ERO1-L, a human protein that favors disulfide bond formation in the endoplasmic reticulum.

Oxidizing conditions must be maintained in the endoplasmic reticulum (ER) to allow the formation of disulfide bonds in secretory proteins. Here we report the cloning and characterization of a mammalian gene (ERO1-L) that shares extensive homology with the Saccharomyces cerevisiae ERO1 gene, required in yeast for oxidative protein folding. When expressed in mammalian cells, the product of the human ERO1-L gene co-localizes with ER markers and displays Endo-H-sensitive glycans. In isolated microsomes, ERO1-L behaves as a type II integral membrane protein. ERO1-L is able to complement several phenotypic traits of the yeast thermosensitive mutant ero1-1, including temperature and dithiothreitol sensitivity, and intrachain disulfide bond formation in carboxypeptidase Y. ERO1-L is no longer functional when either one of the highly conserved Cys-394 or Cys-397 is mutated. These results strongly suggest that ERO1-L is involved in oxidative ER protein folding in mammalian cells.

Protein folding inside living cells is facilitated by the presence of a vast array of molecular chaperones and enzymes (1). For many secretory proteins, the acquisition of the native structure requires the formation of intra-and inter-molecular disulfide bonds (2). This process generally takes place in the endoplasmic reticulum (ER) 1 (3,4). This organelle contains many oxidoreductases, belonging to the protein disulfide isomerase family, which share a characteristic CXXC motif (5). These enzymes are thought to catalyze the formation of native disulfides in cargo proteins. Protein disulfide isomerase is nec-essary for disulfide bond formation in isolated microsomes (6,7) and can be cross-linked to folding or assembly intermediates within the ER lumen (8 -10). The crucial function of protein disulfide isomerase in the isomerization of S-S bonds in folding intermediates has been firmly established (11).
In bacteria, disulfide bonds are formed in the periplasmic space through the assistance of a set of well characterized molecules (for review, see Ref. 12). DsbA, a soluble protein of the periplasmic space, transfers disulfide bonds onto folding polypeptides. DsbA is oxidized by DsbB, an integral membrane protein protruding into the periplasmic space, which in turn transfers electrons to the respiratory chain (13)(14)(15). In respiratory-deficient conditions, DsbA is accumulated in the reduced form while DsbB is trapped in a disulfide-linked intermediate with DsbA (13). Although the primary function of DsbA seems to be the oxidation of periplasmic proteins, DsbC, whose redox state is under the control of DsbD, catalyzes disulfide isomerization. All members of the Dsb family contain at least one CXXC motif.
Although sequence comparisons do not allow the identification of eukaryotic homologs of Dsb genes, the basic principles of disulfide bond formation in the ER of eukaryotic cells and the bacterial periplasmic space are likely to be similar. Indeed, mammalian protein disulfide isomerase has been shown to complement bacterial DsbA mutants, a feature that implies a dithiol oxidant activity for this enzyme (16). Thus, protein disulfide isomerase seems to be capable of the functions that in bacteria are performed by DsbA (oxidation) and DsbC (isomerization), respectively.
One aspect of disulfide bond formation which is still poorly understood is how a suitable redox potential is generated and maintained in the ER. The view that oxidized glutathione (GSSG) is the source of the oxidizing equivalents in the ER (17 and references therein) was challenged recently by the discovery of the Ero1p protein in Saccharomyces cerevisiae (18,19). Ero1p is required for the oxidation of SH groups of both nascent proteins and glutathione in the ER (20). Decreased Ero1p function leads to exaggerated susceptibility of yeast cells to DTT, whereas overexpression confers resistance (18,19).
The levels of functional Ero1p also correlate with the rates of maturation and intracellular transport of disulfide-containing secretory proteins such as carboxypeptidase Y (CPY) and Gas1p, which cannot exit the ER in the reduced form. It has been suggested that Ero1p may fulfill the role of DsbB in the ER of eukaryotic cells (18,19) maintaining protein disulfide isomerase in an oxidized state.
Several observations suggest that in mammalian cells, the ER redox potential is tightly controlled so as to modulate protein secretion (21), degradation (22), and signaling (23,24). Therefore, we undertook a study aimed at identifying and characterizing mammalian genes potentially involved in redox homeostasis within the ER. The existence of genes homologous to S. cerevisiae ERO1 had been reported in several species (19). In the present paper, we describe the isolation and characterization of a human homolog (ERO1-L) encoding a membraneassociated N-glycoprotein of the ER which favors oxidative protein folding in this organelle.

EXPERIMENTAL PROCEDURES
Isolation of Human ERO1-L cDNA Clones-Approximately 3.4 ϫ 10 6 clones from a cDNA library derived from NT2-D1 human embryonal carcinoma cells differentiated by retinoic acid treatment (Stratagene, GmbH, Heidelberg, Germany) were screened using an EcoRI fragment of clone IMAGE: 0612711 (nucleotides 1-832). Positive plaques were purified and analyzed by standard procedures (25). The longest clone (clone F1) was sequenced entirely (accession no. AF081886).
ERO1 and ERO1-L Expression Vectors-For expression in mammalian cells, the complete coding sequence of ERO1-L was PCR amplified from the F1 clone with forward oligonucleotide GTGTTCTAGAGCCG-GAGCTGCAATG and two different reverse oligonucleotides (GTGTG-GTACCATGAATATTCTGTAACAAGT and GTGTGGTACCTTAATGA-ATATTCTGTAACAA) to obtain clones with or without the myc-6his tag of pCDNA3.1 fused in-frame with ERO1-L (pCDNA3.1ERO1-Lmyc and pCDNA3.1ERO1-L, respectively). The two PCR products were digested with XbaI and KpnI and cloned into pCDNA3.1(Ϫ) version A (Invitrogen, San Diego, CA). The inserts from the two clones were sequenced entirely. For expression in yeast, the inserts of pCDNA3.1ERO1-Lmyc and of pCDNA3.1ERO1-L were excised with XbaI and PmeI and cloned in the XbaI and HindIII sites (the latter filled in by Klenow) of pVT102-U (26), to yield pVTERO1-L and pVTERO1-Lmyc. The S. cerevisiae ERO1 (a kind gift of A. Frand and C. Kaiser, MIT, Boston, MA) was cloned in the same vector as follows. The ERO1 gene fused to a triple myc tag was PCR amplified from plasmid pAF82 (18) by the following oligonucleotides: yERO1-FW: GTGTGGATCCATGAGATTAAGAACCGCCATTGC yERO1-RV: GAAGATGGTACCGGTGATAAGTTCAAGAATG. The PCR product was digested with BamHI and KpnI and cloned into pCDNA3.1 to yield pCDNA3.1yERO1-3myc. The ERO1-3myc insert was excised with XbaI and PmeI and cloned in pVT102-U as for ERO1-L.
To generate the COOH-terminal truncated variant of yeast ERO1 (ERO1⌬C), the region of the gene corresponding to the first 444 residues was amplified from pAF82 by oligonucleotides yERO1-FW (see above) and CCACTCTAGAGTCATAACCTTTTCCCGTACATTTTTTCG. The PCR product was digested with BamHI and XbaI and cloned into pVT102-U to yield pVTERO1⌬C.
The mutated EcoRI fragments were then reinserted into pCDNA3.1ERO1-Lmyc and pCDNA3.1ERO1-L. The presence of the proper mutation in all the clones was confirmed by sequencing. Mutated ERO1-L variants were transferred into the yeast vector pVT102-U as described above for wild type (wt) ERO1-L.
Transcription and Translation in Vitro-Transcription reactions were carried out as described (28). Plasmid pCDNA3.1ERO1-Lmyc was linearized with BbsI and transcribed using T7 RNA polymerase (Promega, Southampton, U. K.). Reactions (50 l) were incubated for 2 h at 37°C followed by phenol/chloroform extraction and ethanol precipitation. RNA was resuspended in 50 l of RNase-free water containing 0.5 mM DTT and 40 units of RNasin (Promega).
RNA was translated using a rabbit reticulocyte lysate (FlexiLysate, Promega) for 60 min at 30°C. The translation reaction (25 l) contained 16.5 l of reticulocyte lysate, 0.6 l of 100 mM KCl, 0.5 l of 1 mM amino acids (minus methionine), 15 Ci of L-[ 35 S]methionine (NEN Life Science Products, Driech, Germany), 1 l of in vitro transcribed RNA, and 0.75 l of dog pancreas microsomes (29). After translation, microsomes were isolated by centrifugation for 5 min at 100,000 ϫ g twice through a 0.5 M sucrose cushion. To analyze the glycosylation status of translated products, microsome pellets containing 35 S-labeled ERO1-Lmyc were resuspended in 25 l of water containing 0.1% SDS and 0.5% 2-mercaptoethanol and heated for 2 min at 100°C. The sample was cooled on ice, and 0.5% (w/v) n-octyl glucoside was added to counteract the inhibitory effect of SDS. The sample volume was increased to 50 l, which included incubation buffer (20 mM sodium phosphate, pH 7.5, 0.5 mM phenylmethanesulfonyl fluoride, and 50 mM EDTA), and 1.5 milliunits of Endo-F and peptide N-glycosidase-F (Oxford GlycoSciences, U. K.). Samples were incubated for 16 h at 37°C, resolved by SDS-PAGE, and visualized using a Fujix Bas 2000 Bioimager.
Cells growing logarithmically at 24°C were resuspended at 10 A 600 /ml in SD medium and preincubated at 38°C for 25 min. DTT was added to a final concentration of 62.5 M for the last 10 min of this preincubation. Cells were pulsed with 20 Ci/A 600 Tran 35 S-label mixture (Amersham Pharmacia Biotech, Milano, Italy) for 6 min and chased for different times at 38°C. Samples of 3 A 600 were collected in 20 mM NaN 3 , pelleted, mixed with 50 mg of acid-washed glass beads, and diluted in 100 l of TNET (100 mM Tris-HCl, pH 7.5, 100 mM NaCl, 5 mM EDTA, 1% Triton X-100) containing 1% SDS. Lysis was achieved by vigorous vortexing for 1 min followed by a 1-min incubation on ice; after five such cycles, lysates were heated at 95°C for 5 min, cooled at room temperature, and diluted with 900 l of TNET. After centrifugation, supernatants were precleared with 50 l of 30% protein A-Sepharose and immunoprecipitated with anti-CPY (gift of Dr. H. Riezman, Basel, CH; 1 l/A 600 unit) and protein A-Sepharose for 12 h at 4°C with gentle rotation. Immunoprecipitates were washed four times with 1 ml of TNET, resolved by SDS-PAGE, and analyzed by fluorography.

Isolation of Mammalian Homologs of S. cerevisiae ERO1-
The yeast ERO1 nucleotide sequence (z50178) was used to search for related expressed sequence tags (ESTs) on the EST division of the GenBank data base through the BLAST algorithm (31) accessed through the NCBI web site. EST IMAGE clone 1485-o16, which apparently extended most 5Ј (accession no. AF123887), encodes a protein similar to Ero1p but lacking a translation initiation sequence. The screening of a human cDNA library (see "Experimental Procedures") led to the isolation of four clones, the longest of which, F1 (accession no. AF081886), contains a complete coding sequence for a human homolog of ERO1, which we called ERO1-L (for ERO1-Like). The putative translational start site was established based on the existence of a stop codon 155 base pairs upstream of an ATG located in a good initiation context (32).
The predicted human gene product is a 468-amino acid polypeptide (Fig. 1A), showing 48.5% similarity and 36.5% identity to yeast Ero1p. The algorithm SignalP (33) predicts a signal sequence for ER translocation at the NH 2 terminus, with a potential cleavage site between residues 23 and 24. A putative EF-hand calcium binding domain (34) is present in positions 159 -171. Two N-glycosylation sites are predicted at positions 280 and 384, respectively. Residues 391-397 contain the highly conserved CXXCXXC motif, suggested to be important for bioactivity (18,19).
A murine EST clone was identified (EST IMAGE 1294317) and sequenced (accession no. AF144695) and appears to contain a full-length coding sequence 87.3% identical to human ERO1-L at the nucleotide level. The predicted mouse protein is 92.7% identical to the human sequence. Fluorescence in situ hybridization experiments revealed that ERO1-L is located on human chromosome 14 at q22.1 and in the synthenic 14 C-D region in the mouse (Fig. 1, B and C). To acquire a broader view of the ERO1 gene family in different species, we identified and sequenced an EST (cDNA clone LD02945) coding for a Drosophila melanogaster ERO1-L gene (accession no. AF125280). The sequence of the predicted Drosophila ERO1-L polypeptide is shown in Fig. 1A, together with the products of the Schizosaccharomyces pombe and Arabidopsis thaliana genes.
ERO1-L Is an ER Resident N-Glycoprotein-To determine the intracellular localization of the ERO1-L gene product, human ERO1-L was tagged and expressed in COS-7 cells. The pattern obtained with anti-myc largely overlapped with the distribution of ConA, a lectin that recognizes glycoproteins with terminal mannose residues, and therefore decorates primarily the ER of permeabilized cells ( Fig. 2A). Similar results were obtained in 293T cells (not shown). As an additional marker of the ER we chose erGFP(S65T), a VH-CH1-green fluorescent protein chimeric molecule retained in the ER (35). The co-localization with ConA and erGFP indicates that exogenous ERO1-Lmyc accumulates primarily in the ER. This was confirmed further by endoglycosidase sensitivity assays. A mobility shift was observed when lysates from transfected COS-7 or 293T cells were treated with Endo-H or Endo-F (Fig. 2B). Hence, ERO1-Lmyc molecules are N-glycosylated and remain in an Endo-H-sensitive configuration, consistent with their localization in the ER.

In Vitro Translated ERO1-Lmyc Is Translocated across Microsomal Membranes and Has the Properties of an Integral
Membrane Protein-To ascertain whether human ERO1-Lmyc is a soluble or membrane protein, in vitro transcription/translation assays were performed. Translation in rabbit reticulocyte lysate yielded a 58-kDa band (Fig. 2C, lane 1). In the presence of microsomes, an additional product of slower mobility became detectable (lane 3). This slower migrating polypeptide was sensitive to glycosidases (lane 2), indicating that a fraction of ERO1-Lmyc synthesized in the presence of microsomes undergoes N-linked glycosylation. Translocation into microsomes was confirmed by protease protection experiments (Fig. 2D). The low mobility polypeptide synthesized in the presence of microsomes, which was sensitive to digestion with glycosidases, was not digested by proteinase K (lane 2) unless membranes were solubilized by detergent (Triton X-100, lane  3). The faster migrating polypeptides seen after translation in the presence of microsomes (lane 1) were sensitive to proteolysis (lane 2) and therefore correspond to untranslocated molecules.  (39). Fluorescence in situ hybridization was performed as described (40).
The electrophoretic mobility of ERO1-Lmyc translated in the absence of microsomes was indistinguishable from that of the translocated and deglycosylated translation products (Fig. 2C,  compare lanes 1 and 2), suggesting that the signal peptide of ERO1-Lmyc may not be cleaved. This uncleaved signal sequence could function as a transmembrane domain. To ascertain whether ERO1-Lmyc is indeed integrated into the membrane, primed microsomes were incubated with sodium carbonate alone or in the presence of salt or urea (Fig. 2E). In general, integral membrane proteins are not extracted by these treatments unless detergent is added and therefore fractionate with the pellets, whereas soluble and peripheral membrane molecules accumulate in the supernatants. Clearly, most glycosylated ERO1-Lmyc fractionated with the membranes (Fig.  2E, lanes 3, 5, 7, and 9), unless microsomes were solubilized with detergent (Triton X-100, lanes 11 and 12). Unexpectedly, the faster migrating band was poorly solubilized even in the presence of detergent, suggesting that untranslocated ERO1-Lmyc molecules may form insoluble aggregates.
Taken together, these findings indicate that glycosylated  2 and 3) of dog pancreas microsomes. An aliquot was also incubated with peptide N-glycosidase-F/endoglycosidase F (lane 2). Samples were resolved by SDS-PAGE (12.5% acrylamide). D, ERO1-Lmyc is translocated into the lumen and is resistant to protease digestion. ERO1-Lmyc was translated in vitro as above in the presence of microsomes. After translation, microsomes were isolated by centrifugation at 100,000 ϫ g through a 0.5 M sucrose cushion. Pellets were resuspended in KHM buffer (10 mM potassium acetate, 2 mM magnesium acetate, 20 mM Hepes, pH 7.2), treated with proteinase K alone (lane 2) or with 1% Triton X-100 (lane 3), and resolved by SDS-PAGE as in C. E, ERO1-Lmyc associates strongly with the membrane. ERO1-Lmyc was translated in vitro as in D. Microsomes were isolated by centrifugation at 15,000 ϫ g and resuspended in KHM buffer (lanes 1 and 2), 200 mM sodium carbonate alone (lanes 9 and 10) or supplemented with 50 mM NaCl (lanes 3 and 4), 500 mM NaCl (lanes 5 and 6), or 4 M urea (lanes 7 and 8), or KHM and 1% Triton X-100 (lanes 11 and 12). Samples were then centrifuged at 100,000 ϫ g, divided into pellet (P) or supernatant (S) fractions, and resolved by SDS-PAGE as above. ERO1-Lmyc is either itself an integral membrane protein or forms a strong interaction with an integral membrane protein.
Functional Complementation of the Yeast ero1-1 Mutant by Human ERO1-L-To explore the functional properties of ERO1-L, we expressed the human coding sequence, with or without a myc tag, in the thermosensitive ero1-1 yeast mutant strain (CKY559). These cells do not grow at 37°C and display increased sensitivity to DTT and inefficient ER-Golgi transport of disulfide-rich proteins (18).
First, we tested the ability of human ERO1-L to rescue the temperature sensitivity of the ero1-1 mutant. Thus, ero1-1 was transformed with different vectors and grown at either 24°C or 37°C. Clearly, the human gene allowed growth at 37°C (Fig.  3A). A striking difference between the human and yeast genes is the presence in the latter of a long COOH-terminal region missing from the human sequence (see Fig. 1). The ability of the human sequence to rescue the ero1-1 growth defect suggests that the COOH-terminal domain is not essential for viability. In agreement with this, a mutant of the S. cerevisiae ERO1 gene truncated at position 444 (ERO1⌬C) was able to rescue ero1-1 viability at 37°C.
The high conservation of the CXXCXXC motif in all members of the ERO-1 family suggests that it may play an important functional role. To test this hypothesis, variants of ERO1-L were generated in which the three conserved cysteines were replaced individually by alanines. Mutating cysteines 394 and 397 was detrimental to the ability of ERO1-L to rescue ero1-1 thermosensitivity, whereas the mutant C391A behaved like wt ERO1-L (Fig. 3A). As determined by Western blot analysis (not shown), the expression levels of the three mutants were comparable to those of wt ERO1-Lmyc.
In agreement with a role for Ero1p in maintaining an oxidative environment in the ER, the ero1-1 mutant is more sensitive to DTT than the wt strain (18,19). The expression of human ERO1-L partially alleviated ero1-1 DTT sensitivity (Fig. 3B). Also in this assay, no differences were observed between ERO1 and ERO1⌬C, suggesting that the COOH-terminal region is not crucial for the protective effects toward DTT. As observed in the thermosensitivity assays, mutations in the second and third cysteines of the CXXCXXC motif were detrimental to the activity of ERO1-L, whereas the phenotype of the C391A mutant was indistinguishable from wt ERO1-L.
Lastly, we tested the ability of ERO1-L to complement Golgi transport of CPY. CPY offers a convenient tool for monitoring the redox conditions in the ER because it contains intrachain S-S bonds, whose formation is required for folding and export from the ER (36). After transit through the Golgi, CPY is targeted to the vacuole where it is proteolytically cleaved, yielding the mature (m) form of the polypeptide.
When ero1-1 was transformed with the vector alone and grown under nonpermissive conditions, CPY accumulated in  1-4), ERO1-Lmyc (lanes 5-8), or ERO1 (lanes 9 -12) growing logarithmically at 24°C were resuspended at 10 A 600 /ml in SD medium and incubated at 38°C. After 15 min, DTT was added to the cultures to a final concentration of 62.5 M. After a further 10 min at 38°C, cells were pulse labeled for 6 min with radioactive cysteine and methionine and chased for the indicated times. CPY was immunoprecipitated and resolved by SDS-PAGE under reducing conditions. P1 and M indicate the ER and vacuolar forms of CPY, respectively. B, oxidative maturation of CPY in ero1-1 transformants. Cells were pulse labeled and chased for 30 min as described above before lysis in the presence of 10 mM iodoacetamide. Immunoprecipitates were treated with (ϩ) 100 mM DTT or not (Ϫ) for 5 min at 95°C, alkylated with 400 mM iodoacetamide, and resolved by SDS-PAGE. P1 red and P1 ox indicate ER CPY, reduced or oxidized; M red and M ox indicate vacuolar CPY, reduced or oxidized. the ER in the precursor (p) form (Fig. 4A, lanes 1-4). Transformation with human ERO1-L (lanes 5-8) allowed the transport of CPY along the exocytic pathway, although at a slower rate than yeast ERO1 (lanes 9 -12).
Thus, ERO1-L allows intracellular transport of CPY, otherwise retained in the ER. To determine whether this reflects intrachain disulfide bond formation, we exploited a gel mobility assay that allows discrimination of oxidized and reduced CPY. When ero1-1 was transformed with the vector alone, no differences in the electrophoretic mobility of CPY were observed, suggesting that CPY was retained in the ER in the reduced state (lanes 1 and 2). In contrast, a mobility shift was evident in sec18 (compare lanes 7 and 8), a mutant in which, because of a general ER-Golgi transport defect (37), CPY accumulates in the ER in the oxidized form. As described previously (18), transformation of ero1-1 with S. cerevisiae ERO1 allowed CPY oxidation and cleavage (lanes 5 and 6). When ero1-1 was transformed with ERO1-L, some CPY remained reduced in the ER. However, most of the protein accumulated in the oxidized vacuolar form (lanes 3 and 4), indicating that human ERO1-L is able to promote oxidative protein maturation in the ER.

DISCUSSION
The striking homologies among different species suggest that the ERO1 gene is endowed with an important function maintained across evolution. The human and mouse genes are 87.3% identical at the nucleotide level and 92.7% identical at the protein level. Because they map to synthenic regions (Fig.  1, B and C), they are likely to be orthologous. Blocks of identical residues are distributed throughout the sequences (Fig. 1). The yeast sequence is unique in having a long COOH-terminal tail protruding for 139 residues with respect to the human and mouse sequences. The A. thaliana sequence also displays a COOH-terminal portion, shorter than the yeast tail and unrelated in sequence. Because a truncated ERO1 mutant appears to be active in both thermosensitivity and the DTT sensitivity assays, the function of this COOH-terminal tail remains to be clarified. One of the largest blocks of sequence identity revolves around the CXXCXXC motif. Less conserved are the NH 2terminal portions of the sequences. However, all share hydrophobicity and are predicted to act as ER-targeting signal sequences. Therefore, like the S. cerevisiae Ero1p (18), it is likely that the other members of the family are also localized in the ER.
Several lines of evidence indicate that this is indeed the case for the human ERO1-L gene product. First, immunofluorescence experiments reveal that a transfected myc-tagged ERO1-L expressed in simian COS-7 or human 293T cells colocalizes with two ER markers, a chimeric VH-CH1-GFP (35) and ConA, a lectin that preferentially reacts with N-glycans bearing terminal mannose, characteristic of ER glycoproteins. Second, ERO1-Lmyc undergoes N-glycosylation, a modification restricted to proteins synthesized in the ER. Consistent with an ER localization, the same mobility shift is observed upon treatment with Endo-F and Endo-H, two enzymes that cut all or only immature N-glycans, respectively. Third, in vitro translated ERO1-Lmyc is translocated into microsomal membranes when the latter are present during synthesis and a mobility shift is observed, which is abolished by endoglycosidase digestion, confirming that the newly synthesized polypeptide is N-glycosylated.
Unexpectedly, the mobility of the deglycosylated product is indistinguishable from the protein synthesized in the absence of added microsomal membranes. This suggests that the signal sequence is not cleaved and could act as a transmembrane domain. In support of this, carbonate washing of the membranes in the presence of urea or high salt failed to extract ERO1-Lmyc. Because no other candidate transmembrane re-gions are detectable in the ERO1-L sequence, the conservation of the hydrophobic leader could therefore explain the association of ERO1-L with microsomal membranes. Alternatively, this association might be established by strong interactions with membrane protein(s). Further investigation is required to clarify how ERO1-L associates with membranes and is retained in the ER.
ERO1-L is expressed, in varying amounts, in all cell lines and tissues examined (data not shown), suggesting that this gene may be involved in a general cellular function. The yeast complementation experiments indicate that this function is likely to be related to oxidative protein folding in the ER. Indeed, ERO1-L is able to complement several phenotypic traits of the ero1-1 mutant. ERO1-L completely restores the ability of ero1-1 to grow at 37°C and partially confers resistance to DTT. A clear recovery of function is also observed for the oxidative maturation of CPY in the ero1-1 mutant (Fig. 4). Therefore, ERO1-L can perform, although with reduced efficiency, the oxidative function of Ero1p in yeast cells. The possibility that the human protein may act by rescuing the activity of the defective endogenous ERO-1 yeast gene cannot be formally ruled out. However, the observation that an intact CXXC motif (394 -397) is required for ERO1-L activity argues against a simple "in trans" reactivation of the endogenous ERO1 gene.
Substitutions to alanine of the second or the third cysteine of the ERO1-L CXXCXXC motif (cysteines 394 and 397) are detrimental to ERO1-L activity. In contrast, we were unable to detect loss in activity for the C391A ERO1-L mutant with respect to the wt molecule in any of the assays performed. This is somewhat surprising in view of the conservation of this residue in different species (see Fig. 1A). A prediction of ERO1-L secondary structure with the PREDATOR algorithm (not shown) locates Cys-394 in a loop and Cys-397 at the beginning of an ␣-helix. This arrangement is also found in thioredoxin, DsbA, and protein disulfide isomerase. This observation, together with the mutagenesis data, is consistent with the possibility that cysteines 394 and 397 may constitute a classical CXXC motif (38), crucial for the redox activity of ERO1-L. Beyond the three residues of the CXXCXXC motif, ERO1-L contains additional cysteines, some of which are also highly conserved, which may contribute to the redox activity. Further mutagenesis will be required to elucidate the functional relevance of these residues.
In conclusion, our data suggest that the molecular machine responsible for oxidizing the ER might be remarkably similar in yeast and humans. It will be important to determine whether also in eukaryotes the oxidation of ERO1-L is linked to the respiratory chain as the ultimate electron acceptor system.