In the present study, we have shown that VAMP3, syntaxin13 and SNAP23 form a complex during CHO cell adhesion to FN. While previous studies have documented the existence of a VAMP3-SNAP23-containing SNARE complex , herein is the first evidence that formation of such a complex can be stimulated by cell-ECM interaction. The data are also the first to implicate the function of the endosomal SNARE syntaxin13 in cell adhesion and migration. Inhibition of syntaxin13 impaired cell spreading and migration to an extent that was quantitatively similar to the impairment caused by inhibition of VAMP3 and the effects of inhibiting VAMP3 and syntaxin13 during cell spreading were not additive, suggesting that these SNAREs function along the same trafficking pathway. This conclusion is consistent with our detection of SNAP23 in association with VAMP3 that had been immunoprecipitated from syntaxin13 immunoprecipitate (Fig. 1D). Sequential immunoprecipitation of endogenous proteins is not an efficient process, often providing poor protein recovery, and detection of these interactions here, in response to cell-ECM interaction, is an important observation. Furthermore, evidence of the VAMP3-syntaxin13-SNAP23 complex was obtained through analysis of both endogenous proteins (Fig. 1) and ectopically expressed constructs (Additional file 3), and in both cases the SNARE interactions were enhanced by cell adhesion to ECM substrate.
The cellular importance of the SNAP23-syntaxin13-VAMP3 complex is revealed by data showing that cell spreading and β1 integrin exocytosis were impaired when the function of these SNAREs was disrupted. Our findings are consistent with much evidence indicating that cellular adhesion requires regulated membrane traffic, including the recycling of integrins and the redistribution of integrins to sites of lamellipodium protrusion [3, 5, 6, 35]. Recent studies directly implicate VAMP3 in trafficking of integrins during CHO cell adhesion  and SNAP23 in focal adhesion formation . In the present study, we now report that inhibiting SNAP23 reduced cell spreading by approximately 75%, whether VAMP3 was inhibited or not, consistent with a model in which the functions of these two SNAREs during cell adhesion are part of a common pathway. Overall, these results argue that the SNAP23-syntaxin13-VAMP3 interaction is functionally important to this cell-ECM interaction.
Our findings do not exclude the possibility of other SNAREs being involved in the formation of lamellipodia, but they do suggest that SNAP23 function may be central in this process. It is possible that SNAP23 functions at the plasma membrane and at internal endosomes to facilitate the delivery of secreted and/or recycled components during lamellipodium extension. Other SNAREs (e.g. v-SNAREs in endosomal compartments) may function in compartment-specific pathways, but all these pathways may, in turn, depend on SNAP23 function. If so, then SNAP23CΔ9 would be expected to be a more potent inhibitor of cell motility than other soluble SNARE domains, as was observed. In support of this model, others have reported that a VAMP2-syntaxin13-SNAP25 complex mediates traffic from a sorting endosome to a recycling endosome in neurons [18, 36]. In non-neuronal cells, a similar complex, such as the VAMP3-syntaxin13-SNAP23 complex described here, may be involved in an analogous trafficking pathway. Our findings confirm that blocking syntaxin13, VAMP3 or SNAP23 function interrupts β1 integrin and transferrin trafficking to a perinuclear recycling compartment, causing extensive colocalization with Rab4 (a sorting endosome marker; Additional file 4). These results are the first demonstration of a SNARE complex that functions in transport from a sorting endosome to a recycling endosome in non-neuronal cells.
The effects of SNARE inhibition on cell spreading and migration suggest that SNARE function is required to support the localized extension of the plasma membrane to form lamellipodia. It is interesting to note that, compared to the effects on lamellipodium extension (Fig. 2A), relatively modest effects were observed in cell migration (Fig. 2B). This may be explained by the fact that the cells do retain some capacity to form protrusions, albeit at a reduced rate (see Fig. 3B), and this impaired protrusive capacity supports some degree of cell migration while having a more negative impact on cell spreading. The results of live cell imaging, presented in Fig. 3B, reveal that when the function of syntaxin13, VAMP3 or SNAP23 is disrupted lamellipodia do form, but the protrusion velocity is significantly reduced. It has been demonstrated that protrusion velocity is dependent on dynamic actin polymerization at the edge of lamellipodia, independent of integrin function [37, 38]. Here, we present evidence that PMA-induced F-actin dependent membrane ruffle formation was impaired by blocking SNAP23. Since membrane ruffle formation is not dependent on the attachment of membrane protrusions to a substratum, these results lead to the conclusion that the function of SNAP23 is required for the remodeling of the plasma membrane that is the basis for lamellipodium formation. In this context, SNARE-mediated traffic may be necessary for the insertion of membrane, from intracellular compartments, into the plasma membrane, or the trafficking of actin regulating proteins to the leading edge. It is also possible that other plasma membrane SNAREs (syntaxins 2-4) are involved in remodeling the plasma membrane. For example, syntaxin3 has been shown to be involved in cell membrane expansion . The cells studied here do express syntaxin4 and we are currently developing assays to examine the function syntaxin4 and other SNAREs in cell membrane remodelling events.
The fact that inhibition of VAMP4 or GS15 impaired membrane ruffle formation is consistent with a previous study that suggested PMA-induced ruffle formation is dependent on Arf1 activity . Arf1 is a well known regulator of the biosynthetic-secretory pathway and has also been ascribed important signaling functions leading to actin remodeling [41, 42]. Given the results presented here, it is likely that VAMP4 and GS15 function in the secretory pathway along with Arf1. GS15 localizes to the Golgi, and GS15 function has been linked to retrograde traffic within the Golgi and trans-Golgi network [28, 43]. VAMP4 is known to mediate retrograde traffic from endosomes and secretroy granules to the TGN [23, 44–47], and has been implicated in GLUT4 anterograde traffic from the Golgi . VAMP4 may also be directly involved in trafficking to the plasma membrane as a minor amount of VAMP4 is found at the plasma membrane [our unpublished data and ]. In the present study, our observations indicate that the functions of VAMP4 and GS15 are required for membrane ruffle formation in response to PMA, but the activity of these SNAREs is not necessary for ECM-stimulated lamellipodium extension. Conversely, the functions of VAMP3 and syntaxin13 are required for the efficient formation of lamellipodia during cell spreading on ECM, but not PMA-induced membrane ruffling. Therefore, these data indicate that SNARE-mediated membrane traffic is important for both lamellipodium extension and membrane ruffle formation, but that the SNARE-mediated pathways involved in these two processes are not identical.