CALM-1, the Caenorhabditis elegans calmyrin ortholog, functions with the voltage-gated calcium channels UNC-2 and UNC-36 to regulate neuromuscular junction morphology
Raymond C. Caylor1, Jennifer K. Pirri2, Mark J. Alkema2 and Brian D. Ackley1
Synapses are intercellular junctions that facilitate communication between a neuron and its target. Once formed, synapses undergo regulated changes in size, shape and activity to strengthen, or remove, existing connections. Previously, we have shown the Caenorhabditis elegans voltage-gated calcium channel subunits UNC-2 and UNC-36 mediate changes in GABAergic neuromuscular junctions (NMJs) that occur concurrently with organismal growth. In this study, we show that a gain-of-function mutation in unc-2 caused synaptic overgrowth, resulting in a phenotype resembling the loss of extracellular matrix molecule nidogen. Both unc-2 gain-of-function and nid-1/nidogenloss-of-function phenotypes were suppressed by a loss of function in calm-1, the single C. elegans ortholog of the vertebrate calcium- and integrin-binding proteins (CIB1-4), also known as calmyrins. Expression of calm-1 in the GABAergic motor neurons was able to rescue the synaptic phenotype, indicating calm-1 functions cell autonomously. We found CALM-1 binds to the adaptor protein RACK-1 in a calcium-dependent manner. Loss of rack-1 function caused a nid-1-like phenotype at GABAergic NMJs, and this phenotype was suppressed by removing calm-1. We conclude CALM-1 functions cell autonomously and downstream of UNC-2/UNC-36 to regulate synapse morphology.
Voltage-gated calcium channels (VGCCs) are multimeric proteins that facilitate calcium entry in response to membrane depolarization. Structurally, VGCCs are composed of a voltage-sensing and pore-forming α1 subunits that assemble with β, α2/δ and, occasionally, γ auxiliary subunits, which regulate channel localization and biophysical properties . VGCCs can be classified functionally by their subtype, with the Cav2-like channels residing largely at synapses where they facilitate exocytosis in response to neuronal depolarization. VGCC α1 subunit function is conserved through evolution, as its loss in Caenorhabditis elegans results in decreased neurotransmitter release [2-4] and in Drosophila, where α1 subunit mutations decrease neuronal firing .
Cav2-like VGCCs also regulate synaptic morphology, and this regulation has been found to occur independent of neuronal activity in both C. elegans and Drosophila [6-8]. Moreover, in the vertebrate central nervous system, the α2δ-1 subunit can bind thrombospondin to promote synaptogenesis, and this activity was not changed in the presence of drugs that block calcium channel function . These findings indicate VGCCs can provide an activity-independent signal to shape synaptic morphology. VGCC-derived calcium influx, in addition to neurotransmission, can initiate a myriad of cellular processes, encompassing events ranging from transcriptional regulation to cellular death .
With the ability to translate ionic concentrations into functional outputs, the neuronal calcium sensor (NCS) family is a diverse set of proteins that have evolved to respond to the transient and localized nature of intracellular calcium spikes. NCS proteins possess EF hand domains, which bind calcium and induce protein conformational shifts . These shifts permit NCS members to interact in a calcium-dependent manner with specific target proteins. For example, in photoreceptor cells, calcium activates recoverin and leads to the inhibition of rhodopsin kinase . Additionally, conformational rearrangements in response to calcium binding can, in some NCS family members, lead to the extrusion of a membrane-targeting myristoylation sequence, allowing for a calcium-induced change in protein localization . Thus, NCS proteins are attractive candidates to perform activity-independent morphological changes in response to VGCC function.
We have previously described a regulatory role in synaptic morphology for the C. elegans VGCCs subunits unc-2, the single C. elegans Cav2-type α1 subunit,and unc-36, an auxiliary α2δ subunit. Mutations in the conserved extracellular matrix molecule nidogen result in diffuse, elongated synaptic puncta , and this phenotype requires VGCC function . Moreover, the synaptic morphology VGCC function is specific to unc-2 and unc-36, as the other α1 subunits, egl-19 and cca-1,and a second α2δ subunit, tag-180, encoded in the C. elegans genome did not suppress nid-1 phenotype . Here, we demonstrate a similar synaptic regulatory role for the EF hand containing protein CALM-1, which was identified through a nid-1 suppressor screen. calm-1 is the single C. elegans ortholog of the vertebrate calcium- and integrin-binding proteins (CIB1-4); CIBs also known as calmyrins. CIB proteins can interact with platelet-specific integrins, to modulate platelet aggregation and adhesion [15,16]. Also, CIB1 is a calcium-dependent ligand of InsP3Rs and acts to both activate and inactivate receptor function . CIB proteins also possess the ability to traffic molecules to the plasma membrane via N-terminal myristoylation. For example, CIB1 is capable of transporting sphingosine kinase 1 to the plasma membrane . CALM-1 performs a similar function in C. elegans, as it directs sphingosine kinase (SphK) to synapses in cholinergic neurons, where SphK works to prime synaptic vesicles for release . CIB proteins are also localized in the nervous system, as CIB2 mRNA transcripts and protein are localized in the hippocampus and cortex of rat brains . Additionally, in cultured neurons, calmyrin1 has been shown to regulate neurite extension through its interaction with stathmin2 (SCG10), and an in vitro microtubule polymerization assay showed calmyrin1 inhibits the microtubule-destabilizing effects of stathmin2 . Interestingly, CIB2 was recently discovered as a risk gene in a study looking for rare copy number variants in patients with autism spectrum disorders .
At C. elegans GABAergic NMJs, calm-1 loss of function suppressed the synaptic defects present in nid-1 mutants. Using a gain of function in unc-2, we found calm-1 functioned downstream of UNC-2 VGCCs to regulate synaptic morphology. calm-1 mutants did not have uncoordinated (Unc) locomotion. These results are consistent with UNC-2 VGCCs affecting synaptic morphology independent of synaptic transmission. We also found the receptor for activated C kinase (RACK-1) bound CALM-1 in a calcium-dependent manner. Our work identifies calm-1 as a novel effector of VGCC-dependent regulation of synaptic morphology. Our genetic analyses suggests a balance between VGCC-dependent regulation of morphology and synaptic adhesion provided by nidogen allows for developmental synaptic growth to occur.
calm-1 is a suppressor of nidogen GABAergic neuromuscular junction defects
The D-type GABAergic motor neurons innervate the dorsal and ventral body wall muscles to help coordinate the characteristic sinusoidal movement of Caenorhabditis elegans. The neuromuscular junctions (NMJ) form en passant, along the length of individual axons, leading to one neuron possessing multiple presynaptic areas (Fig. 1A). We visualized NMJs using the presynaptic marker juIs1 (Punc-25::snb-1::GFP), which labels synapses in GABAergic motor neurons with the synaptic vesicle protein synaptobrevin fused to GFP [23,24]. In wild-type (wt)young adult animals, SNB-1::GFP localizes in discrete, evenly spaced puncta (Fig. 1B) with an average of 0.82 ± 0.04 μm2. We have previously demonstrated that loss of the extracellular matrix molecule (ECM) nidogen results in elongated and disorganized appearing puncta (Fig. 1C) with an area (1.52 ± 0.13 μm2, P < 0.05 vs wild type). We have also shown that loss-of-function mutations in the genes encoding the C. elegans voltage-gated calcium channels (VGCC) subunits unc-2 and unc-36 suppress SNB-1::GFP accumulation defects associated with .
To better understand how UNC-2 VGCCs contribute to NMJ morphology, we conducted a genetic screen for modifiers of the synaptic phenotype in nid-1 mutant animals (materials and methods). Briefly, nid-1(cg119);juIs1 animals were reared in the presence of dsRNA directed against genes encoding predicted or verified calcium-binding proteins or ion channels. We specifically examined animals that had grossly normal locomotion, but had alterations from the expected pattern of SNB-1::GFP puncta that form in nid-1(cg119) animals. We identified calm-1 as a potential suppressor of nid-1 in the RNAi screen (Fig. 1D). As calm-1 contained calcium-binding motifs and an N-terminal myristoylation sequence, it was an ideal candidate to link changes in calcium concentration via VGCCs to the morphological architecture of NMJs (Fig. 2A).
We obtained a calm-1 (lof) allele, tm1353, which removes some of the 5’ region adjacent to exon 1, exon 1 and part of the first intron (Fig. 2B), and is likely a complete null for calm-1. The calm-1 single mutants had an effect on SNB-1::GFP puncta size, appearing slightly enlarged with an area of 1.09 ± 0.06 μm2 (P <0.001 vs wt) (Fig. 1E). SNB-1::GFP puncta in nid-1(cg119);calm-1(tm1353) double mutants had an area of 1.09 ± 0.05 μm2, which was significantly less than nid-1(cg119) single mutants (P < 0.05) (Fig. 1F). In contrast to the VGCC mutants, unc-2 and unc-36, or the calmodulin kinase II mutant, unc-43, that we have previously described as nid-1 suppressors , calm-1 mutants did not have an obvious uncoordinated (Unc) phenotype. This is consistent with our previous results that regulation of GABAergic NMJ morphology is, at least in part, due to mechanisms independent of vesicle exocytosis .
calm-1 suppresses synaptic defects due to the loss of the receptor tyrosine phosphatase, ptp-3A
Nidogen interacts with the receptor tyrosine phosphatase ptp-3A/LARto anchor the active zone protein SYD-2 (liprin-α) at the synapse . Spanning from the extracellular matrix to the presynaptic active zone, this protein complex is positioned, and organized, to influence synaptic morphology. ptp-3(mu256) is a strong loss-of-function mutant , and results in the same diffuse, elongated SNB-1::GFP clusters as nidogen mutants, with synaptic areas averaging 1.52 ± 0.09 μm2 (P <0.001 vs wt and P >0.05 vs nid-1) (Fig. 1G). Since we observed calm-1 mutations suppressed nid-1 defects, we wondered if it would behave similarly in ptp-3(mu256) mutants. The calm-1;ptp-3 double mutant suppressed these defects, significantly reducing SNB-1::GFP cluster area to 0.78 ± 0.04 μm2 (P <0.001 vs ptp-3)(Fig. 1H). Previously, we observed similar effects using a mutation in unc-36, the extracellular VGCC α2/δ subunit, in combination with ptp-3, that is the doubles were more like wt than either of the single mutants . Taken together, intact VGCCs and CALM-1 function are required for the defects associated with loss of the nidogen protein complex to be manifested.
Figure 1 Synaptic defects associated with nid-1 and ptp-3 mutants are suppressed by calm-1
(A) Schematic representation of D-type motor neurons. VD neurons have presynaptic areas (green circles) that innervate the ventral body-wall muscles; whereas DD neurons innervate the dorsal muscles. Cell bodies (blue circles) of both neurons reside in the ventral nerve cord. (B) Wild-type animals have punctate shaped SNB-1::GFP areas that are regularly spaced along the nerve cord (arrow). A cell body is indicated by the asterisk. (C) nid-1(cg119) have elongated (arrows) and often irregularly shaped puncta (D) nid-1(cg119) animals reared on calm-1 RNAi have a more wild-type appearance with fewer elongated or disorganized puncta. (E) calm-1 mutants have enlarged synaptic areas when compared to wt. (F) A calm-1(tm1353) allele suppressed the defects observed in nid-1 mutants. (G) ptp-3 mutants show enlarged and disrupted SNB-1::GFP areas. (H) Removing calm-1 from ptp-3 rescued synaptic phenotype back to areas that are similar to the wild type. (I) We quantified the average area of the SNB-1::GFP puncta observed and compared across genotypes. N>180 puncta; *P<0.05, **P<0.001. Statistical significance was calculated by using the Student’s t-test. Error bars represent the Standard Error of the Mean (SEM).
Figure 2 calm-1 encodes a calcium binding protein
(A) A schematic of the domains found in the CALM-1 protein (201 amino acids) by the simple modular architecture research tool . By sequence similarity, CALM-1 is an ortholog of the calcium- and integrin-binding proteins (also known as calmyrin). CALM-1 has three predicted calcium-binding EF-hand motifs: EF1 = 84-112; EF2 = 121-149; EF3 = 162-190), along with a putative myristoylation site (marked by asterisk) position 2 of the protein. Scale bar is equivalent to 25 amino acids. Below the EF hands from CALM-1 are aligned with the consensus sequence, boxed regions indicate conserved calcium binding residues. Note, EF1 lacks a conserved residue (asterisk) suggesting it may not bind calcium. (B) The calm-1 locus consists of four exons, and endogenous expression constructs contained 1.4kb of upstream sequence. The deletion allele tm1353 removes the first coding exon, 5’UTR, and a portion of the first intron. Scale bar = 100bp.
calm-1 functions downstream of unc-2 and unc-36 to regulate neuromuscular junction morphology
calm-1 encodes the C. elegans ortholog of the calcium-and integrin-binding proteins (CIB). CALM-1 has three predicted calcium-binding EF hand motifs (EF hand 1 lacks a required aspartate in position 1 of the canonical EF-hand, and thus it may not be a functional). CALM-1 can be classified as a neuronal calcium sensor (NCS) family member. NCS proteins undergo conformational changes after binding calcium, allowing interactions with calcium-dependent target proteins . One outcome of a conformational change in some NCS proteins, including calmyrin, is the protrusion of an N-terminal myristoylation modification, which drives proteins to the plasma membrane . These characteristics make CALM-1 a candidate to be a downstream mediator of VGCC activity.
In calm-1 mutants, along with suppressing the nid-1 mutation defects, SNB-1::GFP clusters were fewer in number, but larger in size than those in wild-type animals. VGCC subunits unc-36 (1.23 ± 0.06 μm2) and unc-2 (1.26 ± 0.15 μm2) have similar SNB-1 accumulation defects (Fig. 3A,C). These results suggest the loss of CALM-1 is akin, in regulating NMJ morphology, to the reduction of calcium entry through VGCCs. We tested if calm-1 and theVGCC subunits worked redundantly or together to regulate SNB-1::GFP clusters. Double mutant analysis of calm-1;unc-36 (1.30 ± 0.10 μm2; P >0.05 vs unc-36; P >0.05 vs calm-1) and calm-1;unc-2 (1.10 ± 0.7 μm2; P >0.05 vs calm-1; P >0.05 vs unc-2) revealed that the SNB-1::GFP cluster areas were similar to each single mutant (Fig. 3B,D). As the area of the puncta are well below the average size of puncta observed in other genotypes, it is unlikely that there is a threshold effect, and thus, we concluded these genes work in a common pathway to shape presynaptic domains.
One potential mechanism would involve calcium influx through VGCCs leading to activation of CALM-1, which initiates a secondary signaling cascade. Alternatively, CALM-1 could be acting upstream of VGCCs to augment calcium signaling by altering the properties of the channel, as has been described for Frequenin/NCS1 . To understand the epistatic relationship between the VGCCs and CALM-1, we examined a gain-of-function (gof) VGCC allele, unc-2(zf35) . Hyperactivation of UNC-2 resulted in abnormally large and irregularly shaped SNB-1::GFP clusters (1.75 ± 0.17 μm2; P <0.001 vs wt; P <0.001 vs calm-1; P <0.05 vs unc-2), reminiscent of nid-1 mutant puncta (Fig. 3E). Removing CALM-1 from the unc-2(zf35) background reduced the SNB-1::GFP cluster area to 1.27 ± 0.07 μm2 (P <0.05 vs calm-1; P >0.05 vs unc-2 (e55)) (Fig. 3F). This indicated CALM-1 was required for the SNB-1 accumulation defects observed in unc-2(gof) and, thus, likely operates downstream of UNC-2 in regulation of synaptic morphology.
The unc-2(zf35) mutants exhibited a hyperactive locomotor behavior distinct from wild-type animals that was not suppressed by calm-1 loss-of-function (data not shown). This suggests that the activity of the UNC-2 channel was not grossly changed by the removal of calm-1, and that CALM-1 is unlikely to be functioning in synaptic morphology by depressing UNC-2 function. Overall we conclude from these results that calm-1 is likely to function downstream of unc-2, independent of neurotransmission, to regulate synaptic morphology.
We next tested unc-2(zf35);nid-1(cg119) double mutants to understand the interaction between nidogen and hyperactivated voltage-gated calcium channels (Fig. 3G). The unc-2(zf35);nid-1(cg119) double mutants had a SNB-1::GFP area of 1.71 ± 0.19 μm2 (P >0.05 vs cg119), suggesting these two genes likely work in a common genetic pathway to regulate synaptic morphology. Since we find that nid-1 likely functions upstream of unc-2, and the calm-1 mutation is able to suppress both nid-1(cg119) and unc-2(zf35), we conclude that the synaptic morphology defects in nid-1(lof) and unc-2(gof) are due to a common mechanism. A simple model is that NID-1 normally functions to inhibit synaptic growth, and that inhibition is relieved by activation of UNC-2. Since the nid-1(cg119) mutants do not cause a zf35-like locomotor activity, it is unlikely that NID-1 functions to inhibit UNC-2 activity in exocytosis.
Figure 3 calm-1 acts downstream of UNC-2 voltage-gated calcium channels in presynaptic morphology
(A-D) unc-36(e251) (A) and unc-2(e55) (C)have slightly enlarged synaptic areas compared to wild type, although the morphology is not significantly disrupted. unc-36 encodes an α2/δ subunit, and unc-2 encodes the α1 subunit of the VGGCs. unc-36; calm-1 (B) and unc-2;calm-1 (D) double mutants showed sizes similar to each single mutant, indicating they work in a similar pathway to regulate synaptic morphology. (E) An unc-2 gain-of-function allele, (zf35), showed enlarged synaptic areas similar to nid-1 mutants. (F) These defects were reduced in the loss-of-function calm-1 mutant. (G) unc-2(zf35); nid-1(cg119) double mutants are not significantly increased from either single mutant. (H) We quantified the average area of the SNB-1::GFP puncta observed and compared across genotypes. N>180 puncta; *P<0.05. Statistical significance was calculated by using the Student’s t-test. Error bars represent the Standard Error of the Mean (SEM).
CALM-1 localizes to the nervous system and musculature, and is required cell autonomously for proper neuromuscular junction morphology
We generated full-length Pcalm-1::CALM-1::GFP and RFP fusions to assay where the gene was expressed and the protein localized during development. The fusions were deemed functional since they partially rescued an egg-laying-defective (Egl) phenotype that was present in calm-1(tm1353) animals (90% Egl for tm1353 vs 35% for tm1353;lhEx36). CALM-1::GFP was first detectable in embryos at the gastrula stage and expression persisted throughout development and into adulthood. CALM-1 appears to be highly expressed in the nervous system and musculature (Fig.4 A-C). Notably, CALM-1 localization was apparent in the ventral nerve cord, and also appeared in muscle arms, which extend from muscles to the nerve cord. We found that the CALM-1 fusion protein was expressed in both GABAergic and cholinergic neurons that innervate the body wall muscles (Fig. 4D-F).
To test whether CALM-1 functioned cell autonomously an N-terminally tagged mCherry::CALM-1 fusion protein was expressed in the D-type motor neurons using the unc-25 promoter and introduced into the null mutant, tm1353. This construct rescued the SNB-1::GFP accumulation defects present in tm1353, resulting in puncta with an area of 0.88μm2 ± 0.06 (P = 0.01, compared to tm1353 alone; P > 0.05 vs wt) (Fig. 5). These results show that CALM-1 likely functioned in the VD neurons to regulate presynaptic morphology. We compared the subcellular localization of CALM-1::RFP relative to SNB-1::GFP clusters in the VD motor neurons. CALM-1::RFP was found throughout the VD neurons, with expression in the cell body, axon and localizing, partially, to GFP puncta (Fig. 5). Similar to mCherry alone, CALM-1 was not specifically localized to synapses or the regions bordering synaptic areas; rather, CALM-1 appears to be present diffusely throughout VD neurons, suggesting the protein was grossly localized to the cytoplasm under normal conditions.
Figure 4 Expression of calm-1 is observed in both neurons and muscles throughout development
We visualized animals expressing Pcalm-1::CALM-1::GFP (A-C) or Pcalm-1::CALM-1::tagRFP (D-F). (A) An L1 animal exhibits broad CALM-1::GFP expression. (B) Expression was observed in the vulval muscles of adult animals (arrow). (C) Expression was observed in muscles (arrowhead) where it localized primarily in the cytoplasm, but also was found in muscle arms (asterisks). In the nervous system, CALM-1 appeared to be diffusely localized along nerve cords (arrow). (D) We examined the Pcalm-1::CALM-1::RFP localization relative to the GABAergic synaptic marker (SNB-1::GFP) (E). (F) CALM-1 expression was detected in the GABAergic motor neurons (arrow) as well as the adjacent cholinergic motor neurons. Scale bars are 10μm.
Figure 5 calm-1 can function cell autonomously in GABAergic motor neurons to regulate synaptic morphology
(A-D) An mCherry::CALM-1 chimera was expressed in GABAergic motor neurons using the unc-25 promoter.. Its localization in the GABAergic motor neurons is identified by comparing its expression using the synaptic marker juIs1. The colocalization between juIs1 and CALM-1 protein is observed in both cell bodies and in the axons. We found no specific enrichment of CALM-1 in areas labeled by SNB-1::GFP. (D) An enlarged image of the boxed region from (C) demonstrating the lack of enrichment of CALM-1 in SNB-1::GFP puncta. Scale bar in C is 10 mm and D is 1 mm. (E) Punc-25::mCherry::CALM-1 expression in GABA neurons was able to rescue SNB-1::GFP defect of calm-1 mutants. Bar graph of SNB-1::GFP area. N>180 puncta; *P<0.05, **P<0.001. Statistical significance was calculated by using the Student’s t-test. Error bars represent the Standard Error of the Mean (SEM). Scale bar is 10μm.
The intracellular scaffolding molecule RACK-1 is a calcium-dependent binding partner of CALM-1
Our results suggested that calm-1 could be part of a calcium-dependent signal downstream of UNC-2 to regulate changes in synaptic morphology. Since CALM-1 contains predicted EF hand motifs, we hypothesized CALM-1 could bind to target proteins in response to increased calcium at the synapse. We identified potential calcium-dependent interactions using affinity chromatography (materials and methods). We recovered 14 proteins that associated with CALM-1 in a calcium-dependent manner (Table 1). These proteins included multiple ribosomal subunit proteins, a eukaryotic initiation factor and the receptor for activated C kinase (RACK-1). RACK-1 was of interest because it is an intracellular scaffold involved in multiple neuronal processes  and functions in the D-type GABA neurons to control axon pathfinding .
We confirmed that CALM-1 and RACK-1 can physically interact directly in vitro in a calcium-dependent fashion (Fig. 6A). Next, we examined SNB-1::GFP patterning in rack-1(tm2262), which is a strong loss-of-function mutation in rack-1 (Fig 6B). The tm2262 mutants displayed an increase in SNB-1::GFP cluster size, resulting in areas of 1.65 ± 0.21 μm2 ; P <0.001 vs wt(Fig. 6C). The enlarged SNB-1::GFP clusters in rack-1 mutants are suppressed by the removal of calm-1 (1.18 ± 0.09 μm2; P = 0.05 vs rack-1) (Fig. 6D). Additionally, when RACK-1::GFP was expressed in the GABAergic motor neurons, it colocalized with another fluorescent synaptic-vesicle marker, RAB-3::mCherry (Fig. 6E). Together, these results indicate RACK-1 is localized at presynaptic areas, and that it regulated synapse morphology.
Figure 6 CALM-1 binds RACK-1, rack-1 phenocopies nid-1 synaptic defects and is suppressed by loss of calm-1
(A) 6-His-RACK-1 and GST-CALM-1 were mixed at an equimolar ratio in increasing concentrations of calcium (indicated above lanes). After purifying 6-His-RACK-1 using Ni2+-agarose, we probed for the presence of GST-CALM-1 in bound fractions (Ni2+ capture) or unbound (flow through) using an anti-GST antibody. We found that in the presence of high calcium RACK-1 and CALM-1 can directly interact. (B) Gene structure of rack-1.The tm2262 allele has an in-frame deletion that causes a strong loss of function. (C) rack-1(tm2262) had SNB-1::GFP areas that were elongated and disorganized. (D) The defects observed in rack-1 mutants were reduced in calm-1;rack-1 double mutants. (E) RACK-1::GFP expressed specifically in the GABAergic motor neurons colocalized with the synaptic vesicle marker RAB-3::RFP. (F) Bar graph of SNB-1::GFP area. N>180 puncta; *P=0.05, **P<0.001. Statistical significance was calculated by using the Student’s t-test. Error bars represent the Standard Error of the Mean (SEM).
We have shown here the C. elegans calmyrin ortholog CALM-1 plays a role in Cav2/UNC-2 mediated regulation of GABAergic neuromuscular junction (NMJ) morphology in C. elegans. To our knowledge this is the first demonstration that calmyrin-like proteins regulate synaptic morphology. We also provide evidence that calm-1 interacts both physically and genetically with rack-1, a well-described scaffolding protein that integrates multiple signaling pathways. When added to the previously characterized role for calmyrins in the localization of synaptic proteins, e.g. sphingosine kinase, it seems these molecules are emerging as important regulators of synaptic biology. The calm-1 locus encodes an EF hand protein, and we have found that it functions genetically in the same pathway as the voltage-gated calcium channel (VGCC) subunits UNC-2 and UNC-36 to regulate NMJ morphology. Genetic analysis suggests synaptic morphology regulation through VGCCs and CALM-1 is opposite the role of the ECM component nidogen and the leukocyte-common antigen related receptor tyrosine phosphatase PTP-3A. With these contrasting roles, we propose that the calcium signaling components and cellular adhesion molecules work to provide a temporal switch that allows for synaptic stabilization and growth. This supports our previous results that demonstrate synapses form by elongation and division during development in an UNC-2-dependent manner.
calm-1 functions cell autonomously and downstream of VGCCs to regulate synaptic morphology
Nidogen and the tyrosine phosphatase PTP-3A/LAR cooperate to anchor the active zone regulator liprin-α (SYD-2) at presynaptic areas . This complex constricts synaptic material to discrete subcellular regions, as loss-of-function mutations result in a diffusion of synaptic material. We have previously shown mutations in unc-2 and unc-36 suppress the synaptic morphology defects associated with nid-1 and ptp-3, indicating the elongated shape requires proper VGCC function . A simple genetic interpretation is that the NID-1/PTP-3A complex functions to inhibit UNC-2 mediated synaptic expansion required to form new synapses by elongation and budding (Fig. 7). calm-1 acts cell autonomously and downstream of unc-2 and unc-36 to regulate GABAergic NMJ morphology.This is in contrast to a previously described VGCC and calcium-binding protein interaction at Drosophila NMJs. Frequenin/NCS-1 regulates presynaptic bouton number and synaptic firing through the α1 subunit cacophony, presumably activating VGCCs to control morphology . These results underscore the complex relationship between calcium signaling and synaptic morphology. Interestingly, unlike mutations in unc-2, unc-36 and other synaptic transmission mutants, calm-1 mutants are not grossly uncoordinated (Unc). Furthermore, the suppression of the unc-2(gof) synaptic phenotype by calm-1 mutants is not accompanied by the suppression of unc-2(gof) behavioral defects. Together, these results suggest calm-1 functions largely independent of neurotransmission to regulate synaptic structure. Previously, we demonstrated that mutations in unc-13, which is required for synaptic transmission, do not suppress nid-1 synapse morphology defects [6,24]. These results fit into accumulating evidence that synaptic structure can be regulated independent of synaptic firing [6-9].
Figure 7 Model of developmental synaptic growth
A) The genetic interactions we observed suggest that nid-1 and ptp-3 normally function to repress unc-2 function, as loss-of-function mutations in nid-1 or ptp-3 are equivalent to mutations that activate unc-2. Our genetic evidence places calm-1 downstream of unc-2. The unc-2 pathway, including calm-1 is required for the normal pattern of synaptic growth leading to the formation of new synapses from existing ones. Genetic evidence suggests rack-1 is likely functioning in the same pathway as nid-1, but it may also function independently of the NID-1 complex. B) We provide the following model of the developmental changes that occur during synapse addition: (A) Synapses are established de novo and exist in a homeostasis. (B) In response to an as-of-yet unidentified growth signal, presynaptic areas start to increase in size. The increase in size is governed by calcium signaling but to enter the budding phase likely requires increased UNC-2 function and a reduction in nidogen-mediated adhesion. (C) Enlarged synaptic areas transition into two distinct puncta (budding) which requires the UNC-2 channel to be returned to a normal level of function, and likely an increase in nidogen-mediated adhesion. (D) The end result of this process is the establishment of two synaptic connections from a single pre-existing synapse.
Calcium signaling and its role in modulating synaptic adhesion
We essentially observe two distinct phenotypes in the groups of mutants we have described. First, the unc-2, unc-36 or calm-1 lof mutants exhibit slightly enlarged, but morphologically normally puncta. Second, nid-1, ptp-3A, syd-2 or rack-1 lof and unc-2 gof mutants have elongated, morphologically abnormally synapses. The puncta found in the latter group resemble those we have found occur naturally during development, suggesting these molecules contribute to this normal developmental program in opposite fashions. We hypothesize synaptic adhesion and expansion exist in a balance that is regulated by temporal calcium signaling. First, the elongated mutant synaptic puncta resemble the intermediate stage of synaptic growth, regulated by unc-2, observed during wild-type C. elegans development . This suggests proper calcium channel function is required for normal synaptic dynamics. Second, hyperactivated UNC-2 signaling mimics the synaptic defects of loss of molecules required for cellular adhesion, indicating that calcium channel function needs to be regulated for proper synaptic morphology. Finally, removing wild-type calcium signaling from adhesion mutants suppresses the associated elongated defects. These results imply synaptic morphological changes require downregulation of synaptic adhesion molecules to initiate the diffusion of synaptic molecules into adjacent axonal regions. Accompanying downregulation of adhesion molecules, changes in intracellular calcium concentration activate secondary messengers that finalize synapse expansion. The budding of synaptic areas then requires these changes in adhesion and calcium signaling to be reversed, to allow for new synapses to occur (Fig. 7).
Our genetic analysis suggests genes required for synapse expansion are epistatic to synapse adhesion molecules, as double mutants all suppress elongated phenotypes associated with adhesion genes, with calm-1 as the most downstream effector we have identified in this system. The calcium-dependent biochemical interaction between CALM-1 and RACK-1 may indicate a point where the switch in synaptic adhesion may occurs. RACK-1 possesses seven WD40-repeat motifs, which allow it to interact with a diverse range of proteins . Interestingly, calmodulin interacts with another WD-repeat protein, striatin, in neurons, and is responsible for striatin subcellular redistribution , indicating there may be a pattern of interactions between NCS and WD40-repeat proteins. We also identified a number of ribosomal proteins in the CALM-1 calcium-dependent interaction experiment. There is evidence that RACK1 homologs in other organisms can function as translational inhibitors . RACK1 has also shown to interact with the p38/MAPK pathway . In C. elegans the DLK-1/MAPK pathway has been shown to be regulated by the regulator of presynaptic morphology (rpm-1) at the translational level [33,34]. While we do not have evidence that the ribosomal complex proteins can interact directly with CALM-1, or whether any such interaction is via the RACK-1 protein, this is an area of active investigation.
In vertebrate cells Rack1 interacts with PTPμ both at cell contact points and to regulate E-cadherin-mediated adhesion [35,36]. Since lof mutations in C. elegans rack-1 resemble the loss of nid-1 or ptp-3A a simple model would propose that RACK-1 function could stabilize the PTP-3A/NID-1 adhesion complex. This would also be consistent with the results that calm-1 lof suppresses the rack-1 defects. Further work looking at the interaction between RACK-1 and PTP-3, both in the presence and absence of CALM-1, at C. elegans NMJ will provide insight into the molecular mechanisms of synaptic structural modulation.
A similar process of adhesion-dependent dynamic assembly and disassembly that we have described here also occurs at Drosophila NMJs. In that system the cellular adhesion molecule Fasciclin II (FasII) appears to be a critical component. The amount of FasII determines synaptic bouton number in Drosophila, as null mutants have fewer boutons, compared with hypomorphic mutants that possess increased synaptic areas . In contrast with the increase in synaptic growth of FasII mutants, synaptic activity is not altered . These data indicate levels of FasII have a structural role in synaptic structural plasticity. Moreover, in vivo time-lapse imaging of developing larvae found instances of lower levels of FasII at regions of synaptic connections that undergo a budding process that produces new synaptic boutons . Additionally, FasII levels are reduced in the potassium channel double mutant ether-a-go-go Shaker (eag Sh), which results in increased synaptic activity and bouton growth . FasII levels are also reduced in hyperactivated CaMKII mutants that lead to increased synaptic bouton size, and are similar to loss-of-function mutations in synaptic scaffolding molecule Discs large (Dlg) . Thus, synaptic bouton growth seems to require CaMKII phosphorylation of Dlg, preventing Dlg from contacting FasII to stabilize synaptic areas.
In conclusion, the balance between synaptic stability and growth in the C. elegans GABAergic NMJs is modulated by the dynamic interactions that control either synaptic adhesion or expansion. There is a program of synaptic development, which appears to be gated by the UNC-2 VGCCs, which facilitates entry into a process of elongation and division. In this program the adhesion complex regulated by NID-1 likely functions to inhibit elongation and/or promote division. This program therefore would provide both a balance that permits the organism to add synapses during development, or perhaps in response to other cues, in a local fashion.
This work was supported by the National Institutes of Health (NIH) including partial support from the grants GM084491 (MJA) and GM103638 (BDA).
Materials and Methods
Caenorhabditis elegans strains All strains were maintained at 20-22.5°C as previously described  unless otherwise mentioned. The following alleles and double mutants were used in this study: N2 (var. Bristol), juIs1 [Punc-25::SNB-1::GFP], nid-1(cg119), nid-1(cg119);eri-1(mg366)juIs1, ptp-3(mu256);eri-1(mg366)juIs1, calm-1(tm1353), rack-1(tm2262), unc-2(zf35), ptp-3(mu256), unc-2(e55), unc-36(e251), calm-1(tm1353); unc-2(zf35), calm-1(tm1353);unc-36(e251), calm-1(tm1353);ptp-3(mu256), calm-1(tm1353);unc-2(zf35), calm-1(tm1353);rack-1(tm2262). Transgenic animals were generated by germline transformation as previously described .
RNAi screen A library of dsRNA expressing bacterial strains was generated from the Ahringer library . Genes were chose based on text-mining from Wormbase, by looking for genes annotated with a calcium-dependent function, e.g. calcium-binding, etc.,. We used the enhancer of RNAi, eri-1, to potentiate the effect in neurons . Briefly, juIs1, nid-1(cg119) or ptp-3(mu256) animals with the eri-1(mg366) mutation were reared on normal growth media (NGM) supplemented with carbenicillin (25 mg/ml) and IPTG (0.3 mM) seeded with bacteria containing either an empty vector (L4440) or the targeting plasmids. Animals were grown for two generations in the presence of the dsRNA producing bacteria and then visualized the SNB-1::GFP synaptic marker in young adults for deviation from the expected pattern. The screen was designed to find clones that modified the synaptic patterning found in nid-1 mutants, but did not induce gross locomotion defects in animals; we did not specifically exclude analysis of animals with mild locomotor defects (Unc).
Molecular biology A calm-1 genomic fragment (including endogenous promoter) was amplified using pF30A10.1 F1 and F30A10.1 R1 to make pBA245 and recombined into (pDT57) to create Pcalm-1::CALM-1::GFP (pBA251). pBA251 was injected at 45ng/μL into calm-1(tm1353) to create EVL114 (lhEx36). To visualize if CALM-1 was localized in GABA neurons, Pcalm-1::CALM-1::tagRFP was generated by recombining pBA245 into pEVL334 (tag-RFP + let-858 3’ UTR) using LR clonase Gateway System (Life Technologies) to create pEVL395. This was injected into juIs1 at 10ng/μL to create EVL1001. A calm-1 genomic fragment (minus endogenous promoter including the 3’ UTR) was amplified using the following primers: calm-1 ATG F1 and calm-1 3′UTR R1. This product was recombined into pEVL387 (Punc-25::mCherry::unc-43 3’ UTR) using the InFusion enzyme (Clontech) to generate pEVL408 (Punc-25::mCherry::calm-1::calm-1 3′UTR). The pEVL408 plasmid was injected into juIs1 animals at 10ng/μL to create EVL1204 (lhEx389), and crossed into calm-1(tm1353);juIs1 to create EVL1218 for SNB-1::GFP colocalization analysis RACK-1 and RAB-3 colocalization experiment was visualized by injecting pEVL26 (Punc-25::mcherry::rab-3) into lqIs174 (Punc-25::rack-1 gDNA::gfp) at 5 ng/μL to create EVL352. Any additional information or details about plasmids created and used in this report is available upon request.
Image analysis Neuromuscular junction morphology of GABAergic VD neurons was visualized by juIs1 [Punc-25 SNB-1::GFP]. All images were collected on an Olympus FV1000 confocal microscope equipped with Fluoview software. Images were acquired using multi-track parameters for any multi-color images, using a 60X Plan-apochromat objective. Animals were anesthetized using 0.5% phenoxy-propanol (TCI America) in M9 and mounted on 2% agarose pads. Measurements of SNB-1::GFP were as described, with minor modifications . All images were collected using the exact same microscope settings. Briefly, confocal images were projected into a single plane using the maximum projection and exported as a tiff file with a scale bar. Using ImageJ the files were converted to a binary image using the threshold command, so that the binary image resembled the RGB image. A region of interest was drawn around SNB-1 puncta in the ventral nerve cord. The following measurement options were selected: Area, Center of Mass, Circularity, Perimeter, Fit Ellipse, and Limit to Threshold. Scaling was set by measuring the scale bar. The “Analyze Particle” command was used with a minimum of four pixels and no maximum size. The following options were selected: Outline Particles, Ignore Particles Touching Edge, Include Interior Holes and Reset Counter. The resulting measurements were exported to Microsoft Excel for statistical analysis. Comparisons of single mutants and double mutants to the wild type, or specified genotype, were tested by Students two-tailed t-test.
Protein purification and CALM-1 pulldown: A calm-1 cDNA was recombined into pDEST17 (Life Technologies) using LR recombinase (Life Technologies) according to manufacturer’s directions to create pEVL25, and transformed into BL21(DE3) bacteria. Bacteria were grown to OD=600 and induced with 1mM IPTG for three hours at 37°C. Bacteria were collected by centrifugation and pellets frozen in liquid nitrogen. Pellets were lysed and dialyzed into binding buffer (20mM HEPES, 100mM KCL, 2mM CaCl2, 0.1% NP-40, 0.1% Tween-20, pH 7.4). 6His-CALM-1 was purified by affinity chromatography using nickel NTA resin (Qiagen). The CALM-1 column was prepared by immobilizing 6-His-CALM-1 to CNBr-activated sepharose beads (GE Life Sciences) following manufacturer’s directions. N2 animals were grown on HB101 bacteria; washed off plates with M9 and incubated with additional HB101, 500μL streptomycin (100μg/μL), 500μL of 5mg/ml cholesterol and M9 up to 500mL; shaken at room temperature for three days; liquid spun down and supernatant frozen and then lysed with worm lysis buffer (1xPBS, 10% glycerol, 0.1% NP40, 0.1% Tween 20 and 1mM PMSF). The lysate was spun down and supernatant adjusted to a calcium binding buffer (20mM HEPES, 100mM KCL, 2mM CaCl2, 0.1% NP-40, 0.1% Tween-20, pH 7.4) and incubated with the CALM-1-sepharose column overnight. Column was washed with calcium binding buffer, and interacting proteins were removed with 250μL of elution buffer (20mM HEPES, 100mM KCL, 5mM EDTA, 0.1% NP-40, 0.1% Tween-20, pH 7.4) Samples were concentrated by trichloroacetic acid (TCA) precipitation and run on a 7.5% SDS gel. Gel bands were excised and analyzed by MALDI-TOF mass spectrometry (KU Structural Biology Facility).
CALM-1-RACK-1 binding experiments: A calm-1 cDNA was recombined into pDEST15 (Life Technologies) and a rack-1 cDNA was recombined into pDEST17 using LR recombinase (Life Technologies) according to manufacturer’s directions and individually transformed into BL21(DE3) bacteria. Bacteria were grown to OD=600 and then proteins were induced with 1mM IPTG and grown overnight (16 hours) at 37 °C. Bacteria were collected by centrifugation and pellets frozen in liquid nitrogen. Pellets were lysed and purified, GST-CALM-1 via a glutathione matrix (G-Biosciences) and 6-His-RACK-1 via NTA-agarose (Qiagen) according to manufacturer’s directions. After purification, the proteins were and dialyzed into binding buffer lacking calcium (20mM HEPES, 100mM KCL, 0.1% NP-40, 0.1% Tween-20, pH 7.4). Proteins were mixed at 1:1 ratio in binding buffer supplemented with EDTA or Ca2+ as indicated in figure. Protein mixtures were purified via NTA agarose (Qiagen), and eluted using 300 mM imidazole according to manufacturer’s directions. Flow through fractions were collected, and volumes of elution fractions were adjusted to be equal to the flow through. Elutions and flow through fractions were then separated by SDS-PAGE and transferred to nylon membrane for western blotting. We probed for CALM-1 in the elution and flow through fractions using an anti-GST antibody (Life Technologies).
|Table 1. Proteins Identified via CALM-1 Ca2+-dependent Chromatography|
|Gene||WormBase Accession||Brief description|
|RACK-1||WBGene00010556||Receptor of Activated C Kinase|
|EIF-6||WBGene00001234||Eukaryotic initiation factor|
|HSP-70||WBGene00002026||Heat shock protein|
|RLA-0||WBGene00004408||Acidic ribosomal subunit protein P0|
|RPL-2||WBGene00004413||Large ribosomal subunit L8 protein|
|RPL-4||WBGene00004415||Large ribosomal subunit L4 protein|
|RPL-5||WBGene00004416||Large ribosomal subunit L5 protein|
|RPL-7A||WBGene00004419||Large ribosomal subunit L7a protein|
|RPS-3||WBGene00004472||Ribosomal protein, small subunit|
|RPS-6||WBGene00004475||Small (40S) ribosomal subunit S6 protein|
2. Mathews EA, Garcia E, Santi CM, Mullen GP, Thacker C, et al. (2003) Critical residues of the Caenorhabditis elegans unc-2 voltage-gated calcium channel that affect behavioral and physiological properties. J Neurosci 23: 6537-6545.
6. Caylor RC, Jin Y, Ackley BD (2013) The Caenorhabditis elegans voltage-gated calcium channel subunits UNC-2 and UNC-36 and the calcium-dependent kinase UNC-43/CaMKII regulate neuromuscular junction morphology. Neural Dev 8: 10.
9. Eroglu C, Allen NJ, Susman MW, O’Rourke NA, Park CY, et al. (2009) Gabapentin receptor alpha2delta-1 is a neuronal thrombospondin receptor responsible for excitatory CNS synaptogenesis. Cell 139: 380-392.
13. O’Callaghan DW, Ivings L, Weiss JL, Ashby MC, Tepikin AV, et al. (2002) Differential use of myristoyl groups on neuronal calcium sensor proteins as a determinant of spatio-temporal aspects of Ca2+ signal transduction. J Biol Chem 277: 14227-14237.
14. Ackley BD, Kang SH, Crew JR, Suh C, Jin Y, et al. (2003) The basement membrane components nidogen and type XVIII collagen regulate organization of neuromuscular junctions in Caenorhabditis elegans. J Neurosci 23: 3577-3587.
20. Blazejczyk M, Sobczak A, Debowska K, Wisniewska M, Kirilenko A, et al. (2009) Biochemical characterization and expression analysis of a novel EF-hand Ca2+ binding protein calmyrin2 (Cib2) in brain indicates its function in NMDA receptor mediated Ca2+ signaling. Archives of biochemistry and biophysics 487: 66-78.
22. Prasad A, Merico D, Thiruvahindrapuram B, Wei J, Lionel A, et al. (2012) A discovery resource of rare copy number variations in individuals with autism spectrum disorder. G3 (Bethesda, Md) 2: 1665-1685.
24. Jin Y, Jorgensen E, Hartwieg E, Horvitz HR (1999) The Caenorhabditis elegans gene unc-25 encodes glutamic acid decarboxylase and is required for synaptic transmission but not synaptic development. J Neurosci 19: 539-548.
25. Ackley BD, Harrington RJ, Hudson ML, Williams L, Kenyon CJ, et al. (2005) The two isoforms of the Caenorhabditis elegans leukocyte-common antigen related receptor tyrosine phosphatase PTP-3 function independently in axon guidance and synapse formation. J Neurosci 25: 7517-7528.
26. Dason JS, Romero-Pozuelo J, Marin L, Iyengar BG, Klose MK, et al. (2009) Frequenin/NCS-1 and the Ca2+-channel alpha1-subunit co-regulate synaptic transmission and nerve-terminal growth. J Cell Sci 122: 4109-4121.
27. Schumacher J, Hsieh Y-W, Chen S, Pirri J, Alkema M, et al. (2012) Intercellular calcium signaling in a gap junction-coupled cell network establishes asymmetric neuronal fates in C. elegans. Development (Cambridge, England) 139: 4191-4201.
31. Gerbasi VR, Weaver CM, Hill S, Friedman DB, Link AJ (2004) Yeast Asc1p and mammalian RACK1 are functionally orthologous core 40S ribosomal proteins that repress gene expression. Mol Cell Biol 24: 8276-8287.
32. Belozerov VE, Ratkovic S, McNeill H, Hilliker AJ, McDermott JC (2014) In vivo interaction proteomics reveal a novel p38 mitogen-activated protein kinase/Rack1 pathway regulating proteostasis in Drosophila muscle. Mol Cell Biol 34: 474-484.
33. Nakata K, Abrams B, Grill B, Goncharov A, Huang X, et al. (2005) Regulation of a DLK-1 and p38 MAP kinase pathway by the ubiquitin ligase RPM-1 is required for presynaptic development. Cell 120: 407-420.
35. Hellberg CB, Burden-Gulley SM, Pietz GE, Brady-Kalnay SM (2002) Expression of the receptor protein-tyrosine phosphatase, PTPmu, restores E-cadherin-dependent adhesion in human prostate carcinoma cells. J Biol Chem 277: 11165-11173.
36. Mourton T, Hellberg CB, Burden-Gulley SM, Hinman J, Rhee A, et al. (2001) The PTPmu protein-tyrosine phosphatase binds and recruits the scaffolding protein RACK1 to cell-cell contacts. J Biol Chem 276: 14896-14901.
37. Schuster CM, Davis GW, Fetter RD, Goodman CS (1996) Genetic dissection of structural and functional components of synaptic plasticity. I. Fasciclin II controls synaptic stabilization and growth. Neuron 17: 641-654.
38. Schuster CM, Davis GW, Fetter RD, Goodman CS (1996) Genetic dissection of structural and functional components of synaptic plasticity. II. Fasciclin II controls presynaptic structural plasticity. Neuron 17: 655-667.