Multisubunit tethering complexes (MTCs) positively regulate vesicular fusion by as yet unclear mechanism. In this study we provide evidence that the MTC COG enhances the assembly of fusogenic Golgi SNARE complexes and concomitantly prevents nonfusogenic tSNARE interactions. This capability is possibly mediated by multiple direct interactions of COG subunits and specific Golgi SNAREs and SM (Sec1/Munc18) proteins. By using a systematic co-immunoprecipitation analysis, we identified seven new interactions between the COG subunits and components of the Golgi fusion machinery in mammalian cells. Our studies suggest that these multivalent interactions are critical for the assembly of fusogenic SNARE complexes on the Golgi apparatus and consequently for facilitating endosome-to-trans-Golgi network (TGN) and intra-Golgi retrograde transport, and also for coordinating these transport routes.
Tethering factors are a large group of proteins and protein complexes that appear to link transport vesicles to their cognate target membranes. In general, tethering factors can be divided into two major groups: elongated coiled-coil proteins and multisubunit tethering complexes (MTCs) (Bröcker et al., 2010; Sztul and Lupashin, 2006; Whyte and Munro, 2002; Yu and Hughson, 2010). Tethers often cooperate with Rab small GTPases, SNAREs and vesicle coats, and are thought to be involved in vesicular capturing and docking to specific target membranes (Angers and Merz, 2011; Cai et al., 2007). The involvement of tethering factors in vesicular recognition and targeting, and their contribution to membrane fusion specificity have been proposed by numerous studies (Sztul and Lupashin, 2006). Yet, their involvement in the subsequent steps of vesicular fusion has not been fully explored.
Increasing lines of evidence suggest that tethering factors in general, and MTCs in particular, play important role in the assembly of SNARE complexes (Pérez-Victoria and Bonifacino, 2009; Ren et al., 2009; Shorter et al., 2002). Consistent with this view, we previously showed that the MTC Conserved Oligomeric Golgi (COG) positively regulates the assembly of Syntaxin5 (Stx5)-GS28-Ykt6-GS15 and Stx6-Stx16-Vti1a-VAMP4 SNARE complexes (Laufman et al., 2011; Laufman et al., 2009). These Golgi SNARE complexes and their corresponding SM (Sec1/Munc18) proteins Sly1 and Vps45, regulate intra-Golgi and endosome-to-TGN retrograde transport, respectively (Li et al., 2005; Mallard et al., 2002; Rahajeng et al., 2010; Xu et al., 2002). Although, the mechanisms by which COG regulates the assembly of these SNARE complexes have not been fully resolved, our previous studies suggest that the direct interactions between the Cog4 subunit of the complex, Sly1, and Stx5, and between the Cog6 subunit and Stx6, are essential for SNARE complex assembly in mammalian cells (Laufman et al., 2011; Laufman et al., 2009).
The assembly of trans-SNARE complex or SNAREpin consisting of three target SNARE (tSNARE) motifs and one vesicle SNARE (vSNARE) motif into a twisted parallel four-helix bundle is the driving force for membrane fusion. This assembly brings the target membranes and the vesicle close together and catalyzes their fusion (Hong, 2005; Jahn et al., 2003; Malsam et al., 2008; Söllner, 2003). SNAREpin assembly occurs in a stepwise manner and involves the formation of SNARE complex intermediates. The assembly of functional tSNARE complex consisting of three tSNAREs on the target membrane is a pre-requirement for vSNARE binding. SNAREs can be assembled into numerous different complexes in solution, but only few are functional and can drive membrane fusion (McNew et al., 2000). Hence, functional SNARE complexes that drive membrane fusion are called fusogenic, whereas SNARE complexes unable to promote membrane fusion are called non fusogenic. The in vitro assembly of physiological relevant SNARE complexes appears to be very inefficient (Ohya et al., 2009; Stroupe et al., 2009). These features of SNARE complex assembly raise two key questions: how SNAREs are assembled into fusogenic SNARE complexes in intact cells, and how the production of nonfusogenic SNARE complexes is prevented on a specific target membrane, such as the Golgi apparatus.
In this study we show that the COG, a Golgi localized MTC consisting of eight subunits (Chatterton et al., 1999; Loh and Hong, 2002; Podos et al., 1994; Suvorova et al., 2002; Ungar et al., 2002; Ungar et al., 2006; VanRheenen et al., 1999; Whyte and Munro, 2001), enhances the assembly of fusogenic SNARE complexes both in vitro and in intact cells. We provide evidence that COG interacts directly with multiple Golgi SNAREs and SM proteins, and that its Cog4 subunit plays a central role in this interactions network. We further show that depletion of Cog4 expression enhances the formation of nonfusogenic tSNARE interactions. These results suggest that multivalent interactions between the COG subunits and components of the Golgi fusion machinery are directly involved in the assembly of fusogenic SNARE complexes, thereby facilitating specific vesicular fusion events and possibly coordinating different transport routes.
COG interacts with multiple Golgi SNAREs and SM proteins
Previous studies in yeast suggest that COG interacts with several Golgi SNAREs, including Gos1p (GS28), Sec22p, Ykt6p, Sed5p (Syntaxin 5) and Vti1p (Suvorova et al., 2002). Among these interactions, only the interaction of COG with Sed5p in yeast and subsequently with Syntaxin 5 (Stx5) in mammals was further characterized and found to be direct and essential (Laufman et al., 2009; Shestakova et al., 2007). To determine whether COG interacts with additional Golgi SNAREs in mammalian cells, we applied a simple approach that relies on co-immunoprecipitation (co-IP) of the COG subunits and the Golgi SNAREs. We focused on SNAREs that belong to the Stx5-GS28-Ykt6-GS15 and the Stx6-Stx16-Vti1a-VAMP4 SNARE complexes (hereafter referred to as the Stx5 SNARE complex and the Stx6 SNARE complex, respectively) and their corresponding SM proteins, Sly1 and Vps45, respectively. In these assays, we overexpress each of the eight subunits of the COG complex as a Myc-tagged protein in HEK293 cells, immunoprecipitate the COG subunits, and examine their interactions with endogenous SNAREs or SM proteins by western blotting. As shown in Fig. 1, this analysis reliably confirmed previously established interactions including the interaction of the Cog6 subunit and Stx6 (Laufman et al., 2011), Cog4 and Stx5 (Shestakova et al., 2007), and Cog4 and Sly1 (Laufman et al., 2009). Furthermore, it revealed multiple novel interactions that have not been identified before. This sensitive analysis revealed strong interactions between the Cog4 subunit and the tSNAREs Vti1a, Stx16 and GS28, as well as with the SM protein Vps45. Interestingly, the Cog7 subunit also interacts with Stx16, GS28 and Vps45. Reciprocal co-IP experiments of overexpressed Cog4 or Cog7 subunit and their interacting SNAREs or Vps45 further confirmed these new interactions (supplementary material Fig. S1).
We then asked whether these new interactions are direct. For this purpose, we expressed the interacting SNAREs Vti1a, Stx16 and GS28 or the SM protein, Vps45, as recombinant proteins in bacteria, and examined their ability to bind either recombinant full-length Cog4 (Fig. 2A–D) or full-length Cog7 (supplementary material Fig. S2). As shown, Cog4 interacts directly and specifically with the indicated SNAREs as well as with Vps45. Direct interaction between Cog7 and Stx16 or Vps45 was also observed (supplementary material Fig. S2).
Our further characterization of these interactions has been focused on Cog4, which mediates most of the newly identified interactions. We applied several truncated Cog4 mutants and/or SNARE proteins and assessed their interactions by pull-down experiments (Fig. 2D–G). As shown, Cog4 binds the SNARE domain of Vti1a (aa 121–193, Fig. 2E). It also interacts with the SNARE domain of Stx16 (aa 227–302, Fig. 2F), and even more strongly with a fragment consisting of the SNARE domain and its flanking linker region (aa 187–302). Vps45, on the other hand, interacts with the N-terminal region (aa 1–226) of Stx16 (Fig. 2F), suggesting that Stx16 interacts with Cog4 and Vps45 via distinct binding sites; its SNARE motif and its N-terminal region (Fig. 2F), respectively.
Binding analysis of Stx16 (Fig. 2G; supplementary material Fig. S3), Vps45 (Fig. 2D) and Vti1a (supplementary material Fig. S3) to Cog4 truncated mutants revealed that all of them interact with its N-terminal coiled-coil fragment (aa 1–231). However, Vti1a also interacts with the C-terminal fragment of Cog4 (aa 232–785; supplementary material Fig. S3), suggesting the existence of more than one binding site for Vti1a in Cog4. Overall, these results along with our previous studies (Laufman et al., 2009) suggest that the N-terminal region (aa 1–231) of Cog4 interacts directly with multiple Golgi tSNAREs that belong to two distinct SNARE complexes and also with their corresponding SM proteins Sly1 and Vps45.
Cog4 is essential for the assembly of the Stx6 SNARE complex
The multiple interactions of the Cog4 subunit with SNAREs and SM proteins of the two Golgi SNARE complexes: Stx5-GS28-Ykt6-GS15 and Stx6-Stx16-Vti1a-VAMP4 prompt us to examine its influence on SNARE complex assembly. We have previously shown that the Cog4-Sly1 interaction is essential for the assembly of the Stx5 SNARE complex (Laufman et al., 2009), and therefore focused our further characterization on the Stx6 SNARE complex. We first examined the effect of Cog4 depletion by shRNA on the steady-state levels and the subcellular distribution of Stx6, Stx16, Vti1a and VAMP4 in HeLa cells. As seen, the shRNA of Cog4 efficiently downregulated the expression of Cog4 (Fig. 3A), had no marked effects on the steady-state level of its interacting tSNAREs Stx16 or Vti1a, but slightly increased the level of Stx6 (Fig. 3B). Interestingly, the level of the vSNARE, VAMP4, was much higher (∼245%) in Cog4-depleted cells as compared to control or to Cog6-depleted HeLa cells. This increase in the VAMP4 level may reflect an attempt to compensate for the tethering defect of Cog4-depleted cells, as was previously proposed for other vSNARE proteins (Pfeffer, 1996; Sapperstein et al., 1996; VanRheenen et al., 1998).
Subcellular fractionation analysis suggests that depletion of the Cog4 subunit also affects the steady-state distribution of Stx6, Stx16, Vti1a, VAMP4 and Vps45. As shown, depletion of the Cog4 subunit led to redistribution of these SNAREs and Vps45 from the heavy membrane fraction containing the Golgi complex into the light membrane fraction, which is highly enriched in vesicles (Fig. 3C). Consistent with these fractionation results, Stx6, Stx16, Vti1a and VAMP4 could partially be detected in the Golgi of Cog4-depleted HeLa cells, as determined by co-localization with the trans-Golgi network (TGN) marker Golgin 97 or with the Golgi marker p115 (Fig. 3D). It is worth mentioning that Cog4 depletion affects the Golgi morphology and consequently the typical TGN staining of Golgin 97. Nevertheless, Golgin 97 was localized to Golgi-derived structures, whereas Stx6, Stx16, Vti1a, VAMP4 (Fig. 3D) and Vps45 (supplementary material Fig. S4) could partially be detected in these structures, and mainly appeared in punctate cytosolic structures and diffused cytosolic haze. These punctate structures failed to co-localize with the early endosomal marker EEA1 (supplementary material Fig. S5), suggesting that they represent endosome-derived vesicles that failed to tether and fuse with the TGN membranes.
The marked effect of Cog4 depletion on SNAREs distribution (Fig. 3C,D) could influence the assembly of the Stx6 SNARE complex and consequently vesicular fusion at the TGN. To explore this possibility, we treated control and Cog4-depleted HeLa cells with N-ethylmaleimide (NEM), which inhibits NSF and disassembly of SNARE complexes, and assessed the formation of the Stx6 SNARE complex by co-IP experiments. As shown in Fig. 3E, depletion of the Cog4 subunit substantially reduced the interaction between Vti1a and Stx16, the two tSNAREs that bind Cog4 (middle panel). Furthermore, the interactions between VAMP4 and the tSNAREs Vti1a and Stx6 were markedly reduced despite the prominent increase in the steady-state level of VAMP4. Collectively these results suggest that Cog4 is essential for tSNARE complex assembly and consequently for the assembly of the entire Stx6 SNARE complex.
Depletion of the Cog4 subunit inhibits endosome-to-TGN retrograde transport
The profound effects of Cog4 depletion on the subcellular distribution of Stx6, Stx16, Vti1a and VAMP4 and their assembly into fusogenic SNARE complex (Fig. 3C–E), suggest that depletion of Cog4 could also affect endosome-to-TGN retrograde transport. To explore this possibility, we examined the transport of TGN38 from the plasma membrane to the TGN using an antibody uptake assay as we previously described (Laufman et al., 2011). TGN38/46 is a TGN resident protein that constitutively cycles between the TGN and the plasma membrane via early/recycling endosomes (Ghosh et al., 1998). In brief, control and Cog4-depleted HeLa cells were transiently transfected with HA-TGN38, incubated with anti-HA antibodies at 37°C for different time periods, washed, fixed and double immunostained for Golgin 97 and TGN38-HA. As shown in Fig. 4A, 3 min following incubation with anti-HA antibody, TGN38-HA was localized mainly at the plasma membrane in both control and Cog4-depleted cells. At 30 min following anti-HA antibody uptake, TGN38-HA was predominantly localized to the Golgi of the control cells (in 60% of the cells, n = 200) and extensively co-localized with Golgin 97 [co-localization of 65±10% (n = 30)]. In Cog4-depleted cells, however, TGN38-HA failed to co-localize with Golgin 97 [co-localization of 5±2% (n = 30)], and partially co-localized with the early endosomal marker EEA1, suggesting that its transport from early endosomes to the TGN was substantially inhibited. It is worth mentioning that TGN38-HA failed to reach the TGN of Cog4-depleted cells even 2 hours following anti-HA uptake (Fig. 4B). The impaired endosome-to-TGN transport in Cog4-depleted cells was also evident by the steady-state distribution of endogenous proteins that cycle between the endosomes and the TGN including TGN46 (TGN38) and the cation-independent mannose-6-phosphate receptor (CI-MPR). As shown in Fig. 4C, these proteins lost their characteristic Golgi localization and were dispersed throughout the cytosol in Cog4-depleted cells. Co-staining with EEA1 suggests that under steady-state conditions TGN46 is also not localized to early endosomes. Collectively, these results suggest that the Cog4 subunit interacts directly with several components of the endosome-to-TGN retrograde transport machinery including Stx16, Vti1a and Vps45, and that Cog4 is essential for the assembly of the Stx6 SNARE complex and for endosome-to-TGN retrograde transport.
Cog4 facilitates the assembly of fusogenic Stx6 SNARE complex
The ability of Cog4 to interact with multiple components of the Golgi fusion machinery, in most cases via the same N-terminal fragment, suggests that Cog4-SNAREs interactions could facilitate the assembly of fusogenic Golgi SNARE complexes and consequently their regulated fusion events.
To explore this hypothesis, we examined whether Cog4 could enhance the assembly of the Stx6-Stx16-Vti1a tSNARE complex in vitro. For this purpose, we expressed HA-tagged Stx6 and Vti1a in HEK293 cells together with either Myc-Cog4 or with Myc-Vps45 and assessed their interaction with recombinant GST-Stx16 immobilized on glutathione agarose beads by pull-down experiments. As shown in Fig. 5A, Stx6 and Vti1a could bind the immobilized GST-Stx16 but not GST. The presence of Vps45 enhanced their interactions (by 143±10% for Vti1a and by 223±12.3% for Stx6), consistent with previous reports (Struthers et al., 2009). Remarkably, the addition of Cog4 subunit further enhanced the binding of Vti1a and Stx6 to Stx16 (by 240±8.5% and 400±15%, respectively), suggesting that the Cog4 subunit facilitates tSNARE complex assembly.
To further confirm these results, we examined whether Cog4 can also enhance the binding of Stx6 and Stx16 to recombinant GST-Vti1a immobilized on glutathione beads. As shown in Fig. 5B, co-expression of Stx6-HA and Stx16-HA in HEK293 cells together with Myc-Cog4 increased the binding of these tSNAREs to GST-Vti1a by 184±12% and 158±9%, respectively. Altogether, these results suggest that Cog4 enhances tSNAREs assembly in vitro.
We then asked whether Cog4 also influences the assembly of tSNARE complex and subcomplexes in intact cells. To this end, we co-expressed the two tSNAREs Myc-Stx16 and HA-Vti1a or the three tSNAREs Myc-Stx16, HA-Vti1a and HA-Stx6, together with either HA-Vps45 or HA-Cog4 (Fig. 5C). Subsequently, we examined the influence of Cog4 or Vps45 on the Stx16-Vti1a and Stx16-Vti1a-Stx6 interactions by co-IP. As shown, the interaction between Stx16 and Vti1a substantially increased in the presence of Stx6. Vps45, and even more profoundly Cog4, further enhanced the interaction between Stx16 and Vti1a, and also between Stx16, Vti1a and Stx6. These results suggest that Cog4 could enhance the formation of tSNARE complex and also its subcomplexes in intact cells.
To demonstrate that Cog4 directly mediates this effect and can also influence the assembly of the entire Stx6 SNARE complex, we expressed Vti1a, Stx6, VAMP4 and Cog4 as recombinant His-tagged proteins and Stx16 as GST-fusion protein in bacteria. We then assessed the influence of Cog4 on the binding of VAMP4 to immobilized GST-Stx16 in the presence of recombinant Vti1a and Stx6 using pull-down experiment. As shown in Fig. 5D, the binding of VAMP4 was markedly increased in the presence of recombinant Cog4, suggesting that Cog4 directly enhances the assembly of the entire SNARE complex. Collectively, these results suggest that Cog4 enhances the assembly of fusogenic SNARE complexes both in vitro and in intact cells, which are consistent with the proposed role of tethering factors as accelerators of SNARE complex assembly and vesicular fusion (Bröcker et al., 2010).
Cog4 interacts with Golgi SNARE complexes in a mutually exclusive manner and prevents nonfusogenic SNARE interactions
Our results, thus far, suggest that Cog4 plays a central regulatory role in the assembly of fusogenic Golgi SNARE complexes. We hypothesized that its ability to interact with two distinct Golgi SNARE complexes that mediate distinct transport routes might prevent the assembly of nonfusogenic SNARE subcomplexes on the Golgi membrane and could also coordinate different transport events.
To evaluate the influence of Cog4 on the formation of nonfusogenic SNARE subcomplexes on the Golgi membrane, we examined cross-interactions between SNAREs that belong to the two distinct SNARE complexes, Stx5 and Stx6 SNARE complexes, in control and Cog4 knockdown (KD) stable HeLa cell lines. The interactions between the SNAREs were assessed by co-IP experiments. As shown in Fig. 6A, the interaction of Stx5 and its associated SNAREs GS28 and GS15 was markedly reduced in Cog4-depleted cells, consistent with previous studies (Laufman et al., 2009; Shestakova et al., 2007). Remarkably, however, the interaction between Stx5 and Stx6, which belong to two distinct SNARE complexes, was markedly increased (∼twofold) in Cog4-depleted cells as compared to control cells. This was also evident in the reciprocal co-IP experiment, in which we examined the interaction between Stx5 and Stx6 following immunoprecipitation of Stx6 (Fig. 6B). These results suggest that the multiple interactions of Cog4 with Golgi SNAREs not only enhance the assembly of fusogenic Golgi SNARE complexes, but also prevent nonfusogenic SNARE interactions.
The ability of Cog4 to interact with two distinct Golgi SNARE complexes may provide a mechanism for coordinating different Golgi-associated transport routes: intra-Golgi and endosome-to-TGN retrograde transport. We therefore asked whether Cog4 can interact simultaneously with members of the two distinct SNARE complexes, or possibly its interaction with one SNARE complex inhibits its interaction with the other one. To experimentally address these possibilities, we performed in vitro competition-binding assays using recombinant purified SNAREs, SM proteins and Cog4. To examine the influence of Cog4-Stx16 interaction on Cog4-Stx5 interaction, we pre-incubated constant amount of recombinant Cog4 with increasing concentrations of recombinant Stx16, and then assessed the binding of Cog4 to GST-Stx5 immobilized on beads by pull-down experiment (Fig. 6C). Similar strategy was applied for assessing the influence of Cog4-Vps45 interaction on Cog4-Sly1 interaction (Fig. 6D). As shown, pre-incubation of Cog4 with increasing concentrations of Stx16 significantly reduced the binding of Cog4 to GST-Stx5 (Fig. 6C). Likewise, pre-incubation of Cog4 with increasing concentrations of Vps45 significantly reduced the binding of Cog4 to GST-Sly1 (Fig. 6D). We then examined whether the two distinct SNARE complexes can compete for Cog4 binding in intact cells. To this end, we co-expressed Myc-Cog4 with increasing concentrations of HA-Vti1a in HEK293 cells, and then examined the interactions between Myc-Cog4 and endogenous Stx5 by co-IP analysis. As shown in Fig. 6E, increasing amounts of HA-Vti1a significantly reduced the binding of Myc-Cog4 to endogenous Stx5. Altogether these results strongly suggest that the two Golgi SNARE complexes can compete on Cog4 binding and that this competition could provide a mechanism for coordinating intra-Golgi and endosome-to-TGN retrograde transport in intact cells.
In this study we establish a comprehensive interactions map between the different COG subunits and two distinct Golgi SNARE complexes, the Stx5 and Stx6 SNARE complexes using a simple and reliable co-IP analysis (Fig. 1). This analysis revealed seven novel interactions between the COG and components of the Golgi fusion machinery, and shed new light on the complexity and regulation of membrane fusion.
The interactions network that emerged from our studies clearly shows that the Cog4 subunit occupies a central position and mediates multiple interactions; it interacts directly with two tSNAREs of each complex: Stx5 and GS28, and Stx16 and Vti1a, and with their corresponding SM proteins Sly1 and Vps45, respectively, via its N-terminal coiled-coil domain (Figs 1, 2). This symmetry of Cog4 interactions with the two distinct SNARE complexes suggests that Cog4 plays a similar role in their complex assembly. Consistent with this hypothesis, we previously showed that Cog4 contains two adjacent binding sites for Sly1 and Stx5 in its N-terminal fragment, and that Cog4, Sly1 and Stx5 can form a ternary complex. We also showed that Cog4-Sly1 interaction is essential for SNARE complex assembly. Accordingly, we proposed that Cog4 via its simultaneous interactions with Sly1 and Stx5 brings them to close apposition, thereby facilitating SNARE complex assembly and possibly stabilizes the entire complex (Laufman et al., 2009).
In this study, we found that in addition to Stx5 and Sly1, Cog4 also interacts with GS28. These results suggest that Cog4 via its multiple interactions with tSNAREs is essential for the assembly of tSNARE complex, which provides a template for vSNARE binding on the Golgi complex. Indeed, both in vitro and in vivo studies demonstrated that Cog4 can facilitate the assembly of tSNARE complex and subcomplexes (Fig. 5). We could reconstitute the assembly of Stx6-Stx16-Vti1a in the absence or presence of either Vps45 or Cog4, and showed that both Cog4 and Vps45 enhance tSNARE complex assembly. While the stimulatory effect of Vps45 on SNARE complex assembly has been previously shown (Struthers et al., 2009), the influence of Cog4 on tSNARE assembly is shown here for the first time. The entire COG complex, which can bind directly the three tSNAREs, might have an even stronger effect on tSNAREs assembly.
In addition to tSNAREs assembly, we found that Cog4 enhances the assembly of the entire Stx6 SNARE complex in vitro (Fig. 5D). Furthermore, its depletion by shRNA substantially impairs the assembly of this complex in intact cells (Fig. 3E) and consequently attenuates endosome-to-TGN retrograde trafficking (Fig. 4). These results suggest that the Cog4 subunit, and consequently the entire COG complex, positively regulates the assembly of fusogenic Golgi SNARE complexes.
Strikingly, we also found that Cog4 depletion enhances the formation of nonfusogenic tSNARE interactions (Fig. 6A,B). These results suggest that COG positively regulates the assembly of fusogenic SNARE complexes and also prevents the formation of nonfusogenic SNAREs interactions.
The promiscuous interactions of Golgi SNAREs have been previously demonstrated by in vitro binding studies (Tsui and Banfield, 2000). Yet, the number of nonfusogenic SNARE interactions that exist in intact cells remains unknown. Our results clearly show that the amount of such complexes is much higher in the absence of functional COG complex (Fig. 6A,B). The effect of Cog4 depletion on nonfusogenic tSNAREs interactions could be related to its ability to bind the SNARE motif of its interacting tSNAREs (Fig. 2) (Laufman et al., 2011; Shestakova et al., 2007), and therefore to compete with the binding of other Golgi tSNAREs. Alternatively, Cog4 may induce conformational changes in its interacting tSNARE, thereby preventing promiscuous tSNAREs interactions and consequently nonfusogenic assembly. Currently, we cannot exclude either of these possibilities. Nevertheless, the MTC Golgi-associated retrograde protein (GARP) also interacts with the SNARE motifs of Golgi SNAREs via the N-terminal coiled-coil domains of its Vps53 and Vps54 subunits (Pérez-Victoria and Bonifacino, 2009). Hence, it would be interesting to examine the effect of GARP depletion, which also regulates endosome-to-TGN retrograde transport (Pérez-Victoria and Bonifacino, 2009; Pérez-Victoria et al., 2008; Pérez-Victoria et al., 2010) on the formation of nonfusogenic SNARE interactions.
The ability of several subunits of MTCs to interact simultaneously with SNAREs and positively regulate SNARE complex assembly was also proposed for GARP and Dsl1. The Tip20 and Sec39 subunits of the Dsl1 complex, which regulates Golgi-to-ER transport and accelerates SNARE complex assembly in vitro, interact directly with the N-terminal regulatory domains of the tSNAREs Use1 and Sec20. These interactions are necessary for SNARE complex assembly (Diefenbacher et al., 2011; Kraynack et al., 2005; Ren et al., 2009; Tripathi et al., 2009). Similarly, all four subunits of mammalian GARP interact directly with Stx6, Stx16 and VAMP4, and regulate Stx6 SNARE complex assembly (Pérez-Victoria and Bonifacino, 2009; Pérez-Victoria et al., 2010). The Cog4, Cog6 and Cog7 subunits also interact with multiple Golgi SNAREs (Figs 1, 2; supplementary material Figs S1, S2) (Laufman et al., 2011). Hence, it could be that multivalent interactions between subunits of the COG complex or other MTCs and SNAREs primarily function to gather together the SNAREs needed for SNAREpin assembly. This gathering could markedly contribute to the specificity and efficiency of the fusion events, consistent with the proposed function of tethering factors (Bröcker et al., 2010). Yet, COG might also activate SNAREs for assembly, possibly through binding of SM proteins.
Although it is unclear how exactly SM proteins accelerate SNARE complex assembly (Shen et al., 2007; Südhof and Rothman, 2009), their direct interactions with Syntaxins appear to be crucial (Dulubova et al., 2003; Dulubova et al., 2002; Yamaguchi et al., 2002). Among the different SM proteins, Sly1 and Vps45 are known to interact with a short N-terminal sequence of Syntaxins; Stx5 and Stx16, respectively (Dulubova et al., 2002; Yamaguchi et al., 2002). Remarkably, our results indicate that the interactions of Cog4 with Vps45 and with the SNARE motif of Stx16 (Fig. 2E), are highly similar to its interactions with Sly1 and Stx5 (Laufman et al., 2009). These results strongly suggest that COG, SM proteins and Syntaxins employ similar mechanisms to cooperatively regulate the assembly of the two different Golgi SNARE complexes.
Consistent with this hypothesis, we found that the interaction of Cog4 with Vps45 inhibits its interaction with Sly1 (Fig. 6D). Likewise, the interaction of Cog4 with Stx16 or Vti1a inhibits its interaction with Stx5 (Fig. 6C,E). These results suggest that the components of the two SNARE complexes compete on overlapping binding sites in Cog4. Alternatively, it could be that binding of components of one complex induces conformational changes in Cog4 that prevent its interaction with components of the second complex. Regardless the exact binding mode, our results strongly suggest that COG cannot interact simultaneously with the two Golgi SNARE complexes. This selective mode of interactions could provide a mechanism for coordinating intra-Golgi and endosome-to-TGN retrograde transport that is required for maintaining the structural and functional integrity of the Golgi complex.
In summary, our results expose a novel network of interactions between the COG subunits and components of the Golgi fusion machinery, and suggest that MTCs share common mechanisms of SNARE complex assembly. Their multivalent interactions with SNAREs gather the relevant SNAREs, thereby facilitating the assembly of specific SNARE complex. Some MTCs, such as COG and the homotypic fusion and protein sorting (HOPS), also cooperate directly with SM proteins (Krämer and Ungermann, 2011; Laufman et al., 2009; Pieren et al., 2010; Seals et al., 2000), thereby may activate SNARE assembly, control and direct SNARE activity, and ensure the fidelity of intracellular transport.
Materials and Methods
Antibodies, reagents and chemicals
Polyclonal antibodies against Golgin 97, Ykt6, GS15, Stx6, Sly1 and Cog4 were previously described (Laufman et al., 2011; Laufman et al., 2009; Lu et al., 2004; Xu et al., 1997; Zhang and Hong, 2001). Polyclonal antibodies against Stx16, Stx5 and GS28 were a generous gift from Dr B. L. Tang (National University of Singapore, Singapore), Dr S. Somlo (Yale University, New Haven, CT, USA) and Dr Z. Elazar (Weizmann Institute of Science, Rehovot, Israel), respectively. Polyclonal anti-TGN46 antibody was kindly provided by Dr M. Fukuda (Burnham Institute, La Jolla, CA, USA). Polyclonal anti-VAMP4 antibody and monoclonal anti-Stx6 antibody were purchased from Synaptic Systems (Goettingen, Germany) and BD Biosciences (San Jose, CA, USA), respectively. Polyclonal anti-Vti1a and anti-Vps45 antibodies were purchased from Proteintech Group, Inc. (Chicago, IL, USA). Monoclonal anti-CI-MPR antibody and polyclonal anti-EEA1 antibody were purchased from Abcam (Cambridge, MA, USA). Monoclonal anti-α-tubulin antibody and anti-histidine tag antibody were purchased from Sigma-Aldrich (Rehovot, Israel) and AbD Serotec (Kidlington, UK), respectively. Alexa Fluor 488 donkey anti-mouse and anti-rabbit IgGs were purchased from Invitrogen (Grand Island, NY, USA). Cy3 (cyanine 3)-conjugated goat anti-rabbit and goat anti-mouse IgGs as well as the Cy5-conjugated goat anti-rabbit IgG were purchased from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA, USA). Protein-A–agarose beads were purchased from Repligen Corp (Waltham, MA, USA). Talon metal affinity resin was obtained from BD Biosciences (San Jose, CA, USA), whereas the anti-mouse IgG conjugated to agarose beads, glutathione-agarose beads, and Hoechst 33342 were purchased from Sigma-Aldrich (Rehovot, Israel).
The DNA constructs encoding the different Myc-tagged COG subunits and the short hairpin RNA (shRNA) construct targeting human Cog4 have been described previously (Laufman et al., 2009; Loh and Hong, 2002; Loh and Hong, 2004). Full length Cog2, Cog4, Cog7 and truncated Cog4 mutants were produced by subcloning of the corresponding PCR products into either the mammalian pCMV-neomycin-HA expression vector or the pQE30 bacterial expression vector (Qiagen). The following sense and anti-sense oligonucleotide primers have been used for PCR amplifications: Cog2 (aa 1–738): 5′-AAAGATCTGAGAAAAGTAGGATGAACCTGCC-3′ and 5′-ACACCCGGGTTAAGGCTGCTCTGCTGTTGC-3′; Cog4 (aa 1–785): 5′- AAAGGATCCGCGGACCTTGATTCGCCTC-3′ and 5′-ACACTCGAGTTACAGGCGCAGCCTCTTGATATC-3′; Cog4 (aa 1–231): 5′- TTCGGATCCTTGGCGGACCTTGATTCGCC-3′ and 5′- AAACTCGAGCTATCCCTCCTCATGCAAACCC-3′; Cog4 (aa 232–785): 5′-AGAGGATCCTTAAGAAAGTTCTCGGAGTACCTTTGC-3′ and 5′-ACACTCGAGTTACAGGCGCAGCCTCTTGATATC-3′; Cog7 (aa 1–770): 5′-AAAGGATCCGACTTCTCCAAGTTCCTGG-3′ and 5′-AGACCCGGGTCAGTAATTCACACTCCGCATGG-3′.
Mammalian expression vectors encoding the cytosolic fragments of the human Stx6, Stx16, and Vti1a were kindly provided by Dr S. R. Pfeffer (Stanford University School of Medicine, Stanford, CA, USA). These fragments were subcloned into the pCDFDuet-1 bacterial expression vector.
Mammalian expression vectors encoding HA-tagged full-length murine Vti1a and HA-tagged full-length rat Stx16 were kindly provided by Dr G. F. V. Mollard (Bielefeld University, Bielefeld, Germany). Full-length murine Vti1a was subcloned into the mammalian pCMV-neomycin-HA and pCMV-neomycin-Myc expression vectors. Full-length rat Stx16 (aa 1–326) was produced by subcloning of the corresponding PCR product into the mammalian pCMV-neomycin-HA and pCMV-neomycin-Myc expression vectors. The following sense and antisense primers have been used: 5′-AAAGGATCCAATGGCCACCAGGCGTTTAAC-3′ and 5′-AAAGCGGCCGCCTAGCGAGACTTAACAGCGATGAGG-3′.
Truncated rat Stx16 and murine Vti1a mutants were produced by subcloning of the corresponding PCR products into the pGEX-4T-1 (GE Healthcare) bacterial expression vector. The following sense and anti-sense oligonucleotide primers have been used for PCR amplifications: Stx16 (aa 1–302): 5′-AAAGGATCCATGGCCACCAGGCGTTTAAC-3′ and 5′-AAACTCGAGCTACTTCCGATTCTTCTTTTGATACTGTTC-3′; Stx16 (aa 1–226): 5′-AAAGGATCCATGGCCACCAGGCGTTTAAC-3′ and 5′-AAAGCGGCCGCCTACAGCACCAGCTGGTCATC-3′; Stx16 (aa 187–302): 5′-AAAGGATCCATGAAGAATCGAGAGGAAAGATCC-3′ and 5′- AAACTCGAGCTACTTCCGATTCTTCTTTTGATACTGTTC-3′; Stx16 (aa 227–302): 5′-AAAGGATCCGAGCAGAACACACTGGTGG-3′ and 5′- AAACTCGAGCTACTTCCGATTCTTCTTTTGATACTGTTC-3′; Vti1a (aa 1–193): 5′-AAAGGATCCTCTTCCGACTTCGAAGGG-3′ and 5′-AAACTCGAGTCAACGGTTCTGGATGATTCTTCG-3′; Vti1a (aa 1–120): 5′-AAAGGATCCTCTTCCGACTTCGAAGGG-3′ and 5′-AAACTCGAGTCAATCCAGCAGATGTGCCCTC-3′; Vti1a (aa 121–193): 5′-ACAGATCTAACACGGAGAGGCTGGAAAG-3′ and 5′- AAACTCGAGTCAACGGTTCTGGATGATTCTTCG-3′.
The mammalian expression vector encoding GFP-tagged murine Stx6 was kindly provided by Dr J.E. Pessin (Stony Brook University, New York, NY, USA). The full-length murine Stx6 (aa 1–255) was produced by PCR and subcloned into the mammalian pCMV-neomycin-HA vector. The following sense and anti-sense oligonucleotide primers were used: 5′-AAAGGATCCGATGTCCATGGAGGACCCC-3′ and 5′-AAACTCGAGTCACAGCACTAGGAAGAGGATCAGC-3′.
The mammalian expression vector encoding GFP-VAMP4 has been previously described (Zeng et al., 2003). Truncated VAMP4 lacking its transmembrane domain was produced by PCR and subcloned into pCDFDuet-1 bacterial expression vector. The following sense and anti-sense oligonucleotide primers were used: 5′-AACGGATCCATGCCTCCCAAGTTCAAG-3′ and 5′-AGACTCGAGTCAGGCTTTTATTTTGCATCCAC-3′.
The cDNA of murine Vps45 was PCR amplified from a mouse spleen cDNA library, sequenced and cloned into pGEX-4T-1 and pQE30 bacterial expression vectors. The following sense and anti-sense oligonucleotide primers were used: 5′-ACAAGATCTATGAATGTGGTCTTTGCTG-3′ and 5′-AAACTCGAGTCATCTTCTGTTTGCTGACCTTG-3′.
The bacterial expression vector encoding GST-fused GS28 lacking its transmembrane domain was kindly provided by Prof. Z. Elazar (Weizmann Institute of Science, Rehovot, Israel). The cDNA encoding for full-length GS28 was PCR amplified from a mouse spleen cDNA library, sequenced, and subcloned into the pCMV-neomycin-Myc mammalian expression vector. The following sense and antisense oligonucleotide primers were used: 5′-AAAGGATCCAATGGCGGCAGGGACCAGC-3′ and 5′-AAACTCGAGTCAATGGAACGCATACAGCAGC-3′.
Cell culture, transfection and immunofluorescence microscopy
HEK293 and HeLa cells were grown in DME supplemented with 10% fetal bovine serum, 100 µg/ml penicillin, and 100 µg/ml streptomycin. The cells were transfected using the calcium-phosphate method. Stable HeLa cell-lines depleted of the Cog4 subunit were previously described (Laufman et al., 2009). A stable HeLa cell line harboring an empty pSUPER-puro vector was established and used as a control. These stable HeLa cell-lines were used for SNARE complex assembly co-IP assays and for transport assays (Fig. 3E; Fig. 4A; Fig. 6A,B). Transient transfections with the Cog4 shRNA construct or the pSUPER-puro empty vector were performed for all IF localization studies and for steady-state levels and fractionation analysis (Fig. 3A–D; Fig. 4C). In brief, HeLa cells grown on coverslips or in 90-mm tissue culture dishes were transiently transfected with the Cog4 shRNA construct or pSUPER-puro vector. 24 hours after transfection, the cells were either incubated with regular medium for 72 hours and then analyzed by immunofluorescence (Fig. 3D; Fig. 4C) or were incubated with 1 µg/ml puromycin for 72 hours and then analyzed by the indicated biochemical assays (Figs. 3A–C) as described in the corresponding legends.
Immunofluorescence analysis was performed essentially as previously described (Laufman et al., 2011). The specimens were analyzed by a confocal laser-scanning microscope (LSM 510; Carl Zeiss) equipped with a 63×/1.4 oil differential interference contrast M27 objective lens (Plan Apochromat; Carl Zeiss) using the 488-, 543-, and either 405- or 633-nm excitation for fluorescein, Cy3 epifluorescence, and either 4,6-diamidino-2-phenylindole (Hoechst) or Cy5, respectively. Images were acquired using the LSM 510 software.
Cell extracts, subcellular fractionation, immunoprecipitations, and pull-down experiments
For the co-IP assays described in Fig. 1 and supplementary material Figs S1,and S3, cell extracts were prepared by solubilizing HEK293 cells in lysis buffer A (1% Triton X-100, 20 mM Hepes, pH 7.5, 100 mM NaCl, 5 mM MgCl2, 1 mM PMSF, 10 µg/ml leupeptin, and 10 µg/ml aprotinin) for 1 h on ice. Cell lysates were centrifuged at 15,000 g for 15 min at 4°C. Subsequently, supernatants were incubated for 2 hours at 4°C with anti-Myc antibody bound to agarose anti-mouse IgG beads. The beads were then washed three times with lysis buffer, boiled in SDS sample buffer, and separated by SDS-PAGE. SNARE proteins or COG subunits were detected by western blotting using the indicated antibodies. Immunoprecipitation of SNARE complexes (Fig. 3E; Fig. 6A,B) was performed after treatment with NEM essentially as previously described (Pérez-Victoria and Bonifacino, 2009).
Pull-down experiments were performed as previously described (Laufman et al., 2011).
For direct binding assays, His-tagged recombinant proteins were purified from bacteria on a column containing talon metal affinity resin according to the manufacturer's instructions (BD Biosciences). Protein interactions were examined by co-IP (Fig. 2A,B) or pull-down assays (Fig. 2C–E; supplementary material Fig. S2). For the co-IP assays, purified His-Tagged proteins were mixed in HNTG buffer (0.1% Triton, 20 mM Hepes, pH 7.5, 150 mM NaCl and 10% glycerol) and incubated for overnight at 4°C with agarose–Protein-A beads bound to the indicated polyclonal antibodies. The beads were then washed with HNTG buffer, boiled in SDS sample buffer, and analyzed by western blotting. For pull-down assays, purified recombinant His-tagged proteins were incubated in HNTG buffer with either GST or GST-fusion proteins bound to glutathione-agarose beads for overnight at 4°C. The beads were then washed and analyzed as above.
For SNARE complex assembly assays (Fig. 5), lysates of HEK293 cells expressing the indicated SNARE proteins and either Vps45 or Cog4 (Fig. 5A,B) or a mixture of His-tagged purified recombinant proteins (Fig. 5D) were incubated with either GST or GST-fusion proteins bound to glutathione agarose beads for 2 hours at 4°C (purified proteins) or overnight at 4°C (cell lysates). The beads were washed four times, and boiled in SDS sample buffer.
Subcellular fractionation was preformed essentially as we previously described (Laufman et al., 2011).
Densitometric analysis of immunoblots was performed using the ImageJ software (National Institutes of Health).
Transport assays and quantitation
TGN38 antibody uptake assay was performed essentially as we previously described (Laufman et al., 2011). For the quantitative analysis of this assay, confocal images of arbitrary fields from two independent experiments were acquired for both control and Cog4-depleted HeLa cells (n = 200) at each time point using the LSM 510 software. The images were scored visually, and colocalization between the TGN marker Golgin 97 (Fig. 4A, green) and TGN38 (Fig. 4A, red) in the Golgi region was displayed as yellow in the merged image. Cells with a yellow signal in the Golgi were counted as positive. For precise quantitation of TGN localization, the colocalization between Golgin 97 and TGN38 was measured using the colocalization function of the LSM 510 software. Colocalization was determined by measuring the ratio between Golgi-associated TGN38 fluorescence (marked by Golgin 97) to their total fluorescence in the cell. The data shown are mean values±s.d. (n = 30).
Sima Lev is the incumbent of the Joyce and Ben B. Eisenberg Chair of Molecular Biology and Cancer Research.
O.L. designed and performed the experiments and analyzed data. W.H. provided essential reagents and critical comments. S.L. designed experiments and wrote the manuscript. All authors discussed the data.
This work was supported by the Binational Science Foundation (BSF) [grant number 2011404]; and by the Kirk Center for Childhood diseases.