Author response:

eLife Assessment

This study provides fundamental insights by demonstrating that the Nanog mRNA coding sequence (CDS) and 3′UTR domains are spatially segregated and functionally distinct in pluripotent stem cells and blastocysts, with 3′UTR-enriched border cells primarily influencing morphogenesis and CDS-enriched inner cells largely regulating transcription and epigenetic programs. The work opens a novel conceptual avenue for understanding how separable mRNA domains can differentially control cell behavior and differentiation. However, the evidence is incomplete, as key aspects of the molecular nature, biogenesis, and precise characterization of the separated 3′UTR and CDS RNA species, as well as causal links between their perturbation and the observed phenotypes (e.g., via rescue and deeper characterization of 3′UTR elements), remain to be fully established.

We thank the editors and the three reviewers for their careful and constructive engagement with our manuscript. We greatly appreciate the reviewers’ recognition of the conceptual significance of the study and their thoughtful suggestions for strengthening the mechanistic and molecular characterization of the work. We have carefully considered all points raised and outline below the revisions planned for the revised manuscript.

The phenomenon of differential CDS and 3’UTR expression is not unique to Nanog. Independent 3’UTR and CDS expression and differential CDS/3’UTR usage has been observed across multiple genes, tissues, and developmental contexts, including genome-wide (Mercer et al., 2011) and transcriptome scale studies (Kocabas et al., 2025, Ji et al., 2021). Prior studies have proposed that isolated 3’UTRs may arise through regulated RNA processing pathways coupled to exonucleolytic degradation and, in some cases, recapping mechanisms (Malka et al, 2017, Haberman et al., 2024). While the precise molecular mechanisms underlying isolated Nanog CDS and 3’UTR generation remain unresolved, our observations (contained here) support regulated RNA processing models. Our original submission included a brief discussion of this topic; however the revised manuscript will include substantially expanded analyses and discussion of the generation of isolated Nanog CDS and 3’UTR species.

The revised manuscript will address the major concerns regarding:

(1) The molecular nature, biogenesis, and precise characterization of the separated 3′UTR and CDS mRNA species

(2) The causal relationship between perturbation of these RNA species and the observed phenotypes, including additional rescue experiments and deeper computational characterization of putative, functional 3′UTR elements.

Specifically:

(A) New supplementary analyses and schematics designed to further clarify the conceptual and mechanistic framework of the study, including:

(i) Computational examination of the Nanog 3’UTR across all reading frames for open reading frames (ORFs).

(ii) As suggested by Reviewers 1 and 3, single cell traces of Nanog mRNA expression from the full-length mESC dataset used in this study, illustrating distinct transcript isoforms and CDS/3’UTR expression patterns across individual cells, complementing the color-coded tSNE analyses currently presented in Fig. 2.

(iii) Expanded schematic model and analyses addressing possible mechanisms underlying the generation of isolated Nanog CDS and 3’UTR enriched RNA species, including transcript architecture, predicted RNA structural barriers, and exonucleolytic processing models.

(iv) Expanded discussion of the predominantly nuclear localization of the Nanog 3’UTR signal and its implications for transcript biogenesis, processing, and potential noncoding functions.

(B) Correction of all minor labeling errors.

(C) Additional experimental analyses, including:

- Expansion of Nanog 3’UTR overexpression and rescue experiments to include cell spreading assays.

- Expanded analysis of the effects of ROCK pathway inhibitors on colony morphology and cytoskeletal organization.

- Examination of the ability of ROCK inhibition to restore normal embryoid body formation.

Collectively, these planned revisions are intended to strengthen the mechanistic framing, molecular characterization, and broader significance of the study while clarifying the interpretation and scope of the conclusions.

Public Reviews:

Reviewer #1 (Public review):

Summary:

There is evidence that some genes encode mRNAs from which separate processed transcripts may arise, separating the coding sequence (CDS) from the 3'-UTR, and with both mRNA elements remaining stable in the cell. However, the functional consequences of these mRNA fragments have not been firmly established. In the manuscript by Yang et al., the authors probe the mRNA domain architecture of Nanog in the context of embryonic stem cell colonies and blastocysts. The authors detect spatial separation of Nanog CDS-containing mRNA from abundant Nanog 3'-UTR RNAs depending on the cell position in 2D embryonic stem cell colonies or in blastocysts.

Strengths:

The phenotypic analyses of the Nanog mRNA hold promise for revealing distinct roles for the Nanog encoded protein and a separate RNA encompassing the Nanog 3'-UTR.

Weaknesses:

There are a number of questions about the molecular nature of the mRNA species that the authors should address in order for the results to be firmly established, as noted below.

(1) It is not clear how the authors verified that their probes are specific for Nanog CDS or 3'-UTR regions. Especially for the 3'-UTR probe, it is confusing why colonies show green only regions, suggesting only the CDS is present. I would expect the CDS and 3'-UTR probes to colocalize in the interior cells. Is it possible that the 3'-UTR probe is targeting another RNA?

We thank the reviewer for raising the important question of probe specificity. We realize that the data that underlying this concern is the absence of colocalizing between CDS and 3’UTR probes in colony border cells.

The absence of CDS/3’UTR colocalization in colony border cells is not due to probe failure but instead reflects the principal observation underlying the study. If Nanog CDS and 3’UTR sequences were present exclusively as intact full-length transcripts in a strict stoichiometric ratio, Nanog positive cells would be expected to be positive for both probes (appearing yellow). Instead, border cells exhibit strong 3’UTR signal with minimal or absent CDS signal, while adjacent interior cells show the opposite pattern.

The fact that both probes robustly detect signal within the same sample but in spatially distinct cell populations, argues that both probes are functional and that the observed differential localization reflects genuine biological differences in levels of transcript components.

The CDS probe targets ~300 bp within the coding region, while the 3’UTR probe targets ~300 bp within the proximal region of the Nanog 3’UTR. Hybridization specificity was validated as described in the Methods and in our previous studies (Kocabas et al 2015; Ji et al 2021), including negative controls. We additionally now provide a supplemental figure (New Figure 1-figure supplement 2A), highlighting that the Nanog 3’UTR and CDS probes label cell populations distinct from each other, further indicating their specificity.

In addition, full-length scRNA seq datasets from both mouse and human ESCs demonstrate differential CDS/3’UTR expression patterns for Nanog and many other genes. To further clarify this point, the revised manuscript will include single cell transcript traces from mESCs illustrating the distinct Nanog isoforms detected across individual cells (New Figure 2-figure supplement 1A)

(2) It would help for the authors to include a graphic similar to Figure 3, Figure Supplement 1A, that diagrams the location of the CDS and 3'-UTR probes (this should also be done for Oct4 and Sox2). This graphic could also show all potential polyadenylation signals.

We agree that additional schematic clarification would improve readability. The revised manuscript will include schematics showing the locations of the CDS and 3’UTR probes for Nanog, Sox2 and Oct4 (New Fig. 1- figure supplement 1A).

(3) I think, based on the fluorescence patterns, there is evidence that the signal for the Nanog 3'-UTR probe is nuclear (images with DAPI staining), but this is not commented on that I could find. This should be discussed, as nuclear retention has implications for the noncoding function of the 3'-UTR fragment.

The reviewer is correct that the Nanog 3’UTR signal mostly nuclear. Whie this was noted in (the original) Figure 1-figure supplement 2A, we agree that it is possible that mechanistic and functional implications were not sufficiently discussed in the original manuscript. The revised manuscript will include expanded discussion of the relationship between nuclear localization transcript processing, and potential noncoding functions of isolated Nanog 3’UTR species

(4) Figure 2, Figure Supplement 1A needs a better explanation. It's not clear how the reads map to the different regions of the Nanog mature mRNA. The authors should show examples at different ratios of CDS to 3'-UTR. Do the reads have a sharp boundary at the junction of where the isolated 3'-UTR is thought to occur?

We thank the reviewer for this suggestion. The revised manuscript will include new single cell read maps across the Nanog locus from full length mESC scRNA-seq datasets (New Figure 2-figure supplement 1A), illustrating distinct CDS enriched and 3’UTR enriched transcript isoforms across individual cells.

These analyses indicate that some CDS dominant transcripts contain 3’UTR sequence, while many appear to contain little or no detectable 3’UTR sequence. Conversely, many 3’UTR enriched transcripts contain only minimal or truncated CDS sequence. Importantly full CDS and 3’UTR mRNA components are frequently not present in a strict 1:1 ratio, either within individual cells, or across cell populations.

The revised manuscript will also include expanded supplementary analyses integrating transcript architecture, predicted RNA structural barriers, polyadenylation analysis, and single cell coverage patterns to further examine possible mechanisms underlying the generation of isolated Nanog CDS and 3’UTR species (New Figure 2-figure supplement 1B,C).

(5) I looked in the Zenbu browser at human NANOG CAGE mapping in the FANTOM5 dataset. I could not see evidence for substantial capping of a 3'-UTR fragment when filtering for embryonic cell types. Given the strong signal for the 3'-UTR in border cells, I would expect to see evidence for capping if the RNA were indeed capped. This suggests that if it exists, it is likely uncapped and (as noted in point 3) is likely nuclear retained.

Prior studies have reported isolated uncapped and recapped 3’UTR species in multiple systems (Malka et al, 2017; Haberman et al, 2024). We agree that the predominantly nuclear localization and lack of a strong CAGE signal for Nanog are important observations and will expand discussion of these points in the revised manuscript.

(6) Are there predicted polyadenylation signals near the end of the CDS that would generate a short 3'-UTR, and are these signals conserved across mammals?

Computational analysis of the mouse Nanog 3'UTR identifies a single canonical PAS (AATAAA) at position 1074, located at the 3’ end of the annotated 3’UTR and this terminal PAS is conserved across mammals. These analyses will be included as a supplementary figure and discussed further in the revised manuscript section addressing Nanog transcript biogenesis.

(7) It would help to see a zoomed-in view of the region targeted by one of the guide RNAs in the 3'-UTR, and where that site is relative to the polyadenylation signal. Is the polyadenylation signal upstream, i.e., CDS proximal?

This will be provided in the revised manuscript (New Figure 2-figure supplement 1C,i) Two guide RNAs were used to generate the Nanog 3’UTR deletions. The downstream guide is upstream of the terminal polyadenylation signal at nt 1074 to preserve polyadenylation of the remaining Nanog CDS containing transcript.

Consistent with this, all Nanog 3’UTR knockout lines retain normal Nanog protein levels. The revised manuscript will include supplementary schematics showing guide RNA positions relative to the CDS, 3’UTR probes, and terminal PAS.

(8) A final note, the use of green and red together will be challenging for those who are colorblind. Providing a different false color palette would be helpful. 

We appreciate this attention to accessibly. The red/green color combination was chosen to provide the highest contrast between CDS and 3’UTR signals in the in situ hybridization experiments, which is important for visualizing their differential spatial localization. We will ensure that figure legends clearly indicate channel assignments throughout the manuscript.

I am refraining from comments on the cell biology and morphological insights, as they are remote from my core expertise.

Reviewer #2 (Public review):

Summary:

This manuscript shows that the coding sequence (CDS) and 3' untranslated region (3'UTR) of mRNA transcripts from the Nanog gene have distinct expression patterns and functions. In both human and mouse embryonic stem cells colonies and blastocysts, these domains are spatially segregated, with 3'UTR-enriched cells occupying the borders and CDS-enriched cells residing in the interior. CDS mRNA expression is correlated with the expected regulation of transcription and epigenetics associated with the Nanog protein. Interestingly, expression of the 3'UTR appears to play an independent role in cell behavior and colony morphogenesis. Indeed, deletion of the 3'UTR causes specific defects in cell spreading and protrusive activity, with alteration in the localization of adhesion and cytoskeleton-associated proteins. Remarkably, a large proportion of those defects are rescued upon ROCK inhibition. Deletion of either Nanog CDS or 3'UTR leads to distinct modifications in the differentiation competence.

Strengths:

The independent role of 3'UTR mRNA domains, although identified in neurosciences a couple of years ago, is a novel and exciting field relatively unexplored in early development.

The manuscript offers a multilayer series of experiments, in ES cells colony, blastocysts, and embryoid bodies, including imaging, -omics, genetic and pharmacological challenges, and differentiation experiments, thereby unveiling very convincingly the role of Nanog 3'UTR in morphogenesis.

Weaknesses:

The pathways leading to the generation of those distinct transcript domains are unknown. Although the functional differential roles are well demonstrated whether the expression patterns are a cause or a consequence of the cells' localization in the embryo remains to be explored.

We thank the reviewer for these thoughtful comments and for recognizing the potential significance of independent 3’UTR functions in early developmental systems.

Regarding the mechanisms underlying generation of distinct CDS and 3’UTR transcript domains, the revised manuscript will include new supplementary analyses and schematic models addressing possible Nanog transcript processing pathways, as outlined above.

We agree that the relation between spatial location and Nanog 3’UTR expression is an important question. Specifically, it remains unclear whether cells first acquire high Nanog 3’UTR expression and subsequently localize to the colony border or whether border position itself promotes high Nanog 3’UTR expression.

Our current data suggest that both processes may contribute. Deletion of the Nanog 3’UTR does not prevent colonies from establishing border/interior pattern, indicating that high Nanog 3’UTR is not strictly required for border pattern itself. At the same time, Nanog 3’UTR overexpression and rescue experiments increased the likelihood of border localization, suggesting that elevated Nanog 3’UTR expression promotes behaviors associated with border occupancy.

Reviewer #3 (Public review):

Summary:

In this manuscript, Yang et al reported distinct functions of the protein-coding sequence (CDS) and the 3' untranslated region (UTR) in the Nanog mRNA in pluripotent stem cells. They first observed different localization patterns for the CDS and 3' UTR in embryonic stem cells and in blastocyst embryos, and this pattern correlates with cell populations in different pluripotent states based on single-cell sequencing data. To characterize the potentially distinct functions of these regions, the authors generated knockout (KO) cell lines in which either the CDS or the 3' UTR was genetically ablated. These deletions led to different phenotypes in multiple assays. These results provided evidence that the CDS and 3' UTR of an mRNA could have distinct functions. Although these results are potentially interesting, several questions need to be addressed before the validity of their conclusion can be confirmed.

Strengths:

This study provides evidence for distinct functions of the protein-coding sequence and 3' untranslated region of an mRNA in pluripotent stem cells. The concept could be more broadly applied.

Weaknesses:

The initial observation (distinct localization of CDS and 3' UTRs) and the causal relationship between the KO and phenotype need further validation.

Major points:

(1) The authors showed distinct localization patterns of the CDS and 3' UTRs in human and mouse ESCs and blastocysts, and the overlap between their signals was minimal (Figure 1). Does this mean that the CDS and 3' UTR RNAs exist separately? For example, in cells that only showed signals for 3' UTRs, do these RNAs only contain 3' UTRs and lack CDS? Was this confirmed by RNA-seq experiments? If so, how are they generated (i.e., by transcription from a novel promoter or partial degradation of the full-length mRNAs)? This is a key question. Without a clear characterization of these RNAs, the rest of the study cannot be substantiated.

We thank the reviewer for raising this important question, which overlaps substantially with several key points raised by Reviewer #1 concerning the molecular nature and characterization of the Nanog CDS and 3’UTR species.

Colony border cells exhibit strong Nanog 3’UTR signal with minimal detectable CDS signal, while adjacent interior cells show the reciprocal pattern. These observations strongly suggest the existence of distinct Nanog transcript species rather than exclusively full-length transcripts containing stoichiometric amounts of both CDS and 3’UTR sequence.

This conclusion is independently supported by full-length Smart-seq2 scRNA seq datasets from both mouse and human ESCs, which provide transcript coverage across both CDS and 3’UTR regions.

(2) To confirm that the phenotypes of CDS or 3' UTR KO cells were caused by the deleted regions instead of other artifacts, rescue experiments should be performed.

Rescue experiments were included in the original submission (Fig. 4). The revised manuscript will expand these analyses to include cell spreading. We will also include additional ROCK pathway modulation experiments.

(3) As over-expression of the 3' UTR showed a phenotype, important regions within it should be identified, and also the possibility that the 3' UTR contains open reading frame(s) and is translated should be tested.

The revised manuscript will also include supplementary computational analyses of the Nanog 3’UTR, including open reading frame prediction, Kozak scoring, and evolutionary conservation analysis. (New Figure 2-figure supplement 1B). These analyses identify no evidence for strongly supported coding potential within the 3’UTR. Further, isolated Nanog 3’UTR transcripts are largely confined to the nucleus, making active translation unlikely.

The revised manuscript will include new supplementary analyses addressing Nanog transcript structure and possible biogenesis mechanisms (New Figure 2-figure supplement 1C).

References:

ViennaRNA/RNA fold – Lorenz et al 2011 Algorithms Mol Biol 6:26- RNA Secondary Structure stem loop, minimum free energy (MFE) prediction

NCBI BLASTP- Altschul et al (1990) J Mol Biol 215:403- ORF conservation, protein sequence similarity search

NCBI Entrez/Biohthon- Cock et al (2009) Bioinformatics 25:1422- sequence retrieval

PhastCons/UCSC multiz alignments- Siepel et al (2005) Genome Res 15:1034- evolutionary conservation scoring

UCSC Genome Browser- Kent et al. (2002) Genome Res 12:996-1006- conservation track access

Eaton et al (2020) Mol Cell 78:439- Stall model

Brannan et al (2012) Genes Dev 26:2621-Stall model

Addition to Methods.

ORFs (≥10 amino acids) were identified in all three forward frames according to Kozak (1987). Evolutionary conservation was assessed by BLASTP (Altschul et al., 1990) against RefSeq proteins. Poly(A) signals were identified by pattern matching for canonical and non-canonical hexamers. Conserved sequence blocks were obtained from UCSC PhastCons tracks (Siepel et al., 2005). RNA secondary structures were predicted using ViennaRNA RNAfold (Lorenz et al., 2011) with a sliding 80-nt window. The stall model for isolated transcript generation follows Eaton et al. (2020).

💻/thinkpad/🧊/me/📓/2026/5/2/0

indy.dweb.archive-search??q=hermenutic+circle+define@brave

eLife Assessment

There is a need for better and safer dengue virus live attenuated vaccines. This manuscript describes important findings that could lead to the design of a strongly immunogenic, tetravalent live attenuated vaccine for dengue, without the risk of causing antibody-dependent enhancement. However, the experimental evidence presented is incomplete since only constructions of one serotype were tested to prove the principle.

Reviewer #1 (Public review):

Summary:

Dalben et al. grafted the fusion loop mature (FLM) modification, based on a previously reported D2-FLM, to another serotype DENV4, and adapted them to replicate in Vero cells for live attenuated vaccine (LAV) manufacturing while retaining favorable antigenic profiles, generating two new strains: D2-vFLM and D4-vFLM. Deep sequencing revealed adapted mutations at the junction of envelope domains I and II (EDI and EDII), and both D2-vFLM and D4-vFLM showed no evidence of ADE in the presence of FL-targeting Abs. Sera from D2-vFLM immunized mice displayed strong homotypic and reduced heterotypic neutralization compared to wild-type viruses, with minimal to no ADE potential in vitro. Moreover, D2-vFLM immunization completely protected AG129 mice from lethal challenge with mouse-adapted D220. They demonstrate that the FLM modification platform is transferable across serotypes and yields strains with favorable immunogenicity and reduced ADE risk. The FLM approach provides a promising path toward the development of a safer tetravalent DENV LAV.

Strengths:

The authors carried out a series of experiments to generate and characterize two new strains (D2-vFLM and D4-vFLM) of FLM-modified viruses, and showed their antigenic and immunogenic profiles. The observation that the FLM modification platform is transferable across serotypes and yields strains with favorable immunogenicity and reduced ADE risk is interesting.

Weaknesses:

However, one concern is the total number of mutations (including originally introduced and compensatory mutations) in this FLM vaccine platform, and it is not clear regarding the future directions for the proof-of-concept vaccine in this study.

Reviewer #2 (Public review):

Summary:

In this manuscript, YR Dalben et al describe the generation of DENV2 and DENV4 strains with mutations in the fusion loop (FL) of the E protein and pre-membrane (prM) protein to limit potential antibody-dependent enhancement (ADE) resulting from vaccination with live-attenuated vaccines and adapted these strains for growth in Vero cells. They show that the DENV2 version D2-vFLM is immunogenic and generates neutralizing serum against DENV2 and DENV4 after 2 boosts and is protective against lethal challenge. Serum from D2-vFLM also showed no ADE against DENV4.

Strengths:

Overall, the paper is well written and presented, and the data presented support most of the conclusions made. Grafting D2-FLM mutations to DENV4 and adapting both to growth in Vero cells is a good step to show that this method could be used to generate production-level LAV. The growth and stability data are clear and well-conducted.

Weaknesses:

However, there are several weaknesses, mostly in regard to the immunogenicity data, that limit the overall impact. The FLM mutations were only grafted to DENV4 but not to the other Dengue serotypes. The authors acknowledge that this is a proof-of-concept, but generating mutants of the other serotypes would strengthen the idea that this could be used to develop a tetravalent LAV. Immunizations in mice were only performed for D2-vFLM but not D4-vFLM. Immunogenicity data for D4-vFLM would strengthen this work if it shows that it can be immunogenic, protective, and limit ADE, as is shown for D2-vFLM. ADE from D2-vFLM was only tested against DENV4; does it also limit ADE from the other serotypes? This would better show that these mutations do limit ADE across serotypes and not just a single one.

Additionally, some of the immunization data likely need to be repeated:

The authors should describe why they pooled the sera from the mice and whether they purified total IgG or not (Figure 5). They should also probably repeat the challenge experiment since it was 4 mice (D2) against 5 (D2-vFLM), and it is unclear if there is a statistical difference between the results obtained. It is not even mentioned in the Results section (D2 result vs D2-FLM), and thus unclear if using D2-FLM is an improvement in the way the data is currently presented.

Author response:

Public Reviews:

Reviewer #1 (Public review):

Summary:

Dalben et al. grafted the fusion loop mature (FLM) modification, based on a previously reported D2-FLM, to another serotype DENV4, and adapted them to replicate in Vero cells for live attenuated vaccine (LAV) manufacturing while retaining favorable antigenic profiles, generating two new strains: D2-vFLM and D4-vFLM. Deep sequencing revealed adapted mutations at the junction of envelope domains I and II (EDI and EDII), and both D2-vFLM and D4-vFLM showed no evidence of ADE in the presence of FL-targeting Abs. Sera from D2-vFLM immunized mice displayed strong homotypic and reduced heterotypic neutralization compared to wild-type viruses, with minimal to no ADE potential in vitro. Moreover, D2-vFLM immunization completely protected AG129 mice from lethal challenge with mouse-adapted D220. They demonstrate that the FLM modification platform is transferable across serotypes and yields strains with favorable immunogenicity and reduced ADE risk. The FLM approach provides a promising path toward the development of a safer tetravalent DENV LAV.

Strengths:

The authors carried out a series of experiments to generate and characterize two new strains (D2-vFLM and D4-vFLM) of FLM-modified viruses, and showed their antigenic and immunogenic profiles. The observation that the FLM modification platform is transferable across serotypes and yields strains with favorable immunogenicity and reduced ADE risk is interesting.

We thank reviewer 1 for the encouraging comments for our work.

Weaknesses:

However, one concern is the total number of mutations (including originally introduced and compensatory mutations) in this FLM vaccine platform, and it is not clear regarding the future directions for the proof-of-concept vaccine in this study.

Author response table 1.

We summarize the mutations in the FLM platform below.

The maturation mutations are located at the furin cleavage site, which is buried within the membrane or virion. As a result, only five mutations are surface exposed, two of which are in the fusion loop region targeted for removal. Therefore, for a proof-of-concept study, the total number of mutations remains well within the genetic diversity observed among DENV genotypes.

Compensatory mutations may affect overall DENV antigenicity. Notably, one such mutation, K204R, has been reported to alter antigenicity and could contribute to the improved safety profile of the vaccine. However, we have also shown that multiple adaptive pathways can support Vero cell adaptation, and our data indicate that K204R is not absolutely required for this process.

Reviewer #2 (Public review):

Summary:

In this manuscript, YR Dalben et al describe the generation of DENV2 and DENV4 strains with mutations in the fusion loop (FL) of the E protein and pre-membrane (prM) protein to limit potential antibody-dependent enhancement (ADE) resulting from vaccination with live-attenuated vaccines and adapted these strains for growth in Vero cells. They show that the DENV2 version D2-vFLM is immunogenic and generates neutralizing serum against DENV2 and DENV4 after 2 boosts and is protective against lethal challenge. Serum from D2-vFLM also showed no ADE against DENV4.

Strengths:

Overall, the paper is well written and presented, and the data presented support most of the conclusions made. Grafting D2-FLM mutations to DENV4 and adapting both to growth in Vero cells is a good step to show that this method could be used to generate production-level LAV. The growth and stability data are clear and well-conducted.

We thank reviewer 2 for the encouraging comments for our work.

Weaknesses:

However, there are several weaknesses, mostly in regard to the immunogenicity data, that limit the overall impact. The FLM mutations were only grafted to DENV4 but not to the other Dengue serotypes. The authors acknowledge that this is a proof-of-concept, but generating mutants of the other serotypes would strengthen the idea that this could be used to develop a tetravalent LAV.

We selected DENV2 and DENV4 because they are the most genetically divergent. Currently, our data should support the FLM mutations that can be grafted on both DENV2 and DENV4, likely extend to their corresponding genotypes. We agree that the FLM mutations should be evaluated in additional serotypes. We also have promising preliminary data for FLM mutation grafting in DENV1 and are currently applying the same approach to DENV3. We hope to include these results, whether positive or negative, in the revised manuscript.

Immunizations in mice were only performed for D2-vFLM but not D4-vFLM. Immunogenicity data for D4-vFLM would strengthen this work if it shows that it can be immunogenic, protective, and limit ADE, as is shown for D2-vFLM.

We are currently immunizing AG129 mice with DV4 and D4-vFLM, followed by heterotypic challenge with D220. Because DENV vaccine-related hospitalization in clinical trials typically occurs 3 - 4 years after vaccination, we are cautious about whether this experimental design will fully capture the added safety benefit of the FLM mutations. We are also developing a passive immunization model in AG129 mice using diluted DENV4 serum to better mimic long-term waning antibody titers. We will include the future findings in the revised manuscript.

ADE from D2-vFLM was only tested against DENV4; does it also limit ADE from the other serotypes? This would better show that these mutations do limit ADE across serotypes and not just a single one.

We are trying to keep the scope of the paper within DENV2 and DENV4, however, we will perform ADE and neutralization assays for all four serotypes in the revised manuscript.

Additionally, some of the immunization data likely need to be repeated:

The authors should describe why they pooled the sera from the mice and whether they purified total IgG or not (Figure 5).

We used pooled serum, consisting of equal volumes from each mouse, rather than purified IgG. In Figure 5, our goal was to show the overall increase in serum titer after each immunization using cheek-bleed samples from individual animals. Because the available sample volume was limited, we pooled the sera for this analysis. We also measured end-point serum titers for each individual animal.

They should also probably repeat the challenge experiment since it was 4 mice (D2) against 5 (D2-vFLM), and it is unclear if there is a statistical difference between the results obtained. It is not even mentioned in the Results section (D2 result vs D2-FLM), and thus unclear if using D2-FLM is an improvement in the way the data is currently presented.

This experiment was designed to determine whether D2-vFLM protects AG129 mice against homotypic challenge as effectively as DV2-WT. Although the sample size was small, the results support our conclusion. However, we agree with the reviewer that the study should include more animals, and we will increase the group size to n > 8 to 10 in the revised experiment.

transportation costs and trade barriers complicates

They enable a firm to reduce the costs of value creation and/or to create per- ceived value in such a way that premium pricing is possible (e.g., many believe that Apple uses premium pricing for its line of iPhones).

different activities in the countries where they can be done most efficiently or cheaply.

If these create a favorable impression of the firm’s product in the minds of consumers, they increase the price that can be charged for the firm’s product.

In a competitive market, firms must lower prices to attract customers. If the firm were a monopoly, it could charge a much higher price.

A monopoly supplier is a firm that is the only seller in the market. There are no competitors offering similar products.

Consumer surplus is the extra value customers receive when they pay less than what they were actually willing to pay.

I believe this is the non-preprinted version of this article --> 10.1038/s41593-022-01152-z

你们知道可以comment吗？

indy.dweb.archive=Táltos@wikipedia

eLife Assessment

The authors propose a "simplified" model for intrinsically bursting neurons with explicitly controllable parameterization of oscillatory dynamics. The evidence that the modeling approach is generally appropriate and practical for modeling rhythmic bursting neurons and neural circuits is currently incomplete. Based on what the authors present, this model appears to have limited neurobiological relevance and utility but may be useful as a controller for an artificial system, such as in neuro-robotics applications.

Reviewer #1 (Public review):

Summary:

The authors present a simplified neural bursting model with explicitly controllable parameterization of oscillator dynamics designed for neural circuit modeling involved in rhythm generation.

Strengths:

(1) The purpose of the model and applied abstractions are well articulated and justified (2D model, independent parameter control).

(2) Explicit control of burst duration, inter-burst interval, amplitude, resetting-behavior/entrainment. This allows modelers to focus on circuit interactions and is especially useful when details of intrinsic currents and bursting mechanisms are unknown. One could even imagine a scenario where this model would help identify predictions on key underlying burst generation mechanisms.

(3) The model is well described and validated with simulations and comparisons to the base model and one alternative model.

(4) Circuit-level validation is convincing, as it reproduces not only trivial examples.

(5) The underlying mechanism in phase space is well reasoned and justified, extends previous work, e.g., by McKean, by improving usability.

Weaknesses:

(1) The paper heavily relies on numerical demonstrations but does not provide a formal analysis of stability, bifurcations, or entrainment. While appropriate for the intended purposes, a more formal footing could strengthen the model.

(2) Lots of nice demonstrations are shown, but it is less clear how model parameterization was chosen, how behavior depends on parameterization, and in what parameter ranges certain behavior can be expected. A more detailed description of parameterization/exploration of parameter space would greatly benefit anyone using this model in the future.

(3) Some claims on reproduction of prior locomotor CPG model and production of "more biologically realistic activity" by the presented model are overstated. The key feature of the locomotor CPG models cited was that they not only reproduced speed-dependent gait expression of intact mice, but also changes of gait expression after silencing/removal of specific commissural and long propriospinal interneurons (e.g., selective loss of trot after deleting of V0V; changes in gait expression and step-to-step variability after silencing of descending long-propriospinal neurons or ascending V3 LPNs). While likely (at least partially) feasible with the model formulation, the correspondence of these silencing/ablation of neuron classes has not been shown by the model. Importantly, though, it appears that authors didn't show how the model in general behaves under the influence of noise, which is key to reproducing LPN silencing.

Reviewer #2 (Public review):

Summary:

The authors propose a reduced model for intrinsically bursting neurons. The model simply consists of exponential decay of an adaptation variable in a phenomenological silent phase, an exponential growth of that variable in an active phase, and imposed thresholds for jumps between these phases, with some add-ons to allow for effects such as input-dependence.

Strengths:

The model could be used as a controller for an artificial system that needs to switch between on and off states with separate control of state durations. It has some flexibility to allow for variable levels of the activity variable during the active phase. The authors show that the model can be tuned to capture phase response properties of neurons and patterns generated by small networks of neurons.

Weaknesses:

The proposed approach lacks biological relevance, practicality, and originality.

(1) Biological relevance:

Central pattern generators and other bursting neurons use specific physical principles to generate their bursts of activity. These principles place constraints on the tuning of these bursts, including relationships between active and silent phase durations and other properties. By discarding these relationships, the proposed model risks losing key constraints that affect performance in biologically relevant scenarios. The proposed model does not allow for the emergence of interesting dynamical phenomena, which occur naturally in neurons and neuronal networks.

It is also important to note that spikes within bursts can be important and of interest. Biophysical models allow for easy extension to include spikes via fast sodium and potassium currents. The proposed model does not allow for such extensibility.

Finally, as shown in the seminal early-2000s work of Izhikevich, building on fast-slow decomposition work by Rinzel and others, there is a wide variety of possible neuronal bursting patterns. At the very least, several of these have been observed in neuronal recordings. The authors' model is specific to square-wave bursting.

(2) Practicality:

The model makes use of various cut-off functions and other aspects that are implemented as rules. Combining rules with differential equations makes for an awkward modeling framework that is inconvenient to implement, conceptualize, and analyze (e.g., from a bifurcation perspective). Moreover, the authors add more and more adjustments to their basic framework to capture additional features, but these add-ons simply make the model more, and unnecessarily, complicated and awkward. It's worth noting that the authors argue for their model based on the idea that more biophysical models are difficult to tune, yet they compare their model to a biophysical one that they were able to tune to achieve the various patterns that they study. They do not give any indication of how easy or hard it was to tune their own model, nor do they compare simulation times between the two models. I do note that the biophysical model seems to have 22 parameters, whereas the simplified one has 21 in Table 2, which is essentially the same number. Finally, although the authors give some extensions of the model to match observed data, their model does not seem useful for predicting performance in never-before-tested scenarios.

(3) Originality:

As the authors note, the use of low-dimensional, specifically planar, neural models dates back to early authors such as FitzHugh and Nagumo. What the authors fail to acknowledge is that Rinzel, Terman, Kopell, and others did seminal work on neuronal activity, including phenomena such as post-inhibitory rebound and fast threshold modulation, using a relaxation oscillation framework, starting several decades ago. Their work included applications to central pattern generators (e.g., see Terman and collaborators on respiratory CPGs). It is astonishing that the authors don't seem to be aware of this work and do not mention it at all. Moreover, I don't see any advantage of the proposed framework over the earlier relaxation oscillator setting, where many important mechanistic principles have already been analyzed, including extensions to networks. On a related note, even through they propose a piecewise linear model, the authors do not cite the substantial existing work on piecewise linear models (e.g., Hahnloser, Neural Networks, 1998, for an early example; 2024 SIAM Review article by Coombes et al and references therein for much more) including work specifically on bursting, nor do they cite various other previous efforts to capture bursting with simplified models including work on piecewise linear maps by Aguirre et al.

Reviewer #3 (Public review):

This computational modeling study introduces the methodology of replacing bursting neurons in a model circuit with a simplified piecewise-linear model with an "active" and a "quiet" state representing, respectively, the burst of spikes and the inter-burst interval. The shape of the active state loosely represents the intra-burst firing rate. Because (piecewise) linear systems are explicitly solvable, the transitions from quiet to active and vice versa can be calculated explicitly to match exactly what a biophysically realistic model or a biological neuron does in different conditions. The base piecewise-linear model is built to represent a 2D biophysical neuron with a cubic v-nullcline. The simplicity of the model allows for matching the kinetics of more complex models with a tractable simplified set of equations, as exemplified by approximations of burst duration and amplitude, phase-response curves, entrainment, and, finally, mimicking the activities of two CPG circuit models using this simplified representation.

Major comments

(1) The use of piecewise linear approximations to explicitly estimate properties of biophysical neurons is a well-known and common technique. This study adds nothing to the technique in terms of novelty.

(2) Although the model explicitly matches active and inactive durations of a circuit neuron, the dynamics are explicitly "clamped" by the user because the reduced model parameters explicitly depend on the input. There are cases where this is useful, for example, when we are interested in the dynamics of _other_ neurons (B, C, D, ...) within the context of activity, and we "clamp" the dynamics of neuron A. One should note that this is no better than having a look-up table. Effectively, to give a comparison, it is like using a sine wave to represent a pacemaker neuron and explicitly define its frequency at different input levels so that it responds "dynamically". However, the neuron is restricted to what the user puts in, and therefore, calling it a dynamical system is entirely wrong. I am afraid that the use of this crude tool is not described well enough in the manuscript to warn a naïve user not to fall for this trap.

(3) The phase resetting curves are used incorrectly. PRCs are useful when the perturbation is weak (soft), which would demonstrate the nature of the vector field near the limit cycle and therefore inform us of the nature of its stability or instability. A hard PRC would always reset the cycle to the fixed offset from the perturbation phase and is therefore uninformative in understanding dynamics. (It is, however, useful experimentally in identifying which neurons are part of the CPG.) The authors clearly know that the dynamics of the system away from the limit cycle do not conserve those of a biophysical neuron. So what is the point?

(4) I work on the STG, one of the systems exemplified here. Even in the small and relatively regular CPGs of the STG, the definition of the active and quiet parts of a burst is often less clear than what the authors suggest. Bursting neurons often do multiple bursts in a cycle, and therefore, substituting the burst envelope is a subjective matter. This is even more problematic in bursting neurons in the brain, where there is often no quiet period. This should be discussed.

Author response:

We thank the editors and reviewers for their time and feedback. We are encouraged by the feedback that the purpose and abstractions of the model are well articulated and justified, that the explicit control of bursting characteristics is useful, and that the circuit-level validations are convincing.

Before responding to individual reviewer comments, we would like to address the framing in the current assessment that the model "appears to have limited neurobiological relevance and utility but may be useful as a controller for an artificial system, such as in neuro-robotics applications." We respectfully suggest that this framing understates the model's relevance to neuroscience. Specifically, a growing body of literature aims to understand biological motor control by building embodied simulations. Yet, these simulations either use overly simple artificial neural network (ANN) units without dynamics or computationally intensive biophysical ones that are difficult to train. Our model is not intended as a biophysical account of how individual neurons generate bursts at the level of ionic mechanisms or spikes that goal is already well served by the conductance-based and reduced biophysical models we cite. Rather, its contribution is to make intrinsic bursting dynamics readily incorporable into neural circuit models that can be used in complex settings, with parameters that map directly onto quantities that circuit-level neuroscience most often measures and tunes in models (burst duration, duty cycle, amplitude, shape, input dependence). Indeed, Reviewer #1 notes that: "The purpose of the model and applied abstractions are well articulated and justified [...] This allows modelers to focus on circuit interactions and is especially useful when details of intrinsic currents and bursting mechanisms are unknown. One could even imagine a scenario where this model would help identify predictions on key underlying burst generation mechanisms."

We see our work as a neuroscience contribution as much as a neuro-robotics one. Bringing tractable, controllable bursting into this regime allows circuit modelers to study how intrinsic bursting interacts with circuit connectivity without committing to specific biophysical mechanisms, and it lets ANN-style models incorporate a class of dynamics that is biologically pervasive but currently underrepresented. We validated the model against two well-studied biological CPGs (the crustacean pyloric circuit and the mammalian locomotor circuit) precisely because the target use case is biological circuit modeling.

While we remain committed to the belief that bringing bio-inspired neurons with interpretable intrinsic dynamics into ANN-style modeling of biological control systems is a useful contribution as an eLife Methods paper, the reviews have made clear that we have not situated our work clearly enough within the literature. In revision, we will sharpen this positioning in the Introduction and Discussion, and better situate the model relative to both the long tradition of non-spiking relaxation-oscillator and piecewise-linear modeling in neuroscience and also to current trends in simulated control.

Public Reviews:

Reviewer #1 (Public review):

(1) Formal analysis

The paper heavily relies on numerical demonstrations but does not provide a formal analysis of stability, bifurcations, or entrainment. While appropriate for the intended purposes, a more formal footing could strengthen the model.

We agree that a formal dynamical-systems treatment would deepen the work, and we appreciate the reviewer's acknowledgment that the numerical-only approach may nevertheless be appropriate for the intended purposes. Because the model is hybrid (continuous dynamics combined with discrete switching rules), a full formal analysis is non-trivial, and we view it as a substantial follow-up rather than something to fold into the present manuscript. In revision, we will discuss more explicitly the opportunities such formal analysis presents.

(2) Parameter tuning and parameter-space characterization

It is less clear how model parameterization was chosen, how behavior depends on parameterization, and in what parameter ranges certain behavior can be expected.

We agree that this would substantially improve usability, and we will expand this aspect of the paper. The revision will include: (a) more details describing how parameters maps onto observable features of the bursting waveform, (b) recommended parameter ranges and the qualitative behaviors expected at their boundaries, and (c) practical guidance for tuning the model to match observations or embed into circuits.

(3) Locomotor CPG interneuron ablation and noise

The correspondence of these silencing/ablation of neuron classes has not been shown by the model. Importantly, though, it appears that authors didn't show how the model in general behaves under the influence of noise.

The reviewer is right that the cited work establishes validity of the circuit model in large part through silencing/ablation experiments, and we did not reproduce those experiments. We understand those gait expression phenomena to be arising from non-bursting interneuron activations and a robust solution found for connection weights between them. The half-center bursting neurons only see a time-varying input signal, and their response is well-characterized by the constant, pulse, and periodic analyses we perform. As such, we chose to reproduce a few key experiments to retain a focus on our simplified neuron model. We will rephrase the relevant passages to make this scope explicit and ensure that our reproduction claims are appropriately stated. We will also expand on how the model interfaces with noise together with the proposed parameter-space characterization.

Reviewer #2 (Public review):

(1) Biological relevance

Central pattern generators and other bursting neurons use specific physical principles to generate their bursts of activity. These principles place constraints on the tuning of these bursts, including relationships between active and silent phase durations and other properties. By discarding these relationships, the proposed model risks losing key constraints that affect performance in biologically relevant scenarios.

We agree that biophysical models impose constraints that arise from underlying mechanisms. For instance, as input alters the curved shape of nullcline-v in Figure 1, the active/quite phase durations and duty cycle change in constrained ways. The question seems to be if our model is too flexible for instance, making it too easy to achieve desired phase durations, duty cycles, and other input-dependent responses. We see this as a valuable feature of our model, not a bug. Firstly, even if our model may be expressive enough to achieve a variety of response profiles (as in Figure 3—figure supplement 3), the careful modeler will ensure matching to experimental observations. Moreover, in many circuit systems, the relevant biophysical details are often unknown for the specific neurons being modeled as noted by Reviewer #1, and the modelers' primary goal is to reproduce circuit-level activity. Such can be achieved easily with a simplified model, and also with a biophysical model as data becomes available. Finally, we should note that modelers can and do tune the parameters of biophysical models within determined ranges in order to achieve desired phase durations and duty cycles, relaxing constraints somewhat in order to reproduce appropriate activity.

It is also important to note that spikes within bursts can be important and of interest. [...] The authors' model is specific to square-wave bursting.

We agree that spikes are important and interesting in many settings, and we believe that biophysical models would be most appropriate in these cases. In many cases, too, some abstraction and simplification is desirable, and this would not necessarily detract from the model's biological relevance. As we discuss in our high-level comments, we aim to bring intrinsic bursting dynamics into the ANN-style modeling regime that typically neglects intrinsic dynamics altogether. While the simplified model may be limited in some ways, it is nevertheless useful for many common biologically relevant scenarios, as validated by our circuit experiments. Finally, we would note that many of the raised limitations (no intra-burst spike structure, restricted bursting class, abstracted constraints) are shared by the relaxation-oscillator and piecewise-linear traditions that the reviewer cites approvingly, which suggests that our model lies along a familiar abstraction continuum rather than outside it. In revision, we will explicitly acknowledge that the model captures a basic/regular form of bursting within a broader taxonomy, and clarify the conditions under which abstracting the biophysical constraints is appropriate.

(2) Practicality

The model makes use of various cut-off functions and other aspects that are implemented as rules. Combining rules with differential equations makes for an awkward modeling framework

On the modeling framework, we would defend the hybrid formulation (rules + ODE) as our aim is to prioritize usability by modelers, not the simplicity or elegance of equations. While a "pure-ODE" Fitzhugh-Nagumo-style polynomial may seem simple and elegant—with dv/dt = av^3 + bv^2 + cv + d and a, b, c, d parameters as the reviewer has pointed out a lot of complexity can arise from this. Tuning these parameters is far from intuitive, as small changes can produce nonlinear effects and qualitative shifts in behavior. Achieving the right phase durations, input-dependent scaling, waveform amplitude and shape, phase delays, and other characteristics simultaneously to match experimental data is quite cumbersome in the elegant models, not to mention the biophysical models. In contrast, these characteristics are easy to control in our model, because we translate complex dynamical behavior from implicit to explicit and surface a set of interpretable and tunable parameters.

The authors argue for their model based on the idea that more biophysical models are difficult to tune, yet they compare their model to a biophysical one that they were able to tune to achieve the various patterns that they study. They do not give any indication of how easy or hard it was to tune their own model [...] The biophysical model seems to have 22 parameters, whereas the simplified one has 21 in Table 2, which is essentially the same number.

To clarify, we did not tune the biophysical model, but rather copied its parameters from the cited work. We will make this more explicit in the relevant Methods section.

We could not simply specify or tune these parameters because they have complex biological priors that must be derived from experimental data for example, the membrane capacitance (20 pF), ionic conductance and reversal potentials (4.5 nS, -62.5 mV), and many gating kinetics parameters (slopes, midpoints, time constants for sigmoid/bell curves).

It is often the case that such parameters must be estimated in specific preparations then reused and refined over many years. For instance, the biophysical model we compare to borrowed parameters from (Kim et al. 2022), which retuned time constants relative to (Danner et al. 2017), which altered NaP conductance from (Danner et al. 2016), which retuned duty cycles from (Molkov et al. 2015), which adapted from respiratory networks of (Rubin et al. 2008), which used gating kinetics parameters from (Butera et al. 1999). Similarly, the crustacean pyloric circuit model we compare to is from (Alonso and Marder 2020), which augmented the circuit and parameters of (Prinz et al. 2004), which sampled from a database of procedurally generated parameters from (Prinz et al. 2003), which developed parameter priors from the lobster STG experimental work of (Turrigiano et al. 1995). These brief descriptions of the multi-decade lineage of parameter sets omit the substantial parallel and preceding work related their development, but they suffice to demonstrate the incredible science and effort that goes into building biophysical models for particular circuits. Such data is often unavailable and such detail is often undesirable for different research goals, in which case our simplified model is a valuable and practical tool.

The key parameters of our simplified model are observable quantities like active/quiet durations (in seconds), input-dependent duration scaling (as a fraction of intrinsic durations), input strength that induces tonic firing, etc. As such, tuning the bursting neuron parameters for circuit models was easy, with manual tuning from scratch taking less than 1 day. As Table 3 shows, the resulting parameters are often simple, elegant numbers and can be derived directly from observations. For instance, the pyloric PD active and quiet durations (200 ms and 800 ms, respectively) are set using the exact target values that (Alonso and Marder 2020) encode in their objective for a genetic algorithm to tune their model’s biophysical parameters (or rather, a subset of them for tractability).

Thus, the 22-vs-21 comparison is not very informative, because the parameters are not comparable in kind. However, to make it easier to tune our model, we will revise the manuscript to include: (a) more details describing how parameters maps onto observable features of the bursting waveform, (b) recommended parameter ranges and the qualitative behaviors expected at their boundaries, and (c) practical guidance for tuning the model to match observations or embed into circuits.

(3) Originality

What the authors fail to acknowledge is that Rinzel, Terman, Kopell, and others did seminal work on neuronal activity [...] The authors do not cite the substantial existing work on piecewise linear models [...] I don't see any advantage of the proposed framework over the earlier relaxation oscillator setting, where many important mechanistic principles have already been analyzed, including extensions to networks.

We thank the reviewer for these pointers and apologize for the gap in our literature coverage. While we had cited McKean, FitzHugh-Nagumo, Izhikevich, et al. as representative examples of different model classes, we agree that the broader relaxation-oscillator and piecewise-linear traditions deserve more comprehensive treatment including Rinzel, Terman, Kopell, et al. on relaxation-oscillators; and Hahnloser, Coombes, Aguirre, et al. on piecewise-linear models. We will expand the related work discussion and clarify how our contribution is novel and valuable.

To be clear, we do not claim to be the first to use piecewise-linear models for neurons. Our intended contribution is the specific construction a rectangular limit cycle whose horizontal/vertical decoupling permits a closed-form mapping from interpretable parameters to burst features and the demonstration that this construction integrates cleanly into firing-rate circuit models of biological CPGs, which we believe will provide realism for more complex models with learned components.

Moreover, in contrast to many other relaxation-oscillator models including the elegant Fitzhugh-Nagumo-style model we discussed above, our model is not aimed at establishing mechanistic principles or being simple enough to analyze formally. It is a practical tool that affords precise control of many bursting characteristics, which is important for closer alignment between firing-rate circuit models and biological activity. We will state this contribution more precisely in the revision so it is not conflated with a broader novelty claim.

Reviewer #3 (Public review):

(1) Novelty of piecewise-linear approximation

The use of piecewise linear approximations to explicitly estimate properties of biophysical neurons is a well-known and common technique. This study adds nothing to the technique in terms of novelty.

We agree that piecewise-linear approximations of neurons are not themselves novel, and we have not intended to claim otherwise: We cite the McKean model as a direct predecessor and, prompted by Reviewer #2, we will substantially expand citations to the relaxation-oscillator and piecewise-linear traditions (Rinzel, Terman, Kopell, Hahnloser, Coombes, Aguirre, et al.). Our intended contribution is not the use of piecewise-linear pieces per se but the specific construction: a rectangular limit cycle whose horizontal/vertical decoupling permits a closed-form, interpretable mapping from burst features (duration, duty cycle, amplitude, shape, input dependence) to dynamics, and clean integration into firing-rate circuit models of biological CPGs. We will revise the relevant passages so this contribution and the boundaries of our novelty claim are stated precisely.

(2) Dynamical system mechanism

This is no better than having a look-up table [...] The neuron is restricted to what the user puts in, and therefore, calling it a dynamical system is entirely wrong.

We would like to take the opportunity to clarify this point, because the model's behavior is much richer than the lookup-table characterization suggests. The model is closed-loop: trajectories evolve through coupled state variables whose response to time-varying input depends on current state, not on a precomputed table of input-to-output values.

Specifically:

(a) The input represents the net time-varying synaptic drive, not a clamped voltage level;

(b) The adaptation and voltage variables evolve according to coupled differential equations both on and off the limit cycle;

(c) The duration and scale parameters only constrain active/quiet durations at input endpoints (-1, 0, +1), while the response at intermediate inputs is determined by the dynamics and other parameters such as the adaptation time constant, which can qualitatively reshape the constant-input response curve (Figure 3—supplement figure 3);

(d) The response to a transient input depends on the current state for example, excitatory pulses early in the active phase have little effect, as in the biophysical model.

This is a direct result of the simplified model using a similar limit cycle and nullcline structure as the biophysical model’s dynamical system (Figure 1).

(3) PRC usage

The phase resetting curves are used incorrectly. PRCs are useful when the perturbation is weak (soft) [...] A hard PRC would always reset the cycle to the fixed offset from the perturbation phase and is therefore uninformative in understanding dynamics.

We appreciate this point and would like to clarify what we show and why. We present finite (non-infinitesimal) PRCs across a range of input strengths and signs, spanning both the "soft" (weak-perturbation) regime as well as the "hard" (strong-perturbation) regime, rather than focusing on the "hard" regime alone. Importantly, even in the strong-perturbation regime we do not see that pulses "always reset the cycle to the fixed offset from the perturbation phase". In Figure 4, we see that the active phase exhibits a non-resetting region whose size and location depend on parameters. This region governs entrainability and phase-locking offset, and is thus a key aspect of the neuron's dynamics. Moreover, the strong-perturbation regime is also biologically relevant in our circuit examples. For instance, the inhibitory connections within the pyloric CPG are strong enough to cause hard resets, and these resets shape the circuit-level dynamics we reproduce. We will revise the pulse-input section to state these points more explicitly so the rationale is clear for showing PRCs across a range of inputs.

(4) Defining active/quiet phases

The definition of the active and quiet parts of a burst is often less clear than what the authors suggest. Bursting neurons often do multiple bursts in a cycle, and therefore, substituting the burst envelope is a subjective matter. This is even more problematic in bursting neurons in the brain, where there is often no quiet period.

We agree that waveform envelope can be subjective in some preparations, and we can add this caveat to the discussion.

On neurons with no quiet period, we note that this behavior is in fact already supported in our model, as seen in Figure 3: under strong excitatory input, both the biophysical and simplified models enter a regime in which firing rate never reaches zero. As the model can generally be viewed as an abstract limit cycle that maps onto periodic waveforms through the firing function, the quiet phase need not correspond to literal silence.

On more complex waveforms, we could imagine different firing functions that produce richer burst shapes including multi-peak bursts, but we have not tried this explicitly. Of course, for research questions concerned with irregular bursting or spike-to-burst transitions, a lower-level biophysical model would be more appropriate. In revision, we will expand on how the firing function could produce more complex burst shapes.

💻/thinkpad/🧊/me/📓/2026/5/2/0

indy.dweb.archive

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

Learn more at Review Commons

Reply to the reviewers

We thank the reviewers and we are glad that they acknowledge this work to be a timely contribution to a quickly moving field and a valuable tool to generate testable hypothesis. We are pleased that reviewer #2 highlights that “a major strength is the combination of orthogonal evidence types” and that the tool serves to generate novel hypothesis. The revised manuscript will sharpen the positioning of the study within this context. Additional experimental evidence will be provided to address the points raised by reviewers #1 and #3.

Reviewer #1* 1.The authors do not co-IP ARF1. This does not surprise me as small GTPases often hydrolyse their GTP during lysis. *

We agree that this is likely due to transient association and GTP hydrolysis during lysis and will add a section to the manuscript.

There have been a number of ARF1 bioID screens done- have the authors checked if their complex has turned up here?

We will include this in the revised manuscript.

1. I am a bit confused by some of the interpretation about KO and loss of JTB staining. They interpret: "The SYS1 acts as a Golgi recruitment factor for both ARFRP1 and JTB". The ARFRP1 has been published and is a cytosolic protein, so that makes sense. However, the JTB is not cytosolic by a membrane protein, so cannot be "recruited". Now maybe it is retained in the Golgi by this interaction, but if that is the case you would still expect signal on another organelle or the plasma membrane (and we see it isnt degraded in the lysosome due to the western blot). I am confused by the authors model here.

We will clarify the phrasing and will provide a clearer interpretation, also considering the other improved imaging experiments that will be included in the revised manuscript.

4.The authors validate their JTB antibody and confirm the fact that there are not reduced SYS1 levels in the JTBKO- this is very clear (albeit unquantified). What I do not see validated is the SYS1KO. I think this is quite important.

We will validate SYS1 KO using TIDE and/or western blotting.

5.The colocalisation in panel 3D is weak and unclear to me. It is not quantified. It is not clear if there have been 3 repeats.

The revised manuscript will include improved imaging data. We will repeat relevant experiments, include appropriate controls and quantify where necessary.

6.The imaging in figure 3 is not clear in places, and it stands out in a very clear manuscript. I cannot see the JTB in panel F. There are no scale bars. The dynamic range of the image is not utalised. I do not see the stain in the JTB in either of the sys1 KO, i do not see the SYS1-FLAG staining in the complement, and it is not quantified at all. It may all seem trivial, but (to me) this is an absolutely critical bit of biology data to support the informatics.

The revised manuscript will include improved imaging data. We will repeat relevant experiments, include appropriate controls and quantify where necessary.

7.I am a bit unconvinced by the interpretation of it being a retrograde trafficking complex. This is for 2 key reasons- 1) the VSV-G is antrograde (despite unusually they interpret a "severe defect in retrograde transport"). 2) Even if it was only having an effect in the retrograde direction I would still remain a little open minded about it as you can easily mistake trafficking of a protein in one direction for another if an unknown protein (SNARE for example) has defective trafficking.

We used VSVG-KDEL in this assay. This setup specifically measures retrograde trafficking. We will clarify this in the revised manuscript. We will clarify in the Discussion that we confirmed a role in retrograde trafficking but cannot exclude a role in anterograde trafficking

Reviewer #2

Major comment: scope and interpretation of DepMap-derived functional evidence The manuscript could benefit from more clearly defining the scope of the functional evidence used to nominate complexes. The central co-dependency signal is derived from DepMap 24Q2 CRISPR gene-effect profiles, which are primarily cancer cell-line fitness/proliferation data. This is an important limitation because the resulting correlations may preferentially capture complexes or pathways that influence viability in proliferating cancer cells, while missing complexes active in differentiated, tissue-specific, stimulus-dependent, or non-proliferative contexts. Conversely, some correlations may reflect shared cancer-lineage or fitness dependencies rather than direct participation in a stable complex. The authors are appropriately cautious in stating that DepCom is not a complete inventory of human protein complexes, but the title, framing, and resource description could still be read as implying a more general catalogue of functional protein complexes. The authors might consider adding a clearer introduction to DepMap and explicitly discuss how the cancer-cell-line origin of the data affects interpretation of the 518 predicted complexes. This could be addressed without new experiments, for example by adding text early in the Results section explaining what the CRISPR gene-effect scores measure, and by expanding the Discussion to clarify that DepCom represents structurally plausible complexes prioritized by co-dependency across cancer cell lines, rather than an unbiased or context-independent map of human protein complexes. The selection of highlighted examples would also benefit from clearer justification. The peroxisome, actin, WNK/TSC22D2, and Golgi/JASS examples are biologically interesting, but the rationale for choosing them is not always explicit. Were they selected because they were novel, high-confidence, disease-associated, experimentally tractable, or representative of different resource categories? Briefly stating the selection criteria would help readers understand whether these examples are illustrative case studies or representative outcomes of the pipeline.

We agree with the reviewers' assessment that this resource should be viewed as hypothesis-generating and that the overall framing should be improved. We will revise the manuscript at the appropriate sections, according to the more detailed comments of all reviewers.

Minor comments

1. Clarify post-clustering removal of large/problematic protein families and complexes. In the Methods, the authors state that "clusters of histones and keratin clusters, as well as the mito-ribosome, complexes of the electron transport chain and the mediator complex" were removed because of their large sizes. This filtering step would benefit from additional detail. Please specify the criteria used to define these removed clusters, how many clusters/proteins were removed at this stage, and whether removal was based only on size or also on biological/manual curation. It would also be helpful to explain why these proteins or clusters were removed after clustering rather than excluded before graph construction and clustering, since highly connected or compositionally biased protein families could potentially influence neighboring cluster assignments. If available, a brief robustness check showing that pre-removal of these proteins gives similar candidate complexes would strengthen confidence in the clustering procedure.

We will add the requested information to the relevant section. Alongside the manuscript we will also provide lists of the complexes before and after every filtering step

1. Clarify the rationale for excluding complexes larger than 5000 residues. The 5000-residue cutoff is understandable for AF3 computational cost, but the manuscript should briefly state how many candidate complexes were excluded by this cutoff and whether this preferentially removes known large assemblies. This would help readers understand the scope of complexes that DepCom is expected to miss.

Alongside the manuscript we will now also provide lists of the complexes before and after every filtering step.

1. Improve wording in the CAP1/CFL1/WDR1/ACTB example. The sentence "Additionally, CAP1 works in concert with CFL1 to accelerate depolymerisation, though if a four-protein complex consisting of actin, WDR1, CAP1 and CFL1 is relevant is not clear" is difficult to parse. Possible revision might be something like: "Additionally, CAP1 works in concert with CFL1 to accelerate depolymerisation, although it remains unclear whether actin, WDR1, CAP1 and CFL1 form a stable four-protein complex in cells." This more clearly separates known biology from the speculative interpretation of the DepCom prediction.

Wording will be improved.

1. Improve reproducibility details for AF3 predictions. The Methods state that predictions were run using a local AF3 installation, but reproducibility would be improved by reporting relevant AF3 settings, number of seeds/models per complex, whether templates were used, how disordered regions were handled, and whether predictions were repeated for all complexes or only selected examples. This is especially important because the manuscript notes that multiple predictions can yield different subunit arrangements.

We will provide detailed settings in the methods section. Regarding disordered parts: All predictions used full length sequences (canonical UNIPROT ID) for each protein, so disordered residues are included. If disordered regions have low PLDDT and poor PAE, these regions will simply not score as interfaces in AlphaBridge. The one exception where we did crop structures is Figure 2D, but purely for visualization purposes, the full length complex did score in the pipeline (uncropped).

Reviewer #3

Co-essentiality is not the same as physical complex membership. This is the biggest conceptual concern. Genes in the same pathway are co-essential whether or not their products bind. The authors lean on the structural prediction step to filter this out, but that means the entire pipeline rests on AF3+AlphaBridge being correct about who interacts with whom. There is no independent benchmarking shown of how often AlphaBridge calls a true positive vs a false positive at the chosen 0.5 cutoff. Why 0.5? Where does that number come from? A short benchmarking section using known complexes (CORUM 5.0, hu.MAP 2.0, the PDB) would make the choice defensible. Right now it reads as arbitrary.

We thank the reviewer for bringing up the need for such an important clarification. We fully agree that co-essentiality does not equal physical interaction and structure predictions are imperfect. This is precisely the logic underlying our pipeline design, not a limitation we overlooked. The two data sources are used sequentially and serve distinct roles: first, we construct protein sets that are connected through networks of predicted binary physical interactions; then we cluster these based on DepMap correlations, selecting likely physical complexes that display co-essentiality between their components.

In other words, clustering on DepMap data alone would certainly return many spurious correlations: as the referee points out “Co-essentiality is not the same as physical complex membership”. Anchoring the search space with structural predictions substantially reduces this noise. Neither data source alone is sufficient, nor do we claim either is definitively "correct": the value lies in their combination. We hope improved phrasing in the revised manuscript will highlight this better.

Regarding benchmarking AlphaBridge score: we have benchmarked AlphaBridge, in response to reviewer feedback on the original AlphaBridge paper (Structure, Cell Press). In the figure here it is clear that in our benchmark of PDB structures (with

Comparison to existing resources is incomplete. I can't help but wonder what was found here that would not have been possible by analysing existing resources. CORUM 5.0 (7,193 mammalian complexes, ~71% human-derived; Tsitsiridis et al. 2024 NAR), hu.MAP 2.0 (Drew et al. 2021, ~6,965 complexes from >15,000 MS experiments), BioPlex 3.0 (Huttlin et al. 2021, 118,162 interactions in HEK293T), ad the Complex Portal already cover a large fraction of the human complexome. The authors compare to PDB, the original interactome paper, and Complex Portal, but they explicitly skip CORUM and hu.MAP, both of which are central reference resources in this space. Without including these, the "60 complexes unique to DepCom" number is not really meaningful. This needs to be redone properly.

We will add the comparison with Corum and hu-MAP in the revision.

Validation rate is one out of 518. The JASS work is solid, but a single experimentally validated complex out of 518 gives the reader essentially no estimate of how often the rest of the predictions are correct. Even a smaller systematic effort, say IP-MS on five to ten predicted novel complexes in the same cell line, would do an enormous amount to establish how trustworthy the resource is. The authors already have the V5/IP-MS pipeline running. Right now the manuscript implicitly asks the reader to trust 517 predictions on the strength of one validation.

In this paper we validated one out of the 60 complexes we claim are new. Notably we provide new biological data and demonstrate how consulting our resource, or following the same logic of combining functional and structural information, can lead to new exciting discoveries. We note that out of the 518 complexes we list, 69 complexes are exactly mirrored in the PDB and/or Complex Portal, while for another 389 there is partial evidence. Thus, our dataset is amply validated, and at the same time contains data to enable new discoveries. We also note, that following the release of our resource eight months ago, a new high-impact publication “validated” a complex we have independently picked in DepMap (Oosterheert et al, Choreography of rapid actin filament by coronin, cofilin and AIP1, Cell, 2025). We will rephrase relevant sections (also in response to reviewer 2) to increase clarity about validation.

The functional and disease clustering is potentially circular. GO terms and STRING associations are themselves derived in large part from the published literature on protein function, including text mining channels in STRING, much of which is downstream of complex membership. Of course complexes cluster into "DNA repair" and "vesicle trafficking" if you cluster on GO and STRING. The same applies to Open Targets, which integrates GWAS Catalog, ClinVar, literature mining, and other sources. The clustering is fine as a navigation aid for the website, but it is not, as currently presented, an independent validation of anything. I would tone the discussion down accordingly.

We did not mean to present the clustering as an independent validation. We will tone down the discussion accordingly.

AF3 limitations on this class of problem. AF3 itself acknowledges limitations (Abramson et al. 2024, including the December 2024 addendum), and subsequent benchmarking has flagged disordered regions, dynamic/large assemblies, and certain transmembrane systems as known weak points. The JASS complex is largely transmembrane, the WNK1-TSC22D2 example involves disorder-to-order transitions, and several flagship examples involve large multi-domain proteins. The authors acknowledge some of this in passing but should state explicitly which complexes were trimmed, how the trimming choices were made, and whether predictions were repeated with different seeds to check stability. Figure S4 is a good start, but for a resource paper a more systematic seed-stability analysis is warranted.

No complexes were trimmed for the initial AF3 predictions. The WNK1-TSC22D2 example was trimmed and re-predicted only for visualization purposes. We apologize for the misunderstanding and will state this more clearly.

AF3 certainly has limitations. Regarding disordered regions, these will almost always be assigned a poor pLDDT (also if AF3 wrongly folds them into helices). AlphaBridge will not pickup these low pLDDT regions as interfaces. Regarding dynamic assemblies, these might again lead to poor confidence scores and consequently these will not be picked up as interfaces by AlphaBridge. If AF3 confidence metrics are analyzed properly, the main concern for both disordered regions and dynamic assemblies is to miss true positive interactions, rather than finding false positive. As we did not aim to identify all possible human complexes, we consider focusing on the most confidently predicted interactions to be a fair trade off.

While the JASS complex is indeed a membrane protein complex, the predictions are exceptionally confident across multiple seeds (we can provide predictions from multiple seeds for revision), and validates experimentally. Of course, structure predictions are no substitute for experimental structures, as cautioned multiple times throughout the manuscript.

Figure S4 shows that despite the complex overall geometry being flexible, the interaction sites are predicted with high confidence across different poses. Since the aim of this study was to identify proteins interacting with each other, not accurate structures (which need to be solved experimentally), we argue that recomputing all structures with multiple seeds is disproportionately expensive computationally and would delay publication of a timely study while adding little.

Statistics are thin in several places. On the Fisher exact test for Golgi/ER enrichment in V5-JTB IP-MS (Supplemental Table 1), an odds ratio of 2.77 is modest, and there is no comparison to a matched control IP. Is this more than expected by chance against an appropriate background? The IP-MS volcano plots show many significant proteins, but how was the background controlled? On the LLM section, no quantitative evaluation is presented at all and the assessment is admitted to be subjective.

We will qualify the conclusions drawn from the IP-MS experiments. We maintain that together with the additional cell biology data, we build a compelling and convincing picture for this JASS complex.

Experimentally, the background is controlled by measuring enrichment over WT cell lines that have undergone the same IP procedure as the V5-SYS1/JTB expressing cells (lysis, incubation with the anti-V5 conjugated beads, same wash procedure and sample processing), as is the standard in the field. We will clarify in the Methods section. Regarding identification, FDR rate was set to 1% at protein and peptide level and peptide spectrum matches (PSMs) were additionally filtered for SequestHT Xcorr score >1.

We agree with the referee that the LLM interpretation is subjective and cannot be benchmarked. We suggest revising the resource and the paper, only providing structured LLM prompts to facilitate users asking the right questions, but we will not provide the LLM answers as part of the resource.

The 4�ACTB speculation. The authors themselves note the AlphaBridge score declines from 0.9 (1�ACTB) to 0.78 (4�ACTB), yet they speculate about functional implications. This is exactly the kind of post-hoc rationalisation around weak evidence that should either be supported with experiment or removed. Either remove or qualify as speculative.

We will qualify this as speculative

The LLM-assisted analysis. I am genuinely uncomfortable with releasing 76 LLM-generated complex annotations as part of a published resource when the authors openly state these have "not been systematically validated". Putting these summaries on a website with the imprimatur of a peer-reviewed paper will lead to them being cited and reused. At minimum, the website needs prominent warnings on every page where an LLM summary appears, the prompts must be fully reproducible (not just downloadable as JSON), and a small validation table, say 10 complexes scored by a domain expert for accuracy of each claim, should be included as a supplemental figure. As it stands this section reads like an enthusiastic add-on that has not been thought through with the same care as the rest of the work.

We thank the referee for bringing forward this consideration. We agree to remove the LLM answers for the 78 complexes from the manuscript and from the website, to ensure that the outputs cannot be cited. We will provide two different objective structure prompts for download to encourage variety in responses for curious users who want to explore. We will add a prominent disclaimer noting that responses resulting from these prompts cannot be interpreted as facts without validation.

We cannot guarantee reproducibility with modern LLM inference architecture. Even if seeds are kept the same and temperature=0, floating-point non-determinism in GPU operations, distributed inference, and batch effects may lead to different results. Furthermore, models go through many different iterations rapidly. As a consequence, it is impossible for us to guarantee reproducibility

Cutoffs and cluster numbers need stability analysis. The cutoff for the 75th-percentile DepMap correlation (mean of random + 3 SD = 0.147) is reasonable but should be accompanied by an FDR or precision/recall estimate against a labelled reference set. The choice of 20 final clusters in functional clustering (because that gave a peak in silhouette score) and 14 for disease clustering should also be supported by stability analysis, e.g. resampling.

The 75th percentile cutoff is, in our opinion, well justified and sufficient for our purposes. FDR and precision recall need a set of true and false positives. The DepMap correlation clusters are an intermediate step in our pipeline and do not necessarily hold the final complexes. How can intermediate reference DepMap clusters be constructed and defined as true or false positives? Even if we would score clusters that contain a known complex as true positives, how to define false positives? If clusters do not contain a known complex, that does not necessarily mean that these proteins don’t interact, just that they have not been shown to interact yet.

We will run resampling to improve confidence in the choice of cluster number.

Internal numerical consistency. The bioRxiv preprint abstract refers to 354 high-confidence multi-protein complexes, while the body of the manuscript discusses 518 (224 dimers + 294 multimers). The relationship between these numbers should be stated explicitly. Likewise, the breakdown of "60 unique to DepCom" into 41 heterodimers + 19 multimeric should be reconcilable in the figures and tables. The number "9,764 unique seed proteins" should also be clarified to confirm it is the DepCom-internal seed set and not inherited from the Zhang et al. coverage or hu.MAP 2.0 (9,963 proteins). These are easy fixes but matter for a resource paper.

BioRxiv preprint: The preprint that the reviewer read is an older version, which will be updated. .

The 9,764 unique seed proteins is from the Zhang et al paper, and are the human proteins identified to confidently interact with at least one other human protein. We will make this more clear.

Mander's overlap coefficient. The VSV-G(ts045)-KDELR retrograde-transport assay is well established and the experiment is clean, but MOC has been increasingly criticised in the colocalisation literature (Adler & Parmryd 2010, 2021). Best practice is to also report Manders' M1/M2 coefficients or Pearson's correlation alongside MOC. Adding these would be straightforward and would strengthen Fig 4B.

We will improve co-localization measures where appropriate.

Minor comments 1. Page 4: "candidate sets of potential multi-protein complex members". Pick one, they are either candidates or potential, not both.

Will be addressed.

Page 7: "Complex 294... mechanistic basis for CFL1 and WDR1 cooperation has only recently been described". Please update the reference list and language given how recent this is.

Will be addressed.

Page 7: JTB is described as "poorly characterised". This is a bit too strong. JTB has been studied in the context of TGF-β-induced mitochondrial regulation (Kanome et al. 2007), cytokinesis and chromosomal passenger complex association (Platica et al. 2011), the structural characterisation of its extracellular domain (Rousseau et al. 2012), and breast cancer biomarker work (Jayathirtha et al. 2022). A more accurate framing would be "incompletely characterised, with previously reported but functionally unresolved roles". The novelty here is the Golgi connection, which is genuine.

We will rephrase.

Page 8: the citation of Blomen et al. 2015 Science for "Golgi-related synthetic lethality" should be checked against the actual supplementary data of that paper to confirm the JTB attribution is correct.

Will be check.

Figure 1: as in many omics papers, please think of us colourblind readers. The pink-green DepMap correlation scale will be hard for some of us.

The color scheme in use, alongside others, was tested with two colleagues that have different variants of colour blindness and was judged to be the best compromise.

Figure 5A and 5B: 21 and 14 colour-coded clusters respectively in a single UMAP is too much. Consider splitting into separate panels by broad theme or providing an interactive version only.

We will focus on a subsection, and provide the full interactive version on the homepage

Page 11: "manually evaluated the quality of outputs". By whom, blinded to which model produced which output? Methods are silent on this.

As stated above, we will remove the LLM part

Some figures show "hairballs" with very limited informative content. Fig. 1B left panel and the AlphaBridge wheel plots in particular convey relatively little at the size shown.

We will try and find a way to draw the AlphaBridge circular plots in better resolution; we do not however that the reviewer’s observation might be an artefact of the PDF file distributed to reviewers.

The reference list looks a bit thin on prior systematic complexome efforts. BioPlex 3.0 (Huttlin et al. 2021 Cell), hu.MAP 2.0 (Drew et al. 2021 MSB) and CORUM 5.0 (Tsitsiridis et al. 2024 NAR) should all be cited and discussed.

We will include the additional references where appropriate

The discussion section drifts into general comments about AI in science that don't add much. I would cut about a third of it and use the space for a more careful framing of the actual contribution.

We will shorten the discussion section and phrase more carefully.

General assessment Reviewer #3: The strongest aspect of this study is the JASS complex story. The IP-MS, the SYS1-KO rescue experiment, the VSV-G(ts045)-KDELR transport assay, and the orthogonal CRISPR screens with diphtheria and Pseudomonas exotoxins together build a convincing case for JTB as a regulator of Golgi-to-ER retrograde trafficking. This part of the paper is genuinely nice work and would stand on its own. The pipeline itself, combining structural predictions with functional dependency data and filtering with AlphaBridge, is sensible and timely. It is a reasonable demonstration of how confidence filtering should be done at this kind of scale. The main limitations concern the resource framing. After reading the manuscript several times I am still trying to identify the central novel contribution beyond the JASS validation. The interactome predictions are taken from Zhang et al., DepMap is public, AF3 is public, AlphaBridge is the authors' own previously published tool, and GO/STRING/Open Targets/dbPTM are all public. The manuscript is essentially an integrative pipeline plus a website plus one experimentally followed-up complex. The framing oversells what is genuinely new. The authors' own comparison (Fig. S3) shows 60 complexes "unique to DepCom" out of 518, of which 41 are heterodimers and only 19 are multimeric. Nineteen genuinely novel multi-protein complexes is still a contribution but it is a long way from the 354/518 that the abstract and discussion implicitly emphasise. The validation rate (one of 518) and the missing comparisons to CORUM 5.0 and hu.MAP 2.0 are the two issues that most need addressing.

We will rephrase these issue to adjust the framing. We would put forward that the main contribution of this manuscript is to present an integrative framework that combines data from orthogonal sources to highlight the possibility of structure prediction models to serve as a discovery tool. The reviewer identifies correctly (albeit derogatorily) that this is “essentially” an integrative pipeline. But it is an integrative pipeline that combines genetics and computational structure predictions in a novel (to the best of our knowledge) way and surfaces interesting new biology. The biology of the JASS complex goes well-beyond simple validation experiments, and we believe its discovery (based on our data) carries more value that the reviewer attributes to it.

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

Learn more at Review Commons

Referee #3

Evidence, reproducibility and clarity

Summary:

Uckelmann and colleagues combine the recently published binary human interactome predictions from Zhang et al. (2025) with co-essentiality data from DepMap CRISPR screens to nominate sets of proteins that may form higher-order complexes. They cluster proteins around each "seed" using Leiden community detection on the DepMap correlation matrix, run AlphaFold3 on each candidate set, and apply AlphaBridge to retain only those interfaces predicted with confidence. After filtering they arrive at 518 complexes, of which 224 are dimers and 294 are larger assemblies (note: the abstract of the bioRxiv preprint refers to 354 high-confidence complexes, so the relationship between these numbers should be made explicit). They illustrate the resource with a few worked examples (PEX3/16/19/ACBD5, an actin-CFL1-WDR1-CAP1 assembly, a WNK1-NRBP1-TSC22D2 complex), and they experimentally validate one previously uncharacterised assembly that they name JASS (JTB-ARFRP1-SYS1), placing JTB at the Golgi and showing a role for it in Golgi-to-ER retrograde transport. They also provide a web portal (depcom.eu) with PTM mapping, GO/STRING-based functional clustering, Open Targets disease clustering, and LLM-generated executive summaries.

Major comments:

I am supportive of integrating orthogonal datasets in this kind of framework, but I am much less enthusiastic about how the analyses are carried through, and I think there are several issues that need adressing before this work is publishable.

1. Co-essentiality is not the same as physical complex membership. This is the biggest conceptual concern. Genes in the same pathway are co-essential whether or not their products bind. The authors lean on the structural prediction step to filter this out, but that means the entire pipeline rests on AF3+AlphaBridge being correct about who interacts with whom. There is no independent benchmarking shown of how often AlphaBridge calls a true positive vs a false positive at the chosen 0.5 cutoff. Why 0.5? Where does that number come from? A short benchmarking section using known complexes (CORUM 5.0, hu.MAP 2.0, the PDB) would make the choice defensible. Right now it reads as arbitrary.
2. Comparison to existing resources is incomplete. I can't help but wonder what was found here that would not have been possible by analysing existing resources. CORUM 5.0 (7,193 mammalian complexes, ~71% human-derived; Tsitsiridis et al. 2024 NAR), hu.MAP 2.0 (Drew et al. 2021, ~6,965 complexes from >15,000 MS experiments), BioPlex 3.0 (Huttlin et al. 2021, 118,162 interactions in HEK293T), and the Complex Portal already cover a large fraction of the human complexome. The authors compare to PDB, the original interactome paper, and Complex Portal, but they explicitly skip CORUM and hu.MAP, both of which are central reference resources in this space. Without including these, the "60 complexes unique to DepCom" number is not really meaningful. This needs to be redone properly.
3. Validation rate is one out of 518. The JASS work is solid, but a single experimentally validated complex out of 518 gives the reader essentially no estimate of how often the rest of the predictions are correct. Even a smaller systematic effort, say IP-MS on five to ten predicted novel complexes in the same cell line, would do an enormous amount to establish how trustworthy the resource is. The authors already have the V5/IP-MS pipeline running. Right now the manuscript implicitly asks the reader to trust 517 predictions on the strength of one validation.
4. The functional and disease clustering is potentially circular. GO terms and STRING associations are themselves derived in large part from the published literature on protein function, including text mining channels in STRING, much of which is downstream of complex membership. Of course complexes cluster into "DNA repair" and "vesicle trafficking" if you cluster on GO and STRING. The same applies to Open Targets, which integrates GWAS Catalog, ClinVar, literature mining, and other sources. The clustering is fine as a navigation aid for the website, but it is not, as currently presented, an independent validation of anything. I would tone the discussion down accordingly.
5. AF3 limitations on this class of problem. AF3 itself acknowledges limitations (Abramson et al. 2024, including the December 2024 addendum), and subsequent benchmarking has flagged disordered regions, dynamic/large assemblies, and certain transmembrane systems as known weak points. The JASS complex is largely transmembrane, the WNK1-TSC22D2 example involves disorder-to-order transitions, and several flagship examples involve large multi-domain proteins. The authors acknowledge some of this in passing but should state explicitly which complexes were trimmed, how the trimming choices were made, and whether predictions were repeated with different seeds to check stability. Figure S4 is a good start, but for a resource paper a more systematic seed-stability analysis is warranted.
6. Statistics are thin in several places. On the Fisher exact test for Golgi/ER enrichment in V5-JTB IP-MS (Supplemental Table 1), an odds ratio of 2.77 is modest, and there is no comparison to a matched control IP. Is this more than expected by chance against an appropriate background? The IP-MS volcano plots show many significant proteins, but how was the background controlled? On the LLM section, no quantitative evaluation is presented at all and the assessment is admitted to be subjective.
7. The 4×ACTB speculation. The authors themselves note the AlphaBridge score declines from 0.9 (1×ACTB) to 0.78 (4×ACTB), yet they speculate about functional implications. This is exactly the kind of post-hoc rationalisation around weak evidence that should either be supported with experiment or removed. Either remove or qualify as speculative.
8. The LLM-assisted analysis. I am genuinely uncomfortable with releasing 76 LLM-generated complex annotations as part of a published resource when the authors openly state these have "not been systematically validated". Putting these summaries on a website with the imprimatur of a peer-reviewed paper will lead to them being cited and reused. At minimum, the website needs prominent warnings on every page where an LLM summary appears, the prompts must be fully reproducible (not just downloadable as JSON), and a small validation table, say 10 complexes scored by a domain expert for accuracy of each claim, should be included as a supplemental figure. As it stands this section reads like an enthusiastic add-on that has not been thought through with the same care as the rest of the work.
9. Cutoffs and cluster numbers need stability analysis. The cutoff for the 75th-percentile DepMap correlation (mean of random + 3 SD = 0.147) is reasonable but should be accompanied by an FDR or precision/recall estimate against a labelled reference set. The choice of 20 final clusters in functional clustering (because that gave a peak in silhouette score) and 14 for disease clustering should also be supported by stability analysis, e.g. resampling.
10. Internal numerical consistency. The bioRxiv preprint abstract refers to 354 high-confidence multi-protein complexes, while the body of the manuscript discusses 518 (224 dimers + 294 multimers). The relationship between these numbers should be stated explicitly. Likewise, the breakdown of "60 unique to DepCom" into 41 heterodimers + 19 multimeric should be reconcilable in the figures and tables. The number "9,764 unique seed proteins" should also be clarified to confirm it is the DepCom-internal seed set and not inherited from the Zhang et al. coverage or hu.MAP 2.0 (9,963 proteins). These are easy fixes but matter for a resource paper.
11. Mander's overlap coefficient. The VSV-G(ts045)-KDELR retrograde-transport assay is well established and the experiment is clean, but MOC has been increasingly criticised in the colocalisation literature (Adler & Parmryd 2010, 2021). Best practice is to also report Manders' M1/M2 coefficients or Pearson's correlation alongside MOC. Adding these would be straightforward and would strengthen Fig 4B.

Minor comments

1. Page 4: "candidate sets of potential multi-protein complex members". Pick one, they are either candidates or potential, not both.
2. Page 7: "Complex 294... mechanistic basis for CFL1 and WDR1 cooperation has only recently been described". Please update the reference list and language given how recent this is.
3. Page 7: JTB is described as "poorly characterised". This is a bit too strong. JTB has been studied in the context of TGF-β-induced mitochondrial regulation (Kanome et al. 2007), cytokinesis and chromosomal passenger complex association (Platica et al. 2011), the structural characterisation of its extracellular domain (Rousseau et al. 2012), and breast cancer biomarker work (Jayathirtha et al. 2022). A more accurate framing would be "incompletely characterised, with previously reported but functionally unresolved roles". The novelty here is the Golgi connection, which is genuine.
4. Page 8: the citation of Blomen et al. 2015 Science for "Golgi-related synthetic lethality" should be checked against the actual supplementary data of that paper to confirm the JTB attribution is correct.
5. Figure 1: as in many omics papers, please think of us colourblind readers. The pink-green DepMap correlation scale will be hard for some of us.
6. Figure 5A and 5B: 21 and 14 colour-coded clusters respectively in a single UMAP is too much. Consider splitting into separate panels by broad theme or providing an interactive version only.
7. Page 11: "manually evaluated the quality of outputs". By whom, blinded to which model produced which output? Methods are silent on this.
8. Some figures show "hairballs" with very limited informative content. Fig. 1B left panel and the AlphaBridge wheel plots in particular convey relatively little at the size shown.
9. The reference list looks a bit thin on prior systematic complexome efforts. BioPlex 3.0 (Huttlin et al. 2021 Cell), hu.MAP 2.0 (Drew et al. 2021 MSB) and CORUM 5.0 (Tsitsiridis et al. 2024 NAR) should all be cited and discussed.
10. The discussion section drifts into general comments about AI in science that don't add much. I would cut about a third of it and use the space for a more careful framing of the actual contribution.

Significance

General assessment:

The strongest aspect of this study is the JASS complex story. The IP-MS, the SYS1-KO rescue experiment, the VSV-G(ts045)-KDELR transport assay, and the orthogonal CRISPR screens with diphtheria and Pseudomonas exotoxins together build a convincing case for JTB as a regulator of Golgi-to-ER retrograde trafficking. This part of the paper is genuinely nice work and would stand on its own. The pipeline itself, combining structural predictions with functional dependency data and filtering with AlphaBridge, is sensible and timely. It is a reasonable demonstration of how confidence filtering should be done at this kind of scale.

The main limitations concern the resource framing. After reading the manuscript several times I am still trying to identify the central novel contribution beyond the JASS validation. The interactome predictions are taken from Zhang et al., DepMap is public, AF3 is public, AlphaBridge is the authors' own previously published tool, and GO/STRING/Open Targets/dbPTM are all public. The manuscript is essentially an integrative pipeline plus a website plus one experimentally followed-up complex. The framing oversells what is genuinely new. The authors' own comparison (Fig. S3) shows 60 complexes "unique to DepCom" out of 518, of which 41 are heterodimers and only 19 are multimeric. Nineteen genuinely novel multi-protein complexes is still a contribution but it is a long way from the 354/518 that the abstract and discussion implicitly emphasise. The validation rate (one of 518) and the missing comparisons to CORUM 5.0 and hu.MAP 2.0 are the two issues that most need addressing.

Advance:

The advance is incremental rather than conceptual. The idea of intersecting co-essentiality with structural predictions is sensible but not new in spirit, and similar hybrid approaches are now becoming more common in this space (see e.g. EndoMAP.v1, Gonzalez-Lozano et al. 2025 Nature, which the authors do cite). What is new here is the specific implementation, the AlphaBridge filtering layer, and the JASS finding. The technical advance lies in the AlphaBridge filtering step on top of AF3 at a reasonably large scale. The biological advance is the JASS complex and the demonstration that JTB plays a role in Golgi-to-ER retrograde transport, which is genuinely new and well supported.

Audience:

This work will be of interest mainly to specialised audiences in structural proteomics, computational biology of protein complexes, and the protein-protein interaction community. The JASS finding will be of interest to a broader readership in cell biology, particularly those working on Golgi trafficking, ARF/ARL family GTPases, and retrograde transport. The web resource will likely find users among researchers studying specific complexes who want a quick structural hypothesis. I do not think the work, in its current form, will reach broad audiences in the way the authors hope, but a more sober framing would actually help it land better in the specialist community where it belong.

My expertise:

Mass spectrometry-based proteomics, protein-protein interaction mapping, systems biology, structural biology. I have working knowledge but not deep expertise in: structural prediction confidence metrics (AF3, AlphaBridge implementation details), DepMap CRISPR co-essentiality analysis, and Golgi cell biology. I would defer to a computational structural biology or cell biology specialist on the AF3 confidence interpretation details and on the cell biology specifics of the JASS validation.

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

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

The study presents DepCom as a broad resource for discovering human multi-protein complexes by integrating predicted binary interactions, DepMap co-dependency, AF3 modelling, and AlphaBridge filtering. Overall, the computational strategy is well motivated, and the experimental validation of the JTB/SYS1/ARFRP1 complex provides a compelling example of how the resource can generate testable biological hypotheses.

Major comment: scope and interpretation of DepMap-derived functional evidence

The manuscript could benefit from more clearly defining the scope of the functional evidence used to nominate complexes. The central co-dependency signal is derived from DepMap 24Q2 CRISPR gene-effect profiles, which are primarily cancer cell-line fitness/proliferation data. This is an important limitation because the resulting correlations may preferentially capture complexes or pathways that influence viability in proliferating cancer cells, while missing complexes active in differentiated, tissue-specific, stimulus-dependent, or non-proliferative contexts. Conversely, some correlations may reflect shared cancer-lineage or fitness dependencies rather than direct participation in a stable complex. The authors are appropriately cautious in stating that DepCom is not a complete inventory of human protein complexes, but the title, framing, and resource description could still be read as implying a more general catalogue of functional protein complexes. The authors might consider adding a clearer introduction to DepMap and explicitly discuss how the cancer-cell-line origin of the data affects interpretation of the 518 predicted complexes. This could be addressed without new experiments, for example by adding text early in the Results section explaining what the CRISPR gene-effect scores measure, and by expanding the Discussion to clarify that DepCom represents structurally plausible complexes prioritized by co-dependency across cancer cell lines, rather than an unbiased or context-independent map of human protein complexes. The selection of highlighted examples would also benefit from clearer justification. The peroxisome, actin, WNK/TSC22D2, and Golgi/JASS examples are biologically interesting, but the rationale for choosing them is not always explicit. Were they selected because they were novel, high-confidence, disease-associated, experimentally tractable, or representative of different resource categories? Briefly stating the selection criteria would help readers understand whether these examples are illustrative case studies or representative outcomes of the pipeline.

Minor comments

1. Clarify post-clustering removal of large/problematic protein families and complexes.

In the Methods, the authors state that "clusters of histones and keratin clusters, as well as the mito-ribosome, complexes of the electron transport chain and the mediator complex" were removed because of their large sizes. This filtering step would benefit from additional detail. Please specify the criteria used to define these removed clusters, how many clusters/proteins were removed at this stage, and whether removal was based only on size or also on biological/manual curation. It would also be helpful to explain why these proteins or clusters were removed after clustering rather than excluded before graph construction and clustering, since highly connected or compositionally biased protein families could potentially influence neighboring cluster assignments. If available, a brief robustness check showing that pre-removal of these proteins gives similar candidate complexes would strengthen confidence in the clustering procedure. 2. Clarify the rationale for excluding complexes larger than 5000 residues.

The 5000-residue cutoff is understandable for AF3 computational cost, but the manuscript should briefly state how many candidate complexes were excluded by this cutoff and whether this preferentially removes known large assemblies. This would help readers understand the scope of complexes that DepCom is expected to miss. 3. Improve wording in the CAP1/CFL1/WDR1/ACTB example.

The sentence "Additionally, CAP1 works in concert with CFL1 to accelerate depolymerisation, though if a four-protein complex consisting of actin, WDR1, CAP1 and CFL1 is relevant is not clear" is difficult to parse. Possible revision might be something like: "Additionally, CAP1 works in concert with CFL1 to accelerate depolymerisation, although it remains unclear whether actin, WDR1, CAP1 and CFL1 form a stable four-protein complex in cells." This more clearly separates known biology from the speculative interpretation of the DepCom prediction. 4. Improve reproducibility details for AF3 predictions.

The Methods state that predictions were run using a local AF3 installation, but reproducibility would be improved by reporting relevant AF3 settings, number of seeds/models per complex, whether templates were used, how disordered regions were handled, and whether predictions were repeated for all complexes or only selected examples. This is especially important because the manuscript notes that multiple predictions can yield different subunit arrangements.-

Significance

General assessment:

This study presents a timely and useful resource for prioritizing candidate human protein complexes by integrating predicted binary protein-protein interactions, DepMap co-dependency profiles, AlphaFold3 structure prediction, and AlphaBridge confidence filtering. A major strength is the combination of orthogonal evidence types: physical interaction predictions define a tractable search space, functional co-dependency helps identify coherent protein groups, and structure-confidence metrics provide an additional filter on the resulting candidates. The experimental validation of the JTB/SYS1/ARFRP1 complex is also a strong aspect of the study, as it demonstrates that the resource can generate biologically meaningful and experimentally testable hypotheses.

The main limitation is that the resource should be interpreted as a prioritized, hypothesis-generating dataset rather than a comprehensive or context-independent catalogue of human protein complexes. As noted above, the DepMap-derived signal reflects cancer cell-line fitness/proliferation dependencies, and the final complex set is also shaped by the starting interactome, filtering choices, and computational constraints on complex size. These limitations do not undermine the utility of the resource, but they should be clearly framed for readers.

One aspect that could further increase the impact and usability of the study is the DepCom web resource. The searchable table of complexes is already useful, particularly for users who want to query by gene or protein name. However, the website also presents functional and disease-based clustering, and many users may want to search or filter complexes by biological process, GO term, pathway, disease association, or disease cluster. Adding GO-term and disease-association fields to the main table, and allowing users to search/filter by these annotations, would make the resource more discoverable and useful to researchers approaching the dataset from a biological process or disease area rather than from a specific gene.

Advance:

The advance is primarily technical and resource-oriented, with an accompanying functional biological demonstration. The study helps fill a gap between large-scale binary interaction prediction and the more difficult problem of nominating higher-order assemblies. By using functional dependency profiles to prioritize multi-protein combinations before structure prediction, the authors reduce an otherwise intractable search space and generate a set of structurally plausible candidate complexes. The JASS complex and the proposed role of JTB in Golgi-to-ER retrograde trafficking provide a compelling example of biological discovery enabled by the pipeline.

The broader DepCom resource, including predicted complex structures, AlphaBridge interface-confidence information, PTM-interface mapping, functional/disease clustering, and downloadable LLM prompts, should provide useful starting points for follow-up studies. These outputs are best viewed as hypothesis-generating rather than definitive biological annotation, but they represent a valuable extension of existing protein-interaction and structure-prediction resources.

Audience:

The study will likely interest a broad basic-research audience, especially researchers in protein complex biology, structural biology, functional genomics, systems biology, cancer dependency mapping, cell biology, and computational biology. It may also be useful to investigators studying specific pathways or poorly characterized proteins, since the resource provides candidate interaction partners and structural hypotheses that can guide experiments. The translational relevance is more indirect, mainly through disease-association clustering and potential target-discovery applications, but the immediate audience is likely to be basic and computational researchers.

My expertise is in computational protein databases, protein domain classification, structural/evolutionary analysis of proteins, and functional annotation resources, including experience with the ECOD database for evolutionary classification of protein domains. I am less able to evaluate the fine details the experimental cell-biology assays beyond their general interpretation and reporting.

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

Characterising protein complexes is a fundamental goal in modern molecular cell biology. Here, Uckelmann and colleagues have presented a solution to part of this problem. By combining functional clustering with alphafold modelling, they present a high throughput bioinformatic solution. The paper and figures are exceptionally clear and well presented. The conclusions are reasonable, and the data interesting. I am a cell biologist with expertise in molecular machinery of trafficking, so the focus of my review will be on the identification of a new complex, that is proposed to have a role in retrograde trafficking. On the whole I find this a interesting and convincing finding. However I have some comments and questions that I hope may help the authors. I will naturally focus my comments on the cell biology.

1.The authors do not co-IP ARF1. This does not surprise me as small GTPases often hydrolyse their GTP during lysis. 2.There have been a number of ARF1 bioID screens done- have the authors checked if their complex has turned up here? 3.I am a bit confused by some of the interpretation about KO and loss of JTB staining. They interpret: "The SYS1 acts as a Golgi recruitment factor for both ARFRP1 and JTB". The ARFRP1 has been published and is a cytosolic protein, so that makes sense. However, the JTB is not cytosolic by a membrane protein, so cannot be "recruited". Now maybe it is retained in the Golgi by this interaction, but if that is the case you would still expect signal on another organelle or the plasma membrane (and we see it isnt degraded in the lysosome due to the western blot). I am confused by the authors model here. 4.The authors validate their JTB antibody and confirm the fact that there are not reduced SYS1 levels in the JTBKO- this is very clear (albeit unquantified). What I do not see validated is the SYS1KO. I think this is quite important. 5.The colocalisation in panel 3D is weak and unclear to me. It is not quantified. It is not clear if there have been 3 repeats. 6.The imaging in figure 3 is not clear in places, and it stands out in a very clear manuscript. I cannot see the JTB in panel F. There are no scale bars. The dynamic range of the image is not utalised. I do not see the stain in the JTB in either of the sys1 KO, i do not see the SYS1-FLAG staining in the complement, and it is not quantified at all. It may all seem trivial, but (to me) this is an absolutely critical bit of biology data to support the informatics. 7.I am a bit unconvinced by the interpretation of it being a retrograde trafficking complex. This is for 2 key reasons- 1) the VSV-G is antrograde (despite unusually they interpret a "severe defect in retrograde transport"). 2) Even if it was only having an effect in the retrograde direction I would still remain a little open minded about it as you can easily mistake trafficking of a protein in one direction for another if an unknown protein (SNARE for example) has defective trafficking.

Significance

Characterising protein complexes is a fundamental goal in modern molecular cell biology. Here, Uckelmann and colleagues have presented a solution to part of this problem. By combining functional clustering with alphafold modelling, they present a high throughput bioinformatic solution. The paper and figures are exceptionally clear and well presented. The conclusions are reasonable, and the data interesting. I am a cell biologist with expertise in molecular machinery of trafficking, so the focus of my review will be on the identification of a new complex, that is proposed to have a role in retrograde trafficking. On the whole I find this a interesting and convincing finding.

This manuscript provides needed clarification for some unexpected behaviour we can experience with RGI outputs, and outlines some possible improvements notably for better prediction of beta-lactamases.

My major comment is that although the authors point out that only a very few ARG models return high rate of False positive, they do not consider that these genes are actually a high burden for non-experienced users of RGI. It should more clearly emphasized.

The main issue being efflux pumps which can basically be found in every bacterial genome. In the paper, they show for instance that RGI returns ~50% false positive with the adeF reference gene i.e. with % protein similarity ranging between 60% and 70% (and down to 40% when looking at the protein identity score, cf current CARD-R version). At such a low rate, every input genome will return an adeF resistance gene, even when phylogenetically distant (eg. adeF from Acinetobacter baumanni is detected in species from other phyla such as Bacteroides or Campylobacter).

The second critical issue are detections of genes from VAN operons. These operons are of extreme clinical importance since they confer vancomycin resistance in nosocomial pathogens Enterococci and Staphylococci. Their current bitscore cutoff again return a high rate of false positive, with some detections as low as 30% amino acid identity for van Y (result from CARD-R current version). 90% of RGI users will consider that they effectively have a vancomycin resistance gene in their genome (not even the full operon) and find this alarming.

These two issues are in my opinion more urgent to solve than the under-detection of betalactamases, because there is already countless published studies reporting over-inflated numbers of resistance genes (including numerous vancomycin resistance) in genomic and metagenomic assemblies based on RGI's results.

De 'lemon's theorie' (bekend als The Market for Lemons) is een economisch concept uit 1970 van Nobelprijswinnaar George Akerlof. Het verklaart hoe asymmetrische informatie (waarbij de verkoper meer weet over een product dan de koper) kan leiden tot marktfalen, waardoor uiteindelijk alleen slechte producten (citroenen) overblijven.

Hier is hoe het werkt in de praktijk: De casus (tweedehands auto's): De theorie wordt vaak uitgelegd aan de hand van tweedehands auto's. Er zijn twee soorten auto's: goede auto's en slechte auto's (in Amerikaans-Engels 'lemons' of 'citroenen' genoemd).

Informatie-ongelijkheid: De verkoper weet precies of de auto goed of slecht is. De koper kan dit aan de buitenkant niet zien en weet dus niet of hij een goede of een slechte auto koopt.

De gemiddelde prijs: Omdat de koper het risico loopt een 'citroen' te kopen, is hij enkel bereid om de gemiddelde prijs van een auto te betalen.

Het resultaat (marktfalen): De eigenaren van goede auto's vinden deze gemiddelde prijs veel te laag en halen hun auto uit de verkoop. Hierdoor blijven uiteindelijk alleen de slechte auto's over en stort de markt voor goede auto's in elkaar.

Wikipedia +1

Het rationele keuzemodel (of rationele-keuzetheorie) is een theorie uit de economie en sociale wetenschappen die stelt dat mensen weloverwogen, logische beslissingen nemen. Ieder individu kiest de optie die de grootste persoonlijke voordelen (nutsmaximalisatie) en de minste kosten oplevert

eLife Assessment

This study presents potentially important findings linking peripheral inflammation to the remodeling of perinodal adipose tissue and draining lymph nodes, suggesting a mechanism by which local tissue inflammation can reshape LN structure and metabolism. The idea is solid and supported by observations. However, the evidence remains incomplete in parts, as several conclusions rely on correlative weight and cellularity measurements, and macrophage involvement requires further validation.

Reviewer #1 (Public review):

The idea is super interesting, and the subsequent work is potentially significant because it links peripheral inflammation to remodelling of perinodal adipose tissue and draining lymph nodes. This suggests an antigen-independent manner by which local tissue inflammation can communicate with and reshape immune organ structure and tissue metabolism. However, the evidence is suggestive. For instance, many conclusions rely on correlational weight/cellularity relationships, models with confounders (spontaneous wounding; potentially systemic IMQ), and macrophage dependence inferred from a single pharmacologic approach without definitive depletion/lineage or tracer-based causal link.

Major Comments:

(1) "Wounding/fighting" evidence is confounding.

Unless I am mistaken, a large part of the argument for inflammation-driven perinodal fat pad atrophy and LN expansion relies on spontaneous fighting injuries in co-housed CCR2-/- males, including animals "culled...due to excessive wounding." Because wound severity, duration, infection load, stress, and cage dynamics are uncontrolled, isn't it difficult to assign causality to "cutaneous inflammation"?

(2) The "CCR2-independent macrophage" conclusion.

The manuscript interprets persistence/accumulation of macrophages despite reduced inflammatory monocytes as CCR2-independent recruitment or local proliferation. However, CCR2 deficiency can alter immune baselines and long-term tissue remodelling. Perhaps consider bone marrow chimeras (WT to CCR2-/-, CCR2-/- to WT ????) or an inducible CCR2 deletion approach to separate developmental/systemic effects from acute inflammation-driven mechanisms. If "in situ proliferation" is proposed, include a direct readout (e.g., Ki67 in ATMs in the fat pad).

(3) IMQ and systemic effects.

The work relies on topical Aldara/imiquimod as an "inflammation without antigen" driver of distal LN/fat-pad remodelling. But IMQ is well known (and cited by the authors) to enter circulation and drive systemic responses, which could blur whether effects are truly draining-site specific vs systemic metabolic/inflammatory effects. It would be ideal to provide systemic context: plasma cytokines and/or metabolic readouts (e.g., circulating FFAs) to distinguish local vs systemic drivers.

(4) Macrophage dependence is inferred from CSF1R inhibitor treatment.

However, validation of macrophage depletion and specificity is incomplete. The manuscript uses AZD7507 (CSF1R inhibitor) and observes partial rescue of fat pad/LN phenotype while skin severity (PASI) is unaffected. But, to this reviewer, the data shown do not clearly quantify actual macrophage depletion efficiency in the target fat pad, and LN at endpoint, and CSF1R blockade can affect multiple myeloid populations. Therefore, show absolute macrophage counts (and likely other myeloid populations) in fat pad and LN with/without AZD7507 at the analysed timepoints, not only outcome weights. (The methods describe dosing but not endpoint depletion quantification??)

(5) Fat pad atrophy/LN expansion is a correlation.

The paper emphasises negative correlations between fat pad and LN weights/cellularity at baseline and with inflammation. But correlation does not establish whether fat pad lipolysis drives LN expansion, whether LN changes drive fat remodelling, or whether both reflect systemic mediators. Add tissue-level evidence distinguishing true adipocyte loss vs other contributors to "weight change" (e.g., oedema/fibrosis).

(6) Evidence for "fatty acid donation" from fat pad to LN.

The lipid data are described as "exemplary," and the inference that LN fatty acids originate from the fat pad is based on temporal ordering and relative abundance. This does not rule out plasma spillover, LN-intrinsic metabolism, or altered lymph flow.

Reviewer #2 (Public review):

The authors aim to demonstrate skin inflammation is associated with fat pad atrophy and lymph node expansion. They further propose that these phenotypes are driven by the recruitment and lipid metabolism of CCR2-independent macrophages.

The authors took advantage of two skin inflammation models, fight-induced and imauimod-induced skin inflammation and analyzed multiple tissues, including skin, fat pads, and lymph nodes. Using a macropahge-depletion method (e.g., CSF-1R inhibitor), the authors further suggest the inverse correlation between fat pads atrophy and lymph node expansion is macropahge-dependent. While the study identifies this intriguing inverse correlation during skin inflammation, the causal pathway linking fat pad atrophy and lymph nodes enlargement has not been clearly established.

To improve the rigor of the manuscript, the authors address the following concerns;

(1) CCR2-deficient mice showed reduced inflammatory monocytes and monocyte-derived macrophages (PMID:16462739; 16341265). During tissue inflammation, CCR2+ classical monocytes are typically recruited to the injured peripheral tissues, including skin, where they differentiate into monocyte-derived macrophages (PMID:38474365). While inflammatory monocytes were reduced in the skin (Figure 3 d), fat pads (Figure 4a, S2D) of CCR2-deficient mice, macrophage numbers were significantly increased in these mice. It remains unclear whether CCR2-independent macrophages were newly recruited from alternative sources or tissue-resident macrophages underwent local self-proliferation to compensate for the loss of CCR2+ monocyte-derived macrophages.

(2) In line 258, the authors state that there was "a significant reduction in CD11C- CD206+ anti-inflammatory macrophages (Figure 4b i-iii)". However, the quantification data in Figure 4b iii do not appear to show any reduction in anti-inflammatory macrophages in either males or females. Please reconcile this discrepancy between the text and the figure.

(3) Although CD11C and CD206 were historically used as markers of inflammatory and anti-inflammatory markers, respectively. These markers are no longer considered sufficient to define the macrophage polarization state, particularly in adipose tissue, where they are constitutively expressed by resident macrophages (PMID:34210853). Numerous studies have demonstrated substantial macrophage diversity/heterogeneity across iWAT, eWAT, and brown fat tissues. The authors should discuss adipose macrophage diversity beyond the outdated M1/M2 frame.

Same this will also be date-wise amount

I think this is date-wise sum and not total sum

這個沒有 hyperlink

雖然不太重要，但這裡用安裝方式或是拿幾個熱門的來描述會不會比較好?

這裡讀起來微不舒服，改成這樣會不會比較好?「在比較之前,我們先單純看看 remote MCP server 作為 remote server 本身有哪些好處」

這裡只說明了 support Skills 的困難,但沒講 support MCP 為什麼相對簡單。建議補一句對照，感覺這樣可以加深印象以及明顯感受到差別。是說 effort 不用用英文吧 哈哈

typo

Definição de recuperação

Definição de treino em sua forma mais básica

Chrissy Teigen’s situation presents a complication for how this chapter frames crowd harassment (the way that a crowd acts in concert to harass an individual) as a crime committed against the victim of the group. Before she became the target of a “dog pile” (a large number of people harassing one person at once), Teigen had also been very active with online harassment. This makes her case raise a question the authors did not pose here; under what conditions does a crowd's target deserve to be harassed? Therefore, this case resists the victim/perpetrator distinction that is implied within this chapters framework.

When reading the definition of doxing by Wikipedia, I realized that I am able to connect this to my time spent on social media. While, I have not personally been doxed I have seen multiple people online discuss being doxed. I notice that this happens usually to people online who have been cancelled or whom the public does not like at the moment. These users, once they have been cancelled, will often create posts that beg people to stop spreading their personal information on the internet because they feel unsafe.

The Ministry of Health's intervention shows that the brand may face regulatory penalties, including the revocation of its product license for violating advertising guidelines.

Wearing white coats in promotional videos cause the audiens have a false impression of medical accreditation. This is unethical and it could manipulate public trust of the brand.

The advertisement is misleading because it implies that the cosmetics are safe to consume, and this could pose a serious health risk to consumers.

The area of fake crowds causing harassment has an ironic resemblance to the way we found bots were similar to each other -- the same type of automation which produces a manufactured political support base and/or artificially inflates engagement (for example, "I have thousands of supporters!") can be used to create the illusion of an enormous amount of hatred for one person. What is disturbing about this is that it is impossible for the person who is being harassed to know if they are really experiencing a massive amount of hate from many people or simply seeing it because of artificial means.

TEST

changer ceci

test

A ordem de pagamentos deve seguir a ordem cronológica, considerando cada fonte de recursos e cada categoria de contrato.

No entanto, poderá haver alteração da ordem de pagamento em detrimento da ordem cronológica. Para tanto, deve haver justificativa da autoridade competente e <u>posterior</u> comunicação aos órgãos de controle interno <u>e</u> tribunal de contas.

Embora se exija comunicação de tal alteração aos órgãos de controle interno e externo, a referida comunicação, observe, é posterior, não sendo necessária prévia anuência daqueles órgãos para a efetivação da alteração.

1. 内联（Inline）：打破物理边界的利器与代价 既然 HLS 把子函数当黑盒，这就带来一个问题：它无法做全局优化。 比如，如果在小黑盒的末尾有一个乘法，在大黑盒里拿到结果后马上跟了一个加法。因为有函数边界的阻挡，HLS 可能会强行在这两个操作中间插入寄存器（断开操作链接）。

这时候，inline（内联指令） 就派上用场了。 如图 2.11 所示，一旦你在子函数上加了 inline：

HLS 会直接抹除函数的物理边界。

所有的子逻辑全部暴露给顶层，融合成一坨巨大的组合逻辑。

好处： HLS 获得了“上帝视角”，它可以跨界进行资源共享和操作链接（Operation Chaining），从而缩短延迟（Latency），省下一些边界上的寄存器。

坏处与警告： 就像之前提到的“循环完全展开”一样，如果你的顶层逻辑极度复杂，你还把所有子模块都 inline 进来，这会生成一个无比庞大的数据流图。Vivado HLS 会被这海量的约束关系搞得内存溢出、耗费几个小时都综合不出来，甚至导致布线失败。

一句话总结： 层次化结构（不内联）是帮 EDA 工具减负，让代码结构模块化，但可能损失一点点跨界优化的性能；内联（Inline）是帮硬件提速，让逻辑融会贯通，但极容易让综合工具崩溃。合理的架构设计，就是在两者之间找平衡。

This isn't showing bold italics; although in my Quarto file it works :)

Statt bei der Beach Tennis WM habe ich die letzten 1,5 Std damit verbracht, dieses Ding zum Laufen zu bringen.

Etonnant que des élèves aussi brillants en principe puissent se montrer aussi ordinaire sur un sujet ou l'intelligence humaine est en jeu...

fermer la parenthèse

expliquer ce que c'est ?

adolescence

idem, référence à mettre en bleu ?

aussi bien à ?

la référence n'est pas en bleu

Peut-être dire un mot des situations non hétéronormées ?

est-ce qu'il manque un mot ici ?

virgule

Idem ? Il me semble que c'est là toute la difficulté : qu'a-t-on à proposer comme alternatives ? On pourrait aussi renvoyer au chapitre précédent, pour souligner la difficulté à s'assurer du consentement d'autrui

donner qq pistes de réponses possibles ?

donner deux ou trois exemples qui fonctionneraient mieux

attendus

Super ! je pense que c'est très utile de proposer ce type d'activités très concrètes. Peut-être à systématiser dans les autres chapitres ?

sociales

ce dont on a envie ? (avoir envie de)

An endolysin containing 15 distinct domains as highlighted in the abstract seems to be very unlikely, but it is not easy to identify the 15-domain lysin in the manuscript. The authors could use spaed.ca to identify and delineate the domains. Possibly multiple domains are in fact a single domain (something frequently observed in CBDs). The authors can also have a glance on phalp.ugent.be and make a comparison of the lysins found in the gut phageome and the 755k lysins found in this database, largely derived from the metagenomic EnVhogDB database. It would be interesting to know how the phageome differentiates in properties from the full metavirome (as we know it today) in terms of lysin diversity.

I think that social media should have a decent amount of responsibility in reducing individual and crowd harassment. I feel this way because harassment can be very damaging to individuals and create harmful outcomes. Social media platforms can increase their responsibility by having skilled moderators review content that includes harassment, or by having a system that is triggered by certain offensive words.

DOI: 10.7554/eLife.105911

Resource: (MMRRC Cat# 036778-UCD,RRID:MMRRC_036778-UCD)

Curator: @evieth

SciCrunch record: RRID:MMRRC_036778-UCD

What is this?

DOI: 10.7554/eLife.105911

Resource: RRID:IMSR_JAX:000664

Curator: @evieth

SciCrunch record: RRID:IMSR_JAX:000664

What is this?

DOI: 10.3389/fonc.2026.1819923

Resource: (ATCC Cat# TIB-71, RRID:CVCL_0493)

Curator: @evieth

SciCrunch record: RRID:CVCL_0493

What is this?

DOI: 10.3389/fonc.2026.1819923

Resource: (TKG Cat# TKG 0144, RRID:CVCL_F936)

Curator: @evieth

SciCrunch record: RRID:CVCL_F936

What is this?

DOI: 10.1210/jendso/bvag013

Resource: (Cell Signaling Technology Cat# 9271, RRID:AB_329825)

Curator: @evieth

SciCrunch record: RRID:AB_329825

What is this?

DOI: 10.1210/jendso/bvag013

Resource: RRID:AB_777802

Curator: @evieth

SciCrunch record: RRID:AB_777802

What is this?

DOI: 10.1210/jendso/bvag013

Resource: RRID:AB_11157733

Curator: @evieth

SciCrunch record: RRID:AB_11157733

What is this?

DOI: 10.1210/jendso/bvag013

Resource: (Cell Signaling Technology Cat# 2532, RRID:AB_330331)

Curator: @evieth

SciCrunch record: RRID:AB_330331

What is this?

DOI: 10.1210/jendso/bvag013

Resource: (Cell Signaling Technology Cat# 4185, RRID:AB_2169402)

Curator: @evieth

SciCrunch record: RRID:AB_2169402

What is this?

DOI: 10.1210/jendso/bvag013

Resource: (Cell Signaling Technology Cat# 9234, RRID:AB_2269803)

Curator: @evieth

SciCrunch record: RRID:AB_2269803

What is this?

DOI: 10.1158/2767-9764.CRC-25-0761

Resource: (KCB Cat# KCB 200970YJ, RRID:CVCL_0336)

Curator: @evieth

SciCrunch record: RRID:CVCL_0336

What is this?

DOI: 10.1126/sciimmunol.adz8360

Resource: FlowJo (RRID:SCR_008520)

Curator: @evieth

SciCrunch record: RRID:SCR_008520

What is this?

DOI: 10.1126/sciimmunol.adz8360

Resource: Cytek Aurora Spectral Analyzer (RRID:SCR_019826)

Curator: @evieth

SciCrunch record: RRID:SCR_019826

What is this?

DOI: 10.1038/s44319-026-00749-w

Resource: (KCB Cat# KCB 200970YJ, RRID:CVCL_0336)

Curator: @evieth

SciCrunch record: RRID:CVCL_0336

What is this?

DOI: 10.1038/s44319-026-00749-w

Resource: RRID:CVCL_1570

Curator: @evieth

SciCrunch record: RRID:CVCL_1570

What is this?

DOI: 10.1038/s44319-026-00749-w

Resource: RRID:CVCL_2192

Curator: @evieth

SciCrunch record: RRID:CVCL_2192

What is this?

DOI: 10.1038/s42003-026-09885-4

Resource: (IMSR Cat# JAX_031661,RRID:IMSR_JAX:031661)

Curator: @evieth

SciCrunch record: RRID:IMSR_JAX:031661

What is this?

DOI: 10.1038/s42003-026-09885-4

Resource: RRID:IMSR_JAX:007909

Curator: @evieth

SciCrunch record: RRID:IMSR_JAX:007909

What is this?

DOI: 10.1038/s41598-026-44822-9

Resource: (Cell Signaling Technology Cat# 4695, RRID:AB_390779)

Curator: @evieth

SciCrunch record: RRID:AB_390779

What is this?

DOI: 10.1038/s41598-026-44822-9

Resource: (Cell Signaling Technology Cat# 4691, RRID:AB_915783)

Curator: @evieth

SciCrunch record: RRID:AB_915783

What is this?

DOI: 10.1038/s41598-026-44822-9

Resource: RRID:AB_3738540

Curator: @evieth

SciCrunch record: RRID:AB_3738540

What is this?

DOI: 10.1038/s41598-026-44822-9

Resource: RRID:AB_3101815

Curator: @evieth

SciCrunch record: RRID:AB_3101815

What is this?

DOI: 10.1038/s41592-026-03057-2

Resource: (Coriell Cat# GM12878, RRID:CVCL_7526)

Curator: @evieth

SciCrunch record: RRID:CVCL_7526

What is this?

DOI: 10.1038/s41540-026-00697-0

Resource: (ICLC Cat# HTL06001, RRID:CVCL_1511)

Curator: @evieth

SciCrunch record: RRID:CVCL_1511

What is this?

DOI: 10.1038/s41467-026-70645-3

Resource: (Proteintech Cat# 22952-1-AP, RRID:AB_2783651)

Curator: @evieth

SciCrunch record: RRID:AB_2783651

What is this?

DOI: 10.1021/acs.nanolett.5c04791

Resource: Northwestern University NUANCE BioCryo Core Facility (RRID:SCR_021288)

Curator: @bandrow

SciCrunch record: RRID:SCR_021288

What is this?

DOI: 10.1021/acs.nanolett.5c04791

Resource: Northwestern University Keck Interdisciplinary Surface Science Core Facility (RRID:SCR_026360)

Curator: @bandrow

SciCrunch record: RRID:SCR_026360

What is this?

DOI: 10.1021/acs.nanolett.5c04791

Resource: University of Virginia School of Medicine Molecular Electron Microscopy Core Facility (RRID:SCR_019031)

Curator: @bandrow

SciCrunch record: RRID:SCR_019031

What is this?

DOI: 10.1016/j.xgen.2026.101190

Resource: RRID:Addgene_247027

Curator: @evieth

SciCrunch record: RRID:Addgene_247027

What is this?

DOI: 10.1016/j.xcrm.2026.102824

Resource: IMGT - the international ImMunoGeneTics information system (RRID:SCR_012780)

Curator: @nmaralla

SciCrunch record: RRID:SCR_012780

What is this?

DOI: 10.1016/j.xcrm.2026.102821

Resource: SHELX (RRID:SCR_014220)

Curator: @nmaralla

SciCrunch record: RRID:SCR_014220

What is this?

DOI: 10.1016/j.xcrm.2026.102821

Resource: SHELX (RRID:SCR_014220)

Curator: @nmaralla

SciCrunch record: RRID:SCR_014220

What is this?

DOI: 10.1016/j.xcrm.2026.102821

Resource: Gene Set Enrichment Analysis (RRID:SCR_003199)

Curator: @nmaralla

SciCrunch record: RRID:SCR_003199

What is this?

DOI: 10.1016/j.xcrm.2026.102821

Resource: Gene Ontology (RRID:SCR_002811)

Curator: @nmaralla

SciCrunch record: RRID:SCR_002811

What is this?

DOI: 10.1016/j.xcrm.2026.102821

Resource: clusterProfiler (RRID:SCR_016884)

Curator: @nmaralla

SciCrunch record: RRID:SCR_016884

What is this?

DOI: 10.1016/j.molcel.2026.04.021

Resource: RRID:Addgene_12251

Curator: @nmaralla

SciCrunch record: RRID:Addgene_12251

What is this?

DOI: 10.1016/j.molcel.2025.04.017

Resource: RRID:Addgene_60955

Curator: @evieth

SciCrunch record: RRID:Addgene_60955

What is this?

DOI: 10.1016/j.molcel.2025.04.017

Resource: RRID:Addgene_187246

Curator: @evieth

SciCrunch record: RRID:Addgene_187246

What is this?

DOI: 10.1016/j.molcel.2025.04.017

Resource: RRID:Addgene_83969

Curator: @evieth

SciCrunch record: RRID:Addgene_83969

What is this?

DOI: 10.1016/j.neuron.2026.04.037

Resource: LabChart (RRID:SCR_023643)

Curator: @nmaralla

SciCrunch record: RRID:SCR_023643

What is this?

DOI: 10.1016/j.cub.2026.04.059

Resource: RRID:Addgene_184284

Curator: @nmaralla

SciCrunch record: RRID:Addgene_184284

What is this?

DOI: 10.1016/j.cub.2026.04.059

Resource: RRID:IMSR_JAX:038515

Curator: @nmaralla

SciCrunch record: RRID:IMSR_JAX:038515

What is this?

DOI: 10.1016/j.cub.2026.04.059

Resource: (IMSR Cat# JAX_016962,RRID:IMSR_JAX:016962)

Curator: @nmaralla

SciCrunch record: RRID:IMSR_JAX:016962

What is this?

DOI: 10.1016/j.cub.2026.04.059

Resource: (IMSR Cat# JAX_016963,RRID:IMSR_JAX:016963)

Curator: @nmaralla

SciCrunch record: RRID:IMSR_JAX:016963

What is this?

DOI: 10.1016/j.cub.2026.04.059

Resource: RRID:Addgene_40753

Curator: @nmaralla

SciCrunch record: RRID:Addgene_40753

What is this?

DOI: 10.1016/j.chembiol.2026.04.008

Resource: (IMSR Cat# CRL_394,RRID:IMSR_CRL:394)

Curator: @nmaralla

SciCrunch record: RRID:IMSR_CRL:394

What is this?

DOI: 10.1016/j.chembiol.2026.04.008

Resource: (RRID:CVCL_0621)

Curator: @nmaralla

SciCrunch record: RRID:CVCL_0621

What is this?

DOI: 10.1016/j.chembiol.2026.04.008

Resource: (ATCC Cat# HTB-21, RRID:CVCL_1115)

Curator: @nmaralla

SciCrunch record: RRID:CVCL_1115

What is this?

DOI: 10.1016/j.chembiol.2026.04.008

Resource: (BCRC Cat# 60549, RRID:CVCL_0620)

Curator: @nmaralla

SciCrunch record: RRID:CVCL_0620

What is this?

DOI: 10.1016/j.chembiol.2026.04.008

Resource: (IMSR Cat# CRL_088,RRID:IMSR_CRL:088)

Curator: @nmaralla

SciCrunch record: RRID:IMSR_CRL:088

What is this?

DOI: 10.1016/j.chembiol.2026.04.008

Resource: (RRID:CVCL_0617)

Curator: @nmaralla

SciCrunch record: RRID:CVCL_0617

What is this?

DOI: 10.1016/j.chembiol.2026.04.008

Resource: (ATCC Cat# HTB-127, RRID:CVCL_0619)

Curator: @nmaralla

SciCrunch record: RRID:CVCL_0619

What is this?

DOI: 10.1016/j.chembiol.2026.04.008

Resource: (ATCC Cat# HTB-121, RRID:CVCL_2319)

Curator: @nmaralla

SciCrunch record: RRID:CVCL_2319

What is this?

DOI: 10.1016/j.chembiol.2026.04.008

Resource: (ATCC Cat# CRL-2329, RRID:CVCL_1254)

Curator: @nmaralla

SciCrunch record: RRID:CVCL_1254

What is this?

DOI: 10.1016/j.celrep.2026.117069

Resource: RStudio (RRID:SCR_000432)

Curator: @evieth

SciCrunch record: RRID:SCR_000432

What is this?

DOI: 10.1016/j.celrep.2026.117069

Resource: (V.A. Lennon, Mayo Clinic College of Medicine Cat# HuC/D_Lennon, RRID:AB_2813895)

Curator: @evieth

SciCrunch record: RRID:AB_2813895

What is this?

DOI: 10.1016/j.celrep.2026.117069

Resource: (Dagmar Wilhelm; The University of Melbourne Cat# Wilhelm_SRY, RRID:AB_2687957)

Curator: @evieth

SciCrunch record: RRID:AB_2687957

What is this?

DOI: 10.1016/j.celrep.2026.117069

Resource: (TransGenic Cat# KO610, RRID:AB_3696859)

Curator: @evieth

SciCrunch record: RRID:AB_3696859

What is this?

DOI: 10.1016/j.canlet.2026.218593

Resource: (KCB Cat# KCB 200942YJ, RRID:CVCL_0326)

Curator: @nmaralla

SciCrunch record: RRID:CVCL_0326

What is this?

DOI: 10.1016/j.canlet.2026.218593

Resource: (RRID:CVCL_0006)

Curator: @nmaralla

SciCrunch record: RRID:CVCL_0006

What is this?

DOI: 10.1007/s00018-026-06239-5

Resource: (ICLC Cat# HTL13002, RRID:CVCL_C302)

Curator: @evieth

SciCrunch record: RRID:CVCL_C302

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:Addgene_73914

Curator: @scibot

SciCrunch record: RRID:Addgene_73914

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_6812

Curator: @scibot

SciCrunch record: RRID:BDSC_6812

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_8546

Curator: @scibot

SciCrunch record: RRID:BDSC_8546

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_35544

Curator: @scibot

SciCrunch record: RRID:BDSC_35544

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_8547

Curator: @scibot

SciCrunch record: RRID:BDSC_8547

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_1776

Curator: @scibot

SciCrunch record: RRID:BDSC_1776

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_3605

Curator: @scibot

SciCrunch record: RRID:BDSC_3605

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_29445

Curator: @scibot

SciCrunch record: RRID:BDSC_29445

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_32556

Curator: @scibot

SciCrunch record: RRID:BDSC_32556

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_34609

Curator: @scibot

SciCrunch record: RRID:BDSC_34609

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_44581

Curator: @scibot

SciCrunch record: RRID:BDSC_44581

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_32236

Curator: @scibot

SciCrunch record: RRID:BDSC_32236

What is this?

DOI: 10.7554/eLife.111053

Resource: RRID:BDSC_35785

Curator: @scibot

SciCrunch record: RRID:BDSC_35785

What is this?

DOI: 10.7554/eLife.110277

Resource: nVoke System (RRID:SCR_023028)

Curator: @scibot

SciCrunch record: RRID:SCR_023028

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:Addgene_137003

Curator: @scibot

SciCrunch record: RRID:Addgene_137003

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:Addgene_198196

Curator: @scibot

SciCrunch record: RRID:Addgene_198196

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:Addgene_54677

Curator: @scibot

SciCrunch record: RRID:Addgene_54677

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:Addgene_198192

Curator: @scibot

SciCrunch record: RRID:Addgene_198192

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:SCR_027430

Curator: @scibot

SciCrunch record: RRID:SCR_027430

What is this?

DOI: 10.7554/eLife.110277

Resource: MATLAB (RRID:SCR_001622)

Curator: @scibot

SciCrunch record: RRID:SCR_001622

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:Addgene_51310

Curator: @scibot

SciCrunch record: RRID:Addgene_51310

What is this?

DOI: 10.7554/eLife.110277

Resource: (IMSR Cat# JAX_007004,RRID:IMSR_JAX:007004)

Curator: @scibot

SciCrunch record: RRID:IMSR_JAX:007004

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:Addgene_54568

Curator: @scibot

SciCrunch record: RRID:Addgene_54568

What is this?

DOI: 10.7554/eLife.110277

Resource: (IMSR Cat# JAX_024742,RRID:IMSR_JAX:024742)

Curator: @scibot

SciCrunch record: RRID:IMSR_JAX:024742

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:Addgene_54545

Curator: @scibot

SciCrunch record: RRID:Addgene_54545

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:Addgene_122373

Curator: @scibot

SciCrunch record: RRID:Addgene_122373

What is this?

DOI: 10.7554/eLife.110277

Resource: RRID:Addgene_85043

Curator: @scibot

SciCrunch record: RRID:Addgene_85043

What is this?

DOI: 10.7554/eLife.109502

Resource: GraphPad Prism (RRID:SCR_002798)

Curator: @scibot

SciCrunch record: RRID:SCR_002798

What is this?

DOI: 10.7554/eLife.109502

Resource: Neurolucida Explorer (RRID:SCR_017348)

Curator: @scibot

SciCrunch record: RRID:SCR_017348

What is this?

DOI: 10.7554/eLife.109502

Resource: Black Zen software (RRID:SCR_018163)

Curator: @scibot

SciCrunch record: RRID:SCR_018163

What is this?

DOI: 10.7554/eLife.109502

Resource: Neurolucida (RRID:SCR_001775)

Curator: @scibot

SciCrunch record: RRID:SCR_001775

What is this?

DOI: 10.7554/eLife.109502

Resource: pClamp (RRID:SCR_011323)

Curator: @scibot

SciCrunch record: RRID:SCR_011323

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 706-605-148, RRID:AB_2340476)

Curator: @scibot

SciCrunch record: RRID:AB_2340476

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 711-065-152, RRID:AB_2340593)

Curator: @scibot

SciCrunch record: RRID:AB_2340593

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 705-605-147, RRID:AB_2340437)

Curator: @scibot

SciCrunch record: RRID:AB_2340437

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 706-065-148, RRID:AB_2340451)

Curator: @scibot

SciCrunch record: RRID:AB_2340451

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 705-065-147, RRID:AB_2340397)

Curator: @scibot

SciCrunch record: RRID:AB_2340397

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 711-605-152, RRID:AB_2492288)

Curator: @scibot

SciCrunch record: RRID:AB_2492288

What is this?

DOI: 10.7554/eLife.109502

Resource: (Sigma-Aldrich Cat# e2886, RRID:AB_2620165)

Curator: @scibot

SciCrunch record: RRID:AB_2620165

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 706-295-148, RRID:AB_2340468)

Curator: @scibot

SciCrunch record: RRID:AB_2340468

What is this?

DOI: 10.7554/eLife.109502

Resource: (Thermo Fisher Scientific Cat# A48270, RRID:AB_2896336)

Curator: @scibot

SciCrunch record: RRID:AB_2896336

What is this?

DOI: 10.7554/eLife.109502

Resource: (Thermo Fisher Scientific Cat# A32790, RRID:AB_2762833)

Curator: @scibot

SciCrunch record: RRID:AB_2762833

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 712-295-153, RRID:AB_2340676)

Curator: @scibot

SciCrunch record: RRID:AB_2340676

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 711-545-152, RRID:AB_2313584)

Curator: @scibot

SciCrunch record: RRID:AB_2313584

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 711-295-152, RRID:AB_2340613)

Curator: @scibot

SciCrunch record: RRID:AB_2340613

What is this?

DOI: 10.7554/eLife.109502

Resource: (Thermo Fisher Scientific Cat# A32816, RRID:AB_2762839)

Curator: @scibot

SciCrunch record: RRID:AB_2762839

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 705-545-147, RRID:AB_2336933)

Curator: @scibot

SciCrunch record: RRID:AB_2336933

What is this?

DOI: 10.7554/eLife.109502

Resource: (Swant Cat# CB 38, RRID:AB_10000340)

Curator: @scibot

SciCrunch record: RRID:AB_10000340

What is this?

DOI: 10.7554/eLife.109502

Resource: (Jackson ImmunoResearch Labs Cat# 703-545-155, RRID:AB_2340375)

Curator: @scibot

SciCrunch record: RRID:AB_2340375

What is this?

DOI: 10.7554/eLife.109502

Resource: (Frontier Institute Cat# Homer1-Go, RRID:AB_2631104)

Curator: @scibot

SciCrunch record: RRID:AB_2631104

What is this?

DOI: 10.7554/eLife.109502

Resource: (Thermo Fisher Scientific Cat# A48258, RRID:AB_2890547)

Curator: @scibot

SciCrunch record: RRID:AB_2890547

What is this?

DOI: 10.7554/eLife.109502

Resource: (List Biological Cat# 703, RRID:AB_10013220)

Curator: @scibot

SciCrunch record: RRID:AB_10013220

What is this?

DOI: 10.7554/eLife.109502

Resource: RRID:AB_2864792

Curator: @scibot

SciCrunch record: RRID:AB_2864792

What is this?

DOI: 10.7554/eLife.109502

Resource: (Frontier Institute Cat# VGluT3-GP-Af920, RRID:AB_2571856)

Curator: @scibot

SciCrunch record: RRID:AB_2571856

What is this?

DOI: 10.7554/eLife.109502

Resource: (Peninsula Laboratories Cat# T-4103.0050, RRID:AB_518614)

Curator: @scibot

SciCrunch record: RRID:AB_518614

What is this?

DOI: 10.7554/eLife.109502

Resource: RRID:AB_518630

Curator: @scibot

SciCrunch record: RRID:AB_518630

What is this?