key: cord-0252714-gty6b7vz
authors: nan
title: Rab1 and Ca2+ are required for the fusion of carrier vesicles mediating endoplasmic reticulum to Golgi transport
date: 1994-04-02
journal: J Cell Biol
DOI: nan
sha: 153c3a86e8939de3ca0331ea95b68f8bb4a3e395
doc_id: 252714
cord_uid: gty6b7vz

Members of the rab/YPT1/SEC4 gene family of small molecular weight GTPases play key roles in the regulation of vesicular traffic between compartments of the exocytic pathway. Using immunoelectron microscopy, we demonstrate that a dominant negative rab1a mutant, rab1a(N124I), defective for guanine nucleotide binding in vitro, leads to the accumulation of vesicular stomatitis virus glycoprotein (VSV-G) in numerous pre-cis-Golgi vesicles and vesicular-tubular clusters containing rab1 and beta-COP, a subunit of the coatomer complex. Similar to previous observations (Balch et al. 1994. Cell. 76:841-852), VSV-G was concentrated nearly 5-10-fold in vesicular carriers that accumulate in the presence of the rab1a(N124I) mutant. VSV-G containing vesicles and vesicular-tubular clusters were also found to accumulate in the presence of a rab1a effector domain peptide mimetic that inhibits endoplasmic reticulum to Golgi transport, as well as in the absence of Ca2+. These results suggest that the combined action of a Ca(2+)-dependent protein and conformational changes associated with the GTPase cycle of rab1 are essential for a late targeting/fusion step controlling the delivery of vesicles to Golgi compartments.

of small molecular weight GTPases play key roles in the regulation of vesicular traffic between compartments of the exocyfic pathway. Using immunoelectron microscopy, we demonstrate that a dominant negative rabla mutant, rabla(N124I), defective for guanine nucleotide binding in vitro, leads to the accumulation of vesicular stomatitis virus glycoprotein (VSV-G) in numerous pre-cis-Golgi vesicles and vesicular-tubular clusters containing rabl and B-COP, a subunit of the coatomer complex. Similar to previous observations (Balch et al. 1994. Cell. 76:841-852) , VSV-G was concentrated nearly 5-1G-fold in vesicular carriers that aocumulate in the presence of the rabla(N124I) mutant. VSV-G containing vesicles and vesiculartubular clusters were also found to accumulate in the presence of a rabla effector domain peptide mimetic that inhibits endoplasmic reticulum to Golgi transport, as well as in the absence of Ca 2+. These results suggest that the combined action of a Ca2+-dependent prorein and conformational changes associated with the GTPase cycle of rabl are essential for a late targeting/fusion step controlling the delivery of vesicles to Golgi compartments. T HE biochemical mechanisms underlying vesicular transport between compartments of the exocytic and endocytic pathways are poorly understood, but are likely to involve the ordered assembly and disassembly of protein complexes that regulate vesicle budding, targeting, and fusion (reviewed in Pryer et al., 1992; Rothman and Orci, 1992) . It is now evident that a collection of GTPases belonging to the rab/YPT/SEC4, A~ Sad, and beterotrimeric G~, families serve as molecular switches to control diverse functions of the transport machinery (reviewed in Goud and McCatfrey, 1991; Barr et al., 1992; Zerial and Stenmark, 1993; . In particular, transport from the ER to and between Golgi compartments in mammalian cells requires the small GTPase rabl. Neutralizing antibodies (Plutuer et al., 1991; Schwaninger et al., 1992; Davidson and Balch, 1993; Peter et al., 1993) and rabl-specific peptides Balch et al., 1993) block transport in vitro. In addition, trans-dominant mutants with altered guanine nucleotide-binding properties inhibit transport in vivo (Tisdale et al., 1992) and in vitro Davidson and Balch, 1993) .

In a previous study , we established the biochemical and morphological properties of the rabla-($25N) mutant, demonstrating that this mutant is likely to be restricted to the GDP-bound state, a condition that inhibits vesicle budding and the export of vesicular stomatitis virus glycoprotein (VSV-G)~ from the ER. Given previous evidence that the rabla(N124I) mutant is also a dominant inhibitor of transport in vivo (Tisdale et al., 1992) , we have analyzed the effects of this mutant on VSV-G transport in vitro using immunoelectron microscopy. We find that the N124I mutant is defective for guanine nucleotide binding in vitro, indicating that it is able to undergo rapid exchange independent of a guanine nucleotide exchange protein (GEP). This mutant does not inhibit vesicle budding from the ER. Instead, it prevents the fusion of VSV-G containing carrier vesicles to the Golgi stack. In the presence of the N124I mutant or in the absence of Ca 2+ (by chelation with EGTA), VSV-G accumulates in 60-nm carrier vesicles, and clusters of vesicles and small tubules-pre-Golgi intermediates that populate the normal pathway in vitro . The combined results from this and our previous studies on the rabl(S25N) mutant suggest that while a conformational change as-sociated with the exchange of GDP for GTP may be necessary for the recruitment of rabl during vesicle budding, GTP hydrolysis is likely to be critical for a late Ca2+-dependent vesicle targeting/fusion step.

A polyclonai reagent recognizing 1)58 was provided by J. Saraste (Saraste and Sveusson, 199D . A monoclonai reagent recognizing p53 was generously provided by H.-P. Hauri (Biocenter, Basel, Switzerland) . Secondary antibodies were obtained form the foilowing sources: FITC-conjugated goat anti-rabbit F(ab)2 and anti-mouse IgG from Zymed Laboratories, Inc. (South San Francisco, CA); gold-conjugated goat anti-rabbit from Amersham Corp. (Arlington Heights, IL). Rabbit antisera to E-COP were derived from an immunogenic peptide containing the unique EAEG sequence (Duden et ai., 1991) as described (W'flson, B. S., et al., 1994) .

Hiss-tagged rabla(N124D protein was prepared as described (Nuoffer et ai., 1994) .

All materials, assays, and procedures used to analyze transport or the morphological distribution of VSV-G using indirect immunofluorescence were as described previously using permeabilized NRK cells . FOr electron microscopy, incubations were conducted in 35-ram tissue culture dishes (Costar Corp., Cambridge, MA) (Baich et ai., 1994) . The analysis of transport using perforated cells based on the processing of VSV-G to endoglycosidase D- (Beckers et al., 1987) or endoglycosidase H-resistant oligosaccharide forms (Davidson and Balch, 1993; Nuoffer et ai., 1994) were performed as described.

After transport in vitro, permeabilized NRK cells were fixed for 30 rain with 3 % paraformaidehyde and 0.1% glutaraldehyde, pH 7.4, in 0.1 M phosphate buffer, washed briefly in PBS, and processed for immunodiffusion or indirect immunogold labeling of ultrathin cryosections.

For immanodiffusion, cells were washed for 10 min in PBS containing 1% BSA, 0.01 M glycine, and 0.1% saponin, pH 7.4, and then incubated sequentially with a primary monoclonai antibody, P5D4, specific for the cytoplasmic tail of VSV-G (overnight), a rabbit anti-mouse bridging anti-" body (1 h), and 5 or 10 nm gold, goat anti-rabbit IgG conjugate (3 h). After washing, the ceils were fixed (1 h) in 3% paraformaldehyde, 1.5% glutaraldehyde, and 7.5% sucrose in 0.1 M cacodylate buffer, pH 7.4. The cells were then pelleted, postfixed in 1% OsO4 in veronal-acetate buffer, stained in block for 2 h in 0.5% uranyl acetate, pH 6.0, dehydrated, and embedded in el)on. Where indicated, cells were counterstained with tannic acid as described (Orci et ai., 1986) . For immunogold labeling of ultrathin cryosections, ceils were harvested, pelleted, and cryoproteeted by infiltration with a mixture of 2.3 M sucrose in 0.1 M phosphate buffer, pH 7.4, containing 20% polyvinyl pyrrolidone, mounted on aluminum nails, and frozen in liquid nitrogen. Ultrathin cryosections were cut on a Reicbert Ultracult E, equipped with a Reichert FC-4 cryoattachment, collected on formvar/carbon-coated nickel grids, and quenched in 0.01 M glycine, pH 7.4. Sections were then incubated 2 h at room temperature with the anti-VSV-G monoclonai antibody in 10% FCS/PBS, a rabbit anti-mouse bridge (1 h), followed by a 5-nm gold, goat anti-rabbit IgG conjugate (2 h). The grids were then stained in 2 % neutral uranyl acetate (10 rain) and subsequently adsorption-stained with 0.2 % uranyl acetate, 0.2% methyl cellulose, and 3.2% polyvinyl alcohol. Sections were observed on an electron microscope (1200 EX; JEOL U.S. A. Inc., Peabody, MA, Philips Electronic Instruments Co., Mahwall, NJ) .

Electron micrographs were enlarged to a magnification of 50,000-100,000fold. The linear membrane density was determined by tracing of the membrane contour length of each compartment on photographic prints with aid of a measuring pen calibrated to give the total millimeter length for a particular magnification. For both vesicular-tubular clusters (VTCs) and Golgi stacks, only the external membrane length reflecting the distribution of gold particles was traced. The numbers reported are the number of gold particles per millimeter of membrane length of the compartment traced with the pen. In the case of 40-80-nm vesicles, the mean linear density is reported as that calculated for a 60-nm vesicle. The data presented for the distribution of VSV-G in the different intermediates and the Golgi for each time point in Fig. 2 are the combined averages of between two and three separate experiments. The error bar in each of these experiments reflects a number of variables that include the combined averages from different experiments, variation in the degree of infection of individual cells, and differences in rates of transport in vitro of VSV-G in different cells within each experiment. Statistical significance of data was determined by a Student's t test.

We have recently established that VSV-G is sorted and concentrated into 60-rim carrier vesicles upon export from the ER . To examine the role of GTP hydrolysis in protein concentration and vesicle budding from the ER, we first examined the effect of the nonhydrolyzable GTP analogue, GTP3,S, on the export of VSV-G using quantitative immunoelectron microscopy and an assay that efficiently reconstitutes this stage of the secretory pathway in vitro using permeab'dized NRK cells (Plumer et al., 1992) . For this purpose and to synchronize the export of VSV-G from the ER, we used a temperature-sensitive mutant of VSV-G (tsO45) which has a thermoreversible folding defect and is retained in the ER when cells are held at the restrictive temperature (Lafay, 1974) . After shifting the ceils to the permissive temperature (32°C), tsO45 VSV-G is efficiently transported to the Golgi stack in rive (Tisdale et al., 1992) When permcabilized NRK cells containing tsO45 VSV-G in the ER were incubated for 45 min at the permissive temperature in the presence of GTP-rS, and the distribution of VSV-G was analyzed by immunoelectron microscopy, VSV-G was not detected in Golgi elements (Fig. I B) . Instead, VSV-G was found in both carrier vesicles and clusters of vesicles and small tubular elements (VTLS) (Fig. I, A- (Phitner et al., 1990) , or rabla(N124I) (1 /~M), and the mean linear density of VSV-G (number of gold particles per micrometer of membrane contour length) in the ER, ER to Golgi carrier vesicles, VTCs, and the Golgi stack was determined as described in the Materials and Methods. ER to Golgi carrier vesicles were identified as 40-80-nm structures that contained >1 gold particles. The error bars define the SEM reflecting the distribution range of VSV-G in individual compartments used to generate the mean value as indicated in the Results. The total number of gold particles counted for each experimental condition are indicated at the top of the figure. The data presented for the distribution of VSV-G in the different intermediates and the Golgi for each time point are the combined averages of between two and five separate experiments. Within an individual experiment or within individual cells examined at high resolution, the fold change in density of VSV-G over that found in the ER ranged from 5.2 to 22 (nine experiments) in VTCs and 4-to 24-fold (four experiments) in the Golgi stack, nd, none detected. sence of inhibitor , but showed a high content of vesicular profiles, suggesting that they were largely composed of clusters of 60-rim vesicles (Fig. 1 , A and C). Vesicles and VTCs containing VSV-G formed in the absence or presence of GTP3,S also contained smooth dense coats (Fig. I C) . VTCs are the high resolution counterpart to the numerous punctate, pre-Golgi structures previously detected using indirect immunofluorescence in vitro in the presence of GTP~S .

To ascertain whether VSV-G was concentrated in vesicles and VTCs in the presence of GTP3,S to a similar degree as that observed under normal incubation conditions , the mean linear membrane density of VSV-G was determined using immunoelectron microscopy. As shown in Fig. 2 A, the mean linear density of VSV-G in isolated carrier vesicles formed in the presence of GTP'vS was 25.8 + 15.8 gold particles//~m (assuming a 60-nm vesicle). These vesicles bad an average of 5.0 + 3.1 gold particles/vesicle (298 vesicles counted) ( Fig. 3 B, solid bars) . These numbers are comparable to those obtained for vesicles formed after only 5 min of incubation in the absence of inhibitor ( Fig. 3 B, open bars vs solid bars) . The mean linear density of VSV-G in these VTCs was 37.1 + 20.6 gold particles//~m (with an average of 30.8 + 11.8 gold particles/VTC, 44 VTCs counted) (Fig. 2 A) . Even after a 45-min incubation at 32°C in the presence of GTP3,S, the concentration of VSV-G in the ER (6.1 + 3.4 gold particles/inn) ( Fig.  2 A) was not significantly different (P < 0.001) from that ob-served before incubation (5.2 ± 2.1 gold particles/t~m) . These results suggest that while vesicle budding occurs in the presence of GTP3~S, transport components present in the assay may be unable to efficiently recycle in the absence of GTP hydrolysis, thereby reducing the overall extent of export from the ER compared to normal incubation conditions in the absence of the inhibitor.

To assure ourselves that the increased linear density of VSV-G in vesicles or VTCs formed in the presence of GTPTS was not a consequence of an unusual change in its linear distribution throughout the ER membrane before vesicle budding, we assessed the distribution of gold particles as individual particles or groups Of particles on the ER membrane in the presence or absence of the inhibitor (Fig. 3 A) . Even after a 45-rain incubation in the presence of GTP~/S, gold particles remaining in the ER membrane were found to be distributed predominantly as individual particles. 819 out of 1,048 particles or groups of particles detected were scored as individual particles (Fig. 3 A) . In this experiment, the mean distribution of gold particles retained in the ER in the presence of GTP-/S was found to be 1.5 + 1.1 gold particles/ group ( Fig. 3 A, solid bars) . This result is not significantly different (P < 0.01) from our previous results, which documented the distribution of gold particles in the ER before incubation in vitro in the absence of GTP3,S ( Fig. 3 A, open bars) (1.5 + 1.2 gold particles/group) . Thus, incubation in the presence of GTPTS does not lead to a general aggregation of VSV-G along the surface of the ER membrane. Concentration is closely coupled to vesicle budcling. Moreover, the 5-to 10-fold increase in density of VSV-G in vesicles and VTCs in the presence of inhibitor is comparable to the increased concentration of VSV-G in identical structures formed in the absence of inhibitor .

To analyze specifically the role of tab1 in vesicle targeting and fusion, we explored the properties of the rabla mutant, in which the asparagine in position 124 Of the third guanine nucleotide binding motif (NKXD) is substituted by isoleucine. The N124I mutant has a number of properties that distinguish it from those of the rabla(S25N) mutant, which has a preferential affinity for GDP . First, the rabla(N124I) mutant fails to bind both GDP or GTP at detectable levels in vitro (Table I) . The equivalent mutation in H-ras(N116I) is similarly defective for GDP/GTP binding and is transforming, suggesting that it causes constitutive activation a consequence of a GEP-independent, high exchange rate (Walter et al., 1986) . Second, the N124I mutant is a potent inhibitor of ER to Golgi transport when transiently expressed in vivo (Fig. 4, lane c) , even in the absence of the carboxyl-terminal CC motif (Fig. 4, lane d) , suggesting that isoprenylation is not essential for function. However, further truncation of the carboxyl terminus leads to loss of inhibitory activity (Fig. 4 , lane f), reflecting either the requirement for the hypervariable domain for membrane localization or, alternatively, misfolding and instability in vivo. Third, the dominant negative effects of the rabla(N124I) mutant were not sensitive to coexpression with wild type rabla in vivo (Fig. 4 , lane e), suggesting that it functions in an irre- Guanine nuclcofide binding to purified wild type and mutant rabla. Hisetagged wild type and mutant rabla were purified as described , and 31-pmol (0.7-t~10 allquots were incubated for 45 min at 30°C in an assay buffer comprising 50 mM Hepes-potassinm hydroxide, pH 8, 5 mM EDTA, 2 mM MgCI2, 1 mM DTT, 0.1 #g/ml BSA, and either 1.4 #M [3H]GDP (specific activity = 3 Ci/mmol) or 1.1/tM a~P]GTP (specific activity ffi 6 Ci/mmol) (final vol ffi 100 td). Nucleotide exclmnge was termimm.d by the addition of e~cess MgCIz, and bound nuclcotide measured by fiquid scintillation counting after capture of the protein on 0.45/an nitrocellulose membranes (BA85; Schleicher & Schuell, Inc., Keene, Nil).

versible fashion. Finally, the intact rabla(N124I) mutant failed to incorporate pH]geranylgeranylpyrophosphate when incubated in the presence of cytosol prepared from rat liver homogenates (data not shown). Given the previous observation that additional secondary or tertiary structure is essential for the geranylgeranylation of CC-and CXC-containing proteins , it is apparent that the conformation of the N124I mutant is different from that of either wild type rabl or the rabl(S25N) mutant. HeLa cells were infected with the T7 RNA polymerase-recombinant vaccinia virus vTF7-3 and cotran.~ected with plasmids encoding VSV-G and the indicated rabla constructs as described previously (Nuotfer et al., 1994) . After 5-h transfecfion, coils were pulse-labeled with trans-~SS label for 10 min, not chased 0ane a) or chased for 60 min (lanes b-f). VSV-G was immtmoprecipitated, digested with endo H, and the endo H-sensitive, the endo H-resistant intermediate, and the terminally glycosylated forms were separated by SDS-PAGE. Autoradiographs were quantitated by scanning densitometry as described previously (Nuotfer et al., 1994) . Transport is expressed as the percentage of total VSV-G converted to endo H-resistant forms. Lanes a and b show cells transfected with VSV-G only; in lanes c-f, cells were cotransfected with VSV-G and the indicated plasmid(s). In lane e, cells were transfected with both rabla(N124I) and rabla-wt. In all cases, expression levels of wild type and mutant proteins (about two-to eightfold) were confirmed using Western blotting (data not shown).

To assess directly the consequences of the rabla(N124I) mutant on ER to Golgi transport, the effects of the mutant on the processing of VSV-G oligosaccharides to the cis-Golgi, endoglycosidase D-sensitive form was analyzed in vitro. In this case, perforated 15B CHO cells were incubated in the presence of cytosol at the permissive temperature in the presence of recombinant rabla(N124I) prepared from Escherichia coli . As shown in Fig. 5 A, rabla(N124I) strongly inhibited transport in vitro in a dosedependent fashion with half-maximal inhibition occurring at ,~400 ng (final concentration = 0.5 t~M). Inhibition was abolished if the protein was denatured by boiling before addition to the assay. Preincubation of the mutant in the presence of cytosol reduced the amount of mutant protein required for half-maximal inhibition by ,~50%, suggesting that cytosolic factor(s) may participate in establishing the inhibitory phenotype of the rabla(N124I) mutant (data not shown). Analysis of the temporal sensitivity of the transport reaction to the rabla(N124I) mutant revealed that transport could be inhibited only at early, but not late times of incubation (Fig.  5 B) . Cells incubated in the absence of the inhibitor rapidly became insensitive to the subsequent addition of the mutant protein. For example, >50% of maximal transport activity became resistant to the addition of the rabla(N124I) mutant during the first 15 min of incubation (Fig. 5 B, closed squares) , a time point at which <10% of the VSV-G had reached the cis-Golgi compartment and processed to the endoglycosidase D-sensitive form (Fig. 5 B, open squares) . Given that the first 20 min of incubation in vitro is the time period during which VSV-G is sorted and concentrated via 60-nm carrier vesicles (Baich et al., 1994) , these results raised two possibilities. Either the N124I mutant inhibits vesicle budding per se, or the mutant protein (or a rabl-regulated transport factor) needs to be recruited during vesicle formation to function at a later vesicle targeting/fusion step.

To distinguish between these two possibilities, the transport of tsO45 VSV-G was examined in vitro using indirect immunofluorescence after incubating permeabilized NRK cells at 32°C in the presence or absence of mutant protein. As expected, before the start of the incubation, tsO45 VSV-G was restricted to the ER (Fig. 6 A) and could not be detected in the Golgi stack, based on the distribution of the cis/medial Golgi marker o~-1,2 marmosidase II (Man 1/) (Velasco et al., 1993) (Fig. 6 B) . Incubation in the absence (Fig. 6 , C and D) or presence of excess wild-type rabla (Fig. 6 , E-J) led to the redistribution of VSV-G from the ER to pre-Golgi punctate structures that overlapped with p58 (Fig. 6, E [VSV-G] and F [p58]), a marker for pre-Golgi intermediates (Saraste et al., 1987; Saraste and Svensson, 1991) and to the Golgi compartments containing Man II (Fig. 6, G and 5) . However, the N124I mutant, unlike the $25N mutant , did not affect the export of VSV-G from the ER to p58-containing pre-Golgi intermediates ( 

15B CHO cells was measured as described previously (Beckvrs et al., 1987) . Assay mixtures were supplemented with the indicated amounts of rabla wild type (i) or rabla(N124I) (n). The inset shows SDS-PAGE analysis of 2 #g of the purified mutant protein stained with Coomassie blue. (B) Temporal sensitivity of VSV-G transport to rabla(N124I). At the times indicated, the assays were shifted to ice, and the cells collected by centrifugation (3 rain at 500 g). The membranes were resuspcnded in fresh reaction mixture, s supplemented with 1.25/~g rabla(N124D and either maintained on ice (n) or incubated at 320C for the remainder of a total of 90 rain (,l). microscopy, the extent of overlap of VSV-G with p58-containing structures was '~75%, with <5-10% overlap with Man H-containing Golgi compartments.

Using immunoelectron microscopy, we next identified the pre-Golgi structures accumulating VSV-G in the presence of the rabla(N124I) mutant after a 45-min incubation at the permissive temperature in vitro. VSV-G was exported from the ER and accumulated in vesicles and numerous VTCs with a high proportion of vesicular profiles (Fig. 7 , A and C-D).

As shown in Fig. 2 B , the mean linear densities of VSV-G in vesicles (34.5 + 18.8 gold particles//zm, 36 vesicles counted) and VTCs (29.7 + 11.8 gold particles//~m, 37.1 + 17.0 particles/VTC; 44 VTCs counted) in the presence of rabla(N124I) were similar to the density observed in vesicles in the absence of the mutant protein after 15 min incubation at the permissive temperature (29 + 17 gold particles//~m; Balch et ai., 1994) . However, the N124I mutant completely blocked transport of VSV-G to Golgi stacks (Fig. 7 A,/arge arrows). Quantitation revealed that the linear membrane density of VSV-G in Golgi stacks was only 1.2 + 1.3 gold particles//~m (two gold particles per stack, 28 Golgi stacks counted) (Fig. 2 B) . No inhibition of transport to the cis face of the Golgi stack was observed in the presence of excess wild type rabla (Fig. 7 B) , a result consistent with the efficient processing of VSV-G to Golgi forms in the presence of wild type rabla (Fig. 5) . Thus, in contrast to the $25N mutant, N124I mutant inhibits exclusively a late vesicle targeting/fusion step.

Consistent with a role for rabl in vesicle targeting/fusion, we previously observed a late block in transport using a synthetic peptide identical to the effector domain of rabl (Plutnet et al., 1990; Balch et al., 1993) . When examined using indirect immunofluorescence Schwaninger et al., 1992) , incubation of perforated cells with this peptide mimetic leads to the accumulation of VSV-G in punctate pre-Golgi structures. Using immunoelectron microscopy, we found that incubation in the presence of the effector domain peptide led to the accumulation of VSV-G in VTCs morphologically similar to those observed in the presence of the rabla(N124I) mutant (Fig. 1 E) . The mean linear density of VSV-G in these VTCs was 30.5 + 11 gold particles//~m with an average of 26.1 + 10.4 gold particles/VTC (19 VTCs examined) (Fig. 2 C) . Consistent with the efficient inldbition of transport by the peptide, no significant transport to the Golgi stack could be detected (Fig. 2 C) . Moreover, little VSV-G could be detected in independent carrier vesicles (Fig. 2 C) , and the profiles of most VTCs were vesicular rather than tubular in shape (Fig. 1 E) . These results are consistent with the possibility that the peptide may sequester the function of an effector protein required for a late vesicle targeting or fusion step.

We have recently provided biochemical evidence that the nonclathrin coat protein B-COP (Duden et al., 1991; Serafini et al., 1991; Waters et al., 1991; Orci et al., 1993) is essential for export of VSV-G from the ER and may be recruited coordinately with rabl as part of an 18 to 19S precoat particle (Peter et al., 1993) . These results predict that the coat structure present on vesicles and VTCs should contain both rabl and B-COP. Using an antibody specific for B-COP, both vesicles and budding profiles at the tips of cisternae and along the c/s face of the Golgi stack labeled strongly for B-COP in perforated cells incubated for 30 min at 32°C in vitro (Fig.  8 A) . In addition, VTCs containing VSV-G found either adjacent to the nucleus (Fig. 8 B) or in more peripheral locations (not shown) contained B-COP. Similarly, rabl colocalized with VSV-G on the external surface of VTCs (Fig. 8 C) and on the as~medial compartments of Golgi stacks (data not shown). This degree of colocalization is consistent with our previous observations that rabl and B-COP are abundant on pre-Golgi punctate intermediates and Golgi stacks based on indirect immunofluorescence Schwaninger et al., 1992) .

To insure that the VSV-G-and B-COP-containing vesicles observed in Fig. 8 A were not derived from vesicles budding from early Golgi compartments, the distribution of ~-COP was examined in cells incubated in the presence of GTP3,S. As expected, under these conditions, the Golgi stack was devoid of VSV-G (compare Fig. 8 D to Fig. 8 A) , although B-COP containing budding profiles at the tips of Golgi cisternae, as well as numerous B-COP containing vesicles lacking VSV-G, could be detected along the proximal face. Despite the clear transport block, VSV-G containing vesicles and VTCs derived from the ER also labeled strongly for B-COP (Fig. 8, E-H) . The combined results establish that the p58-containing vesicular carriers mediating ER to Golgi traffic in vitro contain both rabl and B-COP.

In yeast, the function of the rabl homologue Yptlp has been reported to be sensitive to cytosolic Ca 2+ (Schmitt et al., 1988) . Moreover, mutant alleles of PRM1, a Ca 2+ ATPase present in the Golgi, are strong suppressors of yptl-1 defects in ER to Golgi transport (Rudolph et al., 1989) . Consistent with a potential role for Ca 2+ in ER to Golgi transport, we have previously demonstrated that EGTA is a potent inhibitor of transport in vitro (Beckers and Balch, 1989; Beckers et al., 1990) .

The effect of incubation of perforated cells at the permissive temperature in the absence of Ca :+ is shown in Fig. 1 . VSV-G was efficiently exported from the ER, but was found to accumulate in tightly packed VTCs (Fig. 1, D and F) . As was the case with GTP3,S (Fig. 1, A-C) , VTCs formed in the presence of EGTA show a high proportion of vesicular profiles compared to tubular profiles with a mean linear membrane density of 31.7 ± 9.4 gold particles//~m (Fig. 2 D) . However, VTCs accumulated in the presence of EGTA had an average value of 75 + 26.4 gold particles/VTC (17 VTCs counted). The latter number is indicative of the larger size of the v'rCs formed in the presence of EGTA compared to those observed under normal transport conditions at the same time point (21 ± 19 gold particles/VTC; Balch et al., 1994) or in the presence of GTP3,S (30.8 + 11.8 gold particles/VTC). Moreover, there was a statistically significant (P = 0.01) decrease in the linear density of VSV-G in the ER (1.5 + 1.62 gold particles/~m membrane length) (Fig. 2 D) , compared to that observed before incubation (5.2 ± 2.1; Balch et al., 1994) or in the presence of GTPTS (Fig. 2 A) , indicating that VSV-G was efficiently exported from the ER in the absence of cytosolic Ca 2+. The combined results suggest that Ca 2+ is required at a late targeting or fusion step, which also requires rabl function.

Incubation of perforated NRK cells in the presence of the rabla(N1241) mutant led to the accumulation of VSV-G in pre-Golgi vesicles and VTCs, efficiently inhibiting delivery to the Golgi stack. Similarly, transient expression of rablb-(N121I) (a functionally equivalent isoform with 92 % identity to rabla) in HeLa cells led to the accumulation of VSV-G in punctate, pre-Golgi structures that overlap with the distribution of the pre-Golgi intermediate marker protein p53 (Tisdale et al., 1992) (Plutner, H., and W. E. Balch, unpublished observation) (Schweizer et al., 1988) . Consistent with these results, the yeast counterpart of rabl, Yptlp, has also been suggested to be required for a late vesicle targeting/fusion step in vivo and in vitro (Segev et al., 1988; Bacon et al., 1989; Kaiser and Schekman, 1990; Rexach and Sehekman, 1990; Segev, 1991) . Moreover, Yptlp has recently been shown to be a component of 50-nm carrier vesicles involved in ER to Golgi transport (Lian and Ferro-Novick, 1993) . Thus, it is apparent that the presence of the N124I mutant blocks transport in a dominant fashion by preventing interaction of the endogenous wild type rabl with a transport component(s) required for the targeting or fusion of transport vesicles with the cis-Golgi compartment.

What is the defect in rabla(N124I) function responsible for this phenotype? The equivalent H-ras (Nll6I) mutant is transforming, suggesting that this mutation leads to constitutive activation (Walter et al., 1986) . Based on the inability of both mutants to bind GDP or GTP in vitro and the structural conservation of ras-like GTPases, the N124I mutation may have a similar effect on rabla. If so, the ability of carrier vesicles to form, but not fuse, may reflect an alteration in the conformation of the rabl mutant which prevents its normal cycling from the GTP-bound to the GDP-bound form. In the absence of such a conformational switch, late events regulating vesicle targeting/fusion fail to be initiated. Since we have not directly demonstrated that the mutant is actively recruited during vesicle budding, two sites of action are currently possible. Either the mutant is recruited during vesicle budding in place of endogenous wild type rabl, or alternatively, the mutant binds directly to a downstream component required for vesicle targeting/fusion, thereby preventing completion of the transport cycle by vesicles containing wild type rabl. Interestingly, the rabl effector domain peptide inhibits transport at an identical step. Whether the component inhibited by rabla(N124I) mutant and that sensitive to the peptide are identical remains to be determined.

Budding from the ER We previously demonstrated that the only concentration step detectable in the transport of VSV-G to the trans Golgi network occurs during vesicle budding from the ER. In the present study, we detected no significant increase in the concentration of VSV-G in vesicles or VTCs accumulating in the presence of the rabla(N124I) mutant or inhibitors that block the fusion of carrier vesicles or VTCs to the Golgi stack. These results are consistent with the interpretation that sorting and concentration of VSV-G is completed upon budding from the ER . They argue against the possibility that concentration may occur as a result of, for example, recycling of membrahe and lipid via additional rounds of budding from tubular elements present in VTCs. The results also provide evidence that the rabla(N124I) mutant, the effector domain peptide and the absence of Ca 2+ lead to the accumulation of a common pre-Golgi intermediate by blocking the normal transport process. Since rabl is also required for transport between Golgi compartments Davidson and Balch, 1993) , steps in which further concentration cannot be detected (Orci et ai., 1986; Balch et al., 1994) , it is presently unclear whether rabl per se is directly involved in protein sorting or concentration. However, a number of other small GTPases including Arf Dascher and Balch, 1994) and Sarl (Nakano and Muramatsu, 1989; Oka et al., 1991; d'Enfert et al., 1991; Barlowe et al., 1993; Kuge et al., 1994) , as well as an unknown G~, (s) protein are essential for export from the ER in mammalian cells. These proteins and/or their associated machineries are currently reasonable candidates for the regulation of steps related to protein sorting and concentration.

Carrier vesicles budding from the ER of NRK cells rapidly (within 5-10 rain) organize into VTCs in vitro , which are similar in structure to putative pre-Golgi elements containing p58 in intact cells (Saraste and Svensson, 1991) . A characteristic of both p53-and p58-containing pre-Golgi intermediates in vivo (Schweizer et al., 1991; Saraste and Svensson, 1991) , or those observed here in vitro, is the presence of short tubular elements. More extensively tubulated or vacuolated pre-Golgi elements have been observed to accumulate nascent membrane and secretory proteins upon incubation at reduced temperature (Saraste and Kuismanen, 1984; Saraste et al., 1986; Saraste and Svens- son, 1991; Schweizer et al., 1988 Schweizer et al., , 1990 Lotti et al., 1992) or during virus infection, particularly using viruses which mature in pre-Golgi intermediates (Tooze et al., 1984; Sodeik et al., 1993; Krijnse-Locker et al., 1994) . We suggest that the morphology of pre-Golgi intermediates observed in different cell lines, in the presence of selected viruses and under different incubation conditions, is likely to vary considerably reflecting rate-limiting step(s) in either vesicle budding from the ER or in the targeting and fusion of peripherally derived 60-nm carrier vesicles to the perinuclear cis-Golgi compartment. Our results are most consistent with the interpretation that transport between the ER and the cis-Golgi compartment involves a single round of vesicle budding/fusion. Since VTCs lack continuity with the ER, lack the abundant ER marker protein CaBP3, and appear subsequent to the budding of individual vesicles, we consider it likely that they represent collection of vesicles en route to the cis-Golgi compartment.

Both ER to Golgi (Beckers et al., 1990) and intra-Golgi transport Davidson and Balch, 1993) require Ca 2÷. In this paper we have shown that the fusion of vesicles or VTCs with the acceptor compartment is the site of Ca 2+ dependence. Accumulation of VTCs was particularly striking in the absence of Ca 2+ and, like those accumulated in the presence of the rabl effector domain peptide, these VTCs frequently appeared to have a higher content of vesicles as opposed to the short, tubular elements observed in VTCs formed under normal incubation conditions, suggesting that Ca 2+ was also required for the fusion of nascent vesicles to generate these tubular elements. Thus, VTCs principally represent a collection of vesicles en route to the perinuclear Golgi. The ability of ER-derived carrier vesicles to form short tubules in VTCs may reflect the welldocumented ability of "like" compartments of the exocytic and endocytic pathways to fuse with each other.

It is now apparent that rab proteins and a family of Ca 2+dependent proteins may specifically regulate the fusion of carrier vesicles mediating traffic between many different compartments of the exocytic pathway (Bennett and Scheller, 1993; S61iner et al., 1993) . For example, rabphilin is a Ca2+-binding protein possibly involved in rab3 function (Shiratakis et al., 1993) . Moreover, several lines of evidence have suggested a link between the requirement for Ca 2+ and rabl function (Schmitt et al., 1988) . The PMR1 gene in yeast encodes a Ca2+-adenosine triphosphatase (ATPase) that colocalizes to early Golgi compartments (Antebi and Fink, 1992) . Pmrl mutants are strong suppressors of yptM defects in ER to Golgi transport (Rudolph et al., 1989) , suggesting that these two components function in the context of a protein complex that participates in vesicle targeting and/or fusion. In some respects, the fusion of VTCs with the cis-Golgi compartment is reminiscent of the rab3/Ca2+-dependent fusion of vesicles to the cell surface during regulated secretion (reviewed in Holz et al., 1991; see Oberhauser et al., 1992; Padtield et al., 1992; Senyshyn et al., 1992) . However, unlike regulated events at the cell surface, which require a transient increase in Ca 2+ as part of an upstream signaling pathway, ER to Golgi vesicles and V'ICs function optimally at normal intracellular (•100 riM) Ca 2+ concentrations (Beckers and Balch, 1989; Rexach and Schekman, 1990) . At this time, we prefer the model in which both rabl and a Ca2+-binding protein facilitate the fusion of pre-Golgi carriers to the cis-Golgi compartment.

The cumulative evidence from this and previous studies (Tisdale et al., 1992; Davidson and Balch, 1993; allows us to propose a tentative working model for the function of rabl in ER to Golgi transport (Fig. 9 ). This model, in principle, is also applicable to vesicular traffic between early Golgi compartments, We propose that rabl is recruited in the GDP-bound conformation, either as part of a 18-19S precoat particle containing/3-COP or in the form of a smaller guanine nucleotide dissociation inhibitor (GDDrabl complex (Fig. 9) . The requirement for/3-COP is derived from our recent observations that a functional pool of rabl involved in vesicle budding is associated with a high molecular weight protein complex containing/3-COP (Peter et al., 1993) . Consistent with these results, we demonstrated in the present study that/3-COP is associated with vesicles and VTCs containing VSV-G en route to the cis face of the Golgi stack in vitro. A similar distribution of B-COP to pre-Golgi carriers has been detected in vivo (I-Iendricks et al., 1993; Oprins et al., 1993; Pepperkok et al., 1993) . GDI is now believed to be associated with the GDP-bound, cytosolic form of rab proteins and to assist in their delivery to the membranes (Soldati et al., 1993; Ullrich et al., 1993) . The possibility that GDI is involved in an early recruitment step is consistent with our observations that a second, low molecular weight pool of rabl, which is likely to represent a GDI-rabl complex (Balch, W. E., and E Peter, unpublished observations) , can also serve as a source for rabl during vesicle budding from the ER (Peter et al., 1993) . The necessity for recruitment of rabl in the GDP-bound form is also consistent with our previous observation that overexpression of rablb(Q67L), a mutant that is likely to be restricted to the GTP-bound form because of defective GTP hydrolysis, is not a dominant inhibitor of transport in vivo (Tisdale et al., 1992) , suggesting that it is excluded from the Hgure 9. A summary of the role of rabl in the regulation of vesicles and VTCs mediating vesicular traffic between the ER and the C-olgi. See text for details. transport machinery (Fig. 9) . Finally, the model is consistent with the observation that recruitment is likely to involve a protein (GEP), given the inhibitory properties of the $25N mutant which markedly attenuates export from the ER by possibly inhibiting rabl-GEP . GTP hydrolysis, presumably mediated by a rabl-specific GTPase-activating protein, is likely to be involved in a late vesicle targeting/fusion step. This interpretation is consistent with the observation that rabl is found on ER to Golgi vesicular carriers. As suggested above, the N124I substitution may restrict rabl to a conformation which allows it to interact in an irreversible fashion with an effector involved in vesicle targeting or fusion (Fig. 9) . While the putative target for rabl involved in targeting/fusion is unknown, recent studies have found that H-ras(N116I) can form a complex with a purified ras-specific GEP (Hwang et al., 1993) . These results raise the possibility that the N124I mutant can at least interact and be recruited by a rabl-GEP during vesicle budcling. However, the mode of interaction of the $25N and N124I mutants with GEP and/or other transport components are clearly different in that the two mutants diverge with respect to their requirements for posttranslational isoprenylation, competition with wild type rabl, and morphological phenotypes. One possibility to account for these results is that while the $25N mutant reduces the overall activity or efficiency of rabl-GEP in promoting wild type rabl recruitment, the N124I mutant adopts a conformation that supports its efficient recruitment during coat assembly and vesicle budding, but renders it unable to undergo a critical conformational change required to initiate vesicle targeting/fusion (Fig. 9) . Since both mutants lead to inhibition of intra-Golgi transport and loss of the Golgi stack in a fashion distinct from that observed for brefeldin A or microtubule depolymerizing agents such as nocodazole ; Balch, W. E., unpublished results), it is apparent that the integrity of the Golgi stack is exceptionally sensitive to the disruption of vesicular traffic between compartments (see Wilson et al., 1994) .

Given the complexities of events involved in vesicle budding and fusion, further experiments are currently in progress to identify rabl-GEP and the targeting/fusion machinery regulated by rabl and rabl-speciiic GTPase-activating protein. With these components in hand, we should be able to refine the above working model, which is largely derived from the use of selected mutants defective in guanine nucleotide exchange and hydrolysis.

The yeast Ca2+-ATPase homologue, PMR1, is required for normal Golgi function and localizes in a novel Oolgilike distribution

The GTP-binding protein Yptl is required for transport in vitro: the Golgi apparatus is defective in yptl mutants

Use of synthetic peptides to study the function of GTPasos regulating vesicular traffic

ADP-ribosylation factor (ARF) is required for vesicular trafficking between the endoplasmic reticulure (ER) and the c/s Go lgi compartment

Vesicular stomatitis virus glycoprotein (VSV-G) is sorted and concentrated during export from the endoplasmic reticulum

Purification and characterization of SARlp, a small GTP-binding protein required for transport vesicle formation from the endoplasmic reticulum

Trimeric G proteins and vesicle formation

Calcium and GTP: essential components in vesicular trafficking between the endoplasmic reticulum and Golgi apparatus

Semi-intact cells permeable to macromolecules: use in reconstimtion of protein transport from the endoplasmic reticulum to the Golgi complex

Sequential intermediates in the transport of protein between the endoplasmic reticulum and the Golgi

The molecular machinery for secretion is conserved from yeast to neurons

Secl2p-dependent membrane binding of the small GTP-binding protein Sarlp promotes formation of transport vesicles from the ER

Dominant inhibitory mutants of ARF1 inhibit ER to Golgi transport and trigger the disassembly of the Golgi apparatus

Differential inhibition of multiple vesicular transport steps between the endoplasmic reticulum and trans Golgi network

/~-COP, a 110-kD protein associated with non-clathrin-coated vesicles and the Golgi complex, shows homology to/3-adaptin

Small GTP-binding proteins and their role in transport

Disruption of endoplasmic reticulum to Golgi transport leads to the accumulation of aggregates containing/~-COP in pancreatic acinar cells

Mechanisms involved in calcium-dependent exocytosis

Probing SDC25-C-domain-induced guanine nucleotide exchange by guanine ring binding domain mutants of v-H-ms

Distinct sets of SEC genes govern transport vesicle formation and fusion early in the secretory pathway

Isoprenoid modification of rab proteins terminating in CC or CXC motifs

Pats (CXXX) and tab (CC/CXC) prenylation signal sequences are unique and functionally distinct

Characterization of the budding compartment of mouse hepatitis virus: evidence that transport from the RER to the Golgi complex requires only one vesicular transport step

Sarl promotes vesicle budding from the endoplasmic reticulum but not G-olgi compartments

Boslp, an integral membrane protein of the endoplasmic reticulum to Golgi transport vesicles, is required for their fusion competence

Envelope proteins of vesicular stomatitis virus: effect of temperature-sensitive mutations in complementation groups HI and V

A novel GTP-binding protein, Sarlp, is involved in transport from the endoplasmic reticulum to the Golgi apparatus. 7

lmmunocytochemical analysis of the transfer of the vesicular stomatitis virus G glycoprotein from the intermediate compartment to the Golgi complex

GTPases: multifunctional molecular switches regulating vesicular tra~c

A GDP-bound form or rabl inhibits protein export from the endoplasmic reticulum and transport between Golgi compartments

Exocytotic fusion is activated by Rab3a peptides

Reconstitution of GTPbinding Sarl protein flmction in endoplasmic reticulum to Golgi transport

B-COP localizes mainly to the cis-Gol~ side in exocrine pancreas

A new type of coated vesicular carrier that appears not to contain clathrin: its possible role in protein transport within the Golgi stack

Budding from Golgi membranes requires the coatomer complex of nonclathrin proteins

A synthetic peptide of the rab3a effector domain stimulates amylase release from permeabillzed pancreatic acixfi

B-COP is essential in biosynthetic membrane transport from the endoplasmic reticulum to the Golgi complex in vivo

B-COP i~ essential for transport of protein from the endoplasmic reticulum to the Golgi in vitro

Rablb regulates vesicular transport between the endoplasmic reticulum and successive Golgi compartments

Morphological analysis of protein transport from the endoplasmic reticulum to Golgi membranes in digitonin permeabilized cells: role of the p58-containing compartment

Synthetic peptides of the rab effector domain inhibit vesicular transport through the secretory pathway

Vesicle-mediated protein sorting

Distinct biochemical requirements for the budding, targeting, and fusion of endoplasmic reticulum-derived transport vesicles

Molecular dissection of the secretory pathway

The yeast secretory pathway is perturbed by mutations in PMR1, a member of a calcium ATPase family

Pre-and post-Golgi vacuoles operate in the transport of Semliki Forest virus membrane glycoproteins to the ceil surface

Temperature-sensitive steps in the transport of secretory proteins through the Golgi complex in exocrine pancreatic cells

Antibodies to rat pancreas Golgi subfractinns-identification of a 58-kD cis-Golgi protein

Distribution of the intermediate elements operating in ER to Golgi transport

Study of a temperatoresensitive mutant of the ras-related YPT1 gene product in yeast suggests a role in the regulation of intraceilular calcium

Sequential transport of protein between the endoplasmic reticulum and successive Golgi compartments in semi-intact cells

Multiple GTP-binding proteins regulate vesicular transport from the endoplasmic reticulum to Golgi membranes

Identification, by a monoclonal antibody, ofa 53-kD protein associated with a tobulo-vesicular compartment at the cis side of the Golgi apparatus

Identification of an intermediate compartment involved in protein transport from endoplesmic reticulum to Golgi apparatus

The isolated endoplasmic reticulum-Golgi intermediate comparUnent exhibits properties that are different from endoplasmic reticulum and cis-Golgi

Mediation of the attachment or fusion step in vesicular transport by the GTP-binding Yptl protein

The yeast GTP-binding YPT 1 protein and a mammalian counterpart are associated with the secretion machinery

Synthetic peptides of the etfector-binding domain of rab enhance secretion from digitonin-permeabilized chromatfin cells. FEBS (Fed. Eur

A coat subunit of Golgi-derived non-clathrin-coated vesicles with homology to the clathrin-coated vesicle coat protein B-adaptin

Rabphilln-3A: a putative target protein for stag p25A/rab3A p25 small GTP-binding protein related to synaptotag-mJn

Assembly of vaccinia virus: role of the intermediate compartment between the endoplasmic reticulum and the Gol~ stacks

Rab GDI: a solubillzing and recycling factor for tab9 protein

SNAP receptors implicated in vesicle targeting and fusion

GTP-bindin 8 mutants of rabl and rab2 are potent inhibitors of vesicular transport from the endoplasmic reticulum to Golgi complex

Replication of coronavirus MHV A59 in sac ceils: determination of the first site of buddin 8 or progeny virinns

Rab GDI as a general regulator for the membrane association of rab proteins

Ceil-type dependent variations in the subcellular distribution of ~-mannosidese I and II

The oncogenic activation of human p21 ~ by a novel mechanism

Coatomer': a cytnsolic protein complex containing subtmits of non-clathrin-coated Golgi transport vesicles

A rabl mutant affecting guanine nuclcotide exchange promotes disassembly of the Golgi apparatus

Rab GTPases in vesicular transport

We thank K. Moremen and K. Howell for providing the antibody reagents used in these studies. We thank J. Saraste and H.-P. Hauri for providing the p58 and p53 antibodies, respectively. We would like to thank Tammy McQuiston for her assistance in photography. TSRI manuscript number 7461-CB.Received for publication 2 November 1993 and in revised form 6 January 1994.