SNAREs (soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptors) are widely accepted to drive all intracellular membrane fusion events. SM (Sec1/Munc18-like) proteins bind to SNAREs and this interaction may underlie their ubiquitous requirement for efficient membrane fusion. SM proteins bind to SNAREs in at least three modes: (i) to a closed conformation of syntaxin; (ii) to the syntaxin N-terminus; and (iii) to the assembled SNARE complex. Munc18-1 exhibits all three binding modes and recent in vitro reconstitution assays suggest that its interaction with the syntaxin N-terminus is essential for neuronal SNARE complex binding and efficient membrane fusion. To investigate the physiological relevance of these binding modes, we studied the UNC-18/UNC-64 SM/SNARE pair, which is essential for neuronal exocytosis in Caenorhabditis elegans. Mutations in the N-terminus of UNC-64 strongly inhibited binding to UNC-18, as did mutations targeting closed conformation binding. Complementary mutations in UNC-18 designed to selectively impair binding to either closed syntaxin or its N-terminus produced a similarly strong inhibition of UNC-64 binding. Therefore high-affinity UNC18/UNC-64 interaction in vitro involves both binding modes. To determine the physiological relevance of each mode, unc-18-null mutant worms were transformed with wild-type or mutant unc-18 constructs. The UNC-18(R39C) construct, that is defective in closed syntaxin binding, fully rescued the locomotion defects of the unc-18 mutant. In contrast, the UNC-18(F113R) construct, that is defective in binding to the N-terminus of UNC-64, provided no rescue. These results suggest that binding of UNC-18 to closed syntaxin is dispensable for membrane fusion, whereas interaction with the syntaxin N-terminus is essential for neuronal exocytosis in vivo.
Intracellular membrane fusion is controlled by the same ubiquitous protein machinery from yeast to the human brain . Key components of this machinery are the SNAREs [SNAP (soluble N-ethylmaleimide-sensitive fusion protein-attachment protein) receptors] . First identified from brain as syntaxin 1, SNAP-25 and VAMP (vesicle-associated membrane protein) , this family of proteins is characterized by possession of one or more signature SNARE motifs, which mediate the interaction of the individual proteins to form a four-helical heteromeric complex . The zippering together of SNAREs localized to the vesicle (v-) and target (t-) membranes as complex formation proceeds is thought to pull the two membranes together and drive the fusion reaction . Indeed, membrane fusion can be reconstituted in vitro between proteoliposomes containing appropriate SNAREs [5,6].
However, SNAREs are not the only evolutionarily conserved proteins required for intracellular membrane fusion. A universal requirement for the SM (Sec1/Munc18-like) protein family in vesicle fusion has been recognized for many years and is supported by genetic studies in a wide variety of organisms, including yeast (SEC1, SLY1, VPS33, VPS45), nematodes (unc-18), flies (ROP) and mice (munc18-1) [7,8]. Their precise function in the membrane fusion process remains unclear, although the observation that SM proteins interact with cognate SNAREs involved in the corresponding membrane traffic steps – particularly syntaxin homologues – suggests that the conserved function of SM proteins in membrane fusion involves interactions with SNAREs.
This appealing idea has proved difficult to reconcile with the divergent binding modes of different SM/SNARE protein interactions, however . For example, neuronal Munc18-1 was originally shown to bind with high affinity to a closed conformation of syntaxin 1a in isolation to form a complex that precludes syntaxin entering the SNARE complex (Mode 1) . In contrast, mammalian Munc18c, yeast Sly1 and Vps45 bind to the extreme N-terminus of their cognate syntaxins, either in isolation or as part of a SNARE complex (Mode 2) [11–15]. A third binding mode is evident in yeast Sec1, which is claimed not to bind efficiently to its cognate syntaxin (Sso1/2) in isolation (but see ), and to only interact with the ternary SNARE complex (Mode 3) [17,18]. Recently, however, evidence has emerged suggesting that individual SM proteins can employ multiple SNARE-binding modes [19,20]. For example, the binary interaction of Munc18-1 with syntaxin 1 and of Munc18c with syntaxin 4 appears to involve both Mode 1 and Mode 2 interactions [21–23]. Similarly, the Mode 3 interaction of Munc18-1 with the SNARE complex requires Mode 2 N-terminal binding to occur [24,25].
The available structural information on SM-SNARE protein binding Modes 1 and 2 has enabled the design of mutations that disable these interactions. The Mode 1 interaction of Munc18-1 with the closed conformation of syntaxin can be inhibited by introduction of mutations that render syntaxin constitutively open . However, yeast and Caenorhabditis elegans expressing such open mutants as the sole copy of their appropriate syntaxin are apparently normal under standard conditions (although a synthetic phenotype can be seen with yeast sec9 and C. elegans unc-13 or unc-10 mutants) [27–29]. Similarly, mutations that disable the Mode 2 interaction between Sly1 and Sed5, and between Vps45 and Tlg2, show no defects in ER (endoplasmic reticulum)–Golgi or vacuolar trafficking in yeast [12,14]. The physiological significance of these SNARE-binding modes for SM function in vivo therefore remains uncertain.
Recent in vitro reconstitution experiments have revealed a role for Munc18-1 in accelerating SNARE-driven liposome fusion . This stimulatory action is eliminated by mutations that impair Mode 2 syntaxin N-peptide binding, which is in turn required for Mode 3 interaction with the SNARE complex. In contrast, mutations that interfere with Mode 1 closed syntaxin binding do not inhibit the stimulation of liposome fusion by Munc18-1 . These in vitro results therefore suggest that the interaction of Munc18-1 with the N-terminus of syntaxin 1 is critical for neuronal exocytosis. We set out to test this in vivo using C. elegans as a model organism. This is a suitable model, as there is high sequence conservation between the worm UNC-18/UNC-64 and the mammalian Munc18-1/syntaxin 1 proteins (59% and 64% identical amino acids respectively) and as all SM proteins examined to date have very similar crystal structures. Furthermore, Munc18-1 can substitute for UNC-18 in vivo , demonstrating functional homology between the worm and mammalian proteins. In the present study we report that mutations in both UNC-18 and UNC-64 that interfere with either Mode 1 or Mode 2 binding strongly reduce the binary interaction of these proteins in vitro. Despite this, worms expressing an UNC-18 mutant defective in closed syntaxin binding are phenotypically normal, indicating that Mode 1 binding is dispensable for the essential function of UNC-18 in neurotransmission. In contrast, worms expressing an UNC-18 mutant defective in N-terminal syntaxin binding are severely unco-ordinated, indicating that Mode 2 binding to syntaxin is essential for UNC-18 function in vivo.
MATERIALS AND METHODS
Unless otherwise stated, all materials were obtained from Sigma.
Plasmid construction and mutagenesis
The unc-18 rescuing construct, Punc-18::unc-18 (a gift from Professor H. Kitayama, Department of Molecular Oncology, Kyoto University, Japan) , was mutated by site-directed mutagenesis to introduce NheI/BamHI and XhoI/NotI sites at the 5′ and 3′ ends of the unc-18 cDNA respectively, to produce pAM-INJ-unc-18. The BamHI–XhoI unc-18 cDNA fragment was then cloned into the pENTR-1A Entry vector (Invitrogen) by standard techniques, producing pENTR1a-unc-18. The unc-18 point mutants (R39C and F113R) were produced by site-directed mutagenesis of pENTR1a-unc-18. Mutant BamHI–XhoI unc-18 cDNA fragments were then transferred to pAM-INJ-unc-18. The N-terminal GST (glutathione transferase)-fusion Destination vector, pG-GEX-6P-1, was created by insertion of a Gateway® Conversion Cassette (Invitrogen) at the (blunted) BamHI site of pGEX-6P-1 (GE Healthcare). Wild-type and point mutant unc-18 cDNAs were transferred from pENTR1a-unc-18 by LR reaction to create pG-GEX-unc-18. All site-directed mutagenesis reactions were performed using the GeneTailor™ method (Invitrogen), with Phusion™ DNA polymerase (Finnzymes).
Recombinant protein expression and purification
Recombinant GST–UNC-18 proteins were expressed from pG-GEX-unc-18 and purified using glutathione–Sepharose (GE Healthcare) as previously described .
Protein interaction studies
GST-tagged proteins (1.5 μg) were bound to glutathione-coated beads (GE Healthcare) for 1 h at 4 °C. Beads were washed with binding buffer (150 mM CH3CO2K, 20 mM Hepes, 1 mM MgCl2 and 1% Tween 20, pH 7.4) before incubation with in vitro translated 35S-radiolabelled UNC-64 prepared from T7-promoter-containing PCR products, as previously described . After washing, bound proteins were eluted by boiling in SDS/PAGE sample buffer. The samples were then subjected to SDS/PAGE and radiolabelled proteins were visualized by autoradiography.
GST-tagged fusion proteins were cleaved with PreScission Protease (GE Healthcare) to remove the GST tag. A Jasco-J810 spectropolarimeter was used for CD measurements in the far UV region from 190 to 260 nm. Spectra were recorded at protein concentrations of approx. 0.25 mg/ml in a cuvette of 1 mm path-length in a temperature-controlled cell holder at 25 °C. Averaged CD signals, corrected for the buffer, were converted into mean residue weighted molar ellipticity as follows :
where C is the protein concentration in mM, θ is the measured ellipticity in millidegrees, n is the number of residues and l is the pathlength in cm.
Protein three-dimensional structure production
All three-dimensional representations were produced in PyMOL (DeLano Scientific). UNC-18 and UNC-64 models were produced from their amino acid sequences using the SWISS-MODEL web server , based on the previously published Munc18-1/syntaxin 1a tertiary structure (PDB entry 3C98). The UNC-18/UNC-64 complex structure was produced by aligning UNC-18 and UNC-64 structures with PDB 3C98 in PyMOL. The UNC-64 N-peptide was too short to be modelled accurately so was produced by mutating the syntaxin 1a N-peptide in PyMOL.
Nematode culture and strains
C. elegans were cultured on 60-mm-diameter NGM (nematode growth medium) agar plates at 20 °C with Escherichia coli OP50 as a food source, using standard methods . The strain used in the present study was unc-18(e81).
Transformation of C. elegans
Germline transformation of unc-18(e81) worms with DNA was performed by microinjection . e81 worms were rescued with pAM-INJ-unc-18. Rescuing cDNA constructs (10 μg/μl) were co-injected with a reporter construct, pTG96 (sur-5::GFP) at 10 μg/μl, to identify transformed worms . The total injected DNA concentration was made up to 130 μg/μl for all injections with pBluescript-SK(+) vector. At least three independent transformed lines were analysed for each mutant construct. Strains were confirmed by single worm PCR using specific primers and GoTaq® DNA polymerase (Promega).
Confirmation of transformation by RT–PCR (reverse transcription–PCR)
Animals were harvested from two 60-mm-diameter NGM plates with chilled M9 buffer (42.3 mM Na2HPO4, 22 mM KH2PO4, 85.6 mM NaCl and 1 mM MgSO4) and centrifuged at 5000 g for 5 min at room temperature (20 °C). The worm pellet was lysed in 400 μl of TRI Reagent™ (Sigma) and RNA was purified according to the manufacturer's instructions. First strand cDNA was produced from 1 μg of total RNA template using AMV reverse transcriptase and random primers (Promega) following the manufacturer's instructions. PCR was then performed with primers specific for the gene of interest, using GoTaq® DNA polymerase (Promega).
Protein extraction and Western blotting
Animals were harvested from 24 60-mm-diameter NGM plates in chilled M9 buffer and allowed to settle on ice for 10 min. After two washes in M9 buffer, the worm pellets were re-suspended in 800 μl of worm lysis buffer [50 mM Tris/HCl, 150 mM NaCl, pH 7.6, 1.5% (w/v) n-octyl glucoside and protease inhibitor cocktail (Sigma)] and lysed at 4000 p.s.i. (1 p.s.i. ≈6.9 kPa) in a One-Shot cell disruptor (Constant Systems). The lysed worm suspension was then incubated at 4 °C with agitation before insoluble material was removed by centrifugation at 15000 g for 30 min at 4 °C. Western blotting was performed using approx. 20 μg of total protein from each strain. Membranes were blotted with anti-GFP (green fluorescent protein) (Clontech) or anti-UNC-18 (, a gift from Professor R. Hosono, Department of Biochemistry, Kanazawa University, Kanazawa, Japan) antibodies, and proteins were visualized using HRP (horseradish peroxidase)-conjugated anti-mouse IgG antibodies or HRP-conjugated anti-actin (clone AC-15).
All assays were performed on young adult hermaphrodite animals from sparsely populated plates. Experiments were conducted in a temperature-controlled room at 20 °C. All results were expressed as means±S.E.M. Significance was tested by the Mann–Whitney U test.
Locomotion was quantified by counting thrashes in solution over a 1 min time period . Young adult hermaphrodites were removed from NGM plates and placed in an untreated 96-well plate containing 100 μl of freshly made Dent's solution (140 mM NaCl, 6 mM KCl, 3 mM CaCl2, 1 mM MgCl2, 10 mM glucose, 10 mM Hepes and BSA at 0.1 mg·ml−1, pH 7.4). A thrash was defined as one complete sinusoidal movement from maximum to minimum amplitude and back again. To control for any variation in environmental factors, the locomotion rates of individual transgenic worm lines being compared were measured alternately.
OP50 E. coli was seeded around the edge of 60-mm-diameter NGM plates before approx. 50 worms were placed in the centre in a ∼10 μl drop of Dent's solution. When the liquid had been absorbed, the plates were shaken briefly to reduce clumping. Plates were scored at intervals for the percentage of worms having reached the bacterial ring. This was repeated three times for each strain.
Pharyngeal pumping assays
Worms were examined at 80× magnification under a dissecting microscope. The number of beats of the pharynx was counted over 1 min. Ten worms per strain were used.
It has recently become clear that binding of mammalian Munc18-1 to syntaxin 1 involves both Mode 1 binding to the closed syntaxin conformation and Mode 2 binding to the syntaxin N-terminal peptide [21,23]. To determine the relative contributions of these distinct interaction mechanisms to their C. elegans orthologues, we designed mutations in UNC-64 to block each mode of binding to UNC-18 (Figure 1A). For Mode 1 closed syntaxin binding, we made the previously described L166A/E167A ‘open’ mutant UNC-64 , itself based on the original L165A/E166A ‘open’ mutant syntaxin 1a . We also created an I234A mutation in the SNARE motif, analogous to the I233A syntaxin 1a mutant that is severely defective in Munc18-1 binding, probably due to impairment of the Mode 1 interaction [10,40,41]. For Mode 2 N-terminal binding, we created a deletion mutant lacking the N-terminal 21 residues (Δ21N), and a point mutant, L9A, based on the L8A syntaxin 1a mutant that prevents Mode 2 binding and stimulation of liposome fusion by Munc18-1  (Figure 1B). Both mutants are predicted to interfere with the binding of the UNC-64 N-peptide to the hydrophobic pocket in domain 1 of UNC-18 [11,24] (Figure 1C). These mutations were introduced into untagged constructs encoding the cytoplasmic domain of UNC-64. Wild-type and mutant 35S-labelled UNC-64 proteins were then made by in vitro transcription/translation and their binding to GST–UNC-18 was determined. Binding of wild-type UNC-64 to GST–UNC-18 was readily observed, whereas negligible binding occurred with GST controls, confirming the specificity of the interaction (Figure 1D). All four UNC-64 mutations inhibited binding, although some specific binding remained with the open mutant (Figures 1D and 1E).
UNC-18 binding is inhibited by UNC-64 mutations that impair either N-terminal or closed conformation binding modes
These data suggested that high-affinity binding of UNC-18 to UNC-64 involves both Mode 1 closed conformation binding and Mode 2 N-peptide binding. If so, reciprocal mutations in UNC-18 that interfere with Mode 1 or 2 binding to UNC-64 should also inhibit the UNC-64/UNC-18 interaction (Figure 1A). To target Mode 1 closed syntaxin binding, we made an R39C mutant, analogous to the well-characterized R39C Munc18-1 mutation , and itself based on the R50C Drosophila Rop mutant . For Mode 2 N-terminal binding, we created a point mutation, F113R, at Site 1 of the putative hydrophobic interaction surface in Domain 1 (Figures 2A and 2B). This was chosen because substitution of the conserved hydrophobic residue at this position for charged amino acids inhibits Mode 2 binding of mammalian Munc18c, yeast Vps45 and Sly1 to their respective syntaxins [12,14,15]. These mutations were introduced into GST–UNC-18 and the effect on binding to 35S-labelled UNC-64 proteins was determined. Binding of wild-type UNC-64 to GST–UNC-18 was greatly reduced by both the R39C and F113R mutations to near background levels (Figures 2C and 2D). To check that the novel F113R mutation does not result in gross conformational defects, we performed CD spectroscopy on the wild-type and F113R recombinant proteins, after removal of their GST tags. This yielded essentially identical and superimposable spectra (Figure 2E), demonstrating that the F113R protein is properly folded and thus ruling out this trivial explanation of the observed impaired binding. Thus results from multiple mutations in both UNC-64 and UNC-18 strongly supports the idea that this high-affinity binary interaction involves both closed conformation and N-terminal syntaxin binding.
UNC-64 binding is inhibited by UNC-18 mutations that impair either N-terminal or closed conformation binding modes
To determine the physiological relevance of each binding mode, we generated C. elegans expressing only wild-type, R39C or F113R UNC-18. This was achieved by germline transformation of unc-18 (e81)-null mutants with wild-type or mutant unc-18 plasmids along with sur5::GFP as a visible reporter . UNC-18 expression from these constructs is controlled by the natural unc-18 promoter , ensuring physiologically relevant regulation of gene expression. To confirm that transformation had taken place, DNA was isolated from the various strains and PCR amplification performed using primers specific for the unc-18 plasmid. This revealed bands of the predicted size for wild-type-, F113R- and R39C-transformed worms, but not for the untransformed e81 mutants (Figure 3A). Primers specific for the F113R mutation only produced a product in F113R-transformed worms, whereas endogenous act-1 (actin) was amplified in all strains, confirming the specificity of the PCR. Analysis of mRNA expression by RT–PCR using primers specific for plasmid-borne unc-18 confirmed that expression occurred only in transformed lines, whereas generic unc-18 primers revealed amplification of the endogenous unc-18 mRNA in all strains as expected (Figure 3B). Finally, we assessed protein expression by Western blotting using an UNC-18 antibody . This identified a band of the predicted size (∼65 kDa) in the wild-type Bristol N2 strain that was absent in the untransformed unc-18 (e81)-null mutant, but reappeared in wild-type-, F113R- and R39C-transformed e81 mutants (Figure 3C). Anti-GFP antibody staining was observed only in transformed worms, whereas actin was present at similar levels in all strains as expected (Figure 3C). From RT–PCR and Western blotting, it appeared that F113R and R39C were expressed at lower levels than in wild-type UNC-18 rescue lines. It is important to note, however, that the two mutants were expressed at the same level.
Transformation of C. elegans unc-18-null mutants with wild-type or mutant unc-18 constructs
Having established C. elegans lines expressing only wild-type, R39C or F113R UNC-18, we next determined their phenotypes. The severe inhibition of synaptic vesicle exocytosis in unc-18-null mutants greatly impairs neuromuscular transmission, resulting in a classical unco-ordinated (Unc) locomotion phenotype . Microscopical observation clearly revealed that the severe Unc phenotype of the e81 mutant was rescued by expression of wild-type UNC-18 as expected (Supplementary movie at http://www.BiochemJ.org/bj/418/bj4180073add.htm; note fluorescent green staining from the SUR-5–GFP marker). Strikingly, R39C-expressing worms appeared indistinguishable from strains rescued with wild-type UNC-18, whereas no rescue was evident with the F113R construct (Supplementary movie). To quantify these effects, thrashing assays in solution were performed, whereby the number of sinusoidal movements over a 1 min period was determined. Untransformed e81 mutants produced a very small number of severely unco-ordinated thrashes and this was not improved by expression of F113R UNC-18, whereas expression of wild-type or R39C UNC-18 was equally effective in restoring thrashing frequency to approx. 100/min (Figure 4A), similar to the Bristol N2 strain (Supplementary Figure S1 at http://www.BiochemJ.org/bj/418/bj4180073add.htm). The results in Figure 4(A) were pooled from at least three independently derived transformed lines for each mutant, but analysis of each individual line produced essentially identical results, indicating that the effects are not due to strain variation (Supplementary Figure S1). To check that these differential quantitative effects on locomotion were not specific to thrashing in solution, we employed a dispersal assay, where the rate at which worms can detect and move to a food source on solid media is determined. Again, wild-type- and R39C-transformed worms were indistinguishable, with complete migration into the food occurring within approx. 1 h (Figure 4B); whereas untransformed and F113R-transformed worms exhibited similarly slow movement, with <20% of worms having encountered the food source even after 2 days. In contrast, there was no significant difference between any of the strains in frequency of pharyngeal pumping (Figure 4C), which is a predominantly myogenic process . This therefore confirms that the severe phenotype of F113R UNC-18 worms is not due to general sickness or to post-synaptic muscle effects, but rather is due to a specific impairment of presynaptic neurotransmission.
UNC-18 function in vivo requires Mode 2 interaction with the UNC-64 N-terminus, but not Mode 1 closed syntaxin binding
Insight into SM protein function has been dramatically increased recently by the realization that individual SM proteins can interact with their cognate SNAREs in at least three ways . Indeed, the binary interaction of mammalian Munc18-1 with syntaxin 1 involves both Mode 1 binding to the closed syntaxin conformation and Mode 2 binding to the syntaxin N-terminal peptide [21,23]. Using multiple mutations that affect UNC-18 interaction with the closed conformation (UNC-64 open, UNC-64 I234A; UNC-18 R39C) and the N-terminus (UNC-64 L9A, Δ21N; UNC-18 F113R) of UNC-64, we demonstrate in the present study that the same is true for the orthologous C. elegans proteins. As mutation of either binding mode severely reduced UNC-64/UNC-18 binding, we conclude that this high-affinity in vitro interaction is the product of the two lower-affinity interaction modes. Interestingly, Mode 1 closed conformation binding has thus far only been demonstrated in SM/SNARE pairs involved in regulated exocytosis, i.e. Munc18-1/syntaxin 1 [10,26,45], Munc18c/syntaxin 4 , and UNC-18/UNC-64 (the present study). In contrast, Mode 2 N-terminal binding is common to diverse SM/SNARE pairs, including those involved in constitutive ER–Golgi and prevacuolar membrane traffic in yeast [11,14,46]. This suggests that Mode 1 binding may have evolved as a regulatory mechanism, consistent with the modulation of this binding mode in higher organisms by post-translational modifications, such as phosphorylation and S-nitrosylation [31,47]. In contrast, Mode 2 binding probably serves a conserved role, possibly by facilitating the Mode 3 SNARE complex interaction that is thought to represent a fundamental function of SM proteins in membrane fusion [16,24].
Despite the requirement for both binding modes for tight binary interaction in vitro, we demonstrate here that Mode 1 closed syntaxin binding is not essential for neuronal SM protein function in C. elegans. The ability of the R39C mutant to fully rescue the unc-18 mutant phenotype is consistent with previous work showing that unc-64 mutants can be rescued by an open UNC-64 construct [29,48] and results from in vitro reconstitution assays, where R39C and wild-type Munc18-1 are equally effective in promoting SNARE-mediated membrane fusion . Although clearly not essential, the Mode 1 interaction plays some role in regulated exocytosis, as shown by the effects of R39C and other Mode-1-defective Munc18-1 mutants on dense-cored granule exocytosis [41,42,47] and the phenotype of the original R50C Drosophila rop mutant that R39C was based upon [43,49]. Indeed, it was recently shown that knockin mice expressing open syntaxin 1b suffer seizures and exhibit alterations in synaptic vesicle releasable pool sizes . Interestingly, these mice exhibit slightly increased synaptic vesicle docking, a phenotype shared by C. elegans expressing open UNC-64 , and Mode 1 binding has recently been implicated in secretory granule docking via a Rab3/Munc18-1 interaction [51,52]. The Mode 1 interaction may therefore serve a regulatory role in vesicle docking, but one that is too subtle to be detected by the behavioural assays used in the present and previous studies [29,48].
A major finding of our work is that only Mode 2 N-terminal syntaxin binding is essential for neuronal SM protein function in vivo. This conclusion is based on our observation that F113R UNC-18 provides no rescue of the Unc phenotype despite being expressed at similar levels to the R39C protein that confers full rescue (Figure 3). Although mutations have previously been introduced into the analogous Site 1 residues in yeast Sly1 (L137R) and Vps45 (L117R) that similarly block tight binding to their cognate syntaxins, these have no observable impact on membrane trafficking [12,14]. The findings of the present study therefore provide the first demonstration of an essential in vivo function for the Mode 2 SM–syntaxin interaction. While this manuscript was in preparation, McEwan and Kaplan independently reported locomotion defects in UNC-64 L9A- and UNC-18 L116K-expressing worms, but not in open UNC-64- or UNC-18 R39C-expressing worms , thus confirming that only the Mode 2 interaction is important for normal synaptic transmission. In addition, they suggested that both interaction Modes were required for the putative chaperone function of UNC-18 in enabling transport of UNC-64 through the secretory pathway, as this was only inhibited in double mutants . However, it is unclear if this represents a general mechanism, as syntaxin 1a trafficking in mammalian cells is unaffected by Mode 2 mutations, requiring only closed conformation binding ; and as syntaxin 1 still traffics to the plasma membrane in Munc18-1-knockout mice . Since single mutations that inhibit UNC-18 interaction with the UNC-64 N-terminus exhibit locomotion defects despite fully rescuing this chaperone function, it seems safe to conclude that the late Mode 2-binding-dependent role of UNC-18 in exocytosis is phenotypically more prominent. This essential role of Mode 2 binding is also consistent with results from in vitro liposome fusion assays, where the analogous L8A syntaxin 1a mutation abolishes the stimulatory effect of Munc18-1 , and from mammalian cell studies in which the inhibitory effects of syntaxin fragments require an intact N-terminus . It is not clear why such dramatic effects of Mode 2 mutations on membrane fusion are seen with neuronal SNARE/SM isoforms from worms and mammals, but not with yeast homologues. Perhaps yet another SNARE-binding mode exists in the yeast proteins that compensates for the loss of Mode 2 binding, as suggested for Vps45 .
In summary, our results from the present study clearly demonstrate an essential physiological role for the Mode 2 interaction of UNC-18 with UNC-64. However, the precise molecular function of this binding mode is unclear. It could be that the binary interaction of UNC-18 with the UNC-64 N-terminus serves a direct role, facilitating a novel function of this heterodimeric complex. However, it seems more likely that the Mode 2 interaction is an essential precursor that enables the distinct Mode 3 interaction with the SNARE complex, as previously suggested [24,25]. Unfortunately, distinguishing between these possibilities is not possible at present, as there is no crystal structure of any Mode 3 SM/SNARE complex upon which to design specific Mode 3-disrupting mutations. Interestingly, SM mutants that are unaffected in Modes 1 and 2 binding yet are functionally defective, potentially due to impairment of Mode 3 binding, have previously been described [14,39,57]. Detailed analysis of the molecular defects in these mutants may enable the design of novel mutational tools required to resolve the outstanding mechanistic questions regarding SM/SNARE interactions and their contribution to synaptic membrane fusion in vivo.
We thank Professor R. Hosono for the gift of the UNC-18 antibody and Professor H. Kitayama for the gift of the unc-18 construct. Thanks also to Professor L. Holden-Dye, Dr N. Hopper and Dr V. O'Connor (University of Southampton, U.K.) for advice on C. elegans techniques. The unc-18(e81) strain used in this work was provided by the Caenorhabditis Genetics Center, which is funded by the NIH (National Institutes for Health) National Center for Research Resources (NCRR).
This work was supported by a research grant from BBSRC (Biotechnology and Biological Sciences Research Council) [grant number BB/D01073X/1] (to A. M. and R. D. B.). P. F. was supported by a Wellcome Trust prize studentship. J. W. B. is a recipient of an RCUK (Research Councils U.K.) Academic Fellowship.