Syntaxin 5 regulates endoplasmic reticulum to Golgi transport.

Syntaxins are a family of vesicular transport receptors that are involved in membrane traffic through both the constitutive and regulated secretory pathways. Syntaxins 1A/B,2,3, and 4 are principally associated with the plasma membrane. Two of the syntaxins, 1A and 1B, have been suggested to be the docking receptors for synaptic vesicles with the presynaptic membrane. The most distant member of the family, syntaxin 5, has been found in the Golgi region and has significant homology (35% identity) with Sed5p, an essential protein in yeast which is required for vesicular transport from the endoplasmic reticulum (ER) to the Golgi stack. Here we present evidence that syntaxin 5 performs an analogous function in ER to Golgi transport in mammalian cells. Transient expression of an hemagglutinin-tagged full-length clone of syntaxin 5 and a truncated mutant lacking the transmembrane domain inhibited the transport of vesicular stomatitis virus glycoprotein to the Golgi stack. Under these conditions, vesicular stomatitis virus glycoprotein accumulated in pre-Golgi intermediates, which were strongly enriched in syntaxin 5. Our results suggest that syntaxin 5 is the functional mammalian homologue of Sed5p and provides evidence for its role in regulating the potential targeting and/or fusion of carrier vesicles following export from the ER.

Two types of vesicular trafficking mechanisms contribute to the secretory pathway of eukaryotic cells: those involved in constitutive secretion, delivering protein to the cell surface through continuous exocytosis; and those in the regulated pathway, which respond to extracellular signals in order to trigger the release of vesicular content. The biochemical components required for transport in both pathways are now thought to have a common molecular mechanism(s1 in which related proteins comprising gene families serve analogous functions in the events dictating vesicle budding, targeting, and fusion between different compartments. SynaptobrevinNAMP' and syntaxin *This work was supported in part by Grant GM 42336 from the National Institutes of Health (to W. E. B.), by Shared Instrumentation Grants RR07273 and RR08176, and by the Lucille P. Markey Charitable Trust. This is Scripps Research Institute Manuscript 8720-CB. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduerthis fact, The abbreviations used are: VAMP, vesicle-associated microtubule lA, originally identified as components of the machinery mediating neurotransmitter release at the synapse (1-31, are archetypes for two of these gene families (4). Members of the VAMP/synaptobrevin and syntaxin families are believed to form complexes which promote vesicle docking and fusion (1,5,6). The VAMP family consists of integral membrane proteins associated with carrier vesicles and includes synaptobrevins 1 and 2 (7), cellubrevin (8), and their putative yeast relatives Boslp (91, Betlp (lo), Sec22p (lo), and Snclp/2p (11). Members of the syntaxin family include syntaxins lA, lB, 2,3,4, and 5 in mammalian cells (12) and Pepl2p (13), Ssol/2p (141, and Sed5p (15) in yeast. Syntaxins contain a large cytoplasmic domain, a single hydrophobic membrane anchor, and a short extracellular or lumenal carboxyl-terminal tail. Syntaxin lAhas been found in the pre-synaptic membrane and, while thought to serve as a component of the docking machinery for synaptic vesicles, is also distributed along the entire surface of the neuron, suggestive of a more general role in protein traffic to the plasma membrane (1,4). In this study we report on the functional properties of mammalian syntaxin 5, demonstrating that it is found on pre-Golgi intermediates and that it is essential for vesicular trafflc between the ER and the Golgi stack in mammalian cells.
EXPERIMENTAL PROCEDURES Materials-All materials were obtained as described previously (16). Antibodies specific for the hemagglutinin (HA) epitope were obtained from I. Wilson (The Scripps Research Institute, La Jolla, CAI. A polyclonal serum recognizing a-l,2-mannosidase I1 (anti-Man 11) was kindly provided by M. Farquhar (University of California, San Diego) (17). Antibodies recognizing the pre-Golgi intermediate marker proteins p53 (18) and p58 (19) were generously provided by H.-P. Hauri (Biocenter, Basel, Switzerland) and J. Saraste (University of Bergen, Oslo, Norway), re-VSV-G (P5D4) (20) was kindly provided by T. Kreis. We thank R. Scheller spectively. A monoclonal antibody specific for the carboxyl terminus of (Stanford University, Stanford, CA) for generously providing us with cDNAclones for syntaxins 1Aand 5 (12) and C . Der (University of North Carolina, Charlottesville, NC) for the pETlld-HA expression vector.
Generation of Expression Constructs-The pETlld HA-tagged syntaxin 1A and 5 expression constructs were generated as follows. The syntaxin 5 and syntaxin 1A wild-type clones were modified by PCR to possess a 5'-NdeI restriction site encoding the initiating methionine and a 3'-BamHI restriction site immediately after the stop codon. The truncated forms of syntaxin 5, 5-11 (amino acids 4-267) and 5 1 6 (amino acids 194-267) were generated by PCR using the original pBluescript construct as a template. For syntaxin 5-11, a 5'-NdeI restriction site was created with the conversion of cysteine at position 3 to an initiating methionine.At the 3'-end, a stop codon and aBamHI restriction site were encoded immediately following amino acid 267. For syntaxin 5-16, a 5'-NdeI restriction site was generated by PCR which encoded an initiating methionine at position 194; at the 3'-end a stop codon and BamHI restriction site were encoded immediately following amino acid 267. PCR reactions were carried out using Pfu polymerase (Stratagene) under standard conditions. PCR reaction products were subcloned into the TA cloning vector pCRII (Invitrogen) and sequences verified by the chain termination method. The syntaxin 1A and 5 full-length, syntaxin 5-11, and syntaxin 5-16 sequences were then introduced asNdeZIBamHI fragments into a version of the pETlld vector containing an in-frame HA epitope tag immediately upstream of the NdeI restriction site for expression from the T7 promoter as described ( 16,21).

RESULTS
The current paradigm for syntaxin function suggests a role in vesicle docking and/or fusion to a specific target membrane (12). While syntaxin 5 shows limited homology with other members of the syntaxin family (12), it has significant homology (35% identity) with Sed5p, a yeast protein that is essential for ER to Golgi transport (15). To determine if syntaxin 5 was functionally homologous to Sed5p, we examined the effects of transient overexpression of a HA-tagged form of syntaxin 5 (HA-syntaxin 5) on the transport of vesicular stomatitis virus glycoprotein (VSV-G) from the ER to the Golgi apparatus in mammalian cells. VSV-G is a type I membrane protein containing two N-linked carbohydrate chains and has served as a marker protein to study the biochemical and molecular basis for transport from the ER in vivo (16,(21)(22)(23) and in vitro (22)(23)(24)(25)(26)(27)(28). Vectorial transport of VSV-G from the ER to and through sequential cis-, medial, and trans-Golgi compartments can be measured by the processing of its two oligosaccharide chains from the high mannose (Man,) endoglycosidase H (endo Hbsensitive form found in the ER and pre-Golgi intermediates to endo H-resistant forms found in the Golgi stack. These processing intermediates can be readily distinguished by their unique electrophoretic mobilities using SDS-polyacrylamide gel electrophoresis (16,23,28). The appearance of the first, transient endo H-resistant R, forms corresponds to the transport of VSV-G to the early cidmedial Golgi compartments where one or both of the oligosaccharide chains becomes processed by the action of resident a-1.2-mannosidases and glycosyltransferases (16,23,26,28,29). Subsequent transport of VSV-G to the trans-Golgi and the trans-Golgi network (TGN) leads to the appearance of the fully processed R, form containing two complex, endo H-resistant oligosaccharides (16,23,26,28,29). The appearance of each of these processing intermediates allows us to characterize the potential differential roles of components required for ER to Golgi and/or intra-Golgi transport (16,23).
To transiently express syntaxin 5 and our marker protein, VSV-G, we utilize a recombinant T7 vaccinia virus system (30). We have used this expression system extensively to demonstrate the inhibitory effects of trans dominant negative mutants of the small GTPases Rabl, Rab2, A r f l , and Sarl on the transport of VSV-G through the early secretory pathway (16,(21)(22)(23)27,28). BHK-21 cells infected with the recombinant vaccinia virus (vTF7-3) were cotransfected with expression vectors carrying the HA-tagged-syntaxin 5 and/or VSV-G genes (16,21). After incubation for 3-6 h to allow time for protein expression, transfected cells were pulse-labeled with ["S]methionine for 10 min, followed by a chase in the presence of unlabeled methionine for 75 min. The expression of HA-syntaxin 5 was monitored by immunoblot analysis using a HA-specific monoclonal antibody. Expression could be detected as early as 3 h after transfection (Fig. 1, inset) with protein continuing to accumulate over the 6-h time course to a level 22-fold that observed at 3 h ( Fig. 1, inset ). Inhibition of VSV-G transport (-40%), as indicated by the accumulation of VSV-G in the R, form and the concomitant reduction in the appearance of the R, form, could be detected after only 3 h of transient expression (Fig. 1 , 3 h ) . Although the concentration of the endogenous, wild-type pool of syntaxin 5 is unknown (due at present to the lack of specific antibody), we have found that the expression levels ofthe small GTPases Sarl, A r f l , and Rabl after only 3 h of transfection generally correspond to only 1-2-fold molar excess over their respective endogenous pools (16,21,23). These results suggest that ER to Golgi transport is very sensitive to even low levels of elevated expression of the full-length HA-syntaxin 5. High levels of expression after 6 h (22-fold t h a t observed at 3 h ) led to nearly complete (-80%') inhibition of transport (Fig. 1, 6 Although transport from the ER to the Golgi stack was strongly inhibited by elevated expression of full-lenflh HAsyntaxin 5, further transport of VSV-G through the G o l p stack was unaffected. For example, cells in which transport was only partially inhibited by low levels of expression of HA-syntaxin 5 (i.e. Fig. 1.3 (16). In contrast, the inability of elevated expression of HA-syntaxin 5 to inhibit intra-Golgi transport is identical to the effects of a trans dominant negative mutant of Sarl (Sarl(Q79L)), a small GTPase that is uniquely required for ER to Golgi but not intra-Golgi transport in mammalian cells (2.71. From these results, it is apparent that elevated expression of HA-syntaxin 5 specifically inhibits the function of c a m e r vesicles mediating ER to Golgi transport. To define the potential role of the transmembrane domain in HA-syntaxin 5 function, we examined the effects of elevated expression of two soluble fragments of the protein. One fragment included the entire cytoplasmic domain, lacking only the transmembrane anchor (HA-syntaxin 5-11, amino acids 4-2671 (Fig.  2.4 ). The second fragment consisted of the region ndjacent to the transmembrane domain that displays significant sequence conservation among known members of the syntaxin family (HAsyntaxin 5-16, amino acids 194-267) (Fig. M ) . Thc HA-syntaxin 5-11 construct was stably expressed only at a reduced level when compared to the wild-type protein. a level comparable to that observed with full-lenflh syntaxin 6 nt 3 h post- Syn-5 Syn-5 Syn-5 Syn-la (071 L) 11

FIG. 2. A truncated form of HA-syntaxin 5 partially inhibits ER to Golgi and intra-Golgi transport.
A, truncated forms of HA-syntnxin 5 (5-11 and 5-16) were prepared as described under "Experimental Procedures." t?, ARFl(Q71L) and the HA-syntaxin constructs were cotransfected with VSV-G (h-f) in BHK-21 cells using the vaccinia transient expression system, incubated for 5 h, labeled for 10 rnin with ["Slmethionine, and chased for 75 rnin in the presence of unlabeled methionine. In the control (CfI, n ) cells were transfected with VSV-G alone. The processing of VSV-G to the R,, R,, and R, forms were quantitated as described (16). The ARFllQ71L) mutant was prepared as described previously (16). transfection (data not shown). However, even at low levels of expression, significant inhibition (35%) of transport could detected (Fig. 2B, d ) , similar to that observed after 3 h of expression ofthe full-length molecule. This result is consistent with the ability of a similar truncation of syntaxin lAto prevent synaptic vesicle fusion when microinjected into PC12 cells (12). Interestingly, elevated expression of the truncated form also led to a detectable accumulation of the R, intermediate (Fig. 2B, compare d to c), suggesting a weak effect on intra-Golgi transport. The absence of the transmembrane domain may lead to mislocalization and the partial interaction of syntaxin 5 with the biochemical machinery involved in transport through the Golgi stack. The small conserved fragment (HA-syntaxin 5-16) was found to be unstable during transient expression (data not shown) and thus had no effect on transport (Fig. 2B, e ) .
As a control for the effects of overexpression of full-length syntaxin 5 on transport, we examined the ability of syntaxin 1A to inhibit ER to Golgi transport. Syntaxin 1A function should be restricted to vesicle targeting to the cell surface (1, 12). As shown in Fig. 2, only at a high level of overexpression of HAsyntaxin 1A (a level comparable to that of syntaxin 5 at 6 h post-infection (Fig. 1, inset)) did we detect a partial (30%) inhibition of the transport and processing ofVSV-G to the form (Fig. 2B, f ) . These results raise the possibility that the overexpression of a syntaxin protein may upset the balance of more general transport factors such as NSF, SNAPS, and SNAP receptors, components that are likely to be involved in function of a11 members of the syntaxin family (4)(5)(6).
The inability of VSV-G to be processed to Golgi forms in the presence of elevated levels of the full-length HA-syntaxin 5 could be explained either by inhibition of export from the ER or by its accumulation in pre-Golgi intermediates. To differentiate between these two possibilities, we examined the transport of VSV-G using indirect immunofluorescence. For these experiments, several different cell lines were transfected with an expression vector carrying a temperature-sensitive form of VSV-G (strain ts045J. Strain ts045 VSV-G is retainrd in t h r cisternal network of the ER when crlls arr transfrcted at the restrictive temperature (39.5 "C) (25. 26). Shift of the cells to the permissive temperature (32 "C) 4 h after transfrction at thr restrictive temperature results in the synchronized rrlease of ts045 VSV-G and eficient transport to the compact, juxtanuclear Golgi stack (25,26). When BHK cells were cotransfected with ts045 VSV-G and the full-length HA-syntaxin 5 for 5 h at the restrictive temperature, VSV-G was retained in the ER (data not shown). This result indicates that elevated levels of HA-syntaxin 5 do not interfere with the normal temperature-sensitivr phrnotypc. of ts045 VSV-G. HA-syntaxin 5. on the other hand, showed n typical steady state distribution to small punctate elements. which were distributed throughout the prriphrral cytoplasm (Fig. 3C, arrowhendsJ, and to the rek6on encompassing the juxtanuclear Golgi stack (Fig. 3C. nrrorcq). Similar results wrrr observed in HeLa cells (data not shown, and wrre consistrnt with the reported distribution of syntaxin 5 in COS cells at 37 "C, where the protein was found to colocalize with P-COP (12). a marker for pre-Golgi intermediates and thr cis rlrmpnts of the Golgi stack (27.33-35). In contrast to thr distribution of HA-syntaxin 5, the distribution of HA-syntaxin 1A was Iaraely restricted to the cell surface as descrihrd prrviously ( 12) (data not shown).
To identify the morphological step in transport inhihitrd hy overexpression of syntaxin 5 , RHK crlls cotransfected with ts045 VSV-G and syntaxin 5 for 4 h a t 39.5 "C wrrr shiftrd to the permissive temperature and incuhatrd for 90 min. In the presence of HA-syntaxin 5, ts045-VSV-G did not accurnulatr in the compact Golgi stack as observed in control cells. In contrast, in nearly 80-90% of the cells examined, tsO45-VSV-G protein accumulated in small punctate structures scattered throughout the peripheral cytoplasm (Fig. 3 A ) and within and around the compact, juxtanuclear Golgi stack (Fig. 3 A , arrow). All of the compartments containing VSV-G strongly overlapped with the distribution of p58 (Fig. 3, compare A to B, arrowheads), a protein preferentially enriched at steady state in pre-Golgi intermediates composed of clusters of vesicles and small tubular elements (25,27,36,37) and in the cis elements of the Golgi stack. No effect of overexpression of HA-syntaxin 5 was observed on the distribution of the compact, perinuclear localization of the Golgi stack as determined by the distribution of the cidmedial marker enzyme a-l,2-mannosidase I1 (17)) (data not shown), Strikingly, HA-syntaxin 5 was also found to precisely overlap with the punctate distribution of p58 (Fig. 3, compare C to D, arrowheads) at both low and high levels of expression. As expected, given the striking colocalization of VSV-G to p58 containing elements in the presence of elevated levels of HA-syntaxin 5 , pre-Golgi intermediates accumulating VSV-G completely overlapped with the distribution of HA-syntaxin 5 (data not shown). Partial overlap of p58 and HA-syntaxin 5 was also observed with the distribution of Man 11, consistent with the partial localization of p58 to cis elements of the Golgi stack (data not shown). Identical results were observed in HeLa cells during transient expression of VSV-G and HA-syntaxin 5 (data not shown). These results reinforce the interpretation from the biochemical analyses that syntaxin 5 plays a specific role in ER to Golgi transport and suggest a site of action in the docking andor fusion of pre-Golgi intermediates. DISCUSSION We have demonstrated that syntaxin 5 is a functional component of the transport machinery involved in the delivery of pre-Golgi intermediates to the Golgi stack in mammalian cells. Given the morphological phenotype associated with overexpression of the protein, it is apparent that the full-length syntaxin 5 has a potent trans dominant negative effect on the function of ER to Golgi carrier vesicles. In contrast, full-length HA-syntaxin 1A had only a weak effect at high levels of expression, indicating the functional specificity of syntaxin 5 . The combined results suggest that syntaxin 5 is likely to be the mammalian homologue of yeast Sed5p ( E ) , a protein originally isolated as a suppressor for loss of Erd2p (the yeast homologue to the mammalian KDEL receptor) (38). Indeed, overexpression of Sed5p also leads to the accumulation of pre-Golgi intermediates in yeast (15). While the biochemical mechanism underlying the trans dominant effects of overexpression of syntaxin 5 is presently unknown, one possibility, among others, is that an excess of the protein restricts the availability of a component(s) involved in a late step in vesicle targeting andor fusion.
Consistent with a role for syntaxin 5 in ER to Golgi transport, we found the protein was distributed to pre-Golgi transport intermediates composed of 60-80-nm carrier vesicles and small tubular elements present in the peripheral cytoplasm, and to the juxtanuclear Golgi region based on the distribution of Man 11. The localization in this region with p53 and p58, which partially overlap with Man 11, is suggestive of distribution to the cis elements of the Golgi stack.
The striking distribution of HA-syntaxin 5 to pre-Golgi vesicular-tubular intermediates is difficult to reconcile with the concept that members of the syntaxin family function exclusively as docking receptors, such as has been suggested for the role of syntaxin 1A at the pre-synaptic membrane (1). This distribution was observed at both low and high levels of expression and is therefore unlikely to be an artifact of the transient expression system being used. In the proposed model for syn-of ER to Golgi Dunsport taxin function (1, 51, we would not expect a protein involved in the docking andor fusion to the target compartment to be a component of the intermediate carrier vesicles en route. One possibility to explain this conundrum is that syntaxin 5 present on ER to Golgi intermediates may be an "inactive" recycling form. If so, then its activation, after encountering a component on the target compartment, would be the critical event dictating vesicle docking and fusion. A second possibility is that the small tubular elements associated with pre-Golgi intermediates may comprise a stable "compartment" to which ER-derived vesicles fuse, although there is no evidence to date to support this interpretation (25). A third possibility is that syntaxin 5 is also involved in the formation or maturation of pre-Golgi intermediates. consistent with this possibility, the early effect of depletion of Sed5p in yeast leads t o the extensive elaboration of the ER, rather than accumulation of transport vesicles (15).