The Contributions of Protein Disulfide Isomerase and Its Homologues to Oxidative Protein Folding in the Yeast Endoplasmic Reticulum*

In vitro, protein disulfide isomerase (Pdi1p) introduces disulfides into proteins (oxidase activity) and provides quality control by catalyzing the rearrangement of incorrect disulfides (isomerase activity). Protein disulfide isomerase (PDI) is an essential protein in Saccharomyces cerevisiae, but the contributions of the catalytic activities of PDI to oxidative protein folding in the endoplasmic reticulum (ER) are unclear. Using variants of Pdi1p with impaired oxidase or isomerase activity, we show that isomerase-deficient mutants of PDI support wild-type growth even in a strain in which all of the PDI homologues of the yeast ER have been deleted. Although the oxidase activity of PDI is sufficient for wild-type growth, pulse-chase experiments monitoring the maturation of carboxypeptidase Y reveal that oxidative folding is greatly compromised in mutants that are defective in isomerase activity. Pdi1p and one or more of its ER homologues (Mpd1p, Mpd2p, Eug1p, Eps1p) are required for efficient carboxypeptidase Y maturation. Consistent with its function as a disulfide isomerase in vivo, the active sites of Pdi1p are partially reduced (32 ± 8%) in vivo. These results suggest that PDI and its ER homologues contribute both oxidase and isomerase activities to the yeast ER. The isomerase activity of PDI can be compromised without affecting growth and viability, implying that yeast proteins that are essential under laboratory conditions may not require efficient disulfide isomerization.

Disulfide bonds provide added stability to extracellular proteins by covalently cross-linking two cysteines. Disulfide formation is often error-prone, particularly in the early stages of folding (1), and pairing the correct cysteines into disulfides requires that any mispaired disulfides must be broken and reformed in a different configuration to reach the native structure. In bacteria, disulfides are formed in the periplasm by an elaborate system of oxidases and isomerases that assure the correct cysteines are connected (2,3). In eukaryotes this posttranslational modification occurs in the endoplasmic reticulum (ER) 1 where a complex set of enzyme catalysts promotes correct disulfide formation. In yeast (4,5) and mammalian cells (6) the oxidizing equivalents for disulfide formation are generated principally by Ero1p. These disulfides in turn are delivered to protein disulfide isomerase (Pdi1p), an essential folding catalyst of the endoplasmic reticulum (7).
Both yeast and mammalian protein disulfide isomerase (PDI) are composed of four domains (termed a, b, b, and a) and an anionic tail (c) (8). The two catalytic domains (a and a) are located at the ends of the molecule, and each contains an active site with the sequence CGHC. The catalytic thioredoxin domains are separated by two non-catalytic thioredoxin domains (b and b) (9) in a multidomain structure (abbac). When the active-site cysteines of PDI are in a disulfide (oxidized) form, the enzyme can introduce disulfides into proteins (oxidase activity) through thiol/disulfide exchange. However, when the active-site cysteines of PDI are present in a dithiol (reduced) form, the active site is able to catalyze the reduction or isomerization of substrate disulfides (10,11).
Although PDI exhibits both its oxidase and isomerase activities in vitro, and is by far the most active disulfide isomerase known, some uncertainty exists about its function in vivo. The PDI1 gene is essential in yeast (7). A Pdi1p mutant with no active-site cysteines and no redox-related activity will not complement the lethal deletion of the PDI1 gene (12). Clearly, a catalytic redox function of Pdi1p, either its oxidase or isomerase activity, is essential for yeast growth and viability.
Surprisingly, mutant forms of Pdi1p that are significantly defective in either the oxidase or isomerase activities will rescue the deletion mutant of PDI1. Mutating a single cysteine in both active sites (CGHS:CGHS) impairs the oxidase activity of PDI, but these mutants retain significant isomerase activity in vitro (11,12). These oxidase-deficient mutants (12,13) restore viability to cells lacking Pdi1p, leading to the suggestion that the oxidase activity is not its essential function (12). However, other evidence implies that the primary function of PDI is as an oxidase. Mutants in the CXXC active site of Pdi1p showed increased sensitivity to DTT that correlated to a decreased rate of folding of a secretory protein (13). Pdi1p also accepts oxidizing equivalents from the ER oxidase, Erolp, and transfers them to substrate proteins (14). Consistent with this view, Frand and Kaiser (14), using a gel-shift assay that modifies each free cysteine of Pdi1p with a 0.5-kDa mass label, showed that most of the cysteines of yeast PDI appear to be oxidized in the yeast ER. In addition, a single catalytic domain of PDI (a or a), which is grossly deficient in isomerase activity (5% of wild type) but has 50% of the oxidase activity of PDI in vitro, will replace yeast PDI and maintain normal growth and viability (15). Titration of the expression levels of the isomerase-deficient PDIa domain shows that yeast requires no more than 6% of PDI isomerase activity but needs 60% or more of its oxidase activity (16).
The imbalanced need for PDI fundamental activities raises several important questions about how disulfide formation and isomerization occur in the ER and whether or not PDI, the most active disulfide isomerase in vitro, even displays this activity in vivo. Redundancies in ER disulfide-forming pathways or compensating changes in the expression level or the oxidation state of PDI might confound the apparent requirements for PDI catalytic activities. There are four PDI1 homologues in the yeast ER (MPD1, MPD2, EUG1, and EPS1) (17). None is essential, and they are all normally expressed at low levels (18). However, all of them will rescue the ⌬pdi1 mutation when overexpressed (19 -22). One of these homologues, Mpd2p, becomes essential when an oxidase-deficient PDI mutant (CGHS: CGHS) replaces wild-type PDI (18).
The availability of an isomerase defective form of yeast Pdi1p (the a domain) that provides the essential function has allowed us to determine how PDI and its ER homologues contribute to disulfide isomerization in the yeast ER. In the experiments described below we use oxidase and isomerase-defective variants of PDI to show that limiting the oxidase activity limits cell growth even in the presence of compensatory mechanisms. Despite the importance of its oxidase activity, assays based on the maturation of yeast carboxypeptidase Y (CPY) suggest that PDI does display its isomerase activity in vivo. Using a more sensitive gel-shift assay to detect reduced PDI active sites, we also find that a significant fraction of the PDI active sites are in the reduced state, capable of catalyzing disulfide isomerization. ER homologues of PDI also provide oxidase and isomerase activity to the yeast ER, but surprisingly, yeast strains that have all the homologues of Pdi1p deleted still show wild-type growth rates even when isomerase-deficient Pdi1p provides the only source of PDI function. High levels of disulfide isomerization are not essential to the survival and growth of yeast, suggesting an evolutionary process in yeast that may select against essential proteins that require disulfide isomerization.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-Saccharomyces cerevisiae strains with complete deletions of PDI1 were obtained from Ron Raines (University of Wisconsin, Madison, WI) (12) and from Norgaard et al. (18). Strains with null mutations of PDI1, EUG1, MPD1, MPD2, and EPS1 alone and in combination are described in Norgaard et al. (18). A wild-type strain (CRY1) was obtained from Steve Elledge (Baylor College of Medicine).
The yeast expression vector YPP414-YPS is derived from plasmid pRS414 (23), which is a centromere-based, yeast-bacterial shuttle vector with a TRP1 yeast selectable marker. The pRS414 was digested with KpnI and EcoRI, and an 885-bp fragment encoding PDI1 promoter was inserted. This was followed by digestion with EcoRI and NotI and insertion of a 129-bp expression cassette fragment. This expression cassette contains BglII/BamHI sites that are in-frame with an N-terminal yeast PDI signal sequence to direct the expressed protein to the ER and a C-terminal yeast ER-retention sequence (HDEL).
The coding sequences of rat PDI, the a and a domains of yeast PDI (yeast PDIa, yeast PDIa) were PCR-amplified to contain a 5Ј-BamHI site and a 3Ј-BglII site. Yeast PDI was PCR-amplified to contain a BamHI site at both the 5Ј and 3Ј terminus. The yeast PDIa was amplified using the oligonucleotides 5Ј-GGG TCC CAG ATC TGA TTC CTC TGT CTT CCA ATT GGT C and 3Ј-AGA TTT AGG ATC CGA CGT CGA AGT GAC C GT TTT CCT TG as primers. Yeast PDIa was amplified using the oligonucleotides 5Ј-AAA TTT AGA TCT CCT GAA GAC TCC GCT G and 3Ј-GCG GAG CGG ATC CTT GCT TGA TCA TGA ATT GGA CAA TG as primers. After digestion with BglII and BamHI (rat PDI and yeast PDIa) or BamHI (yeast PDI), the PCR-amplified fragments were inserted into the corresponding site of YPP414-YPS. The yeast a domain includes amino acids from Pro-Glu-Asp-Ser to Met-IIe-Lys-Gln (amino acids 30 -139) of the yeast PDI sequence, and the yeast a domain includes amino acids from Glu-Asn-Gln-Asp to Phe-Asp-Val-Asp (amino acids 263-377).
Growth Phenotypes-Overnight cultures grown in synthetic complete media with 2% glucose (SC medium) were diluted to ϳ10,000 cells/ml, and aliquots of 30 l were placed onto freshly made SC plates. The plates were incubated for 2 days at 30°C before being photographed. The ability of yeast strains to grow in the presence of different concentrations of DTT was assayed by applying cells in the same way on freshly prepared SC plates containing 0 -3 mM DTT. The plates were photographed after incubation at 30°C for 2 days.
Determination of the Oxidation State of PDI-Maleimide-conjugated polyethylene glycol (Mal-PEG) was obtained from Nektar Therapeutics (San Carlos, CA) and was purified by gel filtration on a PD-10 column (Amersham Biosciences) to remove low molecular weight maleimides. These lower molecular weight maleimides react with free sulfhydryl groups but do not shift the molecular weight, leading to incomplete shifts of the molecular weight on SDS-PAGE in fully reduced controls. Yeast cells complemented by yeast PDI, yeast PDIa, or rat PDI were grown to mid-log phase in yeast extract/peptone/dextrose. Trichloroacetic acid was added to a final concentration of 20% to intact cells that had not been pelleted by centrifugation to avoid potential changes in the redox state. After centrifugation, the insoluble material was resuspended in ice-cold 40% trichloroacetic acid, and cells were lysed by vortexing with glass beads (Sigma). Lysates were removed from the glass beads and centrifuged, and the precipitated protein was washed with ice-cold acetone. The protein was resuspended in 5 mM gel-filtered Mal-PEG in non-reducing SDS sample buffer (3% SDS, 0.2 M Tris-HCl, pH 8, glycerol, bromphenol blue) and incubated for 30 min at room temperature then quenched by the addition of DTT to a final concentration of 50 mM. Reduced and oxidized controls were prepared by resuspending the proteins in either 10 mM DTT or 1 mM 5,5Ј-dithiobis(2nitrobenzoic acid) in non-reducing 2ϫ SDS sample buffer, incubating for 30 min at room temperature, acid-precipitating the protein with trichloroacetic acid (20% final concentration), resuspending in Mal-PEG, and then quenching by the addition of DTT to a final concentration of 50 mM. Samples were resolved by SDS-PAGE on precast 4 -20% Tris-HCl gels (Bio-Rad) and transferred to nitrocellulose. Yeast PDI and yeast PDIa were visualized by probing with a polyclonal anti-yeast PDI antibody (from Jakob Winther) followed by a secondary antibody conjugated to horseradish peroxidase. Rat PDI was visualized by probing with a polyclonal anti-rat PDI antibody followed by a secondary antibody conjugated to horseradish peroxidase. Enhanced chemiluminescence (Amersham Biosciences) was used to detect the PDI species. Band intensities were determined using Scion Image software with film exposures in the linear range of intensities. The number of cysteines present as sulfhydryl groups was determined by multiplying the intensity of each band times the number of modified sulfhydryl groups represented by the position of the band. These intensities were summed and divided by the total intensity to estimate the average number of available sulfhydryl groups in the original protein.
Protein Expression and Purification from Escherichia coli-Yeast PDI and yeast PDIa were PCR-amplified to contain a 5Ј-NdeI site and a 3Ј-XhoI site. After digestion with NdeI and XhoI, the PCR products were inserted between the corresponding sites of pET23a (Novagen, Madison, WI). E. coli strain BL21 (DE3) (Invitrogen) transformed with the appropriate pET23a vector was grown at 37°C in LB media supplemented with 100 g/ml ampicillin to an absorbance of 1.0 at 600 nm and induced by adding isopropyl ␤-D-thiogalactoside to a final concentration of 1 mM. After 4 h the cells were harvested by centrifugation, suspended in ice-cold loading buffer (pH 7.5, 20 mM phosphate, 0.5 mM NaCl, and 100 mM imidazole), and disrupted by sonication. After centrifugation at 12,000 ϫ g for 15 min at 4°C, the supernatant was applied to a HiTrap chelating column (Amersham Biosciences) precharged with nickel and equilibrated with loading buffer. The column was washed with 10 ml of loading buffer and then with 5 ml of elution buffer containing 300 mM imidazole. Eluate fractions containing PDI were pooled and dialyzed against 50 mM Tris-HCl, pH 8.0, and 1 mM EDTA to remove the imidazole. The purity of proteins was analyzed by SDS/PAGE and estimated to be Ͼ90%.
Ribonuclease Refolding Assays-Renaturation of reduced RNase was followed in a continuous assay as described previously (24). The formation of active RNase was measured spectrophotometrically by monitoring hydrolysis of the RNase substrate, cCMP, at 296 nm. Each sample contained 4.5 mM cCMP, 1 mM GSH, 0.2 mM GSSG, 100 mM Tris-HCl, pH 8.0, 1 mM EDTA, 8 M reduced RNase, and 0 -9 M yeast PDI or yeast PDIa. The assay was performed at 25°C and initiated by the addition of reduced RNase.

RESULTS
Complementation of the ⌬pdi1 Deletion with a Single Catalytic Domain of PDI-Isomerase-deficient mutants of PDI support growth when expressed from the inducible-repressible GAL1-10 promoter (16). To compare the abilities of individual yeast Pdi1p catalytic domains to support growth when expressed from the endogenous PDI1 promoter on a low copy (cen) plasmid, genes encoding the a (yeast PDIa) and a (yeast PDIa) domains of yeast PDI, full-length yeast PDI and rat PDI were individually introduced into ⌬pdi1 S. cerevisiae using a plasmid shuffling method (12). Expression of yeast PDIa complemented the ⌬pdi1 deletion with only a slightly reduced growth rate compared with the wild-type yeast PDI (Fig. 1). Rat PDI, despite the fact that its oxidase and isomerase activities are similar to those of yeast PDI in vitro (25), supported significantly slower growth (Fig. 1). The yeast PDIa domain and the individual catalytic domains of rat PDI were not able to support growth when expressed from the PDI1 promoter; however, they all rescued the lethal phenotype when overexpressed from the PDI1 promoter on a multicopy plasmid (2 m) (data not shown).
Activities of yeast PDIaЈ in Vitro-The individual catalytic domains (a and a) of mammalian PDI are active oxidases but have little isomerase activity (15,26). When overexpressed, these mammalian catalytic domains will rescue the lethal deletion of PDI1. Although the yeast protein is predicted to have a domain structure that is similar to the mammalian protein, the catalytic domains of the yeast protein have not been characterized. Full-length yeast PDI and yeast PDIa were expressed in E. coli and purified through a C-terminal His 6 tag. Domain boundaries were predicted by homology modeling of the yeast PDIa domain using 3D-PSSM (27). As with the mammalian catalytic domains, the oxidase activity of yeast PDIa with reduced ribonuclease A as substrate is 50% that of an equivalent molar concentration of wild-type yeast PDI (based on moles of protein rather than active sites), but the isomerase activity is quite low, only 5% of wild type (Fig. 2).
The Role of ER Homologues of PDI-All known disulfide isomerases are members of the thioredoxin family, and MPD1, MPD2, EUG1, and EPS1 are the only four thioredoxin family homologous of PDI1 thought to be expressed in the ER of yeast (17). None of these homologues is essential, but they might provide isomerase and/or oxidase activity that becomes essential when it is not provided by yeast PDI1. If so, the ability of yeast PDIa or rat PDI to support viability may depend on chromosomal copies of other ER homologues. To test this we shuffled the various plasmids containing yeast PDIa, yeast PDI, and rat PDI into strains carrying a pdi1 null mutation in combination with deletions of the homologues. Surprisingly, yeast PDIa behaves exactly like full-length yeast Pdi1p, supporting near wild-type growth in all strains, including the strain in which all homologues are absent (Table I). Rat PDI, however, is able to rescue the ⌬pdi1 deletion only if the homologue, MPD2, is present. All strains rescued by rat PDI display slow growth rates, a result similar to that reported previously with an oxidase-deficient yeast PDI mutant in which both active sites were mutated to CGHS (18) Redox state of PDI in Vivo-The in vivo redox states of yeast PDI, yeast PDIa and rat PDI were examined using a gel-shift assay based on mobility shifts caused by the reaction of a free sulfhydryl group with a Mal-PEG (28). To preserve the oxidation state of these proteins, trichloroacetic acid was directly added to intact cells to quench any thiol-disulfide exchange followed by disruption and treatment with Mal-PEG in the presence of SDS. Mal-PEG alkylation of a single sulfhydryl results in an apparent molecular mass shift of ϳ15 kDa as observed by SDS-PAGE (Fig. 3). The Mal-PEG mass-shift on SDS-PAGE indicates the number of cysteines that are present as sulfhydryl groups. Those that are in disulfide oxidation states will not shift with Mal-PEG treatment. Because yeast PDI has six total sulfhydryl groups, the unshifted band ( Fig. 3) represents PDI molecules in which all the cysteines are in disulfides. When Pdi1p is trichloroacetic acid-extracted from a ⌬pdi1 deletion strain complemented with yeast PDI and treated with Mal-PEG, bands corresponding to zero, and two Mal-PEG additions predominate along with a small amount of PDI shifted by the addition of four Mal-PEG, suggesting that wild-type PDI is partially reduced in vivo (Fig. 3). A similar result is observed in a wild-type yeast strain (CRY1) with only a chromosomal copy of wild-type yeast PDI1 (Fig. 3). In the CRY1 strain, there are 1.7 Ϯ 0.4 (n ϭ 5) free sulfhydryl groups. The predominant bands represent fully oxidized Pdi1p and a species shifted by two Mal-PEG modifications. The band representing four sulfhydryl modifications is somewhat more prominent in the CRY1 strain and represents 16 Ϯ 6% (n ϭ 5) of the total intensity compared with 8 Ϯ 6% (n ϭ 11) in the strain complemented with yeast Pdi1p expressed from a plasmid under control of the PDI1 promoter. However, the difference is not statistically significant (p Ͼ 0.05).
Quantitation of the band intensities can be used to determine how many of the sulfhydryl groups are in each oxidation state. Because a population of active sites may be partially reduced/oxidized, the number of sulfhydryl groups that are available represents an average across all the sites. Approximately 1.3 Ϯ 0.3 (n ϭ 11) sulfhydryl groups (of six total) are found in yeast Pdi1p present in the yeast ER. However, rat PDI is much more reduced. All six of its six sulfhydryl groups are available for Mal-PEG modification (Fig. 3). Because the resolution between 4, 5, and 6 Mal-PEG additions is low, experiments were also performed in which the free SH groups were initially modified by N-ethylmaleimide. After reducing any disulfides with DTT, the sulfhydryl groups originally present as a disulfide were modified with Mal-PEG. The experiment (not shown) confirmed that rat PDI is almost entirely (Ͼ90%) reduced.
Yeast Pdi1p has two cysteines that are located outside the catalytic sites in the a domain. To determine whether the sulfhydryl groups found in yeast PDI are in the active site or located entirely in the non-active site positions, the non-active site cysteines were both mutated to alanines, and the construct was introduced into yeast. Mal-PEG shift experiments clearly show that the number of available sulfhydryl groups in this mutant (1.8 Ϯ 0.2, n ϭ 8) is comparable with that found in the wild-type protein. This suggests that the free cysteines observed in the wild-type PDI must be at one or both of the active sites and that the non-active site cysteines are present as a disulfide (Fig. 3). Homology modeling of the sequence of the yeast PDIa domain (27) places the two non-active site cysteines of the yeast a domain within 3.3 Å of each other, consistent with their presence as a disulfide. The finding that removing the non-catalytic cysteines has little effect on the number of available PDI sulfhydryl groups in the ER suggests that the 1.3 SH groups found in yeast Pdi1p represent reduced active sites. Expressed as the fraction of active sites in the reduced state, the active sites are 32 Ϯ 8% (n ϭ 11) reduced. The redox state of yeast PDIa, which has only two sulfhydryl groups, also reveals the presence of reduced active sites, comprising 43 Ϯ 10% (n ϭ 16) of the total. The Mal-PEG, a PEG-conjugated maleimide, has an average molecular mass of 5 kDa but shifts the protein by an apparent molecular mass of 15 kDa on SDS-PAGE. Proteins were displayed by SDS-PAGE (4 -20% Tris-HCl gel) and visualized by Western blotting using either an antiyeast PDI or an anti-rat PDI antibody followed by incubation with horseradish peroxidase-labeled secondary antibody and detection with ECL. Apparent molecular weight shifts above the unmodified protein indicate the number of free sulfhydryl groups modified. Cysteines present as disulfides are not modified. In each experiment the experimental sample is paired with reduced (red) and oxidized (ox) controls that were treated with 10 mM DTT or 1 mM 5,5Ј-dithiobis(2-nitrobenzoic acid) before Mal-PEG modification. Top, strains expressing yeast PDI (6 SH) from a cen plasmid (left) or in its normal chromosomal location (middle) in the wild-type strain, CRY1. The mutant yeast PDI (4 SH, right) has the two non-active site cysteines mutated to alanines. Middle, rat PDI (6 SH, middle) shown with ladders of rat PDI generated by using mixtures of Mal-PEG and N-ethylmaleimide. Bottom, yeast PDIa (2 SH, bottom).
The Oxidizing Apparatus of the ER-In the yeast ER, rat PDI is much more reduced than yeast PDI or yeast PDIa, raising the possibility that although isomerase activity could be supported by the reduced rat PDI, the oxidase activity might be compromised by a low concentration of oxidized active sites. Sensitivity of the growth rate to DTT is a convenient indicator of the effectiveness of the oxidizing apparatus of the ER (29,30). A ⌬pdi1 strain complemented by yeast PDIa is DTT tolerant up to a concentration of 3 mM DTT, similar to the behavior of strains complemented with yeast PDI. In contrast, ⌬pdi1 supported by rat PDI is considerably more sensitive to DTT, failing to grow at DTT concentrations higher than 0.5 mM (Fig. 4). DTT sensitivity was also examined in strains complemented by either yeast PDI1 or yeast PDIa in which all genes encoding PDI homologues were also deleted. By itself, deletion of all the PDI1 homologues of the ER has no significant effect on either the DTT sensitivity or the growth rate (Fig. 4), suggesting that yeast Pdi1p or yeast PDIa on their own provide the majority of the oxidizing activity.
Protein Folding in the ER-Upon translocation into the ER of wild-type yeast, CPY is oxidized and glycosylated with four core N-glycosyl residues, resulting in the precursor, p1, which has a molecular mass of 67 kDa. When properly folded, including the formation of native disulfide bonds, the p1 form exits the ER and is further modified in the Golgi compartment to give a 69-kDa, p2 form of CPY. This form is sorted to the vacuole, where it is processed to mature CPY (63 kDa) by cleavage of the pro-sequence (31).
Because only correctly folded proteins are allowed to exit the ER, a decrease in the rate of CPY maturation reflects defects in oxidative folding. The maturation of CPY was observed in pulse-chase experiments. As shown in Fig. 5, CPY maturation is strongly dependent on the gene that was introduced to rescue the ⌬pdi1 deletion. Although CPY maturation is almost arrested in cells expressing rat PDI, CPY maturation proceeds at a normal rate in cells rescued by yeast Pdi1p or yeast PDIa, indicating that protein folding is compromised in the presence of rat PDI but is intact as long as the strains are complemented by yeast Pdi1p or yeast PDIa.
Contributors to Isomerase Activity in the Yeast ER-Given the very low isomerase activity of yeast PDIa in vitro, the ability of strains complemented with yeast PDIa to support a near normal rate of CPY maturation (Fig. 5) may depend on other PDI1 homologues of the ER. Consistent with this sugges-tion, the maturation of CPY is seriously compromised when yeast Pdi1p is replaced by the isomerase-deficient yeast PDIa and all four of the ER homologues are also deleted (⌬pdi1, ⌬eug1, ⌬mpd1, ⌬mpd2, and ⌬eps1) (Fig. 5). The seriously compromised CPY folding in this strain can, however, be restored to wild-type behavior by replacing the isomerase-defective PDIa with wild-type yeast Pdi1p (Fig. 5). On the other hand, rat PDI does not support normal CPY maturation rates even when all of the homologues are present, suggesting that rat PDI does not provide oxidative folding activities that are available from yeast PDI even when the other PDI homologues are present.

DISCUSSION
In vitro PDI can catalyze the formation and isomerization of disulfide bonds. The most straightforward approach to deciding which of the two activities of PDI is essential to yeast would be to determine whether mutants that selectively destroy each of the activities support growth. The finding that a single catalytic domain of yeast PDI supports near wild-type growth when expressed at levels similar to the wild-type protein (16) has made it possible to separate the two catalytic activities. In vitro yeast PDIa has relatively high oxidase activity (ϳ50% of wild type) but minimal (ϳ5%) isomerase activity (Fig. 2). When Pdi1p is replaced by this isomerase-defective mutant (yeast PDIa), the oxidative folding apparatus of the ER is normal. The lack of DTT sensitivity in this strain (Fig. 4), the normal CPY maturation rates (Fig. 5), and the absence of a requirement for MPD2, which is needed in oxidase-defective strains (Table I) (14,18), all suggest that yeast PDIa functions as an efficient oxidase in vivo, comparable with wild-type yeast PDI.
The separation of oxidase and isomerase activities is less complete in the oxidase-deficient mutants of PDI that have been studied previously. Mutants with a single active site cysteine (CGHS) are deficient in oxidase activity in vitro (16% of wild type), but they also have lower isomerase activity (30 -50% of wild type) (11). In addition, the isomerase activity of the CGHS mutants of PDI is only observed in a redox buffer (32). Kaiser and Frand (14) find that these mutants can be trapped in a covalent complex with Ero1p. Consequently, the CGHStype mutants have two potential effects in vivo; they have diminished activity of Ero1p itself due to the inhibition by complex formation and an inability to transfer oxidizing equivalents to substrate proteins.
Rat and yeast PDI share relatively low sequence similarity (28% identical, 46% similar) yet have identical catalytic activities in vitro (25). Unlike yeast PDI, which is significantly oxidized in the ER, rat PDI is almost completely reduced (Fig.  3). Differences in the in vivo redox states of the rat and yeast proteins could be due to a number of factors including differences in their integration into the ER-oxidizing transfer pathway, differences in redox potentials, differences in expression level, or even the partial mislocalization of the rat protein to a more reduced compartment. The reduced rat enzyme might be able to provide isomerase activity in the yeast ER, but several lines of evidence suggest that when rat PDI is used to complement the deletion of PDI1, the oxidative activity of the yeast ER is compromised, consistent with the observation of reduced rat PDI. Strains complemented by rat PDI expressed from the PDI1 promoter show slow growth (Fig. 1), increased sensitivity to DTT (Fig. 4), a requirement for MDP2, and compromised CPY maturation (Fig. 5). All of these phenotypes are identical to those observed for strains supported by mutants of PDI with compromised oxidase activity (13,18).
The contribution of PDI to the oxidative folding activities of the ER quality control system can be observed most clearly after eliminating potential compensating mechanisms by de- FIG. 4. DTT sensitivity of yeast strains that express various PDI constructs. Overnight cultures grown in synthetic complete media were diluted to ϳ10,000 cells/ml, and aliquots of 30 l were applied on to SC plates containing various concentration of DTT. The plates were photographed after incubation at 30°C for 2 days.
leting the genes for all four of the non-essential PDI homologues of the ER (⌬pdi1, ⌬mpd1, ⌬mpd2, ⌬eug1, ⌬epsd1). In this strain yeast PDIa, even with its low isomerase activity (ϳ% of wild-type), will support near wild-type growth rates (Table I) when expressed from the endogenous PDI1 promoter (Fig. 3). However, in this strain the maturation of CPY is extremely slow compared with the same strain complemented with wild-type yeast Pdi1p (Fig. 5). Because the oxidase activity of PDIa appears to be intact in vivo, the defect that is responsible for the compromised CPY maturation is most likely isomerization (Fig. 5). Thus, the oxidase activity provided by yeast PDIa appears to be entirely sufficient to provide the essential function of PDI and support normal growth. However, rat PDI (Table I) and a PDI mutant (CGHS:CGHS) (18) with compromised oxidase activity will not rescue the PDI1 deletion in the absence of the ER homologue, Mpd2p. By contrast, yeast cells supplied with PDI mutants with minimal isomerase activity are viable and grow normally, but compromising the PDI oxidase activity has a much more detrimental effect.
As with other critical pathways in yeast, compensation mechanisms involving the presence of alternative ER oxidases and isomerases may exist to supplement the missing functions of PDI. Clearly, the pathway for introducing oxidizing equivalents into the ER is redundant since at least two ER oxidases have been found, Ero1p (4) and Erv2p (33). There are also mechanisms that can supplement the oxidase activity of PDI. When the ER oxidase activity of PDI is diminished by replacing yeast PDI with the rat enzyme, one of the ER homologues, Mpd2p, becomes essential, suggesting that this protein is capable of supplementing the oxidase activity. Winther and coworkers (18) have also found that Mpd2 becomes essential when the oxidase-deficient PDI mutant, CGHS:CGHS, is used to supply yeast PDI function. The involvement of Mpd2p in catalyzing an alternative oxidation pathway is also suggested by the observation that Ero1p can transfer oxidizing equivalents to Mpd2p (14). The expression of PDI1 and the PDI homologues of the ER are all regulated by the unfolded protein response (34) so that their expression and the expression of defective PDI mutants may also depend on feedback from the ER as to the overall quality of the oxidative folding pathway. Although the PDI homologues are normally expressed at low levels (18), their expression could be up-regulated by defects in PDI activity (35).
Despite the fact that the most critical function of PDI is to support oxidation, our results suggest that PDI and its ER homologues also contribute to isomerase activity in vivo. CPY maturation requires disulfide isomerization in vitro (36), and the maturation of this nonessential yeast vacuolar protein provides a convenient assay of ER oxidative folding. In strains missing the other ER homologues of PDI and expressing the isomerase-deficient yeast PDIa, CPY maturation is significantly defective (Fig. 5). Because the oxidase activity expressed by PDIa supports normal growth, the impaired maturation of CPY implies a defect in isomerase activity in the absence of wild-type PDI and its homologues. In this homologue-deficient strain, replacing yeast PDIa with wild-type Pdi1p restores oxidative folding of CPY to normal (Fig. 5). Because the major difference in the in vitro activities of PDI and PDIa lies in the isomerase function, restoration of CPY folding is most likely due to adding the isomerase activity of Pdi1p. This would suggest that PDI, the most active disulfide isomerase in vitro, also displays this activity in vivo.
One or more of the PDI homologues also appears to contribute isomerase activity. A strain expressing PDIa in the absence of other ER homologues cannot support CPY maturation, but when the homologues are present, CPY isomerization is FIG. 5. Deletion of PDI homologues results in a decreased rate of CPY maturation in yeast PDIa-complemented ⌬pdi1 yeast strain. Cells were pulse-labeled with 35 S-labeled amino acid for 15 min and chased with nonradioactive cysteine and methionine for the indicated periods of time. A, CPY was immunoprecipitated, and the samples were resolved by SDS-PAGE. The precursor (p) forms of CPY and the mature (m) form of CPY are indicated. B, quantitation of the band intensity and non-linear least squares fitting of the disappearance of the precursor to a first-order exponential function provides estimates for the halflives for maturation of CPY. normal (Fig. 5). Because adding isomerase activity in the form of wild-type Pdi1p restores CPY maturation, we suggest that one or more of the homologues can also contribute to ER isomerization. As with ER oxidation, the isomerization pathways also appear to display some redundancy.
The maturation of CPY in the homologue-null strain complemented with PDIa has a half-life of about 30 min (Fig. 5), compared with 4 min in a wild-type strain. This slow maturation of CPY may be due to the uncatalyzed (spontaneous) isomerization of CPY, which has been observed in vitro (36), or there may be some unknown additional catalyst of isomerization available to CPY, albeit one that is much less effective than PDI. In contrast to the near normal CPY maturation in strains expressing yeast PDIa in the presence of other ER homologues, Solovyov et al. (16) found that CPY maturation was defective when PDIa was overexpressed from the GAL1-10 promoter (approximately eight times normal PDI expression levels). Thus, the overexpression of the single PDIa domain may interfere with normal ER function. This is consistent with the slow growth rate observed when PDIa, but not yeast PDI, is overexpressed (16).
Catalysis of disulfide isomerization requires a free PDI sulfhydryl group to initiate substrate rearrangements by attack on a substrate disulfide. Isomerase activity requires a reduced PDI active site. Previous gel-shift experiments using 4-acetamido-4Ј-maleimidylstilbene-2,2Ј-disulfonic acid to add 0.5 kDa to the molecular mass for each free sulfhydryl group have suggested that PDI is almost entirely oxidized in the ER (14). Using a reagent, Mal-PEG, which supplies a larger gel shift (15 kDa), reveals that ϳ32% of the PDI active sites are reduced, whereas 68% are oxidized. This means that on average 1.3 free sulfhydryl groups are available for reaction with maleimide (Fig. 3), a gel shift that would be difficult to detect with 4-acetamido-4Ј-maleimidylstilbene-2,2Ј-disulfonic acid. Of the six sulfhydryl groups of PDI, 4.6 (77% of the total cysteines) are in fact in an oxidized state, consistent with the results reported by Kaiser and Frand (14). However, a significant fraction of the active sites are available in the reduced, isomerase-capable state, consistent with a contribution of Pdi1p to substrate isomerization in the ER. The effects of wild-type PDI on the maturation of CPY and the presence of reduced PDI in the ER strongly suggest that PDI provides isomerase activity to the yeast ER.
In contrast to its mammalian counterparts, where the two non-active site cysteines are in the b domain, yeast PDI has two, non-active site cysteines in the a domain. When these cysteines are replaced by alanines, the in vivo redox state shows that mutant yeast PDI has 1.8 free sulfhydryls, comparable with that observed in the wild-type enzyme. Thus, the non-catalytic cysteines in the wild-type yeast enzyme are present as a disulfide, so that the 1.3 free sulfhydryl groups of yeast PDI in the ER are located in the active sites, where they are available to catalyze isomerization.
If the isomerase activity of PDI is dispensable for yeast growth and viability, how important is isomerase activity to yeast in general whether provided by yeast PDI or other isomerization catalysts of the ER? When 95% of the PDI isomerase activity is eliminated by replacing yeast PDI with yeast PDIa and all the ER homologues have been removed as well, the growth rate of the strain is near wild-type despite the fact that the ER oxidative folding activity appears to be severely impaired (Table I, Fig. 5). One might speculate that yeast has evolved to limit the need for disulfide isomerization in the folding of its essential proteins, at least under laboratory conditions. Nonessential substrates containing multiple disulfide bonds, like CPY, may require isomerization, but the folding of essential secretory substrates, which are required for optimal growth or other essential processes, may depend largely on the oxidative power of PDI rather than its isomerase activity. In contrast to yeast, bacterial disulfide formation is not essential. Eliminating the periplasmic system that forms disulfides in bacteria impairs mobility and the assembly of flagella, but it is not lethal (37). In bacteria, the DsbA-DsbB system is responsible for disulfide formation in secretory proteins, and the DsbC-DsbD system catalyzes isomerization. Eukaryotes, like yeast, appear to have become dependent on forming disulfides while consolidating the oxidase and isomerase functions into a single protein, PDI. Moreover, the ER is equipped with an array of PDI homologues that could sustain relatively efficient machinery for native disulfide bond formation when the catalytic functions of PDI are compromised.