncertainty filtering shifts the trade-off between false positives and false negatives would clarify the practical utility of this feature.

Thank you for the suggestion. Your comment prompted us to consider what is the best way to represent the uncertainty and, additionally, what is the best metric to return to users and how to visualize the results. Based on this, we included several new figures (Figure 3H and Supplementary Figures S3-5). We used these figures to select the cutoffs (mean prediction of 0.6, and variance < 0.225) which were then set as the defaults in Squidly, and used in all subsequent analyses. The effect of these cutoffs is most evident in the tradeoff of precision and recall. Hence users may opt to select their own filters based on the mean prediction and variance across the predictions, and these cutoffs can be passed as command line parameters to Squidly. The choice to use a consistent default cutoff selected using the Uni3175 benchmark has slightly improved the reported performance for the benchmarks seen in table 1, and figure 3C. However, our interpretation remains the same.

The excerpt highlights computational efficiency, reporting substantial runtime improvements (e.g., 108 s vs. 5757 s). However, the comparison lacks details on dataset size, hardware/software environment, and reproducibility conditions. Without these details, the speedup claim is difficult to evaluate. Furthermore, it remains unclear whether the reported efficiency gains come at the expense of predictive performance

Thank you for pointing out this limitation in how we presented the runtime results. We have rerun the tests and updated the table. An additional comment is added underneath, which details the hardware/software environment used to run both tools, as well as that the Squidly model is the ensemble version. As per the relationship between efficiency gains and predictive performance, both 3B and 15B models are benchmarked side by side across the paper.

Compared to the tools we were able to comprehensively benchmark, it does not come at a cost. However, we note that the increased benefits in runtime assume that a structure must be folded, which is not the case for enzymes already present in the PDB. If that is the case, then it is likely already annotated and, in those cases, we recommend using BLAST which is superior in terms of run time than either Squidly or a structure-based tool and highly accurate for homologous or annotated sequences.

Given the well-known biases in public enzyme databases, the dataset is likely enriched for model organisms (e.g., E. coli, yeast, human enzymes) and underrepresents enzymes from archaea, extremophiles, and diverse microbial taxa. Would this limit conclusions about Squidly's generalizability to less-studied lineages?

The enrichment for model organisms in public enzyme databases may indeed affect both ESM2 and Squidly when applied to underrepresented lineages such as archaea, extremophiles, and diverse microbial taxa. We agree that this limitation is significant and have adjusted and expanded the previous discussion of benchmarking limitations accordingly (lines 358, 369). We thank the reviewer for highlighting this issue, which has helped us to improve the transparency and balance of the manuscript.

Reviewer #2:

The authors aim to develop Squidly, a sequence-only catalytic residue prediction method. By combining protein language model (ESM2) embedding with a biologically inspired contrastive learning pairing strategy, they achieve efficient and scalable predictions without relying on three-dimensional structure. Overall, the authors largely achieved their stated objectives, and the results generally support their conclusions. This research has the potential to advance the fields of enzyme functional annotation and protein design, particularly in the context of screening large-scale sequence databases and unstructured data. However, the data and methods are still limited by the biases of current public databases, so the interpretation of predictions requires specific biological context and experimental validation.

Strengths:

The strengths of this work include the innovative methodological incorporation of EC classification information for "reaction-informed" sample pairing, thereby enhancing the discriminative power of contrastive learning. Results demonstrate that Squidly outperforms existing machine learning methods on multiple benchmarks and is significantly faster than structure prediction tools, demonstrating its practicality.

Weaknesses:

Disadvantages include the lack of a systematic evaluation of the impact of each strategy on model performance. Furthermore, some analyses, such as PCA visualization, exhibit low explained variance, which undermines the strength of the conclusions.

We thank the reviewer for their comments and feedback. 

The authors state that "Notably, the multiclass classification objective and benchmarks used to evaluate EasIFA made it infeasible to compare performance for the binary catalytic residue prediction task." However, EasIFA has also released a model specifically for binary catalytic site classification. The authors should include EasIFA in their comparisons in order to provide a more comprehensive evaluation of Squidly's performance.

We thank the reviewer for raising this point. EasIFA’s binary classification task includes catalytic, binding, and “other” residues, which differs from Squidly’s strict catalytic residue prediction. This makes direct comparison non-trivial, which is why we originally had opted to not benchmark against EasIFA and instead highlight it in our discussion.

Given your comment, we did our best to include a benchmark that could give an indication of a comparison between the two tools. To do this, we filtered EasIFA’s multiclass classification test dataset for a non-overlapping subset with Squidly and AEGAN training data and <40% sequence identity to all training sets. This left only 66 catalytic residue– containing sequences that we could use as a held-out test set from both tools. We note it is not directly equal as Squidly and AEGAN had lower average identity to this subset (8.2%) than EasIFA (23.8%), placing them at a relative disadvantage.

We also identified a potential limitation in EasIFA’s original recall calculation, where sequences lacking catalytic residues were assigned a recall of 0. We adapted this to instead consider only the sequences which do have catalytic residues, which increased recall across all models. With the updated evaluation, EasIFA continues to show strong performance, consistent with it being SOTA if structural inputs are available. Squidly remains competitive given it operates solely from sequence and has a lower sequence identity to this specific test set.

Due to the small and imbalanced benchmark size, differences in training data overlap, and differences in our analysis compared with the original EasIFA analysis, we present this comparison in a new section (A.4) of the supplementary information rather than in the main text. References to this section have been added in the manuscript at lines 265-268. Additionally, we do update the discussion and emphasize the potential benefits of using EasIFA at lines (353-356).

The manuscript proposes three schemes for constructing positive and negative sample pairs to reduce dataset size and accelerate training, with Schemes 2 and 3 guided by reaction information (EC numbers) and residue identity. However, two issues remain:

(a) The authors do not systematically evaluate the impact of each scheme on model performance.

(b) In the benchmarking results, it is not explicitly stated which scheme was used for comparison with other models (e.g., Table 1, Figure 6, Figure 8). This lack of clarity makes it difficult to interpret the results and assess reproducibility.

(c) Regarding the negative samples in Scheme 3 in Figure 1, no sampling patterns are shown for residue pairs with the same amino acid, different EC numbers, and both being catalytic residues.

We thank the reviewer for these suggestions, which enabled us to improve the clarity and presentation of the manuscript. Please find our point by point response:

(a) We thank the reviewer for highlighting the lack of clarity in the way we have presented our evaluation in the section describing the Uni3175 benchmark. We aimed to systematically evaluate the impact of each scheme using the Uni3175 benchmark and refer to these results at lines 244-258, Additionally, we have adjusted the presentation of this section at lines 244-247 also in line with related comments from reviewer 1 in order to make the intention of this section and benchmark results to allow a comparison of each scheme to baseline models and AEGAN. These results led us to use Scheme 3 in both models for the other benchmarks in Figures 2 and 3. Please let us know if there is anything we can do to further improve the interpretability of Squidly’s performance.

(b) We thank the reviewer for highlighting this issue and improving the clarity of our manuscript. We agree that after the Uni3175 benchmark was used to evaluate the schemes, we did not clearly state in the other benchmarks that scheme 3 was chosen for both the 3B and 15B models. We have made changes in table 1 and the Figure legends of Figures 2 and 3 to state that scheme 3 was used. In addition, we integrated related results into panel figures (e.g. Figures 2 and 3 now show models trained and tested on consistent benchmark datasets) and standardized figure colors and legend formatting throughout. Furthermore, we suspect that the previous switch from using the individual vs ensembled Squidly models during the paper was not well indicated, and likely to confuse the reader. Therefore, we decided to consistently report the ensembled Squidly models for all benchmarks except in the ablation study (Figure 2A). In line with this, we altered the overview Figure 1A, so that it is clearer that the default and intended version of Squidly is the ensemble.

(c) We appreciate the reviewer pointing this out. You’re correct, we explicitly did not sample the negatives described by the reviewer in scheme 3 as our focus was on the hard negatives that relate most to the binary objective.  We do think this is a great idea and would be worth exploring further in future versions of Squidly, where we will be expanding the label space used for hard-negative sampling and including binding sites in our prediction. We have updated the discussion at lines 395-396 to highlight this potential direction.

The PCA visualization (Figure 3) explains very little variance (~5% + 1.8%), but its use to illustrate the separability of embedding and catalytic residues may overinterpret the meaning of the low-dimensional projection. We question whether this figure is appropriate for inclusion in the main text and suggest that it be moved to the Supporting Information.

We thank the reviewer for this suggestion. We had discussed this as well, and in the end decided to include it in the main manuscript. We agree that the explained variance is low. However, when we first saw the PCA we were surprised that there was any separation at all. This then prompted us to investigate further, so we kept it in the manuscript to be true to the scientific story. However, we do agree that our interpretation could be interpreted as overly conclusive given the minimal variance explained by the top 2 PCs. Therefore, we agree with the assessment that the figure, alongside the accompanying results section, is more appropriately placed in the supplementary information. We moved this section (A.1) to the appendix to still explain the exploratory data analysis process that we used to tackle this problem, so that the general thought process behind Squidly is available for further reading.  

Minor Comments:

(1) Figure Quality and Legends a) In Figure 4, the legend is confusing: "Schemes 2 and 3 (S1 and S2) ..." appears inconsistent, and the reference to Scheme 3 (S3) is not clearly indicated.

(b) In Figure 6, the legend overlaps with the y-axis labels, reducing readability. The authors should revise the figures to improve clarity and ensure consistent notation.

The reviewer correctly notes inconsistencies in figure presentation. We have revised the legend of Figure 4 (now 2A) to ensure schemes are referred to consistently and Scheme 3 (S3) is clearly indicated. We also adjusted Figure 6 (now 2c) to remove the overlap between the legend and y-axis labels.  

Conclusion

We thank the reviewers and editor again for their constructive input. We believe the revisions and clarifications substantially strengthened the manuscript and the resource

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

This study used explicit-solvent simulations and coarse-grained models to identify the mechanistic features that allow for the unidirectional motion of SMC on DNA. Shorter explicit-solvent models describe relevant hydrogen bond energetics, which were then encoded in a coarse-grained structure-based model. In the structure-based model, the authors mimic chemical reactions as signaling changes in the energy landscape of the assembly. By cycling through the chemical cycle repeatedly, the authors show how these time-dependent energetic shifts naturally lead SMC to undergo translocation steps along DNA that are on a length scale that has been identified.

Strengths:

Simulating large-scale conformational changes in complex assemblies is extremely challenging. This study utilizes highly-detailed models to parameterize a coarse-grained model, thereby allowing the simulations to connect the dynamics of precise atomistic-level interactions with a large-scale conformational rearrangement. This study serves as an excellent example for this overall methodology, where future studies may further extend this approach to investigated any number of complex molecular assemblies.

We thank the reviewer for careful reading of our manuscript and highlighting the value of our bottom-up multiscale simulation approach.

Weaknesses:

The only relative weakness is that the text does not always clearly communicate which aspects of the dynamics are expected to be robust. That is, which aspects of the dynamics/energetics are less precisely described by this model? Where are the limits of the models, and why should the results be considered within the range of applicability of the models?

We appreciate this insightful comment and agree that it is important to more explicitly describe the robustness and limitations of the simulation model used in this study. In response to this comment, we have revised the Discussion section of our manuscript.

First, to clarify the robust aspects of our model, we have added a new subsection titled “Parametric choices and robustness of simulation model” to the Discussion, which is as follows:

“The switching Gō approach adopted in this study is a powerful tool for providing the relationship between known large-scale conformational changes and the resulting functional and mechanical dynamics of the molecular machine (Brandani and Takada, 2018b; Koga and Takada, 2006b; Nagae et al., 2025). In this study, we mimic conformational change induced by ATP binding and hydrolysis events by instantaneously switching the potential energy function from one that stabilized a given conformation to another that stabilized a different conformation. This drives the protein to undergo a conformational transition toward the minimum of the new energy landscape.

This approach is particularly well suited to investigate whether a given conformational change in a subunit of a molecular machine can produce the overall motion observed, and whether this process is mechanically feasible. Therefore, the fundamental mechanisms identified in this study, i.e., DNA segment capture mechanism, the correlation between step size and loop length, and the unidirectional translocation mechanism originating from the asymmetric kleisin path, can be considered as robust, as they emerge directly from the structural and topological constraints of the SMC-kleisin architecture rather than from tuned parameters.”

Additionally, to more clearly define the limits of our model, we have expanded the "Limitations in current simulations" subsection. Specifically, we have added a detailed discussion regarding the energetics and transition pathways inherent to the switching Gō approach, which is as follows:

“First, use of switching potentials to trigger conformational changes impose a limitation on predictive power for energetics and transition pathways. The switching of potentials is akin to a “vertical excitation” from one energy landscape to another, rather than a thermally activated crossing of an energy barrier. Consequently, the model cannot provide quantitative predictions of the transition rates or the free energy barriers associated with these changes. Furthermore, while the subsequent relaxation follows the new potential landscape, it is not guaranteed to reproduce the unique, physically correct transition pathway. Nevertheless, this simplification is justified because conformational changes within the protein are expected to occur on a much faster timescale than the large-scale motion of the DNA. Thus, this simplification has a limited impact on our main conclusions regarding the functional DNA dynamics driven by these large-scale conformational changes.”

We have not made any additions regarding the timescale and dwell times for each ATP state, as these were already discussed in the original manuscript.

Reviewer #2 (Public review):

Summary:

The authors perform coarse grained and all atom simulations to provide a mechanism for loop extrusion that is involved in genome compaction.

Strengths:

The simulations are very thoughtful. They provide insights into the translocation process, which is only one of the mechanisms. Much of the analyses is very good. Over all the study advances the use of simulations in this complicated systems.

We sincerely thank the reviewer for their thoughtful and encouraging comments.

Weaknesses:

Even the authors point out several limitations, which cannot be easily overcome in the paper because of the paucity of experimental data. Nevertheless, the authors could have done so to illustrate the main assertion that loop extrusion occurs by the motor translocating on DNA. They should mention more clearly that there are alternative theories that have accounted for a number of experimental data.

We thank the reviewer for these constructive suggestions. As the reviewer pointed out, it is important to state more explicitly how the unidirectional DNA translocation revealed in this study relates to the widely recognized loop-extrusion hypothesis of genome organization and situate our findings with the context of major alternative theories.

To address this, we first clarify the relationship between the translocation mechanism we observed and the phenomenon of loop extrusion. We emphasize that our simulations were designed to elucidate the core motor activity of the SMC complex, and we explicitly state our view that loop extrusion is a functional consequence of this motor activity when the complex is anchored to DNA.

Second, as the reviewer also suggested, we addressed alternative models of loop extrusion that also have experimental support in more details. We have revised the Discussion accordingly to provide a more balanced and comprehensive context. Further details are provided in our separate response to the comment below.

Reviewer #3 (Public review):

Summary:

In this manuscript, Yamauchi and colleagues combine all-atom and coarse-grained MD simulations to investigate the mechanism of DNA translocation by prokaryotic SMC complexes. Their multiscale approach is well-justified and supports a segment-capture model in which ATP-dependent conformational changes lead to the unidirectional translocation of DNA. A key insight from the study is that asymmetry in the kleisin path enforces directionality. The work introduces an innovative computational framework that captures key features of SMC motor action, including DNA binding, conformational switching, and translocation.

This work is well executed and timely, and the methodology offers a promising route for probing other large molecular machines where ATP activity is essential.

Strengths:

This manuscript introduces an innovative yet simple method that merges all-atom and coarse-grained, purely equilibrium, MD simulations to investigate DNA translocation by SMC complexes, which is triggered by activated ATP processes. Investigating the impact of ATP on large molecular motors like SMC complexes is extremely challenging, as ATP catalyses a series of chemical reactions that take and keep the system out of equilibrium. The authors simulate the ATP cycle by cycling through distinct equilibrium simulations where the force field changes according to whether the system is assumed to be in the disengaged, engaged, and V-shaped states; this is very clever as it avoids attempting to model the non-equilibrium process of ATP hydrolysis explicitly. This equilibrium switching approach is shown to be an effective way to probe the mechanistic consequences of ATP binding and hydrolysis in the SMC complex system.

The simulations reveal several important features of the translocation mechanism. These include identifying that a DNA segment of ~200 bp is captured in the engaged state and pumped forward via coordinated conformational transitions, yielding a translocation step size in good agreement with experimental estimates. Hydrogen bonding between DNA and the top of the ATPase heads is shown to be critical for segment capturtrans, as without it, translocation is shown to fail. Finally, asymmetry in the kleisin subunit path is shown to be responsible for unidirectionally.

This work highlights how molecular simulations are an excellent complement to experiments, as they can exploit experimental findings to provide high-resolution mechanistic views currently inaccessible to experiments. The findings of these simulations are plausible and expand our understanding of how ATP hydrolysis induces directional motion of the SMC complex.

We thank the reviewer for the thoughtful and encouraging assessment of our work. We appreciate the reviewer’s summary of our key contributions, especially our switching Gō strategy, the segment-capture mechanism of SMC translocation, and the role of kleisin-path asymmetry in ensuring unidirectionality.

Weaknesses:

There are aspects of the methodology and modelling assumptions that are not clear and could be better justified. The major ones are listed below:

(1) The all-atom MD simulations involve a 47-bp DNA duplex interacting with the ATPase heads, from which key residues involved in hydrogen bonding are identified. However, DNA mechanics-including flexibility and hydrogen bond formation-are known to be sequence-dependent. The manuscript uses a single arbitrary sequence but does not discuss potential biases. Could the authors comment on how sequence variability might affect binding geometry or the number of hydrogen bonds observed?

We thank the reviewer for this insightful comment regarding the potential effects of DNA sequence.

The primary biological role of the SMC complex is to organize genome architecture on a global scale; as such, its fundamental interaction with DNA is considered not to be sequence-specific. Our all-atom MD simulations and analysis pipeline were designed to probe the nature of this general interaction. Our approach confirms this rationale: the analysis exclusively identified hydrogen bonds formed between amino acid residues and the phosphate groups of the DNA's sugar-phosphate backbone. As shown in Figs. 1B and 1C, the results confirm that the key stabilizing interactions occur between basic residues on the SMC head surface and the DNA backbone. Since the backbone is chemically uniform, the stable binding mode we characterized is inherently sequence-independent.

While the final bound state is likely sequence-independent, we agree that sequence-dependent properties such as local DNA flexibility or intrinsic curvature could influence the kinetics of the binding process. For example, the rate of initial recognition or the ease of DNA bending on the head surface might vary between AT-rich and GC-rich regions. However, once the DNA is bound, we expect the stable binding geometry and the identity of the key interacting residues to be conserved across different sequences.

Therefore, we are confident that using a single, representative DNA sequence is a valid approach for elucidating the fundamental, non-sequence-specific aspects of SMC-DNA interaction and does not alter the general validity of the translocation mechanism proposed in this work.

(2) A key feature of the coarse-grained model is the inclusion of a specific hydrogen-bonding potential between DNA and residues on the ATPase heads. The authors select the top 15 hydrogen-bond-forming residues from the all-atom simulations (with contact probability > 0.05), but the rationale for this cutoff is not explained. Also, the strength of hydrogen bonds in coarse-grained models can be sensitive to context. How did the authors calibrate the strength of this interaction relative to electrostatics, and did they test its robustness (e.g., by varying epsilon or residue set)? Could this interaction be too strong or too weak under certain ionic conditions? What happens when salt is changed?

Thank you for these comments. We provide our rationale for the parameter choices below.

The contact probability cutoff of 0.05 was chosen to create a comprehensive set of residues that form physically robust interactions with DNA. To establish this robustness, we performed a parallel set of all-atom simulations using a different force field (see Fig. S2). This cross-validation revealed two key points. First, the top six residues (Arg120, Arg123, Ile63, Arg111, Arg62, and Lys56), which include experimentally confirmed DNA-binding sites, consistently exhibited the highest contact probabilities in both force fields, confirming the reliability of our identification. Second, and just as importantly, many residues with lower contact probabilities (e.g., Trp115, Tyr107, Arg105, Ser124, and Ser54) were also consistently detected across both simulations. This reproducibility suggests that these interactions are physically robust and not artifacts of a specific force field. We therefore concluded that a 0.05 cutoff is a well-balanced threshold that ensures the inclusion of not only the primary anchor residues but also the secondary, moderately interacting residues that are crucial for cooperatively stabilizing the DNA. We discussed this point in Method in the revised manuscript, which is as follows:

“The rationale for this cutoff is the physical robustness of the identified interactions; all-atom simulations using a different force field confirmed that the same set of key interacting residues, including both strong and moderate binders, was consistently identified (Fig. S2).”

The strength of the hydrogen bond potential was set to ϵ = 4.0 k​T (≈2.4 kcal/mol), a physically plausible value corresponding to an ideal hydrogen bond. To test the robustness of this parameterization, we performed preliminary simulations where we varied these parameters by (i) reducing the value of ϵ and (ii) restricting the interaction to only the top six anchor residues. In both test cases, while a short DNA duplex (47 bp) could still bind to the ATPase heads, simulations with a long DNA (800 bp) failed to form a stable DNA loop after initial docking. These tests demonstrated that a larger set of cooperative interactions with a physically realistic strength was necessary for the full segment capture mechanism. Our final parameter set (15 residues at ϵ = 4.0 k​T) was thus chosen as the parameter set required to capture both the initial anchoring of DNA and the subsequent cooperative stabilization of the captured loop.

As correctly pointed out, ionic conditions are a critical factor. Our simulations revealed that the salt concentration had a more pronounced effect on the kinetics of the DNA finding its correct binding site rather than on the thermodynamic stability of the final bound state. During our parameter tuning, we found that at physiological salt conditions (150 mM), long-range electrostatic interactions become dominant. This caused the DNA to be non-specifically captured by positively charged patches on the sides of the heads, which are not the functional binding sites. This off-pathway trapping kinetically prevented the DNA from reaching its proper location within the simulation timeframe. In contrast, the high-salt conditions (300 mM) used in this study screen these long-range interactions, suppressing non-specific trapping and allowing the DNA to efficiently explore the protein surface. This enables the correct binding to be established via the specific, short-range hydrogen bonds. Therefore, the ion concentration in our model is more as a crucial kinetic control factor to reproduce correct binding pathway within a realistic simulation timeframe. This point is discussed in the new subsection entitled “Parametric choices and robustness of simulation model”.

(3) To enhance sampling, the translocation simulations are run at 300 mM monovalent salt. While this is argued to be physiological for Pyrococcus yayanosii, such a concentration also significantly screens electrostatics, possibly altering the interaction landscape between DNA and protein or among protein domains. This may significantly impact the results of the simulations. Why did the authors not use enhanced sampling methods to sample rare events instead of relying on a high-salt regime to accelerate dynamics?

We agree that enhanced sampling methods are powerful for exploring rare events. However, many of these techniques require the pre-definition of a suitable, low-dimensional reaction coordinate (RC) to guide the simulation. The primary goal of our study was to discover the DNA translocation mechanism as it emerges naturally from fundamental physical interactions, without imposing a priori assumptions about the specific pathway.

The DNA segment capture process is complex, involving the coordinated motion of a long DNA polymer and multiple protein domains. Defining a simple RC in advance was not feasible and would have carried a significant risk of biasing the system toward an artificial pathway. Therefore, to avoid such bias, we chose to perform direct, unbiased molecular dynamics simulations. Using a physiologically relevant high-salt concentration (300 mM) for Pyrococcus yayanosii was a strategy to accelerate the system's natural dynamics, allowing us to observe these unbiased trajectories within a feasible computational timescale.

Because our current work has elucidated the fundamental steps of this mechanism, we agree that this work provides a foundation for more quantitative analyses. As suggested, future studies using methods like Markov State Model analysis or enhanced sampling techniques, guided by more sophisticated RCs defined from the insights of this work, would be a valuable next step for characterizing the free-energy landscape of the process or longer time scale dynamics.

(4) Only a small fraction of the simulated trajectories complete successful translocation (e.g., 45 of 770 in one set), and this is attributed to insufficient simulation time. While the authors are transparent about this, it raises questions about the reliability of inferred success rates and about possible artefacts (e.g., DNA trapping in coiled-coil arms). Could the authors explore or at least discuss whether alternative sampling strategies (e.g., Markov State Models, transition path sampling) might address this limitation more systematically?

We thank the reviewer for raising this point that is crucial for considering limitations and future directions of our study.

As we noted in a previous response, the primary reason we did not employ such enhanced sampling methods was the limited prior knowledge available to define previously uncharacterized DNA translocation process. Therefore, we first try to define the key conformational states and transitions without the potential bias of a predefined model or reaction coordinate. This approach was successful, as it allowed us to identify critical on-pathway states like “DNA segment capture” and significant off-pathway or kinetically trapped states such as 'DNA trapping' between the coiled-coil arms.

We fully agree that the low success rate observed is a key finding that points to significant kinetic bottlenecks, and that a more systematic analysis is required. Having identified the essential states, applying techniques such as Markov State Models (MSMs) or transition path sampling represents a powerful and logical next step. These methods, using a state-space definition based on our findings, will enable a quantitative characterization of the free-energy landscape and the transition rates between states. This will provide a rigorous understanding of the kinetic factors, such as the depth of the trapped-state energy well, that underlie the low translocation efficiency.

In the revised manuscript, we discuss the application of these advanced sampling methods as a feasible and promising future direction, which is as follows:

“Future studies can leverage the insights from this work to overcome the current timescale limitations. Techniques such as Markov state modeling (Husic and Pande, 2018; Prinz et al., 2011) or enhanced sampling methods (Hénin et al., 2022) may be employed to quantitatively characterize the free-energy landscape and transition rates. Such an approach would provide a rigorous understanding of the kinetic barriers, such as the stability of the trapped state, that govern the efficiency of SMC translocation.”

Reviewer #1 (Recommendations for the authors):

As noted in the public review, there could be a more systematic description of the limits of the model. The model appears to be carefully crafted, though every model has limits. It could be helpful for the general readership to give some idea of which parametric choices are more critical, and which mechanistic features should be robust to minor changes in parameters.

We sincerely thank the reviewer for this constructive comment. We agree that clarifying which aspects of our model is robust and sensitive to specific parameter choices is crucial for the reader's understanding.

We have expanded the Discussion to clarify how specific simulation parameters affect the efficiency and success rate of DNA translocation in our coarse-grained simulations. In particular, we have added a description of the parametric choices for (i) selection and strength of hydrogen bonds, (ii) ionic strength, and (iii) interaction strength between the coiled-coil arms. The discussion can be found in subsection entitled “Parametric choice and robustness of simulation model” in the Discussion, which is as follows:

“On the other hand, the efficiency and success rate of DNA translocation in our simulations are more sensitive to certain parametric choices. For instance, the selection and strength of hydrogen bond-like interactions are a key factor. Our model incorporates specific hydrogen bonds between the upper surface of the ATPase heads and DNA, based on all-atom simulations. These interactions are essential for initiating segment capture; without them, DNA fails to migrate to the correct binding surface. While the identification of these key residues is a robust finding—persisting across different all-atom force fields (Fig. S2)—their strength and number in the coarse-grained potential are critical parameters that directly influence the probability and kinetics of DNA capture. Another critical parameter is the ionic strength. We performed translocation simulations at an ionic strength of 300 mM to accelerate DNA dynamics. At lower concentrations, non-specific electrostatic interactions between DNA and positively charged patches on the sides of the ATPase heads or coiled-coil arm became dominant, hindering the efficient migration of DNA to its functional binding site. Using a higher-than-physiological ionic strength is a justified practice in coarse-grained simulations employing the Debye-Hückel approximation, as it serves as a first-order correction to mimic the strong local charge screening by condensed counterions that is not explicitly captured by the mean-field model (Brandani et al., 2021; Niina et al., 2017b). Finaly, the interaction strength between the coiled-coil arms is also important. In our model, once the arms closed during the transition from the V-shaped to the disengaged state, they remained closed on the simulated timescale, frequently trapping DNA pushed from the hinge and thereby leading to failed translocation. This behavior suggests that the arm–arm interactions may be overestimated. A parameterization that allows for more frequent, transient opening of the arms could increase the success rate of DNA pumping.”

Reviewer #2 (Recommendations for the authors):

This paper reports simulations (all atom and coarse grained) to provide molecular details of loop extrusion. In general, it is a well done paper. There are a few issues that the authors should address.

(1) The study supposes that loop extrusion occurs by translocation. Although they point out alternate models like scrunching (C Dekker; the theory by Takaki is also based on the scrunching model that the authors should mention), they should discuss this further. After all, the Takaki theory does predict several experimental outcomes very accurately. The precise mechanism has not been nailed down - The paper by Terakawa in Science suggests the extrusion is by translocation, but the evidence is not clear.

We thank the reviewer for this insightful comment. We agree that our discussion should briefly acknowledge alternative models such as scrunching. We have therefore revised the manuscript to mention the theory by Takaki et al. (Nat. Commun., 2021), which reproduces several experimental outcomes.

Because our present work specifically addresses the translocation mechanism based on DNA segment capture, we now state that scrunching and related models represent alternative proposals for loop extrusion.

In this revision, we have added discussion to the end of the subsection titled "DNA segment capture as the mechanism of the DNA translocation by SMC complexes." in the Discussion section, which is as follows:

“Turning to loop extrusion mechanisms, alternative mechanisms have been proposed in addition to the DNA-segment capture model. For example, Takaki et al. developed a scrunching-based theory that quantitatively accounts for several experimental observations, including force-velocity relationships and step-size distributions. While our present study focuses on the DNA translocation mechanism via segment capture, it is important to note that scrunching and other models remain plausible alternatives for loop extrusion. The precise mechanism may depends on the specific SMC complex and their subunits and remains to be fully resolved.”

(2) It is unclear how one can say from Figure 4I and J that translocation has taken place. These panels show that the base pair length increases. This should be explained more clearly. They should also simultaneously plot the location of the heads (2D plot).

Thank you for this valuable suggestion. In response to the comment on how translocation is presented in Fig. 4I and J, we have revised the text to make it clear that the SMC complex moves along DNA in subsection entitled “DNA translocation via DNA-segment capture”, as follows:

“Fig. 4I represents the one-dimensional contour coordinate of the DNA molecule, indexed by base pairs (1-800). In this plot, translocation is visualized as a discontinuous shift in the range of base-pair indices that the SMC complex contacts over one complete ATP cycle”

“This translocation is recorded in Fig. 4I as the average coordinate of the kleisin contact region (red dots) jumps from ~400 bp before the cycle to ~600bp after, which corresponds to a translocation event of ~200 bp”

We believe that adding this explanation makes it clearer to readers that Fig. 4I and 4J provide direct evidence for unidirectional translocation of the SMC complex.

(3) The transitions between the states are very abrupt (see Figure 2). Please explain. Also, in which state does extrusion take place? What is the role of the V-shape - is it part of the ATPase cycle?

We thank the reviewer for raising these questions.

In our simulation, we implemented ATP-binding state change by instantaneously switching the structure-based (Gō-type) potential between reference conformations for the disengaged (apo), engaged (ATP-bound), and V-shaped (ADP-bound) states at predetermined times. The system rapidly relaxes along the new funnel-shaped potential energy surface toward its minimum. This rapid relaxation is why the transition appears abrupt in metrics such as the Q-score in Fig.2.

The V-shaped state corresponds to a key ADP-bound intermediate within the ATP hydrolysis cycle. Its primary role in our model is preparatory; it establishes the necessary open geometry that allows for the subsequent "zipping" of the coiled-coil arms. Crucially, unidirectional pumping motion is generated during the transition from the V-shaped state to the disengaged state. That is, the zipping motion of the coiled-coil arm pushes the captured DNA segment forward, resulting in a net translocation along the DNA.

(4) It appears the heads do not move between the disengaged to engaged states. Why not in their model?

Thank you for pointing out the lack of clarity in explanation of the SMC head movement in our simulations.

In our model, the transition from the disengaged to the engaged state involves a dynamic rearrangement of the SMC heads. Specifically, one ATPase head slides (~10 Å) and rotates (~85°) relative to the other ATPase head to re-associate at a new dimer interface. This movement drives the global conformational change of the complex from a rod-like shape to an open ring, a mechanism proposed in a previous structural study (Diebold-Durand et al., Mol. Cell, 2017).

As reviewer 2 noted, this crucial motion, which is reflected in the changing head-head distance and hinge angle in Fig. 2A, was not sufficiently highlighted in the text. We have therefore revised the manuscript to explicitly describe this head rearrangement to improve clarity, which is as follows:

“Upon transition to the engaged state, the two ATPase heads were quickly rearranged to form the new inter-subunit contacts. Specifically, this rearrangement involves one ATPase head sliding by approximately 10 Å and rotating by 85° relative to the other, allowing it to associate through a different interface (Diebold-Durand et al., 2017b). The fractions of formed contacts, Q-scores, that exist at the disengaged (engaged) states quickly decreased (increased) (Fig. 2A, top two plots).”

(5) What is pumping - it has been used in Marko NAR in the DNA capture model. How is that illustrated in the simulations?

We thank the reviewer for raising this point. In the context of the DNA segment-capture model by Marko et al. (NAR, 2019), "pumping" refers to the conceptual process where a DNA loop, captured in an upper compartment of the SMC ring, is transferred to a lower compartment, resulting in net translocation.

Our simulations provide a direct, molecular-resolution visualization of the physical mechanism underlying this concept. We illustrate that the "pumping" action is not a passive transfer but an active, mechanical process driven by a specific conformational change. This occurs during the transition from the V-shaped (ADP-bound) to the disengaged state. As shown in our trajectories, the two coiled-coil arms close in a zipper-like manner, beginning from the hinge and progressing toward the ATPase heads. This zipping motion physically pushes the captured DNA segment from the hinge region toward the kleisin ring.

This process is visualized in our simulations as a clear, unidirectional translocation step (see Figs. 4B–D, 4I, and S6). The result is a net forward movement of the DNA by a distance that corresponds to the length of the initially captured loop, a key prediction of the Marko’s model that we quantify in our step-size analysis (Figs. 4K–L and S8).

To make this point clearer for the reader, we have revised the manuscript. We have explicitly defined this "zipping and pushing" action as the physical basis for the "pumping" mechanism in the subsection titled "Zipping motion of coiled-coil arms pushes the DNA from hinge domain toward kleisin ring", which is as follows:.

“This active, mechanical pushing of the DNA loop, driven by the sequential closing of the coiled-coil arm, constitutes the physical basis of the “pumping” mechanism that drives unidirectional translocation. Our simulations thus provide a concrete, molecular-level visualization for this key step in the DNA segment-capture model.”

(6) The length of DNA simulated is small for understandable reasons. Both experiments and theory show that loop extrusion sizes can be very large, far exceeding the sizes of the SMA complex. Could the small size of DNA be affecting the results?

We thank the reviewer for this important comment. The relationship between our simulated system size and the large-scale phenomena observed experimentally is a key point.

Our study was specifically designed to elucidate the fundamental mechanism of the elementary, single-cycle translocation step at near-atomic resolution. For this purpose, the 800 bp DNA length was sufficient. The observed translocation step size per cycle was 216 ± 71 bp, which is substantially smaller than the total length of the simulated DNA. This confirms that the boundaries of our system did not artificially constrain the core translocation process we aimed to investigate. Therefore, we think that the DNA length used in this study did not systematically bias our main findings regarding the motor mechanism itself.

As the reviewer pointed out, on the other hand, our current setup cannot reproduce the formation of kilobase-scale loops. We hypothesize that these large-scale events are intrinsically linked to the stochastic nature of the ATP hydrolysis cycle, which was simplified in our simulation model. We used fixed durations for each state for computational feasibility. In a more realistic scenario, a stochastically prolonged engaged state would provide a larger duration time for a captured DNA loop to grow via thermal diffusion. This could lead to occasional, much larger translocation steps upon ATP hydrolysis, contributing to the large loop sizes seen experimentally.

(7) Minor point: The first CG model using three sites was introduced in PNAS vol 102, 6789 2005. The authors should consider citing it.

Thank you for this suggestion. We have now cited the paper the reviewer recommended. Please find subsection entitled Coarse-grained simulations in Materials and Methods.

Reviewer #2 (Public review):

Summary:

This work by Clarke, Rittershofer, and colleagues used categorization and discrimination tasks with subjective reports of task regularities. In three behavioral experiments, they found that these subjective reports explain task accuracy and response times at least as well and sometimes better than objective measures. They conclude that subjective experience may play a role in predicting processing.

Strengths:

This set of behavioral studies addresses an important question. The results are replicated three times with a different experimental design, which strengthens the claims. The design is preregistered, which further strengthens the results. The findings could inspire many studies in decision-making.

Weaknesses:

It seems to me that it is important, but difficult to distinguish whether the objective and subjective measures stem from reasonably different mechanisms contributing to behavior, or whether they are simply two noisy proxies to the same mechanism, in which case it is not so surprising that both contribute to the explained variance. The authors acknowledge in the discussion that the type of objective measure that is chosen is crucial.

For instance, the RW model's learning rates were not fitted to participants but to the sequence of stimuli, so they represent the optimal parameter values, not the true ones that participants are using. Is the subjective measure just a readout of the RW model's true state when using the participants' parameters? Relatedly, would the authors consider the RW predictions from participants using a sub-optimal alpha to be a subjective or an objective measure? Do the results truly show the importance of subjective measures, or is it another way of saying that humans are sub-optimal (Rahnev & Denison, 2018, BBS) ... or optimal for other goals. I see the difficulty of avoiding double-dipping on accuracy, but this seems essential to address. This relates to a more general question about the underlying mechanisms of subjective versus objective measures, which is alluded to in the discussion but could be interesting to develop a bit further.

In terms of methods, I did not fully understand the 'RW model expectedness' objective metric in Experiments 2 and 3. VT is defined as the 'model's expectation for the given tone T. A (signed?) prediction error is defined for the expectation update, but it seems that the RW model expectedness used in the figures and statistical models is VT, sign-inverted for unexpected stimuli. So how do we interpret negative values, and how often do they occur? Shouldn't it be the unsigned value that is taken as objective surprise? This could be explained in a bit more detail. Could this be related to the quadratic effect that one can see in Figures 4E and 5E, which is not taken into account in the statistical model? Figures 4A and 5A also seem to show a combination of linear and quadratic effects. A more complete description of the objective measure could help determine whether this is a serious issue or just noise in the data.

Gabor patches in Experiments 2 and 3 seemed to have been presented at quite a sharp contrast (I did not find this info), and accuracy seems to saturate at 100%. What was the distribution of error rates, i.e., how many participants were so close to 100% that there was no point in including them in the analysis?

In the second preregistration, the authors announced that BIC comparisons between the full model and the objective model will test whether subjective measures capture additional variance [...] beyond objective prediction error. This is also the conclusion reached in sections 3.3 and 4.3. The model comparison, however, is performed by selecting the best of three models, excluding the null model. It seems that the full model still wins over the objective model, but sometimes quite marginally. Could the authors not test the significance of the model comparison since models are nested?

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

(1) As part of getting rid of cross-contamination in the bulk data, the authors model the scRNA data, extrapolate it to the bulk data and subtract out "contaminant" cell types. One wonders, however, given that low expressed genes are not represented in the scRNA data, whether the assignment of a gene to one or another cell type can really be made definitive. Indeed, it's possible that a gene is expressed at low levels in one cell, and high levels in another, and would therefore be considered a contaminant. The result would be to throw out genes that actually are expressed in a given cell type. The definitive list would therefore be a conservative estimate, and not necessarily the correct estimate.

We agree that the various strategies we employ do not result in perfect annotation of gene expression. However, despite their limitations, they are significantly better than either the single cell or the bulk data alone. We represent these strengths and shortcomings throughout the manuscript (for example, in ROC curves).

(2) It would be quite useful to have tested some genes with lower expression levels using in vivo gene-fusion reporters to assess whether the expression assignments hold up as predicted. i.e. provide another avenue of experimentation, non-computational, to confirm that the decontamination algorithm works.

We agree that evaluating only highly-expressed genes might introduce bias. We used a large battery of in vivo reporters, made with best-available technology (CRISPR insertion of the fluorophore into the endogenous locus) to evaluate our approaches. These reporters were constructed without bias in terms of gene expression and therefore represent both high and low expression levels. These data are represented throughout the manuscript (for example, in ROC curves). Details about the battery of reporters may be found in Taylor et al 2021. In addition to these reporters, this manuscript also generates and analyzes two other types of gene sets: non-neuronal and ubiquitous genes. Again, these genes are selected without bias toward gene expression, and the techniques presented here are benchmarked against them as well, with positive results.

(3) In many cases, each cell class would be composed of at least 2 if not more neurons. Is it possible that differences between members of a single class would be missed by applying the cleanup algorithms? Such transcripts would be represented only in a fraction of the cells isolated by scRNAseq, and might then be considered not real.

For the data set presented in this manuscript, all cells of a single neuron type were labeled and isolated together by FACS, and sequencing libraries were constructed from this pool of cells. Thus, potential subtypes within a particular type (when that type includes more than one cell) cannot be resolved by this method. By contrast, such subtypes were in some cases resolved in the single cell approach. To make the two data sets compatible with each other, for the single cell data we combined any subtypes together. We state in the Methods:

“For this work, single cell clusters of neuron subtypes were collapsed to the resolution of the bulk replicates (example: VB and VB1 clusters in the single cell data were treated as one VB cluster).”

(4) I didn't quite catch whether the precise staging of animals was matched between the bulk and scRNAseq datasets. Importantly, there are many genes whose expression is highly stage-specific or age-specific so even slight temporal differences might yield different sets of gene expression.

We agree that accurate staging is critically important for valid comparisons between data sets and have included an additional supplemental table with staging metadata for each sample. The staging protocol used for the bulk data set was initially employed to generate scRNA seq data and should be comparable. An additional description of our approach is now included in Methods:

“Populations of synchronized L1s were grown at 23 C until reaching the L4 stage on 150 mM 8P plates inoculated with Na22. The time in culture to reach the L4 stage varied (40.5-49 h) and was determined for each strain. 50-100 animals were inspected with a 40X DIC objective to determine developmental stage as scored by vulval morphology (Mok et al., 2025). Cultures were predominantly composed of L4 larvae but also typically included varying fractions of L3 larvae and adults.”

We have also updated supplementary table 1 to include additional information about each sort including the observed developmental stages and their proportions when available, the temperature the worms were grown at, the genotype of each experiment, and the number of cells collected in FACS.

(5) To what extent does FACS sorting affect gene expression? Can the authors provide some controls?

We appreciate this suggestion. We agree that FACS sorting (and also dissociation of the animals prior to sorting) might affect gene expression, particularly of stress-related transcripts. We note that dissociation and FACS sorting was also used to collect cells for our single cell data set (Taylor et al 2021). We would note that clean controls for this approach can be prohibitively difficult to collect, as the process of dissociation and FACS will inevitably change the proportion of cell types present in the sample, and for bulk sequencing efforts it is difficult even with deconvolution approaches to accurately account for changes in gene expression that result from dissociation and FACS, versus changes in gene expression that result from differences in cell type composition. We regrettably omitted a discussion of these issues in the manuscript. We now write in the Results:

“The dissociation and FACS steps used to isolate neuron types induce cellular stress responsive pathways (van den Brink et al., 2017; Kaletsky et al., 2016, Taylor 2021). Genes associated with this stress response (Taylor 2021) were not removed from downstream analyses, but should be viewed with caution.”

Reviewer #2 (Public review):

The bulk RNA-seq data collected by the authors has high levels of contamination and, in some cases, is based on very few cells. The methodology to remove contamination partly makes up for this shortcoming, but the high background levels of contaminating RNA in the FACS-isolated neurons limit the confidence in cell-specific transcripts.

We agree that these are the limitations of the source data. One of the manuscript’s main goals is to analyze and refine these source data, reducing these limitations and quantifying the results.

The study does not experimentally validate any of the refined gene expression predictions, which was one of the main strengths of the initial CenGEN publication (Taylor et al, 2021). No validation experiments (e.g., fluorescence reporters or single molecule FISH) were performed for protein-coding or non-coding genes, which makes it difficult for the reader to assess how much gene predictions are improved, other than for the gold standard set, which may have specific characteristics (e.g., bias toward high expression as they were primarily identified in fluorescence reporter experiments).

We agree that evaluating only highly-expressed genes might introduce bias. We used a large battery of in vivo reporters, made with best-available technology (CRISPR insertion of the fluorophore into the endogenous locus) to evaluate our approaches. These reporters were constructed without bias in terms of gene expression and therefore represent both high and low expression levels. These data are represented throughout the manuscript (for example, in ROC curves). Details about the battery of reporters may be found in Taylor et al 2021. In addition to these reporters, this manuscript also generates and analyzes two other types of gene sets: non-neuronal and ubiquitous genes. Again, these genes are selected without bias toward gene expression, and the techniques presented here are benchmarked against them as well, with positive results.

The study notes that bulk RNA-seq data, in contrast to scRNA-seq data, can be used to identify which isoforms are expressed in a given cell. However, no analysis or genome browser tracks were supplied in the study to take advantage of this important information. For the community, isoform-specific expression could guide the design of cell-specific expression constructs or for predictive modeling of gene expression based on machine learning.

We strongly agree that these datasets allow for new discoveries in neuronal splicing patterns and regulators, which is explored further in other publications from our group and other research groups in the field. We did not sufficiently highlight these works in the body of our text, and have added a reference in the discussion. “In addition, the bulk RNA-seq dataset contains transcript information across the gene body, which parallel efforts have used to identify mRNA splicing patterns that are not found in the scRNA-seq dataset.” These works can be found in references 26 and 27.

(1) The study relies on thresholding to determine whether a gene is expressed or not. While this is a common practice, the choice of threshold is not thoroughly justified. In particular, the choice of two uniform cutoffs across protein-encoding RNAs and of one distinct threshold for non-coding RNAs is somewhat arbitrary and has several limitations. This reviewer recommends the authors attempt to use adaptive threshold-methods that define gene expression thresholds on a per-gene basis. Some of these methods include GiniClust2, Brennecke's variance modeling, HVG in Seurat, BASiCS, and/or MAST Hurdle model for dropout correction.

We appreciate the reviewer’s suggestion, and would note that the integrated data currently incorporates some gene-specific weighting to identify gene expression patterns, as the single-cell data are weighted by maximum expression for each gene prior to integration with the LittleBites cleaned data. This gene level normalization markedly improved gene detection accuracy, and is discussed in depth in our 2021 Paper “Molecular topography of an entire nervous system”. We previously explored several methods for setting gene specific thresholds for identifying gene expression patterns in the integrated dataset. Unfortunately we found that none of the tested methods out performed setting “static” thresholds across all genes in the integrated dataset, and tended to increase false positive rates for some low abundance genes, where gene-specific thresholding can tend towards calling a gene expressed in at least one cell type when it is actually not expressed in any cell types present. These methods are likely to provide better results for expanded datasets that cover all tissue types (where one might reasonably expect that a gene is likely to be expressed in at least one sample).

(2) Most importantly, the study lacks independent experimental validation (e.g., qPCR, smFISH, or in situ hybridization) to confirm the expression of newly detected lowly expressed genes and non-coding RNAs. This is particularly important for validating novel neuronal non-coding RNAs, which are primarily inferred from computational approaches.

We agree that smFISH and related in situ validation methods would be an asset in this analysis. Unfortunately because most ncRNAs are very short, they are prohibitively difficult to accurately measure with smFISH. Many ncRNAs we attempted to assay with smFISH methods can bind at most 3 fluorescent probes, which unfortunately was not reliably distinguishable from background autofluorescence in the worm. Many published methods for smFISH signal amplification have not been optimized for C. elegans, and the tough cuticle is a major barrier for those efforts.

(3) The novel biology is somewhat limited. One potential area of exploration would be to look at cell-type specific alternative splicing events.

We appreciate this suggestion. Indeed, as we put our source data online prior to publishing this manuscript, two published papers already use this source data set to analyze alternative splicing. Further, these works include validation of splicing patterns observed in this source data, indicating the biological relevance of these data sets.

(4) The integration method disproportionately benefits neuron types with limited representation in scRNA-seq, meaning well-sampled neuron types may not show significant improvement. The authors should quantify the impact of this bias on the final dataset.

We agree that cell-types that are well represented in the single-cell dataset tend to have fewer new genes identified in the Integrated dataset than “rare” cell-types in the single cell data. However we would note that cell-types that are highly abundant in the single-cell data appear to become increasingly vulnerable to non-neuronal false positives, and that integration’s primary effect in high abundance cell-types appears to be reducing the false positive rate for non-neuronal genes. Thus we suggest that integration benefits all cell-types across the spectrum of single-cell abundance. The false positives are likely caused by a side-effect of normalization steps in the single-cell dataset, which is moderated by using the LittleBites cleaned bulk samples as an orthogonal measurement. The benefit of integration for cell-types with low abundance in the single-cell dataset is now quantified, and the benefits of integration for low and high abundance cell-types from the single cell data are described in the following section (p.13):

“To test the stability of LittleBites cleanup across different single-cell reference dataset qualities, we ran the algorithm on a set of bulk samples by first subsetting the corresponding single-cell cluster’s population to 10, 50, 100, or 500 cells. We performed this process 500 times for each subsampling rate for each sample (2000 total runs per sample). We found that testing gene AUROC values are stable across reference cluster sizes (Fig. 2D), suggesting that even if the target cell type is rarely represented in a single cell reference, accurate cleaning is still possible. However, comparing gene level stability across target cluster population levels reveals that low abundance references have higher gene level variance (Fig. 2E), lower purity estimates (Fig. S2F), higher variance in the mean expression across genes (Fig. S2G), and they tend to have lower overall expression (suggesting more aggressive subtraction) (Fig. S2H). This indicates that while binary gene calling is improved even if the reference cluster is small, users should be cautious when using fewer than 100 cells in their single cell reference cluster as the resulting cleanup is less stable.”

(5) The authors employ a logit transformation to model single-cell proportions into count space, but they need to clarify its assumptions and potential pitfalls (e.g., how it handles rare cell types).

We agree that the assumptions and pitfalls of the logit model are key for evaluating its usefulness, especially for cell types that are rarely captured in the single-cell dataset. The assumptions and pitfalls are described in the methods section, but we regretfully omitted any mention of those pitfalls in the results, which we have now rectified.

The description in the methods section is: “We applied this formula to our real single cell dataset and used this equation to transform proportion measures of gene expression into a count space to generate the Prop2Count dataset for downstream analysis and integration with bulk datasets. This procedure allows for proportions data to be used in downstream analyses that work with counts datasets. However, it does limit the range of potential values that each gene can have, with the potential values set as:

As n approaches 0, the number of potential values decreases, which can be incompatible with some downstream models. Thus, caution should be used when applying this transformation to datasets with few cells.”

The new mention in the results is: “However, caution should be taken when using this approach in scRNAseq cases where all replicates of a cell type contain few cells. scProp2Count values are limited to the space of possible proportion values, and so replicates with low numbers of cells will have fewer potential expression “levels” which may break some model assumptions in downstream applications (see Methods).”

(6) The LittleBites approach is highly dependent on the accuracy of existing single-cell references. If the scRNA-seq dataset is incomplete or contains classification biases, this could propagate errors into the bulk RNA-seq data. The authors may want to discuss potential limitations and sensitivity to errors in the single-cell dataset, and it is critical to define minimum quality parameters (e.g. via modeling) for the scRNAseq dataset used as reference.

We appreciate this suggestion, and agree that manuscript would benefit from a description of where the LittleBites method can give poor results. To this end, we subset our single cell reference for individual neurons of interest to the level of 10, 50, 100, or 500 cells (500 iterations per sample rate), and then ran Littlebites, and compared metrics of gene expression stability, sample composition estimates, and AUROC performance on test genes. We found that when fewer than 100 cells for the target cell type are included in the single cell reference, gene expression stability drops (variance between subsampling iterations was much higher when fewer reference cells were used). However, we found that AUROC values were consistently high regardless of how many reference cells were included, but that this stability in AUROC values was paired with lower overall counts in samples with <100 reference cells after cleanup. This indicates that in cases where few reference cells are present, higher AUROC values might be achieved by more aggressive subtraction, which is attenuated when the reference model is more complete. This analysis is shown in figure 2 and figure S2, and described in the results section, recreated here.

“To test the stability of Littlebites cleanup across different single-cell reference dataset qualities, we ran the algorithm on a set of bulk samples by first subsetting the corresponding single-cell cluster’s population to 10, 50, 100, or 500 cells. We performed this process 500 times for each subsampling rate for each sample (2000 total runs per sample). We found that testing gene AUROC values are stable across reference cluster sizes (Fig. 2D), suggesting that even if the target cell type is rarely represented in a single cell reference, accurate cleaning is still possible. However, comparing gene level stability across target cluster population levels reveals that low population references have higher gene level variance (Fig. 2E), lower purity estimates (Fig. S2F), higher variance in the mean expression across genes (Fig. S2G), and they tend to have lower overall expression (suggesting more aggressive subtraction) (Fig. S2H). This suggests that while binary gene calling is improved similarly even if the reference cluster is small, users should be cautious when using less than 100 cells in their single cell reference cluster as the resulting cleanup is less stable.”

(7) Also very important, the LittleBites method could benefit from a more intuitive explanation and schematic to improve accessibility for non-computational readers. A supplementary step-by-step breakdown of the subtraction process would be useful.

We appreciate this suggestion and implemented a step-by-steo breakdown of the subtraction process in the methods section, also copied below. We also updated the graphic representation in figure 2A.

“LittleBites Subtraction algorithm

LittleBites is an iterative algorithm for bulk RNA-seq datasets, that improves the accuracy of cell-type specific bulk RNA-seq sample profiles by removing counts from non-target contaminants (e.g. ambient RNA from dead cells, carry-over non-target cells from FACS enrichment due to imperfect gating). This method leverages single cell reference datasets and ground truth expression information to guide iterative and conservative subtraction to enrich for true target cell-type expression. Using this approach, LittleBites balances subtraction by optimizing using both a single-cell reference, and an orthogonal ground truth reference, moderating biases inherent to either reference.

This algorithm first calculates gene level specificity weights in a single cell reference dataset using SPM (Specificity Preservation Method) (re-add 22, re-add 23). SPM assigns high weights (approaching 1) to genes expressed in single cell types while applying conservative weights to genes with broader expression patterns, which helps to reduce inappropriate subtraction.

The algorithm proceeds in a loop of three steps:

Step 1: Estimate Contamination. Each bulk sample is modeled as the sum of a linear combination of single-cell profiles (target cell type and likely contaminants) using non-negative least squares (NNLS). The resulting coefficients provide the estimate of how much of the sample’s counts come from the target cell-type, and how much comes from each contaminant cell-type.

Step 2: Weighted Subtraction. Each bulk sample is cleaned by subtracting the weighted sum of contaminant single-cell profiles. This subtraction is attempted multiple times (separately) across a series of learning rate weights (usually ranging from 0-1) which moderate the size of the subtraction step (Equation 1). This produces a range of possible “cleaned” sample options for evaluation.

Step 3: Performance Optimization. For each learning rate, the cleaned result is evaluated against a set of ground truth genes by calculating the area under the receiver operating characteristic curve (AUROC). The learning rate that optimizes the AUROC is then selected. When multiple learning rates yielded equivalent AUROC values, the lowest learning rate value is chosen to minimize subtraction.

If the optimal learning rate at Step 3 is 0 (no subtraction option beats the baseline) then the loop is halted. Else, the cleaned bulk profile is returned to Step 1, and the loop continues until the AUROC cannot be improved upon using the single-cell reference modeling.“

(8) In the same vein, the ROC curves and AUROC comparisons should have clearer annotations to make results more interpretable for readers unfamiliar with these metrics.

We agree that the ROC and AUROC metrics need a clearer explanation to make their use and interpretations clearer. We included a description of both metrics, and a suggestion for how to interpret them in the results section, copied below.

“To evaluate the post-subtraction datasets accuracy we used the area under the Receiver-Operator Characteristic (AUROC) score. In brief, we set a wide range of thresholds to call genes expressed or unexpressed, and then compared it to expected expression from a set of ground truth genes. This comparison produces a true positive rate (TPR, the percentage of truly expressed genes that are called expressed), and false positive rate (FPR, the percentage of truly not expressed genes that are called expressed), and a false discovery rate (FDR, the percentage of genes called expressed that are truly not expressed). The Receiver-Operator Characteristic (ROC) is the graph of the line produced by the TPR and FPR values across the range of thresholds tested, and the AUROC is calculated as the sum of the area under that line. A “random” model of gene expression is expected to have an AUROC value of 0.5, and a “perfect” model is expected to have an AUROC value of 1. Thus, AUROCs below 0.5 are worse than a random guess, and values closer to 1 indicate higher accuracy.”

(9) Finally, after the correlation-based decontamination of the 4,440 'unexpressed' genes, how many were ultimately discarded as non-neuronal?

a) Among these non-neuronal genes, how many were actually known neuronal genes or components of neuronal pathways (e.g., genes involved in serotonin synthesis, synaptic function, or axon guidance)?

b) Conversely, among the "unexpressed" genes classified as neuronal, how many were likely not neuron-specific (e.g., housekeeping genes) or even clearly non-neuronal (e.g., myosin or other muscle-specific markers)?

Combined with point 10, see below.

(10) To increase transparency and allow readers to probe false positives and false negatives, I suggest the inclusion of:

a) The full list of all 4,440 'unexpressed' genes and their classification at each refinement step. In that list flag the subsets of genes potentially misclassified, including:

- Neuronal genes wrongly discarded as non-neuronal.

- Non-neuronal genes wrongly retained as neuronal.

b) Add a certainty or likelihood ranking that quantifies confidence in each classification decision, helping readers validate neuronal vs. non-neuronal RNA assignments.

This addition would enhance transparency, reproducibility, and community engagement, ensuring that key neuronal genes are not erroneously discarded while minimizing false positives from contaminant-derived transcripts.

We agree that the genes called “unexpressed” in the single-cell data need more context and clarity. First, we trimmed the list to only include 2,333 genes of highest confidence. Second, for those genes we identified any with published neuronal expression patterns. Identifying genes that were retained as neuronal but are likely non-neuronal in origin is difficult as many markers are expressed in a mixture of neuronal and non-neuronal cell-types, however we used a curated list of putative non-neuronal markers to assess the accuracy of the integrated data (see supplementary table 4), and established that most non-neuronal markers are undetected in the integrated data, with the number of detected genes decreasing as our threshold stringency increases. Of note, a few putative non-neuronal genes remain detected even at high thresholds, indicating that our dataset still retains a small percentage of neuronal false positives. This result has been collected in the new supplementary figure 4F, and addressed in the following text in the results section “Testing against a curated list of non-neuronal genes from fluorescent reporters and genomic enrichment studies, we found that of 445 non-neuronal markers, each gene was detected in an average of 12.5 cells or a median of 3 cells in the single-cell dataset, and an average of 8.7 cells or a median of 1 cell in the integrated dataset, at a 14% FDR threshold.”

We also included a list of “unexpressed” gene identities and tissue annotations as new supplementary tables 16 and 17.

Reviewer #2 (Recommendations for the authors):

The utility of the bulk RNA-seq data would be significantly increased if the authors were to analyze which isoforms are expressed in individual neurons. Also, it would be very useful to know if there are instances where a gene is expressed in several neurons, but different isoforms are specific to individual neurons.

We appreciate this suggestion. Indeed, as we put our source data online prior to publishing this manuscript, two published papers already use this source data set to analyze alternative splicing. Further, these works include validation of splicing patterns observed in this source data, indicating the biological relevance of these data sets. This is now noted in our discussion section “In addition, the bulk RNA-seq dataset contains transcript information across the gene body, which parallel efforts have used to identify mRNA splicing patterns that are not found in the scRNA-seq dataset.” These works can be found in references 26 and 27.

Reviewer #3 (Recommendations for the authors):

(1) Describe the number of L4 animals processed to obtain good-quality bulk RNAseq libraries from the different neuronal types. If the number of worms would be different for different neuronal types, then please make a supplementary table listing the minimum number of worms needed for each neuronal type.

We appreciate the reviewer’s recommendation, and agree that it would be a useful resource to provide suggestions for how many worms are needed per experiment. Unfortunately We did not track the total number of animals for each sample. We aimed to start with 200-300 ul of packed worms for each strain, generally equating to >500,000 worms, but yields of FACS-isolated cells varied among cell types. Because replicates for specific neuron types were also variable in some instances (See additions to supplemental Table 1), yields likely depend on multiple factors. We have previously noted (Taylor et al., 2021), for example, that some cell types were under-represented in scRNA-seq data (e.g, pharyngeal neurons) based on in vivo abundance presumptively due to the difficulty of isolation or sub-viability in the cell dissociation-FACS protocol.

(2) List the thresholds for the parameters used during the FASTQC quality control and the threshold number of reads that would make a sample not useful.

We now include parameters for sample exclusion in the methods section. “Samples were excluded after sequencing if they had: fewer than 1 million read pairs, <1% of uniquely mapping reads to the C. elegans genome, > 50% duplicate reads (low umi diversity), or failed deduplication steps in the nudup package.”

(3) In Figure 5C, include an overlapping bar that shows the total number of genes in each cell type. You may need to use a log scale to see both (new and all) numbers of genes in the same graph. Add supplementary tables with the names of all new genes assigned to each neuronal type.

We agree that this figure panel needed additional context. On further reflection we concluded that figure 5 was not sufficiently distinct from figure 4 to warrant separation, and incorporated some key findings from figure 5 into figure S4.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #1 (Public review):

Weaknesses:

(1) Weaknesses of this study include a proposed mechanism underlying the sexual dimorphism phenotype based on experimentation in only males, and widespread reliance on over-expression when investigating protein-protein interaction and localization. Additionally, a minor weakness is that the text describing the identification of cyp17a2 as a candidate contains errors that are confusing.

We thank the reviewer for these insightful comments, which have helped us improve the manuscript.

(1) Experimentation in males. We focused on male zebrafish for our mechanistic studies to preclude potential confounding effects from female hormones and to directly interrogate the basis of the observed male-biased resistance. As confirmed in the manuscript (lines 151-153), both wild-type and cyp17a2⁻/⁻ males developed normal male sex organs and exhibited comparable androgen levels. This crucial control gives us confidence that the differences in antiviral immunity we observed are a direct consequence of Cyp17a2 loss-of-function, rather than secondary to developmental or hormonal abnormalities. We fully agree that elucidating the mechanism in females represents a valuable and interesting direction for future research.

(2) Over-expression studies. We acknowledge that overexpression approaches can have inherent limitations. To mitigate this and strengthen our conclusions, we complemented these experiments with loss-of-function data from both knockout zebrafish and knockdown cells, as well as validation at the endogenous level (e.g., Fig. 4J and S4C). The consistent results obtained across these diverse experimental models collectively reinforce our conclusion that Cyp17a2 interacts with and stabilizes STING.

(3) We thank the reviewer for pointing out the lack of clarity in the text regarding the selection process of Cyp17a2. We have thoroughly revised the manuscript to provide a precise and accurate description of our methodology. The relevant text is now as follows: “Differential expression analysis identified 1511 upregulated and 1117 downregulated genes (Fig. 2A and Table S2). We then focused on a subset of known or putative sexrelated genes. Among these eight candidates, cyp17a2 exhibited the most significant male-biased upregulation, a finding that was subsequently confirmed by qPCR (Fig. 2B and S1A)” (lines 142-144).

(2) Lines 139-140 describe the data for Figure 2 as deriving from "healthy hermaphroditic adult zebrafish". This appears to be a language error and should be corrected to something that specifies that the comparison made is between healthy adult male and female kidneys.

We thank the reviewer for pointing out this inaccuracy. This was a terminological error, and we have corrected the text to accurately state “transcriptome sequencing was performed on head-kidney tissues from healthy adult male and female zebrafish” (lines 139-140). We have carefully reviewed the manuscript to ensure no similar errors are present.

(3) In Figure 2A and associated text cyp17a2 is highlighted but the volcano plot does not indicate why this was an obvious choice. For example, many other genes are also highly induced in male vs female kidneys. Figure 2B and line 143 describe a subset of "eight sex-related genes" but it is not clear how these relate to Figure 2A. The narrative could be improved to clarify how cyp17a2 was selected from Figure 2A and it seems that the authors made an attempt to do this with Figure 2B but it is not clear how these are related. This is important because the available data do not rule out the possibility that other factors also mediate the sexual dimorphism they observed either in combination, in a redundant fashion, or in a more complex genetic fashion. The narrative of the text and title suggests that they consider this to be a monogenic trait but more evidence is needed.

We thank the reviewer for raising these important points. We have revised the manuscript to clarify the candidate gene selection process and to avoid any implication that the trait is monogenic.

The selection of cyp17a2 was not based solely on its position in the volcano plot (Fig. 2A), but on a multi-faceted rationale. We first prioritized genes with known or putative sex-related functions from the pool of differentially expressed genes. From this subset, cyp17a2 emerged as the lead candidate due to a combination of unique attributes, it exhibited the most significant and consistent male-biased upregulation among the validated candidates (Fig. 2B and S1A); it is a teleost-specific autosomal gene, suggesting a novel mechanism for sexual dimorphism independent of canonical sex chromosomes; and it showed conserved male-biased expression across multiple tissues (Fig. 2C and 2D). Regarding its representation in the volcano plot, cyp17a2 was included in the underlying dataset but was not explicitly labeled in the revised Figure 2A to maintain visual clarity, as the plot aimed to illustrate the global transcriptomic landscape rather than highlight individual genes.

We agree with the reviewer that other genetic factors may contribute to the observed sexual dimorphism. Accordingly, we have modified the text throughout the manuscript to remove any suggestion of a purely monogenic trait. Our functional data position cyp17a2 as a key and sufficient factor, as its knockout in males was sufficient to ablate the antiviral resistance phenotype (Fig. 2E-G), demonstrating a major, nonredundant role without precluding potential contributions from other genes.

The following specific changes have been made to the text.

(1) The title has been revised by replacing “governs” with “orchestrates.” (line 1)  

(2) The abstract now states “the male-biased gene cyp17a2 as a critical mediator of this enhanced response” instead of “which are driven by the male-biased gene Cyp17a2 rather than by hormones or sex chromosomes.” (lines 33-34)

(3) The discussion now states “Our study leverages this unique context to demonstrate that enhanced antiviral immunity in males is mediated by the male-biased expression of the autosomal gene cyp17a2,” removing the comparative phrasing regarding hormones or sex chromosomes. (lines 364-366)

Reviewer #1 (Public review):

Summary:

This study employed a saccade-shifting sequential working memory paradigm, manipulating whether a saccade occurred after each memory array to directly compare retinotopic and transsaccadic working memory for both spatial location and color. Across four participant groups (young and older healthy adults, and patients with Parkinson's disease and Alzheimer's disease), the authors found a consistent saccade-related cost specifically for spatial memory - but not for color - regardless of differences in memory precision. Using computational modeling, they demonstrate that data from healthy participants are best explained by a complex saccade-based updating model that incorporates distractor interference. Applying this model to the patient groups further elucidates the sources of spatial memory deficits in PD and AD. The authors then extend the model to explain copying deficits in these patient groups, providing evidence for the ecological validity of the proposed saccade-updating retinotopic mechanism.

Strengths:

Overall, the manuscript is well written, and the experimental design is both novel and appropriate for addressing the authors' key research questions. I found the study to be particularly comprehensive: it first characterizes saccade-related costs in healthy young adults, then replicates these findings in healthy older adults, demonstrating how this "remapping" cost in spatial working memory is age-independent. After establishing and validating the best-fitting model using data from both healthy groups, the authors apply this model to clinical populations to identify potential mechanisms underlying their spatial memory impairments. The computational modeling results offer a clearer framework for interpreting ambiguities between allocentric and retinotopic spatial representations, providing valuable insight into how the brain represents and updates visual information across saccades. Moreover, the findings from the older adult and patient groups highlight factors that may contribute to spatial working memory deficits in aging and neurological disease, underscoring the broader translational significance of this work.

Weaknesses:

Several concerns should be addressed to enhance the clarity of the manuscript:

(1) Relevance of the figure-copy results (pp. 13-15).

Is it necessary to include the figure-copy task results within the main text? The manuscript already presents a clear and coherent narrative without this section. The figure-copy task represents a substantial shift from the LOCUS paradigm to an entirely different task that does not measure the same construct. Moreover, the ROCF findings are not fully consistent with the LOCUS results, which introduces confusion and weakens the manuscript's coherence. While I understand the authors' intention to assess the ecological validity of their model, this section does not effectively strengthen the manuscript and may be better removed or placed in the Supplementary Materials.

(2) Model fitting across age groups (p. 9).

It is unclear whether it is appropriate to fit healthy young and healthy elderly participants' data to the same model simultaneously. If the goal of the model fitting is to account for behavioral performance across all conditions, combining these groups may be problematic, as the groups differ significantly in overall performance despite showing similar remapping costs. This suggests that model performance might differ meaningfully between age groups. For example, in Figure 4A, participants 22-42 (presumably the elderly group) show the best fit for the Dual (Saccade) model, implying that the Interference component may contribute less to explaining elderly performance.

Furthermore, although the most complex model emerges as the best-fitting model, the manuscript should explain how model complexity is penalized or balanced in the model comparison procedure. Additionally, are Fixation Decay and Saccade Update necessarily alternative mechanisms? Could both contribute simultaneously to spatial memory representation? A model that includes both mechanisms-e.g., Dual (Fixation) + Dual (Saccade) + Interference-could be tested to determine whether it outperforms Model 7 to rule out the sole contribution of complexity.

Minor point: On p. 9, line 336, Figure 4A does not appear to include the red dashed vertical line that is mentioned as separating the age groups.

(3) Clarification of conceptual terminology.

Some conceptual distinctions are unclear. For example, the relationship between "retinal memory" and "transsaccadic memory," as well as between "allocentric map" and "retinotopic representation," is not fully explained. Are these constructs related or distinct? Additionally, the manuscript uses terms such as "allocentric map," "retinotopic representation," and "reference frame" interchangeably, which creates ambiguity. It would be helpful for the authors to clarify the relationships among these terms and apply them consistently.

(4) Rationale for the selective disruption hypothesis (p. 4, lines 153-154).

The authors hypothesize that "saccades would selectively disrupt location memory while leaving colour memory intact." Providing theoretical or empirical justification for this prediction would strengthen the argument.

(5) Relationship between saccade cost and individual memory performance (p. 4, last paragraph).

The authors report that larger saccades were associated with greater spatial memory disruption. It would be informative to examine whether individual differences in the magnitude of saccade cost correlate with participants' overall/baseline memory performance (e.g. their memory precision in the no-saccade condition). Such analyses might offer insights into how memory capacity/ability relates to resilience against saccade-induced updating.

(6) Model fitting for the healthy elderly group to reveal memory-deficit factors (pp. 11-12).

The manuscript discusses model-based insights into components that contribute to spatial memory deficits in AD and PD, but does not discuss components that contribute to spatial memory deficits in the healthy elderly group. Given that the EC group also shows impairments in certain parameters, explaining and discussing these outcomes of the EC group could provide additional insights into age-related memory decline, which would strengthen the study's broader conclusions.

(7) Presentation of saccade conditions in Figure 5 (p. 11).

In Figure 5, it may be clearer to group the four saccade conditions together within each patient group. Since the main point is that saccadic interference on spatial memory remains robust across patient groups, grouping conditions by patient type rather than intermixing conditions would emphasize this interpretation.

DOI: 10.1016/j.cub.2025.09.008

Resource: RRID:BDSC_70746

Curator: @maulamb

SciCrunch record: RRID:BDSC_70746

What is this?

DOI: 10.1038/s41467-025-65708-w

Resource: RRID:BDSC_52884

Curator: @scibot

SciCrunch record: RRID:BDSC_52884

What is this?

DOI: 10.1038/s41467-025-65708-w

Resource: RRID:BDSC_67760

Curator: @scibot

SciCrunch record: RRID:BDSC_67760

What is this?

Reviewer #1 (Public review):

Summary:

The authors present a novel investigation of the movement vigor of individuals completing a synchronous extension-flexion task. Participants were placed into groups of two (so-called "dyads") and asked to complete shared movements (connected via a virtual loaded spring) to targets placed at varying amplitudes. The authors attempted to quantify what, if any, adjustments in movement vigor individual participants made during the dyadic movements, given the combined or co-dependent nature of the task. This is a novel, timely question of interest within the broader field of human sensorimotor control.

Participants from each dyad were labeled as "slow" (low vigor) or "fast" (high vigor), and their respective contributions to the combined movement metrics were assessed. The authors presented four candidate models for dyad interactions: (a) independent motor plans (i.e., co-activity hypothesis), (b) individual-led motor plans (i.e., leader-follower hypothesis), (c) generalization to a weighted average motor plan (i.e., weighted adaptation hypothesis), and (d) an uncertainty-based model of dynamic partner-partner interaction (i.e., interactive adaptation hypothesis). The final model allowed for dynamic changes in individual motor plans (and therefore, movement vigor) based on partner-partner interactions and observations. After detailed observations of interaction torque and movement duration (or vigor), the authors concluded that the interactive adaptation model provided the best explanation of human-human interaction during self-paced dyadic movements.

Strengths:

The experimental setup (simultaneous wrist extension-flexion movements) has been thoroughly vetted. The task was designed particularly well, with adequate block pseudo-randomization to ensure general validity of the results. The analyses of torque interaction, movement kinematics, and vigor are sound, as are the statistical measures used to assess significance. The authors structured the work via a helpful comparison of several candidate models of human-human interaction dynamics, and how well said models explained variance in the vigor of solo and combined movements. The research question is timely and extends current neuroscientific understanding of sensorimotor control, particularly in social contexts.

Weaknesses:

(1) My chief concern about the study as it currently stands is the relatively low number of data points (n=10). The authors recruited 20 participants, but the primary conclusions are based on dyad-specific interactions (i.e., analyses of "fast" vs "slow" participants in each pair). Some of these analyses would benefit greatly, in terms of power, from the addition of more data points.

1a) The distribution of delta-vigor (Fast group vs Slow group) is highly skewed (see Figures 3D, S6D), with over half of the dyads exhibiting delta-vigor less than 0.2 (i.e., less than 20% of unit vigor). Given the relatively low number of dyads, it would be helpful for the authors to provide explicit listings of VigorFast, VigorSlow, and VigorCombined for each of the 10 separate dyads or pairings.

1b) The authors concluded that the interactive adaptation hypothesis provided the best summary of the combined movement dynamics in the study. If this is indeed the case, then the relative degree of difference in vigor between the fast and slow participants in a dyad should matter. How well did the interactive adaptation model explain variance in the dyads with relatively low delta-vigor (e.g., less than 0.2) vs relatively high delta-vigor?

(2) The authors shared the results of one analysis of reaction time, showing that the reaction times of the slow partners and the fast partners did not differ during the initial passive block. Did the authors observe any changes in RT of either the slow or fast partner during the combined (primary task) blocks (KL, KH, etc.)? If the pairs of participants did indeed employ a form of interactive adaptation, then it is certainly plausible that this interaction would manifest in the initial movement planning phase (i.e., RT) in addition to the vigor and smoothness of the movements themselves.

la referencias en Vancouver no van por orden de aparición?

aqui faltan materias tubos cornig, ácidos, los clínex ceca, varias coas que mencionas en la metodología

para la discusión hay valor de este en MX??

Creo que esto debería ser sintetizado dentro del párrafo anterior en tanto se vincula aclarando por qué es importante estudiar factores meritocráticos en una sociedad con tales características. Porque acá creo que queda un poco descontextuallizado. Entiendo que antes ya habíamos hablado de que no convencía tanto este párrafo.

Esto tiene que ser centrado en la escuela

Muy buena reflexión. Me recuerda a cosas que decía mi compañero de LIGO, y creo que de la app https://www.envirohero.org/

Creo que entre medio de este párrafo y el que sigue hay que3 señalar que las autoeficacias tienen un efecto distinto con CIL consistente entre olas y entre paises. Esto porque es el hallazgo más importante y sólido, y queda bien con el relato. De lo contrario no se cacha el por qué de esta separación. El último parrafo lo dedicaría exclusivamente a las relaciones que tienen los constructos con otras variables: NSE, genero, etc etc. De este modo tampoco se mezclan con esta relación que es la principal para este tipo de estudios.

¿De dónde sacaste esta información? No encontre algo en el informe. No mencionaría la teoría o corriente de Bandura acá, pues ni siquiera la presentamos en el apartado pasado. Además después se define de la misma forma que ya se definión arriba. Mejor no reiterar. PRopuesta:

ICILS grounds on elements of Bandura (1993) view of self-efficacy to propose a measurement of how students judge their capabilities to execute courses of action with Informational and Computation Technologies (fraillon_iea_2025a).

Author response:

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

Public reviews:

Reviewer #1 (Public review):

Summary:

Wu and colleagues aimed to explain previous findings that adolescents, compared to adults, show reduced cooperation following cooperative behaviour from a partner in several social scenarios. The authors analysed behavioural data from adolescents and adults performing a zero-sum Prisoner's Dilemma task and compared a range of social and non-social reinforcement learning models to identify potential algorithmic differences. Their findings suggest that adolescents' lower cooperation is best explained by a reduced learning rate for cooperative outcomes, rather than differences in prior expectations about the cooperativeness of a partner. The authors situate their results within the broader literature, proposing that adolescents' behaviour reflects a stronger preference for self-interest rather than a deficit in mentalising.

Strengths:

The work as a whole suggests that, in line with past work, adolescents prioritise value accumulation, and this can be, in part, explained by algorithmic differences in weighted value learning. The authors situate their work very clearly in past literature, and make it obvious the gap they are testing and trying to explain. The work also includes social contexts that move the field beyond non-social value accumulation in adolescents. The authors compare a series of formal approaches that might explain the results and establish generative and modelcomparison procedures to demonstrate the validity of their winning model and individual parameters. The writing was clear, and the presentation of the results was logical and well-structured.

We thank the reviewer for recognizing the strengths of our work.

Weaknesses:

(1) I also have some concerns about the methods used to fit and approximate parameters of interest. Namely, the use of maximum likelihood versus hierarchical methods to fit models on an individual level, which may reduce some of the outliers noted in the supplement, and also may improve model identifiability.

We thank the reviewer for this suggestion. Following the comment, we added a hierarchical Bayesian estimation. We built a hierarchical model with both group-level (adolescent group and adult group) and individual-level structures for the best-fitting model. Four Markov chains with 4,000 samples each were run, and the model converged well (see Figure supplement 7).

We then analyzed the posterior parameters for adolescents and adults separately. The results were consistent with those from the MLE analysis. These additional results have been included in the Appendix Analysis section (also see Figure supplement 5 and 7). In addition, we have updated the code and provided the link for reference. We appreciate the reviewer’s suggestion, which improved our analysis.

(2) There was also little discussion given the structure of the Prisoner's Dilemma, and the strategy of the game (that defection is always dominant), meaning that the preferences of the adolescents cannot necessarily be distinguished from the incentives of the game, i.e. they may seem less cooperative simply because they want to play the dominant strategy, rather than a lower preferences for cooperation if all else was the same.

We thank the reviewer for this comment and agree that adolescents’ lower cooperation may partly reflect a rational response to the incentive structure of the Prisoner’s Dilemma. 

However, our computational modeling explicitly addressed this possibility. Model 4 (inequality aversion) captures decisions that are driven purely by self-interest or aversion to unequal outcomes, including a parameter reflecting disutility from advantageous inequality, which represents self-oriented motives. If participants’ behavior were solely guided by the payoff-dominant strategy, this model should have provided the best fit. However, our model comparison showed that Model 5 (social reward) performed better in both adolescents and adults, suggesting that cooperative behavior is better explained by valuing social outcomes beyond payoff structures.

Besides, if adolescents’ lower cooperation is that they strategically respond to the payoff structure by adopting defection as the more rewarding option. Then, adolescents should show reduced cooperation across all rounds. Instead, adolescents and adults behaved similarly when partners defected, but adolescents cooperated less when partners cooperated and showed little increase in cooperation even after consecutive cooperative responses. This pattern suggests that adolescents’ lower cooperation cannot be explained solely by strategic responses to payoff structures but rather reflects a reduced sensitivity to others’ cooperative behavior or weaker social reciprocity motives. We have expanded our Discussion to acknowledge this important point and to clarify how the behavioral and modeling results address the reviewer’s concern.

“Overall, these findings indicate that adolescents’ lower cooperation is unlikely to be driven solely by strategic considerations, but may instead reflect differences in the valuation of others’ cooperation or reduced motivation to reciprocate. Although defection is the payoff-dominant strategy in the Prisoner’s Dilemma, the selective pattern of adolescents’ cooperation and the model comparison results indicate that their reduced cooperation cannot be fully explained by strategic incentives, but rather reflects weaker valuation of social reciprocity.”

Appraisal & Discussion:

(3) The authors have partially achieved their aims, but I believe the manuscript would benefit from additional methodological clarification, specifically regarding the use of hierarchical model fitting and the inclusion of Bayes Factors, to more robustly support their conclusions. It would also be important to investigate the source of the model confusion observed in two of their models.

We thank the reviewer for this comment. In the revised manuscript, we have clarified the hierarchical Bayesian modeling procedure for the best-fitting model, including the group- and individual-level structure and convergence diagnostics. The hierarchical approach produced results that fully replicated those obtained from the original maximumlikelihood estimation, confirming the robustness of our findings. Please also see the response to (1).

Regarding the model confusion between the inequality aversion (Model 4) and social reward (Model 5) models in the model recovery analysis, both models’ simulated behaviors were best captured by the baseline model. This pattern arises because neither model includes learning or updating processes. Given that our task involves dynamic, multi-round interactions, models lacking a learning mechanism cannot adequately capture participants’ trial-by-trial adjustments, resulting in similar behavioral patterns that are better explained by the baseline model during model recovery. We have added a clarification of this point to the Results:

“The overlap between Models 4 and 5 likely arises because neither model incorporates a learning mechanism, making them less able to account for trial-by-trial adjustments in this dynamic task.”

(4) I am unconvinced by the claim that failures in mentalising have been empirically ruled out, even though I am theoretically inclined to believe that adolescents can mentalise using the same procedures as adults. While reinforcement learning models are useful for identifying biases in learning weights, they do not directly capture formal representations of others' mental states. Greater clarity on this point is needed in the discussion, or a toning down of this language.

We sincerely thank the reviewer for this professional comment. We agree that our prior wording regarding adolescents’ capacity to mentalise was somewhat overgeneralized. Accordingly, we have toned down the language in both the Abstract and the Discussion to better align our statements with what the present study directly tests. Specifically, our revisions focus on adolescents’ and adults’ ability to predict others’ cooperation in social learning. This is consistent with the evidence from our analyses examining adolescents’ and adults’ model-based expectations and self-reported scores on partner cooperativeness (see Figure 4). In the revised Discussion, we state:

“Our results suggest that the lower levels of cooperation observed in adolescents stem from a stronger motive to prioritize self-interest rather than a deficiency in predicting others’ cooperation in social learning”.

(5) Additionally, a more detailed discussion of the incentives embedded in the Prisoner's Dilemma task would be valuable. In particular, the authors' interpretation of reduced adolescent cooperativeness might be reconsidered in light of the zero-sum nature of the game, which differs from broader conceptualisations of cooperation in contexts where defection is not structurally incentivised.

We thank the reviewer for this comment and agree that adolescents’ lower cooperation may partly reflect a rational response to the incentive structure of the Prisoner’s Dilemma. However, our behavioral and computational evidence suggests that this pattern cannot be explained solely by strategic responses to payoff structures, but rather reflects a reduced sensitivity to others’ cooperative behavior or weaker social reciprocity motives. We have expanded the Discussion to acknowledge this point and to clarify how both behavioral and modeling results address the reviewer’s concern (see also our response to 2).

(6) Overall, I believe this work has the potential to make a meaningful contribution to the field. Its impact would be strengthened by more rigorous modelling checks and fitting procedures, as well as by framing the findings in terms of the specific game-theoretic context, rather than general cooperation.

We thank the reviewer for the professional comments, which have helped us improve our work.

Reviewer #2 (Public review):

Summary:

This manuscript investigates age-related differences in cooperative behavior by comparing adolescents and adults in a repeated Prisoner's Dilemma Game (rPDG). The authors find that adolescents exhibit lower levels of cooperation than adults. Specifically, adolescents reciprocate partners' cooperation to a lesser degree than adults do. Through computational modeling, they show that this relatively low cooperation rate is not due to impaired expectations or mentalizing deficits, but rather a diminished intrinsic reward for reciprocity. A social reinforcement learning model with asymmetric learning rate best captured these dynamics, revealing age-related differences in how positive and negative outcomes drive behavioral updates. These findings contribute to understanding the developmental trajectory of cooperation and highlight adolescence as a period marked by heightened sensitivity to immediate rewards at the expense of long-term prosocial gains.

Strengths:

(1) Rigid model comparison and parameter recovery procedure.

(2) Conceptually comprehensive model space.

(3) Well-powered samples.

We thank the reviewer for highlighting the strengths of our work.

Weaknesses:

A key conceptual distinction between learning from non-human agents (e.g., bandit machines) and human partners is that the latter are typically assumed to possess stable behavioral dispositions or moral traits. When a non-human source abruptly shifts behavior (e.g., from 80% to 20% reward), learners may simply update their expectations. In contrast, a sudden behavioral shift by a previously cooperative human partner can prompt higher-order inferences about the partner's trustworthiness or the integrity of the experimental setup (e.g., whether the partner is truly interactive or human). The authors may consider whether their modeling framework captures such higher-order social inferences. Specifically, trait-based models-such as those explored in Hackel et al. (2015, Nature Neuroscience)-suggest that learners form enduring beliefs about others' moral dispositions, which then modulate trial-bytrial learning. A learner who believes their partner is inherently cooperative may update less in response to a surprising defection, effectively showing a trait-based dampening of learning rate.

We thank the reviewer for this thoughtful comment. We agree that social learning from human partners may involve higher-order inferences beyond simple reinforcement learning from non-human sources. To address this, we had previously included such mechanisms in our behavioral modeling. In Model 7 (Social Reward Model with Influence), we tested a higher-order belief-updating process in which participants’ expectations about their partner’s cooperation were shaped not only by the partner’s previous choices but also by the inferred influence of their own past actions on the partner’s subsequent behavior. In other words, participants could adjust their belief about the partner’s cooperation by considering how their partner’s belief about them might change. Model comparison showed that Model 7 did not outperform the best-fitting model, suggesting that incorporating higher-order influence updates added limited explanatory value in this context. As suggested by the reviewer, we have further clarified this point in the revised manuscript.

Regarding trait-based frameworks, we appreciate the reviewer’s reference to Hackel et al. (2015). That study elegantly demonstrated that learners form relatively stable beliefs about others’ social dispositions, such as generosity, especially when the task structure provides explicit cues for trait inference (e.g., resource allocations and giving proportions). By contrast, our study was not designed to isolate trait learning, but rather to capture how participants update their expectations about a partner’s cooperation over repeated interactions. In this sense, cooperativeness in our framework can be viewed as a trait-like latent belief that evolves as evidence accumulates. Thus, while our model does not include a dedicated trait module that directly modulates learning rates, the belief-updating component of our best-fitting model effectively tracks a dynamic, partner-specific cooperativeness, potentially reflecting a prosocial tendency.

This asymmetry in belief updating has been observed in prior work (e.g., Siegel et al., 2018, Nature Human Behaviour) and could be captured using a dynamic or belief-weighted learning rate. Models incorporating such mechanisms (e.g., dynamic learning rate models as in Jian Li et al., 2011, Nature Neuroscience) could better account for flexible adjustments in response to surprising behavior, particularly in the social domain.

We thank the reviewer for the suggestion. Following the comment, we implemented an additional model incorporating a dynamic learning rate based on the magnitude of prediction errors. Specifically, we developed Model 9:  Social reward model with Pearce–Hall learning algorithm (dynamic learning rate), in which participants’ beliefs about their partner’s cooperation probability are updated using a Rescorla–Wagner rule with a learning rate dynamically modulated by the Pearce–Hall (PH) Error Learning mechanism. In this framework, the learning rate increases following surprising outcomes (larger prediction errors) and decreases as expectations become more stable (see Appendix Analysis section for details).

The results showed that this dynamic learning rate model did not outperform our bestfitting model in either adolescents or adults (see Figure supplement 6). We greatly appreciate the reviewer’s suggestion, which has strengthened the scope of our analysis. We now have added these analyses to the Appendix Analysis section (see Figure Supplement 6) and expanded the Discussion to acknowledge this modeling extension and further discuss its implications.

Second, the developmental interpretation of the observed effects would be strengthened by considering possible non-linear relationships between age and model parameters. For instance, certain cognitive or affective traits relevant to social learning-such as sensitivity to reciprocity or reward updating-may follow non-monotonic trajectories, peaking in late adolescence or early adulthood. Fitting age as a continuous variable, possibly with quadratic or spline terms, may yield more nuanced developmental insights.

We thank the reviewer for this professional comment. In addition to the linear analyses, we further conducted exploratory analyses to examine potential non-linear relationships between age and the model parameters. Specifically, we fit LMMs for each of the four parameters as outcomes (α+, α-, β, and ω). The fixed effects included age, a quadratic age term, and gender, and the random effects included subject-specific random intercepts and random slopes for age and gender. Model comparison using BIC did not indicate improvement for the quadratic models over the linear models for α⁺ (ΔBIC_{quadratic-linear} = 5.09), α− (ΔBICquadratic-linear = 3.04), β (ΔBICquadratic-linear = 3.9), or ω (ΔBICquadratic-linear = 0). Moreover, the quadratic age term was not significant for α⁺, α⁻, or β (all ps > 0.10). For ω, we observed a significant linear age effect (b = 1.41, t = 2.65, p = 0.009) and a significant quadratic age effect (b = −0.03, t = −2.39, p = 0.018; see Author response image 1). This pattern is broadly consistent with the group effect reported in the main text. The shaded area in the figure represents the 95% confidence interval. As shown, the interval widens at older ages (≥ 26 years) due to fewer participants in that range, which limits the robustness of the inferred quadratic effect. In consideration of the limited precision at older ages and the lack of BIC improvement, we did not emphasize the quadratic effect in the revised manuscript and present these results here as exploratory.

Author response image 1.

Linear and quadratic model fits showing the relationship between age and the ω parameter, with 95% confidence intervals.<br />

Finally, the two age groups compared - adolescents (high school students) and adults (university students) - differ not only in age but also in sociocultural and economic backgrounds. High school students are likely more homogenous in regional background (e.g., Beijing locals), while university students may be drawn from a broader geographic and socioeconomic pool. Additionally, differences in financial independence, family structure (e.g., single-child status), and social network complexity may systematically affect cooperative behavior and valuation of rewards. Although these factors are difficult to control fully, the authors should more explicitly address the extent to which their findings reflect biological development versus social and contextual influences.

We appreciate this comment. Indeed, adolescents (high school students) and adults (university students) differ not only in age but also in sociocultural and socioeconomic backgrounds. In our study, all participants were recruited from Beijing and surrounding regions, which helps minimize large regional and cultural variability. Moreover, we accounted for individual-level random effects and included participants’ social value orientation (SVO) as an individual difference measure. 

Nonetheless, we acknowledge that other contextual factors, such as differences in financial independence, socioeconomic status, and social experience—may also contribute to group differences in cooperative behavior and reward valuation. Although our results are broadly consistent with developmental theories of reward sensitivity and social decisionmaking, sociocultural influences cannot be entirely ruled out. Future work with more demographically matched samples or with socioeconomic and regional variables explicitly controlled will help clarify the relative contributions of biological and contextual factors. Accordingly, we have revised the Discussion to include the following statement:  “Third, although both age groups were recruited from Beijing and nearby regions, minimizing major regional and cultural variation, adolescents and adults may still differ in socioeconomic status, financial independence, and social experience. Such contextual differences could interact with developmental processes in shaping cooperative behavior and reward valuation. Future research with demographically matched samples or explicit measures of socioeconomic background will help disentangle biological from sociocultural influences.”

Reviewer #3 (Public review):

Summary:

Wu and colleagues find that in a repeated Prisoner's Dilemma, adolescents, compared to adults, are less likely to increase their cooperation behavior in response to repeated cooperation from a simulated partner. In contrast, after repeated defection by the partner, both age groups show comparable behavior.

To uncover the mechanisms underlying these patterns, the authors compare eight different models. They report that a social reward learning model, which includes separate learning rates for positive and negative prediction errors, best fits the behavior of both groups. Key parameters in this winning model vary with age: notably, the intrinsic value of cooperating is lower in adolescents. Adults and adolescents also differ in learning rates for positive and negative prediction errors, as well as in the inverse temperature parameter.

Strengths: 

The modeling results are compelling in their ability to distinguish between learned expectations and the intrinsic value of cooperation. The authors skillfully compare relevant models to demonstrate which mechanisms drive cooperation behavior in the two age groups.

We thank the reviewer’s recognition of our work’s strengths.

Weaknesses:

Some of the claims made are not fully supported by the data:

The central parameter reflecting preference for cooperation is positive in both groups. Thus, framing the results as self-interest versus other-interest may be misleading.

We thank the reviewer for this insightful comment. In the social reward model, the cooperation preference parameter is positive by definition, as defection in the repeated rPDG always yields a +2 monetary advantage regardless of the partner’s action. This positive value represents the additional subjective reward assigned to mutual cooperation (e.g., reciprocity value) that counterbalances the monetary gain from defection. Although the estimated social reward parameter ω was positive, the effective advantage of cooperation is Δ=p×ω−2. Given participants’ inferred beliefs p, Δ was negative for most trials (p×ω<2), indicating that the social reward was insufficient to offset the +2 advantage of defection. Thus, both adolescents and adults valued cooperation positively, but adolescents’ smaller ω and weaker responsiveness to sustained partner cooperation suggest a stronger weighting on immediate monetary payoffs. 

In this light, our framing of adolescents as more self-interested derives from their behavioral pattern: even when they recognized sustained partner cooperation and held high expectations of partner cooperation, adolescents showed lower cooperative behavior and reciprocity rewards compared with adults. Whereas adults increased cooperation after two or three consecutive partner cooperations, this pattern was absent among adolescents. We therefore interpret their behavior as relatively more self-interested, reflecting reduced sensitivity to the social reward from mutual cooperation rather than a categorical shift from self-interest to other-interest, as elaborated in the Discussion.

It is unclear why the authors assume adolescents and adults have the same expectations about the partner's cooperation, yet simultaneously demonstrate age-related differences in learning about the partner. To support their claim mechanistically, simulations showing that differences in cooperation preference (i.e., the w parameter), rather than differences in learning, drive behavioral differences would be helpful.

We thank the reviewer for raising this important point. In our model, both adolescents and adults updated their beliefs about partner cooperation using an asymmetric reinforcement learning (RL) rule. Although adolescents exhibited a higher positive and a lower negative learning rate than adults, the two groups did not differ significantly in their overall updating of partner cooperation probability (Fig. 4a-b). We then examined the social reward parameter ω, which was significantly smaller in adolescents and determined the intrinsic value of mutual cooperation (i.e., p×ω). This variable differed significantly between groups and closely matched the behavioral pattern.

Following the reviewer’s suggestion, we conducted additional simulations varying one model parameter at a time while holding the others constant. The difference in mean cooperation probability between adults and adolescents served as the index (positive = higher cooperation in adults). As shown in the Author response image 2, decreases in ω most effectively reproduced the observed group difference (shaded area), indicating that age-related differences in cooperation are primarily driven by variation in the social reward parameter ω rather than by others.

Author response image 2.

Simulation results showing how variations in each model parameter affect the group difference in mean cooperation probability (Adults – Adolescents). Based on the best-fitting Model 8 and parameters estimated from all participants, each line represents one parameter (i.e., α+, α-, ω, β) systematically varied within the tested range (α±:0.1–0.9; ω, β:1–9) while other parameters were held constant. Positive values indicate higher cooperation in adults. Smaller ω values most strongly reproduced the observed group difference, suggesting that reduced social reward weighting primarily drives adolescents’ lower cooperation.

Two different schedules of 120 trials were used: one with stable partner behavior and one with behavior changing after 20 trials. While results for order effects are reported, the results for the stable vs. changing phases within each schedule are not. Since learning is influenced by reward structure, it is important to test whether key findings hold across both phases.

We thank the reviewer for this thoughtful and professional comment. In our GLMM and LMM analyses, we focused on trial order rather than explicitly including the stable vs. changing phase factor, due to concerns about multicollinearity. In our design, phases occur in specific temporal segments, which introduces strong collinearity with trial order. In multi-round interactions, order effects also capture variance related to phase transitions. 

Nonetheless, to directly address this concern, we conducted additional robustness analyses by adding a phase variable (stable vs. changing) to GLMM1, LMM1, and LMM3 alongside the original covariates. Across these specifications, the key findings were replicated (see GLMM_sup2 and LMM_sup4–5; Tables 9-11), and the direction and significance of main effects remained unchanged, indicating that our conclusions are robust to phase differences.

The division of participants at the legal threshold of 18 years should be more explicitly justified. The age distribution appears continuous rather than clearly split. Providing rationale and including continuous analyses would clarify how groupings were determined.

We thank the reviewer for this thoughtful comment. We divided participants at the legal threshold of 18 years for both conceptual and practical reasons grounded in prior literature and policy. In many countries and regions, 18 marks the age of legal majority and is widely used as the boundary between adolescence and adulthood in behavioral and clinical research. Empirically, prior studies indicate that psychosocial maturity and executive functions approach adult levels around this age, with key cognitive capacities stabilizing in late adolescence (Icenogle et al., 2019; Tervo-Clemmens et al., 2023). We have clarified this rationale in the Introduction section of the revised manuscript.

“Based on legal criteria for majority and prior empirical work, we adopt 18 years as the boundary between adolescence and adulthood (Icenogle et al., 2019; Tervo-Clemmens et al., 2023).”

We fully agree that the underlying age distribution is continuous rather than sharply divided. To address this, we conducted additional analyses treating age as a continuous predictor (see GLMM_sup1 and LMM_sup1–3; Tables S1-S4), which generally replicated the patterns observed with the categorical grouping. Nevertheless, given the limited age range of our sample, the generalizability of these findings to fine-grained developmental differences remains constrained. Therefore, our primary analyses continue to focus on the contrast between adolescents and adults, rather than attempting to model a full developmental trajectory.

Claims of null effects (e.g., in the abstract: "adults increased their intrinsic reward for reciprocating... a pattern absent in adolescents") should be supported with appropriate statistics, such as Bayesian regression.

We thank the reviewer for highlighting the importance of rigor when interpreting potential null effects. To address this concern, we conducted Bayes factor analyses of the intrinsic reward for reciprocity and reported the corresponding BF10 for all relevant post hoc comparisons. This approach quantifies the relative evidence for the alternative versus the null hypothesis, thereby providing a more direct assessment of null effects. The analysis procedure is now described in the Methods and Materials section: 

“Post hoc comparisons were conducted using Bayes factor analyses with MATLAB’s bayesFactor Toolbox (version v3.0, Krekelberg, 2024), with a Cauchy prior scale σ = 0.707.”

Once claims are more closely aligned with the data, the study will offer a valuable contribution to the field, given its use of relevant models and a well-established paradigm.

We are grateful for the reviewer’s generous appraisal and insightful comments.

Recommendations for the authors

Reviewer #1 (Recommendations for the authors):

I commend the authors on a well-structured, clear, and interesting piece of work. I have several questions and recommendations that, if addressed, I believe will strengthen the manuscript.

We thank the reviewer for commending the organization of our paper.

Introduction: - Why use a zero-sum (Prisoner's Dilemma; PD) versus a mixed-motive game (e.g. Trust Task) to study cooperation? In a finite set of rounds, the dominant strategy can be to defect in a PD.

We thank the reviewer for this helpful comment. We agree that both the rationale for using the repeated Prisoner’s Dilemma (rPDG) and the limitations of this framework should be clarified. We chose the rPDG to isolate the core motivational conflict between selfinterest and joint welfare, as its symmetric and simultaneous structure avoids the sequential trust and reputation dependencies/accumulation inherent to asymmetric tasks such as the Trust Game (King-Casas et al., 2005; Rilling et al., 2002).

Although a finitely repeated rPDG theoretically favors defection, extensive prior research shows that cooperation can still emerge in long repeated interactions when players rely on learning and reciprocity rather than backward induction (Rilling et al., 2002; Fareri et al., 2015). Our design employed 120 consecutive rounds, allowing participants to update expectations about partner behavior and to establish stable reciprocity patterns over time. We have added the following clarification to the Introduction:

“The rPDG provides a symmetric and simultaneous framework that isolates the motivational conflict between self-interest and joint welfare, avoiding the sequential trust and reputation dynamics characteristic of asymmetric tasks such as the Trust Game (Rilling et al., 2002; King-Casas et al., 2005)”

Methods:

Did the participants know how long the PD would go on for?

Were the participants informed that the partner was real/simulated?

Were the participants informed that the partner was going to be the same for all rounds?

We thank the reviewer for the meticulous review work, which helped us present the experimental design and reporting details more clearly. the following clarifications: I. Participants were not informed of the total number of rounds in the rPDG. This prevented endgame expectations and avoided distraction from counting rounds, which could introduce additional effects. II. Participants were told that their partner was another human participant in the laboratory. However, the partner’s behavior was predetermined by a computer program. This design enabled tighter experimental control and ensured consistent conditions across age groups, supporting valid comparisons. III. Participants were informed that they would interact with the same partner across all rounds, aligning with the essence of a multiround interaction paradigm and stabilizing partner-related expectations. For transparency, we have clarified these points in the Methods and Materials section:

“Participants were told that their partner was another human participant in the laboratory and that they would interact with the same partner across all rounds. However, in reality, the actions of the partner were predetermined by a computer program. This setup allowed for a clear comparison of the behavioral responses between adolescents and adults. Participants were not informed of the total number of rounds in the rPDG.”

The authors mention that an SVO was also recorded to indicate participant prosociality. Where are the results of this? Did this track game play at all? Could cooperativeness be explained broadly as an SVO preference that penetrated into game-play behaviour?

We thank the reviewer for pointing this out. We agree that individual differences in prosociality may shape cooperative behavior, so we conducted additional analyses incorporating SVO. Specifically, we extended GLMM1 and LMM3 by adding the measured SVO as a fixed effect with random slopes, yielding GLMM_sup3 and LMM_sup6 (Tables 12–13). The results showed that higher SVO was associated with greater cooperation, whereas its effect on the reward for reciprocity was not significant. Importantly, the primary findings remained unchanged after controlling for SVO. These results indicate that cooperativeness in our task cannot be explained solely by a broad SVO preference, although a more prosocial orientation was associated with greater cooperation. We have reported these analyses and results in the Appendix Analysis section.

Why was AIC chosen rather an BIC to compare model dominance?

Sorry for the lack of clarification. Both the Akaike Information Criterion (AIC, Akaike, 1974) and Bayesian Information Criterion (BIC, Schwarz, 1978) are informationtheoretic criterions for model comparison, neither of which depends on whether the models to be compared are nested to each other or not (Burnham et al., 2002). We have added the following clarification into the Methods.

“We chose to use the AICc as the metric of goodness-of-fit for model comparison for the following statistical reasons. First, BIC is derived based on the assumption that the “true model” must be one of the models in the limited model set one compares (Burnham et al., 2002; Gelman & Shalizi, 2013), which is unrealistic in our case. In contrast, AIC does not rely on this unrealistic “true model” assumption and instead selects out the model that has the highest predictive power in the model set (Gelman et al., 2014). Second, AIC is also more robust than BIC for finite sample size (Vrieze, 2012).”

I believe the model fitting procedure might benefit from hierarchical estimation, rather than maximum likelihood methods. Adolescents in particular seem to show multiple outliers in a^+ and w^+ at the lower end of the distributions in Figure S2. There are several packages to allow hierarchical estimation and model comparison in MATLAB (which I believe is the language used for this analysis; see https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1007043).

We thank the reviewer for this helpful comment and for referring us to relevant methodological work (Piray et al., 2019). We have addressed this point by incorporating hierarchical Bayesian estimation, which effectively mitigates outlier effects and improves model identifiability. The results replicated those obtained with MLE fitting and further revealed group-level differences in key parameters. Please see our detailed response to Reviewer#1 Q1 for the full description of this analysis and results.

Results: Model confusion seems to show that the inequality aversion and social reward models were consistently confused with the baseline model. Is this explained or investigated? I could not find an explanation for this.

The apparent overlap between the inequality aversion (Model 4) and social reward (Model 5) models in the recovery analysis likely arises because neither model includes a learning mechanism, making them unable to capture trial-by-trial adjustments in this dynamic task. Consequently, both were best fit by the baseline model. Please see Response to Reviewer #1 Q3 for related discussion.

Figures 3e and 3f show the correlation between asymmetric learning rates and age. It seems that both a^+ and a^- are around 0.35-0.40 for young adolescents, and this becomes more polarised with age. Could it be that with age comes an increasing discernment of positive and negative outcomes on beliefs, and younger ages compress both positive and negative values together? Given the higher stochasticity in younger ages (\beta), it may also be that these values simply represent higher uncertainty over how to act in any given situation within a social context (assuming the differences in groups are true).

We appreciate this insightful interpretation. Indeed, both α+ and α- cluster around 0.35–0.40 in younger adolescents and become increasingly polarized with age, suggesting that sensitivity to positive versus negative feedback is less differentiated early in development and becomes more distinct over time. This interpretation remains tentative and warrants further validation. Based on this comment, we have revised the Discussion to include this developmental interpretation.

We also clarify that in our model β denotes the inverse temperature parameter; higher β reflects greater choice precision and value sensitivity, not higher stochasticity. Accordingly, adolescents showed higher β values, indicating more value-based and less exploratory choices, whereas adults displayed relatively greater exploratory cooperation. These group differences were also replicated using hierarchical Bayesian estimation (see Response to Reviewer #1 Q1). In response to this comment, we have added a statement in the Discussion highlighting this developmental interpretation.

“Together, these findings suggest that the differentiation between positive and negative learning rates changes with age, reflecting more selective feedback sensitivity in development, while higher β values in adolescents indicate greater value sensitivity. This interpretation remains tentative and requires further validation in future research.”

A parameter partial correlation matrix (off-diagonal) would be helpful to understand the relationship between parameters in both adolescents and adults separately. This may provide a good overview of how the model properties may change with age (e.g. a^+'s relation to \beta).

We thank the reviewer for this helpful comment. We fully agree that a parameter partial correlation matrix can further elucidate the relationships among parameters. Accordingly, we conducted a partial correlation analysis and added the visually presented results to the revised manuscript as Figure 2-figure supplement 4.

It would be helpful to have Bayes Factors reported with each statistical tests given that several p-values fall within the 0.01 and 0.10.

We thank the reviewer for this important recommendation. We have conducted Bayes factor analyses and reported BF10 for all relevant post hoc comparisons. We also clarified our analysis in the Methods and Materials section: 

“Post hoc comparisons were conducted using Bayes factor analyses with MATLAB’s bayesFactor Toolbox (version v3.0, Krekelberg, 2024), with a Cauchy prior scale σ = 0.707.”

Discussion: I believe the language around ruling out failures in mentalising needs to be toned down. RL models do not enable formal representational differences required to assess mentalising, but they can distinguish biases in value learning, which in itself is interesting. If the authors were to show that more complex 'ToM-like' Bayesian models were beaten by RL models across the board, and this did not differ across adults and adolescents, there would be a stronger case to make this claim. I think the authors either need to include Bayesian models in their comparison, or tone down their language on this point, and/or suggest ways in which this point might be more thoroughly investigated (e.g., using structured models on the same task and running comparisons: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0087619).

We thank the reviewer for the comments. Please see our response to Reviewer 1 (Appraisal & Discussion section) for details.

Reviewer #2 (Recommendations for the authors):

The authors may want to show the winning model earlier (perhaps near the beginning of the Results section, when model parameters are first mentioned).

We thank the reviewer for this suggestion. We agree that highlighting the winning model early improves clarity. Currently, we have mentioned the winning model before the beginning of the Results section. Specifically, in the penultimate paragraph of the Introduction we state:

“We identified the asymmetric RL learning model as the winning model that best explained the cooperative decisions of both adolescents and adults.”

Reviewer #3 (Recommendations for the authors):

In addition to the points mentioned above, I suggest the following:

(1) Clarify plots by clearly explaining each variable. In particular, the indices 1 vs. 1,2 vs. 1,2,3 were not immediately understandable.

We thank the reviewer for this suggestion. We agree that the indices were not immediately clear. We have revised the figure captions (Figure 1 and 4) to explicitly define these terms more clearly: 

“The x-axis represents the consistency of the partner’s actions in previous trials (t₋₁: last trial; t_−1,2: last two trials; t_−1,2,3: last three trials).”

It's unclear why the index stops at 3. If this isn't the maximum possible number of consecutive cooperation trials, please consider including all relevant data, as adolescents might show a trend similar to adults over more trials.

We thank the reviewer for raising this point. In our exploratory analyses, we also examined longer streaks of consecutive partner cooperation or defection (up to four or five trials). Two empirical considerations led us to set the cutoff at three in the final analyses. First, the influence of partner behavior diminished sharply with temporal distance. In both GLMMs and LMMs, coefficients for earlier partner choices were small and unstable, and their inclusion substantially increased model complexity and multicollinearity. This recency pattern is consistent with learning and decision models emphasizing stronger weighting of recent evidence (Fudenberg & Levine, 2014; Fudenberg & Peysakhovich, 2016). Second, streaks longer than three were rare, especially among some participants, leading to data sparsity and inflated uncertainty. Including these sparse conditions risked biasing group estimates rather than clarifying them. Balancing informativeness and stability, we therefore restricted the index to three consecutive partner choices in the main analyses, which we believe sufficiently capture individuals’ general tendencies in reciprocal cooperation.

The term "reciprocity" may not be necessary. Since it appears to reflect a general preference for cooperation, it may be clearer to refer to the specific behavior or parameter being measured. This would also avoid confusion, especially since adolescents do show negative reciprocity in response to repeated defection.

We thank you for this comment. In our work, we compute the intrinsic reward for reciprocity as p × ω, where p is the partner cooperation expectation and ω is the cooperation preference. In the rPDG, this value framework manifests as a reciprocity-derived reward: sustained mutual cooperation maximizes joint benefits, and the resulting choice pattern reflects a value for reciprocity, contingent on the expected cooperation of the partner. This quantity enters the trade-off between U_cooperation and U_defectionand captures the participant’s intrinsic reward for reciprocity versus the additional monetary reward payoff of defection. Therefore, we consider the term “reciprocity” an acceptable statement for this construct.

Interpretation of parameters should closely reflect what they specifically measure.

We thank the reviewer for pointing this out. We have refined the relevant interpretations of parameters in the current Results and Discussion sections.

Prior research has shown links between Theory of Mind (ToM) and cooperation (e.g., Martínez-Velázquez et al., 2024). It would be valuable to test whether this also holds in your dataset.

We thank the reviewer for this thoughtful comment. Although we did not directly measure participants’ ToM, our design allowed us to estimate participants’ trial-by-trial inferences (i.e., expectations) about their partner’s cooperation probability. We therefore treat these cooperation expectations as an indirect representation for belief inference, which is related to ToM processes. To test whether this belief-inference component relates to cooperation in our dataset, we further conducted an exploratory analysis (GLMM_sup4) in which participants’ choices were regressed on their cooperation expectations, group, and the group × cooperation-expectation interaction, controlling for trial number and gender, with random effects. Consistent with the ToM–cooperation link in prior research (MartínezVelázquez et al., 2024), participants’ expectations about their partner’s cooperation significantly predicted their cooperative behavior (Table 14), suggesting that decisions were shaped by social learning about others’ inferred actions. Moreover, the interaction between group and cooperation expectation was not significant, indicating that this inference-driven social learning process likely operates similarly in adolescents and adults. This aligns with our primary modeling results showing that both age groups update beliefs via an asymmetric learning process. We have reported these analyses in the Appendix Analysis section.

More informative table captions would help the reader. Please clarify how variables are coded (e.g., is female = 0 or 1? Is adolescent = 0 or 1?), to avoid the need to search across the manuscript for this information.

We thank the reviewer for raising this point. We have added clear and standardized variable coding in the table notes of all tables to make them more informative and avoid the need to search the paper. We have ensured consistent wording and formatting across all tables.

I hope these comments are helpful and support the authors in further strengthening their manuscript.

We thank the three reviewers for their comments, which have been helpful in strengthening this work.

Reference

(1) Fudenberg, D., & Levine, D. K. (2014). Recency, consistent learning, and Nash equilibrium. Proceedings of the National Academy of Sciences of the United States of America, 111(Suppl. 3), 10826–10829. https://doi.org/10.1073/pnas.1400987111

(2) Fudenberg, D., & Peysakhovich, A. (2016). Recency, records, and recaps: Learning and nonequilibrium behavior in a simple decision problem. ACM Transactions on Economics and Computation, 4(4), Article 23, 1–18. https://doi.org/10.1145/2956581

(3) Hackel, L., Doll, B., & Amodio, D. (2015). Instrumental learning of traits versus rewards: Dissociable neural correlates and effects on choice. Nature Neuroscience, 18, 1233– 1235. https://doi.org/10.1038/nn.4080

(4) Icenogle, G., Steinberg, L., Duell, N., Chein, J., Chang, L., Chaudhary, N., Di Giunta, L.,Dodge, K. A., Fanti, K. A., Lansford, J. E., Oburu, P., Pastorelli, C., Skinner, A. T.,Sorbring, E., Tapanya, S., Uribe Tirado, L. M., Alampay, L. P., Al-Hassan, S. M.,Takash, H. M. S., & Bacchini, D. (2019). Adolescents’ cognitive capacity reaches adult levels prior to their psychosocial maturity: Evidence for a “maturity gap” in a multinational, cross-sectional sample. Law and Human Behavior, 43(1), 69–85. https://doi.org/10.1037/lhb0000315

(5) Krekelberg, B. (2024). Matlab Toolbox for Bayes Factor Analysis (v3.0) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.13744717

(6) Martínez-Velázquez, E. S., Ponce-Juárez, S. P., Díaz Furlong, A., & Sequeira, H. (2024). Cooperative behavior in adolescents: A contribution of empathy and emotional regulation? Frontiers in Psychology, 15, 1342458. https://doi.org/10.3389/fpsyg.2024.1342458

(7) Tervo-Clemmens, B., Calabro, F. J., Parr, A. C., et al. (2023). A canonical trajectory of executive function maturation from adolescence to adulthood. NatureCommunications, 14, 6922. https://doi.org/10.1038/s41467-023-42540-8

(8) King-Casas, B., Tomlin, D., Anen, C., Camerer, C. F., Quartz, S. R., & Montague, P. R. (2005). Getting to know you: reputation and trust in a two-person economic exchange. Science, 308(5718), 78-83. https://doi.org/10.1126/science.1108062

(9) Rilling, J. K., Gutman, D. A., Zeh, T. R., Pagnoni, G., Berns, G. S., & Kilts, C. D. (2002). A neural basis for social cooperation. Neuron, 35(2), 395-405. https://doi.org/10.1016/s0896-6273(02)00755-9

(10) Fareri, D. S., Chang, L. J., & Delgado, M. R. (2015). Computational substrates of social value in interpersonal collaboration. Journal of Neuroscience, 35(21), 8170-8180. https://doi.org/10.1523/JNEUROSCI.4775-14.2015

(11) Akaike, H. (2003). A new look at the statistical model identification. IEEE transactions on automatic control, 19(6), 716-723. https://doi.org/10.1109/TAC.1974.1100705

(12) Schwarz, G. (1978). Estimating the dimension of a model. The annals of statistics, 461464. https://doi.org/10.1214/aos/1176344136

(13) Burnham, K. P., & Anderson, D. R. (2002). Model selection and multimodel inference: A practical information-theoretic approach (2nd ed.). Springer.https://doi.org/10.1007/b97636

(14) Gelman, A., & Shalizi, C. R. (2013). Philosophy and the practice of Bayesian statistics. British Journal of Mathematical and Statistical Psychology, 66(1), 8–38. https://doi.org/10.1111/j.2044-8317.2011.02037.x

(15) Gelman, A., Carlin, J. B., Stern, H. S., Dunson, D. B., Vehtari, A., & Rubin, D. B. (2014). Bayesian data analysis (3rd ed.). Chapman and Hall/CRC. https://doi.org/10.1201/b16018

(16) Vrieze, S. I. (2012). Model selection and psychological theory: A discussion of the differences between the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). Psychological Methods, 17(2), 228–243. https://doi.org/10.1037/a0027127.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary: 

Zhang and colleagues examine neural representations underlying abstract navigation in the entorhinal cortex (EC) and hippocampus (HC) using fMRI. This paper replicates a previously identified hexagonal modulation of abstract navigation vectors in abstract space in EC in a novel task involving navigating in a conceptual Greeble space. In HC, the authors claim to identify a three-fold signal of the navigation angle. They also use a novel analysis technique (spectral analysis) to look at spatial patterns in these two areas and identify phase coupling between HC and EC. Finally, the authors propose an EC-HPC PhaseSync Model to understand how the EC and HC construct cognitive maps. While the wide array of techniques used is impressive and their creativity in analysis is admirable, overall, I found the paper a bit confusing and unconvincing. I recommend a significant rewrite of their paper to motivate their methods and clarify what they actually did and why. The claim of three-fold modulation in HC, while potentially highly interesting to the community, needs more background to motivate why they did the analysis in the first place, more interpretation as to why this would emerge in biology, and more care taken to consider alternative hypotheses seeped in existing models of HC function. I think this paper does have potential to be interesting and impactful, but I would like to see these issues improved first.

General comments:

(1) Some of the terminology used does not match the terminology used in previous relevant literature (e.g., sinusoidal analysis, 1D directional domain).

We thank the reviewer for this valuable suggestion, which helps to improve the consistency of our terminology with previous literature and to reduce potential ambiguity. Accordingly, we have replaced “sinusoidal analysis” with “sinusoidal modulation” (Doeller et al., 2010; Bao et al., 2019; Raithel et al., 2023) and “1D directional domain” with “angular domain of path directions” throughout the manuscript.

(2) Throughout the paper, novel methods and ideas are introduced without adequate explanation (e.g., the spectral analysis and three-fold periodicity of HC).

We thank the reviewer for raising this important point. In the revised manuscript, we have substantially extended the Introduction (paragraphs 2–4) to clarify our hypothesis, explicitly explaining why the three primary axes of the hexagonal grid cell code may manifest as vector fields. We have also revised the first paragraph of the “3-fold periodicity in the HPC” section in the Results to clarify the rationale for using spectral analysis. Please refer to our responses to comment 2 and 3 below for details.

Reviewer #2 (Public review):

The authors report results from behavioral data, fMRI recordings, and computer simulations during a conceptual navigation task. They report 3-fold symmetry in behavioral and simulated model performance, 3-fold symmetry in hippocampal activity, and 6-fold symmetry in entorhinal activity (all as a function of movement directions in conceptual space). The analyses are thoroughly done, and the results and simulations are very interesting.

We sincerely thank the reviewer for the positive and encouraging comments on our study.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) This paper has quite a few spelling and grammatical mistakes, making it difficult to understand at times.

We apologize for the wordings and grammatical errors. We have thoroughly re-read and carefully edited the entire manuscript to correct typographical and grammatical errors, ensuring improved clarity and readability.

(2) Introduction - It's not clear why the three primary axes of hexagonal grid cell code would manifest as vector fields.

We thank the reviewer for raising this important point. In the revised Introduction (paragraphs 2, 3, and 4), we now explicitly explain the rationale behind our hypothesis that the three primary axes of the hexagonal grid cell code manifest as vector fields.

In paragraph 2, we present empirical evidence from rodent, bat, and human studies demonstrating that mental simulation of prospective paths relies on vectorial representations in the hippocampus (Sarel et al., 2017; Ormond and O’Keefe, 2022; Muhle-Karbe et al., 2023).

In paragraphs 3 and 4, we introduce our central hypothesis: vectorial representations may originate from population-level projections of entorhinal grid cell activity, based on three key considerations:

(1) The EC serves as the major source of hippocampal input (Witter and Amaral, 1991; van Groen et al., 2003; Garcia and Buffalo, 2020).

(2) Grid codes exhibit nearly invariant spatial orientations (Hafting et al., 2005; Gardner et al., 2022), which makes it plausible that their spatially periodic activity can be detected using fMRI.

(3) A model-based inference: for example, in the simplest case, when one mentally simulates a straight pathway aligned with the grid orientation, a subpopulation of grid cells would be activated. The resulting population activity would form a near-perfect vectorial representation, with constant activation strength along the path. In contrast, if the simulated path is misaligned with the grid orientation, the population response becomes a distorted vectorial code. Consequently, simulating all possible straight paths spanning 0°–360° results in 3-fold periodicity in the activity patterns—due to the 180° rotational symmetry of the hexagonal grid, orientations separated by 180° are indistinguishable.

We therefore speculate that vectorial representations embedded in grid cell activity exhibit 3-fold periodicity across spatial orientations and serve as a periodic structure to represent spatial direction. Supporting this view, reorientation paradigms in both rodents and young children have shown that subjects search equally in two opposite directions, reflecting successful orientation encoding but a failure to integrate absolute spatial direction (Hermer and Spelke, 1994; Julian et al., 2015; Gallistel, 2017; Julian et al., 2018).

(3) It took me a few reads to understand what the spectral analysis was. After understanding, I do think this is quite clever. However, this paper needs more motivation to understand why you are performing this analysis. E.g., why not just take the average regressor at the 10º, 70º, etc. bins and compare it to the average regressor at 40º, 100º bins? What does the Fourier transform buy you?

We are sorry for the confusion. we outline the rationale for employing Fast Fourier Transform (FFT) analysis to identify neural periodicity. In the revised manuscript, we have added these clarifications into the first paragraph of the “3-fold periodicity in the HPC” subsection in the Results.

First, FFT serves as an independent approach to cross-validate the sinusoidal modulation results, providing complementary evidence for the 6-fold periodicity in EC and the 3-fold periodicity in HPC.

Second, FFT enables unbiased detection of multiple candidate periodicities (e.g., 3–7-fold) simultaneously without requiring prior assumptions about spatial phase (orientation). By contrast, directly comparing “aligned” versus “misaligned” angular bins (e.g., 10°/70° vs. 40°/100°) would implicitly assume knowledge of the phase offset, which was not known a priori.

Finally, FFT uniquely allows periodicity analysis of behavioral performance, which is not feasible with standard sinusoidal GLM approaches. This methodological consistency makes it possible to directly compare periodicities across neural and behavioral domains.

(4) A more minor point: at one point, you say it’s a spectral analysis of the BOLD signals, but the methods description makes it sound like you estimated regressors at each of the bins before performing FFT. Please clarify. 

We apologize for the confusion. In our manuscript, we use the term spectral analysis to distinguish this approach from sinusoidal modulation analysis. Conceptually, our spectral analysis involves a three-level procedure:

(1) First level: We estimated direction-dependent activity maps using a general linear model (GLM), which included 36 regressors corresponding to path directions, down-sampled in 10° increments.

(2) Second level: We applied a Fast Fourier Transform (FFT) to the direction-dependent activity maps derived from the GLM to examine the spectral magnitude of potential spatial periodicities.

(3) Third level: We conducted group-level statistical analyses across participants to assess the consistency of the observed periodicities.

We have revised the “Spectral analysis of MRI BOLD signals” subsection in the Methods to clarify this multi-level procedure.

(5) Figure 4a:

Why do the phases go all the way to 2*pi if periodicity is either three-fold or six-fold? 

When performing correlation between phases, you should perform a circular-circular correlation instead of a Pearson's correlation.

We thank the reviewer for raising this important point. In the original Figure 4a, both EC and HPC phases spanned 0–2π because their sinusoidal phase estimates were projected into a common angular space by scaling them according to their symmetry factors (i.e., multiplying the 3-fold phase by 3 and the 6-fold phase by 6), followed by taking the modulo 2π. However, this projection forced signals with distinct intrinsic periodicities (120° vs. 60° cycles) into a shared 360° space, thereby distorting their relative angular distances and disrupting the one-to-one correspondence between physical directions and phase values. Consequently, this transformation could bias the estimation of their phase relationship.

In the revised analysis and Figure 4a, we retained the original phase estimates derived from the sinusoidal modulation within their native periodic ranges (0–120° for 3-fold and 0–60° for 6-fold) by applying modulo operations directly. Following your suggestion, the relationship between EC and HPC phases was then quantified using circular–circular correlation (Jammalamadaka & Sengupta, 2001), as implemented in the CircStat MATLAB toolbox. This updated analysis avoids the rescaling artifact and provides a statistically stronger and conceptually clearer characterization of the phase correspondence between EC and HPC.

(6) Figure 4d needs additional clarification:

Phase-locking is typically used to describe data with a high temporal precision. I understand you adopted an EEG analysis technique to this reconstructed fMRI time-series data, but it should be described differently to avoid confusion. This needs additional control analyses (especially given that 3 is a multiple of 6) to confirm that this result is specific to the periodicities found in the paper.

We thank the reviewer for this insightful comment. We have extensively revised the description of the Figure 4 to avoid confusion with EEG-based phase-locking techniques. The revised text now explicitly clarifies that our approach quantifies spatial-domain periodic coupling across path directions, rather than temporal synchronization of neural signals.

To further address the reviewer’s concern about potential effects of the integer multiple relationship between the 3-fold HPC and 6-fold EC periodicities, we additionally performed two control analyses using the 9-fold and 12-fold EC components, both of which are also integer multiples of the 3-fold HPC periodicity. Neither control analysis showed significant coupling (p > 0.05), confirming that the observed 3-fold–6-fold coupling was specific and not driven by their harmonic relationship.

The description of the revised Figure 4 has been updated in the “Phase Synchronization Between HPC and EC Activity” subsection of the Results.

(7) Figure 5a is misleading. In the text, you say you test for propagation to egocentric cortical areas, but I don’t see any analyses done that test this. This feels more like a possible extension/future direction of your work that may be better placed in the discussion.

We are sorry for the confusion. Figure 5a was intended as a hypothesis-driven illustration to motivate our analysis of behavioral periodicity based on participants’ task performance. However, we agree with the reviewer that, on its own, Figure 5a could be misleading, as it does not directly present supporting analyses.

To provide empirical support for the interpretation depicted in Figure 5a, we conducted a whole-brain analysis (Figure S8), which revealed significant 3-fold periodic signals in egocentric cortical regions, including the parietal cortex (PC), precuneus (PCU), and motor regions.

To avoid potential misinterpretation, we have revised the main text to include these results and explicitly referenced Figure S8 in connection with Figure 5a.

The updated description in the “3-fold periodicity in human behavior” subsection in the Results is as follows:

“Considering the reciprocal connectivity between the medial temporal lobe (MTL), where the EC and HPC reside, and the parietal cortex implicated in visuospatial perception and action, together with the observed 3-fold periodicity within the DMN (including the PC and PCu; Fig. S8), we hypothesized that the 3-fold periodic representations of path directions extend beyond the MTL to the egocentric cortical areas, such as the PC, thereby influencing participants' visuospatial task performance (Fig. 5a)”.

Additionally, Figure 5a has been modified to more clearly highlight the hypothesized link between activity periodicity and behavioral periodicity, rather than suggesting a direct anatomical pathway.

(8) PhaseSync model: I am not an expert in this type of modeling, so please put a lower weight on this comment (especially compared to some of the other reviewers). While the PhaseSync model seems interesting, it’s not clear from the discussion how this compares to current models. E.g., Does it support them by adding the three-fold HC periodicity? Does it demonstrate that some of them can't be correct because they don't include this three-fold periodicity?

We thank the reviewer for the insightful comment regarding the PhaseSync model. We agree that further clarifying its relationship to existing computational frameworks is important.

The EC–HPC PhaseSync model is not intended to replace or contradict existing grid–place cell models of navigation (e.g., Bicanski and Burgess, 2019; Whittington et al., 2020; Edvardsen et al., 2020). Instead, it offers a hierarchical extension by proposing that vectorial representations in the hippocampus emerge from the projections of periodic grid codes in the entorhinal cortex. Specifically, the model suggests that grid cell populations encode integrated path information, forming a vectorial gradient toward goal locations.

To simplify the theoretical account, our model was implemented in an idealized square layout. In more complex real-world environments, hippocampal 3-fold periodicity may interact with additional spatial variables, such as distance, movement speed, and environmental boundaries.

We have revised the final two paragraphs of the Discussion to clarify this conceptual framework and emphasize the importance of future studies in exploring how periodic activity in the EC–HPC circuit interacts with environmental features to support navigation.

Reviewer #2 (Recommendations for the authors):

(1) Please show a histogram of movement direction sampling for each participant.

We thank the reviewer for this helpful suggestion. We have added a new supplementary figure (Figure S2) showing histograms of path direction sampling for each participant (36 bins of 10°). The figure is also included. Rayleigh tests for circular uniformity revealed no significant deviations from uniformity (all ps > 0.05, Bonferroni-corrected across participants), confirming that path directions were sampled evenly across 0°–360°.

(2) Why didn’t you use participants’ original trajectories (instead of the trajectories inferred from the movement start and end points) for the hexadirectional analyses? 

In our paradigm, participants used two MRI-compatible 2-button response boxes (one for each hand) to adjust the two features of the greebles. As a result, the raw adjustment path contained only four cardinal directions (up, down, left, right). If we were to use the raw stepwise trajectories, the analysis would be restricted to these four directions, which would severely limit the angular resolution. By instead defining direction as the vector from the start to the end position in feature space, we can expand the effective range of directions to the full 0–360°. This approach follows previous literature on abstract grid-like coding in humans (e.g., Constantinescu et al., 2016), where direction was similarly defined by the relative change between two feature dimensions rather than the literal stepwise path. We have added this clarification in the “Sinusoidal modulation” subsection of the revised method.

(3) Legend of Figure 2: the statement "localizing grid cell activity" seems too strong because it is still not clear whether hexadirectional signals indeed result from grid-cell activity (e.g., Bin Khalid et al., eLife, 2024). I would suggest rephrasing this statement (here and elsewhere). 

Thank you for this helpful suggestion. We have removed the statement “localizing grid cell activity” to avoid ambiguity and revised the legend of Figure 2a to more explicitly highlight its main purpose—defining how path directions and the aligned/misaligned conditions were constructed in the 6-fold modulation. We have also modified similar expressions throughout the manuscript to ensure consistency and clarity.

(4) Legend of Figure 2: “cluster-based SVC correction for multiple comparisons” - what is the small volume you are using for the correction? Bilateral EC?

For both Figure 2 and Figure 3, the anatomical mask of the bilateral medial temporal lobe (MTL), as defined by the AAL atlas, was used as the small volume for correction. This has been clarified in the revised Statistical Analysis section of the Methods as “… with small-volume correction (SVC) applied within the bilateral MTL”.

(5) Legend of Figure 2: "ROI-based analysis" - what kind of ROI are you using? "corrected for multiple comparisons" - which comparisons are you referring to? Different symmetries and also the right/left hemisphere?

In Figure 2b, the ROI was defined as a functional mask derived from the significant activation cluster in the right entorhinal cortex (EC). Since no robust clusters were observed in the left EC, the functional ROI was restricted to the right hemisphere. We indeed included Figure 2c to illustrate this point; however, we recognize that our description in the text was not sufficiently clear.

Regarding the correction for multiple comparisons, this refers specifically to the comparisons across different rotational symmetries (3-, 4-, 5-, 6-, and 7-fold). Only the 6-fold symmetry survived correction, whereas no significant effects were detected for the other symmetries.

We have clarified these points in the “6-fold periodicity in the EC” subsection of the result as “… The ROI was defined as a functional mask of the right EC identified in the voxel-based analysis and further restricted within the anatomical EC. These analyses revealed significant periodic modulation only at 6-fold (Figure  2c; t(32) = 3.56, p = 0.006, two-tailed, corrected for multiple comparisons across rotational symmetries; Cohen’s d = 0.62) …”.

We have also revised the “3-fold periodicity in the HPC” subsection of the result as “… ROI analysis, using a functional mask of the HPC identified in the spectral analysis and further restricted within the anatomical HPC, indicated that HPC activity selectively fluctuated at 3-fold periodicity (Figure 3e; t(32) = 3.94, p = 0.002, corrected for multiple comparisons across rotational symmetries; Cohen’s d = 0.70) …”.

(6) Figure 2d: Did you rotationally align 0{degree sign} across participants? Please state explicitly whether (or not) 0{degree sign} aligns with the x-axis in Greeble space.

We thank the reviewer for this helpful question. Yes, before reconstructing the directional tuning curve in Figure 2d, path directions were rotationally aligned for each participant by subtracting the participant-specific grid orientation (ϕ) estimated from the independent dataset (odd sessions). We have now made this description explicit in the revised manuscript in the “6-fold periodicity in the EC” subsection of the Results, stating “… To account for individual difference in spatial phase, path directions were calibrated by subtracting the participant-specific grid orientation estimated from the odd sessions ...”.

(7) Clustering of grid orientations in 30 participants: What does “Bonferroni corrected” refer to? Also, the Rayleigh test is sensitive to the number of voxels - do you obtain the same results when using pair-wise phase consistency? 

“Bonferroni corrected” here refers to correction across participants. We have clarified this in the first paragraph of the “6-fold periodicity in the EC” subsection of the Result and in the legend of Supplementary Figure S5 as “Bonferroni-corrected across participants.”

To examine whether our findings were sensitive to the number of voxels, we followed the reviewer’s guidance to compute pairwise phase consistency (PPC; Vinck et al., 2010) for each participant. The PPC results replicated those obtained with the Rayleigh test. We have updated the new results into the Supplementary Figure S5. We also updated the “Statistical Analysis” subsection of the Methods to describe PPC as “For the PPC (Vinck et al., 2010), significance was tested using 5,000 permutations of uniformly distributed random phases (0–2π) to generate a null distribution for comparison with the observed PPC”.

(8) 6-fold periodicity in the EC: Do you compute an average grid orientation across all EC voxels, or do you compute voxel-specific grid orientations?

Following the protocol originally described by Doeller et al. (2010), we estimated voxel-wise grid orientations within the EC and then obtained a participant-specific orientation by averaging across voxels within a hand-drawn bilateral EC mask. The procedure is described in detail in the “Sinusoidal modulation” subsection of the Methods.

(9) Hand-drawn bilateral EC mask: What was your procedure for drawing this mask? What results do you get with a standard mask, for example, from Freesurfer or SPM? Why do you perform this analysis bilaterally, given that the earlier analysis identified 6-fold symmetry only in the right EC? What do you mean by "permutation corrected for multiple comparisons"?

We thank the reviewer for raising these important methodological points. To our knowledge, no standard volumetric atlas provides an anatomically defined entorhinal cortex (EC) mask. For example, the built-in Harvard–Oxford cortical structural atlas in FSL contains only a parahippocampal region that encompasses, but does not isolate, the EC. The AAL atlas likewise does not contain an EC region. In FreeSurfer, an EC label is available, but only in the fsaverage surface space, which is not directly compatible with MNI-based volumetric group-level analyses.

Therefore, we constructed a bilateral EC mask by manually delineating the EC according to the detailed anatomical landmarks described by Insausti et al. (1998). Masks were created using ITK-SNAP (Version 3.8, www.itksnap.org). For transparency and reproducibility, the mask has been made publicly available at the Science Data Bank (link: https://www.scidb.cn/s/NBriAn), as indicated in the revised Data and Code availability section.

Regarding the use of a bilateral EC mask despite voxel-wise effects being strongest in the right EC. First, we did not have any a priori hypothesis regarding laterality of EC involvement before performing analyses. Second, previous studies estimated grid orientation using a bilateral EC mask in their sinusoidal analyses (Doeller et al., 2010; Constantinescu et al., 2016; Bao et al., 2019; Wagner et al., 2023; Raithel et al., 2023). We therefore followed this established approach to estimate grid orientation.

By “permutation corrected for multiple comparisons” we refer to the family-wise error correction applied to the reconstructed directional tuning curves (Figure 2d for the EC, Figure 3f for the HPC). Specifically, directional labels were randomly shuffled 5,000 times, and an FFT was applied to each shuffled dataset to compute spectral power at each fold. This procedure generated null distributions of spectral power for each symmetry. For each fold, the 95th percentile of the maximal power across permutations was used as the uncorrected threshold. To correct across folds, the 95th percentile of the maximal suprathreshold power across all symmetries was taken as the family-wise error–corrected threshold. We have clarified this procedure in the revised “Statistical Analysis” subsection of the Methods.

(10) Figures 3b and 3d: Why do different hippocampal voxels show significance for the sinusoidal versus spectral analysis? Shouldn’t the analyses be redundant and, thus, identify the same significant voxels? 

We thank the reviewer for this insightful question. Although both sinusoidal modulation and spectral analysis aim to detect periodic neural activity, the two approaches are methodologically distinct and are therefore not expected to identify exactly the same significant voxels.

Sinusoidal modulation relies on a GLM with sine and cosine regressors to test for phase-aligned periodicity (e.g., 3-fold or 6-fold), calibrated according to the estimated grid orientation. This approach is highly specific but critically depends on accurate orientation estimation. In contrast, spectral analysis applies Fourier decomposition to the directional tuning profile, enabling the detection of periodic components without requiring orientation calibration.

Accordingly, the two analyses are not redundant but complementary. The FFT approach allows for an unbiased exploration of multiple candidate periodicities (e.g., 3–7-fold) without predefined assumptions, thereby providing a critical cross-validation of the sinusoidal GLM results. This strengthens the evidence for 6-fold periodicity in EC and 3-fold periodicity in HPC. Furthermore, FFT uniquely facilitates the analysis of periodicities in behavioral performance data, which is not feasible with standard sinusoidal GLM approaches. This methodological consistency enables direct comparison of periodicities across neural and behavioral domains.

Additionally, the anatomical distributions of the HPC clusters appear more similar between Figure 3b and Figure 3d after re-plotting Figure 3d using the peak voxel coordinates (x = –24, y = –18), which are closer to those used for Figure 3b (x = –24, y = –20), as shown in the revised Figure 3.

Taken together, the two analyses serve distinct but complementary purposes.

(11) 3-fold sinusoidal analysis in hippocampus: What kind of small volume are you using to correct for multiple comparisons?

We thank the reviewer for this comment. The same small volume correction procedure was applied as described in R4. Specifically, the anatomical mask of the bilateral medial temporal lobe (MTL), as defined by the AAL atlas, was used as the small volume for correction. This procedure has been clarified in the revised Statistical Analysis section of the Methods as following: “… with small-volume correction (SVC) applied within the bilateral MTL.”

(12) Figure S5: “right HPC” – isn’t the cluster in the left hippocampus? 

We are sorry for the confusion. The brain image was present in radiological orientation (i.e., the left and right orientations are flipped). We also checked the figure and confirmed that the cluster shown in the original Figure S5 (i.e., Figure S6 in the revised manuscript) is correctly labeled as the right hippocampus, as indicated by the MNI coordinate (x = 22), where positive x values denote the right hemisphere. To avoid potential confusion, we have explicitly added the statement “Volumetric results are displayed in radiological orientation” to the figure legends of all volume-based results.

(13) Figure S5: Why are the significant voxels different from the 3-fold symmetry analysis using 10{degree sign} bins?

As shown in R10, the apparent differences largely reflect variation in MNI coordinates. After adjusting for display coordinates, the anatomical locations of the significant clusters are in fact highly similar between the 10°-binned (Figure 3d, shown above) and the 20°-binned results (Figure S6).

Although both analyses rely on sinusoidal modulation, they differ in the resolution of the input angular bins (10° vs. 20°). Combined with the inherent noise in fMRI data, this makes it unlikely that the two approaches would yield exactly the same set of significant voxels. Importantly, both analyses consistently reveal robust 3-fold periodicity in the hippocampus, indicating that the observed effect is not dependent on angular bin size.

(14) Figure 4a and corresponding text: What is the unit? Phase at which frequency? Are you using a circular-circular correlation to test for the relationship?

We thank the reviewer for raising this important point. In the revised manuscript, we have clarified that the unit of the phase values is radians, corresponding to the 6-fold periodic component in the EC and the 3-fold periodic component in the HPC. In the original Figure 4a, both EC and HPC phases—estimated from sinusoidal modulation—were analyzed using Pearson correlation. We have since realized issues with this approach, as also noted R5 to Reviewer #1.

In the revised analysis and Figure 4a (as shown above), we re-evaluated the relationship between EC and HPC phases using a circular–circular correlation (Jammalamadaka & Sengupta, 2001), implemented in the CircStat MATLAB toolbox. The “Phase synchronization between the HPC and EC activity” subsection of the Result has been accordingly updated as following:

“To examine whether the spatial phase structure in one region could predict that in another, we tested whether the orientations of the 6-fold EC and 3-fold HPC periodic activities, estimated from odd-numbered sessions using sinusoidal modulation with rotationally symmetric parameters (in radians), were correlated across participants. A cross-participant circular–circular correlation was conducted between the spatial phases of the two areas to quantify the spatial correspondence of their activity patterns (EC: purple dots; HPC: green dots) (Jammalamadaka & Sengupta, 2001). The analysis revealed a significant circular correlation (Figure 4a; r = 0.42, p < 0.001) …”.

In the “Statistical analysis” subsection of the method:

“… The relationship between EC and HPC phases was evaluated using the circular–circular correlation (Jammalamadaka & Sengupta, 2001) implemented in the CircStat MATLAB toolbox …”.

(15) Paragraph following “We further examined amplitude-phase coupling...” - please clarify what data goes into this analysis.

We thank the reviewer for this helpful comment. In this analysis, the input data consisted of hippocampal (HPC) phase and entorhinal (EC) amplitude, both extracted using the Hilbert transform from the reconstructed BOLD signals of the EC and HPC derived through sinusoidal modulation. We have substantially revised the description of the amplitude–phase coupling analysis in the third paragraph of the “Phase Synchronization Between HPC and EC Activity” subsection of the Results to clarify this procedure.

(16) Alignment between EC 6-fold phases and HC 3-fold phases: Why don't you simply test whether the preferred 6-fold orientations in EC are similar to the preferred 3-fold phases in HC? The phase-amplitude coupling analyses seem sophisticated but are complex, so it is somewhat difficult to judge to what extent they are correct. 

We thank the reviewer for this thoughtful comment. We employed two complementary analyses to examine the relationship between EC and HPC activity. In the revised Figure 4 (as shown in Figure 4 for Reviewer #1), Figure 4a provides a direct and intuitive measure of the phase relationship between the two regions using circular–circular correlation. Figure 4b–c examines whether the activity peaks of the two regions are aligned across path directions using cross-frequency amplitude–phase coupling, given our hypothesis that the spatial phase of the HPC depends on EC projections. These two analyses are complementary: a phase correlation does not necessarily imply peak-to-peak alignment, and conversely, peak alignment does not always yield a statistically significant phase correlation. We therefore combined multiple analytical approaches as a cross-validation across methods, providing convergent evidence for robust EC–HPC coupling.

(17) Figure 5: Do these results hold when you estimate performance just based on “deviation from the goal to ending locations” (without taking path length into account)? 

We thank the reviewer for this thoughtful suggestion. Following the reviewer’s advice, we re-estimated behavioral performance using the deviation between the goal and ending locations (i.e., error size) and path length independently. As shown in the new Figure S9, no significant periodicity was observed in error size (p > 0.05), whereas a robust 3-fold periodicity was found for path length (p < 0.05, corrected for multiple comparisons).

We employed two behavioral metrics,(1) path length and (2) error size, for complementary reasons. In our task, participants navigated using four discrete keys corresponding to the cardinal directions (north, south, east, and west). This design inherently induces a 4-fold bias in path directions, as described in the “Behavioral performance” subsection of the Methods. To minimize this artifact, we computed the objectively optimal path length and used it to calibrate participants’ path lengths. However, error size could not be corrected in the same manner and retained a residual 4-fold tendency (see Figure S9d).

Given that both path length and error size are behaviorally relevant and capture distinct aspects of task performance, we decided to retain both measures when quantifying behavioral periodicity. This clarification has been incorporated into the “Behavioral performance” subsection of the Methods, and the 2^nd paragraph of the “3-fold periodicity in human behavior” subsection of the Results.

(18) Phase locking between behavioral performance and hippocampal activity: What is your way of creating surrogates here?

We thank the reviewer for this helpful question. Surrogate datasets were generated by circularly shifting the signal series along the direction axis across all possible offsets (following Canolty et al., 2006). This procedure preserves the internal phase structure within each domain while disrupting consistent phase alignment, thereby removing any systematic coupling between the two signals. Each surrogate dataset underwent identical filtering and coherence computation to generate a null distribution, and the observed coherence strength was compared with this distribution using paired t-tests across participants. The statistical analysis section has been systematically revised to incorporate these methodological details.

(19) I could not follow why the authors equate 3-fold symmetry with vectorial representations. This includes statements such as “these empirical findings provide a potential explanation for the formation of vectorial representation observed in the HPC.” Please clarify.

We thank the reviewer for raising this point. Please refer to our response to R2 for Reviewer #1 and the revised Introduction (paragraphs 2–4), where we explicitly explain why the three primary axes of the hexagonal grid cell code can manifest as vector fields.

(20) It was unclear whether the sentence “The EC provides a foundation for the formation of periodic representations in the HPC” is based on the authors’ observations or on other findings. If based on the authors’ findings, this statement seems too strong, given that no other studies have reported periodic representations in the hippocampus to date (to the best of my knowledge).

We thank the reviewer for this comment. We agree that the original wording lacked sufficient rigor. We have extensively revised the 3rd paragraph of the Discussion section with more cautious language by reducing overinterpretation and emphasizing the consistency of our findings with prior empirical evidence, as follows: “The EC–HPC PhaseSync model demonstrates how a vectorial representation may emerge in the HPC from the projections of populations of periodic grid codes in the EC. The model was motivated by two observations. First, the EC intrinsically serves as the major source of hippocampal input (Witter and Amaral, 1991; van Groen et al., 2003; Garcia and Buffalo, 2020), and grid codes exhibit nearly invariant spatial orientations (Hafting et al., 2005; Gardner et al., 2022). Second, mental planning, characterized by “forward replay” (Dragoi and Tonegawa, 2011; Pfeiffer, 2020), has the capacity to activate populations of grid cells that represent sequential experiences in the absence of actual physical movement (Nyberg et al., 2022). We hypothesize that an integrated path code of sequential experiences may eventually be generated in the HPC, providing a vectorial gradient toward the goal location. The path code exhibits regular, vector-like representations when the path direction aligns with the orientations of grid axes, and becomes irregular when they misalign. This explanation is consistent with the band-like representations observed in the dorsomedial EC (Krupic et al., 2012) and the irregular activity fields of trace cells in the HPC (Poulter et al., 2021). ”

Author response:

The following is the authors’ response to the original reviews

A point by point response included below. Before we turn to that we want to note one change that we decided to introduce, related to generalization on unseen tissues/cell types (Figure 3a in the original submission and related question by Reviewer #2 below). This analysis was based on adding a latent “RBP state” representation during learning of condition/tissue specific splicing. The “RBP state” per condition is captured by a dedicated encoder. Our original plan was to have a paper describing a new RBP-AE model we developed in parallel, which also served as the base to capture this “RBP State”. However, we got delayed in getting this second paper finalized (it was led by other lab members, some of whom have already left the lab). This delay affected the TrASPr manuscript as TrASPr’s code should be available and analysis reproducible upon publication. After much deliberation, we decided that in order to comply with reproducibility standards while not self scooping the RBP-AE paper, we eventually decided to take out the RBP-AE and replace it with a vanilla PCA based embedding for the “RBP-State”. The PCA approach is simpler and reproducible, based on linear transformation of the RBPs expression vector into a lower dimension. The qualitative results included in Figure 3a still hold, and we also produced the new results suggested by Reviewer #2 in other GTEX tissues with this PCA based embedding (below). 

We don’t believe the switch to PCA based embedding should have any bearing on the current manuscript evaluation but wanted to take this opportunity to explain the reasoning behind this additional change.

Public Reviews:

Reviewer #1 (Public review):

Summary:

The authors propose a transformer-based model for the prediction of condition - or tissue-specific alternative splicing and demonstrate its utility in the design of RNAs with desired splicing outcomes, which is a novel application. The model is compared to relevant existing approaches (Pangolin and SpliceAI) and the authors clearly demonstrate its advantage. Overall, a compelling method that is well thought out and evaluated.

Strengths:

(1) The model is well thought out: rather than modeling a cassette exon using a single generic deep learning model as has been done e.g. in SpliceAI and related work, the authors propose a modular architecture that focuses on different regions around a potential exon skipping event, which enables the model to learn representations that are specific to those regions. Because each component in the model focuses on a fixed length short sequence segment, the model can learn position-specific features. Another difference compared to Pangolin and SpliceAI which are focused on modeling individual splice junctions is the focus on modeling a complete alternative splicing event.

(2) The model is evaluated in a rigorous way - it is compared to the most relevant state-of-the-art models, uses machine learning best practices, and an ablation study demonstrates the contribution of each component of the architecture.

(3) Experimental work supports the computational predictions.     

(4) The authors use their model for sequence design to optimize splicing outcomes, which is a novel application.

We wholeheartedly thank Reviewer #1 for these positive comments regarding the modeling approach we took to this task and the evaluations we performed. We have put a lot of work and thought into this and it is gratifying to see the results of that work acknowledged like this.

Weaknesses:

No weaknesses were identified by this reviewer, but I have the following comments:

(1) I would be curious to see evidence that the model is learning position-specific representations.

This is an excellent suggestion to further assess what the model is learning. To get a better sense of the position-specific representation we performed the following analyses:

(1) Switching the transformers relative order: All transformers are pretrained on 3’ and 5’ splice site regions before fine-tunning for the PSI and dPSI prediction task. We hypothesized that if relative position is important, switching the order of the transformers would make a large difference on prediction accuracy. Indeed if we switch the 3’ and 5’ we see as expected a severe drop in performance, with Pearson correlation on test data dropping from 0.82 to 0.11. Next, we switched the two 5’ and 3’ transformers, observing a drop to 0.65 and 0.78 respectively. When focusing only on changing events the drop was from 0.66 to 0.54 (for 3’ SS transformers), 0.48 (for 5’ SS transformers), and 0.13 (when the 3’ and 5’ transformers flanking the alternative exon were switched). 

(2) Position specific effect of RBPs: We wanted to test whether the model is able to learn position specific effects for RBPs. For this we focused on two RBPs, FOX (a family of three highly related RBPs), and QKI, both have a relatively well defined motif, known condition and position specific effect identified via RBP KD experiments combined with CLIP experiments (e.g. PMID: 23525800, PMID: 24637117, PMID: 32728246). For each, we randomly selected 40 highly and 40 lowly included cassette exons sequences. We then ran in-silico mutagenesis experiments where we replaced small windows of sequences with the RBP motifs (80 for RBFOX and 80 for QKI), then compared TrASPR’s predictions for the average predictions for 5 random sequences inserted in the same location. The results of this are now shown in Figure 4 Supp 3, where the y-axis represents the dPSI effect per position (x-axis), and the color represents the percentile of observed effects over inserting motifs in that position across all 80 sequences tested. We see that both RBPs have strong positional preferences for exerting a strong effect on the alternative exon. We also see differences between binding upstream and downstream of the alternative exon. These results, learned by the model from natural tissue-specific variations, recapitulate nicely the results derived from high-throughput experimental assays. However, we also note that effects were highly sequence specific. For example, RBFOX is generally expected to increase inclusion when binding downstream of the alternative exon and decrease inclusion when binding upstream. While we do observe such a trend we also see cases where the opposite effects are observed. These sequence specific effects have been reported in the literature but may also represent cases where the model errs in the effect’s direction. We discuss these new results in the revised text.

(3) Assessing BOS sequence edits to achieve tissue-specific splicing: Here we decided to test whether BOS edits in intronic regions (at least 8b away from the nearest splice site) are important for the tissue-specific effect. The results are now included in Figure 6 Supp 1, clearly demonstrating that most of the neuronal specific changes achieved by BOS were based on changing the introns, with a strong effect observed for both up and downstream intron edits.

(2) The transformer encoders in TrASPr model sequences with a rather limited sequence size of 200 bp; therefore, for long introns, the model will not have good coverage of the intronic sequence. This is not expected to be an issue for exons.

The reviewer is raising a good question here. On one hand, one may hypothesize that, as the reviewer seems to suggest, TrASPr may not do well on long introns as it lacks the full intronic sequence.

Conversely, one may also hypothesize that for long introns, where the flanking exons are outside the window of SpliceAI/Pangolin, TrASPr may have an advantage.

Given this good question and a related one by Reviewer #2, we divided prediction accuracy by intron length and the alternative exon length.

For short exons  (<100bp) we find TrASPr and Pangolin perform similarly, but for longer exons, especially those > 200, TrASPr results are better. When dividing samples by the total length of the upstream and downstream intron, we find TrASPr outperform all other models for introns of combined length up to 6K, but Pangolin gets better results when the combined intron length is over 10K. This latter result is interesting as it means that contrary to the second hypothesis laid out above, Pangolin’s performance did not degrade for events where the flanking exons were outside its field of view. We note that all of the above holds whether we assess all events or just cases of tissue specific changes. It is interesting to think about the mechanistic causes for this. For example, it is possible that cassette exons involving very long introns evoke a different splicing mechanism where the flanking exons are not as critical and/or there is more signal in the introns which is missed by TrASPr. We include these new results now as Figure 2 - Supp 1,2 and discuss these in the main text.

(3) In the context of sequence design, creating a desired tissue- or condition-specific effect would likely require disrupting or creating motifs for splicing regulatory proteins. In your experiments for neuronal-specific Daam1 exon 16, have you seen evidence for that? Most of the edits are close to splice junctions, but a few are further away.

That is another good question. Regarding Daam1 exon 16, in the original paper describing the mutation locations some motif similarities were noted to PTB (CU) and CUG/Mbnl-like elements (Barash et al Nature 2010). In order to explore this question beyond this specific case we assessed the importance of intronic edits by BOS to achieve a tissue specific splicing profile - see above.

(4) For sequence design, of tissue- or condition-specific effect in neuronal-specific Daam1 exon 16 the upstream exonic splice junction had the most sequence edits. Is that a general observation? How about the relative importance of the four transformer regions in TrASPr prediction performance?

This is another excellent question. Please see new experiments described above for RBP positional effect and BOS edits in intronic regions which attempt to give at least partial answers to these questions. We believe a much more systematic analysis can be done to explore these questions but such evaluation is beyond the scope of this work.

(5) The idea of lightweight transformer models is compelling, and is widely applicable. It has been used elsewhere. One paper that came to mind in the protein realm:

Singh, Rohit, et al. "Learning the language of antibody hypervariability." Proceedings of the National Academy of Sciences 122.1 (2025): e2418918121.

We definitely do not make any claim this approach of using lighter, dedicated models instead of a large ‘foundation’ model has not been taken before. We believe Rohit et al mentioned above represents a somewhat different approach, where their model (AbMAP) fine-tunes large general protein foundational models (PLM) for antibody-sequence inputs by supervising on antibody structure and binding specificity examples. We added a description of this modeling approach citing the above work and another one which specifically handles RNA splicing (intron retention, PMID: 39792954).

Reviewer #2 (Public review):

Summary:

The authors present a transformer-based model, TrASPr, for the task of tissue-specific splicing prediction (with experiments primarily focused on the case of cassette exon inclusion) as well as an optimization framework (BOS) for the task of designing RNA sequences for desired splicing outcomes.

For the first task, the main methodological contribution is to train four transformer-based models on the 400bp regions surrounding each splice site, the rationale being that this is where most splicing regulatory information is. In contrast, previous work trained one model on a long genomic region. This new design should help the model capture more easily interactions between splice sites. It should also help in cases of very long introns, which are relatively common in the human genome.

TrASPr's performance is evaluated in comparison to previous models (SpliceAI, Pangolin, and SpliceTransformer) on numerous tasks including splicing predictions on GTEx tissues, ENCODE cell lines, RBP KD data, and mutagenesis data. The scope of these evaluations is ambitious; however, significant details on most of the analyses are missing, making it difficult to evaluate the strength of the evidence. Additionally, state-of-the-art models (SpliceAI and Pangolin) are reported to perform extremely poorly in some tasks, which is surprising in light of previous reports of their overall good prediction accuracy; the reasoning for this lack of performance compared to TrASPr is not explored.

In the second task, the authors combine Latent Space Bayesian Optimization (LSBO) with a Transformer-based variational autoencoder to optimize RNA sequences for a given splicing-related objective function. This method (BOS) appears to be a novel application of LSBO, with promising results on several computational evaluations and the potential to be impactful on sequence design for both splicing-related objectives and other tasks.

We thank Reviewer #2 for this detailed summary and positive view of our work. It seems the main issue raised in this summary regards the evaluations: The reviewer finds details of the evaluations missing and the fact that SpliceAI and Pangolin perform poorly on some of the tasks to be surprising. We made a concise effort to include the required details, including code and data tables. In short, some of the concerns were addressed by adding additional evaluations, some by clarifying missing details, and some by better explaining where Pangolin and SpliceAI may excel vs. settings where these may not do as well. More details are given below. 

Strengths:

(1) A novel machine learning model for an important problem in RNA biology with excellent prediction accuracy.

(2) Instead of being based on a generic design as in previous work, the proposed model incorporates biological domain knowledge (that regulatory information is concentrated around splice sites). This way of using inductive bias can be important to future work on other sequence-based prediction tasks.

Weaknesses:

(1) Most of the analyses presented in the manuscript are described in broad strokes and are often confusing. As a result, it is difficult to assess the significance of the contribution.

We made an effort to make the tasks be specific and detailed,  including making the code and data of those available. We believe this helped improve clarity in the revised version.

(2) As more and more models are being proposed for splicing prediction (SpliceAI, Pangolin, SpliceTransformer, TrASPr), there is a need for establishing standard benchmarks, similar to those in computer vision (ImageNet). Without such benchmarks, it is exceedingly difficult to compare models. For instance, Pangolin was apparently trained on a different dataset (Cardoso-Moreira et al. 2019), and using a different processing pipeline (based on SpliSER) than the ones used in this submission. As a result, the inferior performance of Pangolin reported here could potentially be due to subtle distribution shifts. The authors should add a discussion of the differences in the training set, and whether they affect your comparisons (e.g., in Figure 2). They should also consider adding a table summarizing the various datasets used in their previous work for training and testing. Publishing their training and testing datasets in an easy-to-use format would be a fantastic contribution to the community, establishing a common benchmark to be used by others.

There are several good points to unpack here. Starting from the last one, we very much agree that a standard benchmark will be useful to include. For tissue specific splicing quantification we used the GTEx dataset from which we select six representative human tissues (heart, cerebellum, lung, liver, spleen, and EBV-transformed lymphocytes). In total, we collected 38394 cassette exon events quantified across 15 samples (here a ‘sample’ is a cassette exon quantified in two tissues) from the GTEx dataset with high-confidence quantification for their PSIs based on MAJIQ. A detailed description of how this data was derived is now included in the Methods section, and the data itself is made available via the bitbucket repository with the code.

Next, regarding the usage of different data and distribution shifts for Pangolin: The reviewer is right to note there are many differences between how Pangolin and TrASPr were trained. This makes it hard to determine whether the improvements we saw are not just a result of different training data/labels. To address this issue, we first tried to finetune the pre-trained Pangolin with MAJIQ’s PSI dataset: we use the subset of the GTEx dataset described above, focusing on the three tissues analyzed in Pangolin’s paper—heart, cerebellum, and liver—for a fair comparison. In total, we obtained 17,218 events, and we followed the same training and test split as reported in the Pangolin paper. We got Pearson: 0.78 Spearman: 0.68 which are values similar to what we got without this extra fine tuning. Next, we retrained Pangolin from scratch, with the full tissues and training set used for TrASPr, which was derived from MAJIQ’s quantifications. Since our model only trained on human data with 6 tissues at the same time, we modified Pangolin from original 4 splice site usage outputs to 6 PSI outputs. We tried to take the sequence centered with the first or the second splice site of the mid exon. This test resulted in low performance (3’ SS: pearson 0.21 5’ SS: 0.26.). 

The above tests are obviously not exhaustive but their results suggest that the differences we observe are unlikely to be driven by distribution shifts. Notably, the original Pangolin was trained on much more data (four species, four tissues each, and sliding windows across the entire genome). This training seems to be important for performance while the fact we switched from Pangolin’s splice site usage to MAJIQ’s PSI was not a major contributor. Other potential reasons for the improvements we observed include the architecture, target function, and side information (see below) but a complete delineation of those is beyond the scope of this work. 

(3) Related to the previous point, as discussed in the manuscript, SpliceAI, and Pangolin are not designed to predict PSI of cassette exons. Instead, they assign a "splice site probability" to each nucleotide. Converting this to a PSI prediction is not obvious, and the method chosen by the authors (averaging the two probabilities (?)) is likely not optimal. It would be interesting to see what happens if an MLP is used on top of the four predictions (or the outputs of the top layers) from SpliceAI/Pangolin. This could also indicate where the improvement in TrASPr comes from: is it because TrASPr combines information from all four splice sites? Also, consider fine-tuning Pangolin on cassette exons only (as you do for your model).

Please see the above response. We did not investigate more sophisticated models that adjust Pangolin’s architecture further as such modifications constitute new models which are beyond the scope of this work.

(4) L141, "TrASPr can handle cassette exons spanning a wide range of window sizes from 181 to 329,227 bases - thanks to its multi-transformer architecture." This is reported to be one of the primary advantages compared to existing models. Additional analysis should be included on how TrASPr performs across varying exon and intron sizes, with comparison to SpliceAI, etc.

This was a good suggestion, related to another comment made by Reviewer #1. Please see above our response to them with a breakdown by exon/intron length.

(5) L171, "training it on cassette exons". This seems like an important point: previous models were trained mostly on constitutive exons, whereas here the model is trained specifically on cassette exons. This should be discussed in more detail.

Previous models were not trained exclusively on constitutive exons and Pangolin specifically was trained with their version of junction usage across tissues. That said, the reviewer’s point is valid (and similar to ones made above) about a need to have a matched training/testing and potential distribution shifts. Please see response and evaluations described above. 

(6) L214, ablations of individual features are missing.

These were now added to the table which we moved to the main text (see table also below).

(7) L230, "ENCODE cell lines", it is not clear why other tissues from GTEx were not included.

Good question. The task here was to assess predictions in unseen conditions, hence we opted to test on completely different data of human cell lines rather than additional tissue samples. Following the reviewers suggestion we also evaluated predictions on two additional GTEx tissues, Cortex and Adrenal Gland. These new results, as well as the previous ones for ENCODE, were updated to use the PCA based embedding of “RBP-State” as described above. We also compared the predictions using the PCA based embedding of the “RBP-State” to training directly on data (not the test data of course) from these tissues. See updated Figure 3a,b. Figure 3 Supp 1,2.

(8) L239, it is surprising that SpliceAI performs so badly, and might suggest a mistake in the analysis. Additional analysis and possible explanations should be provided to support these claims. Similarly, the complete failure of SpliceAI and Pangolin is shown in Figure 4d.

Line 239 refers to predicting relative inclusion levels between competing 3’ and 5’ splice sites. We admit we too expected this to be better for SpliceAI and Pangolin but we were not able to find bugs in our analysis (which is all made available for readers and reviewers alike). Regarding this expectation to perform better, first we note that we are not aware of a similar assessment being done for either of those algorithms (i.e. relative inclusion for 3’ and 5’ alternative splice site events). Instead, our initial expectation, and likely the reviewer’s as well, was based on their detection of splice site strengthening/weakening due to mutations, including cryptic splice site activation. More generally though, it is worth noting in this context that given how SpliceAI, Pangolin and other algorithms have been presented in papers/media/scientific discussions, we believe there is a potential misperception regarding tasks that SpliceAI and Pangolin excel at vs other tasks where they should not necessarily be expected to excel. Both algorithms focus on cryptic splice site creation/disruption. This has been the focus of those papers and subsequent applications.  While Pangolin added tissue specificity to SpliceAI training, the authors themselves admit “...predicting differential splicing across tissues from sequence alone is possible but remains a considerable challenge and requires further investigation”. The actual performance on this task is not included in Pangolin’s main text, but we refer Reviewer #2 to supplementary figure S4 in the Pangolin manuscript to get a sense of Pangolin’s reported performance on this task. Similar to that, Figure 4d in our manuscript is for predicting ‘tissue specific’ regulators. We do not think it is surprising that SpliceAI (tissue agnostic) and Pangolin (slight improvement compared to SpliceAI in tissue specific predictions) do not perform well on this task. Similarly, we do not find the results in Figure 4C surprising either. These are for mutations that slightly alter inclusion level of an exon, not something SpliceAI was trained on - SpiceAI was trained on genomic splice sites with yes/no labels across the genome. As noted elsewhere in our response, re-training Pangolin on this mutagenesis dataset results in performance much closer to that of TrASPr. That is to be expected as well - Pangolin is constructed to capture changes in PSI (or splice site usage as defined by the authors), those changes are not even tissue specific for the CD19 data and the model has no problem/lack of capacity to generalize from the training set just like TrASPr does. In fact, if you only use combinations of known mutations seen during training a simple regression model gives correlation of ~92-95% (Cortés-López et al 2022). In summary, we believe that better understanding of what one can realistically expect from models such as SpliceAI, Pangolin, and TrASPr will go a long way to have them better understood and used effectively. We have tried to make this more clear in the revision.

(9) BOS seems like a separate contribution that belongs in a separate publication. Instead, consider providing more details on TrASPr.

We thank the reviewer for the suggestion. We agree those are two distinct contributions/algorithms and we indeed considered having them as two separate papers. However, there is strong coupling between the design algorithm (BOS) and the predictor that enables it (TrASPr). This coupling is both conceptual (TrASPr as a “teacher”) and practical in terms of evaluations. While we use experimental data (experiments done involving Daam1 exon 16, CD19 exon 2) we still rely heavily on evaluations by TrASPr itself. A completely independent evaluation would have required a high-throughput experimental system to assess designs, which is beyond the scope of the current paper. For those reasons we eventually decided to make it into what we hope is a more compelling combined story about generative models for prediction and design of RNA splicing.

(10) The authors should consider evaluating BOS using Pangolin or SpliceTransformer as the oracle, in order to measure the contribution to the sequence generation task provided by BOS vs TrASPr.

We can definitely see the logic behind trying BOS with different predictors. That said, as we note above most of BOS evaluations are based on the “teacher”. As such, it is unclear what value replacing the teacher would bring. We also note that given this limitation we focus mostly on evaluations in comparison to existing approaches (genetic algorithm or random mutations as a strawman). 

Recommendations for the authors: 

Reviewer #1 (Recommendations for the authors):

Additional comments:

(1) Is your model picking up transcription factor binding sites in addition to RBPs? TFs have been recently shown to have a role in splicing regulation:

Daoud, Ahmed, and Asa Ben-Hur. "The role of chromatin state in intron retention: A case study in leveraging large scale deep learning models." PLOS Computational Biology 21.1 (2025): e1012755.

We agree this is an interesting point to explore, especially given the series of works from the Ben-Hur’s group. We note though that these works focus on intron retention (IR) which we haven’t focused on here, and we only cover short intronic regions flanking the exons. We leave this as a future direction as we believe the scope of this paper is already quite extensive.

(2) SpliceNouveau is a recently published algorithm for the splicing design problem:

Wilkins, Oscar G., et al. "Creation of de novo cryptic splicing for ALS and FTD precision medicine." Science 386.6717 (2024): 61-69.

Thank you for pointing out Wilkins et al recent publication, we now refer to it as well. 

(3) Please discuss the relationship between your model and this deep learning model. You will also need to change the following sentence: "Since the splicing sequence design task is novel, there are no prior implementations to reference."

We revised this statement and now refer to several recent publications that propose similar design tasks.  

(4) I would suggest adding a histogram of PSI values - they appear to be mostly close to 1 or 0.

PSI values are indeed typically close to either 0 or 1. This is a known phenomenon illustrated in previous studies of splicing (e.g. Shen et al NAR 2012 ). We are not sure what is meant by the comment to add a histogram but we made sure to point this out in the main text: 

“...Still, those statistics are dominated by extreme values, such that 33.2\% are smaller than 0.15 and 56.0\% are higher than 0.85. Furthermore, most cassette exons do not change between a given tissue pair (only 14.0\% of the samples in the dataset, \ie a cassette exon measured across two tissues, exhibit ΔΨ| ≥ 0.15).”

(5) Part of the improvement of TrASPr over Pangolin could be the result of a more extensive dataset.

Please see above responses and new analysis.

(6) In the discussion of the roles of alternative splicing, protein diversity is mentioned, but I suggest you also mention the importance of alternative splicing as a regulatory mechanism:

Lewis, Benjamin P., Richard E. Green, and Steven E. Brenner. "Evidence for the widespread coupling of alternative splicing and nonsense-mediated mRNA decay in humans." Proceedings of the National Academy of Sciences 100.1 (2003): 189-192.

Thank you for the suggestion. We added that point and citation. 

(7) Line 96: You use dPSI without defining it (although quite clear that it should be Delta PSI).

Fixed.

(8) Pretrained transformers: Have you trained separate transformers on acceptor and donor sites, or a single splice junction transformer?

Single splice junction pre-training.

(9) "TrASPr measures the probability that the splice site in the center of Se is included in some tissue" - that's not my understanding of what TrASPr is designed to do.

We revised the above sentence to make it more precise: “Given a genomic sequence context S_e = (s_e,...,s_e), made of  a cassette exon e and flanking intronic/exonic regions, TrASPr predicts for tissue c the fraction of transcripts where exon e is included or skipped over, ΔΨ-_e,c,c’.”

(10) Please include the version of the human genome annotations that you used. 

We used GENCODE v40 human genome hg38- this is now included in the Data section. 

(11) I did not see a description of the RBP-AE component in the methods section. A bit more detail on the model would be useful as well.

Please see above details about replacing RBP-AE with a simpler linear PCA “RBP-State” encoding. We added details about how the PCA was performed to the Methods section.

(12) Typos, grammar:

-   Fix the following sentence: ATP13A2, a lysosomal transmembrane cation transporter, linked to an early-onset form of Parkinson's Disease (PD) when 306 loss-of-function mutations disrupt its function.

Sentence was fixed to now read: “The first example is of a brain cerebellum-specific cassette exon skipping event predicted by TrASPr in the ATP13A2 gene (aka PARK9). ATP13A2 is a lysosomal transmembrane cation transporter, for which loss of function mutation has been linked to early-onset of Parkinson’s Disease (PD)”.

-   Line 501: "was set to 4e−4"(the - is a superscript). 

Fixed

-   A couple of citations are missing in lines 580 and 581.

Thank you for catching this error. Citations in line 580, 581 were fixed.

(13) Paper title: Generative modeling for RNA splicing predictions and design - it would read better as "Generative modeling for RNA splicing prediction and design", as you are solving the problems of splicing prediction and splicing design.  

Thank you for the suggestion. We updated the title and removed the plural form.

Reviewer #2 (Recommendations for the authors):

(1) Appendices are not very common in biology journals. It is also not clear what purpose the appendix serves exactly - it seems to repeat some of the things said earlier. Consider merging it into the methods or the main text. 

We merged the appendices into the Methods section and removed redundancy.

(2) L112, "For instance, the model could be tasked with designing a new version of the cassette exon, restricted to no more than N edit locations and M total base changes." How are N and M different? Is there a difference between an edit location and a base change? 

Yes, N is the number of locations (one can think of it as a start position) of various lengths (e.g. a SNP is of length 1) and the total number of positions edited is M. The text now reads “For instance, the model could be tasked with designing a new version of the cassette exon, restricted to no more than  $N$ edit locations (\ie start position of one or more consecutive bases) and $M$ total base changes.”

(3) L122: "DEN was developed for a distinct problem". What prevents one from adapting DEN to your sequence design task? The method should be generic. I do not see what "differs substantially" means here. (Finally, wasn't DEN developed for the task you later refer to as "alternative splice site" (as opposed to "splice site selection")? Use consistent terminology. And in L236 you use "splice site variation" - is that also the same?).

Indeed, our original description was not clear/precise enough. DEN was designed and trained for two tasks: APA, and 5’ alternative splice site usage. The terms “selection”, “usage”, and “variation” were indeed used interchangeably in different locations and the reviewer was right, noting the lack of precision. We have now revised the text to make sure the term “relative usage” is used. 

Nonetheless, we hold DEN was indeed defined for different tasks. See figures from Figure 2A, 6A of Linder et al 2020 (the reference was also incorrect as we cited the preprint and not the final paper):

In both cases DEN is trying to optimize a short region for selecting an alternative PA site (left) or a 5’ splice site (right). This work focused on an MPRA dataset of short synthetic sequences inserted in the designated region for train/test. We hold this is indeed a different type of data and task then the one we focus on here. Yes, one can potentially adopt DEN for our task, but this is beyond the scope of this paper. Finally, we note that a more closely related algorithm recently proposed is Ledidi (Schreiber et al 2025) which was posted as a pre-print. Similar to BOS, Ledidi tries to optimize a given sequence and adopt it with a few edits for a given task. Regardless, we updated the main text to make the differences between DEN and the task we defined here for BOS more clear, and we also added a reference to Ledidi and other recent works in the discussion section.

(4) L203, exons with DeltaPSI very close to 0.15 are going to be nearly impossible to classify (or even impossible, considering that the DeltaPSI measurements are not perfect). Consider removing such exons to make the task more feasible.

Yes, this is how it was done. As described in more details below, we defined changing samples as ones where the change was >= 0.15 and non-changing as ones where the change in PSI was < 0.05 to avoid ambiguous cases affecting the classification task.  

(5) L230, RBP-AE is not explained in sufficient detail (and does not appear in the methods, apparently). It is not clear how exactly it is trained on each new cellular condition.

Please see response in the opening of this document and Q11 from

Reviewer 1 

(6) L230, "significantly improving": the r value actually got worse; it is therefore not clear you can claim any significant improvement. Please mention that fact in the text.

This is a fair point. We note that we view the “a” statistic as potentially more interesting/relevant here as the Pearson “r” is dominated by points being generally close to 0/1.  Regardless, revisiting this we realized one can also make a point that the term “significant” is imprecise/misplaced since there is no statistical test done here (side note: given the amount of points, a simple null of same distribution yes/no would pass significance but we don’t think this is an interesting/relevant test here). Also, we note that with the transition to PCA instead of RBP-AE we actually get improvements in both a and r values, both for the ENCODE samples shown in Figure 3a and the two new GTEX tissues we tested (see above). We now changed the text to simply state: 

“...As shown in Figure 3a, this latent space representation allows TrSAPr to generalize from the six GTEX tissues to unseen conditions, including unseen GTEX tissues (top row), and ENCODE cell lines (bottom row). It improves prediction accuracy compared to TrASPr lacking PCA (eg a=88.5% vs a=82.3% for ENCODE cell lines), though naturally training on the additional GTEX and ENCODE conditions can lead to better performance  (eg a=91.7%, for ENCODE, Figure 3a left column).”

(7) L233, "Notably, previous splicing codes focused solely on cassette exons", Rosenberg et al. focused solely on alternative splice site choice.

Right - we removed that sentence.. 

(8) L236, "trained TrASPr on datasets for 3' and 5' splice site variations". Please provide more details on this task. What is the input to TrASPr and what is the prediction target (splice site usage, PSI of alternative isoforms)? What datasets are used for this task?

The data for this data was the same GTEx tissue data processed, just for alternative 3’ and 5’ splice sites events. We revised the description of this task in the main task and added information in the Methods section. The data is also included in the repo.

(9) L243, "directly from genomic sequences", and conservation?

Yes, we changed the sentence to read “...directly from genomic sequences combined with related features” 

(10) L262, what is the threshold for significant splicing changes?

The threshold is 0.15 We updated the main text to read the following:

The total number of mutations hitting each of the 1198 genomic positions across the 6106 sequences is shown in \FIG{mut_effect}b (left), while the distribution of effects ($|\Delta \Psi|$) observed across those 6106 samples is shown in \FIG{mut_effect}b (right). To this data we applied three testing schemes. The first is a standard 5-fold CV where 20\% of combinations of point mutations were hidden in every fold while the second test involved 'unseen mutation' (UM) where we hide any sample that includes mutations in specific positions for a total of 1480 test samples. As illustrated by the CDF in \FIG{mut_effect}b, most samples (each sample may involve multiple positions mutated) do not involve significant splicing changes. Thus, we also performed a third test using only  the 883 samples were mutations cause significant changes ($|\Delta \Psi|\geq 0.15 $). 

(11) L266, Pangolin performance is only provided for one of the settings (and it is not clear which). Please provide details of its performance in all settings.

The description was indeed not clear. Pangolin’s performance was similar to SpliceAI as mentioned above but retraining it on the CD19 data yielded much closer performance to TrASPr. We include all the matching tests for Pangolin after retraining in Figure 4 Supp Figure 1. 

(12) Please specify "n=" in all relevant plots. 

Fixed.

(13) Figure 3a, "The tissues were first represented as tokens, and new cell line results were predicted based on the average over conditions during training." Please explain this procedure in more detail. What are these tokens and how are they provided to the model? Are the cell line predictions the average of the predictions for the training tissues?

Yes, we compared to simply the average over the predictions for the training tissues for that specific event as baseline to assess improvements (see related work pointing for the need to have similar baselines in DL for genomics in https://pubmed.ncbi.nlm.nih.gov/33213499/). Regarding the tokens - we encode each tissue type as a possible value and feed the two tissues as two tokens to the transformer.

(14) Figure 4b, the total count in the histogram is much greater than 6106. Please explain the dataset you're using in more detail, and what exactly is shown here.

We updated the text to read: 

“...we used 6106 sequence samples where each sample may have multiple positions mutated (\ie mutation combinations) in exon 2 of CD19 and its flanking introns and exons (Cortes et al 2022). The total number of mutations hitting each of the 1198 genomic positions across the 6106 sequences is shown in Figure 4b (left).”

(15) Figure 5a, how are the prediction thresholds (TrASPr passed, TrASPr stringent, and TrASPr very stringent) defined?

Passed: dpsi>0.1, Stringent: dpsi>0.15, Very stringent: dpsi>0.2 This is now included in the main text.

(16) L417, please include more detail on the relative size of TrASPr compared to other models (e.g. number of parameters, required compute, etc.).

SpliceAI is a general-purpose splicing predictor with 32-layer deep residual neural network to capture long-range dependencies in genomic sequences. Pangolin is a deep learning model specifically designed for predicting tissue-specific splicing with similar architecture as SpliceAI. The implementation of SpliceAI that can be found here https://huggingface.co/multimolecule/spliceai involves an ensemble of 5 such models for a total of ~3.5M parameters. TrASPr, has 4 BERT transformers (each 6 layers and 12 heads) and MLP a top of those for a total of ~189M parameters. Evo 2, a genomic ‘foundation’ model has 40B parameters, DNABERT has ~86M (a single BERT with 12 layers and 12 heads), and Borzoi has 186M parameters (as stated in https://www.biorxiv.org/content/10.1101/2025.05.26.656171v2).  We note that the difference here is not just in model size but also the amount of data used to train the model. We edited the original L417 to reflect that.

(17) L546, please provide more detail on the VAE. What is the dimension of the latent representation?

We added more details in the Methods section like the missing dimension (256) and definitions for P(Z) and P(S). 

(18) Consider citing (and possibly comparing BOS to) Ghari et al., NeurIPS 2024 ("GFlowNet Assisted Biological Sequence Editing").

Added.

(19) Appendix Figure 2, and corresponding main text: it is not clear what is shown here. What is dPSI+ and dPSI-? What pairs of tissues are you comparing? Spearman correlation is reported instead of Pearson, which is the primary metric used throughout the text.

The dPSI+ and dPSI- sets were indeed not well defined in the original submission. Moreover, we found our own code lacked consistency due to different tests executed at different times/by different people. We apologize for this lack of consistency and clarity which we worked to remedy in the revised version. To answer the reviewer’s question, given two tissues ($c,c'$), dPSI+ and dPSI- is for correctly classifying the exons that are significantly differentially included or excluded. Specifically, differential included exons are those for which  $\Delta \Psi_{e,c1,c2} = \Psi_\Psi_{e,c1} - \Psi_{e,c2}  \geq 0.15$, compared to those that are not  ($\Delta \Psi_{e,c1,c2} < 0.05). Similarly, dPSI- is for correctly classifying the exons that are significantly differentially excluded in the first tissue or included in the second tissue ($\Delta \Psi_{e,c1,c2} = \Psi_\Psi_{e,c1} - \Psi_{e,c2}  \leq -0.15$) compared to those that are not  ($\Delta \Psi_{e,c1,c2} > -0.05). This means dPSI+ and dPSI- are dependent on the order of c1, c2. In addition, we also define a direction/order agnostic test for changing vs non changing events i.e. $|\Delta \Psi_{e,c1,c2}| \geq 0.15$ vs $|\Delta \Psi_{e,c1,c2}| < 0.05$. These test definitions are consistent with previous publications (e.g. Barash et al Nature 2010, Jha et al 2017) and also answer different biological questions: For example “Exons that go up in brain” and “Exons that go up in Liver” can reflect distinct mechanisms, while changing exons capture a model’s ability to identify regulated exons even if the direction of prediction may be wrong. The updated Appendix Figure 2 is now in the main text as Figure 2d and uses Pearson, while AUPRC and AUROC refer to the changing vs no-changing classification task described above such that we avoid dPSI+ and dPSI- when summarizing in this table over 3 pairs of tissues . Finally, we note that making sure all tests comply with the above definition also resulted in an update to Figure 2b/c labels and values, where TrASPr’s improvements over Pangolin reaches up to 1.8fold in AUPRC compared to 2.4fold in the earlier version. We again apologize for having a lack of clarity and consistent evaluations in the original submission.

(20) Minor typographical comments:

-   Some plots could use more polishing (e.g., thicker stroke, bigger font size, consistent style (compare 4a to the other plots)...).

Agreed. While not critical for the science itself we worked to improve figure polishing in the revision to make those more readable and pleasant. 

-   Consider using 2-dimensional histograms instead of the current kernel density plots, which tend to over-smooth the data and hide potentially important details. 

We were not sure what the exact suggestion is here and opted to leave the plots as is.

-   L53: dPSI_{e, c, c'} is never formally defined. Is it PSI_{e, c} - PSI_{e, c'} or vice versa?  

Definition now included (see above).

-   L91: Define/explain "transformer" and provide reference. 

We added the explanation and related reference of the transformer in the introduction section and BERT in the method section.  

-   L94: exons are short. Are you referring here to the flanking introns? Please explain. 

We apologize for the lack of clarity. We are referring to a cassette exon alternative splicing event as is commonly defined by the splice junctions involved that is from the 5’ SS of the upstream exon to the 3’ SS of the downstream exon. The text now reads:

“...In contrast, 24% of the cassette exons analyzed in this study span a region between the flanking exons' upstream 3' and downstream 5' splice sites that are larger than 10 kb.”

-   L132: It's unclear whether a single, shared transformer or four different transformers (one for each splice site) are being pre-trained. One would at least expect 5' and 3' splice sites to have a different transformer. In Methods, L506, it seems that each transformer is pre-trained separately. 

We updated the text to read:

“We then center a dedicated transformer around each of the splice sites of the cassette exon and its upstream and downstream (competing) exons (four separate transformers for four splice sites in total).”

-   L471: You explain here that it is unclear what tasks 'foundation' models are good for. Also in L128, you explain that you are not using a 'foundation' model. But then in L492, you describe the BERT model you're using as a foundation model! 

Line 492 was simply a poor choice of wording as “foundation” is meant here simply as the “base component”. We changed it accordingly.

-   L169, "pre-training ... BERT", explain what exactly this means. Is it using masking? Is it self-supervised learning? How many splice sites do you provide? Also explain more about the BERT architecture and provide references. 

We added more details about the BERT architecture and training in the Methods section.

-   L186 and later, the values for a and r provided here and in the below do not correspond to what is shown in Figure 2. 

Fixed, thank you for noticing this.

-   L187,188: What exactly do you mean by "events" and "samples"? Are they the same thing? If so, are they (exon, tissue) pairs? Please use consistent terminology. Moreover, when you say "changing between two conditions": do you take all six tissues whenever there is a 0.15 spread in PSI among them? Or do you take just the smallest PSI tissue and the largest PSI tissue when there is a 0.15 spread between them? Or something else altogether?

Reviewer #2 is yet again correct that the definitions were not precise. A “sample” involves a specific exon skipping “event” measured in two tissues.  The text now reads: 

“....most cassette exons do not change between a given tissue pair (only 14.0% of the samples in the dataset, i.e., a cassette exon measured across two tissues, exhibit |∆Ψ| ≥ 0.15). Thus, when we repeat this analysis only for samples involving exons that exhibited a change in inclusion (|∆Ψ| ≥ 0.15) between at least two tissues, performance degrades for all three models, but the differences between them become more striking (Figure 2a, right column).”

-   Figure 1a, explain the colors in the figure legend. The 3D effect is not needed and is confusing (ditto in panel C).

Color explanation is now added: “exons and introns are shown as blue rectangles and black lines. The blue dashed line indicates the inclusive pattern and the red junction indicates an alternative splicing pattern.” 

These are not 3D effects but stacks to indicate multiple events/cases. We agree these are not needed in Fig1a to illustrate types of AS and removed those. However, in Fig1c and matching caption we use the stacks to  indicate HT data captures many such LSVs over which ML algorithms can be trained. 

-   Figure 1b, this cartoon seems unnecessary and gives the wrong impression that this paper explores mechanistic aspects of splicing. The only relevant fact (RBPs serving as splicing factors) can be explained in the text (and is anyway not really shown in this figure).

We removed Figure 1b cartoon.

-   Figure 1c, what is being shown by the exon label "8"? 

This was meant to convey exon ID, now removed to simplify the figure. 

-   Figure 1e, left, write "Intron Len" in one line. What features are included under "..."? Based on the text, I did not expect more features.

Also, the arrows emanating from the features do not make sense. Is "Embedding" a layer? I don't think so. Do not show it as a thin stripe. Finally, what are dPSI'+ and dPSI'-? are those separate outputs? are those logits of a classification task?

We agree this description was not good and have updated it in the revised version. 

-   Figure 1e, the right-hand side should go to a separate figure much later, when you introduce BOS.

We appreciate the suggestion. However, we feel that Figure 1e serves as a visual representation of the entire framework. Just like we opted to not turn this work into two separate papers (though we fully agree it is a valid option that would also increase our publication count), we also prefer to leave this unified visual representation as is.

-   Figure 2, does the n=2456 refer to the number of (exons, tissues) pairs? So each exon contributes potentially six times to this plot? Typo "approximately". 

The “n” refers to the number of samples which is a cassette event measured in two tissues. The same cassette event may appear in multiple samples if it was confidently quantified in more than two tissues. We updated the caption to reflect this and corrected the typo.

-   Figure 2b, typo "differentially included (dPSI+) or excluded" .

Fixed.

-   L221, "the DNABERT" => "DNABERT".

Fixed.

-   L232, missing percent sign.

-    

Fixed.

-   L246, "see Appendix Section 2 for details" seems to instead refer to the third section of the appendix.

We do not have this as an Appendix, the reference has been updated.

-   Figure 3, bottom panels, PSI should be "splice site usage"? 

PSI is correct here - we hope the revised text/definitions make it more clear now.

-   Figure 3b: typo: "when applied to alternative alternative 3'".

Fixed.

-   p252, "polypyrimidine" (no capitalization).

Fixed.

-   Strange capitalization of tissue names (e.g., "Brain-Cerebellum"). The tissue is called "cerebellum" without capitalization.

We used EBV (capital) for the abbreviation and lower case for the rest.

-   Figure 4c: "predicted usage" on the left but "predicted PSI" on the right. 

Right. We opted to leave it as is since Pangolin and SpliceAI do predict their definition of “usage” and not directly PSI, we just measure correlations to observed PSI as many works have done in the past. 

-   Figure 4 legend typo: "two three".

Fixed.

-   L351, typo: "an (unsupervised)" (and no need to capitalize Transformer).

Fixed.

-   L384, "compared to other tissues at least" => "compared to other tissues of at least".

Fixed.

-   L549, P(Z) and P(S) are not defined in the text.

Fixed.

-   L572, remove "Subsequently". Add missing citations at the end of the paragraph.

Fixed.

-   L580-581, citations missing.

Fixed.

-   L584-585, typo: "high confidince predictions"

Fixed.

-   L659-660, BW-M and B-WM are both used. Typo?

Fixed.

-   L895, "calculating the average of these two", not clear; please rewrite.

Fixed.

-   L897, "Transformer" and "BERT", do these refer to the same thing? Be consistent.  

BOS is a transformer and not a BERT but TrASPr uses the BERT architecture. BERT is a type of transformer as the reviewer is surely well aware so the sentence is correct. Still, to follow the reviewer’s recommendation for consistency/clarity we changed it here to state BERT.

-   Appendix Figure 5: The term dPSI appears to be overloaded to also represent the difference between predicted PSI and measured PSI, which is inconsistent with previous definitions. 

Indeed! We thank the reviewer again for their sharp eye and attention to details that we missed. We changed Supp Figure 5, now Figure 4 Supplementary Figure 2, to |PSI’-PSI| and defined those as the difference between TrASPr’s predictions (PSI’) and MAJIQ based PSI quantifications.

Manque un verbe.

Manque un verbe.

Visual Studio Code n'accepte pas "event" mais accepte "Event" (en tout cas sur mon pc c'est comme ça).

Joint Public Review:

Summary:

This is an excellent, timely study investigating and characterizing the underlying neural activity that generates the neuroendocrine GnRH and LH surges that are responsible for triggering ovulation. Abundant evidence accumulated over the past 20 years implicated the population of kisspeptin neurons in the hypothalamic RP3V region (also referred to as the POA or AVPV/PeN kisspeptin neurons) as being involved in driving the GnRH surge in response to elevated estradiol (E2), also known as the "estrogen positive feedback". However, while former studies used Cfos coexpression as a marker of RP3V kisspeptin neuron activation at specific times and found this correlates with the timing of the LH surge, detailed examination of the live in vivo activity of these neurons before, during, and after the LH surge remained elusive due to technical challenges.

Here, Zhou and colleagues use fiber photometry to measure the long-term synchronous activity of RP3V kisspeptin neurons across different stages of the mouse estrous cycle, including on proestrus when the LH surge occurs, as well as in a well-established OVX+E2 mouse model of the LH surge.

The authors report that RP3V kisspeptin neuron activity is low on estrous and diestrus, but increases on proestrus several hours before the late afternoon LH surge, mirroring prior reports of rising GnRH neuron activity in proestrus female mice. The measured increase in RP3V kisspeptin activation is long, spanning ~13 hours in proestrus females and extending well beyond the end of the LH secretion, and is shown by the authors to be E2 dependent.

For this work, Kiss-Cre female mice received a Cre-dependent AAV injection, containing GCaMP6, to measure the neuronal activation of RP3V Kiss1 cells. Females exhibited periods of increased neuronal activation on the day of proestrus, beginning several hours prior to the LH surge and lasting for about 12 hours. Though oscillations in the pattern of GCaMP fluorescence were occasionally observed throughout the ovarian cycle, the frequency, duration, and amplitude of these oscillations were significantly higher on the day of proestrus. This increase in RP3V Kiss1 neuronal activation that precedes the increase in LH supports the hypothesis that these neurons are critical in regulating the LH surge. The authors compare this data to new data showing a similar increased activation pattern in GnRH neurons just prior to the LH surge, further supporting the hypothesis that RP3V Kiss1 cell activation causes the release of kisspeptin to stimulate GnRH neurons and produce the LH surge.

Strengths:

This study provides compelling data demonstrating that RP3V kisspeptin neuronal activity changes throughout the ovarian cycle, likely in response to changes in estradiol levels, and that neuronal activation increases on the day of the LH surge.

The observed increase in RP3V kisspeptin neuronal activation precedes the LH surge, which lends support to the hypothesis that these neurons play a role in regulating the estradiol-induced LH surge. Continuing to examine the complexities of the LH surge and the neuronal populations involved, as done in this study, is critical for developing therapeutic treatments for women's reproductive disorders.

This innovative study uses a within-subject design to examine neuronal activation in vivo across multiple hormone milieus, providing a thorough examination of the changes in activation of these neurons. The variability in neuronal activity surrounding the LH surge across ovarian cycles in the same animals is interesting and could not be achieved without this within-subjects design. The inclusion and comparison of ovary-intact females and OVX+E2 females is valuable to help test mechanisms under these two valuable LH surge conditions, and allows for further future studies to tease apart minor differences in the LH surge pattern between these 2 conditions.

This study provides an excellent experimental setup able to monitor the daily activity of preoptic kisspeptin neurons in freely moving female mice. It will be a valuable tool to assess the putative role of these kisspeptin neurons in various aspects of altered female fertility (aging, pathologies...). This approach also offers novel and useful insights into the impact of E2 and circadian cues on the electrical activity of RP3V kisspeptin neurons.

An intriguing cyclical oscillation in kisspeptin neural activity every 90 minutes exists, which may offer critical insight into how the RP3V kisspeptin system operates. Interestingly, there was also variability in the onset and duration of RP3V Kisspeptin neuron activity between and within mice in naturally cycling females. Preoptic kisspeptin neurons show an increased activity around the light/dark transition only on the day of proestrus, and this is associated with an increase in LH secretion. An original finding is the observation that the peak of kisspeptin neuron activation continues a few hours past the peak of LH, and the authors hypothesize that this prolonged activity could drive female sexual behaviors, which usually appear after the LH surge.

The authors demonstrated that ovariectomy resulted in very little neuronal activity in RP3V kisspeptin neurons. When these ovarietomized females were treated with estradiol benzoate (EB) and an LH surge was induced, there was an increase in RP3V kisspeptin neuronal activation, as was seen during proestrus. However, the magnitude of the change in activity was greater during proestrus than during the EB-induced LH surge. Interestingly, the authors noted a consistent peak in activity about 90 minutes prior to lights out on each day of the ovarian cycle and during EB treatment, but not in ovariectomized females. The functional purpose of this consistent neuronal activity at this time remains to be determined.

Though not part of this study, the comparison of neuronal activation of GnRH neurons during the LH surge to the current data was convincing, demonstrating a similar pattern of increased activation that precedes the LH surge.

In summary, the study is well-designed, uses proper controls and analyses, has robust data, and the paper is nicely organized and written. The data from these experiments is compelling, and the authors' claims and conclusions are nicely supported and justified by the data. The data support the hypothesis in the field that these RP3V neurons regulate the LH surge. Overall, these findings are important and novel, and lend valuable insight into the underlying neural mechanisms for neuroendocrine control of ovulation.

Weaknesses:

(1) LH levels were not measured in many mice or in robust temporal detail, such as every 30 or 60 min, to allow a more detailed comparison between the fine-scale timing of RP3V neuron activation with onset and timing of LH surge dynamics.

(2) The authors report that the peak LH value occurred 3.5 hours after the first RP3V kisspeptin neuron oscillation. However, it is likely, and indeed evident from the 2 example LH patterns shown in Figures 3A-B, that LH values start to increase several hours before the peak LH. This earlier rise in LH levels ("onset" of the surge) occurs much closer in time to the first RP3V kisspeptin neuron oscillatory activation, and as such, the ensuing LH secretion may not be as delayed as the authors suggest.

(3) The authors nicely show that there is some variation (~2 hours) in the peak of the first oscillation in proestrus females. Was this same variability present in OVX+E2 females, or was the variability smaller or absent in OVX+E2 versus proestrus? It is possible that the variability in proestrus mice is due to variability in the timing and magnitude of rising E2 levels, which would, in theory, be more tightly controlled and similar among mice in the OVX+E2 model. If so, the OVX+E2 mice may have less variability between mice for the onset of RP3V kisspeptin activity.

(4) One concern regarding this study is the lack of data showing the specificity of the AAV and the GCaMP6s signals. There are no data showing that GCaMP6s is limited to the RP3V and is not expressed in other Kiss1 populations in the brain. Given that 2ul of the AAV was injected, which seems like a lot considering it was close to the ventricle, it is important to show that the signal and measured activity are specific to the RP3V region. Though the authors discuss potential reasons for the low co-expression of GCaMP6 and kisspeptin immunoreactivity, it does raise some concern regarding the interpretation of these results. The low co-expression makes it difficult to confirm the Kiss1 cell-specificity of the Cre-dependent AAV injections. In addition, if GFP (GCaMP6s) and kisspeptin protein co-localization is low, it is possible that the activation of these neurons does not coincide with changes in kisspeptin or that these neurons are even expressing Kiss1 or kisspeptin at the time of activation. It is important to remember that the study measures activation of the kisspeptin neuron, and it does not reveal anything specific about the activity of the kisspeptin protein.

(5) One additional minor concern is that LH levels were not measured in the ovariectomized females during the expected time of the LH surge. The authors suggest that the lower magnitude of activation during the LH surge in these females, in comparison to proestrus females, may be the result of lower LH levels. It's hard to interpret the difference in magnitude of neuronal activation between EB-treated and proestrus females without knowing LH levels. In addition, it's possible that an LH surge did not occur in all EB-treated females, and thus, having LH levels would confirm the success of the EB treatment.

(6) This kisspeptin neuron peak activity is abolished in ovariectomized mice, and estradiol replacement restored this activity, but only partially. Circulating levels of estradiol were not measured in these different setups, but the authors hypothesize that the lack of full restoration may be due to the absence of other ovarian signals, possibly progesterone.

(7) Recordings in several mice show inter- and intra-variability in the time of peak onset. It is not shown whether this variability is associated with a similar variability in the timing of the LH surge onset in the recorded mice. The authors hypothesized that this variability indicates a poor involvement of the circadian input. However, no experiments were done to investigate the role of the (vasopressinergic-driven) circadian input on the kisspeptin neuron activation at the light/dark transition. Thus, we suggest that the authors be more tentative about this hypothesis.

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

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary:

Liu et al. provided evidence of the interaction between endocytosis and VAMP8-mediated endocytic recycling of clathrin-mediated endocytosis (CME) cargo through a knockdown approach combined with total internal reflection fluorescence (TIRF) microscopy, western blotting, and functional assays in a mammalian cell line system. They demonstrated that VAMP8 impairs the initial stages of CME, such as the initiation, stabilization, and invagination of clathrin-coated pits (CCPs). VAMP8 indirectly regulates CME by facilitating endocytic recycling. The depletion of VAMP8 alters endosomal recycling, as shown here by the transferrin receptor, towards lysosomal degradation, thereby inhibiting clathrin-coated vesicle (CCV) formation. Overall, I found this study to be highly engaging because of its elucidation of the unexpected role of R-Snare in influencing the levels of cargo proteins within the context of clathrin-mediated endocytosis (CME). This MS will be helpful for researchers in endocytosis and protein trafficking fields. It appears to me that VAMP8 interacts with multiple targets within the endo-lysosomal pathway, collectively influencing the clathrin-mediated endocytosis (CME). Therefore, the contribution of lysosomes in this context should be evaluated. This matter should be addressed experimentally and discussed in the MS before considering publication.

Major comments:

1. Figure 4D demonstrates that the knockdown of VAMP8 leads to an increase in lysosome numbers and lysosomal perinuclear clustering, as evidenced by LAMP1 staining (Figure 5A). Additionally, the knockdown of VAMP8 results in the downregulation of most surface receptors, as illustrated in Figure 3A, which typically follows the lysosomal degradation pathway. The observed reduction in TfR cargo could be attributable to the decreased presence of the Tfn Receptor in siVAMP8-treated cells compared to that in control cells. How do the authors explain this phenomenon? Upon reviewing these observations, I suggest that the mechanism outlined in the manuscript-specifically, "Depletion of VAMP8 skews endosomal recycling of CME cargo, exemplified here by transferrin receptor, toward lysosomal degradation, thereby inhibiting CCV formation"-may serve as a secondary rather than a primary cause. This can be ruled out by the following experiments:
   • Assessment of lysosomal biogenesis markers through RT-PCR or Western blotting following VAMP8 knockdown.
   • Assessment of transferrin receptor stability under VAMP8 knockdown conditions using cycloheximide.
   • Previous studies have indicated that perinuclear clustering of lysosomes is correlated with increased degradative activity. Therefore, assessing the lysosomal perinuclear index in the images presented in Figure 5A (LAMP1) effectively determines the presence or absence of this phenomenon.
2. Given that VAMP8 is implicated in lysosomal fusion events, I hypothesized that VAMP8 undergoes degradation via the lysosomal pathway. However, Figure 4F indicates that there was no restoration of VAMP8 following leupeptin treatment. Could you please provide an explanation for this discrepancy or is it trafficked to proteasomal degradation pathway?
3. Figure 5A and 5C demonstrate that the restoration of TfnR in siVAMP8 under leupeptin conditions was similar to the levels observed in the sicontrol without leupeptin. However, no enhancement in TfnR uptake (Figure 5F) was detected in cells treated with siVAMP8 under leupeptin treatment conditions. How can these observations be reconciled with each other?

Minor comments:

1. The manuscript does not provide details of the western blotting method and quantification criteria.
2. Fig1A &B) - The siVAMP8 #1 blot indicates a reduction exceeding 90%, whereas the bar graph depicts a reduction of 70-80%. It is advisable to elucidate the quantification criteria in the Methods section to prevent potential confusion. Were the protein levels normalized to the loading control?
3. Enhancing the readability of the graph could be achieved by labeling the Y-axis as either 'All CCP' or 'Bonafide CCP' of CME analysis graphs.
4. The legends of panels 1M and N do not correlate with the corresponding figures. Need corrections.
5. Fig 4D- Is the technique employed for electron immunogold staining utilizing a lysosome-specific antibody? How do the authors substantiate their assertion that the darkly stained structures are lysosomes and not other cellular compartments?
6. Electron micrographs of siVAMP8 cells revealed the presence of dark-stained bodies near the plasma membrane. The implications of this observation should be explained in the discussion section.
7. Fig5A- Provide the color code for the merged images.
8. Fig5G- schematic needs to be improved to demonstrate the contribution of increased lysosomal content.

Significance

VAMP8 is an R-SNARE critical for late endosome/lysosome fusion and regulates exocytosis, especially in immune and secretory cells. It pairs with Q-SNAREs to mediate vesicle fusion, and its dysfunction alters immunity, inflammation, and secretory processes. This study revealed that the SNARE protein VAMP8 influences clathrin-mediated endocytosis (CME) by managing the recycling of endocytic cargo rather than being directly recruited to clathrin-coated vesicles. This study advances our understanding of cellular trafficking mechanisms and underscores the essential role of recycling pathways in maintaining membrane dynamics. This is an excellent piece of work, and the experiments were designed meticulously; however, the mechanism is not convincing enough at this point. This MS will surely benefit the general audience, specifically the membrane and protein trafficking and cell biology community.

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

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Liu and colleagues show that the knockdown of VAMP8 impairs CME by downregulating various cargo receptors, including TfnR, by rerouting theses receptors to lysosomes for degradation instead of recycling them to the plasma membrane. The results also imply that the lack of sufficient receptors (CME-cargo and associated endocytic machinery) in turn impairs the initiation/stabilization of nascent CCPs and their subsequent invagination. As shown by specific mutations in CALM, VAMP8 is apparently not directly required for CME. The emloyed cmeAnalysis DASC assay appears to be state of the art. The data are overall convincing. Nevertheless, the authors should address/clarify the following points:

Major comments:

• A rescue experiment of the VAMP8 knockdown using VAMP8 and RFP-VAMP8 should be included to exclude off-target effects and demonstrate the functionality of the RFP-VAMP8 construct.
• Please confirm that the RFP-VAMP8 expression levels in the CALM(WT) and CALM(SNARE) cells are comparable (compare first panels in Fig. 2A and 2B) and provide information about the RFP-VAMP8 expression levels compared to endogenous VAMP8. Figure 2D shows that RFP-VAMP8 is not enriched in CCPs in CALM(SNARE) cells. This raises the questions whether endocytic vesicles in the CALM(SNARE*) background indeed lack VAMP8 or still contain some residual VAMP8 levels. A complete lack of VAMP8 would imply that VAMP8 does not play a major role in determining the fate (fusion partner) of the endocytic vesicles (in the pathways analyzed by the authors). If possible, provide experimental data to solve this issue or discuss this point.

Minor comments:

• Fig. 1 N and M: The figure panels should be switched to fit to the legend, or vice versa.
• In contrast to Table S1, which show a reduction of TfnR by a factor of 1.8, the Western blot analysis (Fig. 3C) shows a 4-fold reduction. Please explain the divergence.
• It is surprising that the TfnR knockdown phenocopies the VAMP8 knockdown. Why does the knockdown of a single receptor affect endocytosis, measured by the eGFP-CLCa recruitment? Compared to other plasma membrane receptors, how abundant is TnfR? If available, please provide references demonstrating that the knockdown of other receptors has similar effects on endocytosis?
• The authors should briefly discuss to which degree the knockdown of VAMP8 may also affect receptor exocytosis, thereby contributing to a reduction of cargo receptors at the plasma membrane and impaired CME.
• VAMP8 has an established role in autophagosome - lysosome flux, favoring the fusion with the lysosomes. In the present study, VAMP8 knockdown seems to reroute receptors for lysosomal degradation in the absence of VAMP8. Please discuss.
• For clarity, the authors may consider to restructure their abstract, directly starting with their finding that "Depletion of VAMP8 skews endosomal recycling of CME cargo, exemplified by ..........

Significance

Overall, this study provides significant insights into the role of VAMP8 in the recycling of receptors to the plasma membrane. The lack of VAMP8 results in rerouting of plasma membrane receptors to lysosomes and thereby indirectly reduces endocytosis. The results will be of broad interest in the field of membrane trafficking. The reviewers field of expertise is membrane trafficking, in particular molecular mechanisms of exocytosis.

Il n'y a, bien sûr, aucun rapport entre l'espace vital nécessaire à la démographie allemande et l'ambition de Trump de s'approprier un espace stratégique. Folie progressiste dégénérée, lamentable fantasme d'Européen débile impuissant.

Author response:

The following is the authors’ response to the previous reviews

Reviewer 1

The authors should clarify the statement regarding the expression in horizontal cells (lines 170-172). In line 170, it is stated that GFP was observed in horizontal cells. Since GFP is fused to Cx36, the observation of GFP in horizontal cells would suggest the expression of Cx36-GFP.

We believe that there appears to be a misunderstanding. GFP is observed in horizontal cells, because the test AAV construct, which consists of the HKamac promoter and a downstream GFP sequence, was used to validate the promoter specificity in wildtype animals. This was just a test to confirm that HKamac is indeed active in AII amacrine cells as previously described by Khabou et al. 2023. This construct was not used for the large scale BioID screen. For these experiments, V5-dGBP-Turbo was expressed under the control of the HKamac promoter as illustrated in Figure 2A.

Fig 7: the legend is missing the descriptions for panels A-C.

We apologize for this mistake. We have missed the label “(A-C)” and added it to the legend.

Supplemental files are not referenced in the manuscript.

We have added a reference for these files in line 221-226.

Reviewer 2

Supplementary Files 1 and 2 are presented as two replicates of the zebrafish proteomic datasets, but they appear to be identical.

This appears to be a misunderstanding. These two replicates contain slightly different hits, although the most abundant candidates are identical.

Reviewer 3

Thank you for the positive comments

Reviewer #1 (Public review):

Summary:

The authors note that there is a large corpus of research establishing the importance of LC-NE projections to medial prefrontal cortex (mPFC) of rats and mice in attentional set or 'rule' shifting behaviours. However, this is complex behavior and the authors were attempting to gain an understanding of how locus coeruleus modulation of the mPFC contributes to set shifting.

The authors replicated the ED-shift impairment following NE denervation of mPFC by chemogenetic inhibition of the LC. They further showed that LC inhibition changed the way neurons in mPFC responded to the cues, with a greater proportion of individual neurons responsive to 'switching', but the individual neurons also had broader tuning, responding to other aspects of the task (i.e., response choice and response history). The population dynamics was also changed by LC inhibition, with reduced separation of population vectors between early-post-switch trials, when responding was at chance, and later trials when responding was correct. This was what they set out to demonstrate and so one can conclude they achieved their aims.

The authors concluded that LC inhibition disrupted mPFC "encoding capacity for switching" and suggest that this "underlie[s] the behavioral deficits."

Strengths:

The principal strength is combining inactivation of LC with calcium imaging in mPFC. This enabled detailed consideration of the change in behavior (i.e., defining epochs of learning, with an 'early phase' when responding is at chance being compared to a 'later phase' when the behavioral switch has occurred) and how these are reflected in neuronal activity in the mPFC, with and without LC-NE input.

Comments on revised version:

In their response to reviewers, the authors say "We report p values using 2 decimal points and standard language as suggested by this reviewer". However, no changes were made in the manuscript: for example, "P = 4.2e-3" rather than "p = 0.004".

In their response to the reviewers, they wrote: "Upon closer examination of the behavioral data, we exclude several sessions where more trials were taken in IDS than in EDS." If those sessions in which EDSIDS. Most problematic is the fact that the manuscript now reads "Importantly, control mice (pooled from Fig. 1e, 1h, Supp. Fig. 1a, 1b) took more trials to complete EDS than IDS (Trials to criterion: IDS vs. EDS, 10 {plus minus} 1 trials vs. 16 {plus minus} 1 trials, P < 1e-3, Supp. Fig. 1c), further supporting the validity of attentional switching (as in Fig. 1c)" without mentioning that data has been excluded.

Author response:

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

Public Reviews:

We thank the reviewers and editors for this peer review. Following the editorial assessment and specific review comments, in this revision we have included new analysis to support the validity of the behavioral task (Reviewer #2). We have improved data presentation by including 1) data points from individual animals (Reviewer #1, #3), 2) updated histology showing the expression of hM4Di in LC neurons as well as LC terminals in the mPFC (Reviewer #3), and 3) more detailed descriptions of methodology and data analysis (Reviewer #1, #2, #3).

Recommendations for the authors:

Reviewer #2 (Recommendations for the authors):

(1) Planned t-tests should be performed in both control and experimental animals to determine if the number of trials needed to reach criterion on the ID is lower than on the ED. Based on the data analyses showing no difference among the control group, the data could be pooled to demonstrate that the task is valid. Reporting all p-values using 2 decimal points and standard language e.g., p < 0.001 would greatly improve the readability of the data. 

Thank you for this suggestion. As pointed out by this reviewer, more trials to reach performance criterion in EDS than IDS is indicative of successful acquisition and switching of the attentional sets. Upon closer examination of the behavioral data, we exclude several sessions where more trials were taken in IDS than in EDS, and our conclusions that DREADD inhibition of the LC or LC input to the mPFC impaired rule switching in EDS remain robust (e.g., new Fig. 1e, 1h). We also pool control and test data (Fig. 1e, 1h, new Supp. Fig. 1a, 1b) to demonstrate the validity of this task (new Supp. Fig. 1c, IDS vs. EDS in the control group, 10 ± 1 trials vs. 16 ± 1 trials, P < 1e-3). The validity of set shifting is also supported by the new Fig. 1c.  

We report p values using 2 decimal points and standard language as suggested by this reviewer.

Relevant to the comments from Reviewer #1 in the public review, we now show individual data points on the bar charts (new Fig. 1e, 1h).  

(2) It may also be helpful to provide the average time between CNO infusion and onset of the ED as well as information about when maximal effects are expected after these treatments.

Systemic CNO injections were administered immediately after IDS, and we waited approximately one hour before proceeding to EDS. Maximal effects of systemic CNO activation were reported to occur after 30 minutes and last for at least 4-6 hours. Both control and test groups received the CNO injections in the same manner. This is now better described in Methods.  

Reviewer #3 (Recommendations for the authors):

(1) Add better histology images showing colocalization of TH and HM4Di. Quantification of colocalization would be optimal.

We now include better histology images (new Fig. 1d) and have quantified the colocalization of TH and HM4Di in the main text (line 115-116).  

(2) If possible, images showing HM4Di expression in mPFC axon terminals would be useful. If these are colocalized with TH immunostaining, that would increase confidence in their identity. This would be much more useful than the images provided in Figure 1C.

We now include new image to show hM4Di expression (mCherry) in LC terminals in the mPFC (new Fig. 1f). However, due to technical limitations (species of the primary antibody), we did not co-stain with TH.

(3) Include behavior of mice from the miniscope experiment in Figure 2 to show they are similar to those from Figure 1.

This is now included in Supp. Fig. 1b.

(4) More details about the processing and segmentation of miniscope data would be helpful (e.g., how many neurons were identified from each animal?). 

We use standard preprocessing and segmentation pipelines in Inscopix data processing software (version 1.6), which includes modules for motion correction and signal extraction. Briefly, raw imaging videos underwent preprocessing, including a x4 spatial down sampling to reduce file size and processing time. No temporal down sampling was performed. The images were then cropped to eliminate post-registration borders and areas where cells were not visible. Prior to the calculation of the dF/F0 traces, lateral movement was corrected. For ROI identification, we used a constrained non-negative matrix factorization algorithm optimized for endoscopic data (CNMF-E) to extract fluorescence traces from ROIs. We identified 128 ± 31 neurons after manual selection, depending on recording quality and field of view. Number of neurons acquired from each animal are now included in Methods. This is now further elaborated in Methods (line 405415).  

(5) Add more methodological detail for how cell tuning was analyzed, including how z-scoring was performed (across the entire session?), and how neurons in each category were classified. 

We have expanded the Methods section to clarify how cell tuning was analyzed (line 419430). Calcium traces were z-scored on a per-neuron basis across the entire session. For each neuron, we computed trial-averaged activity aligned to specific task events (e.g., digging in one of the two ramekins available). A neuron was classified as responsive if its activity showed a significant difference (p < 0.05) between two conditions within the defined time window in the ROC analysis.

(6) For data from Figure 2F it would be very useful to plot data from individual mice in addition to this aggregated representation.

We now include data from individual mice in Supp. Table 1.

(7) I think it would be helpful to move some parts of Figure S1 to the main Figure 1, in particular the table from S1A. 

Fig. S1 is now part of the new Fig. 1.

(8) Clarify whether Figure S2 is an independent replication, as implied, or whether the same test data is shown twice in two separate figures (In Figure 1b and Supplementary Figure 2).

The test group in Fig. S2 (new Fig. S1) is the same as the test group in Fig. 1b (new Fig. 1e), but the control group is a separate cohort. This is now clarified in the figure legends.  

(9) The authors should add a limitations section to the discussion where they specifically discuss the caveats involved in relating their results specifically to NE. This should include the possible involvement of co-transmitters and off-target expression of Cre in other populations.

Thank you for this comment. Previous pharmacology and lesion studies showed that LC input or NE content in the mPFC was specifically required for EDS-type switching processes (Lapiz, M.D. et al., 2006; Tait, D.S. et al. 2007; McGaughy, J. et al. 2008), in light of which we interpret our mPFC neurophysiological effects with LC inhibition as at least partially mediated by the direct LC-NE input.  When discussing the limitations of our study, we now explicitly acknowledge the potential involvement of co-transmitters released by LC neurons (line 253-256).  

(10) The authors should provide details about the TH antibody uses for IHC

We now include more details in immunohistochemistry (line 384-388).

(11) Throughout, it would be helpful to include datapoints from individual animals - these are included in some supplementary figures, but are missing in a number of the main plots.

Reviewer #1 made a similar comment, and we now include individual data points in the figures (e.g., Fig. 1e, 1h).

Author response:

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

eLife Assessment

This study introduces a novel method for estimating spatial spectra from irregularly sampled intracranial EEG data, revealing cortical activity across all spatial frequencies, which supports the global and integrated nature of cortical dynamics. The study showcases important technical innovations and rigorous analyses, including tests to rule out potential confounds; however, the lack of comprehensive theoretical justification and assumptions about phase consistency across time points renders the strength of evidence incomplete. The dominance of low spatial frequencies in cortical phase dynamics continues to be of importance, and further elaboration on the interpretation and justification of the results would strengthen the link between evidence and conclusions.

Public Reviews:

Reviewer #1 (Public review):

Summary:

The paper uses rigorous methods to determine phase dynamics from human cortical stereotactic EEGs. It finds that the power of the phase is higher at the lowest spatial phase.

Strengths:

Rigorous and advanced analysis methods.

Weaknesses:

The novelty and significance of the results are difficult to appreciate from the current version of the paper.

(1) It is very difficult to understand which experiments were analysed, and from where they were taken, reading the abstract. This is a problem both for clarity with regard to the reader and for attribution of merit to the people who collected the data.

We now explicitly state the experiments that were used, lines 715-716.

(2) The finding that the power is higher at the lowest spatial phase seems in tune with a lot of previous studies. The novelty here is unclear and it should be elaborated better.

It is not generally accepted in neuroscience that power is higher at lowest spatial frequencies, and recent research concludes that traveling waves at this scale may be the result of artefactual measurement (Orczyk et al., 2022; Hindriks et al., 2014; Zhigalov & Jensen,2023). The question we answer is therefore timely and a source of controversy to researchers analysing TWs in cortex. While, in our view, the previous literature points in the direction of our conclusions (notably the work of Freeman et. al. 2003; 2000; Barrie et al. 1996), it is not conclusive at the scale we are interested in, specifically >8cm, and certainly not convincing to the proponents of ‘artefactual measurement’.

We have added to a sentence to make this explicit in the abstract, lines 20-22. Please also note previous text at the end of the introduction, lines 140-148 and in the first paragraph of the discussion, lines 563-569.

I could not understand reading the paper the advantage I would have if I used such a technique on my data. I think that this should be clear to every reader.

We have made the core part of the code available on github (line 1154), which should simplify adoption of the technique. We have urged, in the Discussion (lines 653-663), why habitual measurement of SF spectra is desirable, since the same task measured with EEG, sEEG or ECoG does not encompass the same spatial scales, and researchers may be comparing signals with different functional properties. Until reliable methods for estimating SF are available, not dependent on the layout of the recording array, data cannot be analysed to resolve this question. Publication of our results and methods will help this process along.

(3) It seems problematic to trust in a strong conclusion that they show low spatial frequency dynamics of up to 15-20 cm given the sparsity of the arrays. The authors seem to agree with this concern in the last paragraph of page 12. 

The new surrogate testing supports our conclusions. The sEEG arrays would not normally be a first choice to estimate SF spectra, for reasons of their sparsity, which may be why such estimates have not been done before. Yet, this is the research challenge that we sought to solve, and a problem for which there was no ready method to hand. Nevertheless, it is a problem that urgently needed to be solved given the current debate on the origin of large-scale TWs. We have now included detailed surrogate testing of real data plus varying strength model waves (Figure 6A and Supplementary Figure 4). We believe this should convince the reader that we are measuring the spatial frequency spectrum with sufficient accuracy to answer the central research question.

They also say that it would be informative to repeat the analyses presented here after the selection of more participants from all available datasets. It begs the question of why this was not done. It should be done if possible.

We have now doubled the number of participants in the main analyses. Since each participant comprises a test of the central hypothesis, now the hypothesis test now has 23 replications (Supplementary Figures 2 and 3). There were four failures to reach significance due to under-powered tests, i.e., not enough contacts. This is sufficient test of the hypothesis and, in our opinion, not the primary obstacle to scientific acceptance of our results. The main obstacle is providing convincing tests that the method is accurate, and this is what we have focussed on. Publication of python code and the detailed methods described here enable any interested researcher to extend our method to other datasets.

(4) Some of the analyses seem not to exploit in full the power of the dataset. Usually, a figure starts with an example participant but then the analysis of the entire dataset is not as exhaustive. For example, in Figure 6 we have a first row with the single participants and then an average over participants. One would expect quantifications of results from each participant (i.e. from the top rows of GFg 6) extracting some relevant features of results from each participant and then showing the distribution of these features across participants. This would complement the subject average analysis.

The results are now clearly split into sections, where we first deal with all the single participant analyses, then the surrogate testing to confirm the basic results, then the participant aggregate results (Figure 7 and Supplementary Figure 7). The participant aggregate results reiterate the basic findings for the single participants. The key finding is straightforward (SF power decreases with SF) and required only one statistical analysis per subject.

(5) The function of brain phase dynamics at different frequencies and scales has been examined in previous papers at frequencies and scales relevant to what the authors treat. The authors may want to be more extensive with citing relevant studies and elaborating on the implications for them. Some examples below:

Womelsdorf T, et alScience. 2007

Besserve M et al. PloS Biology 2015

Nauhaus I et al Nat Neurosci 2009

We have added two paragraphs to the discussion, in response to the reviewer suggestion (lines 606-623). These paragraphs place our high TF findings in the context of previous research.

Reviewer #2 (Public review):

Summary:

In this paper, the authors analyze the organization of phases across different spatial scales. The authors analyze intracranial, stereo-electroencephalogram (sEEG) recordings from human clinical patients. The authors estimate the phase at each sEEG electrode at discrete temporal frequencies. They then use higher-order SVD (HOSVD) to estimate the spatial frequency spectrum of the organization of phase in a data-driven manner. Based on this analysis, the authors conclude that most of the variance explained is due to spatially extended organizations of phase, suggesting that the best description of brain activity in space and time is in fact a globally organized process. The authors' analysis is also able to rule out several important potential confounds for the analysis of spatiotemporal dynamics in EEG.

Strengths:

There are many strengths in the manuscript, including the authors' use of SVD to address the limitation of irregular sampling and their analyses ruling out potential confounds for these signals in the EEG.

Weaknesses:

Some important weaknesses are not properly acknowledged, and some conclusions are overinterpreted given the evidence presented.

The central weakness is that the analyses estimate phase from all signal time points using wavelets with a narrow frequency band (see Methods - "Numerical methods"). This step makes the assumption that phase at a particular frequency band is meaningful at all times; however, this is not necessarily the case. Take, for example, the analysis in Figure 3, which focuses on a temporal frequency of 9.2 Hz. If we compare the corresponding wavelet to the raw sEEG signal across multiple points in time, this will look like an amplitude-modulated 9.2 Hz sinusoid to which the raw sEEG signal will not correspond at all. While the authors may argue that analyzing the spatial organization of phase across many temporal frequencies will provide insight into the system, there is no guarantee that the spatial organization of phase at many individual temporal frequencies converges to the correct description of the full sEEG signal. This is a critical point for the analysis because while this analysis of the spatial organization of phase could provide some interesting results, this analysis also requires a very strong assumption about oscillations, specifically that the phase at a particular frequency (e.g. 9.2 Hz in Figure 3, or 8.0 Hz in Figure 5) is meaningful at all points in time. If this is not true, then the foundation of the analysis may not be precisely clear. This has an impact on the results presented here, specifically where the authors assert that "phase measured at a single contact in the grey matter is more strongly a function of global phase organization than local". Finally, the phase examples given in Supplementary Figure 5 are not strongly convincing to support this point.

“using wavelets with a narrow frequency band … this analysis also requires a very strong assumption about oscillations, specifically that the phase at a particular frequency (e.g. 9.2 Hz in Figure 3, or 8.0 Hz in Figure 5) is meaningful at all points in time”

Our method uses very short time-window Morlet wavelets to avoid the assumptions of oscillations, i.e., long-lasting sinusoids in the signal, in the sense of sinusoidal waveforms, or limit cycles extending in time. Cortical TWs can only last one or two cycles (Alexander et al., 2006), requiring methods that are compact in the time domain to avoid underreporting the desired phenomena. Additionally, the short time-window Morlet wavelets have low frequency resolution, so they are robust with respect to shifts in frequency between sites. We now discuss this issue explicitly in the Methods (lines 658-674). This means the phase estimation methods used in the manuscript precisely do not have the problem of assuming narrow-band oscillations in the signal. The methods are also robust to the exact shape of the waveforms; the signal needs be only approximately sinusoidal; to rise and fall. This means the Fourier variant we use does not introduce ringing artefact that can be introduced using longer timeseries methods, such as FFT.

“This step makes the assumption that phase at a particular frequency band is meaningful at all times”

This important consideration is entrenched in our choice of methods. By way of explanatory background, we point out that this step is not the final step. Aggregation methods can be used to distinguish between signal and noise. In the simple case, event-locked time-series of phase can be averaged. This would allow consistent (non-noise) phase relations to be preserved, while the inconsistent (including noise) phase relations would be washed out. This is part of the logic behind all such aggregation procedures, e.g., phase-locking, coherence. SVD has the advantage of capturing consistent relations in this sense, but without loss of information as occurs in averaging (up to the choice of number of singular vectors in the final model). Specifically, maps of the spatial covariances in phase are captured in the order of the variance explained. Noise (in the sense conveyed by the reviewer) in the phase measurements will not contribute to highest rank singular vectors. SVD is commonly used to remove noise, and that is one of its purposes here. This point can be seen by considering the very smooth singular vectors derived from MEG (Figure 3F) in this new version of the manuscript. These maps of phase gradients pull out only the non-noisy relations, even as their weighted sums reproduce any individual sample to any desired accuracy.

To summarize, the next step (of incorporating the phase measure into the SVD) neatly bypasses the issue of non-meaningful phase quantification. This is one of the reasons why we do not undertake the spatial frequency estimates on the raw matrices of estimated phase.

We now include a new sub-paragraph on this topic in the methods, lines 831-838.

In addition, we have reworded the first description of the methods with a new paragraph at the end of the introduction, which better balances the description of the steps involved. The two sentences (lines 162-166 highlight the issue of concern to the reviewer.

“there is no guarantee that the spatial organization of phase at many individual temporal frequencies converges to the correct description of the full sEEG signal.”

The correct description of the full sEEG signal is beyond the scope of the present research. Our main goal, as stated, is to show that the hypothesis that ‘extra-cranial measurements of TWs is the result of projection from localized activity’ is not supported by the evidence of spatial patterns of activity in the cortex. Since this activity can be accessed as single frequency band (especially if localized sources create the large-scale patterns), analysis of SF on a TF-by-TF basis is sufficient.

“This has an impact on the results presented here, specifically where the authors assert that "phase measured at a single contact in the grey matter is more strongly a function of global phase organization than local".

We agree with the reviewer, even though we expect that the strongest influences on local phase are due to other cortical signals in the same band. The implicit assumption of the focus on bands of the same temporal frequency is now made explicit in the abstract (lines 31-34).

A sentence addressing this issue had been added to the first paragraph of the discussion (lines 579-582).

Inclusion of cross-frequency interactions would likely require a highly regular measurement array over the scales of interest here, i.e., the noise levels inherent in the spatial organization of sEEG contacts would not support such analyses.

“Finally, the phase examples given in Supplementary Figure 5 are not strongly convincing to support this point.”

We have removed the phase examples that were previously in Supplementary Figure 5 (and Figure 5 in the previous version of the main text), since further surrogate testing and modelling (Supplementary Figure 11) shows the LSVs from irregular arrays will inevitably capture mixtures of low and high SF signals. The final section of the Methods explains this effect in some detail. Instead, the new version of the manuscript relies on new surrogate testing to validate our methods.

Another weakness is in the discussion on spatial scale. In the analyses, the authors separate contributions at (approximately) > 15 cm as macroscopic and < 15 cm as mesoscopic. The problem with the "macroscopic" here is that 15 cm is essentially on the scale of the whole brain, without accounting for the fact that organization in sub-systems may occur. For example, if a specific set of cortical regions, spanning over a 10 cm range, were to exhibit a consistent organization of phase at a particular temporal frequency (required by the analysis technique, as noted above), it is not clear why that would not be considered a "macroscopic" organization of phase, since it comprises multiple areas of the brain acting in coordination. Further, while this point could be considered as mostly semantic in nature, there is also an important technical consideration here: would spatial phase organizations occurring in varying subsets of electrodes and with somewhat variable temporal frequency reliably be detected? If this is not the case, then could it be possible that the lowest spatial frequencies are detected more often simply because it would be difficult to detect variable organizations in subsets of electrodes?

The motivation for our study was to show that large-scale TWs measured outside the cortex cannot be the result of more localized activity being ‘projected up’. In this case, the temporal frequency of the artefactual waves would be the same as the localized sources, so the criticism does not apply.

“while this point could be considered as mostly semantic in nature”

We have changed the terminology in the paper to better coincide with standard usage. Macroscopic now refers to >1cm, while we refer to >8cm as large-scale.

“15 cm is essentially on the scale of the whole brain, without accounting for the fact that organization in sub-systems may occur.”

We can assume that subtle frequency variation (e.g., within an alpha phase binding) is greatest at the largest scales of cortex, or at least not less varying than measurements within regions. This means that not considering frequency-drift effects will not inflate low spatial frequency power over high spatial frequency power. Even so, the power spectrum we estimated is approximately 1/SF, so that unmeasured cross-frequency effects in binding (causal influences on local phase) would have to overcome the strength of this relation for this criticism to apply, which seems unlikely.

“would spatial phase organizations occurring in varying subsets of electrodes and with somewhat variable temporal frequency reliably be detected?”

See our previous comments about the low temporal frequency resolution of two cycle Morlet wavelets. The answer is yes, up to the range approximated by half-power bandwidth, which is large in the case of this method (see lines 760-764).

Another weakness is disregarding the potential spike waveform artifact in the sEEG signal in the context of these analyses. Specifically, Zanos et al. (J Neurophysiol, 2011) showed that spike waveform artifacts can contaminate electrode recordings down to approximately 60 Hz. This point is important to consider in the context of the manuscript's results on spatial organization at temporal frequencies up to 100 Hz. Because the spike waveform artifact might affect signal phase at frequencies above 60 Hz, caution may be important in interpreting this point as evidence that there is significant phase organization across the cortex at these temporal frequencies.

We have now added a sentence on this issue to the discussion (lines 600-602).

However, our reading of the Zanos et al. paper is that the low temporal frequency (60-100Hz) contribution of spikes and spike patterns is negligible compared to genuine post-synaptic membrane fluctuations (see their Figure 3). These considerations come more strongly into play when correlations between LFP and spikes are calculated or spike triggered averaging is undertaken, since then a signal is being partly correlated with itself, or, partly averaged over the supposedly distinct signal with which it was detected.

A last point is that, even though the present results provide some insight into the organization of phase across the human brain, the analyses do not directly link this to spiking activity. The predictive power that these spatial organizations of phase could provide for spiking activity - even if the analyses were not affected by the distortion due to the narrow-frequency assumption - remains unknown. This is important because relating back to spiking activity is the key factor in assessing whether these specific analyses of phase can provide insight into neural circuit dynamics. This type of analysis may be possible to do with the sEEG recordings, as well, by analyzing high-gamma power (Ray and Maunsell, PLoS Biology, 2011), which can provide an index of multi-unit spiking activity around the electrodes.

“even if the analyses were not affected by the distortion due to the narrow-frequency assumption”

See our earlier comment about narrow TFs; this is not the case in the present work.

The spiking activity analysis would be an interesting avenue for future research. It appears the 1000Hz sampling frequency in the present data is not sufficient for method described in Ray & Maunsell (2011). On a related topic, we have shown that large-scale traveling waves in the MEG and 8cm waves in ECoG can both be used to predict future localized phase at a single sensor/contact, two cycles into the future (Alexander et al., 2019). This approach could be used to predict spiking activity, by combining it with the reviewer’s suggestion. However, the current manuscript is motivated by the argument that measured large-scale extra-cranial TWs are merely projections of localized cortical activity. Since spikes do not arise in this argument, we feel it is outside the scope of the present research. We have added this suggestion to the discussion as a potential line of future research (lines 686-688).

Reviewer #3 (Public review):

Summary:

The authors propose a method for estimation of the spatial spectra of cortical activity from irregularly sampled data and apply it to publicly available intracranial EEG data from human patients during a delayed free recall task. The authors' main findings are that the spatial spectra of cortical activity peak at low spatial frequencies and decrease with increasing spatial frequency. This is observed over a broad range of temporal frequencies (2-100 Hz).

Strengths:

A strength of the study is the type of data that is used. As pointed out by the authors, spatial spectra of cortical activity are difficult to estimate from non-invasive measurements (EEG and MEG) due to signal mixing and from commonly used intracranial measurements (i.e. electrocorticography or Utah arrays) due to their limited spatial extent. In contrast, iEEG measurements are easier to interpret than EEG/MEG measurements and typically have larger spatial coverage than Utah arrays. However, iEEG is irregularly sampled within the threedimensional brain volume and this poses a methodological problem that the proposed method aims to address.

Weaknesses:

The used method for estimating spatial spectra from irregularly sampled data is weak in several respects.

First, the proposed method is ad hoc, whereas there exist well-developed (Fourier-based) methods for this. The authors don't clarify why no standard methods are used, nor do they carry out a comparative evaluation.

We disagree that the method is ad hoc, though the specific combination of SVD and multiscale differencing is novel in its application to sEEG. The SVD method has been used to isolate both ~30cm TWs in MEG and EEG (Alexander et al., 2013; 2016), as well as 8cm waves in ECoG (Alexander et al., 2013; 2019). In our opening examples in the results now reiterate these previous related findings, by way of example analysis of MEG data (Figure 3). This will better inform the reader on the extent of continuity of the method from previous research.

Standard FFT has been used after interpolating between EEG electrodes to produce a uniform array (Alamia et al., 2023). There exist well-developed Fourier methods for nonuniform grids, such as simple interpolation, the butterfly algorithm, wavefield extrapolation and multi-scale vector field techniques. However, the problems for which these methods are designed require non-sparse sampling or less irregular arrays. The sEEG contacts (reduced in number to grey matter contacts) are well outside the spatial irregularity range of any Fourierrelated methods that we are aware of, particularly at the broad range of spatial scales of interest here (2cm up to 24cm). This would make direct comparison of these specialized Fourier method to our novel methods, in the sEEG, something of a straw-man comparison.

We now include a summary paragraph in the introduction, which is a brief review of Fourier methods designed to deal with non-uniform sampling (lines 159-162).

Second, the proposed method lacks a theoretical foundation and hinges on a qualitative resemblance between Fourier analysis and singular value decomposition.

We have improved our description of the theoretical relation between Fourier analysis and SVD (additional material at lines 839-861 and 910-922). In fact, there are very strong links between the two methods, and now it should be clearer that our method does not rely on a mere qualitative resemblance.

Third, the proposed method is not thoroughly tested using simulated data. Hence it remains unclear how accurate the estimated power spectra actually are.

We now include a new surrogate testing procedure, which takes as inputs the empirical data and a model signal (of known spatial frequency) in various proportions. Thus, we test both the impact of small amount of surrogate signal on the empirical signal, and the impact of ‘noise’ (in the form of a small amount of empirical signal) added to the well-defined surrogate signal.

In addition, there are a number of technical issues and limitations that need to be addressed or clarified (see recommendations to the authors).

My assessment is that the conclusions are not completely supported by the analyses. What would convince me, is if the method is tested on simulated cortical activity in a more realistic set-up. I do believe, however, that if the authors can convincingly show that the estimated spatial spectra are accurate, the study will have an impact on the field. Regarding the methodology, I don't think that it will become a standard method in the field due to its ad hoc nature and well-developed alternatives.

Simulations of cortical activity do not seem the most direct way to achieve this goal. The first author has published in this area (Liley et. al., 1999; Wright et al., 2001), and such simulations, for both bulk and neuronally based simulations, readily display traveling wave activity at low spatial frequencies (indeed, this was the origin of the present scientific journey). The manuscript outlines these results in the introduction, as well as theoretical treatments proposing the same. Several other recent studies have highlighted the appearance of largescale travelling waves using connectome-based models (https://www.biorxiv.org/content/10.1101/2025.07.05.663278v1; https://www.nature.com/articles/s41467-024-47860-x), which we do not include in the manuscript for reasons of brevity. In short, the emergence of TW phenomenon in models is partly a function of the assumptions put into them (i.e., spatial damping, boundary conditions, parameterization of connection fields) and would therefore be inconclusive in our view.

Instead, we rely on the advantages provided by the way our central research question has been posed: that the spatial frequency distribution of grey matter signal can determine whether extra-cranial TWs are artefactual. The newly introduced surrogate methods reflect this advantage by directly adding ground truth spatial frequency components to individual sample measurements. This is a less expensive option than making cortical simulations to achieve the same goal.

For the same reasons, we include testing of the methods using real cortical signals with MEG arrays (for which we could test the effects of increasing sparseness of contacts, test the effects of average referencing, and also construct surrogate time-series with alternative spectra).

Recommendations for the authors:

Reviewer #2 (Recommendations for the authors):

Major points

Methods, Page 18: "... using notch filters to remove the 50Hz line signal and its harmonics ...": The sEEG data appear to have been recorded in North America, where the line frequency is 60 Hz. Is this perhaps a typo, or was a 50 Hz notch filter in fact applied here (which would be a mistake)?

This has now been fixed in the text to read 60Hz. This is the notch filter that was applied.

Minor points

(1) While the authors do state that they are analyzing the "spatial frequency spectrum of phase dynamics" in the abstract, this could be more clearly emphasized. Specifically, the difference between signal power at different spatial frequencies (as analyzed by a standard Fourier analysis) and the organization of phase in space (as done here) could be more clearly distinguished.

We now address this point explicitly on lines 167-172. We now include at the end of the results additional analyses where the TF power is included. This means that the effects of including signal power at different temporal frequencies can be directly compared to our main analysis of the SF spectrum of the phase dynamics.

(2) Figure 1A-C: It was not immediately clear what the lengths provided in these panels (e.g."> 40 cm cortex", "< 10 cm", "< 30 cm") were meant to indicate. This could be made clearer.

Now fixed in the caption.

(3) Figure 2A: If this is surrogate data to explain the analysis technique, it would be helpful to note explicitly at this point.

This Figure has been completely reworked, and now the status of the examples (from illustrative toy models to actual MEG data) should be clearer.

(4) Figure 4A: Why change from "% explained variance" for the example data in Figure 2C to arbitrary units at this point?

This has now been explicitly stated in the methods (lines 1033-1036).

(5) Page 15: "This means either the results were biased by a low pass filter, or had a maximum measurable...": If the authors mean that the low-pass filter is due to spatial blurring of neural activity in the EEG signal, it would be helpful to state that more directly at this point.

Now stated directly, lines 567-568.

(6) Page 23: "...where |X| is the complex magnitude of X...": The modulus operation is defined on a complex number, yet here is applied to a vector of complex numbers. If the operation is elementwise, it should be defined explicitly.

‘Elementwise’ is now stated explicitly (line 1020).

Reviewer #3 (Recommendations for the authors):

In the submitted manuscript, the authors propose a method to estimate spatial (phase) spectra from irregularly sampled oscillatory cortical activity. They apply the method to intracranial (iEEG) data and argue that cortical activity is organized into global waves up to the size of the entire cortex. If true, this finding is certainly of interest, and I can imagine that it has profound implications for how we think about the functional organization of cortical activity.

We have added a section to the discussion outlining the most radical of these implications: what does it mean to do source localization when non-local signals dominate? Lines 670-681.

The manuscript is well-written, with comprehensive introduction and discussion sections, detailed descriptions of the results, and clear figures. However, the proposed method comprised several ad hoc elements and is not well-founded mathematically, its performance is not adequately assessed, and its limitations are not sufficiently discussed. As such, the study failed to convince (me) of the correctness of the main conclusions.

We now have a direct surrogate testing of the method. We have also improved the mathematical explanation to show that the link between Fourier analysis and SVD is not ad hoc, but well understood in both literatures. We had addressed explicitly in the text all of the limitations raised by the reviewers.

Major comments

(1) The main methodological contribution of the study is summarized in the introduction section:

"The irregular sampling of cortical spatial coordinates via stereotactic EEG was partly overcome by the resampling of the phase data into triplets corresponding to the vertices of approximately equilateral triangles within the cortical sheet."

There exist well-established Fourier methods for handling irregularly sampled data so it is unclear why the authors did not resort to these and instead proposed a rather ad hoc method without theoretical justification (see next comment).

We have re-reviewed the literature on non-uniform Fourier analysis. We now briefly review the Fourier methods for handling irregularly sampled data (lines 155-162) and conclude that none of the existing methods can deal with the degree of irregularity, and especially sparsity, found for the grey-matter sEEG contacts.

(2) In the Appendix, the authors write:

"For appropriate signals, i.e., those with power that decreases monotonically with frequency, each of the first few singular vectors, v_k, is an approximate complex sinusoid with wavenumber equal to k."

I don't think this is true in general and if it is, there must be a formal argument that proves it. Furthermore, is it also true for irregularly sampled data? And in more than one spatial dimension? Moreover, it is also unclear exactly how the spatial Fourier spectrum is estimated from the SVD.

In response to these reviewer queries, we now spend considerably more time in the conceptual set-up of the manuscript, giving examples of where SVD can be used to estimate the Fourier spectrum. We have now unpacked the word ‘appropriate’ and we are now more exact in our phrasing. This is laid out in lines 843-850 of the manuscript. In addition, the methods now describe the mathematical links between Fourier analysis and SVD (lines 851861 and 910-922).

The authors write:

"The spatial frequency spectrum can therefore be estimated using SVD by summing over the singular values assigned to each set of singular vectors with unique (or by binning over a limited range of) spatial frequencies. This procedure is illustrated in Figure 1A-C."

First, the singular vectors are ordered to decreasing values of the corresponding singular values. Hence, if the singular values are used to estimate spectral power, the estimated spectrum will necessarily decrease with increasing spatial frequency (as can be seen in Figure 2C). Then how can traveling waves be detected by looking for local maxima of the estimated power spectra?

TWs are not detected by looking for local maxima in the spectra. Our work has focussed on the global wave maps derived from the SVD of phase (i.e., k=1-3), which also explain most of the variance in phase. This is now mentioned in the caption to Figure 3 (lines 291-294).

Second, how are spatial frequencies assigned to the different singular vectors? The proposed method for estimating spatial power spectra from irregularly sampled data seems rather ad hoc and it is not at all clear if, and under what conditions, it works and how accurate it is.

The new version of the manuscript uses a combination of the method previously presented (the multi-scale differencing) and the method previously outlined in the supplementary materials (doing complex-valued SVD on the spatial vectors of phase). We hope that along with the additional expository material in the methods the new version is clearer and seems less ad hoc to the reviewer. Certainly, there are deep and well-understood links between Fourier analysis and SVD, and we hope we have brought these into focus now.

(3) The authors define spatial power spectra in three-dimensional Euclidean space, whereas the actual cortical activity occurs on a two-dimensional sheet (the union of two topological 2spheres). As such, it is not at all clear how the estimated wavelengths in three-dimensional space relate to the actual wavelengths of the cortical activity.

We define spatial power spectra on the folded cortical sheet, rather than Cartesian coordinates. We use geodesic distances in all cases where a distance measurement is required. We have included two new figures (Figure 5 and Supplementary Figure1) showing the mapping of the triangles onto the cortical sheet, which should bring this point home.

(4) The authors' analysis of the iEEG data is subject to a caveat that is not mentioned in the manuscript: As a reference for the local field potentials, the average white-matter signal was used and this can lead to artifactual power at low spatial frequencies. This is because fluctuations in the reference signal are visible as standing waves in the recording array. This might also explain the observation that

"A surprising finding was that the shape of the spatial frequency spectrum did not vary much with temporal frequency."

because fluctuations in the reference signal are expected to have power at all temporal frequencies (1/f spectrum). When superposed with local activity at the recording electrodes, this leads to spurious power at low spatial frequencies. Can the authors exclude this interpretation of the results?

The new version of the manuscript deals explicitly with this potential confound (lines 454467). First, the artefactual global synchrony due to the reference signal (the DC component in our spatial frequency spectra of phase) is at a distinct frequency from the lowest SF of interest here. The lowest spatial frequency is a function of the maximum spatial range of the recording array and not overlapping in our method with the DC component, despite the loss of SF resolution due to the noise of the spatial irregularity of the recording array. This can be seen from consideration of the SF tuning (Figure 4) for the MEG wave maps shown in Figure 3, and the spectra generated for sparse MEG arrays in Supplementary Figure 5. Additionally, this question led us to a series of surrogate tests which are now included in the manuscript. We used MEG to test for the effects of average reference, since in this modality the reference free case is available. The results show that even after imposing a strong and artefactual global synchrony, the method is highly robust to inflation of the DC component, which either way does not strongly influence the SF estimates in the range of interest (4c/m to 12c/m for the case of MEG).

(5) Related to the previous comment: Contrary to the authors' claims, local field potentials are susceptible to volume conduction, particularly when average references are used (see e.g. https://www.cell.com/neuron/fulltext/S0896-6273(11)00883-X)

Methods exist to mitigate these effects (e.g. taking first- or second-order spatial differences of the signals). I think this issue deserves to be discussed.

We have reviewed this research and do not find it to be a problem. The authors cited by the reviewer were concerned with unacknowledged volume conduction up to 1 cm for LFP. The maximum spatial frequency we report here is 50c/m, or equivalent to 2cm. While the intercontact distance on the sEEG electrodes was 0.5cm, in practice the smallest equilateral triangles (i.e., between two electrodes) to be found in the grey matter was around 2cm linear size. We make no statements about SF in the 1cm range. We do now cite this paper and mention this short-range volume conduction (lines 602-605). The method of taking derivatives has the same problems as source localization methods. They remove both artefactual correlations (volume conduction) and real correlations (the low SF interactions of interest here). We mention this now at lines 667-669. In addition, our method to remove negative SF components from the LSVs ameliorates the effects of average referencing. There are now more details in the Methods about this step (lines 924-947), as well as a new supplementary figure illustrating its effects on signal with a known SF spectrum (MEG, supplementary Figure 6).

(6) Could the authors add an analysis that excludes the possibility that the observed local maxima in the spectra are a necessary consequence of the analysis method, rather than reflecting true maxima in the spectra? A (possibly) similar effect can be observed in ordinary Fourier spectra that are estimated from zero-mean signals: Because the signals have zero mean, the power spectrum at frequency zero is close to zero and this leads to an artificial local maximum at low frequencies.

We acknowledge the reviewer’s mathematical point. We do not agree that it could be an issue, though it is important to rule it out definitively. First, removing the DC component will only produce an artefactual low SF peak if the power at low SF is high. This may occur in the reviewer’s example only because temporal frequency has a ~1/f spectrum. If the true spectrum is flat, or increasing in power with f, no such artificial low SF will be produced (see Supplementary Figure 5G). Additionally,

(1) The DC component is well separated from the low SF components in our method;

(2) We now include several surrogate methods which show that our method finds the correct spectral distribution and is not just finding a maximum at low SFs due to the suggested effect (subtraction of the DC component). Analysis of separated wave maps in MEG (Figures 3 & 4) shows the expected peaks in SF, increasing in peak SF for each family of maps when wavenumber increases (roughly three k=1 maps, three k=2 etc.). A specific surrogate test for this query was also undertaken by creating a reverse SF spectrum in MEG phase data, in which the spectrum goes linearly with f over the SF range of interest, rather than the usual 1/f. Our method correctly finds the former spectrum (Supplementary Figure 5). Additionally, we tested for the effects of introducing the average reference and the effects of our method to remove the DC component of the phase SF spectrum (Supplementary Figure 6). We can definitively rule out the reviewer’s concern.

A related issue (perhaps) is the observation that the location of the maximum (i.e. the peak spatial frequency of cortical activity) depends on array size: If cortical activity indeed has a characteristic wavelength (in the sense of its spectrum having a local maximum) would one not expect it to be independent of array size?

This is only true when making estimates for relatively clean sinusoidal signals, and not from broad-band signals. Fourier analysis and our related SVD methods are very much dependent on maximum array size used to measure cortical signals. This is why the first frequency band (after the DC component) in Fourier analysis is always at a frequency equivalent to 1/array_size, even if the signal is known to contain lower frequency components. We now include a further illustration of this in Figure 3, a more detailed exposition of this point in the methods, and in Supplementary Figure 11 we provide a more detailed example of the relation between Fourier analysis and SVD when grids with two distinct scales are used.

In short, it is not possible, mathematically, to measure wavelengths greater than the array size in broad-band data. This is now stated explicitly in the manuscript (lines 143-144). A common approach in Neuroscience research is to first do narrowband filtering, then use a method that can accurately estimate ‘instantaneous’ phase change, such as the Hilbert transform. This is not possible for highly irregular sEEG arrays.

(7) The proposed method of estimating wavelength from irregularly sampled threedimensional iEEG data involves several steps (phase-extraction, singular value decomposition, triangle definition, dimension reduction, etc.) and it is not at all clear that the concatenation of all these steps actually yields accurate estimates.

Did the authors use more realistic simulations of cortical activity (i.e. on the convoluted cortical sheet) to verify that the method indeed yields accurate estimates of phase spectra?

We now included detailed surrogate testing, in which varying combinations of sEEG phase data and veridical surrogate wavelengths are added together.

See our reply from the public reviewer comments. We assess that real neurophysiological data (here, sEEG plus surrogate and MEG manipulated in various ways) is a more accurate way to address these issues. In our experience, large scale TWs appear spontaneously in realistic cortical simulations, and we now cite the relevant papers in the manuscript (line 53).

Minor comments

(1) Perhaps move the first paragraph of the results section to the Introduction (it does not describe any results).

So moved.

(2) The authors write:

"The stereotactic EEG contacts in the grey matter were re-referenced using the average of low-amplitude white matter contacts"

Does this mean that the average is taken over a subset of white-matter contacts (namely those with low amplitude)? Or do the authors refer to all white-matter contacts as "low-amplitude"? And had contacts at different needles different references? Or where the contacts from all needles pooled?

A subset of white-matter contacts was used for re-referencing, namely those 50% with lowest amplitude signals. This subset was used to construct a pooled, single, average reference. We have rephrased the sentences referring to this procedure to improve clarity (line 202 and 743745).

Reviewer #3 (Public review):

Summary:

This paper reports on an association between body size and the occurrence of species in cities, which is quantified using an 'urban score' that can be visualized as a 'Species Urbanness Distribution' for particular taxa. The authors use species records from the Global Biodiversity Information Facility (GBIF) and link the occurrence data to nighttime lighting quantified using satellite data (Visible Infrared Imaging Radiometer Suite-VIIRS). They link the urban score to body size data to find 'heterogeneous relationship between body size and urban tolerance across the tree'. The results are then discussed with reference to potential mechanisms that could possibly produce the observed effects (cf. Figure 1).

Strengths:

The novelty of this study lies in the huge number of species analyzed and the comparison of results among animal taxa, rather than in a thorough analysis of what traits allow species to persist under urban conditions. Such analyses have been done using a much more thorough approach that employs presence-absence data as well as a suite of traits by other studies, for example, in (Hahs et al. 2023, Neate-Clegg et al. 2023). The dataset that the authors produced would also be very valuable if these raw data were published, both the cleaned species records as well as the body sizes.

The paper could strongly add to our understanding of what species occur in cities when the open questions are addressed.

Weaknesses:

I value the approach of the authors, but I think the paper needs to be revised.

In my view, the authors could more carefully validate their approach. Currently, any weakness or biases in the approach are quickly explained away rather than carefully explored. This concerns particularly the use of presence-only data, but also the calculation of the urban score.

The vast majority of data in GBIF is presence-only data. This produces a strong bias in the analysis presented in the paper. For some taxa, it is likely that occurrences within the city are overrepresented, and for other taxa, the opposite is true (cf. Sweet et al. 2022). I think the authors should try to address this problem.

The authors should compare their results to studies focusing on particular taxa where extensive trait-based analyses have already been performed, i.e., plants and birds. In fact, I strongly suggest that the authors should compare their results to previous studies on the relationship between traits, including body size and occurrences along a gradient of urbanisation, to draw conclusions about the validity of the approach used in the current study, which has a number of weaknesses.

They should be be more careful in coming up with post-hoc explanations of why the pattern found in this study makes sense or suggests a particular mechanism. This reviewer considers that there is no way in which the current study can disentangle the different possible mechanisms without further analyses and data, so I would suggest pointing out carefully how the mechanisms could be studied

More details should be given about the methodology. The readers should be able to understand the methods without having to read a number of other papers.

References:

Hahs, A. K., B. Fournier, M. F. Aronson, C. H. Nilon, A. Herrera-Montes, A. B. Salisbury, C. G. Threlfall, C. C. Rega-Brodsky, C. A. Lepczyk, and F. A. La Sorte. 2023. Urbanisation generates multiple trait syndromes for terrestrial animal taxa worldwide. Nature Communications 14:4751.

Neate-Clegg, M. H. C., B. A. Tonelli, C. Youngflesh, J. X. Wu, G. A. Montgomery, Ç. H. Şekercioğlu, and M. W. Tingley. 2023. Traits shaping urban tolerance in birds differ around the world. Current Biology 33:1677-1688.

Sweet, F. S. T., B. Apfelbeck, M. Hanusch, C. Garland Monteagudo, and W. W. Weisser. 2022. Data from public and governmental databases show that a large proportion of the regional animal species pool occur in cities in Germany. Journal of Urban Ecology 8:juac002.

Une reformulation de la banalité du mal: la pensée est dialogue , y compris à l'intérieur de soi, comme confrontation, comme deux en un. Quand elle s'éteint, il n'y a plus que l'obéissance et la banalité.

C'est ça les intellectuels qui virent la montée du nazisme et ... Eichmann.

Est ce la même chose aujourd'hui, où les intellectuels ont la responsabilité de la montée des populismes ?

ha, you can distract chatbots, as we've seen from the start. This is the classic 'it's not for me but for my mom' train ticket sales automation hangup in response to 'to which destination would you like a ticket', and then 'unknown railway station 'for my mom' in a new guise. And they didn't even expect that to happen? It's an attack service!

Guide pratique : S'engager sans s'épuiser, cultiver un militantisme durable

Introduction : La double facette de l'engagement moderne

Face à une urgence écologique et sociale de plus en plus palpable, nous assistons à une multiplication des formes d'engagement citoyen.

Des actions de désobéissance civile aux initiatives de sensibilisation, en passant par la création de médias indépendants, cet élan collectif est vital pour faire face aux défis de notre époque.

Cependant, cette mobilisation intense expose les individus et les organisations à un risque élevé d'épuisement physique et psychologique, un phénomène souvent désigné sous le nom de « burnout militant ».

Loin d'être un signe de faiblesse, cet épuisement est une conséquence logique d'une lutte exigeante contre des systèmes profondément ancrés.

Ce guide se veut une ressource pragmatique et encourageante, synthétisant les stratégies, les changements de perspective et les leçons partagées par des militants expérimentés pour préserver son énergie et cultiver sa motivation sur le long terme.

En tant que psychologue observant ces dynamiques, ce guide vise à outiller les acteurs du changement pour qu'ils puissent aligner leur action extérieure avec leur résilience intérieure.

--------------------------------------------------------------------------------

1. Comprendre la flamme de l'engagement : Les racines de l'action

Avant de chercher à protéger la flamme de l'engagement, il est fondamental de comprendre ce qui l'a allumée.

Identifier ses motivations profondes, cette « étincelle » initiale qui pousse à l'action, est la première étape pour construire un engagement résilient et authentique.

C'est en se reconnectant à ce « pourquoi » viscéral que l'on peut trouver la force de traverser les moments de doute et de fatigue.

Cette section explore les divers détonateurs de l'action, tels que vécus et partagés par des personnes engagées aux parcours variés.

1.1. L'étincelle initiale : Identifier votre « pourquoi »

Les chemins qui mènent à l'engagement sont multiples, souvent personnels et profondément transformateurs. Ils naissent d'une rencontre entre une sensibilité individuelle et une réalité qui devient intolérable.

• La prise de conscience soudaine : Pour certains, l'engagement naît d'un choc, d'une information qui brise les certitudes.

C'est le cas de l'arboriste-grimpeur Thomas Braille, qui a été « coupé dans ses jambes » en réalisant que l'échéance de l'urgence climatique n'était plus une projection lointaine mais une réalité imminente :

« 20 ans, c'est demain ».

Cette prise de conscience a été catalysée par la peur viscérale pour l'avenir de son fils.

• Le sentiment d'injustice personnel : L'expérience vécue de l'injustice est un moteur puissant et durable.

Pour la réalisatrice Flore Vasseur, le « foyer de la flamme » se trouve dans une injustice personnelle vécue durant l'enfance.

Cette blessure initiale, bien que longtemps enfouie, est devenue la source d'une quête de réparation et d'une sensibilité aiguë aux injustices du monde.

• La passion confrontée à la réalité : L'engagement peut aussi émerger lorsque la passion d'une vie se heurte à l'inaction et à l'absurdité du système.

L'agroclimatologue Serge Zaka, passionné par la météo depuis l'enfance, a basculé dans un engagement public en constatant les impacts concrets du changement climatique (des végétaux brûlés à 46°C) et l'ignorance des décideurs politiques face à des études qu'ils avaient eux-mêmes commandées.

• La quête de cohérence et la fin de la solitude : Parfois, l'engagement est une flamme qui couvait depuis longtemps mais peinait à trouver un exutoire.

Pour Anaïs Terrien, présidente de La Fresque du Climat, un engagement précoce mais solitaire a trouvé un nouvel élan grâce à un outil lui permettant enfin de structurer le dialogue, de briser l'isolement et d'être comprise dans ses préoccupations.

1.2. Le moteur psychologique de l'action

Selon l'analyse de l'écopsychologue Emmanuel Delrieu, l'engagement n'est pas un simple choix intellectuel, mais une transformation profonde qui répond à des mécanismes psychologiques précis.

1. L'interaction des forces : Pour persévérer, un engagement doit mobiliser une synergie de trois types de forces.

Les forces affectives (ce qui nous touche, la sensibilité à la souffrance du monde), les forces comportementales (la capacité à agir et à persévérer dans la durée) et les forces cognitives (la capacité à analyser et à réconcilier les aspects positifs et négatifs de la lutte).

2. La résolution de la dissonance cognitive : S'engager est souvent un moyen de réduire la tension interne entre ses valeurs et les paradigmes dominants de la société (capitalisme, patriarcat, colonialisme).

Face à cette dissonance, l'action permet de « remettre de l'ordre dans sa vie » en alignant ses comportements avec ses convictions profondes.

3. La transformation par l'enracinement : Plus l'engagement est profond, plus l'individu se transforme et se « radicalise », au sens étymologique du terme :

il s'enracine dans ses convictions. Cet enracinement crée des liens, un « mycélium » avec d'autres luttes, renforçant la solidarité et la position de chacun.

Cependant, cette même puissance qui ancre l'individu dans ses convictions le rend aussi plus vulnérable.

En s'alignant si profondément avec sa cause, il s'expose frontalement à la résistance, à l'inertie et à la violence du système qu'il combat, créant un terrain propice à l'usure.

--------------------------------------------------------------------------------

2. Naviguer les tempêtes : Reconnaître et gérer le risque d'épuisement

Loin d'être un échec personnel ou un signe de faiblesse, les moments de fatigue, de doute et même d'effondrement sont des étapes quasi inévitables du parcours militant.

Ils sont le reflet de l'intensité de la lutte et de la violence de ce qui est combattu.

L'enjeu stratégique n'est donc pas d'éviter ces moments à tout prix, mais d'apprendre à en reconnaître les signes avant-coureurs et à y répondre de manière constructive et bienveillante.

2.1. Les symptômes avant-coureurs du burnout militant

Être à l'écoute de soi est la première ligne de défense. Voici quelques signaux d'alerte, basés sur les analyses et témoignages, qui doivent inciter à la prudence :

• Fatigue physique et mentale : Une irritabilité croissante et une fatigue persistante qui ne se résorbe pas avec le repos sont des premiers signes clairs que les réserves d'énergie s'épuisent (Emmanuel Delrieu).

• Perte de sens et envie de retrait : Après une action extrême – 40 jours de grève de la faim suivis d'une grève de la soif – Thomas Braille a ressenti le besoin de s'isoler : « je ne voulais plus voir d'êtres humains ».

Ce sentiment que le sacrifice est vain et que « tout le monde s'en fout » est un symptôme critique.

• Sentiment de submerssion : L'impression que « le vase était presque plein et menaçait de casser » a poussé Anaïs Terrien à annuler ses engagements.

Cette sensation d'être submergé par les responsabilités et les urgences est un indicateur majeur.

• Confrontation à l'indifférence et au cynisme : La frustration face à l'inaction générale, comme l'a vécue Flore Vasseur après les révélations d'Edward Snowden, peut user la motivation et mener à un sentiment d'impuissance destructeur.

2.2. Le burnout comme un cycle, et non comme une fin

Il est crucial de déconstruire l'idée que le burnout est un point final. C'est avant tout un signal et une étape de transformation.

• L'effondrement est un « moment transformatoire nécessaire ».

L'écopsychologue Emmanuel Delrieu insiste : plus on résiste à la fatigue et au besoin de changement, plus l'effondrement est douloureux.

L'accepter comme une étape nécessaire permet de le traverser plus sereinement.

• L'engagement n'est pas linéaire mais cyclique. Il s'apparente à une spirale.

Les phases de « down » ne sont pas des régressions, mais des moments où l'on plonge pour « chercher des forces encore plus grandes d'ancrage ».

Chaque cycle permet de se transformer et de repartir sur des bases plus solides.

• L'erreur est de « toujours vouloir être parfait et aller bien ». Comme le souligne Flore Vasseur, la société nous pousse à masquer nos vulnérabilités.

Or, la libération que représentent les émotions, les larmes et l'acceptation de ses failles est une source de résilience immense.

L'enjeu stratégique est donc de cultiver un réseau de soutien solide, capable de vous accueillir lors de ces phases d'effondrement pour qu'elles deviennent des sources de transformation, et non de destruction.

2.3. Les facteurs aggravants spécifiques au militantisme

Au-delà du surmenage classique, le militantisme expose à des sources de stress uniques qui accélèrent le risque d'épuisement.

1. La violence des attaques personnelles : L'exposition publique s'accompagne souvent d'une violence décomplexée.

Les insultes constantes reçues par Serge Zaka sur son physique (allant jusqu'à la création du sobriquet « Grosaka ») ou sa crédibilité (son chapeau) sont une forme de harcèlement visant à déstabiliser et à user psychologiquement.

2. L'invisibilisation institutionnelle : Comme l'analyse Emmanuel Delrieu, les structures politiques et sociales nient ou minimisent systématiquement les luttes.

Cette non-reconnaissance est une source d'injustice profonde et d'épuisement, car elle oblige à se battre non seulement pour sa cause, mais aussi pour la légitimité même de son combat.

3. La confrontation à la force du système : Les militants se heurtent à la capacité du système à absorber et neutraliser la critique.

Flore Vasseur a constaté que « plus vous tapez dedans, plus il est fort ».

Le système peut transformer la dénonciation en spectacle, la vidant de sa substance et laissant le militant avec un sentiment d'impuissance.

--------------------------------------------------------------------------------

3. Entretenir la flamme : Stratégies pour un engagement durable

Un engagement durable ne se résume pas à la gestion des crises d'épuisement.

Il repose sur la mise en place de stratégies proactives pour nourrir sa motivation, protéger son énergie et construire sa propre résilience.

Les quatre piliers suivants, complémentaires et interdépendants, offrent des pistes concrètes pour y parvenir.

3.1. Stratégie 1 : La force du collectif et du soutien

Le premier et le plus puissant rempart contre l'épuisement est la qualité des liens humains. L'isolement est le terreau du burnout.

• S'appuyer sur le collectif : Anaïs Terrien l'affirme sans détour : elle a été sauvée du burnout par son conseil d'administration.

Le groupe agit comme un filet de sécurité, permettant de prendre le relais lorsque l'un de ses membres flanche.

• Savoir demander de l'aide : Reconnaître ses propres limites et oser demander du soutien n'est pas une faiblesse, mais une compétence stratégique essentielle pour durer.

C'est un acte de confiance envers le collectif.

• Cultiver le « prendre soin du lien » : Comme le propose Emmanuel Delrieu, il est crucial d'instaurer au sein des groupes une pratique active de soutien mutuel.

Cela signifie créer des espaces où la vulnérabilité est acceptée et où l'on prend soin les uns des autres autant que de la cause défendue.

3.2. Stratégie 2 : La justesse de la perspective

La manière dont on perçoit son action et ses objectifs peut radicalement diminuer la pression et le risque d'épuisement.

• Adopter « l'esprit des cathédrales » : Partagée par Flore Vasseur via Edward Snowden, cette métaphore est libératrice.

Elle invite à accepter de ne pas voir le résultat final de ses actions, mais à se concentrer sur sa contribution : poser sa « brique » avec la confiance que d'autres construiront dessus.

• Lutter « pour » plutôt que « contre » : Ce changement de paradigme, également proposé par Flore Vasseur, rend l'engagement plus positif et moins autodestructeur.

Il s'agit de se battre pour un monde désirable, pour la vie, pour l'avenir de ses enfants — des moteurs qui génèrent une énergie positive et renouvelable, à l'inverse de la lutte contre un système qui peut se révéler corrosive.

• Renoncer à l'attente d'un résultat immédiat :

L'attente d'une victoire rapide est l'une des principales sources de dépression et de désillusion pour les militants.

L'esprit des cathédrales aide à se détacher de cette tyrannie du résultat.

3.3. Stratégie 3 : L'alignement et l'action authentique

Un engagement qui dure est un engagement qui vient du cœur, pas de l'ego.

• Se connecter à son injustice profonde : Comme le conseille Flore Vasseur, les blessures personnelles, les humiliations, les trahisons vécues sont le « fioul » le plus durable.

C'est en allant chercher ce qui nous touche viscéralement que l'on trouve une énergie inépuisable.

• S'engager pour se réparer soi-même : Plutôt que de s'engager pour la reconnaissance sociale ou l'image, ce qui mène inévitablement à l'épuisement, l'engagement le plus durable est celui qui est aussi une démarche intime.

Comme l'explique Flore Vasseur, « on y va pour se réparer soi. Ce qu'on vient réparer c'est soi et en se réparant soi on répare le monde ».

• Diversifier ses projets et ses sources d'énergie : Pour ne pas dépendre d'une seule source de gratification, il est sain de « ne pas mettre tous ses œufs dans le même panier », comme le pratique Anaïs Terrien.

Avoir d'autres projets (collectif d'habitation, jardinage, art) permet de se ressourcer et de maintenir un équilibre.

3.4. Stratégie 4 : La culture du soin personnel

Prendre soin de soi n'est pas un luxe ou un acte égoïste ; c'est une condition indispensable pour pouvoir continuer à prendre soin du monde.

• « Faire silence d'humain » : Ce conseil d'Emmanuel Delrieu invite à se reconnecter régulièrement et profondément à la nature, loin du bruit et de l'agitation humaine, pour se ressourcer et retrouver une perspective plus large.

• Se détacher de la peur du jugement : Thomas Braille illustre une source de force immense :

« Je n'ai pas peur du jugement des hommes, j'ai peur uniquement du jugement de mon fils ».

Se libérer de la peur du regard social permet d'agir avec une plus grande liberté et une plus grande force.

• Le plus grand renoncement : renoncer à plaire. Cette phrase puissante de Flore Vasseur résume un acte de libération essentiel.

Un militant ne peut pas plaire à tout le monde. L'accepter, c'est se libérer d'un poids immense.

• Se nourrir de la joie : Malgré les difficultés, l'engagement est aussi une source de joies intenses.

Flore Vasseur rappelle rencontrer « plus souvent des moments de joie quasi extatique que des moments de burnout ».

Le lien, la solidarité et les petites victoires sont des nourritures essentielles.

--------------------------------------------------------------------------------

4. Conclusion : L'engagement, un marathon pour la vie

En définitive, l'engagement sur le long terme s'apparente bien plus à un marathon qu'à un sprint. Les stratégies pour durer ne sont pas des distractions ou des luxes, mais des composantes essentielles de la lutte elle-même.

Prendre soin de soi, cultiver la force du collectif, ajuster sa perspective et agir depuis un lieu d'authenticité sont les conditions de la victoire.

En acceptant la nature cyclique de l'énergie et en se rappelant constamment son « pourquoi », il devient possible non seulement de tenir, mais aussi de s'épanouir dans l'action.

Comme le disait Baden-Powell, cité par Anaïs Terrien, l'objectif n'est peut-être pas de sauver le monde seul et tout de suite, mais plus humblement et plus durablement d'« essayer de laisser le monde un peu meilleur que quand vous êtes arrivé ».

Interesting

ja

7:37 "Anspruchsdenken"

ja, vor allem das "recht auf leben"... 90% aller menschen sind überflüssig aber wollen trotzdem leben. warum?

• Ineffectiveness of the Education System: Traditional education often focuses on delivering vast amounts of material without teaching the actual tools or techniques for effective memorization and information retention.
• Passive vs. Active Learning: Scientific research (notably by Craik and Watkins in 1973) demonstrates that passive repetition has almost zero impact on long-term memory and knowledge building.
• Definition of Passive Reproduction: This refers to "mindless" repetition where information is maintained in short-term memory without any cognitive processing or mental engagement (e.g., repeating a phone number just long enough to dial it).
• Common Mistakes: Rote memorization of facts/dates, highlighting text without deep thought, and copying notes verbatim are largely ineffective and represent a significant waste of time.
• The Importance of Active Engagement: To build lasting knowledge, one must engage with the material through "active learning"—this involves speculating, questioning, drawing connections, and integrating new facts into existing mental models.
• Efficiency and Time Management: Using active methods can allow a learner to process in 15 minutes what might otherwise take 5 to 10 hours using passive, repetitive methods.
• Building Mental Frameworks: True specialization in any field requires effective memory tools to connect isolated bits of information into useful, functional models of knowledge.

I'm just not a scalpel person! Ha!

DOI: 10.1016/j.celrep.2025.116753

Resource: (BioLegend Cat# 105838, RRID:AB_2616739)

Curator: @scibot

SciCrunch record: RRID:AB_2616739

What is this?

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

Learn more at Review Commons

Reply to the reviewers

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

Summary: In this study, the authors used proximity proteomics in U2OS cells to identify several E3 ubiquitin ligases recruited to stress granules (SGs), and they focused on MKRN2 as a novel regulator. They show that MKRN2 localization to SGs requires active ubiquitination via UBA1. Functional experiments demonstrated that MKRN2 knockdown increases the number of SG condensates, reduces their size, slightly raises SG liquidity during assembly, and slows disassembly after heat shock. Overexpression of MKRN2-GFP combined with confocal imaging revealed co-localization of MKRN2 and ubiquitin in SGs. By perturbing ubiquitination (using a UBA1 inhibitor) and inducing defective ribosomal products (DRiPs) with O-propargyl puromycin, they found that both ubiquitination inhibition and MKRN2 depletion lead to increased accumulation of DRiPs in SGs. The authors conclude that MKRN2 supports granulostasis, the maintenance of SG homeostasis , through its ubiquitin ligase activity, preventing pathological DRiP accumulation within SGs.

Major comments: - Are the key conclusions convincing? The key conclusions are partially convincing. The data supporting the role of ubiquitination and MKRN2 in regulating SG condensate dynamics are coherent, well controlled, and consistent with previous literature, making this part of the study solid and credible. However, the conclusions regarding the ubiquitin-dependent recruitment of MKRN2 to SGs, its relationship with UBA1 activity, the functional impact of the MKRN2 knockdown for DRiP accumulation are less thoroughly supported. These aspects would benefit from additional mechanistic evidence, validation in complementary model systems, or the use of alternative methodological approaches to strengthen the causal connections drawn by the authors. - Should the authors qualify some of their claims as preliminary or speculative, or remove them altogether? The authors should qualify some of their claims as preliminary. 1) MKRN2 recruitment to SGs (ubiquitin-dependent): The proteomics and IF data are a reasonable starting point, but they do not yet establish that MKRN2 is recruited from its physiological localization to SGs in a ubiquitin-dependent manner. To avoid overstating this point the authors should qualify the claim and/or provide additional controls: show baseline localization of endogenous MKRN2 under non-stress conditions (which is reported in literature to be nuclear and cytoplasmatic), include quantification of nuclear/cytoplasmic distribution, and demonstrate a shift into bona fide SG compartments after heat shock. Moreover, co-localization of overexpressed GFP-MKRN2 with poly-Ub (FK2) should be compared to a non-stress control and to UBA1-inhibition conditions to support claims of stress- and ubiquitination-dependent recruitment. *

Authors: We will stain cells for endogenous MKRN2 and quantify nuc/cyto ratio of MKRN2 without heat stress, without heat stress + TAK243, with HS and with HS + TAK243. We will do the same in the MKRN2-GFP overexpressing line while also staining for FK2.

*2) Use and interpretation of UBA1 inhibition: UBA1 inhibition effectively blocks ubiquitination globally, but it is non-selective. The manuscript should explicitly acknowledge this limitation when interpreting results from both proteomics and functional assays. Proteomics hits identified under UBA1 inhibition should be discussed as UBA1-dependent associations rather than as evidence for specific E3 ligase recruitment. The authors should consider orthogonal approaches before concluding specificity. *

Authors: We have acknowledged the limitation of using only TAK243 in our study by rephrasing statements about dependency on “ubiquitination” to “UBA1-dependent associations”.

* 3) DRiP accumulation and imaging quality: The evidence presented in Figure 5 is sufficient to substantiate the claim that DRiPs accumulate in SGs upon ubiquitination inhibition or MKRN2 depletion but to show that the event of the SGs localization and their clearance from SGs during stress is promoted by MKRN3 ubiquitin ligase activity more experiments would be needed. *

Authors: We have acknowledged the fact that our experiments do not include DRiP and SG dynamics assays using ligase-dead mutants of MKRN2 by altering our statement regarding MKRN2-mediated ubiquitination of DRiPs in the text (as proposed by reviewer 1).

*- Would additional experiments be essential to support the claims of the paper? Request additional experiments only where necessary for the paper as it is, and do not ask authors to open new lines of experimentation. Yes, a few targeted experiments would strengthen the conclusions without requiring the authors to open new lines of investigation. 1) Baseline localization of MKRN2: It would be important to show the baseline localization of endogenous and over-expressed MKRN2 (nuclear and cytoplasmic) under non-stress conditions and prior to ubiquitination inhibition. This would provide a reference to quantify redistribution into SGs and demonstrate recruitment in response to heat stress or ubiquitination-dependent mechanisms. *

Authors: We thank the reviewer for bringing this important control. We will address it in revisions.

We will quantify the nuclear/cytoplasmic distribution of endogenous and GFP-MKRN2 under control, TAK243, heat shock, and combined conditions, and assess MKRN2–ubiquitin colocalization by FK2 staining in unstressed cells.

* 2) Specificity of MKRN2 ubiquitin ligase activity: to address the non-specific effects of UBA1 inhibition and validate that observed phenotypes depend on MKRN2's ligase activity, the authors could employ a catalytically inactive MKRN2 mutant in rescue experiments. Comparing wild-type and catalytic-dead MKRN2 in the knockdown background would clarify the causal role of MKRN2 activity in SG dynamics and DRiP clearance. *

Authors: We thank the reviewer for this suggestion and have altered the phrasing of some of our statements in the text accordingly.

* 3) Ubiquitination linkage and SG marker levels: While the specific ubiquitin linkage type remains unknown, examining whether MKRN2 knockdown or overexpression affects total levels of key SG marker proteins would be informative. This could be done via Western blotting of SG markers along with ubiquitin staining, to assess whether MKRN2 influences protein stability or turnover through degradative or non-degradative ubiquitination. Such data would strengthen the mechanistic interpretation while remaining within the current study's scope. *

Authors: We thank the reviewer for requesting and will address it by performing MKRN2 KD and perform Western blot for G3BP1.

*

• Are the suggested experiments realistic in terms of time and resources? It would help if you could add an estimated cost and time investment for substantial experiments. The experiments suggested in points 1 and 3 are realistic and should not require substantial additional resources beyond those already used in the study. • Point 1 (baseline localization of MKRN2): This involves adding two control conditions (no stress and no ubiquitination inhibition) for microscopy imaging. The setup is essentially the same as in the current experiments, with time requirements mainly dependent on cell culture growth and imaging. Overall, this could be completed within a few weeks. • Point 3 (SG marker levels and ubiquitination): This entails repeating the existing experiment and adding a Western blot for SG markers and ubiquitin. The lab should already have the necessary antibodies, and the experiment could reasonably be performed within a couple of weeks. • Point 2 (catalytically inactive MKRN2 mutant and rescue experiments): This is likely more time-consuming. Designing an effective catalytic-dead mutant depends on structural knowledge of MKRN2 and may require additional validation to confirm loss of catalytic activity. If this expertise is not already present in the lab, it could significantly extend the timeline. Therefore, this experiment should be considered only if similarly recommended by other reviewers, as it represents a higher resource and time investment.

Overall, points 1 and 3 are highly feasible, while point 2 is more substantial and may require careful planning.

• Are the data and the methods presented in such a way that they can be reproduced? Yes. The methodologies used in this study to analyze SG dynamics and DRiP accumulation are well-established in the field and should be reproducible, particularly by researchers experienced in stress granule biology. Techniques such as SG assembly and disassembly assays, use of G3BP1 markers, and UBA1 inhibition are standard and clearly described. The data are generally presented in a reproducible manner; however, as noted above, some results would benefit from additional controls or complementary experiments to fully support specific conclusions.

• Are the experiments adequately replicated and statistical analysis adequate? Overall, the experiments in the manuscript appear to be adequately replicated, with most assays repeated between three and five times, as indicated in the supplementary materials. The statistical analyses used are appropriate and correctly applied to the datasets presented. However, for Figure 5 the number of experimental replicates is not reported. This should be clarified, and if the experiment was not repeated sufficiently, additional biological replicates should be performed. Given that this figure provides central evidence supporting the conclusion that DRiP accumulation depends on ubiquitination-and partly on MKRN2's ubiquitin ligase activity-adequate replication is essential. *

Authors: We thank the reviewer for noting this accidental omission. We now clarify in the legend of Figure 5 that the experiments with DRiPs were replicated three times.

Minor comments: - Specific experimental issues that are easily addressable. • For the generation and the validation of MKRN2 knockdown in UOS2 cells data are not presented in the results or in the methods sections to demonstrate the effective knockdown of the protein of interest. This point is quite essential to demonstrate the validity of the system used

Authors: We thank the reviewer for requesting and will address it by performing MKRN2 KD and perform Western blot and RT-qPCR.

• * In the supplementary figure 2 it would be useful to mention if the Western Blot represent the input (total cell lysates) before the APEX-pulldown or if it is the APEX-pulldown loaded for WB. There is no consistence in the difference of biotynilation between different replicates shown in the 2 blots. For example in R1 and R2 G3BP1-APX TAK243 the biotynilation is one if the strongest condition while on the left blot, in the same condition comparison samples R3 and R4 are less biotinilated compared to others. It would be useful to provide an explanation for that to avoid any confusion for the readers. * Authors: We have added a mention in the legend of Figure S2 that these are total cell lysates before pulldown. The apparent differences in biotin staining are small and not sufficient to question the results of our APEX-proteomics.

• * In Figure 2D, endogenous MKRN2 localization to SGs appears reduced following UBA1 inhibition. However, it is not clear whether this reduction reflects a true relocalization or a decrease in total MKRN2 protein levels. To support the interpretation that UBA1 inhibition specifically affects MKRN2 recruitment to SGs rather than its overall expression, the authors should provide data showing total MKRN2 levels remain unchanged under UBA1 inhibition, for example via Western blot of total cell lysates. * Authors: Based on first principles in regulation of gene expression, it is unlikely that total MKRN2 expression levels would decrease appreciably through transcriptional or translational regulation within the short timescale of these experiments (1 h TAK243 pretreatment followed by 90 min of heat stress).

• * DRIPs accumulation is followed during assembly but in the introduction is highlighted the fact that ubiquitination events, other reported E3 ligases and in this study data on MKRN2 showed that they play a crucial role in the disassembly of SGs which is also related with cleareance of DRIPs. Authors could add tracking DRIPs accumulation during disassembly to be added to Figure 5. I am not sure about the timeline required for this but I am just adding as optional if could be addressed easily. * Authors: We thank the reviewer for proposing this experimental direction. However, in a previous study (Ganassi et al., 2016; 10.1016/j.molcel.2016.07.021), we demonstrated that DRiP accumulation during the stress granule assembly phase drives conversion to a solid-like state and delays stress granule disassembly. It is therefore critical to assess DRiP enrichment within stress granules immediately after their formation, rather than during the stress recovery phase, as done here.

• * The authors should clarify in the text why the cutoff used for the quantification in Figure 5D (PC > 3) differs from the cutoff used elsewhere in the paper (PC > 1.5). Providing a rationale for this choice will help the reader understand the methodological consistency and ensure that differences in thresholds do not confound interpretation of the results. * Authors: We thank the reviewer for this question. The population of SGs with a DRiP enrichment > 1.5 represents SGs with a significant DRiP enrichment compared to the surrounding (background) signal. As explained in the methods, the intensity of DRiPs inside each SG is corrected by the intensity of DRiPs two pixels outside of each SG. Thus, differences in thresholds between independent experimental conditions (5B versus 5D) do not confound interpretation of the results but depend on overall staining intensity that can different between different experimental conditions. Choosing the cut-off > 3 allows to specifically highlight the population of SGs that are strongly enriched with DRiPs. MKRN2 silencing caused a strong DRiP enrichment in the majority of the SGs analyzed and therefore we chose this way of data representation. Note that the results represent the average of the analysis of 3 independent experiments with high numbers of SGs automatically segmented and analyzed/experiment. Figure 5A, B: n = 3 independent experiments; number of SGs analyzed per experiment: HS + OP-puro (695; 1216; 952); TAK-243 + HS + OP-puro (1852; 2214; 1774). Figure 5C, D: n = 3 independent experiments; number of SGs analyzed per experiment: siRNA control, HS + OP-puro (1984; 1400; 1708); siRNA MKRN2, HS + OP-puro (912; 1074; 1532).

• * For Figure 3G, the authors use over-expressed MKRN2-GFP to assess co-localization with ubiquitin in SGs. Given that a reliable antibody for endogenous MKRN2 is available and that a validated MKRN2 knockdown line exists as an appropriate control, this experiment would gain significantly in robustness and interpretability if co-localization were demonstrated using endogenous MKRN2. In the current over-expression system, MKRN2-GFP is also present in the nucleus, whereas the endogenous protein does not appear nuclear under the conditions shown. This discrepancy raises concerns about potential over-expression artifacts or mislocalization. Demonstrating co-localization using endogenous MKRN2 would avoid confounding effects associated with over-expression. If feasible, this would be a relatively straightforward experiment to implement, as it relies on tools (antibody and knockdown line) already described in the manuscript.

* Authors: We thank the reviewer for requesting and will address it by performing MKRN2 KD, FK2 immunofluorescence microscopy and perform SG partition coefficient analysis.

* - Are prior studies referenced appropriately? • From line 54 to line 67, the manuscript in total cites eight papers regarding the role of ubiquitination in SG disassembly. However, given the use of UBA1 inhibition in the initial MS-APEX experiment and the extensive prior literature on ubiquitination in SG assembly and disassembly under various stress conditions, the manuscript would benefit from citing additional relevant studies to provide more specifc examples. Expanding the references would provide stronger context, better connect the current findings to prior work, and emphasize the significance of the study in relation to established literature *

Authors: We have added citations for the relevant studies.

• *

At line 59, it would be helpful to note that G3BP1 is ubiquitinated by TRIM21 through a Lys63-linked ubiquitin chain. This information provides important mechanistic context, suggesting that ubiquitination of SG proteins in these pathways is likely non-degradative and related to functional regulation of SG dynamics rather than protein turnover. * Authors: The reviewer is correct. We have added to the text that G3BP1 is ubiquitinated through a Lys63-linked ubiquitin chain.

• *

When citing references 16 and 17, which report that the E3 ligases TRIM21 and HECT regulate SG formation, the authors should provide a plausible explanation for why these specific E3 ligases were not detected in their proteomics experiments. Differences could arise from the stress stimulus used, cell type, or experimental conditions. Similarly, since MKRN2 and other E3 ligases identified in this study have not been reported in previous works, discussing these methodological or biological differences would help prevent readers from questioning the credibility of the findings. It would also be valuable to clarify in the Conclusion that different types of stress may activate distinct ubiquitination pathways, highlighting context-dependent regulation of SG assembly and disassembly. * Authors: We thank the reviewer for this suggestion. We added to the discussion plausible explanations for why our study identified new E3 ligases.

• *

Line 59-60: when referring to the HECT family of E3 ligases involved in ubiquitination and SG disassembly, it would be more precise to report the specific E3 ligase identified in the cited studies rather than only the class of ligase. This would provide clearer mechanistic context and improve accuracy for readers. * Authors: We have added this detail to the discussion.

• *

The specific statement on line 182 "SG E3 ligases that depend on UBA1 activity are RBULs" should be supported by reference. * Authors: We have added citations to back up our claim that ZNF598, CNOT4, MKRN2, TRIM25 and TRIM26 exhibit RNA-binding activity.

*- Are the text and figures clear and accurate?

• In Supplementary Figure 1, DMSO is shown in green and the treatment in red, whereas in the main figures (Figure 1B and 1F) the colours in the legend are inverted. To avoid confusion, the colour coding in figure legends should be consistent across all figures throughout the manuscript. *

Authors: We have made the colors consistent across the main and supplementary figures.

• *

At line 79, the manuscript states that "inhibition of ubiquitination delayed fluorescence recovery dynamics of G3BP1-mCherry, relative to HS-treated cells (Figure 1F, Supplementary Fig. 6A)." However, the data shown in Figure 1F appear to indicate the opposite effect: the TAK243-treated condition (green curve) shows a faster fluorescence recovery compared to the control (red curve). This discrepancy between the text and the figure should be corrected or clarified, as it may affect the interpretation of the role of ubiquitination in SG dynamics. * Authors: Good catch. We now fixed the graphical mistake (Figure 1F and S6).

• * Line 86: adjust a missing bracket * Authors: Thank you, we fixed it.

• *

There appears to be an error in the legend of Supplementary Figure 3: the legend states that the red condition (MKRN2) forms larger aggregates, but both the main Figure 3C of the confocal images and the text indicate that MKRN2 (red) forms smaller aggregates. Please correct the legend and any corresponding labels so they are consistent with the main figure and the text. The authors should also double-check that the figure panel order, color coding, and statistical annotations match the legend and the descriptions in the Results section to avoid reader confusion.

* Authors: This unfortunate graphical mistake has been corrected.

• * At lines 129-130, the manuscript states that "FRAP analysis demonstrated that MKRN2 KD resulted in a slight increase in SG liquidity (Fig. 3F, Supplementary Fig. 6B)." However, the data shown in Figure 3F appear to indicate the opposite trend: the MKRN2 KD condition (red curve) exhibits a faster fluorescence recovery compared to the control (green curve). This discrepancy between the text and the figure should be corrected or clarified, as it directly affects the interpretation of MKRN2's role in SG disassembly. Ensuring consistency between the written description and the plotted FRAP data is essential for accurate interpretation. * Authors: We thank the reviewer and clarify in the legend of Figure 3F and the Results the correct labels: indeed faster fluorescence recovery seen in MKRN2 KD is correctly interpreted as increased liquidity in the text.

• *

At lines 132-133, the manuscript states: "Then, to further test the impact of MKRN2 on SG dynamics, we overexpressed MKRN2-GFP and observed that it was recruited to SG (Fig. 3G)." This description should be corrected or clarified, as the over-expressed MKRN2-GFP also appears to localize to the nucleus. * Authors: The text has been modified to reflect both the study of MKRN2 localization to SGs and of nuclear localization.

• *

At lines 134-135, the manuscript states that the FK2 antibody detects "free ubiquitin." This is incorrect. FK2 does not detect free ubiquitin; it recognizes only ubiquitin conjugates, including mono-ubiquitinated and poly-ubiquitinated proteins. The text should be corrected accordingly to avoid misinterpretation of the immunostaining data. * Authors: Thank you for pointing out this error. We have corrected it.

• * Figure 5A suffers from poor resolution, and no scale bar is provided, which limits interpretability. Additionally, the ROI selected for the green channel (DRIPs) appears to capture unspecific background staining, while the most obvious DRIP spots are localized in the nucleus. The authors should clarify this in the text, improve the image quality if possible, and ensure that the ROI accurately represents DRIP accumulation - in SGs rather than background signal. * Authors: We thank the reviewer for pointing the sub-optimal presentation of this figure. We modified Figure 5A to improve image quality and interpretation. Concerning the comment that “the most obvious DRIP spots are localized in the nucleus”, this is in line with our previous findings demonstrating that a fraction of DRiPs accumulates in nucleoli (Mediani et al. 2019 10.15252/embj.2018101341). To avoid misinterpretation, we modified Figure 5A as follows: (i) we provide a different image for control cells, exposed to heat shock and OP-puro; (ii) we select a ROI that only shows a few stress granules; (iii) we added arrowheads to indicate the nucleoli that are strongly enriched for DRiPs; (iv) we include a dotted line to show the nuclear membrane, helping to distinguish cytoplasm and nucleus in the red and green channel. We also include the scale bars (5 µm) in the image.

* Do you have suggestions that would help the authors improve the presentation of their data and conclusions?

• In the first paragraph following the APEX proteomics results, the authors present validation data exclusively for MKRN2, justifying this early focus by stating that MKRN2 is the most SG-depleted E3 ligase. However, in the subsequent paragraph they introduce the RBULs and present knockdown data for MKRN2 along with two additional E3 ligases identified in the screen, before once again emphasizing that MKRN2 is the most SG-depleted ligase and therefore the main focus of the study. For clarity and logical flow, the manuscript would benefit from reordering the narrative. Specifically, the authors should first present the validation data for all three selected E3 ligases, and only then justify the decision to focus on MKRN2 for in-depth characterization. In addition to the extent of its SG depletion, the authors may also consider providing biologically relevant reasons for prioritizing MKRN2 (e.g., domain architecture, known roles in stress responses, or prior evidence of ubiquitination-related functions). Reorganizing this section would improve readability and better guide the reader through the rationale for the study's focus.*

Authors: We thank the reviewer for this suggested improvement to our “storyline”. As suggested by the reviewer, we have moved the IF validation of MKRN2 to the following paragraph in order to improve the flow of the manuscript. We added additional justification to prioritizing MKRN2 citing (Youn et al. 2018 and Markmiller et al. 2018).

• *

At lines 137-138, the manuscript states: "Together these data indicate that MKRN2 regulates the assembly dynamics of SGs by promoting their coalescence during HS and can increase SG ubiquitin content." While Figure 3G shows some co-localization of MKRN2 with ubiquitin, immunofluorescence alone is insufficient to claim an increase in SG ubiquitin content. This conclusion should be supported by orthogonal experiments, such as Western blotting, in vitro ubiquitination assays, or immunoprecipitation of SG components. Including a control under no-stress conditions would also help demonstrate that ubiquitination increases specifically in response to stress. The second part of the statement should therefore be rephrased to avoid overinterpretation, for example:"...and may be associated with increased ubiquitination within SGs, as suggested by co-localization, pending further validation by complementary assays." * Authors: The statement has been rephrased in a softer way as suggested by the reviewer.

• At line 157, the statement: "Therefore, we conclude that MKRN2 ubiquitinates a subset of DRiPs, avoiding their accumulation inside SGs" should be rephrased as a preliminary observation. While the data support a role for MKRN2 in SG disassembly and a reduction of DRIPs, direct ubiquitination of DRIPs by MKRN2 has not been demonstrated. A more cautious phrasing would better reflect the current evidence and avoid overinterpretation. * * *Authors: We thank the reviewer for this suggestion and have altered the phrasing of this statement accordingly.

*Reviewer #1 (Significance (Required)):

General assessment: provide a summary of the strengths and limitations of the study. What are the strongest and most important aspects? What aspects of the study should be improved or could be developed?

• This study provides a valuable advancement in understanding the role of ubiquitination in stress granule (SG) dynamics and the clearance of SGs formed under heat stress. A major strength is the demonstration of how E3 ligases identified through proteomic screening, particularly MKRN2, influence SG assembly and disassembly in a ubiquitination- and heat stress-dependent manner. The combination of proteomics, imaging, and functional assays provides a coherent mechanistic framework linking ubiquitination to SG homeostasis. Limitations of the study include the exclusive use of a single model system (U2OS cells), which may limit generalizability. Additionally, some observations-such as MKRN2-dependent ubiquitination within SGs and changes in DRIP accumulation under different conditions-would benefit from orthogonal validation experiments (e.g., Western blotting, immunoprecipitation, or in vitro assays) to confirm and strengthen these findings. Addressing these points would enhance the robustness and broader applicability of the conclusions.

Advance: compare the study to the closest related results in the literature or highlight results reported for the first time to your knowledge; does the study extend the knowledge in the field and in which way? Describe the nature of the advance and the resulting insights (for example: conceptual, technical, clinical, mechanistic, functional,...).

• The closest related result in literature is - Yang, Cuiwei et al. "Stress granule homeostasis is modulated by TRIM21-mediated ubiquitination of G3BP1 and autophagy-dependent elimination of stress granules." Autophagy vol. 19,7 (2023): 1934-1951. doi:10.1080/15548627.2022.2164427 - demonstrating that TRIM21, an E3 ubiquitin ligase, catalyzes K63-linked ubiquitination of G3BP1, a core SG nucleator, under oxidative stress. This ubiquitination by TRIM21 inhibits SG formation, likely by altering G3BP1's propensity for phase separation. In contrast, the MKRN2 study identifies a different E3 (MKRN2) that regulates SG dynamics under heat stress and appears to influence both assembly and disassembly. This expands the role of ubiquitin ligases in SG regulation beyond those previously studied (like TRIM21).

• Gwon and colleagues (Gwon Y, Maxwell BA, Kolaitis RM, Zhang P, Kim HJ, Taylor JP. Ubiquitination of G3BP1 mediates stress granule disassembly in a context-specific manner. Science. 2021;372(6549):eabf6548. doi:10.1126/science.abf6548) have shown that K63-linked ubiquitination of G3BP1 is required for SG disassembly after heat stress. This ubiquitinated G3BP1 recruits the segregase VCP/p97, which helps extract G3BP1 from SGs for disassembly. The MKRN2 paper builds on this by linking UBA1-dependent ubiquitination and MKRN2's activity to SG disassembly. Specifically, they show MKRN2 knockdown affects disassembly, and suggest MKRN2 helps prevent accumulation of defective ribosomal products (DRiPs) in SGs, adding a new layer to the ubiquitin-VCP model.

• Ubiquitination's impact is highly stress- and context-dependent (different chain types, ubiquitin linkages, and recruitment of E3s). The MKRN2 work conceptually strengthens this idea: by showing that MKRN2's engagement with SGs depends on active ubiquitination via UBA1, and by demonstrating functional consequences (SG dynamics + DRIP accumulation), the study highlights how cellular context (e.g., heat stress) can recruit specific ubiquitin ligases to SGs and modulate their behavior.

• There is a gap in the literature: very few (if any) studies explicitly combine the biology of DRIPs, stress granules, and E3 ligase mediated ubiquitination, especially in mammalian cells. There are relevant works about DRIP biology in stress granules, but those studies focus on chaperone-based quality control, not ubiquitin ligase-mediated ubiquitination of DRIPs. This study seems to be one of the first to make that connection in mammalian (or human-like) SG biology. A work on the plant DRIP-E3 ligase TaSAP5 (Zhang N, Yin Y, Liu X, et al. The E3 Ligase TaSAP5 Alters Drought Stress Responses by Promoting the Degradation of DRIP Proteins. Plant Physiol. 2017;175(4):1878-1892. doi:10.1104/pp.17.01319 ) shows that DRIPs can be directly ubiquitinated by E3s in other biological systems - which supports the plausibility of the MKRN2 mechanism, but it's not the same context.

• A very recent review (Yuan, Lin et al. "Stress granules: emerging players in neurodegenerative diseases." Translational neurodegeneration vol. 14,1 22. 12 May. 2025, doi:10.1186/s40035-025-00482-9) summarizes and reinforces the relationship among SGs and the pathogenesis of different neurodegenerative diseases (NDDs). By identifying MKRN2 as a new ubiquitin regulator in SGs, the current study could have relevance for neurodegeneration and proteotoxic diseases, providing a new candidate to explore in disease models.

Audience: describe the type of audience ("specialized", "broad", "basic research", "translational/clinical", etc...) that will be interested or influenced by this research; how will this research be used by others; will it be of interest beyond the specific field?

The audience for this paper is primarily specialized, including researchers in stress granule biology, ubiquitin signaling, protein quality control, ribosome biology, and cellular stress responses. The findings will also be of interest to scientists working on granulostasis, nascent protein surveillance, and proteostasis mechanisms. Beyond these specific fields, the study provides preliminary evidence linking ubiquitination to DRIP handling and SG dynamics, which may stimulate new research directions and collaborative efforts across complementary areas of cell biology and molecular biology.

• Please define your field of expertise with a few keywords to help the authors contextualize your point of view. Indicate if there are any parts of the paper that you do not have sufficient expertise to evaluate.

I work in ubiquitin biology, focusing on ubiquitination signaling in physiological and disease contexts, with particular expertise in the identification of E3 ligases and their substrates across different cellular systems and in vivo models. I have less expertise in stress granule dynamics and DRiP biology, so my evaluation of those aspects is more limited and relies on interpretation of the data presented in the manuscript.

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

This study identifies the E3 ubiquitin ligase Makorin 2 (MKRN2) as a novel regulator of stress granule (SG) dynamics and proteostasis. Using APEX proximity proteomics, the authors demonstrate that inhibition of the ubiquitin-activating enzyme UBA1 with TAK243 alters the SG proteome, leading to depletion of several E3 ligases, chaperones, and VCP cofactors. Detailed characterization of MKRN2 reveals that it localizes to SGs in a ubiquitination-dependent manner and is required for proper SG assembly, coalescence, and disassembly. Functionally, MKRN2 prevents the accumulation of defective ribosomal products (DRiPs) within SGs, thereby maintaining granulostasis. The study provides compelling evidence that ubiquitination, mediated specifically by MKRN2, plays a critical role in surveilling stress-damaged proteins within SGs and maintaining their dynamic liquid-like properties. Major issues: 1. Figures 1-2: Temporal dynamics of ubiquitination in SGs. The APEX proteomics was performed at a single timepoint (90 min heat stress), yet the live imaging data show that SG dynamics and TAK243 effects vary considerably over time: • The peak or SG nucleation was actually at 10-30 min (Figure 1B). • TAK243 treatment causes earlier SG nucleation (Figure 1B) but delayed disassembly (Figure 1A-B, D). A temporal proteomic analysis at multiple timepoints (e.g., 30 min, 60 min, 90 min of heat stress, and during recovery) would reveal whether MKRN2 and other ubiquitination-dependent proteins are recruited to SGs dynamically during the stress response. It would also delineate whether different E3 ligases predominate at different stages of the SG lifecycle. While such experiments may be beyond the scope of the current study, the authors should at minimum discuss this limitation and acknowledge that the single-timepoint analysis may miss dynamic changes in SG composition. *

Authors: We thank the reviewer for identifying this caveat in our methodology. We now discuss this limitation and acknowledge that the single-timepoint analysis may miss dynamic changes in SG composition.

* Figures 2D-E, 3G: MKRN2 localization mechanism requires clarification. The authors demonstrate that MKRN2 localization to SGs is dependent on active ubiquitination, as TAK243 treatment significantly reduces MKRN2 partitioning into SGs (Figure 2D-E). However, several mechanistic questions remain: • Does MKRN2 localize to SGs through binding to ubiquitinated substrates within SGs, or does MKRN2 require its own ubiquitination activity to enter SGs? • The observation that MKRN2 overexpression increases SG ubiquitin content (Figure 3G-H) could indicate either: (a) MKRN2 actively ubiquitinates substrates within SGs, or (b) MKRN2 recruitment brings along pre-ubiquitinated substrates from the cytoplasm. • Is MKRN2 localization to SGs dependent on its E3 ligase activity? A catalytically inactive mutant of MKRN2 would help distinguish whether MKRN2 must actively ubiquitinate proteins to remain in SGs or whether it binds to ubiquitinated proteins independently of its catalytic activity. The authors should clarify whether MKRN2's SG localization depends on its catalytic activity or on binding to ubiquitinated proteins, as this would fundamentally affect the interpretation of its role in SG dynamics. *

Authors: We thank the reviewer for this experimental suggestion. We will perform an analysis of the SG partitioning coefficient between WT-MKRN2 and a RING mutant of MKRN2.

* Figures 3-4: Discrepancy between assembly and disassembly phenotypes. MKRN2 knockdown produces distinct phenotypes during SG assembly versus disassembly. During assembly: smaller, more numerous SGs that fail to coalesce (Figure 3A-E), while during disassembly: delayed SG clearance (Figure 4A-D). These phenotypes may reflect different roles for MKRN2 at different stages, but the mechanism underlying this stage-specificity is unclear: • Does MKRN2 have different substrates or utilize different ubiquitin chain types during assembly versus disassembly? • The increased SG liquidity upon MKRN2 depletion (Figure 3F) seems paradoxical with delayed disassembly- typically more liquid condensates disassemble faster. The authors interpret this as decreased coalescence into "dense and mature SGs," but this requires clarification. • How does prevention of DRiP accumulation relate to the assembly defect? One would predict that DRiP accumulation would primarily affect disassembly (by reducing liquidity), yet MKRN2 depletion impacts both assembly dynamics and DRiP accumulation. The authors should discuss how MKRN2's role in preventing DRiP accumulation mechanistically connects to both the assembly and disassembly phenotypes. *

Authors: We thank the reviewer and will add to the Discussion a mention of a precedent for this precise phenotype from our previous work (Seguin et al., 2014).

* Figure 5: Incomplete characterization of MKRN2 substrates. While the authors convincingly demonstrate that MKRN2 prevents DRiP accumulation in SGs (Figure 5C-D), the direct substrates of MKRN2 remain unknown. The authors acknowledge in the limitations that "the direct MKRN2 substrates and ubiquitin-chain types (K63/K48) are currently unknown." However, several approaches could strengthen the mechanistic understanding: • Do DRiPs represent direct MKRN2 substrates? Co-immunoprecipitation of MKRN2 followed by ubiquitin-chain specific antibodies (K48 vs K63) could reveal whether MKRN2 mediates degradative (K48) or non-degradative (K63) ubiquitination. *

Authors: The DRiPs generated in the study represent truncated versions of all the proteins that were in the process of being synthesized by the cell at the moment of the stress, and therefore include both MKRN2 specific substrates and MKRN2 independent substrates. Identifying specific MKRN2 substrates, while interesting as a new research avenue, is not within the scope of the present study.

• * Given that VCP cofactors (such as UFD1L, PLAA) are depleted from SGs upon UBA1 inhibition (Figure 2C) and these cofactors recognize ubiquitinated substrates, does MKRN2 function upstream of VCP recruitment? Testing whether MKRN2 depletion affects VCP cofactor localization to SGs would clarify this pathway. * Authors: We thank the reviewer for requesting and will address it by performing MKRN2 KD, VCP immunofluorescence microscopy and perform SG partition coefficient analysis.

• * The authors note that MKRN2 knockdown produces a phenotype reminiscent of VCP inhibition-smaller, more numerous SGs with increased DRiP partitioning. This similarity suggests MKRN2 may function in the same pathway as VCP. Direct epistasis experiments would strengthen this connection. * Authors: This study is conditional results of the above study. If VCP partitioning to SGs is reduced upon MKRN2 KD, which we do not know at this point, then MKRN2/VCP double KD experiment will be performed to strengthen this connection.

* Alternative explanations for the phenotype of delayed disassembly with TAK243 or MKRN2 depletion- the authors attribute this to DRiP accumulation, but TAK243 affects global ubiquitination. Could impaired degradation of other SG proteins (not just DRiPs) contribute to delayed disassembly? Does proteasome inhibition (MG-132 treatment) phenocopy the MKRN2 depletion phenotype? This would support that MKRN2-mediated proteasomal degradation (via K48 ubiquitin chains) is key to the phenotype. *

Authors: We are happy to provide alternative explanations in the Discussion in line with Reviewer #2 suggestion. The role of the proteosome is out of the scope of our study.

• Comparison with other E3 ligases (Supplementary Figure 5): The authors show that CNOT4 and ZNF598 depletion also affect SG dynamics, though to lesser extents than MKRN2. However: • Do these E3 ligases also prevent DRiP accumulation in SGs? Testing OP-puro partitioning in CNOT4- or ZNF598-depleted cells would reveal whether DRiP clearance is a general feature of SG-localized E3 ligases or specific to MKRN2. *

• * Are there redundant or compensatory relationships between these E3 ligases? Do double knockdowns have additive effects? * Authors: Our paper presents a study of the E3 ligase MKRN2. Generalizing these observations to ZNF598, CNOT4 and perhaps an even longer list of E3s, may be an interesting question, outside the scope of our mission.

• * The authors note that MKRN2 is "the most highly SG-depleted E3 upon TAK243 treatment"-does this mean MKRN2 has the strongest dependence on active ubiquitination for its SG localization, or simply that it has the highest basal level of SG partitioning? * Authors: We thank the reviewer for this smart question. MKRN2 has the strongest dependence on active ubiquitination as we now clarify better in the Results.

*Reviewer #2 (Significance (Required)):

This is a well-executed study that identifies MKRN2 as an important regulator of stress granule dynamics and proteostasis. The combination of proximity proteomics, live imaging, and functional assays provides strong evidence for MKRN2's role in preventing DRiP accumulation and maintaining granulostasis. However, key mechanistic questions remain, particularly regarding MKRN2's direct substrates, the ubiquitin chain types it generates, and how its enzymatic activity specifically prevents DRiP accumulation while promoting both SG coalescence and disassembly. Addressing the suggested revisions, particularly those related to MKRN2's mechanism of SG localization and substrate specificity, would significantly strengthen the manuscript and provide clearer insights into how ubiquitination maintains the dynamic properties of stress granules under proteotoxic stress.

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

In this paper, Amzallag et al. investigate the relationship between ubiquitination and the dynamics of stress granules (SGs). They utilize proximity ligation coupled mass spectrometry to identify SG components under conditions where the proteasome is inhibited by a small drug that targets UBiquitin-like modifier Activating enzyme 1 (UBA1), which is crucial for the initial step in the ubiquitination of misfolded proteins. Their findings reveal that the E3 ligase Makorin2 (MKRN2) is a novel component of SGs. Additionally, their data suggest that MKRN2 is necessary for processing damaged ribosome-associated proteins (DRIPs) during heat shock (HS). In the absence of MKRN2, DRIPs accumulate in SGs, which affects their dynamics. Major comments: Assess the knockdown efficiency (KD) for CNOT1, ZNF598, and MKRN2 to determine if the significant effect observed on SG dynamics upon MKRN2 depletion is due to the protein's function rather than any possible differences in KD efficiency. *

Authors: To address potential variability in knockdown efficiency, we will quantify CNOT4, ZNF598, and MKRN2 mRNA levels by RT-qPCR following siRNA knockdown.

* Since HS-induced stress granules (SGs) are influenced by the presence of TAK-243 or MKRN2 depletion, could it be that these granules become more mature and thus acquire more defective ribosomal products (DRIPs)? Do HS cells reach the same level of DRIPs, as assessed by OP-Puro staining, at a later time point? *

Authors: an interesting question. Mateju et al. carefully characterized the time course of DRiP accumulation in stress granules during heat shock, decreasing after the 90 minutes point (Appendix Figure S7; 10.15252/embj.201695957). We therefore interpret DRiP accumulation in stress granules following TAK243 treatment as a pathological state, reflecting impaired removal and degradation of DRiPs, rather than a normal, more “mature” stress granule state.

* Incorporating OP-Puro can lead to premature translation termination, potentially confounding results. Consider treating cells with a short pulse (i.e., 5 minutes) of OP-Puro just before fixation. *

Authors: Thank you for this suggestion. Treating the cell with a short pulse of OP-Puro just before fixation will lead to the labelling of a small amount of proteins, likely undetectable using conventional microscopy or Western blotting. Furthermore, it will lead to the unwanted labeling of stress responsive proteins that are translated with non canonical cap-independent mechanisms upon stress.

* Is MKRN2's dependence limited to HS-induced SGs? *

Authors: We will test sodium arsenite–induced stress and use immunofluorescence at discrete time points to assess whether the heat shock–related observations generalize to other stress types.

*

Minor comments: Abstract: Introduce UBA1. Introduction: The reference [2] should be replaced with 25719440. Results: Line 70, 'G3BP1 and 2 genes,' is somewhat misleading. Consider rephrasing into 'G3BP1 and G3BP2 genes'. Line 103: considers rephrasing 'we orthogonally validated the ubiquitin-dependent interaction' to 'we orthogonally validated the ubiquitin-dependent stress granule localization'. Line 125: '(fig.3C, EI Supplementary fig. 3)' Remove 'I'. Methods: line 260: the reference is not linked (it should be ref. [26]). Line 225: Are all the KDs being performed using the same method? Please specify. *

Authors: The text has been altered to reflect the reviewer’s suggestions.

*Fig.2C: Consider adding 'DEPLETED' on top of the scheme.

Reviewer #3 (Significance (Required)):

The study offers valuable insights into the degradative processes associated with SGs. The figures are clear, and the experimental quality is high. The authors do not overstate or overinterpret their findings, and the results effectively support their claims. However, the study lacks orthogonal methods to validate the findings and enhance the results. For instance, incorporating biochemical and reporter-based methods to measure degradation-related intermediate products (DRIPs) would be beneficial. Additionally, utilizing multiple methods to block ubiquitination, studying the dynamics of MKRN2 on SGs, and examining the consequences of excessive DRIPs on the cell fitness of SGs would further strengthen the research. *

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

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

This study identifies the E3 ubiquitin ligase Makorin 2 (MKRN2) as a novel regulator of stress granule (SG) dynamics and proteostasis. Using APEX proximity proteomics, the authors demonstrate that inhibition of the ubiquitin-activating enzyme UBA1 with TAK243 alters the SG proteome, leading to depletion of several E3 ligases, chaperones, and VCP cofactors. Detailed characterization of MKRN2 reveals that it localizes to SGs in a ubiquitination-dependent manner and is required for proper SG assembly, coalescence, and disassembly. Functionally, MKRN2 prevents the accumulation of defective ribosomal products (DRiPs) within SGs, thereby maintaining granulostasis. The study provides compelling evidence that ubiquitination, mediated specifically by MKRN2, plays a critical role in surveilling stress-damaged proteins within SGs and maintaining their dynamic liquid-like properties.

Major issues:

1. Figures 1-2: Temporal dynamics of ubiquitination in SGs. The APEX proteomics was performed at a single timepoint (90 min heat stress), yet the live imaging data show that SG dynamics and TAK243 effects vary considerably over time:
   • The peak or SG nucleation was actually at 10-30 min (Figure 1B).
   • TAK243 treatment causes earlier SG nucleation (Figure 1B) but delayed disassembly (Figure 1A-B, D). A temporal proteomic analysis at multiple timepoints (e.g., 30 min, 60 min, 90 min of heat stress, and during recovery) would reveal whether MKRN2 and other ubiquitination-dependent proteins are recruited to SGs dynamically during the stress response. It would also delineate whether different E3 ligases predominate at different stages of the SG lifecycle. While such experiments may be beyond the scope of the current study, the authors should at minimum discuss this limitation and acknowledge that the single-timepoint analysis may miss dynamic changes in SG composition.
2. Figures 2D-E, 3G: MKRN2 localization mechanism requires clarification. The authors demonstrate that MKRN2 localization to SGs is dependent on active ubiquitination, as TAK243 treatment significantly reduces MKRN2 partitioning into SGs (Figure 2D-E). However, several mechanistic questions remain:
   • Does MKRN2 localize to SGs through binding to ubiquitinated substrates within SGs, or does MKRN2 require its own ubiquitination activity to enter SGs?
   • The observation that MKRN2 overexpression increases SG ubiquitin content (Figure 3G-H) could indicate either: (a) MKRN2 actively ubiquitinates substrates within SGs, or (b) MKRN2 recruitment brings along pre-ubiquitinated substrates from the cytoplasm.
   • Is MKRN2 localization to SGs dependent on its E3 ligase activity? A catalytically inactive mutant of MKRN2 would help distinguish whether MKRN2 must actively ubiquitinate proteins to remain in SGs or whether it binds to ubiquitinated proteins independently of its catalytic activity. The authors should clarify whether MKRN2's SG localization depends on its catalytic activity or on binding to ubiquitinated proteins, as this would fundamentally affect the interpretation of its role in SG dynamics.
3. Figures 3-4: Discrepancy between assembly and disassembly phenotypes. MKRN2 knockdown produces distinct phenotypes during SG assembly versus disassembly. During assembly: smaller, more numerous SGs that fail to coalesce (Figure 3A-E), while during disassembly: delayed SG clearance (Figure 4A-D). These phenotypes may reflect different roles for MKRN2 at different stages, but the mechanism underlying this stage-specificity is unclear:
   • Does MKRN2 have different substrates or utilize different ubiquitin chain types during assembly versus disassembly?
   • The increased SG liquidity upon MKRN2 depletion (Figure 3F) seems paradoxical with delayed disassembly- typically more liquid condensates disassemble faster. The authors interpret this as decreased coalescence into "dense and mature SGs," but this requires clarification.
   • How does prevention of DRiP accumulation relate to the assembly defect? One would predict that DRiP accumulation would primarily affect disassembly (by reducing liquidity), yet MKRN2 depletion impacts both assembly dynamics and DRiP accumulation. The authors should discuss how MKRN2's role in preventing DRiP accumulation mechanistically connects to both the assembly and disassembly phenotypes.
4. Figure 5: Incomplete characterization of MKRN2 substrates. While the authors convincingly demonstrate that MKRN2 prevents DRiP accumulation in SGs (Figure 5C-D), the direct substrates of MKRN2 remain unknown. The authors acknowledge in the limitations that "the direct MKRN2 substrates and ubiquitin-chain types (K63/K48) are currently unknown." However, several approaches could strengthen the mechanistic understanding:
   • Do DRiPs represent direct MKRN2 substrates? Co-immunoprecipitation of MKRN2 followed by ubiquitin-chain specific antibodies (K48 vs K63) could reveal whether MKRN2 mediates degradative (K48) or non-degradative (K63) ubiquitination.
   • Given that VCP cofactors (such as UFD1L, PLAA) are depleted from SGs upon UBA1 inhibition (Figure 2C) and these cofactors recognize ubiquitinated substrates, does MKRN2 function upstream of VCP recruitment? Testing whether MKRN2 depletion affects VCP cofactor localization to SGs would clarify this pathway.
   • The authors note that MKRN2 knockdown produces a phenotype reminiscent of VCP inhibition-smaller, more numerous SGs with increased DRiP partitioning. This similarity suggests MKRN2 may function in the same pathway as VCP. Direct epistasis experiments would strengthen this connection.
5. Alternative explanations for the phenotype of delayed disassembly with TAK243 or MKRN2 depletion- the authors attribute this to DRiP accumulation, but TAK243 affects global ubiquitination. Could impaired degradation of other SG proteins (not just DRiPs) contribute to delayed disassembly? Does proteasome inhibition (MG-132 treatment) phenocopy the MKRN2 depletion phenotype? This would support that MKRN2-mediated proteasomal degradation (via K48 ubiquitin chains) is key to the phenotype.
6. Comparison with other E3 ligases (Supplementary Figure 5): The authors show that CNOT4 and ZNF598 depletion also affect SG dynamics, though to lesser extents than MKRN2. However:
   • Do these E3 ligases also prevent DRiP accumulation in SGs? Testing OP-puro partitioning in CNOT4- or ZNF598-depleted cells would reveal whether DRiP clearance is a general feature of SG-localized E3 ligases or specific to MKRN2.
   • Are there redundant or compensatory relationships between these E3 ligases? Do double knockdowns have additive effects?
   • The authors note that MKRN2 is "the most highly SG-depleted E3 upon TAK243 treatment"-does this mean MKRN2 has the strongest dependence on active ubiquitination for its SG localization, or simply that it has the highest basal level of SG partitioning?

Significance

This is a well-executed study that identifies MKRN2 as an important regulator of stress granule dynamics and proteostasis. The combination of proximity proteomics, live imaging, and functional assays provides strong evidence for MKRN2's role in preventing DRiP accumulation and maintaining granulostasis. However, key mechanistic questions remain, particularly regarding MKRN2's direct substrates, the ubiquitin chain types it generates, and how its enzymatic activity specifically prevents DRiP accumulation while promoting both SG coalescence and disassembly. Addressing the suggested revisions, particularly those related to MKRN2's mechanism of SG localization and substrate specificity, would significantly strengthen the manuscript and provide clearer insights into how ubiquitination maintains the dynamic properties of stress granules under proteotoxic stress.

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

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

Summary:

In this study, the authors used proximity proteomics in U2OS cells to identify several E3 ubiquitin ligases recruited to stress granules (SGs), and they focused on MKRN2 as a novel regulator. They show that MKRN2 localization to SGs requires active ubiquitination via UBA1. Functional experiments demonstrated that MKRN2 knockdown increases the number of SG condensates, reduces their size, slightly raises SG liquidity during assembly, and slows disassembly after heat shock. Overexpression of MKRN2-GFP combined with confocal imaging revealed co-localization of MKRN2 and ubiquitin in SGs. By perturbing ubiquitination (using a UBA1 inhibitor) and inducing defective ribosomal products (DRiPs) with O-propargyl puromycin, they found that both ubiquitination inhibition and MKRN2 depletion lead to increased accumulation of DRiPs in SGs. The authors conclude that MKRN2 supports granulostasis, the maintenance of SG homeostasis , through its ubiquitin ligase activity, preventing pathological DRiP accumulation within SGs.

Major comments:

• Are the key conclusions convincing?

The key conclusions are partially convincing. The data supporting the role of ubiquitination and MKRN2 in regulating SG condensate dynamics are coherent, well controlled, and consistent with previous literature, making this part of the study solid and credible. However, the conclusions regarding the ubiquitin-dependent recruitment of MKRN2 to SGs, its relationship with UBA1 activity, the functional impact of the MKRN2 knockdown for DRiP accumulation are less thoroughly supported. These aspects would benefit from additional mechanistic evidence, validation in complementary model systems, or the use of alternative methodological approaches to strengthen the causal connections drawn by the authors. - Should the authors qualify some of their claims as preliminary or speculative, or remove them altogether? The authors should qualify some of their claims as preliminary.

1) MKRN2 recruitment to SGs (ubiquitin-dependent): The proteomics and IF data are a reasonable starting point, but they do not yet establish that MKRN2 is recruited from its physiological localization to SGs in a ubiquitin-dependent manner. To avoid overstating this point the authors should qualify the claim and/or provide additional controls: show baseline localization of endogenous MKRN2 under non-stress conditions (which is reported in literature to be nuclear and cytoplasmatic), include quantification of nuclear/cytoplasmic distribution, and demonstrate a shift into bona fide SG compartments after heat shock. Moreover, co-localization of overexpressed GFP-MKRN2 with poly-Ub (FK2) should be compared to a non-stress control and to UBA1-inhibition conditions to support claims of stress- and ubiquitination-dependent recruitment.

2) Use and interpretation of UBA1 inhibition: UBA1 inhibition effectively blocks ubiquitination globally, but it is non-selective. The manuscript should explicitly acknowledge this limitation when interpreting results from both proteomics and functional assays. Proteomics hits identified under UBA1 inhibition should be discussed as UBA1-dependent associations rather than as evidence for specific E3 ligase recruitment. The authors should consider orthogonal approaches before concluding specificity.

3) DRiP accumulation and imaging quality: The evidence presented in Figure 5 is sufficient to substantiate the claim that DRiPs accumulate in SGs upon ubiquitination inhibition or MKRN2 depletion but to show that the event of the SGs localization and their clearance from SGs during stress is promoted by MKRN3 ubiquitin ligase activity more experiments would be needed. - Would additional experiments be essential to support the claims of the paper? Request additional experiments only where necessary for the paper as it is, and do not ask authors to open new lines of experimentation. Yes, a few targeted experiments would strengthen the conclusions without requiring the authors to open new lines of investigation.

1) Baseline localization of MKRN2: It would be important to show the baseline localization of endogenous and over-expressed MKRN2 (nuclear and cytoplasmic) under non-stress conditions and prior to ubiquitination inhibition. This would provide a reference to quantify redistribution into SGs and demonstrate recruitment in response to heat stress or ubiquitination-dependent mechanisms.

2) Specificity of MKRN2 ubiquitin ligase activity: to address the non-specific effects of UBA1 inhibition and validate that observed phenotypes depend on MKRN2's ligase activity, the authors could employ a catalytically inactive MKRN2 mutant in rescue experiments. Comparing wild-type and catalytic-dead MKRN2 in the knockdown background would clarify the causal role of MKRN2 activity in SG dynamics and DRiP clearance.

3) Ubiquitination linkage and SG marker levels: While the specific ubiquitin linkage type remains unknown, examining whether MKRN2 knockdown or overexpression affects total levels of key SG marker proteins would be informative. This could be done via Western blotting of SG markers along with ubiquitin staining, to assess whether MKRN2 influences protein stability or turnover through degradative or non-degradative ubiquitination. Such data would strengthen the mechanistic interpretation while remaining within the current study's scope. - Are the suggested experiments realistic in terms of time and resources? It would help if you could add an estimated cost and time investment for substantial experiments. The experiments suggested in points 1 and 3 are realistic and should not require substantial additional resources beyond those already used in the study. - Point 1 (baseline localization of MKRN2): This involves adding two control conditions (no stress and no ubiquitination inhibition) for microscopy imaging. The setup is essentially the same as in the current experiments, with time requirements mainly dependent on cell culture growth and imaging. Overall, this could be completed within a few weeks. - Point 3 (SG marker levels and ubiquitination): This entails repeating the existing experiment and adding a Western blot for SG markers and ubiquitin. The lab should already have the necessary antibodies, and the experiment could reasonably be performed within a couple of weeks. - Point 2 (catalytically inactive MKRN2 mutant and rescue experiments): This is likely more time-consuming. Designing an effective catalytic-dead mutant depends on structural knowledge of MKRN2 and may require additional validation to confirm loss of catalytic activity. If this expertise is not already present in the lab, it could significantly extend the timeline. Therefore, this experiment should be considered only if similarly recommended by other reviewers, as it represents a higher resource and time investment.

Overall, points 1 and 3 are highly feasible, while point 2 is more substantial and may require careful planning. - Are the data and the methods presented in such a way that they can be reproduced?

Yes. The methodologies used in this study to analyze SG dynamics and DRiP accumulation are well-established in the field and should be reproducible, particularly by researchers experienced in stress granule biology. Techniques such as SG assembly and disassembly assays, use of G3BP1 markers, and UBA1 inhibition are standard and clearly described. The data are generally presented in a reproducible manner; however, as noted above, some results would benefit from additional controls or complementary experiments to fully support specific conclusions. - Are the experiments adequately replicated and statistical analysis adequate?

Overall, the experiments in the manuscript appear to be adequately replicated, with most assays repeated between three and five times, as indicated in the supplementary materials. The statistical analyses used are appropriate and correctly applied to the datasets presented. However, for Figure 5 the number of experimental replicates is not reported. This should be clarified, and if the experiment was not repeated sufficiently, additional biological replicates should be performed. Given that this figure provides central evidence supporting the conclusion that DRiP accumulation depends on ubiquitination-and partly on MKRN2's ubiquitin ligase activity-adequate replication is essential.

Minor comments:

• Specific experimental issues that are easily addressable.
  • For the generation and the validation of MKRN2 knockdown in UOS2 cells data are not presented in the results or in the methods sections to demonstrate the effective knockdown of the protein of interest. This point is quite essential to demonstrate the validity of the system used
  • In the supplementary figure 2 it would be useful to mention if the Western Blot represent the input (total cell lysates) before the APEX-pulldown or if it is the APEX-pulldown loaded for WB. There is no consistence in the difference of biotynilation between different replicates shown in the 2 blots. For example in R1 and R2 G3BP1-APX TAK243 the biotynilation is one if the strongest condition while on the left blot, in the same condition comparison samples R3 and R4 are less biotinilated compared to others. It would be useful to provide an explanation for that to avoid any confusion for the readers.
  • In Figure 2D, endogenous MKRN2 localization to SGs appears reduced following UBA1 inhibition. However, it is not clear whether this reduction reflects a true relocalization or a decrease in total MKRN2 protein levels. To support the interpretation that UBA1 inhibition specifically affects MKRN2 recruitment to SGs rather than its overall expression, the authors should provide data showing total MKRN2 levels remain unchanged under UBA1 inhibition, for example via Western blot of total cell lysates.
  • DRIPs accumulation is followed during assembly but in the introduction is highlighted the fact that ubiquitination events, other reported E3 ligases and in this study data on MKRN2 showed that they play a crucial role in the disassembly of SGs which is also related with cleareance of DRIPs. Authors could add tracking DRIPs accumulation during disassembly to be added to Figure 5. I am not sure about the timeline required for this but I am just adding as optional if could be addressed easily.
  • The authors should clarify in the text why the cutoff used for the quantification in Figure 5D (PC > 3) differs from the cutoff used elsewhere in the paper (PC > 1.5). Providing a rationale for this choice will help the reader understand the methodological consistency and ensure that differences in thresholds do not confound interpretation of the results.
  • For Figure 3G, the authors use over-expressed MKRN2-GFP to assess co-localization with ubiquitin in SGs. Given that a reliable antibody for endogenous MKRN2 is available and that a validated MKRN2 knockdown line exists as an appropriate control, this experiment would gain significantly in robustness and interpretability if co-localization were demonstrated using endogenous MKRN2. In the current over-expression system, MKRN2-GFP is also present in the nucleus, whereas the endogenous protein does not appear nuclear under the conditions shown. This discrepancy raises concerns about potential over-expression artifacts or mislocalization. Demonstrating co-localization using endogenous MKRN2 would avoid confounding effects associated with over-expression. If feasible, this would be a relatively straightforward experiment to implement, as it relies on tools (antibody and knockdown line) already described in the manuscript.

• Are prior studies referenced appropriately?

  • From line 54 to line 67, the manuscript in total cites eight papers regarding the role of ubiquitination in SG disassembly. However, given the use of UBA1 inhibition in the initial MS-APEX experiment and the extensive prior literature on ubiquitination in SG assembly and disassembly under various stress conditions, the manuscript would benefit from citing additional relevant studies to provide more specifc examples. Expanding the references would provide stronger context, better connect the current findings to prior work, and emphasize the significance of the study in relation to established literature
  • At line 59, it would be helpful to note that G3BP1 is ubiquitinated by TRIM21 through a Lys63-linked ubiquitin chain. This information provides important mechanistic context, suggesting that ubiquitination of SG proteins in these pathways is likely non-degradative and related to functional regulation of SG dynamics rather than protein turnover.
  • When citing references 16 and 17, which report that the E3 ligases TRIM21 and HECT regulate SG formation, the authors should provide a plausible explanation for why these specific E3 ligases were not detected in their proteomics experiments. Differences could arise from the stress stimulus used, cell type, or experimental conditions. Similarly, since MKRN2 and other E3 ligases identified in this study have not been reported in previous works, discussing these methodological or biological differences would help prevent readers from questioning the credibility of the findings. It would also be valuable to clarify in the Conclusion that different types of stress may activate distinct ubiquitination pathways, highlighting context-dependent regulation of SG assembly and disassembly.
  • Line 59-60: when referring to the HECT family of E3 ligases involved in ubiquitination and SG disassembly, it would be more precise to report the specific E3 ligase identified in the cited studies rather than only the class of ligase. This would provide clearer mechanistic context and improve accuracy for readers.
  • The specific statement on line 182 "SG E3 ligases that depend on UBA1 activity are RBULs" should be supported by reference.
  • Are the text and figures clear and accurate?
  • In Supplementary Figure 1, DMSO is shown in green and the treatment in red, whereas in the main figures (Figure 1B and 1F) the colours in the legend are inverted. To avoid confusion, the colour coding in figure legends should be consistent across all figures throughout the manuscript.
  • At line 79, the manuscript states that "inhibition of ubiquitination delayed fluorescence recovery dynamics of G3BP1-mCherry, relative to HS-treated cells (Figure 1F, Supplementary Fig. 6A)." However, the data shown in Figure 1F appear to indicate the opposite effect: the TAK243-treated condition (green curve) shows a faster fluorescence recovery compared to the control (red curve). This discrepancy between the text and the figure should be corrected or clarified, as it may affect the interpretation of the role of ubiquitination in SG dynamics.
  • Line 86: adjust a missing bracket
  • There appears to be an error in the legend of Supplementary Figure 3: the legend states that the red condition (MKRN2) forms larger aggregates, but both the main Figure 3C of the confocal images and the text indicate that MKRN2 (red) forms smaller aggregates. Please correct the legend and any corresponding labels so they are consistent with the main figure and the text. The authors should also double-check that the figure panel order, color coding, and statistical annotations match the legend and the descriptions in the Results section to avoid reader confusion.
  • At lines 129-130, the manuscript states that "FRAP analysis demonstrated that MKRN2 KD resulted in a slight increase in SG liquidity (Fig. 3F, Supplementary Fig. 6B)." However, the data shown in Figure 3F appear to indicate the opposite trend: the MKRN2 KD condition (red curve) exhibits a faster fluorescence recovery compared to the control (green curve). This discrepancy between the text and the figure should be corrected or clarified, as it directly affects the interpretation of MKRN2's role in SG disassembly. Ensuring consistency between the written description and the plotted FRAP data is essential for accurate interpretation.
  • At lines 132-133, the manuscript states: "Then, to further test the impact of MKRN2 on SG dynamics, we overexpressed MKRN2-GFP and observed that it was recruited to SG (Fig. 3G)." This description should be corrected or clarified, as the over-expressed MKRN2-GFP also appears to localize to the nucleus.
  • At lines 134-135, the manuscript states that the FK2 antibody detects "free ubiquitin." This is incorrect. FK2 does not detect free ubiquitin; it recognizes only ubiquitin conjugates, including mono-ubiquitinated and poly-ubiquitinated proteins. The text should be corrected accordingly to avoid misinterpretation of the immunostaining data.
  • Figure 5A suffers from poor resolution, and no scale bar is provided, which limits interpretability. Additionally, the ROI selected for the green channel (DRIPs) appears to capture unspecific background staining, while the most obvious DRIP spots are localized in the nucleus. The authors should clarify this in the text, improve the image quality if possible, and ensure that the ROI accurately represents DRIP accumulation - in SGs rather than background signal.

Do you have suggestions that would help the authors improve the presentation of their data and conclusions?

• In the first paragraph following the APEX proteomics results, the authors present validation data exclusively for MKRN2, justifying this early focus by stating that MKRN2 is the most SG-depleted E3 ligase. However, in the subsequent paragraph they introduce the RBULs and present knockdown data for MKRN2 along with two additional E3 ligases identified in the screen, before once again emphasizing that MKRN2 is the most SG-depleted ligase and therefore the main focus of the study. For clarity and logical flow, the manuscript would benefit from reordering the narrative. Specifically, the authors should first present the validation data for all three selected E3 ligases, and only then justify the decision to focus on MKRN2 for in-depth characterization. In addition to the extent of its SG depletion, the authors may also consider providing biologically relevant reasons for prioritizing MKRN2 (e.g., domain architecture, known roles in stress responses, or prior evidence of ubiquitination-related functions). Reorganizing this section would improve readability and better guide the reader through the rationale for the study's focus.
• At lines 137-138, the manuscript states: "Together these data indicate that MKRN2 regulates the assembly dynamics of SGs by promoting their coalescence during HS and can increase SG ubiquitin content." While Figure 3G shows some co-localization of MKRN2 with ubiquitin, immunofluorescence alone is insufficient to claim an increase in SG ubiquitin content. This conclusion should be supported by orthogonal experiments, such as Western blotting, in vitro ubiquitination assays, or immunoprecipitation of SG components. Including a control under no-stress conditions would also help demonstrate that ubiquitination increases specifically in response to stress. The second part of the statement should therefore be rephrased to avoid overinterpretation, for example:"...and may be associated with increased ubiquitination within SGs, as suggested by co-localization, pending further validation by complementary assays."
• At line 157, the statement: "Therefore, we conclude that MKRN2 ubiquitinates a subset of DRiPs, avoiding their accumulation inside SGs" should be rephrased as a preliminary observation. While the data support a role for MKRN2 in SG disassembly and a reduction of DRIPs, direct ubiquitination of DRIPs by MKRN2 has not been demonstrated. A more cautious phrasing would better reflect the current evidence and avoid overinterpretation.

Significance

General assessment: provide a summary of the strengths and limitations of the study. What are the strongest and most important aspects? What aspects of the study should be improved or could be developed?

• This study provides a valuable advancement in understanding the role of ubiquitination in stress granule (SG) dynamics and the clearance of SGs formed under heat stress. A major strength is the demonstration of how E3 ligases identified through proteomic screening, particularly MKRN2, influence SG assembly and disassembly in a ubiquitination- and heat stress-dependent manner. The combination of proteomics, imaging, and functional assays provides a coherent mechanistic framework linking ubiquitination to SG homeostasis. Limitations of the study include the exclusive use of a single model system (U2OS cells), which may limit generalizability. Additionally, some observations-such as MKRN2-dependent ubiquitination within SGs and changes in DRIP accumulation under different conditions-would benefit from orthogonal validation experiments (e.g., Western blotting, immunoprecipitation, or in vitro assays) to confirm and strengthen these findings. Addressing these points would enhance the robustness and broader applicability of the conclusions.

Advance: compare the study to the closest related results in the literature or highlight results reported for the first time to your knowledge; does the study extend the knowledge in the field and in which way? Describe the nature of the advance and the resulting insights (for example: conceptual, technical, clinical, mechanistic, functional,...).

• The closest related result in literature is - Yang, Cuiwei et al. "Stress granule homeostasis is modulated by TRIM21-mediated ubiquitination of G3BP1 and autophagy-dependent elimination of stress granules." Autophagy vol. 19,7 (2023): 1934-1951. doi:10.1080/15548627.2022.2164427 - demonstrating that TRIM21, an E3 ubiquitin ligase, catalyzes K63-linked ubiquitination of G3BP1, a core SG nucleator, under oxidative stress. This ubiquitination by TRIM21 inhibits SG formation, likely by altering G3BP1's propensity for phase separation. In contrast, the MKRN2 study identifies a different E3 (MKRN2) that regulates SG dynamics under heat stress and appears to influence both assembly and disassembly. This expands the role of ubiquitin ligases in SG regulation beyond those previously studied (like TRIM21).
• Gwon and colleagues (Gwon Y, Maxwell BA, Kolaitis RM, Zhang P, Kim HJ, Taylor JP. Ubiquitination of G3BP1 mediates stress granule disassembly in a context-specific manner. Science. 2021;372(6549):eabf6548. doi:10.1126/science.abf6548) have shown that K63-linked ubiquitination of G3BP1 is required for SG disassembly after heat stress. This ubiquitinated G3BP1 recruits the segregase VCP/p97, which helps extract G3BP1 from SGs for disassembly. The MKRN2 paper builds on this by linking UBA1-dependent ubiquitination and MKRN2's activity to SG disassembly. Specifically, they show MKRN2 knockdown affects disassembly, and suggest MKRN2 helps prevent accumulation of defective ribosomal products (DRiPs) in SGs, adding a new layer to the ubiquitin-VCP model.
• Ubiquitination's impact is highly stress- and context-dependent (different chain types, ubiquitin linkages, and recruitment of E3s). The MKRN2 work conceptually strengthens this idea: by showing that MKRN2's engagement with SGs depends on active ubiquitination via UBA1, and by demonstrating functional consequences (SG dynamics + DRIP accumulation), the study highlights how cellular context (e.g., heat stress) can recruit specific ubiquitin ligases to SGs and modulate their behavior.
• There is a gap in the literature: very few (if any) studies explicitly combine the biology of DRIPs, stress granules, and E3 ligase mediated ubiquitination, especially in mammalian cells. There are relevant works about DRIP biology in stress granules, but those studies focus on chaperone-based quality control, not ubiquitin ligase-mediated ubiquitination of DRIPs. This study seems to be one of the first to make that connection in mammalian (or human-like) SG biology. A work on the plant DRIP-E3 ligase TaSAP5 (Zhang N, Yin Y, Liu X, et al. The E3 Ligase TaSAP5 Alters Drought Stress Responses by Promoting the Degradation of DRIP Proteins. Plant Physiol. 2017;175(4):1878-1892. doi:10.1104/pp.17.01319 ) shows that DRIPs can be directly ubiquitinated by E3s in other biological systems - which supports the plausibility of the MKRN2 mechanism, but it's not the same context.
• A very recent review (Yuan, Lin et al. "Stress granules: emerging players in neurodegenerative diseases." Translational neurodegeneration vol. 14,1 22. 12 May. 2025, doi:10.1186/s40035-025-00482-9) summarizes and reinforces the relationship among SGs and the pathogenesis of different neurodegenerative diseases (NDDs). By identifying MKRN2 as a new ubiquitin regulator in SGs, the current study could have relevance for neurodegeneration and proteotoxic diseases, providing a new candidate to explore in disease models.

Audience: describe the type of audience ("specialized", "broad", "basic research", "translational/clinical", etc...) that will be interested or influenced by this research; how will this research be used by others; will it be of interest beyond the specific field?

The audience for this paper is primarily specialized, including researchers in stress granule biology, ubiquitin signaling, protein quality control, ribosome biology, and cellular stress responses. The findings will also be of interest to scientists working on granulostasis, nascent protein surveillance, and proteostasis mechanisms. Beyond these specific fields, the study provides preliminary evidence linking ubiquitination to DRIP handling and SG dynamics, which may stimulate new research directions and collaborative efforts across complementary areas of cell biology and molecular biology.

Please define your field of expertise with a few keywords to help the authors contextualize your point of view. Indicate if there are any parts of the paper that you do not have sufficient expertise to evaluate.

I work in ubiquitin biology, focusing on ubiquitination signaling in physiological and disease contexts, with particular expertise in the identification of E3 ligases and their substrates across different cellular systems and in vivo models. I have less expertise in stress granule dynamics and DRiP biology, so my evaluation of those aspects is more limited and relies on interpretation of the data presented in the manuscript.

V

V

Je vais itérer ma question du chapitre précédent : pourquoi ? Pourquoi faire tout ça, alors qu'on a juste à aller dans notre document .html et écrire ce qu'on veut ? Pourquoi brûler non neurones avec des "${}, innerHTML …" ? J'ai l'impression qu'on veut se débarrasser du fichier HTML et de tout faire avec .js.

On entoure plutôt l'élément qu'on veut ajouter (ici div), non ?

Sur la question de sous-représentation des classes populaires en politique : https://democratiserlapolitique.org/