Channel-independent function of UNC-9/Innexin in spatial arrangement of GABAergic synapses in C. elegans

  1. Ardalan Hendi
  2. Long-Gang Niu
  3. Andrew William Snow
  4. Richard Ikegami
  5. Zhao-Wen Wang
  6. Kota Mizumoto

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This paper describes novel insights into the potential function of innexin proteins, which are electrical synapse-forming proteins with often quite enigmatic in vivo functions. The authors describe here potential functions in synapse tiling. The paper should be of interest to researchers with interests in molecular mechanisms governing nervous system development.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Precise synaptic connection of neurons with their targets is essential for the proper functioning of the nervous system. A plethora of signaling pathways act in concert to mediate the precise spatial arrangement of synaptic connections. Here we show a novel role for a gap junction protein in controlling tiled synaptic arrangement in the GABAergic motor neurons in Caenorhabditis elegans , in which their axons and synapses overlap minimally with their neighboring neurons within the same class. We found that while EGL-20/Wnt controls axonal tiling, their presynaptic tiling is mediated by a gap junction protein UNC-9/Innexin, that is localized at the presynaptic tiling border between neighboring dorsal D-type GABAergic motor neurons. Strikingly, the gap junction channel activity of UNC-9 is dispensable for its function in controlling tiled presynaptic patterning. While gap junctions are crucial for the proper functioning of the nervous system as channels, our finding uncovered the novel channel-independent role of UNC-9 in synapse patterning.

Article activity feed

  1. Version published to 10.7554/elife.80555 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    This is a very interesting paper. In this manuscript Hendi et al. examined how two independent mechanisms, Wnt signalling and gap junction control two critical aspects of neuronal tiling. Here they have quite elegantly used two neighboring GABAergic motor neurons to show while one specific C. elegans Wnt-homolog, EGL-20, regulates the axonal tiling; innexin UNC-9-mediated gap junction at a very specific position on these axons regulate the chemical synapse tiling on these axons. They also performed multiple experiments to show that the UNC-9 gap junctions controls chemical synapse tiling independent of their channel activity.

    Overall, the paper is interesting and would be of general interest for many neuroscience researchers, specifically to those who are studying neuronal tiling and the role of gap junctions. However, there are some concerns with this study.

    Major concerns:

    1. Authors here only looked at the tiling of axons and presynaptic clusters in DD5/DD6 axons. However, these neurites get transformed in L1 from dendrite to axon and subsequently the nature of the synaptic termini also changes from postsynaptic to presynaptic. To say that egl-20/UNC-9 specifically control axonal tiling and GABAergic presynaptic tiling the authors must check the dendritic tiling and tiling of postsynaptic termini. Specifically, a) does UNC-9 channels also affect the postsynaptic patterning in L1? b) what is the time of unc-9 puncta formation? Is it present in the L1 stage or appears at L2 stage only after the fate switch from dendrite to axon? c) does egl-20 also control dendritic tiling in L1?

    We thank the reviewer for their insightful comments. As described in our original manuscript, we could not check the dendritic tiling between DD5 and DD6 at L4 stage due to the inconsistent labeling of DD6 dendrite with our fluorescent marker. As an alternative method, we measured the length of the (ventral) posterior dendrite of DD5 and showed that it is significantly longer in the egl-20(n585) mutant than in wild type at L4 stage. We also measured the length of postsynaptic domains in the DD5 posterior dendrite and showed that it was also longer in the egl-20(n585) mutant than wild type. Furthermore, we show that the UNC-9 localization at the tip of DD6 dendrite is unaffected in the egl-20(n585) mutant, despite the extension of postsynaptic domains. From these observations, we suggested that postsynaptic spines are distributed throughout the dendrite of DD5 in the egl-20(n585) mutant, and it is not regulated by unc-9.

    In the revised manuscript, we included images of wild type and egl-20(n585) animal in which ACR-12::GFP is co-labeled with mCherry::CAAX. In these strains, the expression of mCherry::CAAX and ACR-12::GFP is not detectable in DD6 in most animals. Using these strains, we confirmed that the DD5 postsynaptic sites are present throughout the dendrite of DD5 in both wild type and egl-20(n585) mutant backgrounds (Figure 1- figure supplement 1).

    a) Unfortunately we were not able to quantify postsynaptic patterning at L1 due to the low expression of ACR-12::GFP and mCherry::CAAX at L1 stage.

    b) UNC-9::7×GFP puncta are present at the tiling border of DD neurons on both ventral and dorsal sides throughout the development. In the original manuscript, we only showed the UNC-9 localization at the dorsal side. We believe our limited description of UNC-9 in the dendrites has caused confusion regarding the phenotypes of DD5 posterior dendrite and postsynaptic sites. In the revised manuscript, we have updated the images of UNC-9::7×GFP to show that the puncta are present in both axons and dendrites (Figures 2F-H).

    In the revised manuscript we also show that UNC-9 puncta are present at DD tiling border in L1 animals. We have included images of UNC-9::7×GFP at L1 at the axonal and dendritic tiling borders of DD5 and DD6 in both wild type and egl-20(n585) animals in Figure 2- figure supplement 5.

    c) As described above, we could not quantify dendritic tiling at L1 due to the low expression of our fluorescent makers at the L1 stage.

    1. Authors have shown that the previously known regulators for gap junction formation, NLR-1 and ZOO-1, do not regulate UNC-9 gap junction puncta on DD5/DD6 axons. Since they are cell adhesion molecule and tight junction component, respectively, presynaptic tiling should be checked in these mutants as well. Also, it is not clear whether these proteins are expressed in DD5/DD6 neurons at all. Since, NLR-1 has previously been shown to regulate unc-9 puncta in nerve ring, expression of these genes in DD5/DD6-neurons should be checked before making these conclusions.

    In the revised manuscript, we have included the presynaptic tiling quantification in zoo-1(tm4133); egl-20(n585) and nlr-1(miz202) egl-20(n585) mutants which showed no significant presynaptic tiling defects (Figure 2- figure supplement 1). We also cited a paper (Taylor et al., 2021) that described the expression of zoo-1 and nlr-1 in the DD neurons.

    1. Authors assumed that the relevant gap junction to be an UNC-9 homotypic homomeric channel, but DD neurons also express several other innexins (inx-1, inx-2, inx-10, inx-14 and unc-7). This raises the possibility that unc-9 channel could be heteromeric in nature. Effect of some other expressed innexins on synaptic tiling apart from unc-7 should also be tested.

    We thank the reviewer for their comment. As per their advice, we tested four additional innexins (inx-1, inx-2, inx-10, and inx-14) which have been reported to be expressed in DD neurons and examined their potential role in presynaptic tiling in egl-20(n585) mutant background. We found that none of them showed significant presynaptic tiling defect. In the revised manuscript, we have included this data in Figure 2E.

    1. Effect of unc-9(Del18) and unc-1 double mutant should be tested.

    We knocked out unc-1 using CRISPR/Cas9 genome editing in the egl-20(n585); unc-9(syb3236 [unc-9(ΔN18)]) mutant background and observed no significant presynaptic tiling defect compared with egl-20(n585); unc-9(syb3236 [unc-9(ΔN18)]), which further strengthen our model that the gap junction channel activity of UNC-9 is dispensable for its function in presynaptic tiling. We have included this data in Figure 5D.

    1. Authors have acknowledged the need to study the role of UNC-9 gap junction channels in maintaining the presynaptic pattering. This reviewer appreciates that idea and suggests the authors check whether late expression of UNC-9 is sufficient to rescue the presynaptic pattering defect observed in egl-20; unc-9 double mutant animals.

    We thank the reviewer for their comment. We conducted late rescue experiment using a heat shock promoter to express unc-9 at L2 stage after the presynaptic tiling competes. We did not observe significant rescue in presynaptic tiling defect in two independent transgenic lines of Phsp::unc-9. While we understand that this does not deny the function of unc-9 for the maintenance of presynaptic tiling, this result is consistent with the idea that unc-9 is required for the establishment of presynaptic tiling. We have included this data in Figure 2- figure supplement 4.

    Reviewer #3 (Public Review):

    This interesting paper from Hendi et al. describes a novel mechanism governing synaptic tiling that depends on expression of a gap junction protein at the border between adjacent presynaptic domains of neighboring neurons. The authors define the role of innexin UNC-9 in establishing the spatial arrangement of synapses in adjacent C. elegans GABA motor neurons. They show that axonal tiling is controlled by Wnt signaling. However, synaptic tiling is preserved when axonal tiling is disrupted in egl-20/Wnt mutants. Synaptic and axonal tiling are both disrupted in egl-20; unc-9 double mutants, suggesting these two processes are controlled through distinct molecular mechanisms. The authors find that UNC-9 is localized to the border between axons of adjacent GABA neurons and provide evidence that the function of UNC-9 in tiling does not require its channel function. The experiments are made possible by the development of a new system for labeling adjacent GABA motor neurons that will also be of general use to the field. The studies rule out requirements for either gap junction activity or several other genes previously implicated in gap junction function/localization, but fall short of clearly defining mechanism. Instead, the study provides additional support for channel-independent structural roles of gap junctions in the nervous system.

    The study's conclusions are generally well-supported by the data but more clarification is required in some areas:

    1. Overlaps between DD5 and DD6 dendrites are not evaluated directly. The authors show the extent of labeling in the DD5 dendrite. This should be clarified.

    We thank the reviewer for their comment. As described above, we could not directly quantify dendritic tiling defect between DD5 and DD6 neurons due to the inconsistent expression of mCherry in the dendrite of DD6. Alternatively, we measured the length of DD5 posterior dendrite in wild type and the egl-20(n585) mutant, and found a significant increase in the DD5 posterior dendrite length in the egl-20(n585) mutants. In the revised manuscript, we have edited the text to more clearly explain the defect of DD5 posterior dendrite.

    1. The authors suggest UNC-9 establishes axonal tiling as early as L2 stage, immediately following DD remodeling. However, no data is shown for UNC-9 localization at this developmental stage. It would also be interesting to know whether UNC-9 performs a similar role prior to remodeling, or if UNC-9 itself undergoes redistribution during the remodeling process.

    We thank the reviewer for their comment. As described above, we acknowledge our initial description of UNC-9 localization in the DD neurons was not sufficient. UNC-9 is present at both the axonal and dendritic tiling borders between DD5 and DD6 neurons throughout the larval development.

    In the revised manuscript, we included UNC-9 localization at the axonal and dendritic tiling borders between DD5 and DD6 in both wild type and egl-20(n585) animals at the L1 stage (Figure 2- supplement figure 5). However, we could not determine whether egl-20(n585); unc-9(e101) mutant exhibits presynaptic patterning defect in the ventral axons prior to remodeling at the L1 stage due to the low expression of our axonal and presynaptic markers at L1 stage.

    1. Based on the representative image, UNC-9 abundance appears reduced in unc-104. The authors should quantify.

    We thank the reviewer for their comment. In the revised manuscript, we quantified the signal intensity of UNC-9::7×GFP at the DD5-DD6 axonal tiling border in wild type, egl-20(n585), unc-104(e1265), zoo-1(tm4133) and nlr-1(gk366849). We found that the fluorescent intensity of UNC-9::7×GFP was indeed slightly lower in egl-20(n585) and unc-104(e1265) mutants compared with wild type animals. This result implies that egl-20 and unc-104 have a minor role in UNC-9 localization. Nevertheless, the UNC-9 puncta are always present in all genotypes we examined. The quantification is included in Figure 2- figure supplement 6, and we suggest that the weak presynaptic tiling defect in the egl-20 single mutant could be explained by this reduction of UNC-9 localization (lines 284-285).

    1. The authors show the distribution of muscle NLG-1 mirrors that of RAB-3. While this suggests the altered distribution of RAB-3 reports on synaptic rearrangement, this conclusion would be strengthened by analysis of an active zone marker.

    We agree with the reviewer that examining the co-localization of RAB-3 with an active zone protein would strengthen our conclusion. As such, we expressed BFP::RAB-3 under the DD specific promoter, flp-13, in a transgenic marker strain (wyIs292) that expresses the active zone protein, UNC-10::tdTomato under the GABAergic promoter, unc-25, and NLG-1::YFP expressed under the body wall muscle promoter, unc-129dm (Maro et al., 2015). Using this strain, we show that RAB-3 co-localized with UNC-10 and apposed to the postsynaptic NLG-1 in both wild type and the egl-20(n585); unc-9(e101) mutant. The representative images are included in Figure 2- figure supplement 2.

  3. eLife

    Evaluation Summary:

    This paper describes novel insights into the potential function of innexin proteins, which are electrical synapse-forming proteins with often quite enigmatic in vivo functions. The authors describe here potential functions in synapse tiling. The paper should be of interest to researchers with interests in molecular mechanisms governing nervous system development.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    This is a very interesting paper. In this manuscript Hendi et al. examined how two independent mechanisms, Wnt signalling and gap junction control two critical aspects of neuronal tiling. Here they have quite elegantly used two neighboring GABAergic motor neurons to show while one specific C. elegans Wnt-homolog, EGL-20, regulates the axonal tiling; innexin UNC-9-mediated gap junction at a very specific position on these axons regulate the chemical synapse tiling on these axons. They also performed multiple experiments to show that the UNC-9 gap junctions controls chemical synapse tiling independent of their channel activity.

    Overall, the paper is interesting and would be of general interest for many neuroscience researchers, specifically to those who are studying neuronal tiling and the role of gap junctions. However, there are some concerns with this study.

    Major concerns:

    1. Authors here only looked at the tiling of axons and presynaptic clusters in DD5/DD6 axons. However, these neurites get transformed in L1 from dendrite to axon and subsequently the nature of the synaptic termini also changes from postsynaptic to presynaptic. To say that egl-20/UNC-9 specifically control axonal tiling and GABAergic presynaptic tiling the authors must check the dendritic tiling and tiling of postsynaptic termini. Specifically, a) does UNC-9 channels also affect the postsynaptic patterning in L1? b) what is the time of unc-9 puncta formation? Is it present in the L1 stage or appears at L2 stage only after the fate switch from dendrite to axon? c) does egl-20 also control dendritic tiling in L1?
    2. Authors have shown that the previously known regulators for gap junction formation, NLR-1 and ZOO-1, do not regulate UNC-9 gap junction puncta on DD5/DD6 axons. Since they are cell adhesion molecule and tight junction component, respectively, presynaptic tiling should be checked in these mutants as well. Also, it is not clear whether these proteins are expressed in DD5/DD6 neurons at all. Since, NLR-1 has previously been shown to regulate unc-9 puncta in nerve ring, expression of these genes in DD5/DD6-neurons should be checked before making these conclusions.
    3. Authors assumed that the relevant gap junction to be an UNC-9 homotypic homomeric channel, but DD neurons also express several other innexins (inx-1, inx-2, inx-10, inx-14 and unc-7). This raises the possibility that unc-9 channel could be heteromeric in nature. Effect of some other expressed innexins on synaptic tiling apart from unc-7 should also be tested.
    4. Effect of unc-9(Del18) and unc-1 double mutant should be tested.
    5. Authors have acknowledged the need to study the role of UNC-9 gap junction channels in maintaining the presynaptic pattering. This reviewer appreciates that idea and suggests the authors check whether late expression of UNC-9 is sufficient to rescue the presynaptic pattering defect observed in egl-20; unc-9 double mutant animals.

  5. eLife

    Reviewer #2 (Public Review):

    This work demonstrates a channel-independent function of a gap junction protein, UNC-9/innexin, and they show unc-9 is required for the tilling of presynaptic termini of DD5 and DD6, two neighboring C. elegans GABAergic neurons. This finding joins the list of growing evidence of the channel-independent function of gap junction proteins in neuronal development.

    Strengths:

    1. A new model system to study synapse tilling, an under-studied yet important aspect of neuronal development.
    2. Strong genetic evidence to support the novel function of UNC-9 in DD5-DD6 synapse tilling in egl-20/Wnt mutant background
    3. Careful characterization of how C-terminal GFP tagging and N-terminal 18 Aa affect UNC-9's channel functions.

    Weaknesses:
    One major weakness is that all analyses were carried out in DD5/DD6 neurons in egl-20/Wnt mutant background. It is unclear whether unc-9 plays the same role in wild type or other genetic background, and whether unc-9 can also affect synapse tilling in other neurons.

  6. eLife

    Reviewer #3 (Public Review):

    This interesting paper from Hendi et al. describes a novel mechanism governing synaptic tiling that depends on expression of a gap junction protein at the border between adjacent presynaptic domains of neighboring neurons. The authors define the role of innexin UNC-9 in establishing the spatial arrangement of synapses in adjacent C. elegans GABA motor neurons. They show that axonal tiling is controlled by Wnt signaling. However, synaptic tiling is preserved when axonal tiling is disrupted in egl-20/Wnt mutants. Synaptic and axonal tiling are both disrupted in egl-20;unc-9 double mutants, suggesting these two processes are controlled through distinct molecular mechanisms. The authors find that UNC-9 is localized to the border between axons of adjacent GABA neurons and provide evidence that the function of UNC-9 in tiling does not require its channel function. The experiments are made possible by the development of a new system for labeling adjacent GABA motor neurons that will also be of general use to the field. The studies rule out requirements for either gap junction activity or several other genes previously implicated in gap junction function/localization, but fall short of clearly defining mechanism. Instead, the study provides additional support for channel-independent structural roles of gap junctions in the nervous system.

    The study's conclusions are generally well-supported by the data but more clarification is required in some areas:

    1. Overlaps between DD5 and DD6 dendrites are not evaluated directly. The authors show the extent of labeling in the DD5 dendrite. This should be clarified.
    2. The authors suggest UNC-9 establishes axonal tiling as early as L2 stage, immediately following DD remodeling. However, no data is shown for UNC-9 localization at this developmental stage. It would also be interesting to know whether UNC-9 performs a similar role prior to remodeling, or if UNC-9 itself undergoes redistribution during the remodeling process.
    3. Based on the representative image, UNC-9 abundance appears reduced in unc-104. The authors should quantify.
    4. The authors show the distribution of muscle NLG-1 mirrors that of RAB-3. While this suggests the altered distribution of RAB-3 reports on synaptic rearrangement, this conclusion would be strengthened by analysis of an active zone marker.

  7. Version published to 10.1101/2022.05.16.492157v1 on bioRxiv