Uncovering the electrical synapse proteome in retinal neurons via in vivo proximity labeling

  1. Stephan Tetenborg
  2. Eyad Shihabeddin
  3. Elizebeth Olive Akansha Manoj Kumar
  4. Crystal L Sigulinsky
  5. Karin Dedek
  6. Ya-Ping Lin
  7. Fabio A Echeverry
  8. Hannah Hoff
  9. Alberto E Pereda
  10. Bryan W Jones
  11. Christophe P Ribelayga
  12. Klaus Ebnet
  13. Ken Matsuura
  14. John O’Brien

  • Curated by eLife

    eLife logo

    eLife Assessment

    This is an important study that characterized proteins associated with electrical synapses in zebrafish and mouse retinal neurons using proximity labeling approaches, complemented by biochemical and histological validations. The resulting protein interactome datasets are convincing and reveal novel scaffold proteins at the electrical synapse. Additional quantification and validation would strengthen the work further.

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Abstract

Electrical synapses containing Connexin 36 (Cx36) represent the main means for direct electrical communication among neurons in the mammalian nervous system. However, little is known about the protein complexes that constitute these synapses. In the present study, we applied different BioID strategies to screen the interactomes of Connexin 36 and its zebrafish orthologue Cx35b in retinal neurons. For in vivo proximity labeling in mice, we took advantage of the Cx36-EGFP strain and expressed a GFP-nanobody-TurboID fusion construct selectively in AII amacrine cells. For in vivo BioID in zebrafish, we generated a transgenic line expressing a Cx35b-TurboID fusion under control of the Cx35b promoter. Both strategies allowed us to capture a plethora of molecules that were associated with electrical synapses and showed a high degree of evolutionary conservation in the proteomes of both species. Besides known interactors of Cx36 such as ZO-1 and ZO-2 we have identified more than 50 new proteins, such as scaffold proteins, adhesion molecules and regulators of the cytoskeleton. Moreover, we determined the subcellular localization of these proteins in mouse retina and tested potential binding interactions with Cx36. Amongst these new interactors, we identified signal induced proliferation associated 1 like 3 (Sipa1l3), a protein that has been implicated in cell junction formation and cell polarity, as a new scaffold of electrical synapses. Interestingly, Sipa1l3 was able to interact with ZO-1, ZO-2 and Cx36, suggesting a pivotal role in electrical synapse function. In summary, our study provides the first detailed view of the electrical synapse proteome in retinal neurons, which is likely to apply to electrical synapses elsewhere.

Article activity feed

  1. eLife

    eLife Assessment

    This is an important study that characterized proteins associated with electrical synapses in zebrafish and mouse retinal neurons using proximity labeling approaches, complemented by biochemical and histological validations. The resulting protein interactome datasets are convincing and reveal novel scaffold proteins at the electrical synapse. Additional quantification and validation would strengthen the work further.

  2. eLife

    Reviewer #1 (Public review):

    This study aims to identify the proteins that compose the electrical synapse, which are much less understood than those of the chemical synapse. Identifying these proteins is important to understand how synaptogenesis and conductance are regulated in these synapses.

    Using a proteomics approach, the authors identified more than 50 new proteins and used immunoprecipitation and immunostaining to validate their interaction of localization. One new protein, a scaffolding protein (Sipa1l3), shows particularly strong evidence of being an integral component of the electrical synapse. The function of Sipa1l3 remains to be determined.

    Another strength is the use of two different model organisms (zebrafish and mice) to determine which components are conserved across species. This approach also expands the utility of this work to benefit researchers working with both species.

    The methodology is robust and there is compelling evidence supporting the findings.

  3. eLife

    Reviewer #2 (Public review):

    Summary:

    This study aimed to uncover the protein composition and evolutionary conservation of electrical synapses in retinal neurons. The authors employed two complementary BioID approaches: expressing a Cx35b-TurboID fusion protein in zebrafish photoreceptors and using GFP-directed TurboID in Cx36-EGFP-labeled mouse AII amacrine cells. They identified conserved ZO proteins and endocytosis components in both species, along with over 50 novel proteins related to adhesion, cytoskeleton remodeling, membrane trafficking, and chemical synapses. Through a series of validation studies¬-including immunohistochemistry, in vitro interaction assays, and immunoprecipitation-they demonstrate that novel scaffold protein SIPA1L3 interacts with both Cx36 and ZO proteins at electrical synapse. Furthermore, they identify and localize proteins ZO-1, ZO-2, CGN, SIPA1L3, Syt4, SJ2BP, and BAI1 at AII/cone bipolar cell gap junctions.

    Strengths:

    The study demonstrates several significant strengths in both experimental design and validation approaches. First, the dual-species approach provides valuable insights into the evolutionary conservation of electrical synapse components across vertebrates. Second, the authors compare two different TurboID strategies in mice and demonstrate that the HKamac promoter and GFP-directed approach can successfully target the electrical synapse proteome of mouse AII amacrine cells. Third, they employed multiple complementary validation approaches-including retinal section immunohistochemistry, in vitro interaction assays, and immunoprecipitation-providing evidence supporting the presence and interaction of these proteins at electrical synapses.

    Weaknesses:

    The major weakness of this paper is the insufficient number of replicates in the proteomics datasets. The zebrafish datasets include only two biological replicates, while the mouse dataset has only one high-quality replicate. Due to the limited number of replicates, it is not possible to determine which enriched proteins are statistically significant.

    Additionally, the Neutravidin staining in the TurboID condition is not restricted to where Cx35 is expressed but is broadly distributed throughout the INL and IPL in the zebrafish retina (Figure 1B, bottom). Therefore, it is necessary to include NeutrAvidin staining in non-labeled retinas to verify whether the biotinylated proteins are specifically associated with Cx35 expression. Although the western blot results showed increased protein enrichment in the TurboID condition compared to non-labeled retinas, this does not confirm that the streptavidin pull-down proteins are associated with Cx35.

    Similarly, it is important to include NeutrAvidin staining in both TurboID and non-labeled conditions in the mouse retina to verify that the biotinylated proteins are specifically associated with gap junctions.

  4. eLife

    Reviewer #3 (Public review):

    Summary:

    This study by Tetenborg S et al. identifies proteins that are physically closely associated with gap junctions in retinal neurons of mice and zebrafish using BioID, a technique that labels and isolates proteins in proximal to a protein of interest. These proteins include scaffold proteins, adhesion molecules, chemical synapse proteins, components of the endocytic machinery, and cytoskeleton-associated proteins. Using a combination of genetic tools and meticulously executed immunostaining, the authors further verified the colocalizations of some of the identified proteins with connexin-positive gap junctions. The findings in this study highlight the complexity of gap junctions. Electrical synapses are abundant in the nervous system, yet their regulatory mechanisms are far less understood than those of chemical synapses. This work will provide valuable information for future studies aiming to elucidate the regulatory mechanisms essential for the function of neural circuits.

    Strengths:

    A key strength of this work is the identification of novel gap junction-associated proteins in AII amacrine cells and photoreceptors using BioID in combination with various genetic tools. The well-studied functions of gap junctions in these neurons will facilitate future research into the functions of the identified proteins in regulating electrical synapses.

    The authors have addressed my concerns in the revised manuscript.

  5. eLife

    Author response:

    The following is the authors’ response to the original reviews

    Public Reviews:

    Reviewer #1 (Public review):

    Summary:

    This study aims to identify the proteins that compose the electrical synapse, which are much less understood than those of the chemical synapse. Identifying these proteins is important to understand how synaptogenesis and conductance are regulated in these synapses. The authors identified more than 50 new proteins and used immunoprecipitation and immunostaining to validate their interaction of localization. One new protein, a scaffolding protein, shows particularly strong evidence of being an integral component of the electrical synapse. However, many key experimental details are missing (e.g. mass spectrometry), making it difficult to assess the strength of the evidence.

    Strengths:

    One newly identified protein, SIPA1L3, has been validated both by immunoprecipitation and immunohistochemistry. The localization at the electrical synapse is very striking.
    A large number of candidate interacting proteins were validated with immunostaining in vivo or in vitro.

    Weaknesses:

    There is no systematic comparison between the zebrafish and mouse proteome. The claim that there is "a high degree of evolutionary conservation" was not substantiated.

    We have added a table as supplementary figure 3 that shows a comparison of all candidates. While there are differences in both proteomes, components such as ZO proteins and the endocytosis machinery are clearly conserved.

    No description of how mass spectrometry was done and what type of validation was done.

    We have contacted the mass spec facility we worked with and added a paragraph explaining the mass spec. procedure in the material and methods section.

    The threshold for enrichment seems arbitrary.

    Yes, the thresholds are somewhat arbitrary. This is due to the fact that experiments that captured larger total amounts of protein (mouse retina samples) had higher signal-to-noise ratio than those that captured smaller total amounts of protein (zebrafish retina). This allowed us to use a more stringent threshold in the mouse dataset to focus on high probability captured proteins.

    Inconsistent nomenclature and punctuation usage.

    We have scanned through the manuscript and updated terms that were used inconsistently in the interim revision of the manuscript.

    The description of figures is very sparse and error-prone (e.g. Figure 6).

    In Figure 1B, there is very broad non-specific labeling by avidin in zebrafish (In contrast to the more specific avidin binding in mice, Figure 2B). How are the authors certain that the enrichment is specific at the electrical synapse?

    The enrichment of the proteins we identified is specific for electrical synapses because we compared the abundance of all candidates between Cx35b-V5-TurboID and wildtype retinas. Proteins that are components of electrical synapses, will only show up in the Cx35b-V5-TurboID condition. The western blot (Strep-HRP) in figure 1C shows the differences in the streptavidin labeling and hence the enrichment of proteins that are part of electrical synapses. Moreover, while the background appears to be quite abundant in sections, biotinylation is a rare posttranslational modification and mainly occurs in carboxylases: The two intense bands that show up above 50 and 75 kDa. The background mainly originates from these two proteins. Therefore, it is easy to distinguish specific hits from non-specific background.

    In Figure 1E, there is very little colocalization between Cx35 and Cx34.7. More quantification is needed to show that it is indeed "frequently associated."

    We agree that “frequently associated” is too strong as a statement. We corrected this and instead wrote “that Cx34.7 was only expressed in the outer plexiform layer (OPL) where it was associated with Cx35b at some gap junctions” in line 151. There are many gap junctions at which Cx35b is not colocalized with Cx34.7.

    Expression of GFP in HCs would potentially be an issue, since GFP is fused to Cx36 (regardless of whether HC expresses Cx36 endogenously) and V5-TurboID-dGBP can bind to GFP and biotinylate any adjacent protein.

    Thank you for this suggestion! There should be no Cx36-GFP expression in horizontal cells, which means that the nanobody cannot bind to anything in these cells. Moreover, to recognize specific signals from non-specific background, we included wild type retinas throughout the entire experiments. This condition controls for non-specific biotinylation.

    Figure 7: the description does not match up with the figure regarding ZO-1 and ZO-2.

    It appears that a portion of the figure legend was left out of the submitted version of the manuscript. We have put the legend for panels A through C back into the manuscript in the interim revision.

    Reviewer #2 (Public review):

    Summary:

    This study aimed to uncover the protein composition and evolutionary conservation of electrical synapses in retinal neurons. The authors employed two complementary BioID approaches: expressing a Cx35b-TurboID fusion protein in zebrafish photoreceptors and using GFP-directed TurboID in Cx36-EGFP-labeled mouse AII amacrine cells. They identified conserved ZO proteins and endocytosis components in both species, along with over 50 novel proteins related to adhesion, cytoskeleton remodeling, membrane trafficking, and chemical synapses. Through a series of validation studies¬-including immunohistochemistry, in vitro interaction assays, and immunoprecipitation - they demonstrate that novel scaffold protein SIPA1L3 interacts with both Cx36 and ZO proteins at electrical synapse. Furthermore, they identify and localize proteins ZO-1, ZO-2, CGN, SIPA1L3, Syt4, SJ2BP, and BAI1 at AII/cone bipolar cell gap junctions.

    Strengths:

    The study demonstrates several significant strengths in both experimental design and validation approaches. First, the dual-species approach provides valuable insights into the evolutionary conservation of electrical synapse components across vertebrates. Second, the authors compare two different TurboID strategies in mice and demonstrate that the HKamac promoter and GFP-directed approach can successfully target the electrical synapse proteome of mouse AII amacrine cells. Third, they employed multiple complementary validation approaches - including retinal section immunohistochemistry, in vitro interaction assays, and immunoprecipitation-providing evidence supporting the presence and interaction of these proteins at electrical synapses.

    Weaknesses:

    The conclusions of this paper are supported by data; however, some aspects of the quantitative proteomics analysis require clarification and more detailed documented. The differential threshold criteria (>3 log2 fold for mouse vs >1 log2 fold for zebrafish) will benefit from biological justification, particularly given the cross-species comparison. Additionally, providing details on the number of biological or technical replicates used in this study, along with analyses of how these replicates compare to each other, would strengthen the confidence in the identification of candidate proteins. Furthermore, including negative controls for the histological validation of proteins interacting with Cx36 could increase the reliability of the staining results.

    While the study successfully characterized the presence of candidate proteins at the electrical synapses between AII amacrine cells and cone bipolar cells, it did not compare protein compositions between the different types of electrical synapses within the circuit. Given that AII amacrine cells form both homologous (AII-AII) and heterologous (AII-cone bipolar cell) electrical synapses-connections that serve distinct functional roles in retinal signaling processing-a comparative analysis of their molecular compositions could have provided important insights into synapse specificity.

    Reviewer #3 (Public review):

    Summary:

    This study by Tetenborg S et al. identifies proteins that are physically closely associated with gap junctions in retinal neurons of mice and zebrafish using BioID, a technique that labels and isolates proteins proximal to a protein of interest. These proteins include scaffold proteins, adhesion molecules, chemical synapse proteins, components of the endocytic machinery, and cytoskeleton-associated proteins. Using a combination of genetic tools and meticulously executed immunostaining, the authors further verified the colocalizations of some of the identified proteins with connexin-positive gap junctions. The findings in this study highlight the complexity of gap junctions. Electrical synapses are abundant in the nervous system, yet their regulatory mechanisms are far less understood than those of chemical synapses. This work will provide valuable information for future studies aiming to elucidate the regulatory mechanisms essential for the function of neural circuits.

    Strengths:

    A key strength of this work is the identification of novel gap junction-associated proteins in AII amacrine cells and photoreceptors using BioID in combination with various genetic tools. The well-studied functions of gap junctions in these neurons will facilitate future research into the functions of the identified proteins in regulating electrical synapses.

    Thank you for these comments.

    Weaknesses:

    I do not see major weaknesses in this paper. A minor point is that, although the immunostaining in this study is beautifully executed, the quantification to verify the colocalization of the identified proteins with gap junctions is missing. In particular, endocytosis component proteins are abundant in the IPL, making it unclear whether their colocalization with gap junction is above chance level (e.g. EPS15l1, HIP1R, SNAP91, ITSN in Figure 3B).

    Recommendations for the authors:

    Reviewer #2 (Recommendations for the authors):

    (1) It would be helpful to include a comprehensive summary of the results from the quantitative proteomics analyses, such as the number of proteins detected in each species and the number of proteins associated with each GO term. Additionally, a clear figure or table highlighting the specific proteins conserved between zebrafish and mice would strengthen the evidence for evolutionary conservation of proteins at electrical synapses.

    We have added the raw data we received from our mass spec facility including a comparison of all the candidates for different species. Supplementary figure 3.

    (2) A more detailed description of the number of experimental and/or technical replicates would improve the technical rigor of the study. For example, what was the rationale for using different log2 fold-change cutoffs in mice versus zebrafish? Are the replicates consistent in terms of protein enrichment?

    We have added raw data from individual experiments as a supplement (Excel spreadsheet). We have two replicates from zebrafish and two from mice. The first experiment in mice was conducted with fewer retinas and a different promoter (human synapsin promoter) and didn’t yield nearly as many candidates. We are currently running a third experiment with 35 mouse retinas which will most likely detect more candidates as we have identified currently. We can update the proteome in this paper once the analysis is complete. It is not feasible to conduct these experiments with multiple replicates at the same time, since the number of animals that have to be used is simply too high, especially since very specific genotypes are required that are difficult obtain.

    (3) It would be interesting to determine whether there are differences in the presence of candidate proteins between AII-AII gap junctions and AII-cone bipolar cell gap junctions. Given that the subcellular localization of AII-AII gap junctions differs from that of AII-cone bipolar cell gap junctions (with most AII-AII gap junctions located below AII-cone ones), histological validations of the proteins shown in Figure 6 can be repeated for AII-AII gap junctions. This would help reveal similarities or differences in the protein compositions of these two types of gap junctions.

    Thank you for this suggestion. We had similar plans. However, we realized that homologous gap junctions are difficult to recognize with GFP. The dense GFP labeling in the proximal IPL, where AII-AII gap junctions are formed, does not allow us to clearly trace the location of individual dendrites from different cells. Detecting AII-AII gap junctions would require intracellular dye Injections of neighboring AII cells. Unfortunately, we don’t have a set up that would allow this. Bipolar cell terminals, on the contrary, are a lot easier to detect with markers such as SCGN, which is why we decided to focus on AII/ONCB gap junctions.

    (4) In Figures 1 and 2, it would be helpful to clarify in the figure legends whether the proteins in the interaction networks represent all detected proteins or only those selected based on log2 fold-change or other criteria.

    Thank you for this suggestion! We have added a description in lines 643 and 662.

    (5) In Figure 1A (bottom panel), please include a negative control for the Neutravidin staining result from the non-labeling group.

    We only tested the biotinylation for wild type retinas in cell lysates and western blots as shown in figure 1C, which shows an entirely different biotinylation pattern.

    (6) In Figure 2B, please include the results of Neutravidin staining for both the labeling and non-labeling groups.

    Same comment: We see the differences in the biotinylation pattern on western blots, which is distinct for Cx36-EGFP and wild type retinas, although both genotypes were injected with the same AAV construct and the same dose of biotin. We hope that this provides sufficient evidence for the specificity of our approach.

    (7) In Figure 5B, the sizes of multiple proteins detected by Western blotting are inconsistent and confusing. For example, the size of Cx36 in the "FLAG-SJ2BP" panel differs from that in the other three panels. Additionally, in the "Myc-SIPA1L3+" panel, the size of SIPA1l3 appears different between the input and IP conditions.

    Thank you for pointing this out! The differences in the molecular weight can be explained by dimerization. We have indicated the position of the dimer and the monomer bands with arrows. Especially, when larger amounts of Cx36 are coprecipitated Cx36 preferentially occurs as a dimer. This can also be seen in our previous publication:

    S. Tetenborg et al., Regulation of Cx36 trafficking through the early secretory pathway by COPII cargo receptors and Grasp55. Cellular and Molecular Life Sciences 81, 1-17 (2024). Figure 1D

    The band that occurs above 150kDa in the SIPA1L3 input is most likely a non-specific product. The specific band for SIPA1L3 can be seen in the IP sample, which has the appropriate molecular weight. We often see much better immuno reactivity for the protein of interest in IP samples, because the protein is concentrated in these experiments which facilitates its detection.

    (8) How specific are the antibodies used for validating the proteins in this study? Given that many proteins, such as EPS15l1, HIP1R, SNAP91, GPrin1, SJ2BP, Syt4, show broad distribution in the IPL (Figure 3B, 4A, 6D), it is important to validate the specificity of these antibodies. Additionally, including negative controls in the histological validation would strengthen the reliability of the results.

    We carefully selected the antibodies based on western blot data, that confirmed that each antibody detected an antigen of appropriate size. Moreover, the distribution of the proteins mentioned is consistent with function of each protein described in the literature. EPS15L1 and GPrin1 for instance are both membrane-associated, which is evident in Hek cells. Figure 5C.

    A true negative control would require KO tissue and we don’t think that this is feasible at this point.

    (9) In Figure 7F, the model could be improved by highlighting which components may be conserved between zebrafish and mice, as well as which components are conserved between the AII-AII junction and AII-cone bipolar cell junction?

    Thank you for this suggestion. However, we don’t think that this is necessary as our study primarily focuses on the AII amacrine cell.

    Currently we are unable to distinguish differences in the composition of AII-AII and AII-ONCB junctions as described above.

    (10) Are there any functional measurements that could support the conclusion that "loss of Cx36 resulted in a quantitative defect in the formation of electrical synapse density complex"?

    The loss of electrical synapse density proteins is shown by these immunostaining comparisons. Functional measurements necessarily depend on the function of the electrical synapse itself, which is gone in the case of the Cx36 KO. It is not clear that a different functional measurement can be devised.

    Reviewer #3 (Recommendations for the authors):

    (1) It would be very helpful if there were page and line numbers on the manuscript.

    Line and page numbers have been added.

    (2) Typos in the 3rd paragraph, the sentence 'which is triggered by the influx of Calcium though non-synaptic NMDA...'

    Should it read '... Calcium THROUGH non-synaptic NMDA'?

    We have corrected this typo.

    (3) Figure 1B: please add a description of the top panels, 'Cx36 S293'.

    A description of the top panels has been added to the figure legend in line. Line 639.

    (4) Figure 1C: what do the arrows indicate?

    We apologize for the confusion. The arrows in the western blot indicate the position of the Cx35-V5-TurboID construct, which can be detected with streptavidin-HRP and the V5 antibody. We have added a description for these arrows to the figure legend. See line 641.

    (5) Related to the point in the 'Weakness', there are some descriptions of how well some of the gap junction-associated proteins colocalize with Cx36 in immunostaining. For example, 'In comparison to the scaffold proteins, however, the colocalization of Cx36 with each of these endocytic components, was clearly less frequent and more heterogenous, which appears to reflect different stages in the life cycle of Cx36' and 'All of these proteins showed considerable colocalization with Cx36 in AII amacrine cell dendrites'. It would be nice to see quantification data to support these claims.

    Thank you for this suggestion. We have added a colocalization analysis to figure 3 (C & D). We quantified the colocalization for the endocytosis proteins Eps15l1 and Hip1r. This quantification included a flipped control to rule out random overlap. For both proteins we confirmed true colocalization (Figure 3D).

    (6) In Figure 5B, it would be helpful if there were arrows or some kind in western blottings to indicate which bands are supposed to be the targeted proteins.

    We have added arrows in IP samples to indicate bands representing the corresponding protein.

    (7) In the sentence including 'for the PBM of Cx36, as it is the case for ZO-1', what is PBM?

    The PBM means PDZ binding motif. We have added an explanation for this abbreviation in line 244.

    (8) Please add a description of the Cx35b promoter construct in the Method section.

    The Cx35b Promoter is a 6.5kb fragment. We will make the clone available via Addgene to ensure that all details of the clone can be accessed via snapgene or alternative software.

  6. Version published to 10.7554/elife.105935.2 on eLife
  7. Version published to 10.7554/elife.105935 on eLife
  9. Version published to 10.1101/2024.11.26.625481v5 on bioRxiv
  10. eLife

    eLife Assessment

    This study aims to identify the proteins that compose the electrical synapse, which are much less understood than those of the chemical synapse. The study is useful in terms of both method development and biological advances, as the authors identified more than 50 new proteins and used immunoprecipitation and immunostaining to validate their interaction. However, the current experimental data are considered incomplete, as many key experimental details are missing.

  11. eLife

    Reviewer #1 (Public review):

    Summary:

    This study aims to identify the proteins that compose the electrical synapse, which are much less understood than those of the chemical synapse. Identifying these proteins is important to understand how synaptogenesis and conductance are regulated in these synapses. The authors identified more than 50 new proteins and used immunoprecipitation and immunostaining to validate their interaction of localization. One new protein, a scaffolding protein, shows particularly strong evidence of being an integral component of the electrical synapse. However, many key experimental details are missing (e.g. mass spectrometry), making it difficult to assess the strength of the evidence.

    Strengths:

    One newly identified protein, SIPA1L3, has been validated both by immunoprecipitation and immunohistochemistry. The localization at the electrical synapse is very striking.
    A large number of candidate interacting proteins were validated with immunostaining in vivo or in vitro.

    Weaknesses:

    There is no systematic comparison between the zebrafish and mouse proteome. The claim that there is "a high degree of evolutionary conservation" was not substantiated.

    No description of how mass spectrometry was done and what type of validation was done.

    The threshold for enrichment seems arbitrary.

    Inconsistent nomenclature and punctuation usage.

    The description of figures is very sparse and error-prone (e.g. Figure 6).

    In Figure 1B, there is very broad non-specific labeling by avidin in zebrafish (In contrast to the more specific avidin binding in mice, Figure 2B). How are the authors certain that the enrichment is specific at the electrical synapse?

    In Figure 1E, there is very little colocalization between Cx35 and Cx34.7. More quantification is needed to show that it is indeed "frequently associated."

    Expression of GFP in HCs would potentially be an issue, since GFP is fused to Cx36 (regardless of whether HC expresses Cx36 endogenously) and V5-TurboID-dGBP can bind to GFP and biotinylate any adjacent protein.

    Figure 7: the description does not match up with the figure regarding ZO-1 and ZO-2.

  12. eLife

    Reviewer #2 (Public review):

    Summary:

    This study aimed to uncover the protein composition and evolutionary conservation of electrical synapses in retinal neurons. The authors employed two complementary BioID approaches: expressing a Cx35b-TurboID fusion protein in zebrafish photoreceptors and using GFP-directed TurboID in Cx36-EGFP-labeled mouse AII amacrine cells. They identified conserved ZO proteins and endocytosis components in both species, along with over 50 novel proteins related to adhesion, cytoskeleton remodeling, membrane trafficking, and chemical synapses. Through a series of validation studies¬-including immunohistochemistry, in vitro interaction assays, and immunoprecipitation - they demonstrate that novel scaffold protein SIPA1L3 interacts with both Cx36 and ZO proteins at electrical synapse. Furthermore, they identify and localize proteins ZO-1, ZO-2, CGN, SIPA1L3, Syt4, SJ2BP, and BAI1 at AII/cone bipolar cell gap junctions.

    Strengths:

    The study demonstrates several significant strengths in both experimental design and validation approaches. First, the dual-species approach provides valuable insights into the evolutionary conservation of electrical synapse components across vertebrates. Second, the authors compare two different TurboID strategies in mice and demonstrate that the HKamac promoter and GFP-directed approach can successfully target the electrical synapse proteome of mouse AII amacrine cells. Third, they employed multiple complementary validation approaches - including retinal section immunohistochemistry, in vitro interaction assays, and immunoprecipitation-providing evidence supporting the presence and interaction of these proteins at electrical synapses.

    Weaknesses:

    The conclusions of this paper are supported by data; however, some aspects of the quantitative proteomics analysis require clarification and more detailed documented. The differential threshold criteria (>3 log2 fold for mouse vs >1 log2 fold for zebrafish) will benefit from biological justification, particularly given the cross-species comparison. Additionally, providing details on the number of biological or technical replicates used in this study, along with analyses of how these replicates compare to each other, would strengthen the confidence in the identification of candidate proteins. Furthermore, including negative controls for the histological validation of proteins interacting with Cx36 could increase the reliability of the staining results.

    While the study successfully characterized the presence of candidate proteins at the electrical synapses between AII amacrine cells and cone bipolar cells, it did not compare protein compositions between the different types of electrical synapses within the circuit. Given that AII amacrine cells form both homologous (AII-AII) and heterologous (AII-cone bipolar cell) electrical synapses-connections that serve distinct functional roles in retinal signaling processing-a comparative analysis of their molecular compositions could have provided important insights into synapse specificity.

  13. eLife

    Reviewer #3 (Public review):

    Summary:

    This study by Tetenborg S et al. identifies proteins that are physically closely associated with gap junctions in retinal neurons of mice and zebrafish using BioID, a technique that labels and isolates proteins proximal to a protein of interest. These proteins include scaffold proteins, adhesion molecules, chemical synapse proteins, components of the endocytic machinery, and cytoskeleton-associated proteins. Using a combination of genetic tools and meticulously executed immunostaining, the authors further verified the colocalizations of some of the identified proteins with connexin-positive gap junctions. The findings in this study highlight the complexity of gap junctions. Electrical synapses are abundant in the nervous system, yet their regulatory mechanisms are far less understood than those of chemical synapses. This work will provide valuable information for future studies aiming to elucidate the regulatory mechanisms essential for the function of neural circuits.

    Strengths:

    A key strength of this work is the identification of novel gap junction-associated proteins in AII amacrine cells and photoreceptors using BioID in combination with various genetic tools. The well-studied functions of gap junctions in these neurons will facilitate future research into the functions of the identified proteins in regulating electrical synapses.

    Weaknesses:

    I do not see major weaknesses in this paper. A minor point is that, although the immunostaining in this study is beautifully executed, the quantification to verify the colocalization of the identified proteins with gap junctions is missing. In particular, endocytosis component proteins are abundant in the IPL, making it unclear whether their colocalization with gap junction is above chance level (e.g. EPS15l1, HIP1R, SNAP91, ITSN in Figure 3B).

  14. eLife

    Author response:

    Public Reviews:

    Reviewer #1 (Public review):

    Summary:

    This study aims to identify the proteins that compose the electrical synapse, which are much less understood than those of the chemical synapse. Identifying these proteins is important to understand how synaptogenesis and conductance are regulated in these synapses. The authors identified more than 50 new proteins and used immunoprecipitation and immunostaining to validate their interaction of localization. One new protein, a scaffolding protein, shows particularly strong evidence of being an integral component of the electrical synapse. However, many key experimental details are missing (e.g. mass spectrometry), making it difficult to assess the strength of the evidence.

    Strengths:

    One newly identified protein, SIPA1L3, has been validated both by immunoprecipitation and immunohistochemistry. The localization at the electrical synapse is very striking.
    A large number of candidate interacting proteins were validated with immunostaining in vivo or in vitro.

    Weaknesses:

    There is no systematic comparison between the zebrafish and mouse proteome. The claim that there is "a high degree of evolutionary conservation" was not substantiated.

    We agree that we should have included a comprehensive comparison of proteins captured in the different species. We are assembling this table and it will be included in the revised manuscript. There is, indeed, significant conservation of many of the proteins enriched in both species.

    No description of how mass spectrometry was done and what type of validation was done.

    Since the mass spec was outsourced to a core facility, we had not included methodological details. We have requested these and will include full details in the revised version of the manuscript. In terms of “validation,” enrichment of proteins at electrical synapses was determined based on capture relative to control samples (non-transgenic zebrafish retinas or non-transgenic mouse retinas infected with the dGBP-TurboID virus) captured and processed at the same time. Actual validations based on protein co-localization and pull-downs is the subject of the rest of the manuscript, and could only be done for a fraction of the identified proteins. This type of validation can be pursued in many future studies.

    The threshold for enrichment seems arbitrary.

    Yes, the thresholds are somewhat arbitrary. This is due to the fact that experiments that captured larger total amounts of protein (mouse retina samples) had higher signal-to-noise ratio than those that captured smaller total amounts of protein (zebrafish retina). This allowed us to use a more stringent threshold in the mouse dataset to focus on high probability captured proteins.

    Inconsistent nomenclature and punctuation usage.

    We have scanned through the manuscript and updated terms that were used inconsistently in the interim revision of the manuscript.

    To describe the mass spec procedure, we will get in touch with the mass spec facility and provide the details in the next round of submission.

    The description of figures is very sparse and error-prone (e.g. Figure 6).

    In Figure 1B, there is very broad non-specific labeling by avidin in zebrafish (In contrast to the more specific avidin binding in mice, Figure 2B). How are the authors certain that the enrichment is specific at the electrical synapse?

    The enrichment of the proteins we identified is specific for electrical synapses because we compared the abundance of all candidates between Cx35b-V5-TurboID and wildtype retinas. Proteins that are components of electrical synapses, will only show up in the Cx35b-V5-TurboID condition. The western blot (Strep-HRP) in figure 1C shows the differences in the streptavidin labeling and hence the enrichment of proteins that are part of electrical synapses. Moreover, while the background appears to be quite abundant in sections, biotinylation is a rare posttranslational modification and mainly occurs in carboxylases: The two intense bands that show up above 50 and 75 kDa. The background mainly originates from these two proteins.

    In Figure 1E, there is very little colocalization between Cx35 and Cx34.7. More quantification is needed to show that it is indeed "frequently associated."

    We agree that “frequently associated” is too strong as a statement. We corrected this and instead wrote “that Cx34.7 was only expressed in the outer plexiform layer (OPL) where it was associated with Cx35b at some gap junctions” in line 150. There are many gap junctions at which Cx35b is not colocalized with Cx34.7.

    Expression of GFP in HCs would potentially be an issue, since GFP is fused to Cx36 (regardless of whether HC expresses Cx36 endogenously) and V5-TurboID-dGBP can bind to GFP and biotinylate any adjacent protein.

    Thank you for this suggestion! There should be no Cx36-GFP expression in horizontal cells, which means that the nanobody cannot bind to anything in these cells. Moreover, to recognize specific signals from non-specific background, we included wild type retinas throughout the entire experiments. This condition controls for non-specific biotinylation.

    Figure 7: the description does not match up with the figure regarding ZO-1 and ZO-2.

    It appears that a portion of the figure legend was left out of the submitted version of the manuscript. We have put the legend for panels A through C back into the manuscript in the interim revision.

    Reviewer #2 (Public review):

    Summary:

    This study aimed to uncover the protein composition and evolutionary conservation of electrical synapses in retinal neurons. The authors employed two complementary BioID approaches: expressing a Cx35b-TurboID fusion protein in zebrafish photoreceptors and using GFP-directed TurboID in Cx36-EGFP-labeled mouse AII amacrine cells. They identified conserved ZO proteins and endocytosis components in both species, along with over 50 novel proteins related to adhesion, cytoskeleton remodeling, membrane trafficking, and chemical synapses. Through a series of validation studies¬-including immunohistochemistry, in vitro interaction assays, and immunoprecipitation - they demonstrate that novel scaffold protein SIPA1L3 interacts with both Cx36 and ZO proteins at electrical synapse. Furthermore, they identify and localize proteins ZO-1, ZO-2, CGN, SIPA1L3, Syt4, SJ2BP, and BAI1 at AII/cone bipolar cell gap junctions.

    Strengths:

    The study demonstrates several significant strengths in both experimental design and validation approaches. First, the dual-species approach provides valuable insights into the evolutionary conservation of electrical synapse components across vertebrates. Second, the authors compare two different TurboID strategies in mice and demonstrate that the HKamac promoter and GFP-directed approach can successfully target the electrical synapse proteome of mouse AII amacrine cells. Third, they employed multiple complementary validation approaches - including retinal section immunohistochemistry, in vitro interaction assays, and immunoprecipitation-providing evidence supporting the presence and interaction of these proteins at electrical synapses.

    Weaknesses:

    The conclusions of this paper are supported by data; however, some aspects of the quantitative proteomics analysis require clarification and more detailed documented. The differential threshold criteria (>3 log2 fold for mouse vs >1 log2 fold for zebrafish) will benefit from biological justification, particularly given the cross-species comparison. Additionally, providing details on the number of biological or technical replicates used in this study, along with analyses of how these replicates compare to each other, would strengthen the confidence in the identification of candidate proteins. Furthermore, including negative controls for the histological validation of proteins interacting with Cx36 could increase the reliability of the staining results.

    While the study successfully characterized the presence of candidate proteins at the electrical synapses between AII amacrine cells and cone bipolar cells, it did not compare protein compositions between the different types of electrical synapses within the circuit. Given that AII amacrine cells form both homologous (AII-AII) and heterologous (AII-cone bipolar cell) electrical synapses-connections that serve distinct functional roles in retinal signaling processing-a comparative analysis of their molecular compositions could have provided important insights into synapse specificity.

    Reviewer #3 (Public review):

    Summary:

    This study by Tetenborg S et al. identifies proteins that are physically closely associated with gap junctions in retinal neurons of mice and zebrafish using BioID, a technique that labels and isolates proteins proximal to a protein of interest. These proteins include scaffold proteins, adhesion molecules, chemical synapse proteins, components of the endocytic machinery, and cytoskeleton-associated proteins. Using a combination of genetic tools and meticulously executed immunostaining, the authors further verified the colocalizations of some of the identified proteins with connexin-positive gap junctions. The findings in this study highlight the complexity of gap junctions. Electrical synapses are abundant in the nervous system, yet their regulatory mechanisms are far less understood than those of chemical synapses. This work will provide valuable information for future studies aiming to elucidate the regulatory mechanisms essential for the function of neural circuits.

    Strengths:

    A key strength of this work is the identification of novel gap junction-associated proteins in AII amacrine cells and photoreceptors using BioID in combination with various genetic tools. The well-studied functions of gap junctions in these neurons will facilitate future research into the functions of the identified proteins in regulating electrical synapses.

    Thank you for these comments.

    Weaknesses:

    I do not see major weaknesses in this paper. A minor point is that, although the immunostaining in this study is beautifully executed, the quantification to verify the colocalization of the identified proteins with gap junctions is missing. In particular, endocytosis component proteins are abundant in the IPL, making it unclear whether their colocalization with gap junction is above chance level (e.g. EPS15l1, HIP1R, SNAP91, ITSN in Figure 3B).

  15. Version published to 10.1101/2024.11.26.625481v4 on bioRxiv
  16. Version published to 10.7554/elife.105935.1 on eLife
  17. Version published to 10.1101/2024.11.26.625481v3 on bioRxiv
  18. Version published to 10.1101/2024.11.26.625481v2 on bioRxiv
  19. Version published to 10.1101/2024.11.26.625481v1 on bioRxiv