THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 268, No. 11, Issue of April 15, pp. 7728-7732, 1993 Printed in U.S.A.

Functional Protein by Family*

Replacement of Members of the

the Saccharomyces cereuisiae Trgl/Pdil Mammalian Protein Disulfide Isomerase

(Received for publication, September 24,1992)

Rainer GuntherSg, Mythili Srinivasanll, Sherry HaugejordenY, Michael GreenV, Ina-Maria EhbrechtS, and Hans KuntzelSII From the $Max-Plunck-Institut fur Experimentelle Medizin, Hermann-Rein-Strasse 3, 0-3400 Gottingen, Federal Republic of Germany and the YDepartment of Microbiology, St. Louis University School of Medicine, St. Louis, Missouri 63104

The TRGlIPDIl gene of Saccharomyces cerevisiae is essential for growth and encodes a lumenal endo- plasmic reticulum (ER) glycoprotein that is structur- ally related to thioredoxin and is involved in the secre- tory pathway. We have tested whether the yeast Trgl/ Pdil protein can be replaced in vivo by three members of the mammalian thioredoxin-related protein family, protein disulfide isomerase (PDI), ERp72, and ERp61. Multicopy plasmids containing galactose-inducible ro- dent PDI and ERp72 genes support germination and growth of haploid t rg l null mutants in galactose-con- taining media, whereas the ERp61 gene is inactive. Strains expressing PDI or ERp72 instead of Trgl are thermosensitive. An overproduced mutant Trgl pro- tein lacking the HDEL retention signal supports growth, whereas a truncated version of the protein containing only one thioredoxin-like domain is inac- tive. The mammalian proteins were localized to both the soluble and microsomal membrane fraction of yeast cells. Our observations indicate that the two unglyco- sylated mammalian proteins PDI and ERp72 are ca- pable of replacing at least some of the critical functions of Trgl, in spite of the fact that the three proteins diverge considerably in sequences surrounding the thioredoxin-related domains.

Intracellular protein sorting in eukaryotes involves a re- trieval system which prevents some ER’ proteins from being secreted, by recognizing carboxyl-terminal retention signals such as KDEL, KEEL, or (in yeast) HDEL (1). A multifunc- tional subset of these resident lumenal ER proteins is char- acterized by the presence of two or three thioredoxin-like domains containing the catalytic center -CGHC- instead of the thioredoxin center -CGPC- (2, 3). The family of lumenal thioredoxin-related ER proteins includes the highly conserved PDI subgroup (2, 4-8) together with three more diverged members, ERp61 (9-11), P5 (12), and ERp72, a protein con- taining three instead of two thioredoxin-like domains (13).

* This work was supported by a Deutsche Forschungsgemeinschaft grant (to H. K.) and National Science Foundation Grant DCB.8905270 (to M. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisernent” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Present address: Medizinische Klinik der Universitat Kiel, Schit- tenhelmstr. 12, D-2300 Kiel, Federal Republic of Germany.

11 To whom correspondence should be addressed. Tel.: 0551-3899- 279; Fax: 0551-3899-388.

*The abbreviations used are: ER, endoplasmic reticulum; PDI, protein disulfide isomerase; kb, kilobase pair(s).

The PDI proteins are reported to be involved in protein disulfide isomerization (2,4, 5), prolyl4-hydroxylation (4,5), binding of N-glycosylation sites (6), binding of thyroid hor- mone (4,7), and microsomal lipid transfer (8), whereas ERp72 has a chaperone role for class I1 major histocompatibility (HLA-DR) proteins, preventing them from leaving the ER in the absence of the invariant peptide chain Ii (14). The ERp6l protein is a microsomal thio1:protein-disulfide oxidoreductase (12) that corresponds in its protein sequence to the product of a rat cDNA clone previously misidentified as the (Y form of phosphatidylinositol-specific phospholipase C (10). Recently, the ERp61 protein has been identified as the stress-inducible lumenal ER protein GRP58.’ Several ERp6l isoforms (ER6OA-ERGOH) appear to be thiol-dependent proteases de- grading PDI and calreticulin (15).

A yeast member ( TRGl ) of the thioredoxin-related protein family was cloned by screening a genomic library with a DNA probe specifying the peptide APWCGHCK (16). The gene has been mapped to the left arm of chromosome I11 close to the glucokinase gene GKLl and shown to be essential for growth. Its product has been identified as a lumenal ER glycoprotein involved in the maturation of vacuolar carboxypeptidase (16). The TRGl gene is allelic with PDII, a gene independently cloned by three other groups (17-19). Although the TRGlI PDIl gene product was originally identified by its ability to bind glycosylation site peptides (17), it is now clear that the Trgl protein is not required for the in vivo transfer of oligo- saccharide chains to nascent peptides (16,20, 21).

In order to learn more about i n vivo functions of this interesting class of proteins, we have studied the expression of mammalian genes encoding PDI, ERp72, and ERp61 in the yeast Saccharomyces cerevisiae. Our finding that two of these proteins (murine PDI and ERp72) can replace the growth- essential lumenal ER functions of the yeast Trgl protein indicates that the amino-terminal leader peptides as well as the mature mammalian proteins can be functional in yeast, although their thioredoxin-like domains are contained in rather divergent peptide regions. We also demonstrate that overproduced Trgl protein does not require a lumenal reten- tion signal for vital functions.

EXPERIMENTAL PROCEDURES

Construction of Plasmids-Plasmid pMDl contains the TRGl gene as a 3.8-kb MluI/XhoI fragment inserted between the BamHI and Sal1 site of pBM272 (22) after repairing the MluI and BamHI sites (16). The resulting GALlp-TRG1 gene (the GAL1 promoter fused at

* R. A. Mazzarella, N. Marcus, S. M. Haugejorden, J. M. Balcarek, J. J. Baldassare, B. Roy, L.-J. Li, A. S. Lee, and M. Green, manuscript in preparation.

7728

Expression of Mammalian PDI and ERp72 in Yeast 7729

position -91 of the TRGl gene) was excised as a 2.8-kb EcoRIIPstI fragment and inserted between the respective sites of pEMBL19, to give pMD3. A 2.8-kb EcoRI/SphI fragment containing the GALlp- TRGl gene was excised from pMD3 and inserted between the respec- tive sites of pNW5, a pBR322 derivative containing the HIS3 gene as a 1.6-kb insert (23), to give pMD4. A second BamHI site outside of GALlp-TRGl in pMD4 was eliminated by repair and religation (pMD7), followed by the insertion of a 2.2-kb EcoRI fragment con- taining the yeast origin 2pD (pDS1). pDS2 was obtained by deleting a 1.2-kb BamHI/NcoI fragment from pDSl (Fig. lA). In order to fuse the GALl promoter to the -3 position of TRGl, a synthetic BamHI/ Sal1 linker encoding the first 7 amino acids of the TRGl leader peptide and containing an NruI site between residues 6 and 7 was inserted between the BamHI and SalI sites of pMD7, to give pEG2.

A 1.45-kb BanIlSalI fragment of pMD3 containing most of the TRGl coding region was inserted between the NruI and Sal1 sites of pEG2, after repairing the Ban1 end, to give pEG4. The resulting GALlp-TRGl fusion sequence is shown in Fig. 1B. In order to delete the carboxyl-terminal HDEL retention signal from the TRGI gene product, the Sal1 site in pEG4 was silenced by end repair and religation (16), introducing a stop codon by frame shift (pEG4/dell). The central NcoI fragments of pEG4 were deleted by NcoI digestion and religation to give pEG4/de12.

A BamHI/SstI/XhoI polylinker was introduced at the translational start of a murine cDNA clone encoding PDI/ERp59 (13). A 1.96-kb BamHI/NcoI fragment containing the coding and upstream region of PDI cDNA was inserted between the respective sites of pDS2 to give pGAL-PDI (Fig. IC). A synthetic BamHI/SalI linker encoding the first 10 residues of the ERp61 leader peptide (lo), followed by a XmaI and SstI site, was inserted between the BamHI and SalI site of pDS2; a 1.8-kb XmaIISstI fragment containing the coding region (except the 10 first residues) of a rat ERp6l cDNA clone (kind gift of J. Balcarek) was then inserted between the respective sites of the linker to give pGAL-ERp61 (Fig. 1D). A BamHIIKpnI site was introduced into a murine cDNA clone encoding ERp72 (13). A 2.3-kb BamHI fragment containing the ERp72 coding region was inserted into the BamHI site of pDS2 to give pGAL-ERp72 (Fig. 1E).

Yeast Growth and Genetics-Yeast strains were grown in minimal media containing 6.8 g/liter N-base without amino acids, 2 % glucose (MMGlc), or 2% galactose ("Gal) and auxotrophic requirements. Rich media contained 1% yeast extract, 2% bacto peptone, and 2% glucose (YPD) or 2% galactose (YPGal). Sporulation medium (Sp2) contained 1.5% potassium acetate. All solid media contained 2% agar. The mutant diploid strain RG20 (MATa/MATa TRGl/trgl::URAJ u r d / u r d his3/his3 leu2/+) was transformed with derivatives of plasmid pDS2 (see Fig. 1) as described (16).

were incubated on solid YPGAl and YPD at 30 "C. The genotype of Asci were dissected with a Singer Micromanipulator, and spores

viable hybrid strains was confirmed by Southern hybridization of restricted genomic DNA and by rescue of plasmids into Escherichia coli.

Subcellular Localization of Overexpressed Protein-Yeast strains were grown in YPGal to early log phase. The cells were washed with water and converted to speroblasts by Zymolyase treatment. Subcel- lular fractionation was performed as described previously (16). Poly- clonal antibodies directed against a synthetic carboxyl-terminal Trgl epitope (16), and purified rodent ER proteins (13) were used for Western blotting.

RESULTS

Plasmid-dependent Spore Germination-In order to com- pare the ability of the mammalian thioredoxin-related pro- teins to complement the Trgl deficiency, we have fused their genes to the strong galactose-inducible and glucose-repressi- ble GALl promoter immediately upstream of the translational start site (see Fig. 1 for restriction maps and fusion se- quences). As reported previously (16-19), the yeast Trgl/Pdil protein shares several features with a group of mammalian proteins residing in the lumen of the ER, including an amino- terminal secretory leader peptide, a carboxyl-terminal lu- menal retention signal, and two domains containing the thio- redoxin-related catalytic center CGHC (Fig. 2). A third center is found only in the ERp72 protein (Fig. 2F). The yeast Trgl protein differs from the mammalian relatives by the presence of five N-glycosylation sites, which are all modified in vivo

- .5 ld,

1 GGATCGlT ATG AAG (BamHI) M K

TRG1

B E N

GGATCCGAGCTCGAG ATG CTG BamHl Sstl Xhol M L

PDI

BX ss P ERp61 GGATCC ATG CCT

BamHl M P

B P b E B

GGATCCGGTACC ATG AAG BamHl Kpll M K

1 ERp72

FIG. 1. Restriction maps and fusion sequences of galactose- inducible genes. A, multicopy shuttle vector pEG2. B-E, pEG2 inserts containing the S. cereuisiae TRGI gene ( B ) or cDNA clones encoding murine ERp59/PDI (C), murine ERp72 ( E ) , and rat ERp6l ( D ) . Arrow, GALl promoter. Filled boxes; coding regions. Open boxes, 3"flanking regions. The sequences refer to the junction between the GALl promoter and the translational start. Restriction cleavage sites are abbreviated as follows: E , EcoRI; B, BamHI; N , NcoI; S, SalI; B, BstEII; S, Sstl; P, PuuII.

(16). A single potential N-glycosylation site close to the amino terminus of ERp72 is not used in vivo (13).

The plasmids encoding the galactose-inducible ER proteins were introduced into a heterozygous TRGl/trgl::URA3 dip- loid strain (RG20) by transformation. The haploid progeny of these transformants were analyzed after dissection of tetrad asci and germination on solid media containing either galac- tose (YPGal, induction) or glucose (YPD, repression) (Table I).

The parental strain RG20 produces only two viable and uracil- requiring spores per tetrad on both media, confirming tha t the TRGl gene product is essential for germination and viability (16-19). The introduction of a multicopy galactose- inducible TRGl gene circumvents the lethality of the trgl disruption, since all four spores per tetrad germinate even under glucose repression of the TRGl gene (on YPD). A deletion of the carboxyl-terminal 38 residues, including the lumenal retention signal HDEL (Fig. 2 B ) has been shown previously to be lethal, if the mutant trgl gene was expressed under the control of the ADCl promoter (16). However, the same mutant gene promotes germination and growth of t rgl- disrupted haploid progeny of strain RG27 in galactose media, if fused to the stronger GALl promoter. A truncated variant of Trgl containing only one CGHC center (Fig. 2C) is ex- pressed in the diploid strain RG25 (data not shown) but does not rescue the Trgl deficiency (Table I). The germination pattern of strain RG26 indicates that the mammalian ERp61 protein cannot functionally replace the Trgl protein, since only two TRGl spores per tetrad germinated. However, pre- liminary data suggest that the mature ERp6l protein acquires some rescuing activity if fused to the Trgl leader peptide, although growth in YPGal of haploid Trgl-deficient cells expressing the chimeric protein is extremely slow (data not shown).

The two other members of the mammalian protein family (PDI and ERp72) are functional in yeast in their unmodified

7730 Expression of Mammalian PDI and ERp72 in Yeast

50 aa - Trgl em? I O 0 0 0 1 0 HDEL + Trgl- dl W A I O 0 0 0 I o 1 + Trgl - d2 P HDEL - PDI VZ4 I I I"+ KDEL+

ERp61 W A I 1 1 QEDL -

ERp72 0 I I I I KEEL + FIG. 2. Schematic diagrams of thioredoxin-related lumenal ER proteins. Filled boxes, thioredoxin-like centers (CGHC). Open

circles, N-glycosylation sites. Shaded boxes, secretory leader peptides. Boxes with wavy line, clusters of acidic residues. The carboxyl-terminal lumenal retention signals and the ability to complement the trgl disruption upon galactose induction are indicated at the right.

TABLE I Tetrad analysis of plasmid-containing TRCl/trgI::URA diploid

(RG20) transformants Viable spores per

tetradb

YPGal YPD Strain" Relevant genotype of

plasmid

RG20 No plasmid 2 2 RG22 GALlp-TRG1 4 4 RG27 GALlp-TRGl/A HDEL 4 2 RG25 GALlp-TRGlfA TRX 2 2 RG26 GALlp-ERp61 2 2 RG23 GALlp-PDI 4 2 RG28 GALlp-ERp72 4 2

"Genotype of the parental strain RG20 MATaIMATa TRGII

*At least twenty tetrads were analyzed. All viable spore pairs were trg1::URAS urd /ura3 hisSfhis3 leuZ/+.

uracil-requiring and contained the wild type TRGl gene.

state (Table I). Although only two TRGl spores per tetrad (derived from diploid strains RG23 and RG28) could germi- nate under glucose repression (on YPD), up to four viable spores were obtained on YPGal plates. All viable tetrads analyzed produced two large colonies containing uracil-re- quiring TRGl haploids and two small plasmid-containing trgl::URA3 haploid colonies, with a 2:2 segregation of mating type and leu2 markers. Since all plasmids were mitotically stable in a trgl::URA3 background, the hybrid strains express- ing mammalian proteins instead of Trgl could be propagated on nonselective media (YPGal).

Growth Properties of Hybrid Strains-A TRGl wild type strain (HK37) and three plasmid-complemented trgl-dis- rupted mutant strains were plated on solid galactose (YPGal) and glucose (YPD) media and grown a t 30 and 37 "C (Fig. 3).

A. YPGal, 30" C B. YPGal, 37O C

On YPGal at 30 "C the four haploid strains differ only mod- erately in their growth rates. Two strains expressing PDI and ERp72, respectively, however, are thermosensitive at 37 "C on YPGal and completely arrest on YPD even at the permis- sive temperature. The doubling times of the four strains are given in Table 11.

The tgl-disrupted strain RG22-1A, which is dependent on the galactose-inducible multicopy TRGl gene for growth has approximately a 1.5-fold lower rate of growth in YPGal than the TRGl wild type strain HK37. This suggests that the galactose-induced overexpression of Trgl has a slight inhibi- tory effect. The growth rate at 30 'C in YPGal is further reduced 2-2.3-fold, if the single copy TRGl gene is replaced by the multicopy mammalian PDI and ERp72 gene, respec- tively. A comparison of the growth kinetics at 30 and 37 "C clearly indicates that the TRGl wild type strain as well as the TRGl-overproducing trgl::URA3 strains are unimpaired by the elevated temperature, whereas the PDI- and ERp72- expressing strains are thermosensitive. The PDI strain, RG23-6C, continues to grow with a long doubling time of about 38 h, whereas the ERp72 strain, RG2R4C, completely arrests under the same conditions.

Cellular Localization of Mammalian Protein.? in Yeast- Yeast strains containing a single chromosomal copy of the TRGl gene produce three proteins, which are recognized by polyclonal antibodies directed against amino- and carboxyl- terminal Trgl epitopes, a TRGI-encoded glycoprotein doublet (gp70/72) localized in the lumen of the ER and a cross- reacting cytosolic 48-kDa protein of unknown genetic orifin (16). The overexpression of a galactose-inducible multicopy TRGl gene in strain RG22-1A produces a group of closely spaced protein bands recognized by anti-Trgl (Fig. 4), includ-

C. YPD, 30°C

FIG. 3. Growth of yeast strains on solid media. Patterns of strains is indicated in diagram. The relevant genotypes are as follows: wild type TRGl (HK37), trgl::URAS, pGALp-TRG1 (RG22-1A), trgI::URA3, pGALp-PDI (RG2:1-6C), trgI::URAd, and pGALp-ERp72 (RC2R- 4C).

Expression of Mammalian PDI and ERp72 in Yeast 7731

TABLE I1 Growth parameters of haploid yeast strains

Growth (doubling time in h )

Strain Relevant genotype" YPD,

30°C: 3 7 ° C 30"C

HK37 TRGI 2.6 2.8 1.5 RC22-1A trgl::URA3,pGALIp-TRGI 4.0 4.0 12 RG23-6C trgI::URA3,pGALlp-PDI 6.0 38 Arrest RG28-4C trfI::URA3,pGALIp-ERp72 5.4 Arrest Arrest

Nuclear background MATa, urd?, his3, ku2.

kDa A B

97 - 66 - 45 - 31 -

FIG. 4. Immunoblot analysis of galactose-induced Trgl pro- tein recognized by anti-Trgl. Strain RG22-1A was grown in liquid YPGal to mid-exponential phase, homogenized, and fraction- ated as described (17). A, membrane fraction (100,000 X g pellet); I?, soluble fraction (100,000 X g supernatant). Size standards are as follows: phosphorylase b, 97 kDa; bovine serum albumin, 66 kDa; ovalbumin, 45 kDa; carbonic anhydrase, 31 kDa.

kDa A B C D E F

97 a 66 . 45 - 31.

FIG. 5. Immunoblot analysis of galactose-induced PDI pro- tein (strain RG23-BC), using anti-ERp59 (A-C). A, soluble fraction (100,000 X g supernatant). R, membrane fraction (100,000 X gpellet). C, crude lysate (500 X g supernatant). L), crude lysate protein of RG23-6C treated with anti-Trgl. E and F, crude lysate protein of HK37 (wild type) treated with anti-Trgl (E) and anti-ERp59 ( F ) . For further details see the legend to Fig. 4.

ing the fully glycosylated gp72 as the upper band and a weak p59 band previously identified as the unglycosylated Trg l precursor (16). The four intermediate bands (molecular masses between 62 and 70 kDa) appear to represent partially core-glycosylated Trgl chains which are converted to the 59- kDa apoprotein by N-glycosidase F treatment (data not shown, see Ref. 16). Since most of the immunoreactive gp60- gp72 protein is found in the membrane fraction, we conclude that the majority of overexpressed Trgl protein is targeted to the ER but is not completely N-glycosylated, probably due to the saturation of the core glycosylation machinery by over- produced substrate protein.

Antibodies to murine PDI recognize a single 59-kDa species in strain RG23-6C expressing a galactose-inducible cDNA clone, localized both in the membrane and soluble fraction (Fig. 5, lanes A-C). Crude lysate protein of the same strain was also treated with anti-Trgl (lane D), confirming the absence of the TRGI-encoded gp70/72 (see lane E ) in the

kDa A B C D E

97 . 66 .

45 ,

31.

FIG. 6. Immunoblot analysis of galactose-induced ERp72 protein (strain RG2R-4C). using anti-ERp72 (A-C). A , soluble fraction (100,000 X R supernatant). R, memhrane fraction (1OO.of)() X g pellet). C, crude lysate (500 X g supernatant 1. I), crude lysate protein of RG2R4C treated with anti-Trgl. E , crude lysate protein of HK37 (wild type) treated with anti-ERp'i2. For further details see the legend to Fig. 4.

ERp59-expressing strain. These data also prove that the cytosolic p48 component cross-reacting with anti-Trgl is not derived from the TRGl gene product. Furthermore, yeast proteins do not cross-react with anti-PDI (Fig. 5, lune F).

A more complex protein pattern was seen when subcellular fractions from strain RG28-4C, expressing murine ERp72. were analyzed with anti-ERp72 antibody (Fig. 6). The slowest migrating major species (apparent molecular mass 80 kDa) is significantly larger than the expected ERp72 gene product, pointing to a post-translational modification of the murine protein in yeast. However, treatment of RC28-4C protein with N-glycosidase F did not alter the mobility of the immu- noreactive proteins (data not shown), and the observed size increment of 8 kDa is too large to be explained by a heterol- ogous core glycosylation of murine ERp72 at the single target site predicted by the ERp72 sequence (13). The protein could possibly be modified by 0-glycosylation, as has been observed in the case of the homologous ERp72/CaRP2 protein from rat liver:"

The immunoreactive proteins also include a 72-kDa species corresponding in size to the unmodified ERp72 protein, as well as some smaller polypeptides possibly derived by prote- olysis. The 80-kDa protein is clearly enriched in the mem- brane fraction (see l a n e s A and R ) , suggesting the efficient targeting of the heterologous protein to the ER in yeast.

DISCUSSION

In oitro studies have suggested a variety of functions for the PDI family of mammalian thioredoxin-related lumenal ER proteins (reviewed in Ref. 20). PDI has been identified as the /3 subunit of prolylhydroxylase ( 5 ) , as a component of a lipid-transporting protein complex (a), as a thyroid hormone- binding protein (7), and as a protein recognizing N-glycosyl- ation sites (6). However, more recent studies have now shown that PDI is not required for transfer of oligosaccarides to nascent proteins or acceptor peptides (25, 26), and that pep- tide binding is not limited to peptides containing N-glycosyl- ation acceptor sites (27, 28).

Furthermore, an in oioo protein disulfide isomerase func- tion of PDI and other thioredoxin-related ER proteins has not been demonstrated directly, and there is no evidence so far for the in oioo formation of incorrect disulfide bonds during folding of secretory proteins (29, 30).

Similarly, the in oioo functions of ERp72 and ERp61 cannot be deduced from in oitro data, which in some cases have led to misleading conclusions (10, 31). ERp72 is not a deoxycy-

H.-D. Soling, personal communication.

7732 Expression of Mammalian PDI and ERp72 in Yeast

tidine kinase, but a chaperone-like protein preventing certain target proteins from leaving the ER (14). ERp61 is not a cytosolic phosphoinositidase, but a lumenal ER protein hav- ing thio1:protein-disulfide oxidoreductase activity in vitro (11). Although a group of ER proteins sharing peptide se- quences with ERp61 have been shown to degrade PDI and calreticulin in uitro, it is not clear whether these proteins have proteolytic functions in vivo (15). ERp72, together with the glucose-regulated GRP94 protein, have recently been shown to co-immunoprecipitate with class I1 major histocom- patibility molecules in human cells expressing HLA-DR, in the absence of the invariant chain Ii. This suggests a role of ERp72 in retaining the HLA-DR molecules in the ER, per- haps by retrieving them from a cis- or pre-Golgi salvage compartment (14).

A growth-essential yeast member of the thioredoxin-related lumenal ER protein family (TrglIPdil) has been cloned re- cently and characterized (16-19). The product of the single- copy TRGIIPDII gene is the only known glycoprotein of this family and shares about equal sequence similarity within the thioredoxin-related domains of PDI, ERp61, and ERp72 (2, 10, 13, 16). Although the TRGlIPDIl product appears to be identical with a microsomal protein affinity-labeled by a glycosylation site peptide (17), the Trgl protein is clearly not required for N-glycosylation in viuo, since Trgl-depleted cells accumulate core-glycosylated precursors of the vacuolar car- boxypeptidase CPY (16).

Our observation that the Trgl function is required for the maturation of the thiol-containing vacuolar protease CPY (16) gives a first hint for a possible thiol-specific chaperone function in vivo, and the results of this study prove now that two mammalian proteins (PDI and ERp72) can replace the growth-essential in vivo function(s) of Trgl. A comparison of the growth rates observed under inducing conditions indicated that ERp72 is slightly more effective than PDI, but neither is as effective as the homologous yeast protein. Western blot analysis of membrane and soluble fractions showed that ERp72 is enriched in microsomal membranes, whereas about equal amounts of PDI are found in crude membranes and in the cytosolic fraction.

We conclude that the three lumenal ER proteins tested in this series of experiments, yeast Trgl, mammalian PDI, and ERp72, are functionally exchangeable, in spite of the fact that they show very little sequence similarity in regions outside the thioredoxin-like domains (13, 16).

Our data also imply that the amino-terminal leader peptides of the two mammalian proteins PDI and ERp72 are recog- nized by the yeast ER translocation system. We do not yet know whether the thermosensitivity of yeast strains express- ing PDI and ERp72 is caused by a block of translocation or by the impairment of lumenal functions at the restrictive temperature. In order to address this question, we have begun to replace the mammalian leader peptides by the analogous Trg l element.

The heterologous carboxyl-terminal lumenal retention sig- nals of the fully induced mammalian proteins (KDEL, KEEL, and QEDL, see Fig. 2) are probably not critically required for lumenal ER functions in yeast, since a Trgl variant lacking the HDEL signal is functional if sufficiently overproduced.

One of the mammalian proteins, ERp61, did not rescue the Trgl deficiency, although the GALlp-ERp61 fusion gene was transcribed in the heterozygous TRGlltrgl::URA3 diploid strain RG26 (data not shown). Our failure to detect a galac- tose-inducible 61-kDa protein by Western blot analysis using

anti-ERp61 antibodies might be explained by rapid proteoly- sis. Indeed, members of the ERp61 subfamily (ER6OA- ERGOH) have been reported to show thiol group-related pro- teolytic activity against themselves and PDI or calreticulin (15).

Sequence alignments of lumenal ER thioredoxin-related proteins have suggested that the yeast single-copy gene TRGl encodes a protein evolutionary related to the mammalian PDI family (16). Our complementation experiments now add a functional aspect to this relationship. The thioredoxin-related protein family may have a chaperone-like function that is specific for thiol-containing secretory proteins as they fold by interacting with the polypeptide backbone and by recognizing only cysteine side chains (29), preventing them from forming incorrect disulfide bonds (30). In yeast these functions may be performed by the product of a single gene, TRGIIPDII. In mammalian cells, however, a group of more specialized pro- teins including PDI, ERp72, and, perhaps, ERp61 may inter- act with different sets of target proteins in the ER.

Acknowledgment-We thank F. Benseler for DNA synthesis.

REFERENCES

2. Edman, J. C., Ellis, L., Blacher, R. W., Roth, R. A., and Rutter, W. J. 1. Pelham, H. R. B. (1989) Annu. Reu. Cell Biol. 5 , 1-23

(1985) Nuture 317. 267-270 3. 4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

Holmgren, A. (1968) Bur. J. Biochem. 6,475-484 Parkkonen, T., Kivirikko, K. 1.. and Pihlajaniemi, T. (1988) Biochem. J.

256. 1005-1011 Tasanen, K., Parkkonen, T., Chow, L., Kivirikko, K. I., and Pihlajaniemi,

Geetha-Habib, M., Noiva, R., Kaplan, H. A,, and Lennarz, W. J.(1988) Cell T. (1988) J. Biol. Chem. 263, 16218-16224

54. 1053-1060 Cheng, S., Gong, Q., Parkinson, C., Robinson, E. A,, Appella, E., Merlino,

Wetterau, J. R., Combs, K. A,, Spinner, S. N., and Joiner, B. J.(1990) J.

Lewis, M. J., Mazzarella, R. A,, and Green, M. (1986) Arch. Biochem.

Bennett, C. F., Balcarek, J. M., Varrichio, A., and Crooke, S. T. (1988)

Srivastava, S. P., Chen, N., Liu, Y. , and Holtzman, J. L. (1991) J. Biol.

~I ~~

G. T., and Pastan, I. (1987) J. Biol. Chem. 262 , 11221-11227

Biol. Chem. 265,9800-9807

Biophys. 245 , 389-403

Nuture 334,268-270

Chem. 266. 20337-20344 Chaudhuri, M: M., Tonin, P. N., Lewis, W. H., and Srinivasan, P. R. (1992)

Mazzarella, R. A,, Srinivasan, M., Haugejorden, S. M., and Green, M.

Schaiff, W. T., Hruska, K. A,, McCourt, D. W., Green, M., and Schwartz,

Biochern. J. 281,645-650

(1990) J. Biol. Chem. 265,1094-1101

B. D. (1992) J. Exp . Med. 176,657-666

Lehle. L.. and Kiintzel. H. (195

18. Farquhar, R., Honey, N., Murant, S. J., Bossier, P., Schultz, L., Montgom- ery, D., Ellis, R. W., Freedman, R. B., and Tuite, M. F. (1991) Gene (Amst.) 108.81-89

19. Tachikawa, H., Miura, T., Katakura, Y., and Mizunaga, T. (1991) J. Biochern. (Tokyo) 110,306-313

20. Noiva R. and Lennarz, W. J. (1992) J. Biol. Chem. 267,3553-3556 21. te Heisen', S., Janetzky, B., Lehle, L., and Aebi, M. (1992) EMBO J. 11 ,

22. Johnston M., and Davis, R. (1984) Mol. Cell. Biol. 4 , 1440-1448 23. Mink, M,', Basak, A. N., and Kuntzel, H. (1990) Mol. & Gen. Genet. 223 ,

24. Hillson, D. A,, Lambert, N., and Freedman, R. B. (1984) Methods Erzzymol.

25. Bulleid, N. J., and Freedman, R. B. (1990) EMBO J. 9,3527-3532 26. Noiva, R., Kaplan, H. A,, and Lennarz, W. J. (1991) Proc. Nutl. Acad. Sci.

27. Morjana, N. A,, and Gilbert, H. F. (1991) Biochemistry 30,49 85-4990 28. Noiva. R.. Kimura. H.. Roos. J.. and Lennarz. W. J. (1991) J. B i d . Chem.

. l" .

2071-2075

107-113

107, 281-294

U. S. A. 8 8 , 1986-1990

. .

29. Weisman, J. S., and Kim, P. S. (1991) Science 253, 1386-1393 30. Sep l , M. S., Bye, J. M., Sambrook, J. F., and Gething, M. H. (1992)

31. Huang, S. H., Tomich, J. M., Wu, H., Jong, A., and Holcenberg, J.

32. Semenza, J. C., Hardwick, K. G., Dean, N., and Pelham, H. R. B.

33. Andres, D. A., Dickerson, I. M., and Dixon, J. E. (1990) J. Biol.

34. Braakman, I., Helenius, J., and Helenius, A. (1992) EMBO J. 11 ,

266 , 19645-19649

lol. 118 , 227-244

J. Biol. Chem. 266, 5353

Cell 61, 1349-1357

265,5952-5955

1722

J. Cell

(1991)

(1990)

Chem.

1717-