Binding of Sly1 to Sed5 enhances formation of the yeast early Golgi SNARE complex

In eukaryotic cells, major intracellular protein trafficking is mediated by vesicular transport, and this has been well studied in the past decade. Docking and fusion of the transport vesicles to the target organelle are mediated by several proteins. In vitro studies using liposomes showed that the membrane proteins SNAREs (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) form the central physical fusion machinery, and the combination of SNAREs determines the compartmental specificity(Weber et al., 1998;McNew et al., 2000). In addition to SNAREs, other proteins are also needed to fulfil membrane fusion in vivo. In Saccharomyces cerevisiae, proteins of the Ypt, Uso1 and Sec1 families have important roles(Nichols and Pelham, 1998;Yoda and Noda, 2001).

Members of the Sec1 family have strong physical interactions with target membrane (t-)SNAREs. Munc18/nSec1 binds to syntaxin 1a, a neural plasma membrane t-SNARE (Hata et al.,1993), and Sly1 binds to Sed5, a yeast Golgi t-SNARE(Søgarrd et al., 1994). Recently, the X-ray crystal structure of the Munc18-syntaxin 1a complex was reported (Misura et al.,2000). However, experimental data indicate that the interaction between Sec1 family proteins and t-SNAREs differs depending on their localization and species. Munc18 is not included in the so-called 20S SNARE complex (Söllner et al.,1993), whereas Sly1 is coprecipitated with the similar complex from the yeast sec18-1 cells(Søgarrd et al., 1994). Typical t-SNAREs have three helical regions for protein-protein interaction. Munc18 binds to the helical regions H1 and H3 of syntaxin 1a and inhibits the binding between syntaxin 1a and VAMP, a neural vesicular (v-)SNARE(Pevsner et al., 1994), as well as the binding between syntaxin 1a and α-SNAP(Hayashi et al., 1994). But Sly1 binds only to the H1 region of Sed5 and does not bind to the rest of Sed5 and therefore does not interfere with the interaction between Sed5 and the yeast α-SNAP Sec17 (Kosodo et al.,1998). Moreover, Sec1 does not interact with uncomplexed Sso1, a yeast plasma membrane t-SNARE (Carr et al.,1999).

Genetic studies in S. cerevisiae have indicated that a number of ER-Golgi secretory genes have interactions with SLY1. A single copy of the gain-of-function mutation SLY1-20 was originally identified as a suppressor mutation that enables the ER-Golgi vesicular transport to occur in the absence of YPT1 (Dascher et al., 1991). Interestingly, it could also suppress the temperature-sensitive defects caused by mutations in many other ER-Golgi secretory genes including uso1-1 and bet1-1(Ossig et al., 1991;Sapperstein et al., 1996;Kito et al., 1996). The sly1^ts allele was identified in a screen for a mutant that arrests transcription of the genes for ribosomal proteins and RNAs at the restrictive temperature (Mizuta and Warner, 1994). The sly1^ts mutant grows normally at 25°C but arrests growth when the temperature is shifted to 37°C and accumulates the ER-form precursor of vacuolar enzymes. Multicopy USO1, BET1 or HSD1 genes could suppress the temperature sensitivity of sly1^ts(Kosodo et al., 2001).

Barlowe and colleagues have reproduced the ER-Golgi transport in vitro by using semi-intact cells and temperature-sensitive mutants. They proposed that membrane fusion of the transport vesicles can be classified into tethering,docking and fusion steps. Mutant proteins of Sly1, Bet1, Bos1, Sed5 and a yeast NSF Sec18 do not interfere with tethering, but block docking and fusion steps (Barlowe, 1997;Cao et al., 1998). They have also showed that Sly1, Sed5 and Ypt1 are required on the acceptor Golgi membrane (Cao and Barlowe,2000).

Thus the role of Sec1 family proteins, including Sly1, in membrane fusion has not been fully understood yet (Jahn,2000). Although the binding between t-SNAREs and Sec1 family proteins has an important role, the function of the binding is unclear(Carr, 2001). From this point of view, we focused on the interaction between Sly1 and the fusion machinery in this report.

The yeast strains used in this study are given inTable 1. They were grown in a YEPD (1% Bacto yeast extract (Difco), 2% Bacto peptone (Difco), 2% glucose) or in an SD (0.5% (NH₄)₂SO, 0.17% Bacto yeast nitrogen base without amino acids (Difco), 2% glucose and appropriate supplements) medium. An SG (2% galactose and 1% raffinose instead of glucose in SD) medium was used to induce genes under the GAL1 promoter. Solid media were made with 2% agar. Escherichia coli DH5α (F^-, supE44ΔlacU169 ø80lacZΔM15 hsdR17 recA1 endA1 gyrA96 thi-1 relA1) was used in plasmid propagation, and E. coli BL21(F^-, ompT hsdS gal dcm) was used in the preparation of recombinant proteins. E. coli was grown in an LB (1% Bacto tryptone(Difco), 0.5% Bacto yeast extract, 0.5% NaCl) medium.

The plasmids used in this study are listed inTable 2 and were constructed as follows. The DNA fragments encoding various proteins were amplified either from genomic DNA or from plasmid DNAs harboring the gene by polymerase chain reaction (PCR) with Pfu DNA polymerase (Stratagene). The myc-encoding sequence was subcloned from pKT10mycC (Matsui et al., 1996), and the HA-encoding sequence was from pYN168(Kosodo et al., 2001). The recombinant proteins were produced as glutathione S-transferase (GST) fusion proteins in pGEX4T-3 (Pharmacia), as maltose-binding protein (MBP) fusion proteins in pMAL-c2 (New England Biolabs) or as Strep-tagged fusion proteins in pASK-IBA2 (Genosys) in E. coli. For protein production in yeast, a low-copy vector pRS416 or multicopy vectors pRS426(Sikorski and Hieter, 1989)and pYES2.0 (Invitrogen) were used.

Production and purification of the GST- and MBP-tagged proteins were done as described previously (Kosodo et al.,1998). Strep-tagged Sly1/Sly1^ts was produced in E. coli, extracted and bound to StrepTactin-Sepharose (Sigma) and then eluted with 2.5 mM desthiobiotin after extensive washing. The concentration of all recombinant proteins was determined by a protein assay kit (BioRad), and proteins were stored with 40% glycerol at -20°C.

Binding assays of recombinant proteins were carried out as described previously (Kosodo et al.,1998).

Yeast cells were grown to an OD₆₀₀ of 0.6 to 0.8 in YEPD or SD medium. They were harvested by centrifugation, washed in distilled water and resuspended in an ice-cold B88 buffer (150 mM KOAc, 5 mM Mg(OAc)₂,20 mM HEPES, 200 mM sorbitol, pH 6.8) with PIC (1 mM phenylmethylsulfonyl fluoride (PMSF), 1 μg/ml 1,10-phenanthroline, 1 μg/ml pepstatinA, 1μg/ml aprotinin, 1 μg/ml leupeptin). One-half volume of acid-washed glass beads was added to the cell suspension in a glass tube, which was then vortexed eight times for 30 seconds, with a 30 second incubation on ice between each burst. The crude lysate was centrifuged at 500 gfor 3 minutes at 4°C to remove unbroken cells. The supernatant was used as a yeast lysate in this study.

The yeast lysate was centrifuged at 10,000 g for 10 minutes at 4°C to generate the supernatant (S10) and pellet (P10) fractions. The S10 was centrifuged at 100,000 g for 60 minutes at 4°C to generate the high-speed supernatant (S100) and pellet (P100) fractions. The same amount of each fraction was resuspended in the Laemmli sample buffer(Laemmli, 1970) and applied to SDS-polyacrylamide gel electrophoresis (PAGE). Immunodecoration of western blotting was performed using indicated primary antibodies followed by horseradish-peroxidase-coupled goat anti-mouse or anti-rabbit IgG as secondary antibodies at a 1:5000 dilution. The bound antibodies were visualized by enhanced chemiluminescence (Pierce).

Affinity-purified rabbit anti-Sed5 or the monoclonal anti-myc antibody 9E10 was used for immunoprecipitation. Triton X-100 was added to the yeast lysate to a 1% final concentration. The lysate was incubated on ice for 5 minutes to solubilize membranes and centrifuged at 10,000 g for 5 minutes to remove debris. After the lysate was mixed with an antibody and incubated at 4°C for 1 hour, Protein A-Sepharose beads (Amersham) were added, and the mixture was incubated at 4°C for 2 hours. Beads were washed five times with B88 buffer with 0.5% Tween 20 and boiled in the Laemmli sample buffer for 1 minute. The supernatants were analyzed by SDS-PAGE and western blotting.

The sly1^ts strain was transformed with pYN110(6myc-SED5 under SED5 promoter on a CEN plasmid) or with pKD201 (3HA-BET1 under GAL1 promoter on a 2μ plasmid). The cells were grown to an OD₆₀₀ of 0.5 in 800 ml SD medium and, as for the sly1^ts/pKD201 strain, they were further incubated in 200 ml SG medium to produce 3HA-Bet1 protein for 2 hours. After harvesting the cells by centrifugation, lysates were prepared by vortexing with glass beads in 8 ml ice-cold B88 buffer. The protein concentration of both lysates was adjusted to 8.0 mg/ml, and the aliquots (850 μl each) were stored at-80°C. The lysates from both strains were defrosted before the reaction and centrifuged at 5000 g for 5 minutes to remove the cytoskeleton and aggregates before the reaction. The cleared lysates of two strains (600 μl) were mixed and then divided into eight aliquots (150μl, each including 1.2 mg protein) quickly on ice. The recombinant Strep-tagged Sly1 or Sly1^ts protein (2.4 μg) was added to each mix. The reaction was started by transferring the mix to a 35°C bath and incubated up to the indicated time. In case of the experiments inFig. 6, 1 mM ATP or 1 mM AMP-PNP (final concentration) was added during the reaction. Other conditions for each reaction are described in the figure legends. At the end of the reaction, the mixtures were diluted by adding 850 μl of B88 buffer with 1%Triton X-100 and subjected to immunoprecipitation by the anti-myc antibody. Anti-Sly1, anti-myc and anti-HA antibodies were used for western blotting. Quantification of the signals was performed by a luminoimage analyzer(LAS-1000plus, Fuji film).

Sly1 strongly binds to the yeast Golgi t-SNARE Sed5(Søgarrd et al., 1994),and therefore this interaction is believed a priori to be crucial for their function. However, this needs to be confirmed by experimental evidence. A temperature-sensitive sly1^ts mutant with a single amino-acid substitution (R266K in Sly1) arrests ER-Golgi transport at the restrictive temperature (Mizuta and Warner, 1994; Cao et al.,1998). We sought to determine whether the interaction between Sed5 and Sly1 was altered in the mutant or not. The SLY1 alleles from the wild-type, sly1^ts mutant and gain-of-function SLY1-20 mutant (Dacher et al., 1991) were cloned, ligated with the sequence encoding a hexameric myc tag at their C-termini and introduced into the sly1^ts mutant. The tagged products were functional because the sly1^ts cells carrying SLY1-6myc or SLY1-20-6myc did not show temperature sensitivity. We prepared the lysate containing Sly1-6myc, Sly1-20-6myc or Sly1^ts-6myc at the permissive temperature of 25°C. After solubilizing the membrane with 1%Triton X-100, the myctagged protein was collected by immunoprecipitation, and the amount of coprecipitated Sed5 was examined by SDS-PAGE and western blotting using anti-Sed5 antiserum. The amount of Sed5 bound to Sly1^ts-6myc was much less than that bound to Sly1-6myc or to Sly1-20-6myc (Fig. 1A). This result suggests that the Sly1^ts protein has a reduced binding affinity for Sed5. However, it may be claimed that the reduction in the amount of Sed5 bound to Sly1^ts was a result of the decreased frequency of ER-Golgi transport in the sly1^ts mutant.

To demonstrate the protein interaction directly, we adopted a coprecipitation assay using recombinant proteins. Glutathione S-transferase(GST) was fused at the N-terminus of Sly1 or Sly1^ts protein, and maltose-binding protein (MBP) was fused at the N-terminus of Sed5▵TMD(amino acids 1-324 of Sed5), Sed5H1 (amino acids 15-39), Sed5H2 (amino acids 146-169) or Sed5H3 (amino acids 252-311), each of which carried a myc tag at the C-terminus. The fusion proteins were produced in E. coli,purified by affinity resins and used in the binding assay at 4°C. As shown in Fig. 1B, a considerable amount of GST-Sly1 was recovered in the immunoprecipitate of MBP-Sed5▵TMD-myc (lane 1), whereas much less GST-Sly1^ts could be recovered (lane 2). GST-Sly1 was coprecipitated with MBP-Sed5H1-myc, which contains the Sly1-binding helix of Sed5 (lane 3;Kosodo et al., 1998), but a detectable amount of GST-Sly1^ts was not coprecipitated (lane 4). Neither GST-Sly1 nor GST-Sly1^ts bound to MBP-Sed5H3-myc (lanes 5 and 6). From these results, we conclude that Sly1^ts has a defect in binding to Sed5.

The Sly1 protein is not predicted to have a transmembrane domain, and the recombinant Sly1-fusion proteins produced in E. coli were soluble. By contrast, nearly 60% of Sly1 localized on the Golgi membrane in yeast cells(Lupashin et al., 1996). This membrane localization of Sly1 has been explained by its interaction with Sed5(Grabowski and Gallwitz, 1997;Cao and Barlowe, 2000). Because Sly1^ts turned out to be defective in binding to Sed5, we examined its intracellular localization by differential centrifugation. The lysate with myc-tagged Sly1, Sly1-20 or Sly1^ts was separated into the membrane fractions, P10 or P100, and the soluble fraction S100. Sly1, Sed5, Scs2 [an ER membrane protein (Kagiwada et al.,1998)], Van1 [a Golgi membrane protein(Hashimoto and Yoda, 1997)]and Pgk1 [a soluble cytosolic protein(Schneiter et al., 1999)] in these fractions were detected by western blotting. As shown inFig. 2, we could not find any difference between distributions of the myc epitope in these strains. The distributions of Sed5 and Van1 were similar. It is unlikely that this distribution was caused simply by contamination because Pgk1 was only detected in one fraction — S100. This result suggests that the yeast membrane has other factors than Sed5 to localize more than half of the total Sly1 protein on the membrane.

Using an in vitro assay, it was shown that a combination of the Q-SNAREs(Sed5, Bos1 and Sec22) and an R-SNARE (Bet1) constitute a potential minimum fusion machinery in the ER-Golgi transport, although other combinations may exist (McNew et al., 2000). Thus, it was important to see whether Sly1 binds only to free Sed5 or whether it also binds to Sed5 in the SNARE complex. We found that the sly1^ts mutation is partially suppressed by introduction of multicopy BET1 (Kosodo et al.,2001). This observation suggests that Bet1 may have an important role among v-SNAREs, in concert with Sly1. Therefore, we focused our study on the interaction with Bet1 and Sly1. First, we examined whether Sly1 is in the Bet1-Sed5 SNARE complex in the normal yeast cells. The 20 S complex found in the sec18-1 cell at the nonpermissive temperature contained Sed5,Bet1 and Sly1 proteins as well as Bos1 and Sec17. In case of exocytosis in neuron, it was shown that v-SNARE(VAMP2) was excluded from the n-Sec1(Munc-18)/t-SNARE(syntaxin1) complex(Pevsner et al., 1994).

The genomic SLY1 was replaced with SLY1-6myc in the wild-type strain. The cells grew well without any detectable defect. The lysate was prepared from them, with 1% Triton X-100 added to solubilize the membrane, and then divided into three parts. Immunoprecipitation was performed using an anti-myc monoclonal antibody, an affinity-purified anti-Sed5 antibody or, as a control, preimmune serum from the rabbit from which we raised the anti-Sed5 antiserum. As shown in Fig. 3, Bet1 was detected in the immunoprecipitate of anti-Sed5 (lane 2). This indicates the presence of the Bet1-Sed5 SNARE complex in the lysate. Sly1-6myc was also detected in this immunoprecipitate. In accordance with this, both Bet1 and Sed5 were detected in the immunoprecipitate of Sly1-myc(lane 3). Because Sly1 does not bind to Bet1 directly, this indicates the presence of a complex containing these three proteins. Preimmune serum precipitated none of these proteins (lane 1). These results strongly suggest that Sly1 still binds to the Sed5-Bet1 SNARE complex. As the H1 helix of Sed5 binds to Sly1 and the H3 helix binds to Bet1(Kosodo et al., 1998;Sacher et al., 1997),simultaneous binding of Sed5 to both Sly1 and Bet1 would be possible. This Bet1-Sed5 interaction was suggested to be susceptible to the action of dissociating ATPase, because Bet1 was not detected in similar experiments when 0.5 mM ATP was added to the immunoprecipitation mix (data not shown).

As a complex containing Sly1, Sed5 and Bet1 was detected, Sly1 may somehow be involved in the formation of the Bet1-Sed5 SNARE complex. We constructed an in vitro assay system to evaluate the effect of Sly1 in this reaction. The lysate of the sly1^ts mutant was selected as the basal condition, because Sly1 is an essential protein and its depletion by other means have obvious shortcomings. The sly1^ts mutant was transformed either with 6myc-SED5 on a CEN plasmid (strain I) or with a GAL1p-3HA-BET1 construct on a 2μ plasmid (strain II). These strains were grown at 25°C, and the expression of 3HA-BET1in strain II was induced in SG medium for 2 hours. The cells were harvested,disrupted by vortexing with glass beads, centrifuged at 4000 gfor 5 minutes and the supernatants (S4) were stocked at -80°C. Almost all endogenous Sly1^ts protein was recovered in S4 (data not shown). Sly1 or Sly1^ts protein with Strep tag at the C-terminus was produced in E. coli and purified by affinity resin to a single band in SDS-PAGE. The lysates of strains I and II were mixed and incubated with or without adding a purified recombinant protein. Then, the mixture was diluted 6.7-fold by B88 buffer on ice, and Triton X-100 was added to 1% of the final concentration to dissolve the membranes. The monoclonal anti-myc antibody was used to precipitate 6myc-Sed5, and 3HA-Bet1, exogenous Sly1/Sly1^ts-Strep and endogenous Sly1^ts in the precipitate were detected by western blotting.

First, we measured the effect of temperature on the cell-free system. 0°C serves as a control for basal reactions. At 15°C, both wild-type and sly1^ts strains grew slowly without apparent difference(data not shown). At 25°C, the wildtype grew slightly faster than the sly1^ts mutant (data not shown). 35°C is a restrictive temperature for the sly1^ts strain and the vesicle transport between ER-Golgi stops at this temperature(Mizuta and Warner, 1994). The reaction mix was kept at these temperatures for 20 minutes. At 0°C, no increase of the amount of Bet1 was found when either Sly1-Strep or Sly1^ts-Strep was added (Fig. 4, lanes 1 and 2). At 15°C, a slight increase in the amount of 3HA-Bet1 was detected in the precipitates by adding either protein(Fig. 4, lanes 3 and 4). At 25°C, either protein gave a considerable increase, but Sly1-Strep had a stronger effect (Fig. 4, lanes 5 and 6). At 35°C, Sly1^ts-Strep had little effect(Fig. 4, lanes 7 and 8). The activity of recombinant proteins in the assay system may have a good accordance with yeast growth.

The temperature dependence of the binding of 3HA-Bet1 to 6myc-Sed5(Fig. 4) also indicates that the Bet1-Sed5 complex was formed during the incubation at each temperature and not in the later process after the dilution and addition of Triton X-100. When Triton X-100 was added before mixing the lysates, we could not find any increase in the amount of 3HA-Bet1 (data not shown).

We further evaluated the time course of the change in the amount of coprecipitated proteins after the temperature shift from 0°C to 35°C. During the incubation at 35°C, the amount of coprecipitated 3HA-Bet1 clearly increased by adding Sly1-Strep protein to the mixture(Fig. 5A, lanes 1-3). Addition of the same amount of Sly1^ts-Strep did not give such an increase(Fig. 5A, lanes 4-6). The exogenous Sly1-Strep protein was found in the precipitates, but Sly1^ts-Strep was hardly detected in them (lanes 1-6). We detected a faint band of endogenous Sly1^ts protein in all lanes (compare withFig. 4). Most of the endogenous Sly1^ts protein bound to 6myc-Sed5 before cell disruption was not released in the present condition and might prevent exogenous Sly1^ts-Strep from replacing its position.

The amount of protein in the immunoprecipitate was measured using a luminoimage analyzer. At 0°C, no difference was found in the amount of 3HA-Bet1 irrespective of whether Sly1-Strep or Sly1^ts-Strep was added. The SNARE-mediated membrane fusion was temperature dependent and did not occur at 0°C (Weber et al.,1998). At 35°C, the increased amount of 3HA-Bet1 with Sly1-Strep was 17 times of that with Sly1^ts-Strep(Fig. 5B). These indicate that Sly1-Strep stimulated formation of the Bet1-Sed5 trans-SNARE complex in a time-and temperature-dependent manner.

Next, we addressed the question of whether the trans-SNARE complex is formed from the cis-SNARE complex or whether its dissociation by NSF/Sec 18 ATPase is required to prime complex formation. If 1 mM ATP was added into the reaction mix, the amount of coprecipitable 3HA-Bet1 decreased(Fig. 6, lanes 2 and 6), and it was almost under the limit of detection in the case of Sly1^ts-Strep. This indicates that ATPase Sec18 that dissociates the cis-SNARE complex was active in our reaction mix. If the unhydrolyzable analog AMPPNP was added instead of ATP, the result was the same as the result without its addition (compare lane 1 with 3, and lane 5 with 7). This indicates that trans-SNARE complex formation in our system did not require ATP hydrolysis. However, if the lysate was incubated in the presence of ATP with Sly1-Strep or Sly1^ts-Strep and then AMPPNP was added, the amount of 3HA-Bet1 significantly increased (lane 4 or 8). This result indicates that the dissociated SNAREs present at the time of lysate preparation participated in the formation of the trans-SNARE complex under the standard conditions. Probably ATP in the lysate was soon exhausted by reactions that consumed ATP. We guess that the pre-incubation of lysate with ATP increased the amount of dissociated SNAREs by the action of the NSF/Sec18 ATPase and thus the amount of the newly formed SNARE complex increased. AMPPNP probably inhibited further action of Sec18 ATPase, which would also dissociate the newly formed SNARE complex after vesicle fusion. We conclude that dissociation of pre-formed SNARE complex is also a prerequisite for trans-SNARE complex formation in the presence of Sly1 protein.

In this report, we have demonstrated that less Sly1^ts protein binds to Sed5 than wild-type Sly1, and this is because of the reduced binding affinity of the mutant Sly1^ts protein for Sed5. We have also shown that Sly1 is included in a SNARE complex containing Bet1 and Sed5 in vivo. Finally, we have shown that addition of purified Sly1 enhances the formation of the trans-SNARE complex in a cell-free system. This is the first demonstration that Sly1 stimulates SNARE complex formation.

GLUT4 vesicle trafficking is abolished at the restrictive temperature in a 3T3L1 mutant cell that has the same point mutation in Munc18c as the yeast sly1^ts allele(Thurmond and Pessin, 2000). Thurmond and Pessin showed the mutant Munc18c had a reduced affinity for the t-SNARE syntaxin 4. On the basis of our results and their report, we conclude that a single amino-acid substitution R266K in Sly1 renders its binding activity to t-SNARE much lower and temperature labile, and consequently vesicle transport is abolished at a nonpermissive temperature. Therefore, the binding of the Sec1 family protein to t-SNARE would be necessary to form the correct SNARE complex.

Sly1 is one of the Sec1 family proteins in S. cerevisiae. The Sec1 family proteins have long been implied as regulators of the membrane fusion process in vesicular transport because of their high affinity for t-SNAREs that trigger membrane fusion (Halachmi and Lev, 1996; Nichols and Pelham,1998). The Munc 18/n-Sec1 protein was first identified and purified from mammalian brain owing to its high binding affinity to the t-SNARE syntaxin 1 (Hata et al.,1993). Munc18 was thought to be a negative regulator of membrane fusion, since it prevented synaptobrevin and SNAP from binding to syntaxin 1(Pevsner et al., 1994). However, the negative regulator model does not fully explain why the protein is essential for neurotransmitter release and intracellular vesicle transport(Ossig et al., 1991;Verhage et al., 2000). By contrast, it has also been proposed that the Sec1 family proteins have a positive or chaperone-like role rather than a negative role(Grabowski and Gallwitz, 1997;Yang et al., 2000;Jahn, 2000). Our results strongly suggest that Sly1, at least, has a positive role in SNARE complex formation, which is followed by membrane fusion.

We have previously reported that Sly1 had no effect on binding of Bet1 to Sed5 using recombinant protein without a transmembrane domain(Kosodo et al., 1998) (K.Y.,N.Y., A.H. and Y.K., unpublished). Recently, it has been reported that the transmembrane domain of syntaxin 1 is critical for its proper interaction with other proteins (Lewis et al.,2001). Actually, Sly1 did not have an effect on the binding after membrane solubilization by Triton X-100 in our cell free system (data not shown). The lipid bilayer and transmembrane domain of SNAREs therefore are probably indispensable for the correct SNARE complex formation and functional role of Sly1. To confirm this, we have shown for the first time the direct role of Sec1/Munc18 family in trans-SNARE complex assembly by a novel in vitro assay system using recombinant Sly1 proteins and intact SNAREs with a transmembrane domain embedded in the membrane in the yeast lysate. This method will be applied to address the role of other fusion-mediating proteins in yeast ER-Golgi vesicle transport such as Uso1, Ypt1, Yip1, Yif1, Sec22, Bos1 or Ykt6 in SNARE complex formation with Sly1 function.

The complex containing Sed5, Bet1, Sly1 as well as Bos1 and Sec17 was discovered in the detergent extract of sec18-1 mutant cells at the nonpermissive temperature (Søgarrd et al., 1994). Although this 20 S complex was originally regarded as a docking complex, it is now considered as a remnant of the fusion machinery. Dissociation of this cis-SNARE complex by the ATPase Sec18 is a pre-requisite for activation and priming of SNAREs to be used in the next round of membrane fusion. Our present data confirmed that a complex containing Sed5, Bet1 and Sly1 exists in the normally growing wild-type yeast cell(Fig. 3), which is also consistent with our previous results using a binding experiment with recombinant proteins (Kosodo et al.,1998). After submission of our manuscript, Peng and Gallwitz reported the formation of a complex containing the Sec22/Sed5/Bos1/Bet1 SNARE complex, Sly1 and Sec17 by mixing recombinant proteins(Peng and Gallwitz, 2002). They suggested that the specificity of the SNARE complex is determined by protein-protein interaction of the components. However, the SNARE complex assembly occurred extremely slowly as it took hours, and importantly the kinetics were not influenced by the presence or absence of Sly1. These apparent contradictions with our present data are probably caused by the absence of membrane in their system. The kinetics of our system are closer to the process occurring in the cell, and the trans-SNARE complex assembly did not occur in our system after the addition of Triton X-100. It is likely that Sly1 helps to assemble the trans-SNARE complex when the SNAREs have transmembrane domains and are anchored to the membrane.

The Sec-1 family may function at two different stages. The Sec1 family protein may act upon t-SNAREs before SNARE complex formation followed by the interaction of v-SNARE and t-SNARE. In this case, the Sec1 family protein would change or fix t-SNAREs into an effective conformation. The other is that the Sec1 family protein may bind to t-SNARE after SNARE complex formation, as is the case in yeast exocytosis. It has been reported that yeast Sec1 does not bind to a single Sso1 molecule, a t-SNARE of plasma membrane, but binds to the v-t-SNARE complex (Grote et al.,2000). We suggest that Sly1 acts before SNARE complex formation because the amount of Sly1 binding to Sed5 is almost the same before and after SNARE complex formation in our cell-free assay system(Fig. 5A, lane 1 or 3). Probably, Sly1 binds to Sed5 before SNARE complex formation, which would,consequently, induce Sed5 into an efficient binding state for Bet1.

SNAREs themselves are the minimal fusion machinery, and trans-SNARE complex formation drives membrane fusion. Thus formed cis-SNARE complexs must be dissociated by the action of NSF/Sec18 ATPase in combination with SNAP/Sec17 before they are used in the next round of membrane fusion. Our data(Fig. 6) implies that the dissociation of cis-SNARE complex is a pre-requisite for stimulation of trans-SNARE complex assembly by Sly1.

SNAREs are considered to be in a cycle of assembly and disassembly in the cell. Solubilization of membrane by detergent stopped trans-SNARE complex assembly, although disassembly could occur if ATP was supplied. The amount of Bet1-Sed5 complex reflects the amount of SNARE complex present at the time of the detergent extraction. During this study, we found that the amount of Bet1 in the Sed5 immunoprecipitate was significantly less when Sly1 is fully active than when it is less active [the sly1^ts mutant was compared with the parent, and the sly1^ts/vector was compared with the sly1^ts/SLY1 plasmid (K.Y., N.Y., A.H. and Y.K., unpublished)]. Apparently, the amount of Bet1-Sed5 complex in detergent extract decreases when Sly1 is more active. So, Sly1 may stimulate disassembly of the cis-SNARE complex by the action of NSF/Sec18 ATPase as well as the assembly of trans-SNARE complex in the presence of the membrane. To examine this biochemically, it will be necessary to develop a cell-free system in which SNAREs are at a specific stage of the assembly/disassembly cycle so that the reaction can be started simultaneously. Genetic data may support the idea that Sly1 may be involved in both assembly and disassembly of SNARE complex. Synthetic lethality of mutations generally suggests that processes that the mutant gene products concern are closely related. When the sec18-1 and sly1^ts mutants were crossed and the progeny was processed for tetrad analysis, the cells that have both mutant alleles did not grow even at the permissive temperature for each single mutation (K.Y., N.Y., A.H. and Y.K.). Thus, the mutant alleles sec18-1 and sly1^ts were synthetic lethal.

Bryant and James recently reported that Vps45, one of the yeast Sec1-family proteins, is essential for ternary SNARE complex formation(Bryant and James, 2001). During vesicle transport between the late Golgi and the endosome, the t-SNARE Tlg2 and its binding partners Tlg1 and Vti1 form a ternary complex. In the absence of Vps45, which binds to Tlg2, Tlg2 suffers rapid degradation, and even if the degradation is prevented through abolition of proteasome activity Tlg2 cannot form a ternary complex. Bryant and James suggest that Sec1 family proteins have chaperone-like activity on t-SNAREs and play an essential role in the activation process that allows t-SNARE to participate in ternary complex formation. Although their observations are mainly made on the basis of immunoprecipitation from cell lysates of null mutants, they are essentially consistent with our results in that Sec1 family proteins enhance ternary SNARE complex formation.

There are four Sec1 family genes in S. cerevisiae(Aalto et al., 1992). Sec1 acts in vesicular transport to the plasma membrane, Vps33 or Vps45 to the vacuole or endosome and Sly1 to the Golgi apparatus(Halachmi and Lev, 1996). Although Sec1 family proteins have no transmembrane domain, a part of them localizes on the organelle membranes as a part of protein complexes. Vps33 is a component of `C-Vps complex' on the vacuolar membrane with Vps11, Vps16,Vps18, Vps39 and Vps41 (Wurmser et al.,2000). Vps45 interacts with Vps21 and Vac1 and localizes on the endosomal membrane (Tall et al.,1999). As shown in Fig. 2, in spite of different binding affinities for Sed5, both Sly1 and Sly1^ts, Sec1 proteins similarly localize on the membrane. This suggests that Sly1 localization is not only dependent on Sed5, but also that other factors have a crucial role. Sly1 might be a part of a certain protein complex, as other Sec1 family proteins are, and might settle on Golgi membrane through this complex.

Our findings with Sly1 and observations of other investigators suggest that Sec1 family proteins generally localize on the target membranes independently from the SNARE proteins that they bind to. SNARE proteins cycle between organelle membranes. For example, Sed5 mainly localizes on the Golgi membrane,but it cycles between the ER and Golgi membranes (Wooding et al., 1998;Cho et al., 2000). There are probably some mechanisms to induce SNARE-dependent membrane fusion and to mix the content of cargo exactly on the Golgi membrane. When Sly1 regulates trans-SNARE complex formation as we have described, Sly1 should localize on the area where membrane fusion is expected to occur. In other words, the area of membrane fusion would not be determined without the correct localization of Sec1 family proteins on the proper membranes.

We thank K. Mizuta for providing the sly1^ts strain. We thank S. Kagiwada, J. Rothman and S. Ferro-Novick for the antibodies. We thank A. Takatsuki and H. Hashimoto for contributing to the discussion. This work was supported by a grant-in-aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan, grants for`Bioarchitect Research' from the Institute of Physical and Chemical Research(RIKEN), a grant from the Kato Memorial Bioscience Foundation (to Y.N.), a grant from the Nagase Science and Technology Foundation (to K.Y.) and a grant from the Japan Society for the Promotion of Science (JSPS) (to Y.K.).