Role of Vesicle-Associated Syntaxin 5 in the Assembly of Pre-Golgi Intermediates

See allHide authors and affiliations

Science  30 Jan 1998:
Vol. 279, Issue 5351, pp. 696-700
DOI: 10.1126/science.279.5351.696


Syntaxins are thought to function during vesicular transport as receptors on the target membrane and to contribute to the specificity of membrane docking and fusion by interacting with vesicle-associated receptors. Here, syntaxin 5 (Syn5) was shown to be an integral component of endoplasmic reticulum–derived transport vesicles. This pool, but not the target, Golgi-associated Syn5 pool, was essential for the assembly of vesicular-tubular pre-Golgi intermediates and the delivery of cargo to the Golgi. The requirement for vesicle-associated Syn5 in transport suggests a reevaluation of the basis for operation of the early secretory pathway.

Syntaxin 5 is an essential membrane protein that mediates transport between the endoplasmic reticulum (ER) and Golgi (1, 2). It belongs to the syntaxin family of SNAP-receptors (SNAREs), which are believed to function as target membrane receptors (t-SNAREs) that mediate docking with vesicle-associated receptors (v-SNAREs) of the synaptobrevin family (3, 4). At steady state, Syn5 appears principally associated with the cis face of the Golgi (1,5, 6). To determine whether its localization is dynamic, we took advantage of the observation that proteins recycling between the ER and Golgi preferentially accumulate in pre-Golgi intermediates when cells are incubated at reduced temperature (15°C) (7, 8). Pre-Golgi intermediates consist of vesicular-tubular elements closely juxtaposed to ER export sites (9) that play a critical role in the recycling of transport components back to the ER (10, 11). Recycling requires the COPI vesicle coat complex that is distinct from the COPII coat that drives export from the ER (12, 13).

Incubation at 15°C led to a marked redistribution of Syn5 from the Golgi region at 32°C (Fig. 1, A and B) to pre-Golgi intermediates scattered throughout the peripheral cytoplasm (Fig. 1, C and D). The distribution of Syn5 at 15°C overlapped with that of p58 (Fig. 1D, inset), a molecule that actively recycles between the ER and Golgi (7, 10,11), and vesicular stomatitis virus glycoprotein (VSV-G), a transmembrane cargo molecule that accumulates in pre-Golgi intermediates at 15°C (10) (Fig. 1C, inset). Furthermore, a mutant of the Sar1 guanosine triphosphatase, Sar1[GDP], which blocks COPII coat assembly and budding (14), caused the Golgi pool of Syn5 to redistribute to the ER like p58 (10,15). Thus, Syn5 appears to recycle rapidly between the ER and pre- or cis-Golgi compartments in a manner analogous to Syn1 in the synapse, which is present on both vesicles and the cell surface (16).

Figure 1

Syn5 recycles between the ER and Golgi compartments. The distribution of Syn5 in vivo was examined by indirect immunofluorescence after incubation of NRK cells either at 32°C (A and B) or 15°C (60 min) (C and D) (43). Cells were infected with the tsO45 strain of vesicular stomatitis virus at the restrictive temperature (39.5°C) to accumulate VSV-G in the ER, before the temperature shift. Upon shift to 32°C, Syn5 overlapped with VSV-G and the Golgi marker α-1,2 mannosidase II (Man II) [arrows in (A) to (D), insets]. At 15°C, Syn5 accumulated in pre-Golgi intermediates as indicated by its overlap with VSV-G and p58 [arrowheads in (A) to (D), insets]. This accumulation could be detected as early as 15 min after shift to 15°C (not shown). No change in the distribution of the Golgi processing enzyme α-1,2-mannosidase II was observed under these conditions (15). Bar: 20 μm.

To examine the requirement for Syn5 in ER-to-Golgi transport in vitro, we added an affinity-purified Syn5-specific antibody prepared against the soluble cytoplasmic domain of Syn5 [Syn5-11 (17)] to an assay that reconstitutes the transport of VSV-G in permeabilized cells (18, 19). The antibody recognized both 42- and 35-kD isoforms (20, 21) of Syn5 on immune blots of whole-cell extracts (15). The intact antibody or Fab fragments inhibited ER–to–cis-Golgi transport in a dose-dependent manner as measured by acquisition of sensitivity to endoglycosidase D (endo D) (Fig. 2A). Addition of a soluble cytoplasmic domain fragment of Syn5 (Syn5-11) (1) also partially inhibited transport (Fig. 2B), consistent with its effects in vivo (1). Inhibition of transport by the antibody to Syn5 (anti-Syn5) was specific as it was blocked by preneutralization with Syn5-11 (Fig. 2A, inset).

Figure 2

Syn5 is required for ER-to-Golgi transport at a downstream vesicle targeting or fusion step. Anti-Syn5 or its Fab fragments (A) or Syn5-11 (B) were added to semi-intact cell transport assays (44) at the concentrations indicated. Transport was followed by measuring the conversion of VSV-G to the endoglycosidase D (endo D)–sensitive cis Golgi form. (A, inset) The assay was either not supplemented (lane a), or supplemented with 1 μg of anti-Syn5 alone (lane b) or 1 μg of anti-Syn5 (lane c) or Fab (lane d) pre-neutralized with Syn5-11 (1 μg). The amounts of VSV-G processed are reported relative to untreated controls in which ∼80% of total VSV-G was transported. (C) Microsomes were incubated for 5 min at 32°C in the absence or presence of either 1 μg of Sar1[GDP] (14) or 1 μg of anti-Syn5 (45). The relative amounts of ER-derived vesicles released into the MSS were determined by quantitative immune blotting for VSV-G (45). Mean values of duplicate samples are shown with the indicated error range.

The ability of Syn5-specific reagents to inhibit ER-to-Golgi transport in permeabilized cells could reflect either inhibition of vesicle formation or their delivery to pre- or cis-Golgi compartments or both. To distinguish between these possibilities, we used a microsome-based assay that reconstitutes the formation of COPII-coated, ER-derived vesicles (14). This assay uses differential centrifugation to separate the slowly sedimenting ER-derived vesicles released into the medium speed supernatant (MSS) from the rapidly sedimenting ER and Golgi membranes that are recovered in the medium speed pellet (MSP). Anti-Syn5 had no effect on the appearance of VSV-G in the MSS, in contrast to a control reaction containing Sar1[GDP], which blocked the formation of VSV-G–containing vesicles (14) (Fig. 2C). Thus, ER-derived vesicle formation does not require Syn5 function.

Because Syn5 recycled between the ER and Golgi (Fig. 1), we tested whether it was a component of ER-derived vesicles. We prepared affinity-purified vesicles using an antibody specific for the cytoplasmic tail of VSV-G (14) and examined their composition by immune blotting. The ER-derived vesicles generated under normal incubation conditions contained Syn5 and its interacting protein, Sly1, along with VSV-G and p58 (Fig.3A, lane b), and the vesicles contained the mammalian ER-to-Golgi v- SNAREs mSec22b, membrin, and rBet1 (20) (Fig. 3B, lane b). In controls, the levels of these proteins recovered on beads were reduced (>90%) when anti–VSV-G was omitted from the immunoisolation procedure (Fig. 3, A and B, lane a) or when COPII vesicle formation was blocked with Sar1[GDP] (Fig. 3, A, lane c, and B, lane d). Incubation in the presence of anti-Syn5 (Fig.3A, lane d) or Fab fragments (Fig. 3A, lane e) blocked the incorporation of both Syn5 and Sly1 into vesicles without affecting the recovery of VSV-G and p58. In contrast to the almost complete (>99%) removal of Syn5 from vesicles in the presence of anti-Syn5, the recovery of membrin was only partially reduced (∼50%), and rBet1 and mSec22b were unaffected (Fig. 3B). Thus, vesicles generated in the presence of antibody were not depleted of v-SNARES, providing us with a test for the role of vesicle-associated Syn5 in ER-to-Golgi transport.

Figure 3

Syn5 can assemble into a 20S SNARE complex on free ER-derived vesicles. (A and B) ER-derived vesicles were immunoisolated on magnetic beads with the use of an antibody against the cytoplasmic tail of VSV-G (14, 45). The vesicles were solubilized in gel sample buffer and analyzed by immune blotting for Syn5, Sly1, VSV-G, p58, and the v-SNAREs mSec22b, rBet1, and membrin (20) as indicated. The 35- and 42-kD isoforms of Syn5 were detected with roughly equal intensity. The 42-kD isoform is shown. The vesicle formation reactions were either not supplemented (A and B, lanes a and b), or supplemented with 2.5 μg of Sar1[GDP] (A, lane c; B, lane d), anti-Syn5, (A, lane d; B, lane c), or Fab (A, lane e). The amount of Syn5 recovered on the vesicles in lane b represented 4 to 5% of the total Syn5 present in the microsome fraction. This corresponds to the recovery of recycling p58 (∼5%) (14) and reflects both the efficiency of vesicle formation in vitro and the immunoisolation procedure. (C) The ability of Syn5 on affinity-purified ER-derived vesicles to interact with Myc-tagged NSF as measured by coimmunoprecipitation (46). Detergent extracts of the vesicles were incubated under 20Scomplex assembly (lanes a to d) or disassembly (lanes e and f) conditions in the presence of recombinant α-SNAP and magnetic bead–anti-Myc conjugates alone (lane a) or pre-bound to Myc-NSF (lanes b to f). The incubations were supplemented with the anti-Syn5 Fab (lanes c and f) or a control Fab (lane d). The recoveries of NSF, Syn5, and anti-Myc heavy chain (Ab HC) were measured by SDS-PAGE analysis of the washed immunoprecipitates followed by immune blotting. (D) Glycerol density gradient analysis of Syn5 present on affinity-purified vesicles. Detergent extracts of the vesicles were incubated under assembly or disassembly conditions in the presence of recombinant α-SNAP and Myc-NSF (46). The control was incubated under assembly conditions in the absence of NSF and SNAP. The reactions were run on linear glycerol density gradients then fractionated (lanes 1 to 14). The fractions were precipitated, then analyzed by SDS-PAGE followed by immune blotting for Syn5. The location of 4.6S (albumin) and 19S2-macroglobulin) are indicated by arrows.

According to the SNARE hypothesis, the interaction of syntaxins with their cognate v-SNAREs upon vesicle docking generates a 7Scomplex and leads to the binding ofN-ethyl-maleimide–sensitive factor (NSF) and SNAPs and the formation of a transient 20S SNARE complex (22). More recent studies, however, suggest that NSF function is complete before vesicle docking (23) and that SNARE complexes can assemble on purified synaptic vesicles (16, 24) and clathrin-coated vesicles (25). To test whether Syn5 could participate in SNARE complex assembly on undocked ER-derived vesicles, we prepared detergent extracts of affinity-purified COPII vesicles and incubated them in the absence or presence of Myc-tagged NSF (Myc-NSF) and α-SNAP. In control incubations lacking Myc-NSF, Syn5 was not immunoprecipitated with anti-Myc (Fig. 3C, lane a) and sedimented as a 5S to 7S complex (Fig. 3D, upper panel). Under conditions that promote the stable assembly of 20S complexes [by preventing adenosine triphosphate (ATP) hydrolysis by NSF], Syn5 (20 to 25% of total) coimmunoprecipitated with NSF (Fig. 3C, lane b) and sedimented as a ∼20Sparticle (Fig. 3D, middle panel). The interaction of Syn5 with NSF was blocked in the presence of the anti-Syn5 Fab fragment, but not a control Fab (Fig. 3C, lanes c and d). Immunoprecipitation with anti-Syn5 showed that Syn5 was associated with membrin in both the 7S and 20S complexes (15) as demonstrated previously (20). mSec22B and rBet1 are also likely to be components of these complexes because interactions between Syn5 and these v-SNAREs and their yeast homologs are well documented (20, 26-28). Under disassembly conditions (supporting ATP hydrolysis by NSF), Syn5 (Fig. 3C, lane e) was not recovered in the anti-Myc immunoprecipitate and association with membrin was lost (15). Accordingly, the 20Spool of Syn5 shifted to a low molecular weight form (Fig. 3D, lower panel), reflecting complete dissolution of the 20S complex as observed previously (20, 22). Thus, ER-derived vesicles can support formation of a 20S complex before vesicle docking.

To determine whether Syn5 is required for vesicle docking, we supplemented the microsome-based assay with anti-Syn5 and monitored the levels of VSV-G in the MSS with increasing time of incubation. Vesicles released from sedimentable ER membranes during the first 10 to 15 min of incubation are subsequently redirected to the MSP as a result of docking and fusion with rapidly sedimenting pre- or cis-Golgi membranes (14). Anti-Syn5 promoted the accumulation of VSV-G in the MSS, like Sar1[GTP], a mutant protein that prevents docking by blocking uncoating of ER-derived vesicles (Fig.4A) (14). Thus, VSV-G released from the ER in the presence of anti-Syn5 appeared to remain associated with slowly sedimenting vesicles that fail to dock with the acceptor compartment.

Figure 4

The vesicular pool of Syn5 is required for ER-to-Golgi transport. (A) Microsomes were incubated for the indicated times at 32°C (45) in the absence (□) or presence of 1 μg of either Sar1[GTP] (○) or the Syn5-specific antibody (•). The levels of ER-derived vesicles recovered in the MSS were quantitated by immune blotting for VSV-G. (B) Two-stage vesicle consumption assays (45). Control incubations (lanes a and b) in which microsomes were incubated for 10 min at 32°C, then either maintained on ice (lane a) or incubated for 60 min in stage 2 with acceptor Golgi membranes (45) (lane b). The amount of VSV-G processed to the endo D–sensitive form was determined. Lanes c to e: Stage 1 incubations were performed in the presence of anti-Syn5 alone (2.5 μg) (lane c) or anti-Syn1 (2.5 μg) (lane e), or anti-Syn5 pre-neutralized with Syn5-11 (lane d). Lanes f and g: Vesicles generated in the absence of antibody in stage 1 were recovered in the MSS, washed, and reincubated in stage 2 (60 min, 32°C) in the presence of acceptor Golgi membranes (47) and either anti-Syn5 alone (5 μg) (lane f) or pre-neutralized with Syn5-11 (lane g). (C) (•) Vesicles synthesized in the presence of anti-Syn5 in stage 1 were incubated in the absence of antibody in stage 2. (○) Golgi membranes (47) were pretreated with anti-Syn5, then neutralized with Syn5-11 before stage 2 incubations containing vesicles generated in stage 1 in the absence of antibody. The amounts of anti-Syn5 used for pretreatment of either the microsomes in stage 1 or the Golgi before stage 2 were adjusted to obtain equivalent molar ratios of antibody to total Syn5 protein (35- and 42-kD isoforms). The amounts of VSV-G processed are reported relative to untreated controls in which ∼30% of total VSV-G was transported. Mean values of duplicate samples are shown with the indicated error range.

To assess directly the requirement for the vesicular pool of Syn5 in targeting to the cis-Golgi compartment, we monitored VSV-G transport using a two-stage assay (14). In this assay, ER-derived vesicles released into the MSS during stage 1 were washed and reincubated in the presence of cytosol, ATP, and purified acceptor Golgi membranes (stage 2). When anti-Syn5 was added only to stage 1, the resulting Syn5-depleted vesicles were unable to deliver VSV-G to the cis-Golgi compartment in stage 2 efficiently (Fig. 4B, lane c). Inhibition was neutralized by preincubation of anti-Syn5 with Syn5-11 (Fig. 4B, lane d). No inhibition was observed when the vesicles were generated in the presence of a control antibody specific for Syn1 (Fig.4B, lane e). Inhibition was also observed when vesicles produced in stage 1 were isolated, pretreated with anti-Syn5 Fab, and reincubated in the absence of antibody in stage 2; parallel incubation with a control Fab recognizing the cytoplasmic tail of VSV-G had no effect (15). Thus, the vesicle-associated pool of Syn5 is required for delivery of cargo to the cis-Golgi.

To analyze the role of Syn5 in vesicle targeting or fusion further, we incubated vesicles formed in stage 1 reactions lacking antibody with acceptor Golgi in the presence of anti-Syn5 in stage 2. Inhibition was again observed (Fig. 4B, lane f), and preincubation with Syn5-11 neutralized the inhibition (Fig. 4B, lane g). The transport block could reflect inhibition of the Syn5 pool associated either with ER-derived vesicles or that which is present on the acceptor compartment or both. Therefore, we preincubated the Golgi membranes with anti-Syn5 at molar ratios (anti-Syn5:protein) that were inhibitory when added to stage 1 incubations. Before incubation in stage 2, the antibody was neutralized with Syn5-11 at concentrations sufficient to bind excess antibody, but not to interfere with transport. Fusion was only marginally (<10%) inhibited compared with the 70 to 75% block observed when Syn5-depleted vesicles were added to untreated Golgi membranes (Fig. 4C). In immunoprecipitation experiments, the Golgi membrane–associated Syn5 pool, like the ER pool, was fully accessible to antibody under standard transport reaction conditions (15). Addition of Syn5-11 to antibody-treated microsomes before stage 1 incubations at 32°C did not abolish inhibition of transport, indicating that Syn5-11 did not dissociate the antibody from membranes (15). Thus, we could not demonstrate a requirement for the cis-Golgi pool of Syn5 in transport.

Because vesicle-associated Syn5 was required for ER-to-Golgi transport, we investigated whether it participates in the assembly of pre-Golgi intermediates in vitro. Incubation in the presence of cytosol and ATP for 15 min at 32°C leads to transport of VSV-G from the ER (Fig.5A) into numerous punctate pre-Golgi intermediates (Fig. 5B) before delivery to the Golgi at later time points (19, 29, 30). Anti-Syn5 (Fig.5C) or Fab fragments (15) inhibited the appearance of VSV-G–containing intermediates. This was confirmed by electron microscopy in combination with quantitative morphometry (9). Whereas intermediates were readily detected in control incubations (676 detected in ∼40 cells), they were largely undetectable in permeabilized cells incubated in the presence of anti-Syn5 (<5% of control value in ∼400 cells). Because vesicle budding from the ER was unaffected by the antibody (Fig. 2), it was apparent that vesicles released from the ER in permeabilized cells failed to coalesce into detectable pre-Golgi intermediates in the absence of Syn5 function.

Figure 5

Syn5 is required for assembly of pre-Golgi intermediates. Permeabilized cells containing VSV-G in the ER (30) were prepared and either held on ice (A) or incubated in the presence of cytosol and ATP in the absence (B) or presence of anti-Syn5 (C). The distribution of VSV-G was determined by indirect immunofluorescence (43). Arrows indicate ER; arrowheads indicate pre-Golgi intermediates. Bar: 20 μm.

The requirement for a functional syntaxin family member on ER-derived vesicles is supported by evidence demonstrating that SNARE complexes form on other vesicular carriers (16, 25) and, in the case of synaptic vesicles, that Syn1 is the target for inactivation by the botulinin C1 toxin (24). It is now apparent that it may be difficult to predict the function of syntaxins from their steady-state morphological distributions. Several lines of evidence now demonstrate that the transient formation of a 20Sintermediate mediated by NSF and SNAPS may promote the disassembly of syntaxins from the SNARE complex, thereby priming membranes for subsequent docking and fusion (23,31-34). Our results are consistent with these observations. Thus, activation of vesicle-associated syntaxin may be a more general feature of vesicular traffic than previously anticipated.

The demonstration that Syn5 functions on ER-derived vesicles provides insight into the functional basis for operation of the early secretory pathway. Instead of ER-to-Golgi transport being mediated directly by COPII vesicles as suggested by studies in yeast (35), our data now provide biochemical evidence that Syn5 may be required for the fusion of COPII vesicles to generate tubular elements of pre-Golgi intermediates. This conclusion is consistent with the ability of intermediates to form de novo in the cell periphery to collect cargo molecules such as VSV-G (9, 36) and move en bloc in a microtubule-dependent manner to the Golgi complex (7, 36). The role for Syn5 in pre-Golgi intermediate formation may be analogous to the syntaxin requirement in yeast homotypic vacuole fusion (31). That the generation of tubular recycling intermediates requires Syn5 function and is critical for transport is consistent with the possibility that their assembly to form the cis-Golgi network (36) provides the foundation for the subsequent maturation of the Golgi (37).

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: webalch{at}


View Abstract

Navigate This Article