INTRODUCTION

The N-ethylmaleimide-sensitive fusion protein (NSF) and the soluble NSF attachment proteins (SNAPs) are required for vesicular transport at multiple steps (1-6). However, transit through the secretory pathway requires a mechanism to ensure the fidelity of transport vesicles docking to their correct target membrane. The known molecular machinery mediating this process consists of SNAP receptors (SNAREs) localized to vesicles (v-SNAREs)¹ that form a specific match with t-SNAREs on target membranes (SNARE hypothesis) (7, 8). The first members of this protein family were initially described in the neuronal synapse. Vesicle-associated membrane protein (VAMP)/synaptobrevin (9) and synaptotagmin I (10) on synaptic vesicles form a complex with syntaxin 1 (11) and SNAP-25 (12) localized on the presynaptic plasma membrane (7, 11, 13). Paired v-t-SNARE complexes form a scaffold for the high affinity binding of SNAP and NSF. (8, 14, 15). ATP hydrolysis by NSF causes disassembly of the SNARE complex, a prerequisite for membrane fusion to occur.

The SNARE hypothesis predicts that v- and t-SNAREs are compartment-specific and that they bind each other directly and specifically. The interaction of synaptic SNAREs has been tested and studied in detail (8, 13, 16-18). Genetic and biochemical analyses in the yeast Saccharomyces cerevisiae has suggested specific localization and functionality. Compartment-specific v-SNAREs and t-SNAREs have been identified, and their inactivation results in distinct secretion-deficient phenotypes (see Ref. 19 for review). Golgi to plasma membrane transport is mediated by Snc1p and Snc2p (20), two homologous v-SNAREs localized to post Golgi vesicles that interact specifically with the t-SNAREs Sso1p and Sso2p (21), located on the plasma membrane. Inactivation of Snc1p and Snc2p or Sso1p and Sso2p results in the accumulation of free 80-100 nm post Golgi transport vesicles. At the ER-Golgi transport step, the VAMP homologues Sec22p, Bos1p, and Bet1p (22-24) are required for the docking and/or fusion of ER-derived transport vesicles with the Golgi both in vitro and in vivo (25, 26). Although controversial data exist about the localization of Bet1p in transport vesicles (26, 27), Sec22p has been both found in COPI and COPII vesicles (28), as well as ER-Golgi transport vesicles devoid of cargo (29). These three ER-Golgi v-SNAREs interact with the t-SNARE Sed5p, which has been localized to the cis side of the Golgi stack in animal cells (30, 31). Formation of v-t-SNARE complexes is tightly regulated and controlled by several proteins, including small GTP binding proteins belonging to the Rab family and members of the Sec1p protein family (19, 32-35). v-t SNARE complexes do not form when these proteins or the SNAREs themselves are inactivated (36-38). In contrast, well defined stoichiometric complexes accumulate when fusion is blocked by a temperature-sensitive sec18 mutant in vivo. This method was employed to immunoisolate distinct SNARE complexes containing Sed5p, utilizing an anti Sed5p antibody. These complexes contained Sec17p (yeast -SNAP), Sly1p (a Sec1p family member), Sed5p, the ER-Golgi v-SNAREs Bos1p, Bet1p, and Sec22p, and three novel proteins, p28, p14, and p26 (36).

Here we report the identification of the putative SNAREs p28, p14, and p26 and the detailed characterization of p26, a unique SNARE with close homologues throughout the eukaryotic lineage, each characterized by an isoprenylation signal.

EXPERIMENTAL PROCEDURES

Yeast strains (Table I) were routinely grown in rich media (1% yeast extract, 2% peptone, and either 2% dextrose-YPD, 8% dextrose-YPD8, or 2% raffinose and 0.5% galactose-YPRG) for wild type cells, MSY60, and JMY89. Synthetic complete media included 0.67% yeast nitrogen base (Difco) and 2% dextrose (SCD) or 2% raffinose, 0.5% galactose (SCRG), and the appropriate auxotrophic supplements (Bio101).

Table I. Yeast and bacterial strains Strain name Genotype Source W3031A MATa ade2-1 leu2-3,112 ura3-1 trp1-1 his3-11,15 S. Esposito W3031B MAT ade2-1 leu2-3,112 ura3-1 trp1-1 his3-11,15 S. Esposito RSY271 MAT his4-619 ura3-52 sec18-1 R. Schekman RSY279 MAT his4-619 ura3-52 sec22-3 R. Schekman RSY955 MAT leu2 sec32-1(bos1ts) R. Schekman MSY41 MATa/ ade2-1/ade2-1 leu2-3,112/leu2-3,112 ura3-1/ura3-1 trp1-1/trp1-1 his3-11,15/his3-11,15 YKT6::LEU2/YKT6 This study MSY54 MAT ade2-1 leu2-3,112 ura3-1 trp1-1 his3-11,15 pMS154(GAL1-10, YKT6, URA3, CEN6) This study MSY60 MAT ade2-1 leu2-3,112 ura3-1 trp1-1 his3-11,15YKT6::LEU2 pMS220(GAL1-10, YKT6/GOS1TM, URA3, CEN6) This study JMY14 MAT ade2-1 leu2-3,112 ura3-1 trp1-1 his3-11,1 pMS237(GAL1-10, mycYKT6, URA3, CEN6) This study JMY15 MATa ade2-1 leu2-3,112 ura3-1 trp1-1 his3-11,1 pMS238(GAL1-10, mycYKT6-C196S-C197S, URA3, CEN6) This study JMY64 MATa/ ade2-1/ade2-1 leu2-3,112/leu2-3,112 ura3-1/ura3-1 trp1-1/trp1-1 his3-11,15/his3-11,15 YKT6::LEU2/YKT6 This study JMY89 MAT ade2-1 leu2-3,112 ura3-1 trp1-1 his3-11,15 YKT6::LEU2 pJM172(GAL1, Hs_YKT6, URA3, 2 µm) This study XL1-Blue F::TN10 proA+B+ lacIq (lacZ)M15/recA1 endA1 gyrA96 (NaIr) thi hsdR17 (rkmk+)supE44 relA1 lac TG-1 F traD36 laclq [lacZ]M15 proA+B+/supE [hsdM-mcrB]5 (rkmkMcrB) thi (lac-proAB)

One copy of the YKT6 gene (nucleotides 1-600) was replaced with a 2.1-kilobase fragment encoding the LEU2 gene in the W303 diploid strain. W303 was transformed with a PCR fragment containing the LEU2 gene flanked by 260 nucleotides of 5-nontranslated DNA and 900 nucleotides of 3-nontranslated DNA to generate MSY41. Integration at the YKT6 locus was confirmed by Southern blotting.

pRRY20 is a derivative of pRS316 (39) with an 0.8-kilobase EcoRI-BamHI fragment containing the GAL1-GAL10 promoter. pMS154 contains the YKT6 open reading frame driven by the GAL1 promoter with 400 base pairs of YKT6 3-untranslated sequence. It was made by inserting the YKT6 gene, which was cloned by PCR using the oligonucleotides YKT6-1 and YKT6-4 (Table II) and yeast genomic DNA (CLONTECH), into the BamHI SacII sites of pRRY20. pMS156, which expresses Gos1p, was generated in a similar manner utilizing oligonucleotides P28MYC and P28SAC to produce the GOS1 gene and 100 base pairs of 3-untranslated sequences by PCR. pMS167, was made by inserting a 0.8-kilobase BglII-SacII PCR fragment (oligonucleotides P28MYC2 and P28SAC), encoding myc-Gos1p, into BamHI/SacII cut pRRY20. pMS219, which expresses the Ykt6p/Gos1TM(204-223) chimera, was made by fusing the cytoplasmic domain of Ykt6p to the transmembrane domain of Gos1p. A BamHI-MluI PCR fragment (oligonucleotides YKT6-1 and P26P28) encoding amino acid 1-191 of Ykt6p was cloned into BamHI/MluI-cut pMS156. A natural MluI site occurs just preceding the Gos1p transmembrane domain. pMS222, expressing the Ykt6p-C196S-C197S mutant, was constructed by inserting a 0.6-kilobase BamHI-SacII PCR fragment (oligonucleotides YKT6-1 and P26CC-SS) into pRRY20. The following constructs were made to express amino-terminally myc-tagged proteins: pMS237 (myc-Ykt6p/Gos1TM) and pMS238 (myc-Ykt6p-C196S-C197S) were constructed by inserting the BamHI-SacII fragments of pMS219 or pMS222, respectively, into BamHI/SacII-cut pMS167. pMS130, which expresses His₆-Ykt6p in bacteria, was constructed by ligating a PCR product (oligonucleotides YKT6-1 and YKT6-2) containing the YKT6 ORF into pQE30 (Qiagen).

Table II. Oligonucleotides Name Oligonucleotide sequence p28SAC CCC CGC GGT TTT ACT CCA ATT AGA AGA p28MYC CCC AGA TCT GAA CAA AAA CTT ATT TCT GAA GAA GAC TTG GGA TCC ATG AGC TCA CAA CCG TCT TTC P28MYC2 CCC AGA TCT ATG GAA CAA AAA CTT ATT TCT GAA GAA GAC TTG GGA TCC ATG AGC TCA CAA CCG TCT TTC KOP26-1 GGA GAA TTC GAC AGA TTA CCA TGA KOP26-2 CGC GGA TCC ATC CAC ACA CTA TTT TCA GCA KOP26-3 CCG CCG CGG TTT TTA AAC CAA AAT TCG GCT CCT P26-P28 CCC GAA TTC ACG CGT TTT TCG AAT TGG ATT TTT TAG CTT G P26 CCSS CCC CCG CGG CAT CTA CAT GAT GAT GGA AGA CGA ATT GGA T YKT6-1 GAG GAT CCA TGA GAA TCT ACT ACA TCG GT YKT6-2 CCG GAT CCA AGC TTA TGC ATC TAC ATG ATG ATGC YKT6-4 TCC CCG CGG TTT CTT ACA CTA ATT ATA LEU2-BAM CGC GGA TCC CAA CAA TTA CAT CAA AAT hp26-R TTA GGG CCC TCA CAT GAT GGC ACA GCA TGA hp26-F TGC TCT AGA ATG AAG CTG TAC AGC CTC AGC

The human YKT6 gene was cloned by PCR (oligonucleotides hp26-R and hp26-F) from a human pancreatic cDNA library (CLONTECH), put into pBluescript KSII⁺ (pJM167), and sequenced by standard dideoxy sequencing with Sequenase (Stratagene). The HsYKT6 gene was passaged through pGEM-3Z (Promega) to pickup restriction sites (pJM168), and then a 630-base pair fragment was cut out of pJM168 as a HindIII-SacI fragment and ligated into HindIII-SacI cut pYES2 (Invitrogen), resulting in pJM172. This construct expresses HsYKT6, under GAL1 control, on a multicopy 2-µm vector. A control vector was generated (pJM175) by placing the 650-base pair yeast YKT6 gene in pYES2.

A 500-ml pre-culture of JMY89 was grown overnight in YPRG media. At time 0, 200 A₆₀₀ units of cells were harvested by centrifugation at 1,500 × g for 5 min. One liter of YPRG or YPD8 (8% glucose) was inoculated at 0.1 A₆₀₀/ml. Aliquots were taken at 2-h intervals to measure cell density. Additionally, 20 A₆₀₀ units were harvested, washed in water, and frozen as cell pellets in liquid nitrogen. At 10 and 24 h an additional 10 A₆₀₀ units were recovered for electron microscopic analysis. Glass bead extracts were prepared from the 2-h time points, and equal amounts of the extracts were analyzed with the anti-carboxypepidase Y monoclonal antibody 10A5-B5 (Molecular Probes) at a dilution of 1:1000.

The Multiple Sequence Alignment was performed using PILEUP from the University of Wisconsin Genetics Computer Group sequence package (40), and the shading was done using BOXSHADE.² The sequences represented in Fig. 1 are: Sc_Ykt6p is S. cerevisiae chromosome XI ORF YKL196c (Z28196); Hs_Ykt6p represents several overlapping human ESTs (THC111483) (H23796, H₂O958, H18232, R76979, H23795, H18270, and H40165); Ce_Ykt6p is amino acids 519-720 of the ORF B0361.8 from the cosmid B0361 (U00031); Ec_Ykt6p is Euplotes crassus ORF1 (M73025); At_Ykt6p is an EST from Arabidopsis thaliana cDNA clone 151D5T7 (T76779), Mm_Ykt6p is an EST from Mus musculus cDNA clone 403610 (W82343); and Rn_Ykt6p is EST110382 (H33932) from Rattus norvegicus.

Recombinant, His₆-tagged Ykt6p was purified from inclusion bodies in Escherichia coli (XL1-Blue) expressing pMS130 using nickel-nitrilotriacetic acid-agarose according to the manufacturer's instructions. Two New Zealand White Rabbits (215 and 216) were immunized by subcutaneous injection of 200 µg each of His₆-Ykt6p to generate anti-Ykt6p antisera.

Proteins separated by SDS-PAGE were electroblotted onto nitrocellulose, and the visualized bands were processed for internal amino acid sequence analysis as described (41). Briefly, membrane bound proteins were digested in situ with trypsin, and the resulting peptides were separated by narrowbore reverse phase high pressure liquid chromatography. Selected peptides were then subjected to chemical microsequencing and matrix-assisted laser desorption ionization mass spectrometry, also as described previously (42, 43).

The 14-kDa band (p14) in the Sed5p immunoprecipitate from extracts of sec18-1 at the nonpermissive temperature was identified by chemical sequencing. The peptides LATFR, NINQEIGD, AVSDSXVINQMTDSLGXMFT (where X indicates no positive identification of an amino acid) are perfect matches to the published Sft1p sequence (44).

Five tryptic peptides from the 28-kDa band (p28) in the Sed5p immunoprecipitate were also analyzed by chemical sequencing/MALDI mass spectrometry. The sequences YSTFAQTTSXEQ, XQFHXQSNVLNXANN, XKEILQDH, IPGVNQLIMK, and LIXQAXET are nearly identical to those present in the predicted 25.4-kDa protein (223 amino acids) translated from the ORF YHL031c.³ Mass analysis on the last peptide gave an m/z value that was within 0.04% of the calculated theoretical molecular mass [MH⁺] for the predicted YHL031c tryptic peptide, which contains the limited sequences, so confirming the identity of the entire peptide.

Wild type cells (W3031A), cells expressing myc-Ykt6p-C196S-C197S (JMY14), or cells expressing myc-Ykt6p/Gos1TM (JMY15) were grown to mid to late log phase in YPD (W3031A) or SCRG with the appropriate auxotrophic supplements. 50 A₆₀₀ units of cells were harvested, and extracts were prepared according to Hardwick and Pelham (46). 100 µl of each supernatant was aliquoted into four polyallomer microultracentrifuge tubes. These aliquots were extracted for 1 h on ice with 900 µl of lysis buffer alone, 0.1 M Na₂CO₃, pH 11.0, or lysis buffer containing either 1 M NaCl or 1% Triton X-100. The extract was separated into membrane and soluble fractions by centrifugation at 60,000 rpm (157,000 × g_max) in a Beckman TLA100.3 rotor for 15 min at 4 °C. Centrifugation at 70,000 rpm (213,000 × g_max) for 1 h produced similar results (data not shown). The pellets were resuspended in the same volume and composition as the supernatants, and all samples were trichloroacetic acid precipitated. Equal percentages of the supernatant and pellet were resolved by SDS-PAGE, transferred to nitrocellulose, and immunodecorated with the anti-Ykt6p antisera (1:2000) or the 9E10 monoclonal antibody (1:500) (47) for myc-tagged constructs.

Triton X-114 (Sigma) was precondensed according to the method of Bordier (48). Cells were spheroplasted as described previously (49). The spheroplast pellet was washed with 1 M sorbitol, and 25 A₆₀₀ units were resuspended in 2% (v/v) Triton X-114 in 10 mM Tris-Cl, pH 7.4, 150 mM NaCl and then extracted on ice for 2-2.5 h. Detergent-insoluble material was removed by a 16,000 × g centrifugation at 4 °C for 10 min. The clarified supernatant was transferred to a new microfuge tube and clouded at 37 °C for 10 min, followed by a 10-min 16,000 × g centrifugation at room temperature. The aqueous phase was transferred to a new tube and the aqueous, and detergent phases were back-extracted three times according to the method of Brusca and Radolf (50).

10 A₆₀₀ units of yeast cells per sample were isolated by centrifugation, washed with water, and resuspended in 1 ml of 2% potassium permanganate for 45 min at room temperature. Following this incubation, the cells were pelleted and washed three times with 1.0 ml of water, and 1.0 ml of 70% ethanol was layered onto the pellet. This pellet was further dehydrated in a grades series of ethanol and embedded in epon resin. Thin sections (~60 nm) were treated with 5% uranyl acetate followed by 0.4% lead citrate and visualized on a JOEL 1200EX transmission electron microscope operating at 80 kV.

RESULTS

The identities of p28, p26, and p14 were determined by microsequencing and mass spectrometry. The isolated Sed5p complex was resolved by SDS-PAGE and blotted on to nitrocellulose. p26 (Ykt6p) was previously reported to be encoded by the ORF YKL196c on chromosome XI (36). Microsequencing and mass spectrometry of tryptic peptides derived from the 28-kDa band revealed that p28 is identical to the 223-amino acid protein with a predicted molecular mass of 25,394 Da encoded by the ORF YHL031c on chromosome VIII. This protein, named Gos1p (Golgi SNARE) is a SNARE protein with a carboxyl-terminal membrane anchor.⁴ Similarly, peptide sequence obtained from the 14-kDa protein led to the identification of p14 as a 97-amino acid (10,960 Da) protein located on chromosome XI. p14 is identical to the product of the SFT1 gene, a gene identified independently by Pelham and colleagues as a high copy suppressor of a sed5-1^ts allele (44).

BLAST searches of various data bases revealed that Ykt6p has homologues in many species including man and that all possess a high degree of sequence similarity (Fig. 1A) (51). The yeast protein is 47% identical in amino acid sequence to the human protein, HsYkt6p, whereas the Caenorhabditis elegans ORF is 57% identical to HsYkt6p. Hydropathy analysis by the method of Kyte and Doolittle (52) predicts that each of these proteins is very hydrophilic in nature (Fig. 1B) and does not possess a transmembrane spanning domain. However, all of the Ykt6p homologues contain a carboxyl-terminal CAAX motif suggesting that these proteins are post-translationally modified by the addition of a 15- or 20-carbon isoprenoid. Ykt6p from S. cerevisiae terminates in the sequence CIIM, which should specify the addition of a farnesyl moiety (53). This characteristic is peculiar for a SNARE molecule, although post-translational addition of hydrophobic moieties, in particular palmitoylation, has been reported for other SNAREs (54-56). Sequence comparisons of Ykt6p with defined SNAREs show a significant similarity to the SEC22 gene product as well as a lower but statistically significant similarity to the late acting v-SNAREs Snc1p and Snc2p.

To address the function of Ykt6p, one copy of YKT6 was replaced with the LEU2 gene in the diploid strain W303 (Fig. 1C). Sporulation of this strain (MSY41) resulted in only two viable progeny, both of which were Leu. This result demonstrates that YKT6 serves an essential function either in germination, vegetative growth, or both. The YKT6 strain could be recovered by an extra-chromosomal copy of the YKT6 gene expressed from the galactose inducible GAL1-10 promoter. Additionally, an amino-terminal epitope-tagged copy of YKT6 was completely functional (Table III).

Table III. Growth of YKT6 deletion strains recovered with various plasmid Strain Protein Carboxyl terminus Rescue of null strain Doubling time in YPRGa min W3031A 150  (3) W3031B 130  (2) JMY89 HsYkt6p KTARKQNSCCAIM ++ 196  (3) MSY54 Ykt6p KQAKKSNSCCIIM +++ 142  (3) MSY56 myc-Ykt6p KQAKKSNSCCIIM +++ ND Ykt6p-C196S-C197S KQAKKSNSSSIIM   ND MSY60 Ykt6p/Gos1TM KQAKKSNSKNAFVLATITTLCILFLFFTW +++ 152  (4) a Number of replicates; ND, not determined.

The human YKT6 gene (HsYKT6) was identified by comparisons of the yeast sequence with the data base of human ESTs. The complete sequence of the open reading frame encoding human Ykt6p was obtained from the data base by splicing together several overlapping ESTs. The human cDNA was cloned by PCR from a human pancreatic cDNA library, and its sequence was confirmed. The HsYKT6 gene was put under the control of GAL1 promoter in a multicopy vector and transformed into the YKT6/YKT6 heterozygous diploid strain JMY64. All of the resulting spores from this strain were viable when grown on rich media with galactose as a carbon source. Analysis of these progeny showed that all of the Leu⁺ spores were also Ura⁺, confirming that viability was plasmid-dependent. The HsYKT6 complemented YKT6 strain (JMY89) grew approximately 25% slower in liquid culture than wild type (Table III). This is the first documented case of a human SNARE functionally complementing its yeast counterpart.

The unique feature distinguishing Ykt6p from other described SNAREs is the absence of a transmembrane spanning domain and the presence of a carboxyl-terminal CAAX box sequence (CCIIM) specifying isoprenylation. Because isoprenylation of an otherwise hydrophilic protein seems to be the only mode to anchor Ykt6p to the membrane, we determined the functional importance of this post-translational modification. To this end, a mutant was generated that eliminates the isoprenylation signal by mutating both cysteines at residues 196 and 197 within the CCIIM sequence to serine. This construct, Ykt6p-C196S-C197S, could not rescue the YKT6 deletion (Table III). Expression of this construct was confirmed in wild type cells (Fig. 2 and data not shown). To determine if a bona fide transmembrane domain could functionally substitute for isoprenylation, we replaced the CCIIM at the carboxyl terminus of Ykt6p by the membrane spanning region of Gos1p. Interestingly, this hybrid, termed Ykt6p/Gos1TM, also rescued the YKT6 deletion (Table III).

Next, we determined the membrane association of Ykt6p, Ykt6p-C196S-C197S, and Ykt6p/Gos1TM. For this purpose these proteins were expressed in a wild type background (W3031B), some with an amino-terminal myc-tag to facilitate protein detection. Yeast extracts from the different strains were treated with buffer, high salt, carbonate, or Triton X-100 and separated into membrane and soluble fractions by ultracentrifugation. Fig. 2 shows that wild type Ykt6p was unusual in that greater that 50% of the protein was not membrane-associated when extracts were prepared in buffer alone. This soluble material was not due to harsh preparation of the extract because gently osmotically lysed spheroplasts displayed similar behavior. However, the membrane-associated material behaved as an integral membrane protein in that it remained membrane-associated after 1 M NaCl or 0.1 M sodium carbonate pH 11.0 treatment. As expected, Ykt6p-C196S-C197S was completely soluble under all conditions tested, whereas Ykt6p/Gos1TM behaves as an integral membrane protein, being released only by Triton X-100 treatment (Fig. 2).

Finally, we wanted to determine the prenylation status of the soluble pool of Ykt6p. In yeast, in contrast to mammalian cells, it is currently not possible to test for the presence of an isoprenyl group by growing yeast in presence of radiolabeled isoprenoid precursors (57). Therefore we chose to assay this modification based on an increased hydrophobicity with the attached prenyl group. Total spheroplast extracts were made using the detergent Triton X-114. The cleared extract was phase partitioned, and the aqueous and detergent phases were resolved by SDS-PAGE and analyzed by immunoblotting with an anti-Ykt6p antibody. Fig. 3 shows that the majority of Ykt6p partitions into the Triton X-114 detergent phase. To demonstrate that the presence of Ykt6p in the Triton X-114 detergent phase was a function of the proposed prenylation, wild type cells expressing a myc-tagged Ykt6p-C196S-C197S protein were detergent extracted and phase partitioned. In contrast to the wild type protein, myc-Ykt6p-C196S-C197S partitioned exclusively to the aqueous phase (Fig. 3B). Importantly, soluble Ykt6p, prepared simply by breaking total yeast in the absence of detergent and removing membranes by ultracentrifugation (Fig. 2), is found in the Triton X-114 detergent phase (Fig. 3C), suggesting that the soluble pool is also modified by isoprenylation.

These experiments strongly indicate that wild type Ykt6p is anchored in membranes by a lipid attached to the CAAX box cysteine(s). Soluble Ykt6p also contains this modification. Our results also show that membrane anchoring is necessary for Ykt6p function and that replacing the lipid-anchor present in the wild type protein with a proteinaceous membrane anchor will support the essential function of Ykt6p.

The presence of Ykt6p in the isolated Sed5p complex suggests that Ykt6p might be involved in ER-Golgi or intra-Golgi transport. To determine the site of Ykt6p action, Ykt6p was depleted in the strain MSY54. This strain contains a disrupted genomic copy of YKT6 and a plasmid borne YKT6 gene driven by the GAL1-10 promoter. Glucose strongly represses transcription from the GAL1-10 promoter (58), and the expression level of YKT6 can be manipulated by growth in different carbon sources. Surprisingly this strain continued to grow unabated in glucose containing media with glucose concentrations up to 8% maintained for several weeks. This phenomena is likely attributable to small amounts of transcription of the gene in glucose, resulting in the production of sufficient Ykt6p to allow growth. Immunoblots of extracts from wild type cells and MSY54 grown in glucose show that Ykt6p is reduced at least 10-fold in MSY54 but is still present in detectable amounts (data not shown).

Because it was not possible to reduce Ykt6p to a level that would slow cell growth, we replaced the plasmid-borne Ykt6p with its human homologue, assuming that the homologue might work less effectively than Ykt6p itself. Indeed this strain (JMY89) ceased growing when shifted to media containing glucose as the only carbon source as shown in Fig. 4. To determine whether depletion of HsYkt6p function influences transport along the secretory pathway, the processing of a well characterized vacuolar protein, carboxypeptidase Y (CPY), was analyzed (Fig. 4). CPY is translocated into the ER, where it receives core oligosaccharides generating the p1 form, transits the Golgi where the core sugars are elongated yielding the p2 form and is proteolytically processed in the vacuole to the mature form. When HsYkt6p is expressed in the presence of galactose, the majority of CPY is found in its mature form, and small amounts of both p1CPY and p2CPY are detectable, consistent with the observed reduced growth rate. During HsYkt6p depletion induced by glucose, the p1 form of CPY steadily accumulates. A small amount of p2CPY also appears to persist until it is masked by the increased p1CPY signal. This may suggest that additional transport steps at the level of the Golgi are also affected by the loss of Ykt6p function. At later time points, mature CPY also appears to migrate slightly faster in SDS-PAGE. It is unclear why this occurs but might be explained by partial proteolysis or changed processing of the covalently attached sugar side chains of CPY.

Next, the HsYkt6p-depleted cells were analyzed morphologically by electron microscopy. Fig. 5 shows wild type cells grown in 8% glucose (W3031A; Fig. 5A), the YKT6 strain expressing HsYkt6p in galactose (JMY89; Fig. 5B), and HsYkt6p-depleted cells grown in 8% glucose (JMY89; Fig. 5, C and D). There is a marked accumulation of 50 nm diameter transport vesicles in the HsYkt6p-depleted cells (Fig. 5, C and D) suggesting that Ykt6p is required vesicle docking and/or fusion. These vesicles seem to be dispersed throughout the cytoplasm and are not clustered as seen in a sec18-1^ts or a sec17-1^ts strain (45).⁵ Another morphological consequence of depleting HsYkt6p is a striking accumulation of ER membranes. This phenotype is demonstrated in Fig. 5C, where a large network of ER membranes continuous with the nuclear envelope is observed. In addition, a generalized exaggeration of undefined membranes and vesicular structures of various sizes are visible. Fragmentation of the vacuole, a phenotype often associated with a secretory defect, is also frequently observed. Under permissive growth conditions in the presence of galactose, the morphology of the YKT6 strain expressing HsYkt6p is essentially the same as in wild type cells (Fig. 5B). Occasionally, exaggerated membranes are seen in some cells in the presence of galactose, which would be consistent with the partial complementation of the human gene.

Finally, we asked if Ykt6p could suppress the phenotype of certain temperature-sensitive secretion mutants. Overexpression of Ykt6p suppressed the temperature-sensitive phenotype of two SEC22 alleles, sec22-1 (44), and sec22-3 as well as a temperature-sensitive Bos1p mutant (sec32-1) (data not shown); both genes encode v-SNAREs found on ER-derived transport vesicles. It suppressed neither mutants in sec12, the GTP-exchange factor for Sar1p involved in budding of COPII-coated transport vesicles, nor mutants in sec18, the NSF homologue, which is involved in vesicle consumption at several intracellular transport steps (data not shown). This provides indirect evidence for an involvement of Ykt6p at the level of the SNARE complex in ER-Golgi transport.

DISCUSSION

Vesicular traffic through the secretory pathway requires specific pairwise interaction of targeting molecules on vesicles and on the destination membrane to maintain the cellular compartmentalization and directionality of transport. This recognition mechanism is provided by cognate v-SNARE and t-SNARE interactions. In this report we characterize a new v-SNARE, Ykt6p, involved in ER-Golgi transport, increasing the number of potential v-SNAREs at this transport step to four: Bet1p, Bos1p, Sec22p, and Ykt6p. Ykt6p has several hallmarks of a SNARE molecule. It was isolated from an assembled v-t-SNARE complex containing Sed5p, many other SNAREs, and Sec17p and shows significant sequence homology to members of the VAMP/synaptobrevin family like Sec22p. Additionally, Ykt6p, present in a detergent extract of total yeast membranes, interacts directly or as a part of a v-t-SNARE complex with recombinant GST-Sec17p (data not shown). A striking feature distinguishing Ykt6p from other SNAREs is the presence of a CAAX box providing a signal for isoprenylation and thereby mediating membrane attachment. In all other known v-SNAREs, hydrophobic peptide sequence function as membrane anchors. In this respect, all of the Ykt6p homologues, ranging from yeast to man, are unique SNARE molecules.

A chromosomal deletion of YKT6 is lethal, but it can be rescued by expression of wild type Ykt6p, amino-terminally myc-tagged Ykt6p, the human Ykt6p homologue, and a Ykt6p chimera, whose isoprenylation signal has been substituted by a proteinaceous membrane anchor. Ykt6p is present at ~0.05% of total protein, but cells are completely viable at levels at least 10-fold below this under the conditions we tested. This would indicate that wild type yeast can adapt to low Ykt6p levels. However, Ykt6p function is still clearly required for viability, because these cells could not lose the URA3-based plasmid on 5-fluroorotic acid plates after many generation in glucose media.

Interestingly, Ykt6p was found in two cellular pools, one in the cytoplasm and another associated with membranes. Triton X-114 phase partitioning would suggest that both the membrane bound and soluble Ykt6p are prenylated, raising the question how the hydrophobic group is masked in the cytosol. Currently, we cannot answer this question, but our data clearly indicate that the membrane-associated pool of the Ykt6p represents the functional form. Mutant Ykt6p, which lacks the isoprenylation signal, cannot recover the YKT6 deletion. In addition, the chimera Ykt6p/Gos1pTM, containing the Gos1p membrane spanning region instead of the isoprenylation signal, was functionally active, and a cytoplasmic pool of this chimera was not detectable. It was not possible to localize the membrane-bound population of Ykt6p by immunofluorescence microscopy to a distinct intracellular membrane, because fluorescence due to the cytosolic pool of Ykt6p resulted in too large of a background signal.

Biochemical studies and the presence of Ykt6p in the isolated Sed5p complex indicate that Ykt6p is involved, at the very least, in ER to Golgi transport. These observations suggest a localization within these compartments. Depletion of Ykt6p function stops cell growth and manifests a transport block at the level of the endoplasmic reticulum, as shown by the accumulation of the p1 form of CPY (Fig. 4). Genetic evidence confirms an ER-Golgi function for Ykt6p. Overexpression of Ykt6p suppresses the temperature-sensitive phenotype of two ER-Golgi v-SNAREs, Sec22p (sec22-1 and sec22-3), and Bos1p (sec32-1). Additionally, a temperature-sensitive allele of Uso1p (uso1-1) can be suppressed by all four ER-Golgi SNARE molecules including Ykt6p (38). Morphological data obtained from yeast cells containing reduced levels of the human homologue of Ykt6p, which replaces the endogenous Ykt6p, gave a heterogeneous phenotype. The depleted cells accumulate 50 nm transport vesicle and exaggerated ER membrane further supporting a Ykt6p function in ER-Golgi transport. Similarly vesicle and ER accumulation has been observed in Sed5p-depleted yeast (46). It is not clear if the exaggerated ER is the primary or secondary effect of the transport block. However, it seems likely that vesicles initially accumulate, and when Ykt6p becomes limiting, production of new vesicles is reduced, leading to the ER accumulation. This would be consistent with a mechanism, which couples vesicle generation to v-SNAREs packaging and therefore ensures generation of docking and fusion competent vesicles. It would also be consistent with a selective block in anterograde transport.

We cannot exclude that Ykt6p might also act at transport steps other than ER-Golgi because a cytoplasmic pool of Ykt6p could provide an interacting partner for several SNAREs localized to multiple compartment. Further analysis of Ykt6p localization and distribution will help to determine its precise role in vesicle targeting and fusion.