Tomosyn Inhibits Synaptic Vesicle Priming in Caenorhabditis elegans

Caenorhabditis elegans TOM-1 is orthologous to vertebrate tomosyn, a cytosolic syntaxin-binding protein implicated in the modulation of both constitutive and regulated exocytosis. To investigate how TOM-1 regulates exocytosis of synaptic vesicles in vivo, we analyzed C. elegans tom-1 mutants. Our electrophysiological analysis indicates that evoked postsynaptic responses at tom-1 mutant synapses are prolonged leading to a two-fold increase in total charge transfer. The enhanced response in tom-1 mutants is not associated with any detectable changes in postsynaptic response kinetics, neuronal outgrowth, or synaptogenesis. However, at the ultrastructural level, we observe a concomitant increase in the number of plasma membrane-contacting vesicles in tom-1 mutant synapses, a phenotype reversed by neuronal expression of TOM-1. Priming defective unc-13 mutants show a dramatic reduction in plasma membrane-contacting vesicles, suggesting these vesicles largely represent the primed vesicle pool at the C. elegans neuromuscular junction. Consistent with this conclusion, hyperosmotic responses in tom-1 mutants are enhanced, indicating the primed vesicle pool is enhanced. Furthermore, the synaptic defects of unc-13 mutants are partially suppressed in tom-1 unc-13 double mutants. These data indicate that in the intact nervous system, TOM-1 negatively regulates synaptic vesicle priming.


Introduction
Membrane fusion is mediated by the interactions of cognate SNARE (soluble NSF attachment protein receptor) proteins associated with vesicle and target membranes [1,2]. Synaptic vesicle exocytosis is a highly specialized form of membrane fusion in which calcium triggers fusion of synaptic vesicles with the plasma membrane, resulting in neurotransmitter release. Prior to vesicle fusion, the plasma membrane Q-SNAREs syntaxin-1a and SNAP-25 assemble with the vesicle-associated R-SNARE synaptobrevin-2 (a.k.a. VAMP-2) to form a stable coiled-coil complex known as the SNARE complex [3,4]. The assembly of the SNARE complex in trans is thought to bring the vesicle into close apposition with the plasma membrane, and may drive the fusion reaction [5]. Several synaptic proteins have been implicated in the regulation of this fusion process through their SNARE interactions, including the recently identified protein tomosyn [6].
Tomosyn is a 130 kDa soluble protein first isolated from rat cerebral cytosol as a syntaxin-binding partner capable of disrupting Munc18-syntaxin-1a complexes [6]. There are two paralogous tomosyn genes in the mammalian genome (tomosyn-1 and À2) that give rise to seven tomosyn isoforms through differential splicing [7,8]. All mammalian tomosyn isoforms have two recognizable domains, an N-terminal domain rich in WD40 repeats and a C-terminal SNARE domain with high sequence homology to the R-SNARE domain of synaptobrevin [9,10]. WD40 repeats are known to form beta propellers that act as protein interacting modules, although binding partners of the tomosyn WD40 repeats have yet to be identified. The tomosyn R-SNARE domain interacts with syntaxin and SNAP-25 to form a tomosyn SNARE complex that does not contain synaptobrevin but still binds to the putative calcium sensor synaptotagmin [6]. The biophysical properties of the tomosyn SNARE complex resemble those of the SNARE complex: both form at similar rates, have strong hysteresis during folding/unfolding transitions, exhibit alpha helicity, and are disassembled by NSF [10]. Consistent with these properties, the crystal structure of the core tomosyn SNARE complex reveals a four-alpha helical arrangement between the SNARE domains of tomosyn, SNAP-25, and syntaxin that is similar to the structure of the synaptobrevin-containing SNARE complex [11]. As predicted from the similarity in crystal structures, synaptobrevin is unable to displace tomosyn from the tomosyn SNARE complex (and vice versa), without prior NSF disassembly [11]. These data imply that formation of tomosyn SNARE complexes may preclude synaptobrevin-containing SNARE complex assembly, and therefore negatively regulate vesicle exocytosis. Consistent with this model, overexpression of vertebrate tomosyn reduces depolarization-induced dense-core granule fusion from PC12 cells [6,10], chromaffin cells [12], insulin-mediated exocytosis of GLUT4-containing vesicles from adipocytes [13], insulin release from beta cells [14], and synaptic transmission in cultured superior cervical ganglion neurons [15]. Although these data support a negative regulatory role for tomosyn in vesicle fusion, tomosyn RNA interference (RNAi) experiments have yielded mixed results. In cultured superior cervical ganglion neurons, tomosyn RNAi inhibited evoked release [15], whereas tomosyn RNAi in mouse beta cells enhanced exocytosis [14].
To address the role of tomosyn in synaptic transmission, in this study we directly assayed the physiological phenotype of tomosyn loss-of-function mutants at the Caenorhabditis elegans neuromuscular junction (NMJ). Our results indicate that tomosyn inhibits synaptic transmission through actions that regulate the size of the readily releasable vesicle pool.

tom-1 Encodes the C. elegans Tomosyn Homolog
The C. elegans genome encodes a single tomosyn gene [16], tom-1, that shares significant identity (;33%) with isoforms of mammalian tomosyn-1 and tomosyn-2 ( Figure 1A). Like vertebrate tomosyn, C. elegans tom-1 encodes multiple alternatively spliced isoforms, TOM-1(A,B,C) which share the Cterminal coiled-coil motif resembling the R-SNARE domain of C. elegans synaptobrevin (SNB-1) ( Figure 1B). The TOM-1B isoform is much smaller than either TOM-1A or TOM-1C, and lacks the N-terminal WD40 repeats. To examine whether the common C-terminal coiled-coil motif of TOM-1 (TOM-1Ct) interacts with syntaxin and SNAP-25, we performed in vitro pull-down assays using recombinant C. elegans proteins. We compared the behavior of synaptobrevin and TOM-1Ct in complex assembly assays with syntaxin-GST (UNC-64) [17] and SNAP-25 (RIC-4) [18]. Both SNB-1 and TOM-1Ct formed complexes, but did so inefficiently ( Figure 1C). Systematic replacement of each protein in the SNARE complex assay revealed that all assays containing the C. elegans SNAP-25 formed inefficiently, while all other mixed species formed complexes efficiently ( Figure S1). When the TOM-1Ct was mixed with C. elegans syntaxin-GST and vertebrate SNAP-25, syntaxin-GST co-precipitated TOM-1Ct and SNAP-25 ( Figure  1D). TOM-1Ct did not form a stable binary complex with syntaxin ( Figure S2). TOM-1Ct formed complexes as efficiently as SNB-1 forms complexes, but the TOM-1-containing complex was SDS sensitive, while the SNB-1-containing complex was SDS resistant. Density traces of the Coomassie blue-stained protein incorporated into the mixed-species TOM-1 complex gave densitometry ratios of 1:1.3:1.2 when divided by the molecular mass for each of the fusion proteins ( Figure 1D), consistent with the 1:1:1 stoichiometry reported for the vertebrate tomosyn complex [10]. These data confirm that C. elegans TOM-1 has the ability to form pseudo-SNARE complexes similar to those of vertebrate tomosyn.
To directly test whether mutations in tom-1 alter synaptic transmission, we recorded synaptic responses from the NMJs of dissected worms. Evoked responses were elicited by applying a depolarizing stimulus to the ventral nerve cord and recorded from voltage-clamped postsynaptic body wall muscles. In 5 mM Ca 2þ saline the evoked current amplitudes were not significantly altered in tom-1 mutants (2,327 6 107 pA, n ¼ 47, for WT vs. 2,443 6 136 pA, n ¼ 37, for tom-1(ok285), p ¼ 0.5; and 2,560 6 283 pA, n ¼ 11, for tom-1(nu468), p ¼ 0.37) (Figure 2A and 2B). However, the charge integral, a measure of the total ion flux during the evoked response (18.1 6 0.97 pC, n ¼ 49, for WT vs. 38.2 6 2.7 pC, n ¼ 30, for tom-1(ok285), p , 0.0001; and 42.6 6 5.6 pC, n ¼ 8, tom-1(nu468), p , 0.0001) ( Figure 2B), was greatly increased in tom-1 mutants due to a prolonged postsynaptic response (half-time-evoked decay: 4.6 6 0.17 ms, n ¼ 47, for WT vs. 12 6 1.1 ms, n ¼ 30, for tom-1(ok285), p , 0.0001; and 12.3 6 1.7 ms, n ¼ 8, for tom-1(nu468), p , 0.0001) ( Figure 2B). The enhanced evoked charge integral of tom-1 mutants in the absence of an increase in evoked amplitude could reflect saturation of the postsynaptic receptor field under the relatively high calcium (5 mM) recording conditions used. To test this possibility, we measured the evoked responses of tom-1 mutants in 0.5 mM Ca 2þ . In lower calcium, the evoked response amplitudes of tom-1 mutants were still comparable to the WT (1,100 6 108 pA, n ¼ 13, for WT vs. 1,410 6 178 pA, n ¼ 7, for tom-1(ok285), p . 0.1; and 1,533 6 202 pA, n ¼ 8, for tom-1(nu468), p . 0.05); however, the response duration remained significantly prolonged (half-time decay: 2.7 6 0.17 ms for WT vs. 4.2 6 0.57 ms for tom-1(ok285), p ¼ 0.016; and 4.2 6 .37 ms for tom-1(nu468), p ¼ 0.0014), again resulting in increased charge integrals (6.9 6 0.82 pC for WT vs. 10.7 6 1.2 pC for tom-1(ok285), p ¼ 0.018; and 13.3 6 2 pC for tom-1(nu468), p ¼ 0.006). These data suggest that the enhanced release observed in tom-1 mutants is due primarily to a prolongation of the evoked response. elegans Tomosyn Homolog (A) Gene structure of the three C. elegans TOM-1 isoforms. All isoforms were confirmed by expressed sequence tags 39 to the 6th exon and the upstream 59 region common to TOM-1A, and TOM-1C was determined by 59 RACE. The two long isoforms TOM-1A and TOM-1C contain N-terminal WD40 repeats and a C-terminal SNARE domain. The short isoform (TOM-1B) contains only the SNARE domain. tom-1(nu468) is a G to A change in W212 resulting in an early stop predicted to disrupt isoforms TOM-1A and TOM-1C [16]. The mutation in tom-1(ok285) is a 1,580-bp deletion that removes part of exon 10 and all of exons 11 through 13. tom-1(ok285) disrupts TOM-1A but, by RT-PCR can produce a mRNA capable of encoding an alternative isoform of TOM-1C lacking exons 11-13 and containing a partial exon 10 and an extra eight amino acids. (B) Amino acid alignment of the R-SNARE domain of rat tomosyn-1, mouse tomosyn-2, C. elegans TOM-1, and C. elegans synaptobrevin-1. Identity is shown as black boxes and similarity as gray boxes. Numbers below indicate helical layers formed during SNARE complex assembly. (C) TOM-1Ct forms tomosyn SNARE complexes with syntaxin and SNAP-25 in vitro to the same extent as C. elegans synaptobrevin (SNB-1) forms synaptobrevin SNARE complexes. C. elegans syntaxin::GST, or GST alone was incubated with SNAP-25-His6 and either His6T7-tagged TOM-1Ct, or His6T7-tagged SNB-1. Complexes were isolated using glutathione agarose beads. SDS resistance of complexes was assayed by heating half of the pulldown before electrophoresis. Proteins were separated on SDS-PAGE and transferred to nitrocellulose and detected with anti-GST, anti-His6, and anti-T7 antibodies. Since evoked responses under our recording conditions largely reflect cholinergic transmission, we next asked whether the prolonged evoked response of tom-1 mutants could be rescued by expressing TOM-1 protein specifically in cholinergic neurons. Recordings from tom-1(nu468) mutants possessing an integrated TOM-1A cDNA array (jaIs1052) reduced the excitatory postsynaptic current amplitude when compared to both WT or tom-1(nu468) (1,094 6 105 pA, n ¼ 8, for jaIs1052, p 0.0006, relative to tom-1(nu468)) and reversed the increased response duration to WT levels (4.7 6 0.74 ms, n ¼ 7, for jaIs1052, p ¼ 0.0019, relative to tom-1(nu468)), resulting in a significant decrease in total charge integral (9.2 6 1.3 pC, n ¼ 7) relative to both WT (p ¼ 0.0001) and tom-1(nu468) (p , 0.0001) (Figure 2A and 2B). We attribute the decrease in release in jaIs1052 relative to WT to overexpression of TOM-1A based on real-time PCR measurements of mRNA levels that indicate tom-1A mRNA levels are 11 6 1.3 (n ¼ 3) times that of the WT. Together, these data suggest that TOM-1A is required presynaptically to regulate synaptic transmission. This observation is consistent with the neuron-specific expression pattern of TOM-1 based on a tom-1 promoter::GFP fusion construct [16].

Neuronal Architecture of tom-1 Mutants
Vertebrate tomosyn has been implicated in the regulation of neurite outgrowth in mammalian cultured neurons [19]. Therefore, we next addressed whether the enhanced evoked release in C. elegans tom-1 mutants could be due to aberrant neuronal connectivity. To examine the neuronal cytoarchitecture of tom-1 mutants, we crossed in an integrated array (juIs14) expressing cytoplasmic green fluorescent protein (GFP) under the cholinergic neuron specific promoter Pacr-2 [20]. Analysis of axonal left/right orientation, axon targeting, and axon fasciculation revealed no differences in the neuronal morphology of tom-1(ok285);juIs14 relative to the WT ( Figure 3A and 3B), suggesting that the tom-1 phenotype is not associated with any discernable innervation defects.

Ultrastructural Analysis of tom-1 Mutants
To test whether the enhanced release in tom-1 mutants resulted from increased synaptic vesicle biogenesis or altered vesicle distribution within synapses, we examined synaptic ultrastructure by electron microscopy (EM) (Figure 4). Recent advances in high-pressure freeze and freeze substitution fixation techniques to prepare worms for EM have greatly improved the quality of ultrastructural data obtained in C. elegans [22]. The advantage of this technique over conventional fixation is that by virtue of the instantaneous freezing of the worms and gradual substitution with fixative, there is less fixation-induced osmotic shock that normally results in cell shrinkage and possible redistribution of synaptic vesicles to the plasma membrane. Using this technique, we found that while the average number of vesicles per synaptic profile did not differ from the WT in tom-1(ok285) mutants (17.45 6 0.86 vesicles/profile, n ¼ 86 profiles, for WT vs. 17 6 0.81 vesicles/ profile, n ¼ 74 profiles, for tom-1(ok285)), vesicle localization was affected. Specifically, tom-1(k285) mutants exhibited a dramatic and significant (p , 0.0001) increase in the number of vesicles contacting the plasma membrane (15.6% 6 0.75%, n ¼ 74 synaptic profiles, for tom-1(ok285) vs. 8.5% 6 0.3%, n ¼ 250 synaptic profiles, for WT) ( Figure 4A and 4B). In WT synapses, contacting vesicles were preferentially localized within ;150 nm of the presynaptic density ( Figure 4C and 4D), whereas in tom-1(ok285) mutants the increase in membrane-contacting vesicles were distributed throughout the terminal ( Figure 4C and 4D).
We next examined the ultrastructure of jaIs1052 ( Figure 5). Since TOM-1A expression was restricted to the cholinergic motor neurons of tom-1(nu468) mutants in jaIs1052, we compared the contacting vesicle pool of cholinergic synapses in jaIs1052 versus tom-1(nu468) mutants ( Figure 5A and 5B). Consistent with the reduced evoked response, cholinergic synapses in jaIs1052 had fewer contacting vesicles relative to tom-1(nu468) (7.9% 6 0.4% , n ¼ 56 profiles, for jaIs1052 vs. 15.8% 6 0.6 %, n ¼ 49 profiles, for tom-1(nu468), p , 0.0001) ( Figure 5B). As an internal control for the specificity of the TOM-1A rescue, we also examined the number of contacting vesicles in the GABAergic synapses of jaIs1052 worms ( Figure  5D and 5E). The number of contacting vesicles was not significantly different (p . 0.5) between jaIs1052 GABA synapses (11.4% 6 1%, n ¼ 20) and tom-1(nu468) (13.7% 6 0.8 %, n ¼ 19 profiles) ( Figure 5E). Similarly, the distribution of vesicles in cholinergic synapses was reduced throughout the terminal in jaIs1052 cholinergic synapses ( Figure 5C), but not in the GABAergic synapses of the same worms ( Figure 5F). These data establish that vesicle contact is regulated by TOM-1 in both cholinergic and GABAergic synapses in C. elegans. Since the formation of tomosyn SNARE complexes competes with the assembly of synaptobrevin-containing SNARE complexes, it has been proposed that tomosyn may inhibit synaptic vesicle priming. Therefore, the enhanced release and increased contacting vesicle pool in tom-1 mutants could reflect an increase in the size of the readily releasable pool. To directly measure the size of the readily releasable pool in tom-1 mutants, we recorded hyperosmotic responses at the neuromuscular junction. The hyperosmotic responses of both tom-1 alleles were significantly increased relative to the WT (18.4 6 2.8 pC, n ¼ 13, for WT vs. 34.6 6 5.1 pC, n ¼ 6, for tom-1(n468), p ¼ 0.0125; and 28.8 6 4.8 pC, n ¼ 10 for tom-1(ok285), p ¼ 0.049) (Figure 6), indicating that loss of TOM-1 results in an increased primed vesicle pool.
This reduction is consistent with the proposed role of UNC-13 in promoting SNARE complex formation and thus, vesicle apposition with the plasma membrane.

Summary
We have examined the role of tomosyn in regulated synaptic transmission by analyzing two hypomorphic mutants of the C. elegans tomosyn homolog, tom-1 [16]. Our results demonstrate that loss of TOM-1A and C isoforms enhances evoked release at the neuromuscular junction. In the absence of a complete tom-1 null, the role of the remaining isoform, TOM-1B, remains to be elucidated. These results are consistent with a previously reported increase in sensitivity of tom-1 mutants to the toxic effects of the acetylcholine esterase inhibitor, aldicarb, which is indicative of increased cholinergic transmission [16]. The enhanced neurotransmission observed electrophysiologically correlates with a redistribution of synaptic vesicles to the plasma membrane in tom-1 mutant synapses. We further demonstrate that the ability of synaptic vesicles to contact the plasma membrane is regulated by the priming factor, UNC-13. When we generated tom-1 unc-13 double mutants we observed partial suppression of the unc-13 priming defect and concomitant restoration of the plasma membrane-contacting vesicle pool. Consistent with previous studies of dense core vesicle fusion [6,10,12,13] and insulin release [14], these data suggest that tomosyn acts as a negative regulator of synaptic vesicle priming.

What Is the Molecular Mechanism by which Tomosyn Negatively Regulates Vesicle Priming?
Priming is thought to involve formation of SNARE complexes between synaptobrevin, syntaxin, and SNAP-25, which bring the vesicle membrane into close apposition with the plasma membrane [5,23]. Like vertebrate tomosyn [10], we demonstrate that the C-terminal of C. elegans TOM-1 also has the ability to form a complex with the SNARE domains of syntaxin and SNAP-25, suggesting this is a conserved tomosyn interaction. Vertebrate tomosyn competes with synaptobrevin for binding to syntaxin and SNAP-25 on the cytosolic surface of PC12 cell membrane sheets [10] as well as in in vitro biochemical studies [6]. By precluding vesicle-associated synaptobrevin from assembly into SNARE complexes, tomosyn is thus proposed to limit vesicle priming. This model is supported by several studies in which tomosyn overexpression has been shown to inhibit dense-core granule fusion [6,10,12] and neuronal exocytosis [15]. Furthermore, in Figure 6. tom-1(ok285) Suppresses the Synaptic Defects of unc-13(s69) Mutants (A) Representative images of unc-13(s69) and tom-1(ok285)unc13(s69). (B) The ratio of plasma membrane-contacting vesicles per profile for WT (black), unc-13(s69) (red), tom-1(ok285)unc13(s69) (maroon), and tom-1(ok285) (purple) scale bar ¼ 200 nm. (C) Comparison of unc-13(s69) and tom-1(ok285)unc-13(s69) plasma membrane-contacting vesicle distribution relative to the PD in 30-nm bins. (D) Representative NMJ recordings demonstrate that the evoked response absent in unc-13(s69) mutants is partially restored in tom-1(ok285)unc13(s69) double mutants. The average evoked charge integral of the tom1 unc-13 double mutants is graphed relative to WT and tom-1(ok285) mutant responses. (E) Representative recordings of hyperosmotic responses demonstrate that the readily releasable pool of vesicles is increased in tom-1(ok285) relative to WT. The hyperosmotic response absent in unc-13(s69) mutants is also partially restored in tom-1(ok285)unc13(s69) double mutants. The mean total charge integral for synaptic events in the first second of the hyperosmotic response is graphed. Data expressed as mean 6 SEM. DOI: 10.1371/journal.pbio.0040261.g006 chromaffin cells, tomosyn overexpression specifically inhibits the fast exocytotic burst corresponding to the fusioncompetent primed granule pool [12]. In contrast to overexpression data, RNAi-mediated knockdown of tomosyn has produced mixed results. In cultured neurons tomosyn RNAi causes a reduction in synaptic vesicle release [15], where as RNAi in beta cells results in enhanced exocytosis [14]. This former study follows a previous report indicating that tomosyn RNAi profoundly inhibits directed neurite outgrowth, which could affect synaptogenesis under these culture conditions, resulting in fewer functional synapses [19]. In the context of an intact animal examined here, loss of tomosyn function does not appear to be deleterious for neurite outgrowth and synaptogenesis. Therefore, the enhanced neurotransmission we observe in C. elegans tom-1 mutants is more consistent with RNAi in beta cells and appears to be attributable to functional changes in release properties rather than aberrant neuronal cytoarchitecture. Specifically, phenotypic analysis of C. elegans tom-1 mutants supports the hypothesis that tomosyn acts as a negative regulator of synaptic vesicle priming in vivo.
In contrast to the C. elegans tom-1 mutant ultrastructural phenotype, tomosyn overexpression in chromaffin cells was not associated with any changes in granule distribution based on conventional EM [12]. In TOM-1A-overexpressing synapses, we also observe near-normal numbers of contacting vesicles. Therefore, there is no conflict between the ultrastructural data reported for tomosyn overexpression and that of the present study. However, it remains to be seen whether loss of tomosyn in chromaffin cells is associated with increased granule association with the plasma membrane.
UNC-13 is a member of a conserved family of synaptic proteins implicated in vesicle priming [24][25][26]. The mammalian homolog Munc13-1 is thought to interact with the Nterminus of syntaxin [27]. Syntaxin can adopt a closed state in solution that occludes the syntaxin SNARE domain (the H3 domain) required for SNARE complex formation [28]. UNC-13 binding to the N-terminus of syntaxin has been proposed to stabilize the syntaxin open configuration, increasing accessibility of the H3 domain [27]. Partial suppression of the C. elegans unc-13 priming defect by overexpression of a constitutively open form of syntaxin supports this model [29]. Furthermore, mutations that disrupt UNC-13/syntaxin interactions in vitro have been shown to reduce release in both C. elegans and chromaffin cells [30,31]. However, a Munc13 domain capable of partially rescuing priming in Munc13 KO mice, fails to bind syntaxin [32], suggesting alternative molecular mechanisms could also account for the priming function of Munc13. Here we show that tom-1(ok285) can also partially suppress the unc-13(s69) mutant phenotype. We postulate that in unc-13 mutants, any transitions of syntaxin to the open state are prevented from forming functional SNARE complexes through the binding of tomosyn. It is also possible that the presence of UNC-13 normally excludes tomosyn from interacting with syntaxin and SNAP-25. In either case, we propose that in tom-1 unc-13 double mutants, syntaxin has an increased probability of assembling into SNARE complexes because tomosyn no longer precludes synaptobrevin binding to the plasma membrane SNAREs.
Why Is the Postsynaptic Response Prolonged in C. elegans tom-1 Mutants?
Many of the contacting synaptic vesicles in tom-1(ok285) mutants are found distal to the presynaptic specialization. If, as at vertebrate NMJs [33], calcium channels are localized to the presynaptic specialization in C. elegans, calciumtriggered fusion of distal vesicles may be delayed relative to the proximal vesicles, resulting in a prolonged muscle response. It is also possible that the increased distance between distal neurotransmitter release sites and postsynaptic receptors prolongs the postsynaptic response in tom-1 mutant muscles.

How Might Tomosyn Restrict the Membrane Localization of Synaptic Vesicles?
One mechanism by which tomosyn may restrict synaptic vesicle fusion has recently emerged from the study of neurite outgrowth following tomosyn RNAi [19]. In cultured hippocampal neurons, knockdown of tomosyn increases neurite sprouting and branching while reducing neurite extension. This regulation is proposed to involve a highaffinity interaction between tomosyn and phosphorylated syntaxin (Ser14), which colocalize to the palms of growth cones [19]. The phosphorylation of syntaxin (Ser14p) by serine/threonine kinase (ROCK) leads to a five-fold increase in binding of tomosyn relative to synaptobrevin. Synaptobrevin-associated vesicles are thus predicted to be excluded from priming in growth cone palms, fusing instead at the leading edge of the growth cone, promoting neurite extension. Whether a similar mechanism functions to spatially restrict synaptic vesicle fusion to the active zone region in synaptic terminals remains to be investigated. However, syntaxin Ser14p is found throughout the rat cortex and appears to be excluded from regions rich in synaptic vesicles [34]. Therefore, it seems entirely possible that tomosyn may act to restrict exocytosis at synaptic terminals. Since syntaxin can be phosphorylated at Ser14 by both ROCK [19] and casein kinase II [34], this could also present a possible mechanism to regulate synaptic efficacy by affecting the level of tomosyn-mediated inhibition of vesicle priming.
The interaction between tomosyn and syntaxin is also regulated by protein kinase A-dependent phosphorylation of tomosyn in the variable linker between the WD40 domain and the C-terminal SNARE homology domain. Specifically, phosphorylation of tomosyn reduces the binding affinity of the tomosyn-syntaxin interaction [15]. Furthermore, protein kinase A-dependent synaptic facilitation appears to act in part through the phosphorylation of tomosyn, suggesting again that the interaction between tomosyn and syntaxin negatively regulates neurotransmitter release.
In summary, tomosyn interacts with syntaxin and SNAP-25 to form tomosyn SNARE complexes predicted to be nonfusogenic, which compete with SNARE complex formation and thus inhibit vesicle priming. Tomosyn perturbation analysis in C. elegans provides the first in vivo evidence in support of this model. Tomosyn mutants not only exhibit enhanced release but also suppress the priming defects of unc-13 mutants. Future studies can now address whether the regulation of tomosyn function by kinases provide mechanisms for synaptic plasticity.
In vitro biochemistry. Purification of recombinant proteins SNAREs lacking TM domains. Vectors for C. elegans SNAP-25 and syntaxin were constructed as follows: the entire SNAP-25 coding region was inserted into pHO4d [35], the syntaxin cytoplasmic domain (aa 1-266) was cloned into pGEX-2T, and the SNB-1 cytoplasmic domain (aa 1-88) was cloned into pRSETC. The TOM-1 SNARE motifcoding region (TOM-1Ct; aa 1,137-1,211) was subcloned into pRSETC and purified as an N-terminally his6-tagged protein. Rat SNAP-25-His6 purified as described in [5] was provided by Phyllis Hanson (Washington University School of Medicine, Saint Louis, Missouri, United States). Proteins were purified using Ni-NTA resin or glutathione agarose, followed in some cases by further purification on a Mono S column using fast protein liquid chromatography.
Methodology. Fusion proteins were batch-purified on Ni-NTAagarose according to the manufacturer's protocols (Qiagen, Valencia, California, United States) under native conditions with modified buffers. The lysis buffer consisted of 20 mM HEPES (pH 7.4), 500 mM KCl, 0.1 mM PMSF, 0.1% ßME, 5% glycerol; and 0-5 mM imidazole. The wash and elution buffer consisted of 20 mM HEPES (pH 7.9), 200 mM KCl, 0.1 mM PMSF, 0.1% ßME, 5% glycerol, 0.5% Triton X-100, and 1-300 mM imidazole. The dialysis buffer consisted of 10 mM HEPES (pH 7.9), 140 mM KCl, and 0.5% Triton X-100. The syntaxin-GST fusion protein and GST alone were batch-purified with glutathione agarose under native conditions [36]. The buffers used differed from those used in the Ni-NTA-agarose purifications only as follows: the lysis buffer contained no imidazole and 100 mM EDTA; the wash and elution buffer contained no imidazole, 2 mM EDTA, and 15 mM reduced glutathione. The purity of the fractions was verified by SDS-PAGE, and the concentration was determined by the Bradford assay.
TOM-1 complex formation assays. For in vitro complex formation, ;1.6 lM of each of C. elegans syntaxin-GST (or GST), rat SNAP-25-His 6 , and C. elegans His 6 -TOM-1Ct (or -His 6 -SNB-1) were incubated from 2 h to overnight at 4 8C with rocking in D-buffer. The proteins were added to an aliquot of pre-equilibrated glutathione-agarose slurry and incubated at 4 8C for 1 h with rocking. The matrix was washed four times with 400 ll D-buffer. SDS-sample buffer was added, and half the reaction was heated (95 8C for 2 min) before analysis on SDS-PAGE. Proteins were detected by Coomassie Blue staining. D-buffer consisted of 10 mM HEPES (pH 7.9), 140 mM KCl, 0.5% Triton X-100, and 0.1 mM PMSF. Integrated densities were calculated using ImageJ (rsb.info.nih.gov/ij/). Complexes assembled using C. elegans SNAP-25-His 6 were performed similarly, but after transfer to nitrocellulose, proteins were detected by immunoblotting using anti-GST, anti-His 6 (SNAP-25, TOM-1Ct, and SNB-1), and anti-T7 (TOM-1Ct and SNB-1) antibodies.
Behavioral assays. Thrashing assays in M9 solution were performed by quantifying the number of body bends/min. For head tap responses, worms were allowed a 1-min settling period after placement in the assay chamber with food, before recording responses to a head tap for 20 s. Brood size was quantified by plating ten L4 hermaphrodites of each genotype onto individual plates and counting the number of progeny by removing hatched larvae and embryos for 3-4 d. Defecation rate was scored under a dissecting microscope using ten young adult worms for each genotype. For each worm, the duration of three consecutive defecations was measured and averaged. The cycle was defined as an interval between two expulsions of the gut content after visible contraction of the body wall muscles.
Real-time PCR. C. elegans total RNA was isolated using a Trizol reagent as described by the manufacturer (Invitrogen, Carlsbad, California, United States). Total RNA (0.5 lg) was used for reverse transcription using the SuperScript III First Strand Synthesis Kit (Invitrogen). Absence of genomic DNA was confirmed by PCR with primers designed to amplify the GFP coding region. GFP is a high copy coinjection marker, and we did not detect any genomic DNA contamination. For amplification of tomosyn isoform A, PCR primers were designed to amplify a unique exon. Dynamin was used as a reference for calibration. Real-time PCR was then performed by fluorescent detection and quantification of SYBR green-labeled PCR product using MJ Research Opticon2 real-time thermocycler (Bio-Rad, Hercules, California, United States). For quantification of relative initial transcript levels, the cycle threshold (C t ) values for isoform A and a dynamin control were determined for each sample. The amount of tom-1A mRNA is reported as a ratio relative to the calibrator (dynamin). After RT-PCR, the products were subjected to agarose gel electrophoresis and sequencing to confirm the specificity of amplified products.
Immunohistochemistry. Immunohistochemistry was performed using a modified method of the whole worm fixation [37]. Briefly, the worms were frozen on dry ice between two slides, which were then split apart. Frozen animals were fixed in 4% slushy formaldehyde in PBS. Following fixation, animals were spun down at 1,600 rpm. Antibodies against UNC-29 (kindly provided by Dr. K. G. Miller) and UNC-17 (kindly provided by Dr. J. B. Rand) were used at a final dilution of 1:200 in PBS and 0.5% Triton X-100 with 3% BSA. Antirabbit and anti-mouse TRITC-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, Pennsylvania, United States) were used at a 1:500 dilution for 4 h at 4 8C. Images were obtained with a 603 objective using an Olympus Optical (Tokyo, Japan) FV-500 laser-scanning confocal microscope. Puncta size and quantity were quantified from image projections of the ventral nerve cord region upstream of the vulva using ImageJ.
Generation of TOM-1A rescuing lines. jaIs1052 strains expressing an integrated array of TOM-1A in tom-1(nu468) mutants were generated by irradiating animals expressing an extrachromasomal array of TOM-1A cDNA and a Pmyo-2::gfp coinjection marker under the cholinergic neuronal unc-17 promoter (kindly provided by Dr. J. Kaplan) using a Cs source (total dose of 1,800 rad).
Freeze substitution was performed in a Reichart AFS machine (Leica, Oberkochen, Germany) as previously described [38] using tannic acid (0.1%) and 0.5% gluteraldehyde fixative introduced over 4 d followed by 2% osmium. Fixed animals were then washed and embedded in Araldite 502 over a 48-h period at 60 8C.
Serial sections were cut at a thickness of 40-50 nm, collected on formvar-covered carbon coated copper grids (EMS, FCF2010-Cu), and counterstained in 2% or 2.5% aqueous uranyl acetate for 4 min, followed by Reynolds lead citrate for 2 min. Images were obtained on a JEOL JEM-1220 transmission electron microscope (JEOL, Tokyo, Japan) operating at 80 kV. Micrographs were collected using a Gatan digital camera (Pleasanton, California, United States).
Morphometric analysis of WT, tom-1(ok285), tom-1(nu468), tom-1(ok285)unc-13(s69), unc-13(s69), and (jaIs1052) TOM-1A rescued animals were performed from ventral nerve cord serial sections. The analysis was performed blind. Images were quantified using NIH Image software. A synapse was defined as a set of serial sections containing a presynaptic specialization and two flanking sections from both sides without presynaptic specialization. Several morphometric measurements were obtained: the number of vesicles per profile, the distance from each vesicle membrane perpendicular to the plasma membrane, and the distance to the proximal edge of the presynaptic specialization. Figure S1. Coomassie Blue-Stained PAGE Gel of Mixed-Species SNARE Complex Assembly Assays Complex assays were performed using C. elegans His 6 -tagged SNAREs and rat His 6 -tagged SNAREs. Purified syntaxin, SNAP-25, and synaptobrevin were mixed, incubated to allow complex formation, and separated on a PAGE gel to visualize complexes. One aliquot of each sample was loaded without heating (rt), and the other was heated to 95 8C for 5 min (b) prior to loading. The position of complexes, syntaxin, SNAP-25, and C. elegans synaptobrevin (which migrates just above the dye front) are labeled on the gel. Unfortunately, vertebrate (v) synaptobrevin migrates in the dye front and is not identifiable on the gel. Various combinations of C. elegans (c) and rat protein (v) mixes were tried. All reactions containing purified C. elegans SNAP-25 failed to form complexes efficiently, while all other combinations form complexes. The positions of molecular weight markers are on the left. Methods: C. elegans His 6 -tagged SNAREs and rat SNAP-25 were purified as described in the main text. Rat His 6 -tagged synaptobrevin and syntaxin were a gift from Michael Crowder (Washington University School of Medicine) and purified as described in [39]. To form complexes, purified syntaxin, SNAP-25, and synaptobrevin were mixed, incubated overnight at 48 C, mixed with SDS-containing sample buffer, and run on a 4%-12% PAGE gel. Found at DOI: 10.1371/journal.pbio.0040261.sg001 (292 KB JPG). Figure S2. C. elegans TOM-1Ct Does Not Form a Stable Binary Complex with Syntaxin A Coomassie-stained gel shows GST pull-down assays performed to eliminate the possibility that TOM-1Ct forms a complex with syntaxin in the absence of SNAP-25. C. elegans GST syntaxin or GST alone was incubated with either rat SNAP-25 (vSNAP-25) or C. elegans tomosyn C-terminal domain (TOM-1Ct). Glutathione agarose beads were added to collect complexes. The supernatant was removed, and the beads were washed extensively. Aliquots of the supernatant (sup) and the protein eluted from the glutathione agarose (beads) were separated on SDS-PAGE and stained for protein with Coomassie Blue. Although syntaxin bound to SNAP-25 in the absence of TOM-1Ct or synaptobrevin, TOM-1Ct did not interact with syntaxin in the absence of SNAP-25. Molecular weights of markers are labeled on the left. Methods: proteins and GST pulled-down assays were performed as described in the main text.