Trans-synaptic linking of calcium channels and post-synaptic receptors at a dyadic synapse

Abstract

Polyadic synapses (where each active zone is juxtaposed with multiple post-synaptic targets) are found in many brain regions and are implicated in learning and memory, yet little is known about how they function. To address this question, we analyze C. elegans cholinergic (ACh) motor neurons, which form dyadic synapses with two targets (a body muscle and a GABAergic motor neuron). Decreasing ACh receptors in either target elicits a retrograde decrease in pre-synaptic type 2 voltage activated calcium channels (CaV2), thereby decreasing ACh release. By contrast, blocking GABA motor neuron differentiation (thereby eliminating one target) results in ectopic clustering of GABA_A receptors at ACh neuromuscular junctions, which elicits a retrograde increase in pre-synaptic strength. These results suggest that pre-synaptic CaV2 channels are linked to post-synaptic receptors in both targets, which allows each target to modify transmission to both targets.

Introduction

A great deal is known about how synapses form, their functional properties, and how they are modified by plasticity. For example, many recent studies suggest that pre-synaptic release sites and post-synaptic densities (PSDs) are aligned across synapses in nanocolumns (1-3). Trans-synaptic alignment is proposed to decrease the delay of post-synaptic responses, to increase the efficiency of post-synaptic receptor activation following SV fusion, and to mediate retrograde regulation of pre-synaptic release probability (1, 2, 4, 5). These effects are thought to be mediated by trans-synaptic tethering of active zone (AZ) proteins to post-synaptic receptor complexes (6-9). Several papers show that AZ molecules (e.g. Neurexin and α2δ2) immobilize receptors in PSDs (10-13). In C. elegans, post-synaptic GABA_A receptors recruit voltage-activated calcium channels (CaV) to pre-synaptic AZs (5). All of these studies focus on simple (one to one) synapses, where a single nerve terminal releases neurotransmitter on a single target; however, recent studies suggest that simple synapses account for just a fraction of the connections in many brain regions.

Large scale connectome studies are available for many brain regions, in many model systems. These studies have revealed that individual synaptic contacts often have multiple pre- and post-synaptic partners. In the mouse retina, cone bipolar axons form divergent synapses with 2, 3, and 4 post-synaptic partners (14). In the fly olfactory system, many cell types form divergent synapses with three or more post-synaptic partners (15). In the fly optic lamina, each photoreceptor active zone contacts four post-synaptic partners (16). Finally, in the fly mushroom body (MB), a majority of Kenyon cell (KC)-MB output neuron (MBON) synapses have a convergent structure where a single MBON post-synaptic site is apposed to two or more pre-synaptic KC partners (17). These results suggest that multiple contact synapses play significant roles in many circuits.

Despite their prevalence, few studies have analyzed the function and plasticity of multiple contact synapses. In general terms, convergent (i.e. many to one) synapses integrate multiple inputs on a single target. Divergent (i.e. one to many) synapses, on the other hand, broadcast synchronized, equal output to multiple targets. New divergent multiple contact synapses are formed in rat hippocampal LTP (18), rabbit eyeblink conditioning (19), and mouse auditory (20) and contextual (21) fear conditioning, suggesting that divergent synapse formation plays an important role in learning and memory. For these reasons, there is significant interest in determining how multiple contact synapses function. Many questions remain to be answered about divergent synapses. Are active zones linked to receptor clusters in all targets? Do all targets contribute to retrograde regulation of presynaptic strength? Can one target alter transmission to other targets? Do divergent synapses exhibit homeostatic plasticity?

To address these questions, we analyze a divergent dyadic synapse in the C. elegans locomotion circuit (Fig. 1A). C. elegans body muscles receive direct input from excitatory (cholinergic, hereafter ACh) and inhibitory (GABAergic, hereafter GABA) motor neurons (MNs) (22). Different MN classes provide input to ventral (ACh, VA and VB; GABA, VD) and dorsal (ACh, DA and DB; GABA, DD) body muscles (Fig. 1A). ACh MNs provide excitatory input to this circuit at dyadic (ACh-D/M) synapses (22) (Fig. 1B), where each ACh MN bouton contacts two post-synaptic partners, a muscle arm (a dendrite-like protrusion from a body muscle) and a spine-like protrusion from a D-type GABA MN dendrite (22-24). ACh-D/M synapses mediate contralateral inhibition, whereby ACh MNs innervating ventral muscles (VA and VB) simultaneously activate DD neurons (thereby relaxing dorsal muscles) (Fig. 1A) (22). Analogous connections couple-dorsal contractions (by DA and DB neurons) to ventral relaxation (by VD neurons). In this manner, ACh-D/M synapses facilitate sinusoidal locomotion (25). Here we compare the mechanisms controlling ACh-D/M transmission to muscles and GABA MNs.

Figure 1. Recording ACh-D/M transmission to GABA MNs.

(A-B) Diagrams illustrating the C. elegans locomotion circuit (A) and the structure of an ACh-D/M synapse (B) are shown. (C-F) Spontaneous mEPSCs and EPSCs evoked by photo-activation of ACh MNs were recorded from DD neurons. Representative traces (C), mean mEPSC amplitude (D), mean mEPSC rate (E), and mean evoked EPSC amplitude (F) are shown for the indicated genotypes. DD mEPSC amplitudes, mEPSC rates, and evoked EPSC amplitudes were all dramatically decreased in both acr-12(ok367) and unc-3(e151) COE mutants. These results confirm that the DD EPSCs correspond to post-responses occurring at ACh-D/M synapses. Sample sizes for each genotype are shown in each panel. Statistical differences were determined by one way ANOVA with Tukey’s multiple comparisons test. Values that differ significantly are indicated (****, p <0.0001). Error bars indicate SEM.

Results:

Recording transmission to GABA MNs at ACh-D/M synapses

In prior studies, ACh-D/M transmission to GABA MNs has been assayed by calcium imaging (23, 24, 26). Here we assess transmission to GABA MNs by recording excitatory post-synaptic currents (EPSCs) from DD neurons (Fig. 1C). Spontaneous miniature EPSCs (mEPSCs) and large EPSCs evoked by photo-activating ACh MNs were readily detected in these recordings (Fig. 1C). To confirm that these currents occur at ACh-D/M synapses, we repeated these recordings in acr-12 mutants (which lack a D neuron ACh receptor, AChR) (27, 28) and in unc-3 COE mutants (which lack a transcription factor required for ACh MN differentiation) (29). DD mEPSC amplitude and rate (Fig. 1C-E) and evoked EPSC amplitude (Fig. 1C,F) were significantly decreased in both acr-12 and unc-3 COE mutants. These results suggest that DD EPSCs correspond to post-synaptic responses at ACh-D/M synapses.

Pre-synaptic NRX-1/NRXN and Post-synaptic UNC-40/DCC cluster ACR-12 receptors at ACh-D/M post-synapses

Receptors are clustered at synapses by binding to synaptic scaffold proteins. To identify the scaffolds that cluster ACR-12 at ACh-D/M synapses, we analyzed the localization and function of endogenous ACR-12 receptors. Using a split GFP tagged allele, acr-12(nu782 GFP₁₁), we reconstituted endogenous ACR-12 fluorescence in VD dorsal cord neurites. Endogenous ACR-12 function was assessed by recording DD mEPSCs. An mEPSC corresponds to the current evoked by fusion of a single synaptic vesicle; consequently, DD mEPSC amplitude measures the function of endogenous ACR-12 receptors clustered at ACh-D/M post-synapses.

Prior work suggested that the adhesion molecule NRX-1/NRXN (acting in ACh MNs) is required to cluster transgene expressed ACR-12 at ACh-D/M synapses (23, 26). Consistent with these studies, endogenous ACR-12 levels at VD post-synapses (Fig. 2A,B) and DD mEPSC amplitudes (Fig. 2C,D) and rates (Fig. 2C,E) were significantly reduced in nrx-1 null mutants. By contrast, ACR-12 levels at VD post-synapses (Fig. 2A,B), DD mEPSC amplitudes (Fig. 2D), and DD mEPSC rates (Fig. 2E) were unaffected in nlg-1 mutants, which lack a post-synaptic NRX-1 ligand, consistent with prior work (23).

Figure 2. NRX-1/NRXN and UNC-40/DCC are required for ACR-12 clustering at ACh-D/M post-synapses.

(A-B) Post-synaptic ACR-12 fluorescence was analyzed in VD neurons. ACR-12(nu782 GFP₁₁) fluorescence was reconstituted by GFP_1-10 expression in GABA neurons. DA/DB pre-synaptic terminals in the dorsal cord were identified by mCherry-tagged UNC-57/Endophilin fluorescence. Representative images of ACR-12 puncta at VD post-synaptic elements (defined by the UNC-57 signal) (A) and mean post-synaptic ACR-12 puncta fluorescence (B) are shown in the indicated genotypes. (C-E) DD mEPSCs were recorded in the indicated genotypes. Representative traces (C), mean mEPSC amplitudes (D), and mean mEPSC rates are shown. ACR-12 post-synaptic levels in VD neurons, DD mEPSC amplitudes, and DD mEPSC rates were significantly reduced in nrx-1(wy1155) and unc-40(e271) mutants but were unaffected in nlg-1(ok259) mutants. The ACR-12 puncta intensity and mEPSC defects seen in unc-40 mutants were rescued by a transgene expressing UNC-40 in GABA neurons (GABA rescue). Sample sizes for each genotype are shown in each panel. Statistical differences were determined by one way ANOVA with Tukey’s multiple comparisons test. Values that differ significantly are indicated (ns, p >0.05; *, p <0.05; ***, p <0.001; ****, p <0.0001). Error bars indicate SEM. Scale bar indicates 5 μm.

We next asked if ACR-12 receptors are clustered by the scaffolds for muscle acetylcholine receptors (AChRs): LIN-2/CASK, FRM-3/FARP, and UNC-40/DCC (30, 31). ACR-12 levels at VD post-synapses (SI Appendix, Fig. S1A,B) and DD mEPSC amplitudes (SI Appendix, Fig. S1C,D) were unchanged in both frm-3 FARP and in lin-2 CASK mutants. DD mEPSC rate was unaffected in frm-3 FARP mutants but was significantly reduced in lin-2 CASK mutants (SI Appendix, Fig. S1E), potentially indicating decreased ACh release from ACh MNs in lin-2 mutants. Thus, unlike muscle AChRs, ACR-12 clustering at ACh-D/M synapses is not mediated by FRM-3/FARP and LIN-2/CASK.

Next, we asked if the netrin receptor UNC-40/DCC clusters ACR-12 at ACh-D/M synapses. A potential role for UNC-40/DCC is suggested by studies showing that UNC-40/DCC promotes synaptic clustering of body muscle AChRs (30) and GABA_A receptors (32). To confirm UNC-40/DCC’s requirement for muscle AChRs, we recorded muscle mEPSCs (SI Appendix, Fig. S2A). For these experiments, we used two unc-40 alleles: unc-40(null) and unc-40(mKO), a muscle specific knockout. The amplitude and rate of muscle mEPSCs were significantly reduced in both unc-40(null) and unc-40(mKO) mutants (SI Appendix, Fig. S2B,C), consistent with the decreased post-synaptic AChR levels seen in a prior study (30).

Next, we asked if UNC-40/DCC is required for ACR-12 synaptic clustering. Consistent with this idea, unc-40 mutants had dramatically reduced ACR-12 levels at VD post-synapses (Fig. 2A,B), decreased DD mEPSC amplitude (Fig. 2D), and decreased DD mEPSC rate (Fig. 2E). The post-synaptic ACR-12 level and mEPSC defects in unc-40 mutants were all rescued by a transgene expressing UNC-40 in GABA neurons (Fig. 2B,D,E). To confirm that decreased mEPSC amplitudes are caused by decreased post-synaptic AChR levels, we also analyzed ACR-12 puncta in DD neurons, finding that puncta intensity was significantly reduced in nrx-1 and unc-40 mutants but was unaffected in nlg-1 mutants (SI Appendix, Fig. S3A,B). Thus, analysis of endogenous ACR-12 receptor localization and function strongly support the idea that ACR-12 clustering at ACh-D/M synapses requires pre-synaptic NRX-1 and post-synaptic UNC-40/DCC.

Decreasing AChRs in either target cell decreases pre-synaptic UNC-2/CaV2 abundance

We recently showed that post-synaptic UNC-49/GABA_A receptors recruit UNC-2/CaV2 channels to GABA MN active zones (5). Prompted by these results, we asked if post-synaptic AChRs have similar retrograde effects at ACh-D/M synapses. UNC-2 protein levels at ACh MN active zones was assessed by reconstituting UNC-2(nu586 GFP₁₁) fluorescence in DA/DB neurons (Fig. 3). In parallel, UNC-2 function was assessed by recording voltage-activate calcium current from VB neurons (Fig. 4). The impact of muscle AChRs on pre-synaptic UNC-2 was assessed by analyzing lin-2 CASK mutants which have dramatically reduced post-synaptic AChR levels in body muscles (30, 31) but have no effect on ACR-12 synaptic clustering in DD/VD neurons (Fig. S1). For this analysis, we used two alleles, lin-2(null) and a muscle specific knockout lin-2(mKO). UNC-2/CaV2 fluorescence at DA/DB active zones (Fig. 3A-B) and UNC-2/CaV2 current in VB neurons (Fig. 4A-B) were significantly reduced in both lin-2 mutants. The effect of VD/DD AChRs was assessed by analyzing acr-12 mutants and unc-40(GABA specific knockout, GKO). The UNC-2/CaV2 signal at DA/DB active zones (Fig. 3C-D) was significantly reduced in both acr-12 and unc-40(GKO) mutants. Similarly, VB neuron UNC-2/CaV2 current was significantly reduced in acr-12 mutants (Fig. 4C-D). Larger decreases in UNC-2/CaV2 fluorescence (Fig. 3E-F) and current (Fig. 4E-F) were observed in unc-40 null mutants and in unc-29; acr-16; acr-12 triple mutants (hereafter AChR triple knockouts, ATKO), which reduce AChRs in both post-synaptic partners. These results suggest that AChRs in both targets act in a retrograde manner to increase presynaptic UNC-2/CaV2 levels at ACh-D/M synapses.

Figure 3. Pre-synaptic UNC-2/CaV2 levels at ACh-D/M synapses are decreased in mutants lacking post-synaptic AChRs.

Pre-synaptic UNC-2 fluorescence was analyzed in DA/DB MNs. UNC-2(nu586 GFP₁₁) fluorescence was reconstituted by expressing GFP_1-10 in DA/DB neurons (using the unc-129 promoter). DA/DB pre-synaptic terminals were identified by the UNC-57/mCherry signal. Representative images (A,C,E) and mean UNC-2 puncta intensity in DA/DB active zones (B,D,F) are shown in the indicated genotypes. UNC-2 levels in DA/DB active zones are significantly decreased in: (A-B) lin-2(syb1019) and lin-2(muscle specific KO, mKO) mutants, which lack muscle post-synaptic AChRs; (C-D) acr-12(ok367) and unc-40(GABA specific knockouts, GKO), which lack VD/DD AChRs; and (E-F) unc-40(e271 null) and unc-29;acr-16;acr-12 triple knockouts (ATKO), which lack AChRs in both targets. lin-2(flex) and unc-40(flox) are WT controls (lacking CRE expression) for lin-2(mKO) and unc-40(GKO), respectively. Sample sizes for each genotype are shown in each panel. Statistical differences were determined as follows: one way ANOVA with Tukey’s multiple comparisons test (B,D); Kruskal-Wallis test with Dunn’s multiple comparisons test (F). Values that differ significantly are indicated (**, p <0.01; ***, p <0.001; ****, p <0.0001). Error bars indicate SEM. Scale bar indicates 5 μm.

Figure 4. UNC-2/CaV2 currents in VB motor neuron are decreased in mutants lacking post-synaptic AChRs.

Voltage-activated UNC-2/CaV2 currents were recorded from VB motor neurons. Representative traces (at 0 mV) (A,C,E), and mean current amplitude (at 0 mV) (B,D,F) are shown in the indicated genotypes. UNC-2/CaV2 currents in VB neurons are significantly decreased in: (A-B) lin-2(syb1019) and lin-2(muscle specific KO, mKO) mutants, which lack muscle post-synaptic AChRs; (C-D) acr-12(ok367), which lack VD/DD AChRs; and (E-F) unc-40(e271) and unc-29;acr-16;acr-12 triple knockouts (ATKO), which lack AChRs in both post-synaptic targets. lin-2(flex) is the WT controls (lacking CRE expression) for lin-2(mKO). Sample sizes for each genotype are shown in each panel. Statistical differences were determined as follows: one way ANOVA with Tukey’s multiple comparisons test (B,F); Mann-Whitney test (D). Values that differ significantly are indicated (*, p <0.05; ****, p <0.0001). Error bars indicate SEM.

Next, we asked if other VB neuron ion channels are disrupted in unc-40 DCC mutants (SI Appendix, Fig. S4). We found that VB neuron EGL-19/CaV1 current (SI Appendix, Fig. S4A-B), input resistance (a measure of leak currents) (SI Appendix, Fig. S4C), resting membrane potential (RMP) (SI Appendix, Fig. S4D), and voltage-activated potassium currents (I_K) (SI Appendix, Fig. S4E-G) were unaffected in unc-40 mutants. These results suggest that the absence of post-synaptic AChRs did not broadly alter VB neuron ion channels and excitability.

ACh-D/M synaptic defects in unc-40 DCC mutants are not a consequence of early developmental defects.

How does UNC-40/DCC regulate ion channel clustering at ACh-D/M synapses? One possibility is that UNC-40/DCC is required for an earlier aspect of development that is required for subsequent ACh-D/M synapse formation. Consistent with this idea, UNC-40/DCC functions as a netrin receptor, which promotes diverse early aspects of neural development such as cell migration and axon guidance (33, 34). Body muscle post-synaptic elements occur along extensions of the plasma membrane (termed muscle arms) that extend into the ventral and dorsal nerve cords (22). UNC-40/DCC is also required for muscle arm extension to the nerve cords (35). These results suggest that ACh-D/M synaptic defects in unc-40 mutants could arise from earlier developmental defects.

To test this possibility, we asked when during development UNC-40/DCC is required for ACR-12 clustering at ACh-D/M synapses. For this analysis, we used heat inducible CRE expression (using the hsp-16.2 promoter) combined with a CRE-activated unc-40 allele, unc-40(nu856 FLEX Off) (described in the Methods). ACh-VD/M synapses in the dorsal cord are formed in stage 2 larvae (L2) (26). Consequently, we compared ACR-12 clustering in unc-40(nu856) adults with and without heat shock during the fourth larval stage (L4) (SI Appendix, Fig. S5). Like unc-40 null mutants, unc-40(nu856) adults without heat shock had significantly reduced ACR-12 clustering at VD post-synapses (SI Appendix, Fig. S5A-B). By contrast, after a 6 hour 30^oC heat shock during L4, unc-40(nu856) adults had significantly increased ACR-12 post-synaptic clustering in VD neurons compared to no heat shock controls (SI Appendix, Fig. S5A-B). These results demonstrate that UNC-40/DCC can function after axon outgrowth, after muscle arm migration, and after synapse formation to promote ACR-12 clustering at ACh-D/M synapses.

NRX-1 and UNC-10/RIM are also altered in unc-40 DCC mutants

To further investigate how UNC-40/DCC modifies the molecular composition of pre-synaptic terminals, we analyzed several other active zone proteins (Fig. 5). Pre-synaptic levels of NRX-1/NRXN, UNC-10/RIM, SYD-2/liprin-α, and RAB-3 and were analyzed using CRISPR alleles carrying fluorescent tags in the corresponding endogenous genes. These proteins are labelled at all synapses. To isolate the ACh-D/M signal, puncta fluorescence was quantified at DA/DB pre-synapses, identified by expressing mCherry tagged UNC-57/Endophilin in these neurons. Using this strategy, we found that NRX-1 and UNC-10/RIM levels at DA/DB synapses were significantly reduced in unc-40 mutants (Fig. 5A and B). By contrast, RAB-3 and SYD-2/liprin-a levels at DA/DB pre-synapses were unaltered in unc-40 mutants (Fig. 5C and D). These results suggest that a subset of AZ proteins is altered in unc-40 DCC mutants.

Figure 5. Analysis of other active zone proteins in mutants lacking post-synaptic AChRs.

The abundance of several AZ proteins was quantified in DA/DB dorsal cord axons of unc-40(e271) and unc-29; acr-16; acr-12 triple knockouts (ATKO), which lack AChRs in both post-synaptic targets. Each AZ protein was visualized using CRISPR alleles containing fluorescent protein tags in the corresponding endogenous gene. DA/DB pre-synaptic regions were identified by expressing mCherry-tagged UNC-57/Endophilin in these neurons (using the unc-129 promoter). Representative images of DA/DB pre-synapses (defined by the UNC-57 signal) (left) and mean AZ puncta intensity (right) are shown for NRX-1/NRXN (A), UNC-10/RIM (B), RAB-3 (C), and SYD-2/liprin-α (D). Statistical differences were determined as follows: one way ANOVA with Tukey’s multiple comparisons test (A); unpaired t test with Welch’s correction (B,D); Mann-Whitney test (C). Significant differences are indicated (ns, p >0.05; *, p <0.05; ***, p <0.001). Sample sizes for each genotype are shown in each panel. Error bars indicate SEM. Scale bar indicates 5 μm.

NRX-1/NRXN and UNC-10/RIM were previously shown to control pre-synaptic UNC-2/CaV2 levels (36, 37), consistent with NRXN and RIM effects in other organisms (38-40). Thus, changes in NRX-1 and UNC-10 could contribute to UNC-40/DCC’s impact on pre-synaptic UNC-2/CaV2 levels. We did several experiments to test this possibility. First, to confirm NRX-1’s role, we analyzed UNC-2 levels in nrx-1 mutants. Pre-synaptic UNC-2 levels in DA/DB neurons were significantly reduced (34% decrease) in nrx-1(wy1155) mutants (which lack all NRX-1 isoforms), while a smaller (albeit significant) decrease (20%) was seen in nrx-1(nu485) mutants (which lack only the long NRX-1α isoform) (SI Appendix, Fig. S6A-D). Second, we asked if NRX-1 controls post-synaptic UNC-40/DCC levels. Consistent with this idea, UNC-40 levels at ACh-VD post-synapses were significantly decreased in nrx-1(null) mutants but were unaffected in other mutants that have decreased post-synaptic AChRs (SI Appendix, Fig. S6E-F). This pattern of mutual interdependence suggests that synaptic NRX-1 and UNC-40 levels are coordinately controlled by trans-synaptic interactions. Third, we asked if UNC-40/DCC’s impact on pre-synaptic NRX-1 is mediated by changes in post-synaptic AChRs. To test this idea, we analyzed ATKO mutants and found that pre-synaptic NRX-1 levels were unaffected (Fig. 5A). Collectively, these results suggest that NRX-1 and UNC-40 could be physically linked across the synapse. This trans-synaptic interaction could be mediated by direct binding of NRX-1 to UNC-40, as has been reported for the corresponding human proteins (41). Alternatively, NRX-1 and UNC-40/DCC could be linked indirectly via associated proteins.

Blocking ACh release has no effect on UNC-2/CaV2 levels in ACh neurons

The decreased UNC-2/CaV2 levels observed in mutants lacking post-synaptic AChRs could represent a compensatory response triggered by decreased ACh transmission. This seems unlikely because the effect we observe (decreased presynaptic UNC-2/CaV2) is anti-homeostatic, i.e. it would exacerbate the transmission defect rather than mitigate it. Nonetheless, it remains possible that the UNC-2/CaV2 defect is a secondary consequence of decreased ACh transmission. To address this possibility, we analyzed UNC-2 levels in ACh MNs in mutants with decreased vesicular ACh transporter (UNC-17/VAChT) function (SI Appendix, Fig. S7). Neither UNC-2/CaV2 puncta intensity in DA/DB neurons (SI Appendix, Fig. S7A-B) nor VB neuron UNC-2/CaV2 current (SI Appendix, Fig. S7C-D) was significantly altered in unc-17 VAChT mutants. These results suggest that decreased ACh transmission is unlikely to account for the decreased pre-synaptic UNC-2/CaV2 levels seen in mutants lacking post-synaptic AChRs.

AChR effects on UNC-2/CaV2 are input specific

Decreased UNC-2/CaV2 levels in ACh neurons could reflect circuit wide changes triggered by inactivating post-synaptic ACh receptors. To address this possibility, we analyzed UNC-2/CaV2 levels in GABA MNs (SI Appendix, Fig. S8). We found that DD neuron UNC-2/CaV2 puncta intensities were unaltered in ATKO mutants (SI Appendix, Fig. S8A-B). These results suggest that post-synaptic AChRs selectively regulate the strength of ACh MN inputs, without altering input from GABA MNs.

ACh release is decreased in mutants lacking post-synaptic AChRs

If pre-synaptic UNC-2/CaV2 levels are decreased in post-synaptic AChR mutants, ACh release should also be reduced at ACh-D/M synapses. To test this idea, we recorded mEPSCs and evoked EPSCs from both post-synaptic partners (Fig. 6A and E). DD mEPSC rate (Fig. 6B) and evoked EPSC amplitudes (Fig. 6D) were significantly reduced in both lin-2(null) and lin-2(mKO) mutants, which lack muscle AChRs. By contrast neither lin-2 mutation altered DD neuron mEPSC amplitude (Fig. 6C), suggesting that these defects are not caused by decreased DD neuron AChR levels. Similarly, in acr-12 mutants (which lack a VD/DD AChR), muscle mEPSC rate (Fig. 6F) and evoked EPSC amplitudes (Fig. 6H) were significantly reduced and these defects were rescued by a transgene restoring ACR-12 expression in GABA neurons (SI Appendix, Fig. S9A,B,D). Here again, muscle mEPSC amplitudes were unaffected in acr-12 mutants (Fig. 6G; SI Appendix, Fig. S9C), suggesting that these defects are not caused by decreased muscle AChR levels. To confirm the role of AChRs expressed in GABA MNs, we analyzed unc-40(GKO) mutants, finding that muscle evoked EPSC amplitudes were significantly reduced (SI Appendix, Fig. S9A,D). Collectively, these results suggest that: 1) ACh release at ACh-D/M synapses is reduced when post-synaptic AChRs are decreased; 2) each post-synaptic partner modifies transmission to the second partner; and 3) ACh-D/M synapses function as dyads, where the same active zone releases ACh on both targets.

Figure 6. ACh release is reduced in mutants lacking post-synaptic AChRs.

Evoked EPSCs (elicited by photo-activation of ACh MNs) and mEPSCs were recorded from DD neurons (A-D) and body muscles (E-H). Representative evoked EPSC (red) and mEPSC (black) traces (A,E), mean mEPSC rate (B,F), mean mEPSC amplitude (C,G), and mean evoked EPSC amplitude (D,H) are shown for the indicated genotypes. DD mEPSC rates and evoked EPSC amplitudes were significantly decreased in lin-2(syb1019, null) and lin-2(muscle specific KO, mKO) mutants, which lack muscle post-synaptic AChRs, while DD mEPSC amplitudes were unaffected. Body muscle mEPSC rates and evoked EPSC amplitudes were significantly decreased in acr-12(ok367) mutants, which lack VD/DD AChRs, while muscle mEPSC amplitudes were unaffected. These results indicate that eliminating AChRs in either target decreases ACh release from motor neurons. lin-2(flex) is the WT control (lacking CRE expression) for lin-2(mKO). Sample sizes for each genotype are shown in each panel. Statistical differences were determined as follows: one way ANOVA with Tukey’s multiple comparisons test (B,C,D); unpaired t test with Welch’s correction (G,H); Mann-Whitney test (F). Values that differ significantly are indicated (ns, p >0.05; *, p <0.05; **, p <0.01; ****, p <0.0001). Error bars indicate SEM.

Mutants lacking GABA MNs have increased transmission at ACh-M synapses.

Thus far, our results suggest that both target cells contribute to retrograde increases in presynaptic UNC-2/CaV2 at ACh-D/M synapses. Next, we asked if compensatory mechanisms maintain ACh release when one target is eliminated. To test this idea, we analyzed unc-30 PITX2 mutants, which lack a transcription factor required for VD/DD cell fate specification (42), for changes in presynaptic UNC-2/CaV2 levels (Fig. 7) and ACh release (Fig. 8) at the resulting ACh-M synapses. Surprisingly, pre-synaptic UNC-2/CaV2 fluorescence (in DA/DB neurons) (Fig. 7A,B) and VB neuron UNC-2/CaV2 currents (Fig. 7E,F) were both significantly increased in unc-30 PITX2 mutants. Increased presynaptic UNC-2/CaV2 levels were accompanied by significantly increased muscle mEPSC rates (Fig. 8A,B) and larger evoked EPSC amplitudes (Fig. 8A,D). By contrast, muscle mEPSC amplitudes were unchanged in unc-30 PITX2 mutants (Fig. 8A,C), suggesting that increased mEPSC rate and evoked EPSC amplitude are not a consequence of increased muscle AChRs. Collectively, these results strongly support the idea that eliminating one target (VD/DD neurons) triggers a compensatory strengthening of transmission to the remaining target (body muscles) by increasing presynaptic UNC-2/CaV2 levels.

Figure 7. Pre-synaptic UNC-2/CaV2 levels in ACh motor neurons are increased in unc-30 PITX2 mutants.

DA/DB pre-synapses in the dorsal cord were identified by expressing tagBFP tagged UNC-57 (using the unc-129 promoter). UNC-2/CaV2(GFP₁₁) puncta intensity in dorsal cord DA/DB axons (A-B) and UNC-2/CaV2 currents in VB neurons (C-D) are compared in the indicated genotypes. Representative images (A) and mean UNC-2 puncta intensities (B), representative traces of VB neuron UNC-2/CaV2 current (at 0 mV) (C), and mean current amplitudes (at 0 mV) (D) are shown. Sample sizes for each genotype are shown in each panel. Statistical differences were determined as follows: Kruskal-Wallis test with Dunn’s multiple comparisons test (B); one way ANOVA with Tukey’s multiple comparisons test (D,F). Values that differ significantly are indicated (ns, p >0.05; **, p <0.01; ****, p <0.0001). Error bars indicate SEM. Scale bar indicates 5 μm.

Figure 8. ACh release is increased at ACh-M synapses in unc-30 PITX2 mutants.

Evoked EPSCs (elicited by photo-activation of ACh MNs) and mEPSCs were recorded from body muscles of the indicated genotypes. Representative evoked EPSC (red) and mEPSC (black) traces (A), mean mEPSC rate (B), mean mEPSC amplitude (C), and mean evoked EPSC amplitude (D) are shown. Body muscle mEPSC rates and evoked EPSC amplitudes were significantly increased in unc-30(e151) mutants, which lack VD/DD neurons, while mEPSC amplitudes were unaffected. These results suggest that unc-30 mutants have increased ACh release. The increased ACh release seen in unc-30 mutants was eliminated in unc-49;unc-30 double mutants, which lack muscle UNC-49/GABA_A receptors. By contrast, muscle mEPSCs and evoked EPSCs were unaltered in unc-25 GAD and unc-49/GABA_A single mutants. These results support the idea that the increased ACh release seen in unc-30 mutants is a consequence of mis-incorporation of UNC-49/GABA_A receptors into ACh-M post-synapses. Sample sizes for each genotype are shown in each panel. Statistical differences were determined by one way ANOVA with Tukey’s multiple comparisons test. Values that differ significantly are indicated (ns, p >0.05; *, p <0.05; ****, p <0.0001). Error bars indicate SEM.

What is the mechanism for this increase in ACh-M transmission? We did several experiments to address this question. Increased ACh release could result from decreased GABA transmission in unc-30 PITX2 mutants. Contrary to this idea, UNC-2/CaV2 fluorescence, UNC-2/CaV2 current, muscle mEPSCs, and muscle evoked EPSCs were all unaffected in unc-25 GAD mutants (which lack the GABA biosynthetic enzyme) (43) (Fig. 8 and 9). Alternatively, unc-30 PITX2 mutants could express a novel post-synaptic signal that potentiates pre-synaptic strength. Consistent with this idea, a prior study suggested that UNC-49/GABA_A receptors are ectopically localized to ACh-M synapses in unc-30 PITX2 mutants (44), which could increase presynaptic strength (5). Several results support this model. First, we confirmed that endogenous UNC-49 levels at ACh-M post-synapses were significantly increased in unc-30 PITX2 mutants (SI Appendix, Fig. S10), as previously reported (44). Second, in otherwise WT animals, mutations inactivating UNC-49/GABA_A had no effect on DA/DB presynaptic UNC-2/CaV2 levels (Fig. 7A,B), VB neuron UNC-2/CaV2 current (Fig. 7C,D), muscle mEPSC rate and amplitude (Fig. 8A-C), and muscle evoked EPSC amplitude (Fig. 8A,D). These results show that UNC-49/GABA_A receptors have no retrograde effects on pre-synaptic strength at ACh-D/M synapses in WT animals. By contrast, inactivating UNC-49/GABA_A receptors in unc-30 PITX2 mutants significantly decreased ACh-M pre-synaptic strength, as indicated by decreased DA/DB presynaptic UNC-2/CaV2 levels (Fig. 7A,B), decreased VB neuron UNC-2/CaV2 current (Fig. 7E,F), decreased muscle mEPSC rate (Fig. 8A-B), and decreased muscle evoked EPSC amplitude (Fig. 8A,D). Decreased ACh-M synaptic strength in unc-49; unc-30 double mutants is unlikely to result from decreased muscle AChRs, because muscle mEPSC amplitudes were unaltered in these double mutants (Fig, 8C). Collectively, these results strongly support the idea that increased pre-synaptic UNC-2/CaV2 levels and increased ACh release seen in unc-30 PITX2 mutants result from misincorporation of UNC-49/GABA_A receptors into ACh-M post-synaptic elements.

Discussion:

Our results lead to five principal conclusions. First, UNC-40/DCC clusters ACR-12 receptors at ACh-D/M post-synapses. Second, AChRs in both targets potentiate ACh release by recruiting UNC-2/CaV2 to ACh MN active zones. Third, each target influences transmission to the second target. Fourth, post-synaptic UNC-49/GABA_A receptors can promote retrograde recruitment of pre-synaptic UNC-2/CaV2 channels at both GABA and ACh NMJs. And fifth, like the C. elegans GABAergic NMJ, the ACh-D/M synapse does not exhibit pre-synaptic homeostatic plasticity (PHP). Below we discuss the significance of these findings.

Pre-synaptic NRX-1 and post-synaptic UNC-40/DCC cluster ACR-12 at ACh-D/M post-synapses.

Prior studies (analyzing transgene expressed ACR-12) suggest that ACR-12 clustering at ACh-D/M synapses is dramatically reduced in nrx-1 mutants but is unaffected in nlg-1 mutants (23, 26). These studies also showed that NRX-1 functions in ACh MNs to promote ACR-12 synaptic clustering. Our results are largely consistent with these prior studies; however, the magnitude of the ACR-12 decrease in nrx-1 mutants that we observed (~30%) was smaller than that seen in the prior work (~70% decrease). Our results and prior work also agree that NLG-1 is not the post-synaptic NRX-1 ligand mediating ACR-12 synaptic clustering.

Prior studies showed that UNC-40/DCC is required for synaptic clusters of muscle AChRs and GABA_A receptors (30, 32). Here, we show that UNC-40/DCC (acting in GABA MNs) plays an essential role in clustering ACR-12 at post-synapses. Imaging endogenous ACR-12 and analysis of DD mEPSC amplitudes indicate similar decreases in synaptic ACR-12 levels in unc-40(e271) mutants (~70% decrease) and these defects were rescued by a transgene restoring UNC-40 expression in GABA neurons. Smaller ACR-12 defects were previously reported in unc-40(e1430) mutants (~25% decrease) (23). Differences in the severity of unc-40 defects could result from analysis of transgenic versus endogenous ACR-12, differences in the unc-40 alleles analyzed, or differences in how ACR-12 fluorescence was quantified. Despite these quantitative differences, both studies indicate that UNC-40/DCC plays a significant role in promoting ACR-12 synaptic clustering. We find that UNC-40 can act in L4 larvae (after ACh-D/M synapse formation) to promote ACR-12 synaptic clustering. ACh-D/M synapses are not eliminated in unc-40 mutants, as indicated by the clustering of several AZ proteins. Collectively, these results suggest that decreased ACR-12 synaptic clustering in unc-40 mutants is not a consequence of earlier developmental defects.

We speculate that ACR-12 synaptic clustering could be mediated by direct trans-synaptic binding of pre-synaptic NRX-1 to post-synaptic UNC-40/DCC. Consistent with direct binding, a prior proteomic study reported that human DCC directly binds several NRXN isoforms (41). Also consistent with this idea, NRX-1 and UNC-40 are mutually dependent for their respective pre- and post-synaptic localization. Further experiments will be required to determine if trans-synaptic coordination of NRX-1 and UNC-40 levels are mediated by direct binding or if NRX-1 and UNC-40/DCC are indirectly linked via associated proteins.

Pre-synaptic UNC-2/CaV2 is trans-synaptically coupled to AChRs in both targets.

In simple (one to one) synapses, many studies suggest that pre-synaptic release sites and post-synaptic receptor clusters are aligned across synapses in nanocolumns (1-3). This nanocolumn alignment is proposed to be mediated by trans-synaptic tethering of AZ proteins to post-synaptic receptor complexes (6,7). Consistent with this hypothesis, we previously reported that post-synaptic GABA_A receptors in body muscles trans-synaptically recruit CaV2 channels to the active zones of contacting GABA MNs at NMJs, thereby potentiating GABA release (5). Here we extend these findings to a dyadic synapse. At ACh-D/M dyads, AChRs in both targets recruit CaV2 channels to ACh MN active zones, thereby increasing ACh release. In this manner, each target promotes transmission to both targets. By contrast, when GABA MN differentiation is blocked (in unc-30 PITX2 mutants), muscle GABA_A receptors are mis-incorporated into the resulting ACh-M post-synapses, which increases CaV2 levels at ACh MN AZs and increases ACh release. These results demonstrate that post-synaptic GABA_A receptors can function as a retrograde signal at both GABA and ACh synapses. We propose that the retrograde regulation of pre-synaptic strength seen in C. elegans results from trans-synaptic linking of UNC-2/CaV2 channels to clustered post-synaptic receptors, consistent with the trans-synaptic nanocolumn alignment seen in mammalian synapses.

Two recent studies question the nanocolumn alignment of release sites and post-synaptic receptors (45, 46). In these elegant studies, the ultrastructure of synapses formed by cultured hippocampal neurons was determined by cryo-electron tomography. In these structures, morphologically docked synaptic vesicles (SVs) in AZs are not aligned across the synapse with post-synaptic scaffolds nor with post-synaptic receptor clusters. In both C. elegans and hippocampal synapses, docked SVs are arrayed across a broad AZ domain (spanning hundreds of nanometers) (47, 48). Consequently, docked SVs vary greatly in their distance from the nearest calcium channel and in their release probabilities. For this reason, the nanocolumn hypothesis does not predict that docked SVs will be precisely aligned with post-synaptic receptor clusters. Instead, the nanocolumn model predicts that pre-synaptic CaV2 clusters (which should accurately predict release sites) are aligned with post-synaptic receptor clusters, consistent with our results at C. elegans ACh and GABA synapses. Interestingly, post-synaptic receptor clusters were laterally displaced (by ~60 nm) from the location of a docked SV in the cryo-ET structures (46). It is possible that this displaced alignment results from exclusion of docked SVs from the area occupied by pre-synaptic CaVs.

Implications for understanding synaptic plasticity.

At many synapses (including human, mouse, fly, and frog NMJs), disrupting post-synaptic receptors triggers compensatory responses (termed presynaptic homeostatic plasticity, PHP) (49-55). PHP restores synaptic function by boosting neurotransmitter release through a variety of mechanisms (56, 57). In C. elegans, disrupting post-synaptic receptors (at both GABA and ACh NMJs) elicits an anti-homeostatic response whereby presynaptic CaV2 abundance, CaV2 current, and neurotransmitter release are diminished (this paper and (5)). Thus, our current and previous results suggest that neither the GABA nor the ACh NMJs of C. elegans exhibit an effective form of PHP. These results reinforce the idea that PHP is not universal and that some animals (e.g. C. elegans) lack the ability to perform homeostatic adjustments of presynaptic function. In fact, it is possible that PHP evolved as a mechanism to offset the anti-homeostatic effects of CaV2 linking to post-synaptic receptors.

Ultrastructural studies suggest that there is a trans-synaptic matching of pre- and post-synaptic sizes across many brain regions (58, 59). Our results suggest that pre- and post-synaptic sizes and strengths are also matched at dyadic synapses. We speculate that mutations disrupting the trans-synaptic link between CaV channels and post-synaptic receptors will prevent matching of pre- and post-synaptic sizes and strengths.

Materials and Methods:

Animals

C. elegans strains were cultivated at room temperature (~22°C) on agar nematode growth media seeded with OP50 bacteria. Unless otherwise stated, the wild-type animal refers to the Bristol N2 strain. Alleles and strains used in this study are listed in SI Appendix, Table S1. Transgenic animals were prepared by microinjection. Single copy transgenes were isolated by the MoSCI and miniMoS techniques (60, 61).

CRISPR alleles

CRISPR alleles were isolated as described (62). Briefly, Cas9 protein and guide RNAs were ordered from IDT. Repair templates shorter than 200bp consisted of ss ULTRAMER oligos from IDT. Longer repair templates were PCR amplified from a plasmid and melted before adding to the injection mix. The injection mix included the pRF4 rol-6(gf) plasmid. Injected animals were singled and 96 F1 progeny were singled from plates that contained rollers, allowed to starve out the plate, and were then screened by PCR for the expected change. Split GFP constructs are as described (63).

Imaging endogenous synaptic proteins.

Pre- and post-synaptic puncta were identified by co-localization with fluorescent protein (mCherry or tagBFP)-tagged UNC-57/Endophilin, which associates with synaptic vesicles (64). The acr-12(nu782 GFP₁₁) allele contains 7 GFP₁₁ tags inserted into the TM3-4 cytoplasmic loop (after codon 406). ACR-12(GFP₁₁) fluorescence at ACh-D/M synapses was reconstituted with a miniMOS transgene co-expressing GFP_1-10 in GABA neurons (unc-25 promoter) and UNC-57/mCherry in DA/DB neurons (unc-129 promoter, nuSi645) or in VB neurons (ceh-12 promoter, nuSi709). Intensity of post-synaptic ACR-12 puncta (identified by close apposition with UNC-57) was quantified using FIJI. The unc-2(nu569 mNG) allele contains an mNG tag inserted after codon 2013 of UNC-2B. The unc-2(nu850 scarlet) allele contains a scarlet i3 tag inserted after codon 2013 of UNC-2B. The unc-2(nu586 GFP₁₁) allele contains 7 GFP₁₁ tags inserted after codon 2013 of UNC-2B, as described (5). UNC-2(nu586 GFP₁₁) fluorescence was reconstituted with a miniMOS transgene expressing GFP_1-10::SL2::UNC-57/mCherry in GABA neurons (nuSi285, unc-47 promoter) or DA/DB neurons (nuSi250, unc-129 promoter). The unc-40(nu864 Scarlet) allele contains scarlet-i3 inserted after codon 1381 of the endogenous unc-40 locus. The unc-49(nu829 mNG) allele contains an mNG tag inserted after codon 392 of the unc-49B isoform. The nrx-1(nu742 mNG) allele contains mNeonGreen inserted at codon 1515 of NRX-1A, which labels both long and short NRX-1 isoforms.

Tissue specific knockouts.

Tissue specific knockouts were performed using CRE inactivated alleles. CRE recombinase was expressed by miniMOS transgenes in muscles (nuSi491, myo-3 promoter), in GABA neurons (nuSi642, unc-25 promoter), or following heat shock (nuSi769, hsp-16.2 promoter). Tissue specific lin-2 knockouts were performed with the lin-2 (nu743 flex) allele, as described (31). Tissue specific unc-40 knockouts were made using the unc-40(nu842 flox) allele, which contains LoxP sites in introns 11 and 14 of the endogenous unc-40 locus.

Heat shock induced unc-40 rescue.

Heat shock induced unc-40 rescue was performed with a heat inducible CRE transgene (nuSi769, hsp-16.2 promoter) combined with a CRE-activated unc-40 allele, unc-40(nu856 FLEX Off). The unc-40(nu856) allele contains a stop cassette in intron 11 in the OFF configuration (i.e. in the same orientation as the unc-40 gene). The stop cassette is flanked by FLEX sites (which are modified loxP sites that mediate CRE induced inversions) (65), as previously described (31, 66). In this manner, unc-40(nu856) expression is induced following heat shock. ACR-12 synaptic clustering was compared in unc-40(nu856) day 2 adults that had been subjected to L4 heat shock (30°C for 6 hours) and in nonheat shocked controls.

Fluorescence imaging

Worms were immobilized on 10% agarose pads with 3 μl of 0.1 μm diameter polystyrene microspheres (Polysciences 00876-15, 2.5% w/v suspension). The dorsal nerve cord just anterior to the vulva was imaged. Images were taken with a Nikon A1R confocal, using a 60X/1.49 NA oil objective. Image volumes spanning the dorsal nerve cord were collected (20-30 planes/volume, 0.4 mm between planes, and 0.14 mm/pixel). Maximum intensity projections for each volume were auto-thresholded, and puncta were identified as round fluorescent objects (area > 0.1 mm²), using analysis of particles. Mean fluorescent intensity in each punctum was analyzed in the raw images. Pre-synaptic regions of interest (ROIs) were identified by localization of a fluorescently tagged synaptic vesicle marker (UNC-57/Endophilin) expressed in either the ACh or GABA motor neurons. Using CRISPR tagged alleles (see Table S1), the intensity of endogenously expressed active zone markers in the UNC-57 ROIs were quantified. All image analysis was done using FIJI.

Electrophysiology

Whole-cell patch-clamp measurements were performed using an Axopatch 200B amplifier with pClamp 10 software (Molecular Devices). The data were sampled at 10 kHz and filtered at 5 kHz. All recordings were performed at room temperature (~19-21 °C). Neurons were identified for patching by transgenes expressing fluorescent proteins in VB neurons (nuSi768, mKate2, ceh-12 promoter) or DD neurons (nuSi652, CRE, flp-13 promoter; flex inverted mKate2, unc-25 promoter). Muscles were identified for patching by bright field imaging.

VB neuron CaV recordings

The bath solution contained (in mM): TEA-Cl 140, CaCl₂ 5, MgCl₂ 1, 4-AP 3, glucose 10, sucrose 5, and HEPES 15 (pH 7.4, 330 mOsm). The pipette solution contained (in mM): CsCl 140, TEA-Cl 10, MgCl₂ 5, KOH 20, Tris 5, CaCl₂ 0.25, sucrose 36, EGTA 5 and Na₂ATP 5, Na₂GTP 1, HEPES 10 (adjusted to pH 7.2, 320 mOsm with CsOH). For UNC-2/CaV2 currents, 10mM nemadipine (an EGL-19/CaV1 inhibitor) was added to the bath solution. For EGL-19/CaV1 currents, recordings were done in unc-2 mutants. The voltage-clamp protocol consisted of −60mV for 50ms, −90mV for 100 ms, and test voltage (from −60mV to +60mV) for 200 ms. Depolarizing steps were initiated immediately after patching (<30 seconds). When repeated sets of depolarizing steps were done in a single neuron, I_CaV currents were stable for the first 3 sets and declined thereafter. To minimize the impact of I_CaV run down, only the first set of depolarizing steps were used for our analysis. In figures, we show inward currents evoked at 0 mV, which elicits the peak UNC-2 current.

VB neuron I_k recordings

The bath solution contained (in mM): NaCl 140, KCl 5, CaCl₂ 5, MgCl₂ 5, dextrose 11 and HEPES 5 (pH 7.2, 320 mOsm). The pipette solution contained (in mM): Kgluconate 120, KOH 20, Tris 5, CaCl₂ 0.25, MgCl₂ 4, sucrose 36, EGTA 5 and Na2ATP 4 (pH 7.2, 320 mOsm). The voltage-clamp protocol consisted of −60mV for 50ms, −90mV for 100 ms, and test voltage (from −60mV to +60mV) for 200 ms.

VB neuron RMP and input resistance recordings

The bath solution contained (in mM): NaCl 140, KCl 5, CaCl₂ 5, MgCl₂ 5, dextrose 11 and HEPES 5 (pH 7.2, 320 mOsm). The pipette solution contained (in mM): Kgluconate 120, KOH 20, Tris 5, CaCl₂ 0.25, MgCl₂ 4, sucrose 36, EGTA 5 and Na2ATP 4 (pH 7.2, 320 mOsm). RMP was measured in current clamp by holding at 0 pA for 20 seconds. Input resistance was measured in current clamp by measuring the membrane potential change evoked by injecting −10 pA.

DD EPSC recordings

DD EPSCs were recorded in dissected adults superfused in an extracellular solution containing 127 mM NaCl, 5 mM KCl, 26 mM NaHCO₃, 1.25 mM NaH₂PO₄, 10 mM glucose, 5mM sucrose, 5 mM CaCl₂, 4 mM MgCl₂, and 10 mM nemadipine at 22 °C. Nemadipine was included in the bath solution to isolate the EPSCs elicited by UNC-2/CaV2 channels. The pipette solution contained 105 mM CH₃O₃SCs, 10 mM CsCl, 15 mM CsF, 4mM MgCl₂, 5mM EGTA, 0.25mM CaCl₂, 10mM HEPES, and 5mM Na₂ATP, 1mM Na₂GTP adjusted to pH 7.2 using CsOH. Whole-cell recordings were carried out at −60mV to record mEPSCs. Stimulus-evoked EPSCs were obtained by photo-activating ACh motor neurons (expressing ChIEF) with a 3 ms light pulse (Lumencor 470 nm LED).

Muscle EPSC recordings

Body muscle EPSCs were recorded as previously described (67). Dissected adults were superfused in an extracellular solution containing 127 mM NaCl, 5 mM KCl, 26 mM NaHCO₃, 1.25 mM NaH₂PO₄, 10 mM glucose, 5mM sucrose, 1 mM CaCl₂, 4 mM MgCl₂, and 10 mM nemadipine at 22 °C. Nemadipine was included in the bath solution to isolate the EPSCs elicited by UNC-2/CaV2 channels. The pipette solution contained 105 mM CH₃O₃SCs, 10 mM CsCl, 15 mM CsF, 4mM MgCl₂, 5mM EGTA, 0.25mM CaCl₂, 10mM HEPES, and 5mM Na₂ATP, 1mM Na₂GTP adjusted to pH 7.2 using CsOH. Whole-cell recordings were carried out at −60mV to record mEPSCs. Stimulus-evoked EPSCs were obtained by photo-activating ACh motor neurons (expressing ChIEF) with a 3 ms light pulse (Lumencor 470 nm LED).

Statistical methods

Specific statistical tests are indicated in each figure legend. Data graphing and statistics were performed in GraphPad Prism 9. No statistical method was used to select sample sizes. Data shown in each figure represent contemporaneous measurements from mutant and control animals over a period of 6 months. For electrophysiology, data points represent mean values for individual neuron or muscle recordings (which were considered biological replicates). For imaging studies, data points represent mean puncta fluorescence values in individual animals (which were considered biological replicates). All data obtained in each experiment were analyzed, without any exclusions.

Significance Statement.

Although many synapses have multiple post-synaptic partners, nearly everything known about synaptic function has been determined by analyzing transmission from pre-synaptic cells to one target. Here we compare transmission to two targets at a dyadic synapse. We show that receptors in each target controls transmission to both targets via retrograde regulation of CaV2 levels at presynaptic active zones. These results provide new insights into the function and plasticity of multiple contact synapses.