Tetraspanin CD82 reduces the formation of CADM1 oligomers

  1. Elisa Lamottke
  2. Harry Warner
  3. Sjoerd van Deventer
  4. Fabian Schwerdtfeger
  5. Linde Lanting
  6. Eric G. Huizinga
  7. Annemiek B. van Spriel
  8. Piet Gros

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

CD82 is a member of the tetraspanin protein superfamily and known as a metastasis suppressor. We identified cell adhesion molecule 1 (CADM1) as an interaction partner of CD82. CADM1 mediates cell adhesion by forming cis and trans oligomers that connect membranes. We show that CD82 reduces the formation of CADM1 oligomers when solubilized by detergent, on liposomes and in cellulo using Jurkat T-cells. Our data is consistent with a 1:1 complex of CD82 and CADM1 in cis that leaves the CADM1 trans -interaction site accessible. Cryo-electron microscopy of the CD82:CADM1 heterodimer suggests an interaction site between the large-extracellular loop of CD82 and an Ig-like domain of CADM1. Consistently, liposomes coupled with CADM1 ectodomain show reduced clustering when reconstituted with CD82. We hypothesize that CD82 may affect spacing of the transmembrane helices of CADM1, possibly by interacting with the extracellular Ig-like domains and hence disrupting CADM1 oligomerization between membranes.

Article activity feed

  1. Review Commons

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

    Learn more at Review Commons

    Reply to the reviewers

    We thank the reviewers for the constructive criticism. All issues are addressed in the following point-by-point response.

    We addressed a major issue of the reviewers by adding a domain-deletion study. This study shows that Ig-like domain 3 of CADM1 is responsible for the interaction. Thus, we now propose a clear mechanism for CD82 regulation of CADM1 adhesion function.

    Point-by-point description of the revisions

    Reviewer #1:

    Major Comments:

    • Using a variety of elegant and sophisticated studies, the authors clearly present data that strongly support their conclusions of the interaction between CADM1 and CD82. Additionally, they speculate on the specific interaction sites/mechanism through which CD82 may affect the spacing of CADM1, which is supported by cryo EM. However, they do not go on to demonstrate a physiological role for this inhibitory interaction.

    In this MS we identify an interaction between CADM1 and CD82. Though we find an inhibitory activity, indeed we do not present *in vivo *data supporting a physiological role.

    Our observation of CADM1-CD82 interaction in HEK cells and Jurkat Tcells, co-expression of the two proteins in several tissues (based on human protein atlas) and shared relevance in metastasis support a physiological role.

    Minor Comments:

    • In Figure 4, there seems to be a significant size difference between WT and CD82KO Jurkat cells. If this is the case, this needs to be described a bit more. Additionally, how does this potential size difference impact the quantification of the percentage space covered by CADM1 clusters? Also, for these experiments, it is more appropriate to use the mean of each independent replicate rather than simply counting each cell as an n. Including these data as super plots so the reader can better understand the spread and reproducibility of the data would be ideal. Best practices suggest that statistical analysis should be performed on data from independent experiments (Lord et al 2020 PMID: 32346721).

    With regard to the size difference, this may indeed represent a phenotype associated with CD82 knockout. A modest difference in cell size was observed, although this did not reach statistical significance (Reviewer Fig. A). As all quantitative analyses in Figure 4 were performed at the cell surface, cell volume was not a parameter used in the primary analysis. To provide a more complete visualisation of the cellular context, we additionally assessed 3D cell dimensions and used these measurements to examine whether the observed clustering phenotype was influenced by cell size. Restricting the analysis to cells with a Z-projected area of 100-150 µm² yielded the same trends in percentage space covered (Reviewer Fig. B), cluster density (Reviewer Fig. C), and cluster size (Reviewer Fig. D). These findings support the conclusion that the differences reported in Figure 4 are maintained independently of the modest differences in cell size observed. We have not included these additional analyses in the revised manuscript, as they do not alter the overall conclusions and may distract from the main message, particularly as the size difference could represent an indirect consequence of CD82 loss. We would, however, be happy to include these data should the editor feel they would strengthen the manuscript. In addition, we have replaced the representative image of the CD82 KO cell with one that more accurately reflects the modest size difference observed.

    We have now updated the graphs (visualizing the independent experiments) to be in-line with Lord et al. 2020.

    Reviewer #2:

    __ __ 2.1 Line 101-102- In MS analysis, Table 1, several common proteins are present in all three bands analyzed by MS including CD82. Authors should explain why CD82 is observed in 250 KDa band in relatively high abundance?

    The intensity of CD82 in the 250-kDa band is 30-50x lower than observed in the other two bands (90-100 and 100-110 kDa).

    The detection of some CD82 in the 250-kDa band likely originates from partial aggregation due to freeze thawing or temporary storage at 4 degrees. Reducing sample buffer was added to all samples but samples were not boiled and partial aggregation is visible in SDS-PAGE gel.

    2.2 Line 118-119, authors should make a better explanation why the lower MW CADM1 band is not observed in co-expression, especially that lower MW band is also absent in CADM1-Strep purification FigS2b? Although the migration pattern of CD82 and CADM1 is similar to glycosylated proteins authors should confirm this band doubling is due to glycosylation. It is crucial because a great portion of the manuscript relies on protein separation on SEC that can be greatly impacted by glycosylation state of the proteins.

    A lower band was observed in the SDS-PAGE gel in fig. S2b (Now Fig. S3b). We now clearly indicate the double band with a bracket instead of the single arrow.

    Regarding the N-glycosylation of CD82, Wang et al. (https://doi.org/10.1016/j.jprot.2011.11.013) demonstrate band shifts of the upper band of CD82 by side-directed mutagenesis.

    In case of CADM1, we tried to validate the bands with additional experiments not included in the manuscript. If the reviewer finds it necessary this figure can be included. We demonstrated that at least the upper band at 90 kDa purified with CD82-gfp-strep shifts to 45 kDa, the expected MW of CADM1, when mutating all N-glycosylation sites to Glu. However, this mutant showed low expression and low stability. To further confirm that the double-band pattern is caused by glycosylation, we treated CADM1(ECD) and CADM1(F42S, ECD) with O-glycosidase, neuroaminidase and PNGaseF. The SDS-PAGE gels show a double band before and a single band after digestion.

    With respect to N-glycan variants in SEC:

    We often observed narrow double peaks for CD82-strep purifications (Fig. S6A, S8B,G), likely caused by the two N-glycosylation variants. However, when fused to eGFP, we can no longer distinguish the two states on SEC (although we still see a double band on SDS-PAGE gel). Consistently, we observed single peaks for different constructs of CADM1 in SEC, while observing double bands on SDS-PAGE gel. Likely, the two variants migrate very close to each other and can only be separated when the added MW of glycans is relatively high. We therefore don’t think that the large shifts of peaks for CADM1 or CADM1-CD82 complex can be caused by N-glycosylation.

    2.3 Line 120-considering the lower yield of CADM1 in co-expression studies and presence of CD82 protein in MS analysis of band 3, it raises the question whether the peak eluted at 10 mL in F-SEC chromatogram of co-expressed proteins is from a legit CD82-CADM1 complex or it is simply CD82 oligomerization/interaction with other proteins. I would suggest authors purify GFP-CD82 and CADM individually and run the mixture over F-SEC to confirm complex formation rather than running co-purified protein. The same experiment should be done with CD82-CADM1mixture without fusing GFP to rule out the possible influence of GFP on dimerization.

    The detection of CD82 in the 250 kDa is likely due to aggregation (see 2.1).

    Indeed, based on SEC results resulting from our small-scale expressions (Fig. S2E) we cannot conclude the composition of the 10 mL peak. In our large-scale experiment, we put single fractions of the SEC on gel and only see bands of CADM1 and CD82 for earlier peaks (Fig.1A-B). From this, we conclude the 12.5 mL peak in fig. 1A is composed of CADM1 and CD82.

    Mixing two detergent-solubilized membrane proteins to analyze complex formation is not straight forward due to the possible interference of micelles.

    2.4 Figure 3. The color scheme used makes it really difficult to follow curves for different concentrations. I would suggest using distinct colors for each concentration.

    We changed the colors of graphs showing different concentrations in figure 2, figure 3, S7 and S8 to a better distinguishable color palette.

    2.5 274-275. Authors argue that the CADM clusters occupy more surface area in CD82KO cells compared to normal cells. From figure 4C. it seems that CD82KO cells are smaller in size than normal cells? Is that true? If that's the case, authors should provide evidence that the expression/surface area is the same for both normal and CD82KO cells before making the conclusion. If the expression level remains the same between two cell types but the size of KO cell is smaller, it is expected that CADM will occupy more surface area on KO cells. Consequently, the clusters are expected to be bigger. Therefore, the smaller cluster size in normal cells is not directly caused by CD82 but by smaller cell size due to CD82KO.

    With regard to the size difference, this may indeed represent a phenotype associated with CD82 knockout. A modest difference in cell size was observed, although this did not reach statistical significance (Reviewer Fig. A). As all quantitative analyses in Figure 4 were performed at the cell surface, cell volume was not a parameter used in the primary analysis. To provide a more complete visualisation of the cellular context, we additionally assessed 3D cell dimensions and used these measurements to examine whether the observed clustering phenotype was influenced by cell size. Restricting the analysis to cells with a Z-projected area of 100-150 µm² yielded the same trends in percentage space covered (Reviewer Fig. B), cluster density (Reviewer Fig. C), and cluster size (Reviewer Fig. D). These findings support the conclusion that the differences reported in Figure 4 are maintained independently of the modest differences in cell size observed. We have not included these additional analyses in the revised manuscript, as they do not alter the overall conclusions and may distract from the main message, particularly as the size difference could represent an indirect consequence of CD82 loss. We would, however, be happy to include these data should the editor feel they would strengthen the manuscript. In addition, we have replaced the representative image of the CD82 KO cell with one that more accurately reflects the modest size difference observed.

    2.6 Line 205-207. Authors show that CADM1(ECD) is not interacting with CD82 and conclude that interaction between transmembrane helices is required. It raises the question that how in cryoEM, authors could capture interaction between Ig domain of CADM1 and CD82 between two different micelles? Providing such details about experimental setup helps readers to adopt the method for structural determination of elusive complexes.

    For the cryo EM sample a complex of full-length CADM1 and CD82 was purified. The 2D classes we observe showing two connected micelles are likely caused by interaction of the CADM1 ectodomain, forming a complex of two dimers of CADM1 and CD82 in one micelle (CD82-CADM1)—(CADM1-CD82).

    Reviewer #3:

    Major comments:

    3.1 In Figure 1, the second co-purification is much less efficient than the first (1C vs 1A). Perhaps removing Figure 1C and the associated SEC trace (or moving to the supplement) would make for an easier-to-understand figure panel. I was initially confused about the low efficiency, and it raised some doubts while I was first interpreting this figure. The large-scale purification in 1F and 1G looks much more convincing than what is shown in 1C.

    We moved all small-scale experiments into a supplemental figure (S2) and truncated the text where possible to leave the main focus on the large-scale purification.

    3.2 I found the experiments in Figure 2 very hard to follow, and the accompanying text needs improvement before publication. Why was CADM1 ECD rather than full-length used for SEC-MALS? It seems like the authors are trying to claim that CD82 reduces CADM1 oligomerization based on a smaller shoulder peak at ~10-12 mL with the fusion construct compared to CADM1 alone. However, it looks like 10 mL is the void volume of this column- wouldn't another explanation be the CADM1 sample simply has a higher percentage of misfolded or aggregated protein, and it not necessarily an "oligomer"?

    We modified the text describing the results in Fig. 2 to improve readability.

    Performing SEC-MALLS with membrane proteins is not straight forward, due to the presence of empty micelles causing an unstable baseline and the possibility that micelles fuse or separate. Therefore, we only performed SEC-MALLS with soluble CADM1(ECD) to have a clear readout.

    The void volume of the column is at ca. 8.5 mL EV. The void peak is indeed barely visible as we use the protein directly after the SEC for purification without concentrating. In the SECs for protein purification the void peak at 8.5 mL is more visible (Fig. S3B,D). The same column has been used for both experiments connected to different systems (AKTA and HPLC).

    Further, we think the 10 mL peaks are complexes instead of aggregation because we see a shift to higher EV after dilution during the SEC after reinjection as expected for concentration dependent complex formation.

    3.3 The liposome assay is a clever way to assess if CD82 alters CADM1 oligomerization. However, details of how this assay was performed need clarification, and I am not convinced of the authors' conclusions given the current presentation. In the S4A, the authors show that CADM1 ECD His does not co-purify with CD82, but then CADM1 ECD His was used in liposome experiments to assess how CD82 changes clustering. Why was full-length CADM1 not used? For quantification, the average liposome radius is reported, but to me, a more robust quantification would be the percentage of liposome area contacting another liposome (i.e., ~35 nm distance between two liposomes), if the authors are trying to assess how CD82 affects CADM1 clustering. How does the average liposome radius reflect CADM1 clustering?

    Reconstituting both membrane proteins in full-length would indeed be the more native approach. However, we chose to anchor CADM1(ECD)-his to the surface of the liposome for several practical reasons. We used CADM1(ECD)-his because it gave us the possibility to “titrate” different CADM1 concentration against one sample of reconstituted liposomes. We cannot estimate how much CD82 is successfully reconstituted into the liposome. When we reconstitute full length CADM1 mixed with CD82 it was difficult to validate if both proteins and in which ratio they are reconstituted.

    We were trying to use cryo-EM pictures to quantify cluster sizes using the grey gradient but this proved difficult even using a single grid with consistent imaging settings because of the ice gradients over the grid and ice contamination. Observed clusters are 3-dimensional and we can only observe single liposomes at the very edge of it.

    The clusters we observed when looking and reconstituted liposomes looked very different from non-reconstituted liposomes. We added examples to figure S5. The clusters are overall less dense and liposomes cluster but not with a consistent distance as we observe it for WT CADM1.

    The DLS reads out the particle sizes and shows two peaks at 100 and 300 nm after adding CADM1(ECD)-His. We interpret this as CADM1 connecting micelles as observed on cryoEM images and therefore causing an increase in average particle size.

    3.4 The AlphaFold model shown in main text Figure 5 is not high confidence, and I suggest moving this to the supplement. To validate that the LEL of CD82 interacts with Ig1 of CADM1 as suggested by the EM density, the authors should perform mutagenesis or a domain swap with a different tetraspanin, and also binding assays with Ig1, Ig2, and Ig3 of CADM1 to validate this claim in the absence of high-resolution structural data.

    We removed the AlphaFold predictions from figure 5 and added it to the other predictions in Fig. S13. We did not see interactions of soluble CADM1(ECD) (Fig. S12) or single Ig domains (not included in MS) with purified CD82. Instead, we added a domain-truncation study in Fig. 5. The study shows the same level of co-purifications of CD82-StrepII3 and CADM1-His, CADM1ΔIg1 and CADM1Δ1-2 but strongly reduced co-purification when deleting all 3 Ig like domains. No CADM1-His constructs were detected in absence of CD82-StrepII3 and expression controls are shown in Fig. S12. The additional data made the overall interaction mechanism more clear and we rephrased it in the discussion.

    3.5 The EM processing workflow figure (S10) is missing many important details. FSC curve needs to be shown for the final map, and details like what was used for motion correction, CTF estimation, and how 31k particles became 45k particles (were these classes used as templates for picking?), and how many classes were used in heterogeneous refinement (and what those volumes looked like) should be included.

    We created a new processing figure. The figure includes the FSC curve and particle distribution for the finale volume. We added the first ab initio and heterogeneous refinement steps for each round of particle picking and marked volumes of which particles were selected for further processing. We indicated the number of selected and total particles for the final 2D classes and then indicated which group of particles was used and the percentage of particles in the resulting 3D volumes instead of indicating the particle number for each volume for better overview. We added what jobs were used for motion correction and CTF estimation.

    3.6 A short model figure at the end of the manuscript would be helpful to summarize the authors' findings.

    We added a figure displaying how CD82 hinders the tight formation of CADM1 multimers between membranes as Fig. 6.

    Minor comments:

    __ __3.7 Line 67: "In cancer, CADM1 can function as an oncogene as well as a tumor suppressor by regulating different signaling cascades, such as the Hippo pathway that promotes cell proliferation [27]." I found this sentence confusing- I think it is worth including another sentence or two explaining how CADM1 can have both tumor suppressor or promoter effects for readers that aren't familiar with CADM1 biology.

    We expended this part of the introduction to better explain the different functions of CADM1 and how it acts as a tumor suppressor and promotor simultaneously using examples:

    “Cell adhesion molecule 1 (CADM1) is a widely expressed adhesion molecule that has a dual function in cell signaling and adhesion and is involved in several biological functions. CADM1 overexpression promotes recruitment of T cells or mast cells in autoinflammatory diseases, like type 1 diabetes, asthma or neuroinflammatory diseases [24, 25, 26]. In most cancers, low CADM1 expression is associated with more aggressive tumors as it promotes apoptosis and inhibits proliferation [27].However, in T-cell lymphoma high CADM1 expression is associated with poor outcomes as its overexpression can lead to increased organ infiltration of tumors. The role of CADM1 in signaling and adhesion might explain its dual association with poor or good cancer prognosis and in both cases the change in expression level makes it a suitable therapeutic target and a biomarker [28, 29]. First in vivo studies using antibody-drug conjugates or chimeric-antigen receptors show a specific response against cancer cells and are considered promising candidates for clinical trials [30, 31, 32]. Considering these recent successes, it is crucial to understand the different functions and regulation of CADM1.”

    3.8 The introduction spends time discussing glycosylation sites on CD82 and CADM1. When I first read the intro, I thought glycosylation would be a focus of some of the experiments, but it wasn't. Instead, I think the paper would be stronger if you spent a bit more time in the intro talking about what is known about CADM1- as some examples, are there other interaction partners known to regulate its oligomerization? Therapeutic targeting is briefly mentioned, but has there been any success or progress in that area?

    The glycosylation explains why we consistently observe double bands in our SDS-PAGE gels and therefore needs to be mentioned in the introduction.

    We added more information about CADM1 as a cancer therapy target and recent developments in the introduction (see 3.6).

  2. Review Commons

    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

    Tetraspanin CD82 reduces the formation of CADM1 oligomers Lamottke et al. present data identifying CADM1 as a partner of the tetraspanin CD82. They then characterize this interaction using binding experiments, liposome assays, microscopy, and single-particle cryo-EM. Overall, I found this work to be an interesting read, and in general, the experiments were well explained and well controlled. Overall, nice work and an interesting finding! Below are my comments for improvements:

    Major comments:

    • In Figure 1, the second co-purification is much less efficient than the first (1C vs 1A). Perhaps removing Figure 1C and the associated SEC trace (or moving to the supplement) would make for an easier-to-understand figure panel. I was initially confused about the low efficiency, and it raised some doubts while I was first interpreting this figure. The large-scale purification in 1F and 1G looks much more convincing than what is shown in 1C
    • I found the experiments in Figure 2 very hard to follow, and the accompanying text needs improvement before publication. Why was CADM1 ECD rather than full-length used for SEC-MALS? It seems like the authors are trying to claim that CD82 reduces CADM1 oligomerization based on a smaller shoulder peak at ~10-12 mL with the fusion construct compared to CADM1 alone. However, it looks like 10 mL is the void volume of this column- wouldn't another explanation be the CADM1 sample simply has a higher percentage of misfolded or aggregated protein, and it not necessarily an "oligomer"?
    • The liposome assay is a clever way to assess if CD82 alters CADM1 oligomerization. However, details of how this assay was performed need clarification, and I am not convinced of the authors' conclusions given the current presentation. In the S4A, the authors show that CADM1 ECD His does not co-purify with CD82, but then CADM1 ECD His was used in liposome experiments to assess how CD82 changes clustering. Why was full-length CADM1 not used? For quantification, the average liposome radius is reported, but to me, a more robust quantification would be the percentage of liposome area contacting another liposome (i.e., ~35 nm distance between two liposomes), if the authors are trying to assess how CD82 affects CADM1 clustering. How does the average liposome radius reflect CADM1 clustering?
    • The AlphaFold model shown in main text Figure 5 is not high confidence, and I suggest moving this to the supplement. To validate that the LEL of CD82 interacts with Ig1 of CADM1 as suggested by the EM density, the authors should perform mutagenesis or a domain swap with a different tetraspanin, and also binding assays with Ig1, Ig2, and Ig3 of CADM1 to validate this claim in the absence of high-resolution structural data.
    • The EM processing workflow figure (S10) is missing many important details. FSC curve needs to be shown for the final map, and details like what was used for motion correction, CTF estimation, and how 31k particles became 45k particles (were these classes used as templates for picking?), and how many classes were used in heterogeneous refinement (and what those volumes looked like) should be included.
    • A short model figure at the end of the manuscript would be helpful to summarize the authors' findings.

    Minor comments:

    • Line 67: "In cancer, CADM1 can function as an oncogene as well as a tumor suppressor by regulating different signaling cascades, such as the Hippo pathway that promotes cell proliferation [27]." I found this sentence confusing- I think it is worth including another sentence or two explaining how CADM1 can have both tumor suppressor or promoter effects for readers that aren't familiar with CADM1 biology
    • The introduction spends time discussing glycosylation sites on CD82 and CADM1. When I first read the intro, I thought glycosylation would be a focus of some of the experiments, but it wasn't. Instead, I think the paper would be stronger if you spent a bit more time in the intro talking about what is known about CADM1- as some examples, are there other interaction partners known to regulate its oligomerization? Therapeutic targeting is briefly mentioned, but has there been any success or progress in that area?

    Significance

    This work reports a new direction interaction between CADM1 and CD82, which is interesting and signficant because relatively few direct interactions between tetraspanins and their partners are well characterized. The manuscript can be improved by relating this new interaction discovery back to how this may help CADM1 biology or developing new agents targeting CADM1, which I mention above in my comments.

  3. Review Commons

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

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    The manuscript by Lamottke et al. characterizes the interaction of CD82 with CADM1 protein. Although, the data provides important insight into how CD82 regulates CADM1 clustering, I have few questions that needs to be addressed to improve the manuscript for readers before acceptance of the paper.

    Line 101-102- In MS analysis, Table 1, several common proteins are present in all three bands analyzed by MS including CD82. Authors should explain why CD82 is observed in 250 KDa band in relatively high abundance?

    Line 118-119, authors should make a better explanation why the lower MW CADM1 band is not observed in co-expression, especially that lower MW band is also absent in CADM1-Strep purification FigS2b? Although the migration pattern of CD82 and CADM1 is similar to glycosylated proteins authors should confirm this band doubling is due to glycosylation. It is crucial because a great portion of the manuscript relies on protein separation on SEC that can be greatly impacted by glycosylation state of the proteins.

    Line 120-considering the lower yield of CADM1 in co-expression studies and presence of CD82 protein in MS analysis of band 3, it raises the question whether the peak eluted at 10 mL in F-SEC chromatogram of co-expressed proteins is from a legit CD82-CADM1 complex or it is simply CD82 oligomerization/interaction with other proteins. I would suggest authors purify GFP-CD82 and CADM individually and run the mixture over F-SEC to confirm complex formation rather than running co-purified protein. The same experiment should be done with CD820-CADM1mixture without fusing GFP to rule out the possible influence of GFP on dimerization.

    Figure 3. The color scheme used makes it really difficult to follow curves for different concentrations. I would suggest using distinct colors for each concentration.

    Line 274-275. Authors argue that the CADM clusters occupy more surface area in CD82KO cells compared to normal cells. From figure 4C. it seems that CD82KO cells are smaller in size than normal cells? Is that true? If that's the case, authors should provide evidence that the expression/surface area is the same for both normal and CD82KO cells before making the conclusion. If the expression level remains the same between two cell types but the size of KO cell is smaller, it is expected that CADM will occupy more surface area on KO cells. Consequently, the clusters are expected to be bigger. Therefore, the smaller cluster size in normal cells is not directly caused by CD82 but by smaller cell size due to CD82KO.

    Line 205-207. Authors show that CADM1(ECD) is not interacting with CD82 and conclude that interaction between transmembrane helices is required. It raises the question that how in cryoEM, authors could capture interaction between Ig domain of CADM1 and CD82 between two different micelles? Providing such details about experimental setup helps readers to adopt the method for structural determination of elusive complexes.

    Significance

    This solid study provides new insight into the interaction between CD82 and CADM1, how self-assembly and clustering is regulated using a complemenatry biochemical approaches. While the structural details remain wanting, the sound biochemical work lends confidence in the study's findings.

    The greater community interested in tetraspanin biochemistry will appreciate this work.

    We have experience understanding related plasma membrane remodeling proteins (from the Prom family) and their interactions with other binding partners.

  4. Review Commons

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

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Summary: In this study, the authors identify cell adhesion molecule 1 (CADM1) as an interaction partner of tetraspanin CD82. They used a combination of biochemical, cellular and structural studies to support this conclusion and identify an inhibitory effect of CD82 on CADM1 oligomerization.

    Major Comments: Using a variety of elegant and sophisticated studies, the authors clearly present data that strongly support their conclusions of the interaction between CADM1 and CD82. Additionally, they speculate on the specific interaction sites/mechanism through which CD82 may affect the spacing of CADM1, which is supported by cryo EM. However, they do not go on to demonstrate a physiological role for this inhibitory interaction.

    Minor Comments: In Figure 4, there seems to be a significant size difference between WT and CD82KO Jurkat cells. If this is the case, this needs to be described a bit more. Additionally, how does this potential size difference impact the quantification of the percentage space covered by CADM1 clusters? Also, for these experiments, it is more appropriate to use the mean of each independent replicate rather than simply counting each cell as an n. Including these data as super plots so the reader can better understand the spread and reproducibility of the data would be ideal. Best practices suggest that statistical analysis should be performed on data from independent experiments (Lord et al 2020 PMID: 32346721).

    Significance

    General comments useful to editors and readers: This is a well written manuscript with compelling data that supports the overall conclusion of a protein-protein interaction between CADM1 and CD82. The abstract could be strengthened by the addition of a bit more biological rationale for the presented work and some context of impact.

    Advance: Through compelling structural work, the authors establish an interaction between CADM1 and CD82 and suggest based on the cryo-EM that the interaction occurs between the CD82(LEL) and an Ig-like domain of CADM1. The manuscript does not provide a physiological impact of the identified interaction, as this was suggested to be outside the scope of the current work. As such this limits the overall advance in knowledge some to a thorough description of a protein-protein interaction that impacts protein clustering.

    Audience: This is largely a structural study evaluating protein-protein interactions and the impact of a membrane scaffold protein on the clustering of a cell adhesion molecule. Thus, it is well suited for structural biologists and cell biologists studying cell adhesion molecules, membrane scaffold proteins and their regulation.

    Describe your expertise: My expertise is in cell biology, so admittedly, some of the structural data was difficult from me to critically evaluate.

  5. Version published to 10.64898/2025.12.12.693935 on bioRxiv