SNAP-23 and syntaxin-2 localize to the extracellular surface of the platelet plasma membrane

Platelets represent an unusual model for studying membrane trafficking and regulated exocytosis. They are anucleate cells that are shed from maturing megakaryocytes. They acquire α-granules and dense granules via regulated delivery of preformed organelles to developing proplatelet ends.¹  Although receptor-mediated^2,3  and pinocytotic⁴  endocytosis have been observed in platelets, constitutive coupled endocytosis-exocytosis cycles that occur in nucleated cells in general^5-8  and hematopoietic cells in particular⁹  have not been documented in platelets. Another distinctive characteristic of the platelet is that its limiting membrane is characterized by a system of tunneling invaginations of the plasma membrane termed the open canalicular system (OCS).^10,11  Evidence that the OCS is open to the extracellular environment is derived from reports using cell-impermeant tracers.¹² 

The SNARE proteins that direct membrane fusion events leading to platelet granule release have been studied.¹³  The tSNAREs SNAP-23^14-16  and syntaxin-2, -4, and -7^14-17  are found in platelets. Platelets also contain gene products of the VAMP family of vSNAREs,¹⁴  including VAMP-3 and VAMP-8,^18,19  which participate in platelet granule secretion.^14,19,20  Antibodies directed at syntaxin-2 and -4 inhibit granule secretion from permeabilized platelets.^14,16,21  Anti-SNAP-23 antibody and a SNAP-23 blocking peptide also inhibit granule secretion.^16,21  These functional data provide compelling evidence that SNARE proteins are essential in mediating the membrane fusion events involved in the secretion of platelet granules.

Unlike many membrane-associated proteins, SNARE proteins lack a signal sequence required for cotranslational insertion into membranes of the endoplasmic reticulum.^22,23  Rather, membrane insertion occurs posttranslationally. The targeting of SNARE proteins to their respective intracellular compartments differs between individual SNARE proteins. SNAP-23 and its homolog SNAP-25 lack a membrane-spanning domain. Association of SNAP-25 with membranes requires a membrane-targeting module, located between the 2 α-helices, that participates directly in SNARE protein complex formation.²⁴  Association of SNAP-23 and SNAP-25 with membranes may also be facilitated by palmitoylation of the membrane-targeting module²⁵  and/or by association with syntaxin isoforms.²⁶  Syntaxins are tail-anchored (type IV) membrane-binding proteins that contain a carboxy-terminal hydrophobic domain inserted into the lipid bilayer.^27,28  Syntaxin-2 localization is dictated in part but not exclusively by this carboxy-terminal transmembrane domain.^27-29  VAMP also contains a carboxy-terminal hydrophobic domain and inserts into membranes of the endoplasmic reticulum in an ATP-dependent manner following translation.²³  The actual mechanisms whereby these tail-anchored SNARE proteins become correctly oriented within membranes and sorted to specific subcellular compartments have not been determined in detail.

We have previously studied the subcellular distribution of 3 SNARE proteins in resting platelets.³⁰  These studies showed that VAMP-3 is found primarily on platelet granule membranes. Most SNAP-23 is located on plasma membranes with the rest distributed between membranes of the OCS and granular membranes. Syntaxin-2 is more equally distributed among the different membrane compartments. This arrangement of SNARE proteins provides a molecular basis for secretion of α-granules via the plasma membrane and OCS as well as for homotypic α-granule secretion. To further evaluate the contribution of SNARE protein distribution to platelet granule secretion, we assessed the subcellular localization of SNARE proteins in activated platelets. Unexpectedly, these studies demonstrated that SNAP-23 and syntaxin-2 are expressed on the extracellular surface of platelets. Further evaluation by flow cytometry, enzymatic degradation, and immunofluorescence microscopy confirmed these results. These studies show the localization of SNARE proteins following platelet activation and represent the first demonstration of SNARE proteins on an extracellular plasma membrane surface.

Approval was obtained from the Beth Israel Deaconess Medical Center institutional review board, Boston, MA, for these studies. Informed consent was obtained in accordance with the Declaration of Helsinki.

All buffer constituents, RGDS, and sulforhodamine were purchased from Sigma (St Louis, MO). Trypsin was purchased from Promega Biosciences (San Luis Obispo, CA). Complete Protease Inhibitor Cocktail was obtained from Roche Diagnostics (Alameda, CA). Sepharose 2B and ¹⁴C-serotonin were obtained from Amersham Pharmacia Biotech (Piscataway, NJ). Botulinum toxin C light chain was obtained from List Biologics (Campbell, CA). Digitonin and α-toxin were obtained from Calbiochem (San Diego, CA). Transfected Escherichia coli expressing His-tagged recombinant acyl-protein thioesterase 1 (APT1) was kindly provided by Dr Thomas Michel.^31,32  All solutions were prepared using water purified by reverse-phase osmosis on a Millipore Milli-Q Purification Water System (Bedford, MA).

Anti-SNAP-23 antibody was developed by immunizing rabbits with a peptide corresponding to the C terminus of SNAP-23 (DRIDIANARAKKLIDS). Antibody was purified from immune serum by affinity chromatography and characterized. The purified antibody recognized recombinant SNAP-23 and identified a single protein with an apparent molecular mass of approximately 23 kDa when assayed in immunoblot analysis of platelet lysates (data not shown). A second anti–SNAP-23 antibody generated by immunization of rabbits with full-length SNAP-23 was purchased from Abcam (Cambridge, MA). Anti–syntaxin-2 antibody is an affinity-purified goat polyclonal antibody obtained from Santa Cruz Biotechnology (Santa Cruz, CA) that is directed against a 19 amino acid portion toward the C terminus of the cytoplasmic domain of syntaxin-1A. The antibody reacts with syntaxin-1A, -1B, -2, and -3. Of these syntaxin isoforms, only syntaxin-2 is found in platelets.¹⁶  The antibody does not react with syntaxin-4 or -7. A second anti–syntaxin-2 antibody directed against a 19 amino acid portion of the N terminus of rat syntaxin-2 from Calbiochem was used for immunoblot analyses. Anti–VAMP-3 antibody was generated by immunizing with a peptide consisting of the 12 N-terminal amino acids of VAMP-3.^18,30  Phycoerythrin (PE)–conjugated mouse monoclonal anti–P selectin antibody AC1.2 was purchased from Becton Dickinson (San Jose, CA).

Blood from healthy donors who had not ingested aspirin in the 2 weeks prior to donation was collected by venipuncture into 0.4% sodium citrate. Citrate-anticoagulated blood was centrifuged at 200g for 20 minutes to prepare platelet-rich plasma. Platelet-rich plasma was then used for ultrastructural studies. For platelet secretion studies, platelets were purified from platelet-rich plasma by gel filtration using a Sepharose 2B column equilibrated in PIPES/EGTA buffer.^14,30  Final gel-filtered platelet concentrations were 1 × 10⁸ to 2 × 10⁸ platelets per milliliter.

Purified human platelets were incubated with buffer, 100 μM SFLLRN, or 0.2 μM phorbol 12-myristate 13-acetate (PMA) fixed in 4% paraformaldehyde, and prepared for sectioning as previously described.^30,33  Immunonanogold staining and processing for electron microscopy was performed at room temperature on cryostat sections mounted on glass slides as described previously except for the following modifications. Glass slides with cryostat sections were incubated in primary polyclonal rabbit antibody directed against the C-terminal end of SNAP-23 at a dilution of 1:50; in an affinity-purified goat polyclonal antibody against a C-terminal portion of syntaxin-1A, -1B, -2, and -3 at a dilution of 1:10 in 0.02 M PBS; or in a rabbit polyclonal antibody directed against the 12 N-terminal amino acids of VAMP-3 at a dilution of 1:30 to 1:50 for 60 minutes. Following incubation with primary antibodies, samples were incubated in the secondary antibody (affinity-purified Fab′ fragment from goat antirabbit IgG conjugated with 1.4 nm nanogold for SNAP-23 and VAMP-3 staining or affinity-purified Fab′ fragment from rabbit antigoat IgG conjugated with 1.4 nm nanogold for syntaxin-2 staining) (Nanoprobes, Stony Brook, NY). Samples were subsequently developed with HQ silver enhancement solution (Nanoprobes, Stony Brook, NY) for 6 to 10 minutes in the darkroom, processed as previously published,^30,33,34  and studied in a CM 10 Philips electron microscope (Eindhoven, the Netherlands).

The following 4 controls were performed to ensure the specificity of immunostaining: (1) replacement of primary antibody by an irrelevant rabbit IgG or goat IgG; (2) omission of specific primary antibody; (3) omission of the secondary antibody; and (4) omission of the HQ silver enhancement solution.

Flow cytometry was performed on platelet samples using a Becton Dickinson FACSCalibur flow cytometer. For analysis of SNARE protein surface expression by flow cytometry, 20 μL of gel-filtered platelets (0.5 × 10⁸ to 1 × 10⁸/mL) were incubated with buffer alone, SFLLRN, or PMA for 20 minutes. To compare staining of SNARE proteins in permeabilized and nonpermeabilized platelets, platelets were incubated in the presence or absence of 3 μg/mL digitonin for 30 minutes. For staining, platelets were incubated with either nonimmune antibody or the indicated anti-SNARE protein antibodies for 30 minutes. Binding of anti-SNARE protein antibodies to platelets was detected using FITC-labeled goat antirabbit secondary antibodies. Platelet samples were analyzed by flow cytometry immediately after a 30-minute exposure to secondary antibodies. Values are reported as percent control compared with signal detected in unstimulated samples exposed to nonimmune antibodies. Data were analyzed using CellQuest software on a MacIntosh PowerPC.

For enzymatic treatments, gel-filtered platelets were incubated with 20 μM indomethacin to pacify the platelets. For trypsin digestion of platelet surface proteins, platelets (0.5 mL; 2 × 10⁹/mL) were exposed to 5 μg/mL trypsin for 4 hours at 37°C. Following this incubation, platelets were incubated with an inhibitor cocktail. Inhibition of trypsin was verified using S-2238 chromogenic substrate from Chromogenix (Milan, Italy). Following verification of trypsin inhibition, platelets were evaluated by flow cytometry. For assessing release of SNAP-23 by APT1, the enzyme was first incubated with 2 μM dithiothreitol (DTT) to augment acyl-thioesterase activity.³¹  Pacified, intact platelets (1 mL; 2 × 10⁹/mL) were incubated in the presence or absence of 10 μg/mL APT1 for 2 hours at 30°C. For evaluating release of syntaxin-2 by botulinum toxin C light chain, pacified concentrated platelets (1 mL; 2 × 10⁹/mL) were incubated in the presence or absence of 5 μg/mL botulinum toxin C light chain overnight at 37°C. For both preparations, platelet suspensions were pelleted at 100 000g for 2 hours to remove microparticles. Supernatants were lyophilized, resuspended in 100 μL sample buffer, and evaluated for SNAP-23 or syntaxin-2 by immunoblotting. For all studies, adequacy of pacification by indomethacin was assessed by monitoring ¹⁴C-serotonin release³³  following incubation with trypsin or by monitoring P selectin surface expression^14,35,36  following incubation with either APT1 or botulinum toxin C light chain. Platelet membrane integrity was evaluated following each enzymatic treatment by testing for permeation to sulforhodamine into platelets by flow cytometry.³⁷ 

Platelets (0.5 × 10⁸ to 1 × 10⁸/mL) treated with indomethacin were exposed to buffer alone or 100 μM SFLLRN and subsequently incubated with buffer alone or 5 μg/mL trypsin for 4 hours at 37°C. Platelets were then labeled using anti–SNAP-23 or anti–syntaxin-2 antibodies, washed in PBS, stained with FITC-labeled goat antirabbit secondary antibodies, and washed again in PBS. Platelets were subsequently plated on polylysine-coated glass slides and visualized using an Olympus AX70 Provis microscope (Olympus, Melville, NY) with a 60×1.4 numerical aperture oil-immersion lens. Digital images were captured using a Roper CoolSNAP HQ CCD camera (Roper Scientific, Ottobrun, Germany) and analyzed using Slidebook software (Intelligent Imaging Innovations, Denver, CO) in a 1392 × 1040 format.

Supernatants were prepared by centrifugation at 100 000g for 2 hours from concentrated gel-filtered platelets (1 mL; 2 × 10⁹/mL) incubated overnight in the presence of 10 μg/mL APT1 and 5 μg/mL botulinum toxin C light chain. Supernatants were concentrated 5-fold to 200 μL and mixed with 100 μL anti-SNAP-23–labeled or anti-syntaxin-2–labeled protein A–Sepharose beads. The reaction mixture was incubated overnight at 4°C with agitation. The beads were collected, washed, and evaluated by immunoblot analysis.

Platelet lysates or lyophilized supernatants were diluted in sample buffer at 95°C for 5 minutes. Proteins were then separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Immunoblotting was performed using FITC-labeled secondary antibodies and visualized using fluorescence detection on a Typhoon 9400 Molecular Imager (Amersham).

Initial studies were directed at evaluating the localization of the SNARE proteins SNAP-23, syntaxin-2, and VAMP-3 following platelet activation. Platelets were exposed to either buffer alone, a thrombin receptor activating peptide, SFLLRN, which acts through protease-activated receptor-1 (PAR1), or PMA to elicit activation. For localization, a preembedding technique using immunonanogold was employed because this technique preserves fine ultrastructural detail without compromising antigen labeling.^30,34  Labeling of resting platelets with antibody directed at SNAP-23 demonstrated that approximately half of the label was localized to the plasma membrane and the remaining label was distributed between membranes of the OCS and granule membranes (Figure 1A). These results were consistent with a previously published report of SNAP-23 localization in platelets.³⁰  Close inspection of SNAP-23 labeling demonstrated that a portion of the plasma membrane label appeared to be on the extracellular surface of the platelet (Figure 1A, arrows). Quantification of SNAP-23 staining demonstrated that 97% of resting platelets expressed at least some SNAP-23 on their extracellular surface. Following platelet activation with SFLLRN, platelets demonstrated the pseudopod formation and shape change typically observed following stimulation of the PAR1 receptor.^33,38  Localization of SNAP-23 to the extracellular surface was much more prominent following exposure to SFLLRN (Figure 1B, arrows). In addition, activated platelets stained more heavily with SNAP-23 antibody than their resting counterparts. Platelets exposed to PMA for 10 minutes demonstrated less marked shape change (Figure 1C).³⁹  However, they demonstrated increased SNAP-23 staining and, in particular, abundant SNAP-23 localization on the extracellular surface of the platelet (Figure 1C, arrows). Quantification of staining demonstrated that 100% of activated platelets expressed SNAP-23 on their extracellular surface. Control samples processed with either substitution of irrelevant rabbit IgG for immune primary antibody or omission of immune primary antibody demonstrated no significant staining (data not shown). Similarly, no significant staining was observed when either secondary antibody or the HQ silver enhancement solution was omitted from the processing. These results suggest that SNAP-23 resides on the extracellular surface of resting and, more dramatically, activated platelets.

Previous localization studies of syntaxin-2 in resting platelets have demonstrated that this SNARE protein is relatively evenly distributed between plasma membrane, granule membranes, and membranes of the OCS.^30,40  Evaluation of resting platelets confirmed this distribution (Figure 2A). A small portion of syntaxin-2 label appeared to be on the outside of platelets (Figure 2A, arrows). Quantification demonstrated that 67.7% of platelets expressed at least some syntaxin-2 on their extracellular surface. Following platelet activation with SFLLRN, syntaxin-2 labeling on the plasma membrane increased (Figure 2B). In addition, more label appeared on the extracellular side of the plasma membrane (Figure 2B, arrows). Localization of syntaxin-2 to the extracellular surface was also visualized in platelets exposed to PMA (Figure 2C, arrows). Quantification showed that 89.8% and 95% of platelets exposed to SFLLRN and PMA, respectively, demonstrated syntaxin-2 on their extracellular surface. Control samples processed with either substitution of irrelevant goat IgG for immune primary antibody or omission of immune primary antibody, secondary antibody, or HQ silver enhancement solution demonstrated no significant staining (data not shown). These results suggest localization of syntaxin-2 to the extracellular surface of platelets.

VAMP-3 has previously been found to localize primarily to granule membranes in resting platelets.³⁰  Immunonanogold staining of platelets exposed to buffer alone demonstrated staining of platelet granules without significant staining of platelet plasma membranes (Figure 3A). In platelets activated with SFLLRN, most VAMP-3 remained associated with intracellular membranes (Figure 3B). Expression of VAMP-3 on the extracellular surface was not significant. Similarly, most VAMP-3 remained associated with granule membranes following activation of platelets with PMA without significant extracellular localization of VAMP-3 (Figure 3C). Control samples processed with either substitution of irrelevant goat IgG for immune primary antibody or omission of immune primary antibody, secondary antibody, or HQ silver enhancement solution demonstrated no significant staining (data not shown). These results demonstrate that, unlike SNAP-23 and syntaxin-2, VAMP-3 is not found on the extracellular side of the plasma membrane.

Expression of SNARE proteins on the extracellular surface of cells has not previously been observed. Electron microscopy using the immunonanogold technique is potentially ambiguous with regard to plasma membrane topology and cannot alone conclusively determine antigen localization to the extracellular surface. To determine whether the apparent extracellular localization of SNAP-23 and syntaxin-2 was an artifact of the immunonanogold technique or represented actual extracellular expression of SNARE proteins, we evaluated intact, unfixed platelets for surface expression of SNAP-23, syntaxin-2, and VAMP-3 by flow cytometry. For evaluation of SNAP-23 surface expression, platelets were incubated with either buffer alone, 100 μM SFLLRN, or 0.2 μM PMA; stained with either nonimmune antibody or antibody directed at the C-terminal end of SNAP-23; and analyzed by flow cytometry. Platelets stimulated with SFLLRN or PMA demonstrated greater than a 2-fold increase in SNAP-23 staining compared with resting platelets (Fig 4A). In contrast, resting platelets failed to stain significantly better with the anti–SNAP-23 antibody than with nonimmune antibody. Similar results were obtained with antibodies directed against full-length SNAP-23 (data not shown). Staining of activated platelets with an antibody directed at a peptide derived from the C-terminal cytoplasmic domain of syntaxin-2 was increased by approximately 2-fold compared with staining with nonimmune antibody (Figure 4B). Similar results were obtained using an antipeptide antibody directed at the N terminus of syntaxin-2 (data not shown). Resting platelets failed to demonstrate significant staining with anti–syntaxin-2 antibodies. Antibody directed at the N terminus of VAMP-3 failed to stain the extracellular surface of either resting or activated platelets (Figure 4C).

Several analyses were performed to assess the detection of tSNAREs on intact platelets by flow cytometry. Evaluation of histograms demonstrated that labeling of activated platelets with anti–SNAP-23 antibody (Figure 4D) or anti–syntaxin-2 antibody (Figure 4E) represented a uniform increase in labeling of platelets and not labeling of an isolated subpopulation. Flow cytometry studies were performed in the presence of 1 mM RGDS to assess whether increased staining of activated platelets with anti-tSNARE antibodies resulted from platelet aggregation. No loss of staining was observed in the presence of RGDS (data not shown). Flow cytometry experiments evaluating staining of intact platelets with antibody directed at the intracellular protein G_αq failed to demonstrate any signal (Figure S1, available on the Blood website; see the Supplemental Materials link at the top of the online article), confirming that platelets do not become nonspecifically permeabilized during the staining procedure. Platelets fixed directly after stimulation with agonists demonstrated equal or increased staining compared with unfixed platelets, indicating that manipulations following platelet stimulation were not responsible for SNARE protein staining (data not shown). Taken together, the flow cytometry data support the supposition that SNAP-23 and syntaxin-2, but not VAMP-3, reside on the extracellular surface of activated platelets.

We next sought to assess what percentage of SNAP-23 and syntaxin-2 localize to the extracellular surface of the platelet plasma membrane. For these studies, we compared anti-tSNARE antibody binding to intact versus digitonin-permeabilized platelets using flow cytometry. Staining of both resting and SFLLRN-stimulated platelets was evaluated. Consistent with previous results, binding of anti–SNAP-23 and anti–syntaxin-2 antibodies to resting, intact platelets increased little, if any, compared with that of nonimmune antibody. In contrast, binding of anti–SNAP-23 and syntaxin-2 antibodies to SFLLRN-stimulated platelets increased 2.4-fold plus or minus 0.4-fold and 2.4-fold plus or minus 1.0-fold, respectively, compared with binding of nonimmune antibodies (Table 1). Binding of anti-tSNARE antibodies increased substantially in both intact and stimulated platelets permeabilized with digitonin, with binding in stimulated platelets increased more than 9-fold compared with that of nonimmune controls. These data demonstrate that although a detectable percentage of these tSNAREs is present on the platelet extracellular surface, the majority of SNAP-23 and syntaxin-2 is intracellular even following platelet activation.

To further verify the extracellular surface localization of SNARE proteins, we determined whether binding of anti–SNAP-23 and anti–syntaxin-2 antibodies to intact platelets is sensitive to trypsin exposure. Platelets were stimulated with either SFLLRN or PMA and then exposed to trypsin at 37°C for 4 hours. Trypsin was then neutralized, and platelets were stained using the anti-SNARE protein antibodies. These studies demonstrated that exposure of intact platelets to trypsin prevented binding of anti–SNAP-23 and anti–syntaxin-2 antibodies to platelets (Figure 5). No binding of anti–VAMP-3 antibodies was observed. Trypsin exposure did not affect binding of nonimmune IgG (data not shown). These results confirm trypsin-sensitive localization of SNAP-23 and syntaxin-2 to the extracellular surface of activated platelets.

Activated platelets demonstrate staining of SNAP-23 and syntaxin-2 on the extracellular surface by both immunonanogold staining and by flow cytometry. Detection of SNAP-23 and syntaxin-2 on the surface of resting platelets is less clear from these studies. Flow cytometry failed to show significant staining of either SNAP-23 or syntaxin-2 in resting platelets. In contrast, the more sensitive technique of immunonanogold staining with electron microscopy demonstrates expression of SNAP-23 and, to a lesser extent, syntaxin-2 on the extracellular surface of resting platelets (Figures 1,2). To evaluate whether these SNARE proteins reside on the extracellular surface of resting platelets, we exposed resting platelets to trypsin prior to activation. Because trypsin could cleave these SNARE proteins regardless of whether their antigenic epitopes were blocked, we reasoned that trypsinization of resting platelets would decrease SNAP-23 and syntaxin-2 exposure following activation only if there were significant expression of these SNARE proteins on the surface of resting platelets. Although trypsin exposure itself activated untreated platelets as indicated by dense granule release, platelets incubated with 20 μM indomethacin remained resting following exposure to trypsin (Figure 6A). Evaluation of membrane integrity using sulforhodamine³⁷  demonstrated that trypsin exposure did not render the platelet membrane permeable (Figure 6B). Following incubation with resting platelets, trypsin was neutralized using a protease inhibitor cocktail and neutralization confirmed using the chromogenic substrate S-2238. Platelets were then exposed to PMA and evaluated for SNAP-23 and syntaxin-2 by flow cytometry and immunoblot analysis. Trypsinization of pacified, resting platelets inhibited detection of either SNAP-23 or syntaxin-2 following stimulation with PMA (Figure 6C). In contrast, stimulation of trypsinized platelets with PMA resulted in a 16.8 plus or minus 4.8-fold increase in P selectin surface expression, indicating that platelets remained responsive to PMA.

Immunofluorescence microscopy was performed to further assess whether trypsin-sensitive SNAP-23 and syntaxin-2 resides on the surface of resting platelets. Indomethacin-treated platelets were exposed to either buffer alone or SFLLRN and then incubated in the presence or absence of trypsin. Intact platelets were then stained with anti–SNAP-23 or anti–syntaxin-2 antibodies, plated on polylysine-coated glass slides, and visualized by immunofluorescence microscopy. Staining of activated platelets was markedly brighter than staining of resting platelets (Figure 6D). However, staining of resting platelets with anti–SNAP-23 or anti–syntaxin-2 antibody was significantly greater than that with nonimmune antibody—increased 3.2-fold ± 0.7-fold for SNAP-23 (based on quantification of 3762 platelets) and 2.1-fold plus or minus 0.6-fold for syntaxin-2 (based on quantification of 1353 platelets). These data support the supposition that a population of SNAP-23 and syntaxin-2 is exposed on the extracellular surface of resting platelets.

To further evaluate the possibility that SNAP-23 and syntaxin-2 are located on the extracellular surface of resting platelets, we used enzymes to interfere with their ability to associate with membranes. SNAP-23 binds membranes via a central membrane binding domain that contains 5 palmitoyl moieties.^41,42  Acyl-protein thioesterase 1 (APT1) specifically removes palmitate from palmitoylated proteins.^32,43,44  Incubation with APT1 did not activate platelets, as evidenced by lack of P selectin surface expression (data not shown), or compromise membrane integrity, as evidenced by the fact that platelets incubated with APT1 remained impermeant to sulforhodamine (Figure 7A). Following incubation of pacified, resting platelets with APT1, platelets were pelleted by ultracentrifugation. High-speed pelleting was used to remove platelet microparticles from the supernatants. Supernatants were then evaluated for SNAP-23 by immunoblot analysis. Incubation of resting platelets with APT1 resulted in the release of SNAP-23 into the supernatant (Figure 7B). Botulinum toxin C light chain is the catalytic subunit of botulinum toxin C, which acts by specifically cleaving certain syntaxin isoforms, including syntaxin-2.⁴⁵  Botulinum toxin C light chain cleaves syntaxin-2 near the C terminus, removing the transmembrane domain and allowing syntaxin-2 to dissociate from the membrane. Incubation with botulinum toxin C neither activated platelets (data not shown) nor compromised membrane integrity (Figure 7A). A syntaxin-2 cleavage product was observed following immunoblot analysis of pacified, resting platelets incubated with botulinum toxin C light chain (Figure S2). This cleavage product was subsequently isolated from supernatants of botulinum toxin C light chain–treated platelets (Figure 7B). These results provide independent confirmation that SNAP-23 and syntaxin-2 are present on the surface of resting platelets.

SNAP-23 and syntaxin isoforms are capable of forming binary complexes.⁴⁶  Furthermore, interactions of SNAP-23 and syntaxins are thought to influence the trafficking of these SNARE proteins during protein sorting.⁴⁷  We therefore sought to determine whether extracellular SNAP-23 and syntaxin-2 form a complex. We exposed platelets to APT1 and botulinum toxin C light chain, pelleted platelets and microparticles, and immunoprecipitated SNAP-23 or cleaved syntaxin-2 from supernatants. Immunopreciptation of SNAP-23 or syntaxin-2 was detected by immunoblot analysis (Figure 7C). Antibody directed at SNAP-23 immunoprecipitated both SNAP-23 and syntaxin-2 from the supernatants. Conversely, antibody directed at syntaxin-2 immunoprecipitated both SNAP-23 and syntaxin-2. Nonimmune antibody failed to immunoprecipate either SNAP-23 or syntaxin-2 (data not shown). These results demonstrate that a complex of SNAP-23 and syntaxin-2 is found in supernatants following exposure of intact platelets to APT1 and botulinum toxin C light chain, indicating that extracellular SNAP-23 and syntaxin-2 retain the ability to bind to one another.

We have used preembedding immunonanogold staining to evaluate SNARE protein localization prior to and following platelet stimulation. This technique enables preservation of platelet ultrastructure, providing clear demarcation of membrane boundaries and demonstrating many of the hallmarks of platelet activation, including pseudopod formation and degranulation. Stimulation of platelets with either SFLLRN or PMA resulted in increased plasma membrane localization of SNAP-23 and syntaxin-2 (Figures 1–2). Total staining of SNAP-23 and syntaxin-2 was also increased following platelet activation. Increased total staining does not result from generation of new antigen because platelets were fixed 10 minutes following activation and could not have generated new protein in time. Increased staining likely results from enhanced epitope availability. We have previously demonstrated disassembly of the trimeric SNARE protein complex in platelets following activation with SFLLRN.¹⁴  SNARE protein complex disassembly may expose SNARE protein epitopes and enable enhanced antibody binding. In addition, SNARE proteins interact with several chaperones, including N-ethylmaleimide–sensitive fusion protein, α-SNAP, γ-SNAP, Munc-18c, Rab isoforms, and others.^15,17,48  Increased staining may result from release of SNAP-23 and/or syntaxin-2 from chaperone proteins. In addition to increased overall staining of SNAP-23 and syntaxin-2, immunonanogold staining suggested the presence of these SNARE proteins on the extracellular surface of platelets.

The observation that immunonanogold staining showed SNAP-23 and syntaxin-2 localized to the platelet surface was an unexpected finding that required confirmation using alternative strategies. Several observations are consistent with the assertion that a population of SNAP-23 and syntaxin-2 is present on the extracellular surface of the platelet plasma membrane. Flow cytometry of intact, activated platelets demonstrated staining of SNAP-23 and syntaxin-2 (Figure 4). In contrast, neither immunonanogold staining nor flow cytometry demonstrated expression of VAMP-3 on the extracellular surface. Exposure of activated intact platelets to trypsin resulted in a reduction of SNAP-23 and syntaxin-2 as measured by flow cytometry (Figure 5) or immunofluorescence microscopy (Figure 6), providing further evidence that a population of these proteins resides on the extracellular surface. More specific enzymatic treatments using APT1 and botulinum toxin C also resulted in loss of SNAP-23 and syntaxin-2 from intact platelets (Figure 7). Comparing staining of intact, SFLLRN-activated platelets with that of digitonin-permeabilized platelets suggests that nearly 20% of SNAP-23 and syntaxin-2 is localized to the plasma membrane. To our knowledge this is the first report to demonstrate the presence of SNARE proteins on the extracellular surface of a cell.

How do platelet SNAP-23 and syntaxin-2 localize to the platelet surface? The marked increase in staining of surface SNAP-23 and syntaxin-2 by both immunonanogold localization and flow cytometry initially suggested to us that the phenomenon is activation dependent. One possibility is that a population of SNAP-23 and syntaxin-2 is sequestered in platelet granules and exposed following activation-induced granule release. This mechanism of activation-dependent antigen exposure is observed with several platelet proteins, including P selectin and CD63.^49,50  However, neither P selectin nor CD63 are found on the surface of resting platelets. In contrast, immunonanogold localization and immunofluorescence microscopy showed SNAP-23 and syntaxin-2 on the resting platelet surface. In addition, we were not able to detect SNAP-23 or syntaxin-2 in the granules of resting platelets. Furthermore, a population of SNAP-23 and syntaxin-2 was released from resting, intact platelets following exposure to APT1 or botulinum toxin C light chain, respectively. SNAP-23 and syntaxin-2 on the extracellular surface of resting, intact platelets were also susceptible to cleavage by trypsin. These observations are inconsistent with the supposition that SNAP-23 and syntaxin-2 translocate from intracellular platelet stores to the extracellular surface only upon activation.

Another potential explanation for the presence of SNAP-23 and syntaxin-2 on the platelet surface is that these proteins traverse the plasma membrane. Certain amino acid sequences termed protein transduction domains have been described to engender peptides and proteins with the ability to cross lipid bilayers,^51-57  albeit with relatively low efficiency.⁵⁸  Of note, syntaxin-2 possesses a potential protein transduction motif in its linker domain (amino acids 141 to 150) and a second adjacent to its transmembrane domain (amino acids 251 to 264). Palmitoylation has also been used to confer peptides with the ability to cross lipid bilayers.^59-61  SNAP-23 and SNAP-25 contain a central membrane binding domain containing 5 or 4 palmitate moieties, respectively.²⁴  This domain directs membrane targeting of SNAP-25 and is itself fusogenic.⁴¹  Whether protein transduction domains or a palmitoylated membrane binding domain endow syntaxin-2 or SNAP-23, respectively, with the ability to traverse a lipid bilayer has not been evaluated.

A third possibility is that a population of SNAP-23 and syntaxin-2 is sorted to an extracytoplasmic compartment during protein trafficking in megakaryocytes. Syntaxin isoforms are type IV integral membrane proteins that are targeted to membranes following translation rather than cotranslationally. Their intracellular trafficking is directed by both their cytoplasmic and C-terminal transmembrane domains.^27-29,62  However, the targeting of type IV integral membrane proteins is not directed by a particular consensus sequence, and relatively little is known about the determinants of membrane orientation of syntaxin isoforms. Similarly, relatively little is known about the membrane interactions of SNAP-23 that occur during protein trafficking. Association of SNAP-25 and syntaxin isoforms in the endoplasmic reticulum and Golgi of nucleated cells has been demonstrated, and this association alters the membrane trafficking of both proteins.^47,63  It is possible that a population of SNAP-23 and syntaxin-2 forms a complex following translation that is preferentially trafficked to an extracytosolic compartment, while unassociated or chaperone-associated counterparts (eg, Munc-18 isoform-associated syntaxins) are trafficked to intracellular compartments. SNAP-23 and sytnaxin-2 released from intact platelets by exposure to APT1 and botulinum toxin C light chain were coimmunoprecipitated by either anti–SNAP-23 or anti–syntaxin-2 antibodies (Figure 7). One interpretation of this observation is that at least some extracellular SNAP-23 is in complex with syntaxin-2 on the platelet surface. Alternatively, SNAP-23 and syntaxin-2 may have formed a complex in solution following cleavage. In either case, this observation demonstrates that extracellular tSNAREs retain the ability to bind one another. Future studies evaluating trafficking of SNAP-23 and syntaxin-2 during megakaryocyte development will be required to determine whether syntaxin-2 influences the localization, and particularly the extracellular surface localization, of SNAP-23.

The observation that a population of SNAP-23 and syntaxin-2 exists on the extracellular surface of platelets has certain implications in the study of SNARE protein biology. The extracellular surface population of SNARE proteins may respond differently to platelet activation than intracellular SNARE proteins. The fact that intracellular and extracellular populations of SNARE proteins exist in platelets will need to be considered in interpreting studies evaluating the regulation of SNARE proteins in activated platelets. This consideration may also impact on studies directed at identifying SNARE protein binding partners. The role of SNARE proteins on the extracellular surface of platelets remains speculative. Platelet-platelet fusion has previously been observed,^64,65  and extracellular SNARE proteins could function in this process. This possibility is particularly intriguing in light of recent studies demonstrating that HeLa cells engineered to express extracellular SNARE proteins fuse to one another.^66,67  It is not known whether SNARE proteins naturally occur on the extracellular surface of cells other than platelets. Relatively few microscopy studies have been performed with adequate resolution to identify extracytosolic orientation of these membrane-associated proteins. Alternatively, the unique mechanism of platelet formation raises the possibility that localization of SNARE proteins to the extracellular surface is a platelet-specific phenomenon that could be exploited to enhance our understanding of membrane dynamics during platelet formation.

This work was supported by National Institutes of Health (NIH) grants AI33372 (A.M.D.), AI44066 (A.M.D.), and HL63250 (R.F.). R.F. is a recipient of an American Society of Hematology Junior Faculty Scholar Award, a Special Project Award from Bayer Healthcare, and a Grant-In-Aid from the American Heart Association.

We thank Tracy Sciuto, Rita A. Monahan-Earley, and Kathyrn Pyne for photographic assistance.

National Institutes of Health

Contribution: R.F. conceived and designed research, analyzed and interpreted data, and drafted the manuscript; N.R. performed all studies not directly related to electron microscopy; D.F. performed all electron microscopy; and A.M.D. supervised electron microscopy, analyzed and interpreted electron micrographs, and edited the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Robert Flaumenhaft, Division of Hemostasis and Thrombosis, Department of Medicine, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Boston, MA, 02115; e-mail: rflaumen@bidmc.harvard.edu.