QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Vol. 19, Issue 2, 485-497, February 2008

This Article Abstract Full Text (PDF) All Versions of this Article: E07-02-0148v1 19/2/485    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Google Scholar Articles by Lam, A. D. Articles by Stuenkel, E. L. PubMed PubMed Citation Articles by Lam, A. D. Articles by Stuenkel, E. L.

SNARE-catalyzed Fusion Events Are Regulated by Syntaxin1A–Lipid Interactions

Alice D. Lam^*,, Petra Tryoen-Toth, Bill Tsai, Nicolas Vitale, and Edward L. Stuenkel^*,

*Department of Molecular and Integrative Physiology, Neuroscience Program, and Department of Cell and Developmental Biology, University of Michigan, Ann Arbor, MI 48109; and Département Neurotransmission et Sécrétion Neuroendocrine, Unité Mixte de Recherche 7168/LC2, Centre National de la Recherche Scientifique/Université Louis Pasteur, 67084 Strasbourg, France

Submitted February 21, 2007; Revised October 29, 2007; Accepted November 5, 2007
Monitoring Editor: Patrick Brennwald

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Membrane fusion is a process that intimately involves both proteins and lipids. Although the SNARE proteins, which ultimately overcome the energy barrier for fusion, have been extensively studied, regulation of the energy barrier itself, determined by specific membrane lipids, has been largely overlooked. Our findings reveal a novel function for SNARE proteins in reducing the energy barrier for fusion, by directly binding and sequestering fusogenic lipids to sites of fusion. We demonstrate a specific interaction between Syntaxin1A and the fusogenic lipid phosphatidic acid, in addition to multiple polyphosphoinositide lipids, and define a polybasic juxtamembrane region within Syntaxin1A as its lipid-binding domain. In PC-12 cells, Syntaxin1A mutations that progressively reduced lipid binding resulted in a progressive reduction in evoked secretion. Moreover, amperometric analysis of fusion events driven by a lipid-binding–deficient Syntaxin1A mutant (5RK/A) demonstrated alterations in fusion pore dynamics, suggestive of an energetic defect in secretion. Overexpression of the phosphatidic acid–generating enzyme, phospholipase D1, completely rescued the secretory defect seen with the 5RK/A mutant. Moreover, knockdown of phospholipase D1 activity drastically reduced control secretion, while leaving 5RK/A-mediated secretion relatively unaffected. Altogether, these data suggest that Syntaxin1A–lipid interactions are a critical determinant of the energetics of SNARE-catalyzed fusion events.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Membrane fusion is a process that underlies compartmentalization within all eukaryotic cells, and allows for the many critical and diverse physiological functions in higher organisms. Despite the essential and ubiquitous nature of this process, a considerable energetic expenditure is required to overcome the electrostatic repulsion between opposing lipid bilayers and to deform and ultimately rupture these bilayers (Chernomordik and Kozlov, 2003; Cohen and Melikyan, 2004). As a result, substantial effort has been placed on defining the molecular machinery that overcomes this energetic barrier to accomplish regulated and rapid membrane fusion.

SNARE (soluble n-ethylmaleimide-sensitive fusion factor attachment protein receptor) proteins have now been identified as the minimal protein machinery required for membrane fusion (Jahn and Scheller, 2006). Their critical role is supported by multiple lines of evidence, including that SNARE proteins are sufficient to drive membrane fusion when reconstituted into liposomes in vitro (Weber et al., 1998) and that cleavage of SNARE proteins by clostridial toxins (Schiavo et al., 1992; Blasi et al., 1993a,b), as well as genetic mutations resulting in loss of SNARE protein function (Broadie et al., 1995; Littleton et al., 1998; Saifee et al., 1998), strongly inhibit neurotransmitter release. Currently, the role of SNARE proteins in membrane fusion is believed to be predominantly mechanical. During neurotransmitter release, nucleation and zippering of a highly stable SNARE core complex formed from two plasma membrane SNARE proteins, Syntaxin1A (Syn1A) and SNAP25, and a vesicle membrane SNARE protein, VAMP2, is believed to generate the energy required to bring opposing membranes to a state of proximity that initiates the fusion event (Jahn and Scheller, 2006). As such, the majority of studies in this field have focused on characterizing SNARE protein–protein interactions, although the fusion process, by definition, must also involve lipids.

That the specific lipid composition of the membrane can have profound consequences on the energetic barrier for membrane fusion has been demonstrated both theoretically as well as in several in vitro membrane fusion systems (Chernomordik et al., 1995; Melia et al., 2006; Vicogne et al., 2006; Zimmerberg and Gawrisch, 2006), although this topic has been difficult to study in the context of secretion from live cells. Local sequestration of specific bioactive lipids may provide energetic proclivity to fusion via enhancement of protein–lipid interactions essential for priming and fusion (for instance, PI(4,5)P2-synaptotagmin interactions) (Bai et al., 2004). Outside of protein–lipid interactions, the shapes of individual lipids can also have dramatic effects on membrane curvature and on the energetics of fusion (Chernomordik et al., 1993; Chernomordik and Kozlov, 2003). For example, cone-shaped lipids (e.g., phosphatidylethanolamine, diacylglycerol, or phosphatidic acid) spontaneously form negative membrane curvatures (Epand et al., 1996; Leikin et al., 1996; Kooijman et al., 2005). The presence of these lipids in the contacting leaflets of merging membranes has been shown to favor the formation of the stalk and hemifusion intermediates that are likely to underlie membrane fusion (Kozlovsky and Kozlov, 2002), and as such, these lipids are often described as being "fusogenic." Effects of spontaneous curvature are a highly local phenomenon (Chernomordik and Kozlov, 2003), however, and in order for fusogenic lipids to exert such effects on exocytosis, they must be directly localized to the sites at which membrane fusion occurs. As the minimal machinery for membrane fusion, SNARE proteins physically define the sites at which exocytosis occurs, and therefore, would be ideal molecular partners to bind and sequester fusogenic or bioactive lipids, thus reducing the energy barrier at exocytotic sites.

The purpose of this study was to determine whether Syn1A, a plasma membrane Q-SNARE, forms functional interactions with structural or bioactive lipids which might then exert a controlling influence over the fusion event. Syn1A (Bennett et al., 1992) is a 35-kDa SNARE protein that is fully anchored to the plasma membrane via a C-terminal transmembrane domain. Several recent reports suggest that Syn1A may form key interactions with lipids. First, biochemical isolation of lipid rafts demonstrated that Syn1A localizes to cholesterol-dependent membrane microdomains; disruption of these rafts resulted in an inhibition of stimulated exocytosis in PC-12 cells (Chamberlain et al., 2001; Lang et al., 2001). Second, optical studies in unroofed PC-12 cells (Lang et al., 2001) and TIRF (total internal reflection fluorescence) studies in intact MIN6 cells (Ohara-Imaizumi et al., 2004) demonstrated that Syn1A forms cholesterol-dependent clusters within the plasma membrane, which were preferential sites at which vesicle docking and fusion occurs. Moreover, Syn1A clusters partially colocalized with PI(4,5)P2 clusters within the plasma membrane, and vesicle fusion could be correlated with the extent of cluster colocalization (Aoyagi et al., 2005). Third, analysis of single fusion events by carbon fiber amperometry demonstrated that Syn1A's transmembrane domain forms part of the fusion pore (Han et al., 2004). Fourth, FRAP (fluorescence recovery after photobleaching) analysis of Syn1A in reconstituted polymer-supported lipid bilayers demonstrated an increase in the immobile fraction of Syn1A in the presence of acidic phospholipids (Wagner and Tamm, 2001). Lastly, Syn1A contains a polybasic juxtamembrane region which, although unstructured, was recently found to be inserted into the lipid bilayer (Kim et al., 2002). Although these data indicate that Syn1A forms key interactions with acidic phospholipids, the specificity of these interactions and the physiological significance has not, to date, been directly evaluated.

Here, we identify a novel and specific interaction between Syn1A and the fusogenic lipid phosphatidic acid, in addition to demonstrating interactions of Syn1A with multiple phosphoinositol lipids, including PI(4,5)P2. Progressive neutralizing mutations within the conserved, polybasic juxtamembrane region in Syn1A reduce Syn1A's affinity for acidic phospholipids in a graded manner that correlates with the graded reduction in the secretory function of these mutants. Using carbon fiber amperometry, we demonstrate that this reduction in secretory function results from a decrease in fusion event frequency; moreover, successful fusion events that occurred demonstrated significantly longer fusion pore durations and smaller fusion pore diameters compared with control. Importantly, the secretory defect seen with a lipid-binding–deficient Syn1A (5RK/A) could be completely rescued by overexpression of the phosphatidic acid–generating enzyme, phospholipase D1 (PLD1), whereas knockdown of PLD1 activity strongly reduced control secretion without affecting secretion from Syn1A 5RK/A-expressing cells. We therefore propose that Syn1A–lipid interactions are critical in determining the energetics of SNARE-mediated fusion events. Sequence alignment across multiple SNARE protein families demonstrates a high level of conservation of the polybasic juxtamembrane region (Weimbs et al., 1998), suggesting that this lipid-binding role may be a common and important one for many SNARE proteins in membrane fusion.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Antibodies
The following antibodies were used: -Syntaxin1A clone HPC-1 (Sigma, St. Louis, MO); -Syntaxin1A clone 78.3 (Synaptic Systems, Göttingen, Germany), -GST (Amersham Pharmacia Biotech, Piscataway, NJ); and -GFP (Clontech, Palo Alto, CA), donkey -mouse HRP and donkey -rabbit HRP (Jackson ImmunoResearch Laboratories, West Grove, PA).

Clones
Rat Syntaxin1A was used for all syntaxin constructs. Site-directed mutagenesis was carried out using the Quickchange Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA). The GST-Syn1A(252-265) peptide construct was made by ligating an annealed oligo (F, TCGAAAGAAGGCCGTCAAGTACCAGAGCAAGGCACGCAGGAAGAAGTAG; R, AGCTCTACTTCTTCCTGCGTGCCTTGCTCTGGTACTTGACGGCCTTCTT) with 5' and 3' overhangs into the pGex-KG vector cut at the XhoI and HindIII sites of the multiple cloning site (MCS) downstream of glutathione S-transferase (GST). Fluorophore-tagged proteins were constructed using the Cre-recombinase–based Creator system (Clontech), in which syntaxin1A, SNAP25, and Munc18-1 were subcloned into recipient fluorophore vectors (pEGFP-C1, or monomeric mutants of pECFP-C1, pEcYFP-C1 [citrine], and red fluorescent protein (RFP) vectors) containing LoxP sequences at the C-terminal end of the fluorophore sequence. For the mRFP-Syntaxin1A-pHluorin construct, the pHluorin (with an associated N-terminal 17 amino acid linker) was PCR subcloned from synaptopHluorin and ligated into the C-terminal end of syntaxin1A (with a mutated stop codon). The PLD1 mammalian expression vector (pCGN-PLD1) encoding human wild-type (WT) PLD1 was as previously described (Hammond et al., 1997). The siRNA-PLD1 construct used was a single bicistronic plasmid that expresses both human growth hormone (hGH) and an small interfering RNA (siRNA) targeted to PLD1, as previously described (Zeniou-Meyer et al., 2007). The sequence fidelity of all new constructs was confirmed by DNA sequencing (University of Michigan DNA Sequencing Core).

Cell Culture
PC-12 cells were cultured in 10% CO₂, in DMEM supplemented with 10% horse serum, 5% fetal bovine serum, penicillin (100 U/ml), streptomycin (100 µg/ml), and 1% gentamicin (10 µg/ml). HEK293-S3 cells were cultured in 5% CO₂, in RPMI 1640 with L-glutamine, supplemented with 10% fetal bovine serum, penicillin (100 U/ml), streptomycin (100 µg/ml), 0.4 mg/ml hygromycin, and 0.6 mg/ml geneticin. PC-12 cells and HEK293-S3 cells were transfected using Lipofectamine 2000 (Invitrogen) according to manufacturer's protocol. Bovine adrenal chromaffin cells were isolated and cultured as described previously (Armstrong and Stuenkel, 2005).

Purification of GST-Fusion Proteins
BL21 DE3 cells were transformed with pGex KG vectors coding for the soluble forms of Syn1A (aa 1-267 or 252-267) or specific soluble mutants of Syn1A. Cells were grown to an OD600 of 0.4–0.6, induced with 0.1 mM IPTG, and grown for 5–8 h with shaking at 23°C. Cells were harvested by centrifugation for 15 min at 5000 x g (Beckman, Fullerton, CA, JA-14 rotor), and resuspended in PBST buffer (in mM: 16 Na₂HPO₄, 4 NaH₂PO₄, 150 NaCl, 2 EDTA, 1% TX-100, pH 7.3) supplemented with protease inhibitors and 1% β-mercaptoethanol. Cells were then mechanically lysed using a French Press at 15,000 psi. GST-fusion proteins were purified from the cell lysate using glutathione-Sepharose 4B beads (Amersham), according to the manufacturer's protocol. For select experiments, the fusion protein was cleaved from the GST moiety with 1.4 NIH units of thrombin (Amersham) for 30 min at 25°C, and the cleaved GST was removed by incubation with glutathione-Sepharose beads for 1 h at room temperature. Purity of each protein was determined by fractionation by SDS-PAGE followed by Coomassie staining and was estimated to be 90%.

Protein Lipid Overlay
"PIP Strips" (nitrocellulose membranes prespotted with 100 pmol each of 15 defined lipids; phosphatidylinositol phosphate [PIP]) were purchased from Echelon Biosciences (Salt Lake City, UT) and binding overlay experiments were carried out according to the manufacturer's protocol. For phosphatidic acid (PA) binding overlay experiments, stock solutions of dipalmitoyl-PA (Avanti Polar Lipids, Alabaster, AL) solubilized in 2:1:0.8 MeOH:chloroform:H₂O were spotted onto Hybond C nitrocellulose and allowed to dry. The nitrocellulose was then blocked for 1 h in TBS-T + 0.1% ovalbumin before initial protein–lipid binding and then was incubated with a molar excess of specific Syn1A mutants or control protein in TBS-T + 0.1% ovalbumin overnight at 4°C. The nitrocellulose was then washed extensively with TBS-T + 0.1% ovalbumin, and the bound protein was detected using antibodies, followed by an ECL reaction. Digital images were taken using a Bio-Rad Fluor-S-Max Imager (Richmond, CA). Integrated densities at each PA spot (four replicates for each PA spot, for each mutant or control protein tested) were measured and normalized to the maximum integrated density for each protein treatment.

Liposome Flotation Binding Analysis
All lipids were purchased from Avanti (Alabaster, AL). Liposomes (small unilamellar vesicles) were made by mixing stock solutions of porcine brain phosphatidylcholine (PC), porcine brain phosphatidylethanolamine (PE), and dipalmitoyl-phosphatidic acid or oleoyl-lysophosphatidic acid (PA/LPA) in chloroform at a 3:2:1 (wt/wt) ratio (in mole % composition, equivalent to 49% PC, 33% PE, and 18% PA for PA-containing liposomes; for LPA-containing liposomes, mole % composition is 45% PC, 30% PE, and 25% LPA). Lipid mixtures were spun dry at 25°C in a tabletop centrifuge and resuspended in liposome buffer (in mM: 50 HEPES, 250 sucrose, 150 potassium acetate). The suspended liposome solution was placed in a bath sonicator for 30 min at 4°C, followed by incubation on a thermomixer to equilibrate overnight at 4°C. For assaying binding of protein to the liposomes, 100 ng of purified soluble Syn1A or specific Syn1A mutants (aa 1-267) was mixed with liposomes and liposome buffer in a total reaction volume of 20 µl, and the reaction mixture was placed on a thermomixer at 37°C for 1 h at 800 rpm. Reaction mixtures were loaded under a 40/30/20/10% sucrose density flotation gradient and centrifuged at 390,000 x g for 1 h at 4°C (Beckman rotor TLA-100) to separate liposomes from unbound protein. Equal volume fractions were collected from the top of each flotation gradient and loaded onto an SDS-PAGE gel, and Syn1A immunoreactivity in each fraction was detected via Western blot analysis.

Coimmunoprecipitations
HEK 293-S3 cells were plated into six-well plates and transiently transfected with WT or mutant Syntaxin1A, Munc18-1, and EGFP-SNAP25. Control transfections used Munc18-1 and EGFP-SNAP25, but excluded the Syntaxin1A. After 48 h of expression, cells were rinsed in ice-cold physiological saline solution (PSS1, in mM: 140 NaCl, 5 KCl, 10 HEPES, 10 glucose, 5 NaHCO₃, 1 MgCl₂, 2.2 CaCl₂, pH 7.3), scraped into ice-cold lysis buffer (in mM: 20 Tris, pH 7.4, 1 EDTA, and 2% sucrose, supplemented with a Mammalian Protease Arrest protease inhibitor cocktail; G-Biosciences, St. Louis, MO), dounce-homogenized by 35 strokes, and centrifuged at 800 x g to remove the nuclei. Subsequently, 1.5 vol immunoprecipitation (IP) buffer (in mM: 150 Tris, pH 7.4, 1 MgCl₂, 0.1 EGTA, 2% Triton X-100, with protease inhibitors) were added to each tube, and tubes were incubated on ice for 30 min to allow solubilization of membranes. Protein concentrations and volumes of the lysates were then equalized across conditions, and IP was carried out with a 2-h incubation with -Syntaxin1A (clone 78.3) at 4°C with rotation. Immunopure protein G beads (Pierce, Rockford, IL) were then added to each sample, and the incubation continued for 1 h. The beads were pelleted by centrifugation (1500 x g for 2 min at 4°C), washed twice in IP buffer, and washed a final time in PBS. Syntaxin1A and EGFP-SNAP25 immunoreactivity was determined in each sample using SDS-PAGE fractionation and Western blot analysis, probing with -Syntaxin1A (clone HPC-1) and -GFP, followed by an ECL reaction. Digital images were taken using a Bio-Rad Fluor-S-Max Imager, and integrated densities of the EGFP-SNAP25 and Syn1A signals were determined. For each transfection condition, the ratio of the integrated densities of enhanced green fluorescent protein (EGFP)-SNAP25 to Syn1A in the immunoprecipitated fraction was determined. All ratios were then normalized to the ratio from the condition containing WT Syn1A, to allow comparison and quantification across experiments.

hGH Secretion Assay
PC-12 cells were plated onto 24-well plates and cotransfected with specific constructs, including hGH, the light chain of the botulinum C neurotoxin (BoNT-C), Munc18-1, and full-length Syntaxin1A (WT or mutant forms). The total DNA concentration was held constant across treatments, by addition of a neomycin control plasmid. At 48–72 h following transfection, cells were rinsed for 6 min in a physiological saline solution (PSS2, in mM: 145 NaCl, 5.6 KCl, 15 NaHEPES, 0.5 MgCl₂, 2.2 CaCl₂, 5.6 glucose, 0.5 NaAscorbate, 2 mg/ml BSA, pH 7.3), followed by a 6-min stimulation with 70 mM K⁺ (same as PSS2, with equimolar substitution of K⁺ for Na⁺). PSS2 containing the secreted hGH was collected, and cells were lysed (lysis buffer, in mM: 0.2 EDTA, 10 HEPES, 1% Triton X-100, pH 7.4) to determine percent of total hGH content secreted. hGH content was measured using an hGH ELISA kit (Roche Diagnostics, Alameda, CA). Each experiment was performed with quadruplicate replicates for each treatment, and each experiment was repeated a minimum of three independent times. hGH secretion experiments using lysophosphatidylcholine (LPC), PLD1 overexpression, and siRNA-PLD1 were completed in France and thus were carried out on a different strain of PC-12 cells than in all other experiments. For experiments utilizing external application of LPC, 1 µM palmitoyl-lysophosphatidylcholine (Avanti Polar Lipids) was added during both the rinse and the stimulation period.

Imaging
Conventional Fluorescence Microscopy. Cells were transferred from media to PSS1 for imaging. Cellular fluorescence was imaged at 25°C using an Olympus 60x 1.2 NA PlanApo water objective on an Olympus IX71 microscope (Melville, NY) coupled to a TILL Photonics (Gräfelfing, Germany) polychrome monochromator for illumination. Appropriate dichroic mirrors and emission filters were used for each fluorophore imaged. Image capture was performed using a TILL-Imago QE camera under the control of TILL software.

Confocal Imaging. Confocal images were taken on an Olympus FluoView 500 Laser Scanning Confocal Microscope, using an LD405 laser, 60x 1.2 NA objective and a pinhole aperture of 260 µm.

Fluorescence Resonance Energy Transfer (FRET) Imaging. A detailed description of the FRET three-cube sensitized emission imaging methodology can be found in our previous publication (Liu et al., 2004). In addition to three different excitation/emission images (excitation/emission in nm): 436/465 (donor excitation, donor emission); 436/535 (donor excitation, acceptor emission), and 500/535 (acceptor excitation, acceptor emission), background and shade correction images were also taken. Image correction and analysis were performed offline using the method of FRET stoichiometry (Hoppe et al., 2002; Liu et al., 2004) and implemented with a custom written IGOR macro (Wavemetrics).

FURA Imaging. PC-12 cells transfected with full-length Syn1A (K253I or K253I-5RK/A), Munc18-1, BoNT-C, and RFP (to allow identification of transfected cells) were loaded for 20 min at 37°C with 3 µM Fura2 AM (Molecular Probes) in PSS1 for calcium imaging. Cells were then rinsed with PSS1, and after 15 min, time-lapse fluorescence images were acquired with excitation fluorescence alternating between 340 and 380 nm and emission was acquired at 510 nm. Fura ratios were calculated as F340/F380.

Amperometry
At 48 h before recording, bovine chromaffin cells were transfected with Syn1A (K253I or K253I-5RK/A), BoNT-C light chain, and GFP, using biolistic bombardment (Gene Gun, Bio-Rad). Conditions for biolistic transfection, and preparation of gold beads were as suggested by the manufacturer, with 2 µg DNA/mg gold beads. Cells were replated onto poly-L-lysine–coated glass coverslips 24 h before recording. Carbon fiber electrodes (5-µm; ALA Scientific, Westbury, NY) held at +650 mV were used for amperometric recordings. Cells were bath perfused with PSS1. Secretion was stimulated by local application of 100 mM K⁺ (same composition as PSS1, with K⁺ substituted equimolar for Na⁺) for 60 s, and currents were recorded during these 60 s, in addition to a 5-s prestimulus baseline recording. Currents were collected using an Axopatch 200 A amplifier modified for extended voltage output (Axon Instruments, Foster City, CA), filtered at 2 kHz, and sampled at 4 kHz. No digital filtering was applied. Currents were analyzed using an Igor XOP written by Eugene Mosharov (Columbia University) and available online (http://cumc.columbia.edu/dept/neurology/sulzer/download.html; Mosharov and Sulzer, 2005). For spike frequency analysis, only spikes with amplitudes >10 pA above the root mean square (RMS) noise level were used. Prespike foot (PSF) analysis was limited only to those PSF whose amplitudes were <30 pA and whose durations were >1 ms (>4x sampling frequency).

Data Analysis and Statistics
Statistical analyses were carried out using GraphPad Prism software. For most comparisons, unpaired t tests were used; statistical significance was designated at a p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Syn1A Binds with High Affinity to the Fusogenic Lipid Phosphatidic Acid
We initially set out to determine whether Syn1A specifically interacts with lipids that might decrease the energetic requirements for membrane fusion or exert other important functions in exocytosis. In these experiments, bacterially expressed soluble Syn1A (aa 1-267) was overlaid in decreasing concentrations onto PIP strips. Figure 1A demonstrates that Syn1A bound to multiple acidic phospholipids in a dose-dependent manner. Although Syn1A bound with highest apparent affinity to the fusogenic lipid, PA, interactions with several PIPs, including PI(3)P, PI(4)P, PI(5)P, PI(3,4)P2, PI(4,5)P2, and PI(3,4,5)P3, were also observed (Figure 1B). Notably, all of the lipids to which Syn1A bound were acidic, although Syn1A did not interact with every acidic phospholipid tested (e.g., phosphatidylserine [PS] or LPA). Furthermore, Syn1A exhibited a greater apparent affinity for the monophosphate lipid PA than the polyphosphate inositol lipids, which contain a greater negative valence compared with PA. These data suggest that although electrostatic interactions have an important role in mediating Syn1A's lipid interactions, other structural features may ultimately underlie the specificity.

View larger version (35K): [in this window] [in a new window]   Figure 1. Syn1A directly binds a subset of acidic phospholipids that includes the fusogenic lipid phosphatidic acid. (A) Protein–lipid overlay demonstrating concentration-dependent binding between soluble Syn1A(1-267) and multiple acidic phospholipids. Left, schematic representing the arrangement of lipids spotted on the nitrocellulose. Filled circles indicate those lipids for which an interaction was demonstrated. Right, -Syn1A immunoreactivity from four representative protein–lipid overlay experiments, in which decreasing concentrations of Syn1A were overlaid onto nitrocellulose membranes spotted with 100 pmol each of indicated phospholipids. (B) Averaged level of interaction of Syn1A for each phospholipid relative to that of PA. Integrated immunoreactive density associated with each lipid on the overlays was averaged from three independent experiments where [Syn1A] = 3 nM and was normalized to the integrated density for PA. Bars, means ± SEM. (C) Representative Western blots of Syn1A immunoreactivity in fractions collected from liposome flotation assays. Bound Syn1A floats with liposomes to the top of the gradient.

The Polybasic Juxtamembrane Region in Syn1A Comprises a Lipid Interaction Domain
Sequence analysis of Syn1A revealed a polybasic juxtamembrane region within Syn1A that is highly conserved across multiple species (Figure 2A, top). To determine if this region is responsible for Syn1A's interactions with acidic phospholipids, a series of progressive neutralizing mutations was generated within this region, in which one (R262A), two (R262A/R263A), or all five (5RK/A) basic residues were neutralized to alanines (Figure 2A, bottom). Mutant proteins were purified and tested for lipid-binding capacity using nitrocellulose blots on which PA had been spotted in increasing amounts. Binding curves were fit using the Hill equation, which allowed determination of the apparent binding affinity (EC₅₀) of each protein for PA (Figure 2B). Progressive neutralizing mutations resulted in a progressive reduction in apparent binding affinity to PA, with the 5RK/A mutant demonstrating the greatest reduction in apparent binding affinity (EC₅₀ 3.2 x 10⁵ fmol), followed by the R262A/R263A mutant (EC₅₀ 6.1 x 10⁴ fmol), and lastly, the R262A mutant (EC₅₀ 2.9 x 10³ fmol), which demonstrated binding similar to the WT protein (EC₅₀ 2.0 x 10³ fmol). Thus, complete neutralization of the polybasic juxtamembrane region in Syn1A resulted in a greater than 2 log shift in EC₅₀.

View larger version (44K): [in this window] [in a new window]   Figure 2. A polybasic juxtamembrane sequence in Syn1A comprises the lipid-binding domain. (A) Top, sequence alignment of juxtamembrane regions of Syn1A across species. Note the five highly conserved basic residues between positions 260-265 (NCBI accession numbers: HS: AAA53519; MMul: NP_001028037; RN: AAF23017; MMus: NP_058081; LP: AAK70494; LS: AAO83845; DM: A55673; CR: AAY54265). Bottom, sequence alignments of Syn1A juxtamembrane mutants used in this study. (B) Progressive neutralization of Syn1A juxtamembrane basic residues correlates to a progressive reduction in apparent binding affinity for PA. Binding between Syn1A mutants and PA was determined using protein–lipid overlays (n = 4), in which increasing amounts of PA were spotted on nitrocellulose. Soluble Syn1A(1-267) was at a fivefold molar excess to PA. Integrated Syn1A immunoreactivity was measured at each PA spot, normalized for each mutant, and the data were fitted with a Hill equation. (C) The 5RK/A Syn1A mutant exhibits significantly reduced binding to PA-containing liposomes compared with WT Syn1A. Percentage of Syn1A bound to liposomes was defined as the percentage of Syn1A immunoreactivity in the top three gradient fractions relative to the total Syn1A immunoreactivity. (D) Representative protein–lipid overlay demonstrating that Syn1A 5RK/A has reduced binding to all acidic phospholipids tested. (E) Representative protein–lipid overlay demonstrating that a peptide of Syn1A's juxtamembrane region (aa 252-265) closely recapitulates the lipid binding of the full-length soluble Syn1A.

Additional Structural Determinants Underlie Syn1A's Lipid-binding Properties
To determine whether structure within the juxtamembrane basic region was important for lipid binding, we constructed a Syn1A RKRK mutant, in which the order of the juxtamembrane basic residues was rearranged, while the overall charge was maintained. The apparent PA binding affinity of the RKRK mutant (EC₅₀ 2.6 x 10³ fmol) was indistinguishable from WT, indicating that Syn1A–lipid interactions can tolerate small structural changes within the juxtamembrane region (Figure 2B).

To establish whether domains outside the polybasic juxtamembrane region affect Syn1A's lipid-binding specificity, we synthesized a peptide corresponding to this region (aa 252-265), as well as the corresponding 5RK/A peptide. Protein–lipid overlays demonstrated that the WT peptide closely recapitulated Syn1A's lipid-binding profile, whereas the 5RK/A mutation abrogated the peptide's ability to bind acidic phospholipids (Figure 2E). One notable difference between the WT peptide and Syn1A (1-267) was that the WT peptide demonstrated a high apparent affinity for PS, a lipid for which Syn1A (1-267) exhibited almost no binding (compare Figure 2, D and E). Importantly, lipid binding to Syn1A (1-259), in which the polybasic juxtamembrane region had been truncated, was largely eliminated (data not shown). Thus, although other regions in Syn1A may modify its lipid-binding profile, the basic juxtamembrane residues are clearly required for these interactions.

Full-Length Syn1A Juxtamembrane Mutants Traffic Normally to the Plasma Membrane in Live Cells
To assess the relevance of Syn1A–lipid interactions in an in vivo situation, full-length, juxtamembrane mutant Syn1A proteins were next studied in living cells. Initial experiments examined whether these mutant Syn1A proteins were expressed and targeted properly. Enhanced cyan fluorescent protein (ECFP)-tagged, full-length Syn1A mutants were generated and transiently transfected into PC-12 cells. All Syn1A constructs were cotransfected with Munc18-1, which facilitated high levels of targeting of Syn1A to the plasma membrane regions. Figure 3A demonstrates that the fluorescence signal associated with both the WT and 5RK/A ECFP-Syn1A proteins trafficked normally to the plasma membrane region, as determined by confocal microscopy.

View larger version (57K): [in this window] [in a new window]   Figure 3. Syn1A 5RK/A targets to plasma membrane regions in PC-12 cells. (A) Representative confocal fluorescence and corresponding DIC images of PC-12 cells transiently transfected with eCFP-tagged Syn1A(WT) and Syn1A (5RK/A) together with Munc18-1. (B) Representative epifluorescence images of PC-12 cells transiently transfected with mRFP-Syn1A-pHluorin (WT or 5RK/A) and Munc18. The pHluorin signal (left) reports only on the pool of Syn1A present at the cell surface, whereas the mRFP signal (right) reports on the entire pool of exogenous Syn1A within the cell. (C) Average pHluorin and mRFP fluorescence intensity for PC-12 cells transfected as in B. Bars, means ± SEM. Numbers above each bar represent the number of cells used to calculate the average for each condition. (D) Scatterplot of mean pHluorin intensity versus mean mRFP intensity for each individual cell. Cells demonstrating moderate Syn1A expression levels (defined as a mean RFP signal between 20 and 170, a range within which >85% of the cells analyzed fell) were used to generate linear fits for each data set, as indicated by the solid lines. Dotted lines represent the 95% confidence intervals of the linear fits, which overlap for the WT and 5RK/A data sets.

Full-Length Syn1A Juxtamembrane Mutants Demonstrate Intact Protein–Protein Interactions within Live Cells
Having determined that the full-length juxtamembrane neutralization mutants of Syn1A were capable of trafficking correctly, we next asked whether these mutants were capable of forming appropriate protein–protein interactions in vivo. The above data suggested that both WT and 5RK/A Syn1A constructs interacted similarly with Munc18-1, as coexpression of these constructs with Munc18-1 greatly facilitated trafficking of both constructs to the plasma membrane. To confirm the normal Munc18-1 interaction properties between these Syn1A constructs, we used a sensitized emission FRET approach to compare binding of the CFP-tagged WT Syn1A or 5RK/A mutant to citrine-Munc18-1 in live cells. That this FRET stoichiometry approach is an accurate reporter of the Syn1A-Munc18-1 interaction was established in our earlier publication (Liu et al., 2004). Figure 4A demonstrates that both WT and 5RK/A proteins associated with Munc18-1 similarly, across a wide range of molar ratios. This result is quantified in Figure 4B, where at equimolar ratios (molar ratio between 0.9 and 1.1), the 5RK/A mutant exhibited a FRET efficiency (ED) with Munc18-1 that was similar to that of the WT protein (WT = 24.2 ± 0.01, 5RK/A = 24.8 ± 0.01). In contrast, a Syn1A mutant (I233A) that was previously shown to have reduced binding to Munc18-1 (Kee et al., 1995), demonstrated a reduced FRET efficiency (I233A = 5.6 ± 0.01) compared with WT. Thus, neutralizing mutations within the juxtamembrane region of Syn1A do not appear to affect Syn1A's interaction with Munc18-1.

View larger version (41K): [in this window] [in a new window]   Figure 4. Syn1A 5RK/A interacts normally with Munc18-1 and SNAP25. (A) Analysis of direct CFP-Syn1A-citrine-Munc18-1 interactions by sensitized emission FRET imaging. Graph shows a pixel-by-pixel plot of FRET efficiency versus molar ratio (Syn1A:Munc18-1). Both Syn1A (WT) and Syn1A (5RK/A) bind Munc18-1 with similar FRET efficiency. By comparison, Syn1A I233A demonstrates strongly reduced FRET efficiencies with Munc18-1 compared with WT. Solid horizontal line demonstrates background FRET values determined from coexpression of CFP and citrine fluoroproteins. (B) Average FRET efficiency for each Syn1A construct, at equimolar ratios of Syn1A and Munc18-1 (0.9 < ratio < 1.1; see vertical lines in B). Bars, mean ± SEM. (C) Representative immunoblot demonstrating co-IP of EGFP-SNAP25 with Syn1A from HEK293-S3 cells transiently transfected with Syn1A (WT, L205A/E206A, or 5RK/A), Munc18-1, and EGFP-SNAP25. Note that Syn1A 5RK/A pulls down similar amounts of EGFP-SNAP25 compared with WT Syn1A. In contrast, a mutant of Syn1A (L205A/E206A) previously shown to have reduced binding to SNARE proteins demonstrates a reduced amount of EGFP-SNAP25 pulldown. (D) Integrated densities of Syn1A and EGFP-SNAP25 bands were measured to determine a ratio of EGFP-SNAP25 to Syn1A for each condition. Ratios within a single experiment were then normalized to that of WT Syn1A to allow for comparison across experiments. Bars, means ± SEM, from three independent experiments.

Taken together, these data demonstrate that, despite profound differences in lipid-binding capabilities, the 5RK/A mutant Syn1A behaves nearly identically to the WT Syn1A in live cells, with respect to expression levels, membrane trafficking, and forming appropriate protein–protein interactions.

Use of BoNT-C Knockdown of Syn1A to Specifically Isolate the Functional Phenotypes of Exogenous Syn1A Constructs
To test whether disruption of Syn1A–lipid interactions would have a functional effect on regulated exocytosis, we transfected Syn1A juxtamembrane neutralization mutant constructs into live secretory cells (PC-12 cells or bovine adrenal chromaffin cells). To reduce potentially confounding effects from endogenous Syn1A in these cells, we transfected the cells with the light chain of BoNT-C, which cleaves Syn1A and precludes it from mediating membrane fusion (Schiavo et al., 1995). Syntaxin 4 was previously reported to be resistant to cleavage by BoNT-C and to contain an Ile in place of Lys at residue 253 of the BoNT-C cleavage site (Schiavo et al., 1995). We therefore tested whether a similar mutation (K253I) in Syn1A, would generate a BoNT-C–resistant Syn1A. For this analysis, PC-12 cells were cotransfected with an N-terminally tagged CFP-Syn1A and Munc18-1, with or without BoNT-C. Cells were then imaged using conventional fluorescence microscopy. In the absence of BoNT-C, both the WT Syn1A and Syn1A_K253I constructs targeted to plasma membrane regions (Figure 5A). In the presence of BoNT-C, however, the WT Syn1A signal was redistributed as a diffuse cytosolic signal within the cells, indicating cleavage by BoNT-C, whereas the Syn1A_K253I signal distributed to the plasma membrane, indicating resistance to cleavage by BoNT-C (Figure 5A). These experiments were quantified by scoring at least 100 random cells from each condition (while blinded to the conditions) as demonstrating either a cytosolic or membrane fluorescence distribution. Importantly, in the presence of BoNT-C, the percent of cells demonstrating a cytosolic fluorescence distribution was 63% for WT Syn1A, compared with only 5% for Syn1A_K253I.

View larger version (21K): [in this window] [in a new window]   Figure 5. BoNT-C knockdown allows isolation of functional effects of exogenous Syn1A constructs in live secretory cells. (A) Top, representative epifluorescence images demonstrating that Syn1A (K253I) is resistant to cleavage by BoNT-C. Images compare the subcellular distribution of fluorescent signal in PC-12 cells transfected with CFP-Syn1A (WT) and CFP-Syn1A (K253I) with/without the light chain of BoNT-C. Munc18-1 was coexpressed in each condition. Bottom, normalized fluorescence intensity profiles, as determined by averaging line scans taken from the outside to the middle of each cell, corresponding to the expression conditions shown above. Line scans were normalized, the peaks were aligned, and scans were averaged for each condition. (B) Functional rescue of BoNT-C knockdown in elevated K+ evoked (70 mM, 6 min) hGH secretion, by cotransfection with Syn1A (K253I) and Munc18-1. Note that S1A (K253I) or Munc18-1 alone did not rescue secretion, nor did Syn1A (WT) cotransfected with Munc18-1. Bars, means ± SEM.

Neutralizing Mutations within Syn1A's Polybasic Juxtamembrane Region Result in a Progressive Inhibition of Syn1A's Secretory Function
To compare the ability of full-length Syn1A juxtamembrane neutralization mutants to rescue secretion, we used the BoNT-C knockdown assay in cotransfected PC-12 cells. Figure 6A demonstrates that neutralizing mutations within Syn1A's polybasic juxtamembrane domain resulted in a progressive decrease in secretory function. When normalized to the extent of rescue seen with Syn1A_K253I, the Syn1A_K253I(5RK/A) mutant was only capable of rescuing secretion to 67 ± 3% of the Syn1A_K253I control level (n = 20). The Syn1A_K253I(R262A/R263A) mutant rescued secretion to 77 ± 4% of the Syn1A_K253I control level (n = 16), whereas the Syn1A_K253I(R262A) mutant exhibited a phenotype similar to Syn1A_K253I, rescuing secretion to 97 ± 4% (n = 16) of the Syn1A_K253I control level. Considering that the residual baseline secretion after BoNT-C knockdown accounts for roughly 45% of the rescued control secretion (Figures 5B and 6A, left, dotted line), the actual deficit in secretion resulting from neutralization of Syn1A's juxtamembrane region was quite substantial, with the Syn1A_K253I(5RK/A) mutant rescuing secretion to only 43% of the Syn1A_K253I control (Figure 6A, right). Importantly, this decline in secretory function of the juxtamembrane Syn1A mutants correlates well with the decrease in apparent affinity of these mutants for binding phosphatidic acid (Figure 2B) and presumably, with their affinity for other acidic phospholipids as well.

View larger version (24K): [in this window] [in a new window]   Figure 6. Neutralization of Syn1A's polybasic juxtamembrane region results in a decrease in evoked secretion. (A) Progressive neutralizing mutations of Syn1A's polybasic juxtamembrane domain result in a graded reduction in evoked hGH secretion. Transfected PC-12 cells were stimulated with 70 mM K+ for 6 min. Data were normalized to control rescue (Syn1AK253I) for each experiment and pooled across at least three independent experiments for each condition. Dotted line represents the baseline level of remaining secretion following BoNT-C treatment. Bars, means ± SEM. Asterisks indicate statistical significance (p < 0.001) from control. Graph at right shows the same data as on the left, but with the average residual BoNT-C baseline secretion for each experiment subtracted from each condition before normalization to control rescue. Note that when the data are corrected for residual BoNT-C secretion, the actual effect of the Syn1A neutralization mutants becomes substantially larger. (B) Evoked [Ca2+] dynamics reported by changes in Fura2 fluorescence intensity ratio (F340 nm/380 nm) demonstrate that depolarization-induced (100 mM K+ for 5 s) calcium fluxes are similar between control and 5RK/A cells. Left, averaged ratio traces for control and 5RK/A cells. Horizontal bar indicates period of stimulus application. Averaged baseline (middle) and evoked change (right) in F340 nm/380 nm ratio for control and 5RK/A cells. (C) Representative amperometric recordings from bovine adrenal chromaffin cells transfected with Syn1AK253I or Syn1AK253I (5RK/A) together with BoNT-C. Exocytotic secretion was evoked by a 60-s application of locally perfused 100 mM K+. Amperometric spikes were detected and quantified by an automated analysis software program written by Eugene Mosharov (Columbia University). (D) Comparison of averaged number of spikes per cell in control and 5RK/A cells. Bars, mean ± SEM. Only spikes >10 pA in amplitude above RMS baseline were considered for analysis. (E) Cumulative spike frequency distribution, normalized for the total number of spikes for control (Syn1AK253I) and Syn1AK253I (5RK/A) cells. The distributions were not significantly different.

To probe the temporal resolution and analyze the dynamics of individual fusion events, we next used carbon fiber amperometry. For these single-cell experiments, bovine adrenal chromaffin cells were used rather than PC-12 cells, as we found the exocytotic responses produced by these cells to be far more robust than those of PC-12 cells. Chromaffin cells were biolistically transfected with BoNT-C, either Syn1A_K253I (control) or Syn1A_K253I(5RK/A), and GFP. Coexpression of Munc18-1 was not necessary in these experiments, as we have previously shown that Syn1A can traffic to the plasma membrane in chromaffin cells without the need for Munc18-1 overexpression (Gladycheva et al., 2007). Transfected cells were stimulated to secrete by local perfusion of 100 mM K⁺ for 60 s, during which time amperometric spikes were recorded. Representative amperometric traces are displayed in Figure 6C. In agreement with the hGH data above, the average number of spikes (fusion events) per cell was substantially reduced in the 5RK/A cells compared with control (control: 33.6 ± 4.0 spikes/cell, n = 86; 5RK/A: 24.3 ± 2.6, n = 82; Figure 6D). However, the frequency distribution of the spikes, when normalized to the total number of spikes for each condition, was identical between control and 5RK/A cells (Figure 6E). This suggests that, rather than affecting a specific subpopulation of vesicles, the decrease in fusion events observed in the 5RK/A condition results from a generalized decrease in fusogenicity.

Syn1A's Polybasic Juxtamembrane Region Regulates Fusion Pore Dynamics
We hypothesized that the generalized decrease in fusogenicity seen in 5RK/A cells might also be manifest in individual fusion events, particularly with regards to fusion pore dynamics. Many amperometric spikes contain a PSF, which is believed to represent the initial flux of catecholamine through the fusion pore, before dilation of the fusion pore and full fusion of the vesicle with the plasma membrane (Chow et al., 1992; Zhou et al., 1996). Analysis of PSF can thus elucidate details surrounding the late stages of vesicle fusion, especially those regarding the formation and expansion of the fusion pore. Changes in PSF amplitudes are believed to represent changes in fusion pore diameter, whereas changes in PSF duration are thought to represent changes in stability of the fusion pore and kinetics of fusion pore expansion.

Representative PSF are shown in Figure 7A. Interestingly, the Syn1A_K253I(5RK/A) cells demonstrated decreased PSF amplitudes, in addition to increased PSF durations, compared with control Syn1A_K253I (control) PSF (Figure 7, B and C). In other words, the fusion pores in the 5RK/A cells were not only smaller in diameter, but also took longer to expand to full fusion. The mean PSF amplitude for control cells was 7.66 ± 0.27 pA, which was slightly reduced to 6.18 ± 0.24 pA for 5RK/A cells (n = 584 control PSF, n = 576 5RK/A PSF; Figure 7B, left). The mean PSF duration for control cells was 7.28 ± 0.74 ms, which was lengthened to 12.33 ± 1.05 ms in 5RK/A cells (n = 584 control PSF, n = 576 5RK/A PSF; Figure 7C, left). Qualitatively, these results also held true under a more stringent analysis scheme, in which the median PSF parameters for each cell were first determined, and the medians then were averaged across cells. Median PSF amplitudes were 6.16 ± 0.43 and 4.42 ± 0.22 pA for control and 5RK/A cells, respectively (n = 66 control cells, n = 78 5RK/A cells; Figure 7B, right). Median PSF durations were 5.44 ± 0.54 ms for control, compared with 8.23 ± 0.95 ms for 5RK/A cells (n = 66 control cells, n = 78 5RK/A cells; Figure 7C, right). In all analysis schemes, the differences between control and 5RK/A PSF parameters were statistically significant (p < 0.05). Thus, neutralization of Syn1A's polybasic juxtamembrane region results in a decrease in fusion pore diameter, as well as a delay in fusion pore expansion. As formation and expansion of the fusion pore have been modeled to be among the most energetically expensive steps in the membrane fusion process (Chernomordik and Kozlov, 2003; Cohen and Melikyan, 2004), these data suggest that energetic inefficiencies in the fusion process may, in part, account for the secretory defects observed with the 5RK/A mutant.

View larger version (21K): [in this window] [in a new window]   Figure 7. Neutralization of Syn1A's polybasic juxtamembrane region results in a decrease in fusion pore diameter and a lengthening of fusion pore duration. (A) Representative PSF for control (Syn1AK253I) or Syn1AK253I (5RK/A) expression conditions (+ BoNT-C) during stimulation by 60-s application of locally perfused 100 mM K+. (B) Left, mean PSF amplitudes for control and 5RK/A cells, from all PSF >1 ms in duration and <30 pA in amplitude (n = 584 control PSF analyzed, n = 576 5RK/A PSF analyzed). Right, averaged median PSF amplitudes (n = 66 control cells analyzed, n = 78 5RK/A cells analyzed). In both cases, the PSF from 5RK/A cells demonstrate statistically significant decreases in amplitude (p < 0.05). Bars, mean ± SEM. (C) Left, mean PSF durations for control and 5RK/A cells, from all PSF >1 ms in duration and <30 pA in amplitude (n = 584 control PSF analyzed, n = 576 5RK/A PSF analyzed). Right, averaged median PSF durations (n = 66 control cells analyzed, n = 78 5RK/A cells analyzed). In both cases, the PSF from 5RK/A cells demonstrate statistically significant lengthening in duration (p < 0.05). Bars, mean ± SEM.

To determine whether the secretory defect in the 5RK/A cells might be related to the energetics of fusion, we first tested the effects of externally applied LPC (1 µM), an inverted-cone–shaped lipid that facilitates fusion pore formation and expansion by inducing positive curvature to the outer leaflet of the membrane bilayer. In these experiments, using an LPC concentration that was low enough so as not to affect stimulated secretion from the Syn1A_K253I control cells, we observed a statistically significant and substantial (>50%) increase in secretion from the Syn1A_K253I(5RK/A) cells (38.6 vs. 59.8% rescued secretion in the absence or presence of LPC, respectively; Figure 8). This result suggests that the exocytotic defect in the 5RK/A mutant occurs at a late step in fusion and is likely energetic in nature.

View larger version (12K): [in this window] [in a new window]   Figure 8. Functional phenotypes of Syn1AK253I control and 5RK/A mutant are differentially regulated by manipulation of phosphatidic acid levels in live PC-12 cells. Graph demonstrates averaged results from hGH secretion assays utilizing the BoNT-C knockdown and rescue. External application of 1 µM LPC has no effect on control secretion, but results in a partial rescue of the 5RK/A phenotype. Overexpression of PLD1 results in near-complete rescue of the 5RK/A phenotype to control secretion levels. On the other hand, knockdown of PLD1 activity via a targeted siRNA construct results in a drastic decrease in control secretion, while having little to no effect on secretion from 5RK/A cells. Results are normalized to the untreated Syn1AK253I control rescue in each experiment to allow comparison across multiple experiments, and data were pooled from at least three independent experiments. Bars, mean ± SEM. Letters above bars indicate pairs of conditions in which the differences are statistically significant (p < 0.05).

Importantly, we found that overexpression of PLD1 resulted in a near complete phenotypic rescue of the 5RK/A mutant [117 vs. 103% rescued secretion, for PLD1-treated Syn1A_K253I control cells or PLD1-treated Syn1A_K253I(5RK/A) cells, respectively; Figure 8]. Also of substantial interest was the finding that siRNA-mediated knockdown of PLD1 drastically reduced secretion from Syn1A_K253I control cells, while having little effect on secretion from Syn1A_K253I(5RK/A) cells [74% reduction in secretion for Syn1A_K253I control cells, compared with a 24% reduction for Syn1A_K253I(5RK/A) cells, in the presence of siRNA-PLD1; Figure 8]. These data provide strong evidence that the loss of interaction between Syn1A and phosphatidic acid largely accounts for the secretory defects seen with the Syn1A 5RK/A mutation and rule out the possibility that this secretory defect resulted from disruption of untested Syn1A–protein interactions or from alterations in Syn1A structure. More importantly, these results demonstrate that Syn1A–lipid interactions are critically important in regulating the overall levels of evoked secretion in live neuroendocrine cells.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
SNARE complexes comprise the minimal protein machinery required for membrane fusion; hence most studies have focused on defining the regulation of their assembly and disassembly. However, the fusion process also involves lipids, and the specific lipid composition of membranes undergoing fusion has profound consequences on the energetic barrier for fusion (Chernomordik et al., 1995). A central and unresolved issue is whether specific mechanisms exist to recruit and sequester fusogenic and/or bioactive lipids at preferential sites of membrane fusion.

In this study, we establish that the Q-SNARE Syntaxin1A (Syn1A) forms functional interactions with acidic phospholipids and that these interactions facilitate membrane fusion. Syn1A specifically interacted with the fusogenic lipid, phosphatidic acid, as well as with multiple polyphosphoinositides, including PI(4,5)P2. Neutralizing mutations within a highly conserved, polybasic juxtamembrane region of Syn1A (aa 252-265) progressively reduced Syn1A's ability to bind lipids, while leaving intact these mutants' abilities to traffic correctly to the plasma membrane and to form appropriate protein–protein interactions. Development of a novel BoNT-C knockdown and rescue assay allowed the secretory function of exogenous mutant Syn1A constructs to be studied in isolation from endogenous WT Syn1A and demonstrated that progressive neutralization of Syn1A's juxtamembrane region resulted in a progressive decrease in secretory function. Moreover, amperometric analysis uncovered a lengthening in fusion pore duration and a decrease in fusion pore diameter in fusion events catalyzed by a lipid-binding–deficient Syn1A (5RK/A), suggesting an energetic defect in fusion. Importantly, we found that the inhibition of secretion observed with the Syn1A 5RK/A mutant could be completely rescued by overexpression of PLD1, and that knockdown of PLD1 activity strongly inhibited control secretion, while having little effect on 5RK/A secretion. Altogether, these data suggest that Syn1A–lipid interactions play a key role in regulating the energetics of membrane fusion.

We propose that Syn1A–lipid interactions function both structurally as well as electrostatically to reduce the energetic barrier for fusion specifically at sites of exocytosis. This energy barrier can be modeled as a series of membrane intermediates formed during the merging of two lipid bilayers (Chernomordik and Kozlov, 2003; Cohen and Melikyan, 2004). First, the membranes are brought within proximity to establish a region of dehydrated contact. A fusion stalk forms as the initial connection between the membranes, and radial expansion of this stalk yields a hemifusion diaphragm, where contacting leaflets of the bilayers have merged, whereas the distal leaflets remain separate. Importantly, the presence of negative curvature–favoring lipids in the contacting leaflets greatly reduces the energetic requirements for stalk and hemifusion formation (Kozlovsky and Kozlov, 2002). PA, under physiological conditions, exhibits negative spontaneous curvature approaching that of PE (Kooijman et al., 2005). Thus, a first structural function of the Syn1A–lipid interaction may be to localize PA to sites of fusion, thereby generating negative curvature to facilitate stalk formation and stabilize the hemifusion intermediate.

Beyond hemifusion, generation of lateral tension within the hemifusion diaphragm leads to membrane rupture and formation of a fusion pore, whereby the distal leaflets of the two bilayers merge. Expansion of this fusion pore results in full membrane collapse and fusion. These steps are believed to be the most energetically expensive in membrane fusion (Chernomordik and Kozlov, 2003; Cohen and Melikyan, 2004). It has been proposed that the lateral tension required for formation and expansion of the fusion pore may be generated by electrostatic interactions between fusogenic proteins and acidic lipids (Zimmerberg and Gawrisch, 2006). Indeed, generation of the lateral tension to drive fusion pore expansion may be a second electrostatic function of the Syn1A–lipid interaction. This is supported by our amperometry results, which demonstrated that neutralization of Syn1A's polybasic juxtamembrane region resulted in a delay in fusion pore expansion and a decrease in fusion pore diameter.

In this study, we demonstrate a novel and specific interaction of Syn1A with PA, a lipid known to exert substantial effects on regulated exocytosis. PLD1, an enzyme that generates PA from PC, potently enhances regulated exocytosis at a late stage in vesicle fusion, in a variety of cell types including PC12 cells, adrenal chromaffin cells, adipocytes, and neurons (Humeau et al., 2001; Vitale et al., 2001; Hughes et al., 2004; Huang et al., 2005; Zeniou-Meyer et al., 2007). Whereas specific effectors for the PLD1-generated PA have remained elusive, our results strongly suggest that the facilitatory effects of PLD on secretion may be mediated in part by Syn1A-PA interactions. Namely, we found that the Syn1A 5RK/A mutant, which lacks the ability to bind PA, demonstrated reduced levels of secretion. Overexpression of PLD1 rescued this defect, presumably because the increased local concentrations of PA generated by PLD1 compensated for the 5RK/A's inability to sequester PA. Similarly, knockdown of PLD1 activity resulted in the inhibition of control secretion, because in the absence of PA, Syn1A's function becomes similar to that of the 5RK/A mutant. Accordingly, knockdown of PLD1 activity had little effect on Syn1A 5RK/A, because this mutant cannot normally bind PA.

We have focused largely on Syn1A-PA interactions, but our finding that Syn1A directly binds PI(4,5)P2 is also of interest. Quantitative in vitro binding assays suggest that a peptide of Syn1A's basic juxtamembrane region may bind to PI(4,5)P2 with higher affinity than to PA (S. McLaughlin, personal communication). PI(4,5)P2 is perhaps the most notable lipid signal for regulated exocytosis, given its requirement in the ATP-dependent priming of vesicles (Eberhard et al., 1990; Hay et al., 1995; Wiedemann et al., 1996). Endogenous PI(4,5)P2 clusters partially colocalize with Syn1A clusters in membrane sheets of PC12 cells, and exocytosis was promoted at sites of Syn1A-PI(4,5)P2 cluster colocalization (Aoyagi et al., 2005). Of substantial interest is that PA and PI(4,5)P2 participate in a positive feed-forward cycle that results in the local generation of both lipids: PA positively regulates PIP5KI, an enzyme that generates PI(4,5)P2 (Jenkins et al., 1994), and in turn, PI(4,5)P2 positively regulates PLD1, which generates PA (Liscovitch et al., 1994). That such specific membrane microdomains can have profound implications on the energetics of SNARE-mediated fusion events is a concept conserved down through yeast. Indeed, it was shown that a yeast SNARE complex (Spo20p-Sso1p-Snc1p) that normally mediates fusion at the prospore membrane was insufficient to drive fusion at the plasma membrane (Coluccio et al., 2004). This insufficiency could be overcome by overexpression of MSS4p (a yeast PI5K). Interestingly, the effect of MSS4p was mediated via recruitment of Spo14p (a yeast PLD), to the plasma membrane. It was suggested that Spo14p-mediated generation of PA at the plasma membrane reduced the energetic requirements for fusion such that the energy released by the Spo20p-Sso1p-Snc1p complex became sufficient to drive fusion at the plasma membrane.

Although the current report is the first to establish the identity of specific Syn1A–lipid interactions, map the interacting domain, and determine direct functional effects of these interactions, it is clear that SNARE–lipid interactions are emerging as significant functional interactions that underlie membrane fusion (Quetglas et al., 2000; Hu et al., 2002; Kweon et al., 2003a,b; Vicogne et al., 2006). Of substantial importance is that sequence alignment across multiple families of SNARE proteins demonstrates high conservation of the lipid-interacting polybasic juxtamembrane region (Weimbs et al., 1998). Therefore, the concept of membrane fusion driven by SNARE proteins must be enlarged to encompass the possibility that SNAREs also function to spatially sequester bioactive lipids as a means to alter the energetic requirements for fusion.

	ACKNOWLEDGMENTS

 
We thank Dr. Stuart McLaughlin (SUNY) and Dr. Joshua Zimmerberg (NIH) for helpful discussion; Drs. Ronald Holz and Mary Bittner (Univ. of Michigan) for secretion assay advice, PC-12 cells, bovine adrenal glands, and other reagents; Matthew D'Andrea-Merrins for molecular biology advice; and Rishi Chaudhuri and Ray Wu for technical assistance. This work was supported by grants from the National Institutes of Health (R01 NS039914 and NS053978, E.L.S.; F31 NS053263, A.D.L.), as well as from the Agence Nationale de la Recherche (ANR-05-BLAN-0326-01, N.V.), and by the Association pour la Recherche sur le Cancer (Grant 4051, N.V.). A.D.L. was also supported in part by a Medical Scientist Training Grant from the National Institutes of Health (GM007863).

	Footnotes

 
This was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E07-02-0148) on November 14, 2007.

Address correspondence to: Alice D. Lam (adlam{at}umich.edu)

Abbreviations used: BoNT-C, botulinum neurotoxin type C; ECL, enhanced chemiluminescence; FRET, fluorescence resonance energy transfer; hGH, human growth hormone; LPA, lysophosphatidic acid; PA, phosphatidic acid; PC, phosphatidylcholine; PC-12, pheochromocytoma-12; PE, phosphatidylethanolamine; PS, phosphatidylserine; PI(4,5)P2, phosphatidylinositol (4,5)-bisphosphate; PLD1, phospholipase D1; PSF, prespike foot; PSS, physiological saline solution; SNAP25, synaptosomal-associated protein of 25 kDa; Syn1A, Syntaxin1A; VAMP2, vesicle-associated membrane protein-2.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Aoyagi, K., Sugaya, T., Umeda, M., Yamamoto, S., Terakawa, S., and Takahashi, M. (2005). The activation of exocytotic sites by the formation of phosphatidylinositol 4,5-bisphosphate microdomains at syntaxin clusters. J. Biol. Chem 280, 17346–17352.[Abstract/Free Full Text]

Armstrong, S. M., and Stuenkel, E. L. (2005). Progesterone regulation of catecholamine secretion from chromaffin cells. Brain Res 1043, 76–86.[CrossRef]