- Email: [email protected]
Expression of genes encoding vesicular stomatitis and sindbis virus glycoproteins in yeast leads to formation of disulfide-linked oligomers
VIROLOGY
153,150-154
Expression
(1986)
of Genes Yeast DUANZHI
Department
Encoding
Vesicular
Leads to Formation WEN,
MINGXIAO
of Microbiology
Stomatitis
and Sindbis
of Disulfide-Linked
DING,
AND MILTON
Virus Glycoproteins
in
Oligomers
J. SCHLESINGER~
and Immunology, Washingto?z University School of il4edicine, St. Louis, Missouri 6.3110
Received Februav
24, 1986; accepted May 12, 1986
Sacchu~omycea cerevisiae strains transformed with plasmids containing cDNAs coding for the glycoproteins of vesicular stomatitis or Sindbis viruses can be induced to produce large amounts of glycosylated virus glycoproteins. Studies reported here show that these proteins form high molecular weight disulfide-linked oligomers in the yeast endoplasmic reticulum. Oligomers were also found for two genetically altered forms of VSV G; one of these was lacking the membrane anchor domain and the other had the cysteine in the cytoplasmic tail replaced with serine. These oligomers can be separated from the bulk of yeast proteins by brief high-speed centrifugation of yeast extracts prepared by boiling cells with 1% sodium dodecyl sulfate. Treatment with thiol-reducing agents converts the oligomers to soluble monomeric forms, and this procedure leads to a substantial purification of glycoproteins from bulk yeast protein. @ 1986 Academic Press, IX
We have recently developed an efficient system for the expression of Sindbis and vesicular stomatitis virus (VSV) glycoproteins in yeast (1). The cDNAs encoding the VSV glycoprotein (G) and the Sindbis virus structural genes (the 26 S RNA coding for the capsid and glycoproteins El and E2) were fused to the yeast galactose kinase gene promoter (2) and inserted into a yeast Escherichia coli shuttle vector such that expression of the virus glycoproteins was only detectable when cells were grown in the presence of galactose. The virus glycoproteins were glycosylated but not fatty acylated and comprised about 2 to 3% of total yeast protein synthesis. Half of the VSV G protein was detected by surface labeling yeast cells, indicating that some of this protein was transported through the yeast secretory pathway. Cells expressing the Sindbis virus genes produced free capsid and glycosylated forms of El and the E2 precursor, ~62. Thus, the cotranslational proteolytic processing that normally occurs during expression of the Sindbis virus structural genes (reviewed in Ref. (8)) ‘To whom dressed.
requests
for
reprints
0042~6822/86 $3.00 Copyright All rights
Q 1966 by Academic Press, Inc. of reproduction in any form reserved.
should
was found in the yeast cell. However, the post-translational processing of p62 to E2, which takes place in the Golgi vesicles, was not detected in yeast. In addition, only low amounts of the Sindbis virus glycoproteins were transported in yeast. After induction of virus protein synthesis in galactose medium, yeast cells grew very slowly suggesting that formation of the foreign protein was blocking essential yeast cellular activities. In an attempt to determine why these proteins were inhibitory to cell growth, we examined the properties of the virus glycoproteins and report here that these glycoproteins formed disulfide-linked oligomers in the yeast endoplasmic reticulum. We used essentially the same procedures as previously described (1) for inducing and detecting synthesis of virus proteins in yeast strains transformed with plasmids containing the virus cDNAs. Yeast extracts were prepared by treating 35S-methionine labeled cells with 1% SDS and vortexing with glass beads. After boiling, the extracts were diluted and analyzed by immunoprecipitation with virus specific antiglycoprotein antibodies. Extracts from cells carrying the VSV G cDNA showed a single
be ad-
150
SHORT
COMMUNICATIONS
band in SDS-PAGE when p-mercaptoethanol was included in the sample buffer and this band had a mobility identical to authentic VSV G (lanes 1 and 2, Fig. 1). However, the immunoprecipitated VSV G prepared in the absence of reducing agent entered the stacking portion of slab gel but did not move further into the gel (lane 3, Fig. 1). This apparent aggregate of the G protein was not an artifact of the immunoprecipitation procedure; it also appeared in the crude extract of yeast before treatment with antisera (lane 4, Fig. 1). We could show that this high molecular weight material was the G protein by excising the top of the gel, incubating it with reducing agent, and rerunning the sample (lane 5, Fig. 1). This aggregate was not detected in extracts of yeast grown under conditions that did not express the VSV G cDNA (lane 6, Fig. 1). When the VSV G was prepared from VSV-infected, [%]methionine-labeled chicken embryo fibroblasts by using sample buffers lacking a reducing agent, no high molecular weight material was detected (lane 7, Fig. 1). If we made extracts of the labeled, infected chicken cells in the presence of unlabeled yeast cells that had been induced by VSV G synthesis and did not add a thiol reducing agent, we found only the monomeric form of G in immuneprecipitates (lane 8, Fig. 1). These last experiments show that oligomer formation did not occur in infected vertebrate cells and, furthermore, the disulfide-linked oligomers were not the result of procedures used for preparation of the extracts for gel analysis. The VSV G cytoplasmic domain of 29 amino acids contains one cysteine residue, and this amino acid is the site of acylation by fatty acid (4). This cysteine was not acylated in the yeast cells (I) and we considered the possibility that lack of acylation might allow for oligomer formation. A test of this hypothesis could be performed by expressing cDNAs encoding G proteins that either had the cysteine replaced by a serine (4) or were entirely lacking the membrane and cytoplasmic domains of this protein (5). These cDNAs were kindly provided by J. Rose (Salk Institute) and were
151 12
34
56
7
6
FIG. 1. Oligomers of VSV glycoprotein produced in yeast. Yeast cells carrying the expression vector pYMS-4 (see Ref. I) for VSV G were induced in 5% galactose minimum medium and labeled with 10 &i/ ml of [35S]methionine (Amersham, 1306 Ci/mm) for 4 hr. Cells were harvested and extracts prepared for immunoprecipitation. Lane 1, lysate from VSV-infected chicken embryo fibroblasts (CEF) labeled with [?S]methionine for 10 min at 4 hr postinfection; lane 2, extracts of yeast immunoprecipitated with anti-G antibody and protein A-Sepharose beads. The immuneprecipitates were treated for 3 min at 100” with SDS-PAGE sample buffer in the presence of 5% pmercaptoethanol; lane 3, same as lane 2, but the immuneprecipitates were treated with electrophoresis sample buffer in the absence of fl-mercaptoethanol; lane 4, yeast extracts were treated directly with electrophoresis sample buffer in the absence of fl-mercaptoethanol; lane 5, the top portion of gel from lane 4 was treated with /3-mercaptoethanol; lane 6, same as lane 4, hut the cells were grown under noninducing conditions in 5% glycerol medium; lane 7, lysate from VSV-infected CEF cells was immunoprecipitated with anti-G antibody, and immuneprecipitates were treated with electrophoresis sample buffer in the absence of p-mercaptoethanok lane 8, labeled VSV-infected CEF cells were mixed with unlabeled yeast cells grown in galactose and extracts immunoprecipitated with antiG antibody. The immuneprecipitates were treated with sample buffer in the absence of jzl-mercaptoethanol.
inserted into our plasmid in place of the normal G cDNA. For this construction, the G cDNAs were excised from the original
152
SHORT
COMMUNICATIONS
plasmid by BanzHI digestion and isolated from low-melting agarose gels. They were ligated into our pYMS-3 vector that had been cut with BumHI. The ligation mixtures were used to transform E. coli and ampicillin-resistant clones were screened by colony hybridization with nick-translated %P-labeled VSV G cDNA as a probe (refer to Ref. (I) for experimental details). These plasmids were used to transform yeast (1). Extracts were prepared from yeast that were grown in galactose medium to induce synthesis of the modified G proteins. We found that the modified G proteins also formed disulfide-linked oligomers, although some monomeric G (about
123456
FIG. 2. Oligomerization of modified VSV G expressed in yeast. Cells carrying the expression vector for either the normal VSV-G cDNA or the mutated cDNAs were labeled with [8SS]methionine in 5% galactose medium and extracts immunoprecipitated with anti-G antibody. They were treated with sample buffer in the absence (lanes 1,2,3) or presence (lanes 4,5,6) of @mercaptoethanol. Lanes 1 and 4, yeast carrying the normal G cDNA; lanes 2 and 5, yeast carrying G cDNA with the cytoplasmic cysteine residue changed to a serine; lanes 3 and 6, yeast carrying G cDNA with the cytoplasmic domain and membrane anchor region removed. The radioactivity for immuneprecipitates of normal G in lanes 1 and 4 was about l/3 of that for the modified G’s in lanes 2,3,5, and 6.
10% of the total) was detected (Fig. 2). Thus, oligomer formation was not the result of defective acylation of the cytoplasmic cysteine and, in fact, occurred with G polypeptides not tethered to membranes. The site of oligomer formation in the yeast secretory pathway could be determined by analyzing G protein expressed in the various yeast set mutants (6). When VSV G cDNA was expressed in the set 18 mutant that is blocked between the endoplasmic reticulum and Golgi, we detected only oligomers of G (data not presented). Thus, interdisulfide bonds apparently form during the folding of the polypeptide chain in the endoplasmic reticulum. It is interesting that this aberrant folding did not appear to affect attachment of the oligosaccharide chains which were present on the G isolated from yeast (1). The size of the G oligomers was estimated by measuring a sedimentation coefficient of the immunoprecipitated protein resuspended in low concentrations of SDS and centrifuged in a lo-30% sucrose gradient in a buffer of 0.05 M Tris-HCI, pH 7.4, 0.1 h4 NaCl, 1 mM EDTA. E. coli Sgalactosidase (S = 16) was used as an internal standard and the S value of the G determined by the method of Martin and Ames (7). Two major peaks of oligomeric G were detected with S values of 10 S and 30 S, although there were smaller amounts of labeled protein sedimenting between these values (Fig. 3). These results suggest that the oligomers are heterogeneous in size. The 10 S fraction accounted for 45% of the total G sample and corresponds to a mol wt of 260,000, assuming the protein is primarily spherical in shape and binds little detergent. Based on the results noted in Fig. 1, lane 4, it seemed possible to use the property of the oligomer formation as a method for rapid purification of the “foreign” protein from yeast cells. After induction of G, yeast cells were harvested and extracts prepared in the absence of reducing agent but in the presence of 1% SDS. These extracts were boiled and centrifuged in a microfuge (15,000 9) for 5 min. The supernatant fraction was diluted fivefold with our normal
SHORT
-<
13
; 0.5 2 9 0.4
25
3 0.3 P 0v) 0.2 3 g 0.1
li
10 FRACTION
15 NUMBER
20
153
COMMUNICATIONS
25
FIG. 3. Sucrose gradient centrifugation of the yeastderived VSV G protein oligomers. [86S]Methionine-labeled oligomers were mixed with E. edi @-galactosidase (100 units) and centrifuged through a 10-30s sucrose gradient in a Spinco SW41 at 40,000 rpm for 4 hr at 4’. Fractions (0.4 ml) were collected from the bottom of the tube and portions were analyzed for trichloroacetic acid precipitable radioactivity (@ 0) and for @-galactosidase activity (0 0) (16). The gradient is noted by the dashed line. The calculated S values for the radioactive peaks are 10 S (fraction 22) and 30 S (fraction 14), based on an S value for fi-galactosidase of 16 S.
immunoprecipitation buffer (1% Triton X100, 1% sodium deoxycholate, 0.1% SDS, 0.15 M NaCl, 0.05 M Tris-HCl, pH 7.2) and centrifuged in a Beckman Spinco 65A rotor at 45K rpm for 1 hr. About 70% of the total immunoprecipitable G protein in the initial yeast extract was recovered in the pellet. Of the radiolabeled protein in this pellet fraction, 50% was the VSV G. Based on Coomassie blue staining of SDS-PAGE gels, 30% of the protein was VSV G. Thus, this one step achieves a 20- to 50-fold enrichment of the glycoprotein. We are currently studying further steps for obtaining homogenous material in high yields. The formation of glycoprotein oligomers was not unique to the VSV G, and yeast cells transformed with plasmids carrying the cDNA for Sindbis virus glycoproteins also produced oligomeric forms of the proteins (data not presented). When tested in the set 18 mutant defective in transport beyond the endoplasmic reticulum, the Sindbis glycoproteins also appeared as disulfide-linked aggregates. Why these glycoproteins form inter rather than intradisulfide bonds in the
yeast endoplasmic reticulum is not clear. Formation of native S-S bonds in proteins is an early step in the maturation of secretory proteins and appears to be catalyzed by a protein disulfide isomerase (reviewed in Ref. (8)). This enzyme is widespread and shows a broad specificity for protein substrates. We suggest that this enzyme is limiting in amounts in the yeast endoplasmic reticulum and the excessive amounts of virus glycoproteins made as a result of the induction via the galactose promoter prevent effective action by this disulfide isomerase. The prevalence of disulfide-linked trimers or tetramers of G may indicate that newly made G polypeptides normally associate in the mammalian cell ER to form noncovalent oligomers. There are several reports which show formation of abnormal protein aggregates when foreign proteins are expressed in yeast or E. coli (g-15). In E. coli, these aggregates sometimes appear as inclusion bodies and they inhibit normal cell growth. In some of these aggregates, abnormal disulfide bonds are found; in others, the aggregates can be solubilized by SDS, indicating noncovalent interactions. As described here for the VSV G, formation of these aggregates has facilitated isolation of the foreign proteins from the bulk of cellular proteins, and in the case of the hepatitis surface antigen the aggregates were potent immunogens (15). We are testing the VSV G aggregate for its immunogenicity and ability to induce neutralizing antibodies. ACKNOWLEDGMENTS We thank Dr. J. Rose for plasmids containing the VSV-G cDNAs, Dr. R. Schekman for the S. cereu-isiae set mutants, and Dr. C. Rice for kindly providing antibodies. This study was supported by a Public Health Service Grant AI 19494.
REFERENCES 1. WEN, D., and SCHLESINGER, M. J., Proc. Nat1 Acud Sci USA, in press (1986). 2. JOHNSTON, M., and DAVIS, R. W., MOL! CeU Biol. 4, 1440-1448 (1984).
154
SHORT
COMMUNICATIONS
8. SCHLESINGER, M. J., In “Virology” (B. N. Fields et al, eds.), pp. 1921-1932. Raven Press, New York, 1985. 4 ROSE, J. K., ADAMS, G. A., and GALLIONE, C. J., Proc Nat1 Ad Sci USA 81.2050-2054 (1984). 5. ROSE, J. K., and BERGMANN, J. E., Cell 30, ‘753-762 (1982). 6. SCHEKMAN, R., Trends Biochm. Sci 7, 243-246 (1982). 7. MARTIN, R. G., and AMES, B. N., J. Biol. Chem 236, 1372-1379 (1961). 8. FREEDMAN, R. B., Trends Bioch,em Sci. 9,438-441 (1984). 9. EMTAGE, J. S., ANGAL, S., DOEL, M. T., HARRIS, T. J. R., JENKINS, B., LILLEY, G., and LOWE, P. A., Proc. Nat1 Ad Sci USA 80, 3671-3675 (1983).
lo. GOFF, C. G., MOIR, D. T., KOHNO, T., GRAVIUS, T. C., SMITH, R. A., YAMASAKI, E., and TAUNTONRIGBY, A., Gae 27,35-46 (1984). II. SCHONER, R. G., ELLIS, L. F., and SCHONER, B. E., Biotechnology 3,151-X4 (1985). 12. SHARMA, S., and GODSON, G. N., science 228,879882 (1985). 19. SIMONS, G., REMAUT, E., ALLET, B., DEVOS, R., and FIERS, W., Gene 28.55-64 (1984). 14 SMITH, R. A., DUNCAN, M. J., and MOIR, D. T., Science229,1219-1224(1985). 15. WAMPLER, D. E., LEHMAN, E. D., BOGER, J., McALEER, W. J., and SCOLNICK, E. M., Proc. NatL Acd Sci. USA 82,6830&M (1985). 16. MILLER, J. H., “Experiments in Molecular Genetics.” Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, 1982.
153,150-154
Expression
(1986)
of Genes Yeast DUANZHI
Department
Encoding
Vesicular
Leads to Formation WEN,
MINGXIAO
of Microbiology
Stomatitis
and Sindbis
of Disulfide-Linked
DING,
AND MILTON
Virus Glycoproteins
in
Oligomers
J. SCHLESINGER~
and Immunology, Washingto?z University School of il4edicine, St. Louis, Missouri 6.3110
Received Februav
24, 1986; accepted May 12, 1986
Sacchu~omycea cerevisiae strains transformed with plasmids containing cDNAs coding for the glycoproteins of vesicular stomatitis or Sindbis viruses can be induced to produce large amounts of glycosylated virus glycoproteins. Studies reported here show that these proteins form high molecular weight disulfide-linked oligomers in the yeast endoplasmic reticulum. Oligomers were also found for two genetically altered forms of VSV G; one of these was lacking the membrane anchor domain and the other had the cysteine in the cytoplasmic tail replaced with serine. These oligomers can be separated from the bulk of yeast proteins by brief high-speed centrifugation of yeast extracts prepared by boiling cells with 1% sodium dodecyl sulfate. Treatment with thiol-reducing agents converts the oligomers to soluble monomeric forms, and this procedure leads to a substantial purification of glycoproteins from bulk yeast protein. @ 1986 Academic Press, IX
We have recently developed an efficient system for the expression of Sindbis and vesicular stomatitis virus (VSV) glycoproteins in yeast (1). The cDNAs encoding the VSV glycoprotein (G) and the Sindbis virus structural genes (the 26 S RNA coding for the capsid and glycoproteins El and E2) were fused to the yeast galactose kinase gene promoter (2) and inserted into a yeast Escherichia coli shuttle vector such that expression of the virus glycoproteins was only detectable when cells were grown in the presence of galactose. The virus glycoproteins were glycosylated but not fatty acylated and comprised about 2 to 3% of total yeast protein synthesis. Half of the VSV G protein was detected by surface labeling yeast cells, indicating that some of this protein was transported through the yeast secretory pathway. Cells expressing the Sindbis virus genes produced free capsid and glycosylated forms of El and the E2 precursor, ~62. Thus, the cotranslational proteolytic processing that normally occurs during expression of the Sindbis virus structural genes (reviewed in Ref. (8)) ‘To whom dressed.
requests
for
reprints
0042~6822/86 $3.00 Copyright All rights
Q 1966 by Academic Press, Inc. of reproduction in any form reserved.
should
was found in the yeast cell. However, the post-translational processing of p62 to E2, which takes place in the Golgi vesicles, was not detected in yeast. In addition, only low amounts of the Sindbis virus glycoproteins were transported in yeast. After induction of virus protein synthesis in galactose medium, yeast cells grew very slowly suggesting that formation of the foreign protein was blocking essential yeast cellular activities. In an attempt to determine why these proteins were inhibitory to cell growth, we examined the properties of the virus glycoproteins and report here that these glycoproteins formed disulfide-linked oligomers in the yeast endoplasmic reticulum. We used essentially the same procedures as previously described (1) for inducing and detecting synthesis of virus proteins in yeast strains transformed with plasmids containing the virus cDNAs. Yeast extracts were prepared by treating 35S-methionine labeled cells with 1% SDS and vortexing with glass beads. After boiling, the extracts were diluted and analyzed by immunoprecipitation with virus specific antiglycoprotein antibodies. Extracts from cells carrying the VSV G cDNA showed a single
be ad-
150
SHORT
COMMUNICATIONS
band in SDS-PAGE when p-mercaptoethanol was included in the sample buffer and this band had a mobility identical to authentic VSV G (lanes 1 and 2, Fig. 1). However, the immunoprecipitated VSV G prepared in the absence of reducing agent entered the stacking portion of slab gel but did not move further into the gel (lane 3, Fig. 1). This apparent aggregate of the G protein was not an artifact of the immunoprecipitation procedure; it also appeared in the crude extract of yeast before treatment with antisera (lane 4, Fig. 1). We could show that this high molecular weight material was the G protein by excising the top of the gel, incubating it with reducing agent, and rerunning the sample (lane 5, Fig. 1). This aggregate was not detected in extracts of yeast grown under conditions that did not express the VSV G cDNA (lane 6, Fig. 1). When the VSV G was prepared from VSV-infected, [%]methionine-labeled chicken embryo fibroblasts by using sample buffers lacking a reducing agent, no high molecular weight material was detected (lane 7, Fig. 1). If we made extracts of the labeled, infected chicken cells in the presence of unlabeled yeast cells that had been induced by VSV G synthesis and did not add a thiol reducing agent, we found only the monomeric form of G in immuneprecipitates (lane 8, Fig. 1). These last experiments show that oligomer formation did not occur in infected vertebrate cells and, furthermore, the disulfide-linked oligomers were not the result of procedures used for preparation of the extracts for gel analysis. The VSV G cytoplasmic domain of 29 amino acids contains one cysteine residue, and this amino acid is the site of acylation by fatty acid (4). This cysteine was not acylated in the yeast cells (I) and we considered the possibility that lack of acylation might allow for oligomer formation. A test of this hypothesis could be performed by expressing cDNAs encoding G proteins that either had the cysteine replaced by a serine (4) or were entirely lacking the membrane and cytoplasmic domains of this protein (5). These cDNAs were kindly provided by J. Rose (Salk Institute) and were
151 12
34
56
7
6
FIG. 1. Oligomers of VSV glycoprotein produced in yeast. Yeast cells carrying the expression vector pYMS-4 (see Ref. I) for VSV G were induced in 5% galactose minimum medium and labeled with 10 &i/ ml of [35S]methionine (Amersham, 1306 Ci/mm) for 4 hr. Cells were harvested and extracts prepared for immunoprecipitation. Lane 1, lysate from VSV-infected chicken embryo fibroblasts (CEF) labeled with [?S]methionine for 10 min at 4 hr postinfection; lane 2, extracts of yeast immunoprecipitated with anti-G antibody and protein A-Sepharose beads. The immuneprecipitates were treated for 3 min at 100” with SDS-PAGE sample buffer in the presence of 5% pmercaptoethanol; lane 3, same as lane 2, but the immuneprecipitates were treated with electrophoresis sample buffer in the absence of fl-mercaptoethanol; lane 4, yeast extracts were treated directly with electrophoresis sample buffer in the absence of fl-mercaptoethanol; lane 5, the top portion of gel from lane 4 was treated with /3-mercaptoethanol; lane 6, same as lane 4, hut the cells were grown under noninducing conditions in 5% glycerol medium; lane 7, lysate from VSV-infected CEF cells was immunoprecipitated with anti-G antibody, and immuneprecipitates were treated with electrophoresis sample buffer in the absence of p-mercaptoethanok lane 8, labeled VSV-infected CEF cells were mixed with unlabeled yeast cells grown in galactose and extracts immunoprecipitated with antiG antibody. The immuneprecipitates were treated with sample buffer in the absence of jzl-mercaptoethanol.
inserted into our plasmid in place of the normal G cDNA. For this construction, the G cDNAs were excised from the original
152
SHORT
COMMUNICATIONS
plasmid by BanzHI digestion and isolated from low-melting agarose gels. They were ligated into our pYMS-3 vector that had been cut with BumHI. The ligation mixtures were used to transform E. coli and ampicillin-resistant clones were screened by colony hybridization with nick-translated %P-labeled VSV G cDNA as a probe (refer to Ref. (I) for experimental details). These plasmids were used to transform yeast (1). Extracts were prepared from yeast that were grown in galactose medium to induce synthesis of the modified G proteins. We found that the modified G proteins also formed disulfide-linked oligomers, although some monomeric G (about
123456
FIG. 2. Oligomerization of modified VSV G expressed in yeast. Cells carrying the expression vector for either the normal VSV-G cDNA or the mutated cDNAs were labeled with [8SS]methionine in 5% galactose medium and extracts immunoprecipitated with anti-G antibody. They were treated with sample buffer in the absence (lanes 1,2,3) or presence (lanes 4,5,6) of @mercaptoethanol. Lanes 1 and 4, yeast carrying the normal G cDNA; lanes 2 and 5, yeast carrying G cDNA with the cytoplasmic cysteine residue changed to a serine; lanes 3 and 6, yeast carrying G cDNA with the cytoplasmic domain and membrane anchor region removed. The radioactivity for immuneprecipitates of normal G in lanes 1 and 4 was about l/3 of that for the modified G’s in lanes 2,3,5, and 6.
10% of the total) was detected (Fig. 2). Thus, oligomer formation was not the result of defective acylation of the cytoplasmic cysteine and, in fact, occurred with G polypeptides not tethered to membranes. The site of oligomer formation in the yeast secretory pathway could be determined by analyzing G protein expressed in the various yeast set mutants (6). When VSV G cDNA was expressed in the set 18 mutant that is blocked between the endoplasmic reticulum and Golgi, we detected only oligomers of G (data not presented). Thus, interdisulfide bonds apparently form during the folding of the polypeptide chain in the endoplasmic reticulum. It is interesting that this aberrant folding did not appear to affect attachment of the oligosaccharide chains which were present on the G isolated from yeast (1). The size of the G oligomers was estimated by measuring a sedimentation coefficient of the immunoprecipitated protein resuspended in low concentrations of SDS and centrifuged in a lo-30% sucrose gradient in a buffer of 0.05 M Tris-HCI, pH 7.4, 0.1 h4 NaCl, 1 mM EDTA. E. coli Sgalactosidase (S = 16) was used as an internal standard and the S value of the G determined by the method of Martin and Ames (7). Two major peaks of oligomeric G were detected with S values of 10 S and 30 S, although there were smaller amounts of labeled protein sedimenting between these values (Fig. 3). These results suggest that the oligomers are heterogeneous in size. The 10 S fraction accounted for 45% of the total G sample and corresponds to a mol wt of 260,000, assuming the protein is primarily spherical in shape and binds little detergent. Based on the results noted in Fig. 1, lane 4, it seemed possible to use the property of the oligomer formation as a method for rapid purification of the “foreign” protein from yeast cells. After induction of G, yeast cells were harvested and extracts prepared in the absence of reducing agent but in the presence of 1% SDS. These extracts were boiled and centrifuged in a microfuge (15,000 9) for 5 min. The supernatant fraction was diluted fivefold with our normal
SHORT
-<
13
; 0.5 2 9 0.4
25
3 0.3 P 0v) 0.2 3 g 0.1
li
10 FRACTION
15 NUMBER
20
153
COMMUNICATIONS
25
FIG. 3. Sucrose gradient centrifugation of the yeastderived VSV G protein oligomers. [86S]Methionine-labeled oligomers were mixed with E. edi @-galactosidase (100 units) and centrifuged through a 10-30s sucrose gradient in a Spinco SW41 at 40,000 rpm for 4 hr at 4’. Fractions (0.4 ml) were collected from the bottom of the tube and portions were analyzed for trichloroacetic acid precipitable radioactivity (@ 0) and for @-galactosidase activity (0 0) (16). The gradient is noted by the dashed line. The calculated S values for the radioactive peaks are 10 S (fraction 22) and 30 S (fraction 14), based on an S value for fi-galactosidase of 16 S.
immunoprecipitation buffer (1% Triton X100, 1% sodium deoxycholate, 0.1% SDS, 0.15 M NaCl, 0.05 M Tris-HCl, pH 7.2) and centrifuged in a Beckman Spinco 65A rotor at 45K rpm for 1 hr. About 70% of the total immunoprecipitable G protein in the initial yeast extract was recovered in the pellet. Of the radiolabeled protein in this pellet fraction, 50% was the VSV G. Based on Coomassie blue staining of SDS-PAGE gels, 30% of the protein was VSV G. Thus, this one step achieves a 20- to 50-fold enrichment of the glycoprotein. We are currently studying further steps for obtaining homogenous material in high yields. The formation of glycoprotein oligomers was not unique to the VSV G, and yeast cells transformed with plasmids carrying the cDNA for Sindbis virus glycoproteins also produced oligomeric forms of the proteins (data not presented). When tested in the set 18 mutant defective in transport beyond the endoplasmic reticulum, the Sindbis glycoproteins also appeared as disulfide-linked aggregates. Why these glycoproteins form inter rather than intradisulfide bonds in the
yeast endoplasmic reticulum is not clear. Formation of native S-S bonds in proteins is an early step in the maturation of secretory proteins and appears to be catalyzed by a protein disulfide isomerase (reviewed in Ref. (8)). This enzyme is widespread and shows a broad specificity for protein substrates. We suggest that this enzyme is limiting in amounts in the yeast endoplasmic reticulum and the excessive amounts of virus glycoproteins made as a result of the induction via the galactose promoter prevent effective action by this disulfide isomerase. The prevalence of disulfide-linked trimers or tetramers of G may indicate that newly made G polypeptides normally associate in the mammalian cell ER to form noncovalent oligomers. There are several reports which show formation of abnormal protein aggregates when foreign proteins are expressed in yeast or E. coli (g-15). In E. coli, these aggregates sometimes appear as inclusion bodies and they inhibit normal cell growth. In some of these aggregates, abnormal disulfide bonds are found; in others, the aggregates can be solubilized by SDS, indicating noncovalent interactions. As described here for the VSV G, formation of these aggregates has facilitated isolation of the foreign proteins from the bulk of cellular proteins, and in the case of the hepatitis surface antigen the aggregates were potent immunogens (15). We are testing the VSV G aggregate for its immunogenicity and ability to induce neutralizing antibodies. ACKNOWLEDGMENTS We thank Dr. J. Rose for plasmids containing the VSV-G cDNAs, Dr. R. Schekman for the S. cereu-isiae set mutants, and Dr. C. Rice for kindly providing antibodies. This study was supported by a Public Health Service Grant AI 19494.
REFERENCES 1. WEN, D., and SCHLESINGER, M. J., Proc. Nat1 Acud Sci USA, in press (1986). 2. JOHNSTON, M., and DAVIS, R. W., MOL! CeU Biol. 4, 1440-1448 (1984).
154
SHORT
COMMUNICATIONS
8. SCHLESINGER, M. J., In “Virology” (B. N. Fields et al, eds.), pp. 1921-1932. Raven Press, New York, 1985. 4 ROSE, J. K., ADAMS, G. A., and GALLIONE, C. J., Proc Nat1 Ad Sci USA 81.2050-2054 (1984). 5. ROSE, J. K., and BERGMANN, J. E., Cell 30, ‘753-762 (1982). 6. SCHEKMAN, R., Trends Biochm. Sci 7, 243-246 (1982). 7. MARTIN, R. G., and AMES, B. N., J. Biol. Chem 236, 1372-1379 (1961). 8. FREEDMAN, R. B., Trends Bioch,em Sci. 9,438-441 (1984). 9. EMTAGE, J. S., ANGAL, S., DOEL, M. T., HARRIS, T. J. R., JENKINS, B., LILLEY, G., and LOWE, P. A., Proc. Nat1 Ad Sci USA 80, 3671-3675 (1983).
lo. GOFF, C. G., MOIR, D. T., KOHNO, T., GRAVIUS, T. C., SMITH, R. A., YAMASAKI, E., and TAUNTONRIGBY, A., Gae 27,35-46 (1984). II. SCHONER, R. G., ELLIS, L. F., and SCHONER, B. E., Biotechnology 3,151-X4 (1985). 12. SHARMA, S., and GODSON, G. N., science 228,879882 (1985). 19. SIMONS, G., REMAUT, E., ALLET, B., DEVOS, R., and FIERS, W., Gene 28.55-64 (1984). 14 SMITH, R. A., DUNCAN, M. J., and MOIR, D. T., Science229,1219-1224(1985). 15. WAMPLER, D. E., LEHMAN, E. D., BOGER, J., McALEER, W. J., and SCOLNICK, E. M., Proc. NatL Acd Sci. USA 82,6830&M (1985). 16. MILLER, J. H., “Experiments in Molecular Genetics.” Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, 1982.