Complexin maintains vesicles in the primed state in C. elegans

The cpx-1 isoform functions at the nematode neuromuscular junction

Two complexin genes, cpx-1 and cpx-2, are predicted in the C. elegans genome (Figure 1a, S1, S2). Worm cpx-1 exhibits greater identity to the fly and mouse complexins than cpx-2 (39% identity between CPX-1 and mouse Complexin-1 versus 32% identity between CPX-2 and mouse Complexin-1, (Figure 1, S2). For cpx-1, highest conservation is observed in the central helical domain (75%), which binds the SNARE complex, whereas the N-terminal facilitory domain, the accessory domain and the C-terminal domains each exhibit about 30% identity. The domains of complexin are defined based on sequence conservation across multiple animal phyla (See Figure S1 legend for full description of how the domains were defined). To assay for complexin function at neuromuscular junctions, we analyzed mutations in cpx-1 and cpx-2. cpx-1(ok1552) deletes the first of two exons of the gene which includes the N-terminus, the accessory domain, the central helix and part of the C-terminal domain. cpx-2(tm1871) deletes part of the accessory domain, the central helix and part of the C-terminal domain of cpx-2 (Figure 1b, S2). Complexins in mice and Drosophila are required for viability; complexin-1 complexin-2 double knockout mice die a few hours after birth and complexin null flies die before adult eclosion [17,19]. By contrast, C. elegans cpx-1(ok1552) and cpx-2(tm1871) mutants are viable and grow to adulthood; similarly cpx-1(ok1552) cpx-2(tm1871) double mutants develop into fertile adults.

Nevertheless, cpx-1(ok1552) mutants are uncoordinated. To quantify body movement, we performed thrashing assays. Animals placed in a drop of water will swim by rapid reciprocal contractions of the body; locomotory activity can be quantified by counting flexures. Wild-type animals are capable of maintaining high rates of thrashing, whereas cpx-1(ok1552) worms exhibit reduced thrashing in liquid (Figure 1c). cpx-2(tm1871) mutants exhibit no swimming defect, suggesting that cpx-1, but not cpx-2, is important for the regulation of locomotion (Figure S2). To test whether loss of complexin results in defective neurotransmission, we tested the sensitivity of the complexin mutants to the acetylcholinesterase inhibitor aldicarb. Loss of cpx-1, but not cpx-2, resulted in hypersensitivity to aldicarb, suggesting there is excessive neurotransmitter release at the neuromuscular junction (Figure 1d; S2). The thrashing and aldicarb hypersensitivity phenotypes of cpx-1 mutants can be fully rescued by expressing the cpx-1 gene under its own promoter or under a pan-neuronal promoter, demonstrating that complexin is functioning in the nervous system (Figure 1d, S3). The aldicarb hypersensitivity defect can also be rescued by expressing cpx-1 in just the acetylcholine neurons demonstrating that excess neurotransmitter release arises from a cell-autonomous defect in the motor neurons (Figure S3a). However, thrashing was not rescued by expressing cpx-1 specifically in the acetylcholine motor neurons suggesting that cpx-1 is required more broadly in the nervous system for normal locomotion (Figure 1c). cpx-1 cpx-2 double mutants are not more severe than the cpx-1 single mutant for the thrashing or aldicarb assays demonstrating that cpx-1 plays the dominant role for the locomotory phenotypes (Figure S2). Together, these data suggest that cpx-1 functions in the motor neurons to prevent excess neurotransmitter release.

Complexins in other species are expressed mostly in neuronal tissue [4–6,19]. To determine the expression pattern of the C. elegans complexin genes, we tagged the N-termini of CPX-1 and CPX-2 with GFP. Each fusion construct includes the promoter, all introns and the native 3’ UTR (Figure 1b, S2). The cpx-1 transgenic reporter is widely expressed in the nervous system, including all the ventral cord motor neurons (Figure 1b). The CPX-1 protein is clustered into puncta along the dorsal and ventral nerve cords suggesting that the protein is likely localized to synapses (Figure 1b inset). Because the transgene rescues the cpx-1 aldicarb hypersensitivity phenotype, the tagged protein is functional (Figure S3). Since no obvious phenotype was detected for cpx-2 it was not possible to assay rescue. However the expression pattern suggests that cpx-2 is more narrowly expressed, including in a few neurons in the head and tail ganglia and in the coelomocyte scavenger cells, but not in the motor neurons of the ventral nerve cord (Figure S2). The expression patterns and behavioral phenotypes suggest that cpx-1 is the primary complexin functioning at neuromuscular junctions.

Loss of cpx-1 has contradictory effects on synaptic transmission

Loss of complexin results in a developmental defect in Drosophila neurons: the number of synaptic varicosities and active zones are nearly doubled [19]. It is has been suggested that the increase in tonic release leads to synaptic hypertrophy in Drosophila. To determine if there is a developmental defect at C. elegans synapses, we examined synaptic markers in cpx-1(ok1552) mutants. Normal numbers of neuromuscular junctions were observed in cpx-1 mutants as assayed by counting synaptic vesicle clusters (synaptobrevin-GFP) or dense projections (α-liprin/SYD-2::GFP) along the ventral nerve cord (Figure 2a, b). Normal postsynaptic receptor fields were observed in cpx-1 mutants by labeling the muscle GABA receptor (UNC-49::GFP; Figure 2c). Finally, axon outgrowth appeared normal in cpx-1(ok1552) mutants as determined by labeling the GABA motor neurons (Figure S3). It is possible that the synaptic phenotypes of cpx-1 mutants arose from long-term compensatory changes that occurred during development. To examine a developmental role for cpx-1 we expressed cpx-1(+) under the control of two different heat-shock promoters in cpx-1(ok1552) mutant animals (Figure S3). In the absence of heat shock, the transgenes failed to rescue the aldicarb hypersensitivity of cpx-1(ok1552); after heat shock in adult animals normal aldicarb sensitivity was restored. Taken together these results suggest that the development of the nervous system in a cpx-1 mutant is grossly normal and that the uncoordinated phenotype and hypersensitivity to aldicarb are due to defects in synaptic function rather than caused by an increase in synaptic number.

Hypersensitivity to aldicarb suggests excessive neurotransmitter release at the neuromuscular junction. Three mechanisms of neurotransmitter release could be responsible for the increase: evoked, tonic or spontaneous release. Evoked release is generated by an action potential or a depolarizing stimulus from an electrode. The transient depolarization causes an influx of calcium at the synapse and stimulates the fusion of vesicles in the readily releasable pool. The coherent release of neurotransmitter produces a large postsynaptic current. Tonic release occurs even in the absence of a stimulus. If the membrane potential of the neuron at rest is sufficiently depolarized, a low level of calcium influx leads to sporadic vesicle fusion and miniature postsynaptic currents. Note that tonic release still requires calcium. In the absence of extracellular calcium, vesicle fusion is very rare and is called spontaneous fusion.

We measured these three types of neurotransmitter release using voltage-clamp recordings from neuromuscular junctions in cpx-1(ok1552) mutants [21]. At 5 mM calcium, tonic release in cpx-1 mutants was the same as in the wild type (Figure 3a, b). Tonic release is sensitive to extracellular calcium, so mini frequencies were measured over a range of calcium concentrations (Figure 3b). To measure spontaneous release, we recorded minis in the absence of extracellular calcium. Note that even in the absence of extracellular calcium that some minis are observed in C. elegans; these residual minis however are dependent on calcium from internal stores [22]. Interestingly, the cpx-1(ok1552) mutants still exhibited a >40% increase in mini frequency compared to the wild type at 0 mM calcium. These results suggest that CPX-1 is required to block spontaneous calcium-independent fusion of synaptic vesicles.

To evaluate evoked release, we stimulated the ventral nerve cord with an electrode. Evoked release was reduced by 94% in cpx-1(ok1552) animals, demonstrating that the readily releasable pool was almost completely absent (Figure 3c, d). Once again, the defect in evoked release could be rescued by expressing cpx-1(+) in the nervous system of the mutant strain (Figure 5). These data suggest that in addition to an inhibitory role, complexin is playing a positive role in fusion.

Complexin plays dual roles in exocytosis

One possible interpretation of these data is that complexin plays one role at the synapse: it blocks spontaneous fusion of synaptic vesicles. In this model the absence of the evoked release is caused by a depletion of primed vesicles by spontaneous fusion. To assay for a depletion of docked vesicles, we examined the number and distribution of synaptic vesicles using electron microscopy. At a gross level the synapses seem normal in cpx-1 mutants: The morphology is normal (Figure 4a), synaptic vesicles have a normal diameter (Figure S4c) and vesicles are clustered around the dense projection (Figure S4a). However, the total number of synaptic vesicles is reduced to about 50% in cpx-1(ok1552) mutants compared to the wild type (Figure 4b). Moreover, the number of docked vesicles is severely reduced: there is a 70% reduction of docked vesicles at acetylcholine synapses and a 75% reduction of docked vesicles at GABA synapses (Figure 4c). Docking at the plasma membrane is reduced along the entire active zone (Figure S4b).

The ultrastructural data are superficially consistent with complexin having a single function: it blocks spontaneous fusion. If this model were true then one would expect that weak mutations would lead to an increase in spontaneous fusion and a concomitant decrease in evoked release; that is, the phenotypes should always be reciprocal. The first 15 amino acids of murine complexin are required for stimulating neurotransmission [12,23,24]. Specifically, the methionine and lysine residues at positions 5 and 6 in mouse complexin are believed to mediate the stimulatory effect [23] and these residues are conserved between mice and worms (Figure 1a). We deleted the 15 amino acids at the N-terminus of the complexin gene (‘ΔN-cpx-1’, Figure 5a). To insure that constructs were stable and not over-expressed, we inserted the transgenes as single copies into the cpx-1(ok1552) genome using Mos-mediated single copy insertion [25]. A cpx-1(+) gene expressed in neurons rescues the locomotory phenotypes, evoked responses and also inhibits the spontaneous fusion of vesicles (Figure 1c, d, Figure 5). The N-terminal deletion construct partially rescues the locomotory phenotype (Figure S5). Electrophysiological recordings demonstrate that spontaneous fusion is increased in the N-terminal deletion allele compared to the wild type (Figure 5b, c), suggesting that the N-terminus is required to inhibit spontaneous fusion. Surprisingly, the N-terminal deletion causes an increase rather than a decrease in evoked release (Figure 5d, e). These data tell us two important facts about complexin: (1) Unlike in the mouse, the N-terminal domain in the worm inhibits fusion, both evoked and spontaneous. (2) The negative and positive roles of complexin are separable. The N-terminal domain is required to inhibit fusion; the stimulatory functions must reside elsewhere. We conclude that complexin plays dual roles during vesicle fusion.

Blocking fusion

Our data demonstrate that the complexin protein arrests SNARE-mediated fusion. First, in dissected preparations, spontaneous vesicle fusion is increased at cpx-1 neuromuscular junctions. Second, increased exocytosis occurs in undissected animals as well: complexin mutant animals exhibit an increased sensitivity to aldicarb, a drug that blocks the degradation of acetylcholine. The increased rate of release is not simply an increase in tonic release. Tonic release depends on extracellular calcium; whereas spontaneous release is independent of extracellular calcium. Thus, complexin may inhibit the spontaneous fusion of vesicles from both the evoked and tonic pools. An inhibitory role for complexin in tonic fusion is consistent with observations in both flies and mice. Complexin mutants in Drosophila exhibit elevated rates of spontaneous fusion as measured by field potentials at the neuromuscular junction of third instar larvae [19]. Similarly, knockdown of complexin in mouse cortical neurons by RNA interference led to a four-fold increase in the rate of spontaneous fusion [12].

An increased consumption of synaptic vesicles should lead to a depletion of vesicles at mutant synapses. In C. elegans spontaneous fusion occurs at high enough levels to deplete both docked and cytoplasmic pools of vesicles as observed by electron microscopy. There is a 70% decrease in the number of vesicles docked at active zones in cpx-1 mutants, presumably because vesicles spontaneously fuse after the priming of the SNARE complex. The loss of complexin from the C. elegans neuromuscular junction even depletes the reserve pool of vesicles by 50%, indicating that the high rates of fusion are outpacing the ability of the cell to replenish vesicles at the synapse. Although an increase in spontaneous fusion is not observed in mouse knockouts, there is a reduction in the readily releasable pool, that is docked and primed vesicles, as measured by applying moderately hyperosmotic media [23].

The inhibitory role of complexin in synaptic vesicle fusion is borne out by biochemical studies of membrane fusion. Complexin can inhibit SNARE-mediated fusion of whole cells [10] and also block SNARE-mediated fusion of liposomes [9]. The block of SNARE-mediated fusion appears to involve the accessory domain of complexin which may prevent synaptobrevin from fully zippering the SNARE complex [9–11].

Stimulating exocytosis

Complexin is also required for calcium-stimulated exocytosis in C. elegans: evoked currents are reduced by 94% in the complexin mutant. These data are also consistent with data from flies and mice. In Drosophila, complexin mutants exhibit an approximately 30–40% decrease in evoked potential [19]. In mouse cultured neurons, disruption of complexin by mutation or RNAi knockdown causes an approximately 70% decrease in evoked responses compared to the wild type [12,17,18]. Thus, loss of evoked synaptic vesicle exocytosis is a common feature of complexin mutants in all organisms studied.

How can complexin facilitate evoked release when it clearly inhibits spontaneous fusion of synaptic vesicles? There are two possible models for the role of complexin in evoked release: In the first model, complexin has only one function – to inhibit spontaneous vesicle fusion; the defect in evoked release is an indirect effect of depletion of the vesicle pool through spontaneous fusion. In the second model, complexin also has a domain that facilitates evoked release. In short, in this second model complexin has dual functions: to inhibit spontaneous release of the tonic pool and to facilitate fusion from the evoked pool.

Stimulating by inhibiting

In the inhibitory model, rapid consumption of vesicles via spontaneous fusion depletes vesicles from the evoked pool (Figure 6, middle). Thus, a role for complexin in stimulating exocytosis is an illusion of an empty readily releasable pool. An attractive aspect of this model is its simplicity: complexin would have a single, uncomplicated function in synaptic vesicle exocytosis. If this were true, mutations that partially disrupt the inhibitory function should disrupt evoked responses proportionally – the phenotypes should always be linked – losing vesicles to constitutive fusion should deplete the readily releasable pool. By contrast, mutations in the N-terminal domain increase spontaneous release but do not reduce evoked release (Figure 5). In fact, evoked release is even higher than in the wild type. Similarly, removal of the C-terminal prenylation motif of fly complexin increases spontaneous release and enhances evoked release [26]. Together, these data suggest that the synapses in worms and flies can maintain high rates of tonic release without depleting the evoked vesicle pool.

A stimulatory domain

In the dual function model, complexin inhibits spontaneous release from the tonic pool, but facilitates evoked release. We found in C. elegans that complexin is essential for evoked release, potentially supporting a dual role for complexin. In the mouse, the N-terminus of complexin is required for evoked responses possibly to promote a late step in the fusion process [20,12,23]. Surprisingly, when we deleted these residues in C. elegans complexin, evoked release was not disrupted. These data suggest that residues involved in promoting fusion are located in a different region of the molecule than the N-terminus in C. elegans.

Stimulating by docking

How then might complexin facilitate evoked release at a molecular level? An alternative model is that complexin facilitates evoked release via the central helix (Figure 6). The binding of the central helix domain of complexin to the partially zippered SNARE complex could prevent unzippering and the reversal of the primed state [7,8]. In an in vitro vesicle docking assay, binding by complexin stabilizes the trans configuration [15]. In the absence of complexin, trans-SNARE complexes are reversible and primed vesicles would become unprimed (Fig. 6, below). Such reversal of priming would contribute to the reduction of docked vesicles observed at the C. elegans neuromuscular junction and account for the unusually severe defect in evoked release. In this model, the function of complexin is to maintain the primed state, and the interactions of the alpha helical regions of complexin with the SNARE complex acts essentially as a fourth SNARE protein. Again, the elegance of this model is its simplicity – complexin stabilizes the docked state. However, we cannot exclude the possibility that complexin plays a more direct role in evoked release, for example via regulatory interactions with synaptotagmin.

Species diversity

How can we reconcile contradictory data for complexin mutants in different experimental systems? In the mouse, complexin primarily stimulates evoked release, whereas in flies it primarily blocks spontaneous release. In the worm, complexin plays a prominent role in both functions: complexin is essential for evoked release and it blocks spontaneous fusion. However, these differences among species are ones of degree: inhibitory and stimulatory functions for complexin are found in each organism. Given the strong sequence conservation, it is likely that complexin is playing similar roles at all synapses, its character shaped by the other players at the synapse. It seems that complexin encompasses schizophrenic roles in exocytosis – it both blocks and stimulates fusion.

Constructs and strains

All constructs were made using standard molecular biology techniques and Gateway technology (Invitrogen). For a full description of how all constructs were made see the online supplemental materials and methods. Rescuing or reporter constructs were either microinjected into the worms gonads to produce progeny carrying extrachromosomal arrays or single copy insertions of rescuing or mutant complexins were produced using MosSCI as previously described [25]. For a full description of the methodologies used to create transgenic C. elegans see the online experimental procedures.

Behavioral assays

Thrashing assays were performed as previously described [27] and are fully described in supplemental online experimental procedures. For the aldicarb assays, 20–25 worms were added to the center of room-temperature aldicarb plates and scored for paralysis after four hours. The assays are fully described in the online experimental procedures. For heat-shock experiments, strains were heat shocked for one hour at 34°C and then allowed to recover at room temperature for six hours prior to assay. For heat shock aldicarb assays, 20–25 worms were added to plate containing 1 mM aldicarb and scored for paralysis every fifteen minutes.

Confocal Microscopy

Worms were immobilized in 2% phenoxypropanol in M9 solution on freshly made 2% agarose pads. Images were acquired on a Pascal LSM5 confocal microscope (Zeiss, Inc.) with a 63× 1.4NA oil objective. For imaging of the dorsal nerve cord worms were imaged with the dorsal surface facing towards the objective. For each of the synaptic markers a Z-stack was taken of a 70 µm section of nerve cord and the total number of GFP positive puncta were counted.

Electrophysiology

Young adult hermaphrodite animals were used for electrophysiological analysis. Miniature and evoked postsynaptic currents (mPSCs and ePSCs) at the neuromuscular junction were recorded as previously described [21,28]. For a full description of electrophysiological methods and analysis see the supplemental experimental procedures online. Data were imported into Origin, version 7.5 (OriginLab), for graphing and statistical analysis. Unpaired t-test was used for statistical comparisons. A P value of < 0.05 is considered statistically significant.

Electron microscopy

N2 and cpx-1(ok1552) worms were fixed in parallel for electron microscopy using high-pressure freezing followed by substitution of solvent-based fixatives as previously described [29–31]. For a full description of the electron microscopy methods see supplemental experimental procedures online.

All morphometry was conducted blind to genotype and included a matched wild-type worm that was fixed in parallel. The number of synaptic vesicles (~30 nm in diameter) in each synaptic profile was counted, and their distances from the dense projection and plasma membrane were measured. Docked vesicles were defined as touching the plasma membrane. Analysis included the acetylcholine neurons VA and VB and the GABA neuron VD.

Reagents

All chemicals were purchased from Sigma-Aldrich and all molecular reagents from Invitrogen and New England Biolabs, unless otherwise stated. Worm strains not generated in the lab were ordered from the Caenorhabditis genetics center (http://www.cbs.umn.edu/CGC/).