The N-terminal cysteine pair of yeast sulfhydryl oxidase Erv1p is essential for in vivo activity and interacts with the primary redox centre Go¨ tz Hofhaus 1 , Jeung-Eun Lee 1 , Ivo Tews 2 , Beate Rosenberg 3 and Thomas Lisowsky 3 1 Institut fu ¨ r Biochemie und Biologisch-Medizinisches Forschungszentrum and 3 Botanisches Institut, Heinrich-Heine-Universita ¨ t Du ¨ sseldorf, Germany; 2 Biochemiezentrum Heidelberg, Germany Yeast Erv1p is a ubiquitous FAD-dependent sulfhydryl oxidase, located in the intermembrane space of mito- chondria. The dimeric enzyme is essential for survival of the cell. Besides the redox-active CXXC motif close to the FAD, Erv1p harbours two additional cysteine pairs. Site- directed mutagenesis has identified all three cysteine pairs as essential for normal function. The C-terminal cysteine pair is of structural importance as it contributes to the correct arrangement of the FAD-binding fold. Variations in dimer formation and unique colour changes of mutant proteins argue in favour of an interaction between the N-terminal cysteine pair with the redox centre of the partner monomer. Keywords: sulfhydryl oxidase; mitochondrial Erv1p; redox- active CXXC; dimer formation; cysteine mutants. Disulfide bonds are important for the structure and function of proteins in eukaryotes [1], prokaryotes [2] and even viruses [3]. Several enzymes are known to catalyse dithiol– disulfide transfer reactions between proteins, but enzymes like sulfhydryl oxidases that are capable of synthesizing disulfide bonds de novo are less common [4]. In general, these enzymes exist as homodimers, which depend on FAD as a cofactor, use oxygen as final electron acceptor and contain a CXXC motif that is involved in the primary redox-reaction [4,5]. The Saccharomyces cerevisiae protein Erv1p (essential for respiration and vegetative growth; encoded by the gene ERV1) and the human homologue Alrp (augmenter of liver regeneration) are sulfhydryl oxidases in the intermembrane space of mitochondria [6]. They are found in a large number of different cell-types and tissues [7]. Their activity is essential for the survival of the cell, for the biogenesis of mitochondria and for the supply of cytoplasmic proteins with mitochondrially assem- bled iron–sulfur clusters [6]. However, their natural sub- strate proteins are not known today. In yeast, a second sulfhydryl oxidase, termed Erv2p, has been identified in the endoplasmic reticulum [8,9]. The N-terminal parts of Erv1p and Erv2p are very distinct (see Fig. 1). In contrast, the C-terminal parts of Erv1p and Erv2p, which include the redox-active centre and the FAD-binding domain, are similar (30% identity) [10]. Recently, the structure of a proteolytic fragment of Erv2p, corresponding to the conserved region, was solved (Fig. 1, 2). A CGC motif at the very C-terminus was described as part of a flexible arm that exchanges the de novo synthesized disulfide bridge with substrate proteins [11]. Members of the mitochondrial Erv1/ Alr protein family lack such a CGC motif (Fig. 1), but instead consistently contain an additional CXXC motif in the N-terminal domain. In order to assess the roles of all individual cysteine residues of Erv1p, we changed them to serines by site-directed mutagenesis. By working with yeast Erv1p, this analysis was limited to six cysteines instead of the nine present in human Alrp and in addition, the mutated yeast proteins can be easily checked for in vivo activity. Materials and methods Strains and plasmids The following strains of S. cerevisiae were used: JRY 675 (MATa, ura – , his4-519, Dleu2) served as wild-type and the erv1-ts strain was pet492–6 A (MATa,ura3-52,Dleu2; pet492ts) [12]. The yeast strain with one disrupted copy of ERV1 was JRY 675, 2n (MATa/MATa, ura3-52/ura3-52, his4-519/his4-519, Dleu2/Dleu2, ERV1/erv1::LEU2) [12]. Plasmids were: pET-24a(+) (Novagen) and pRS416 [13]. Escherichia coli strains were: DH5-a [14] and BL21(+) (Novagen). Gene constructs of ERV1 and in vitro mutagenesis The complete yeast ERV1 gene and a shorter fragment encoding the 15 kDa C-terminus (DN-Erv1p) have been cloned into the hexahistidyl-tag vector pET-24a(+) as described previously [15]. The pET-24a(+) harboring the complete ERV1 gene was used for site-directed mutagenesis (PCR-based site-directed mutagenesis kit Excite; Strata- gene) with the following primers: C30S (forward 5¢-CGATCATGTAACACCCTAC-3¢/reverse 5¢-CGAAG Correspondence to G. Hofhaus, Institut fu ¨ r Biochemie und Biologisch- Medizinisches Forschungszentrum, Heinrich-Heine-Universita ¨ t Du ¨ sseldorf, Universita ¨ tsstraße 1, 40225 Du ¨ sseldorf, Germany. Fax: + 49 211 81 15310, Tel.: + 49 211 81 15189, E-mail: hofhaus@uni-duesseldorf.de (Received 20 December 2002, revised 4 February 2003, accepted 12 February 2003) Eur. J. Biochem. 270, 1528–1535 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03519.x GTTTGCCATCTTCG-3¢), C33 (forward 5¢-CTACTT GACTTTCAGTACGTGACC-3¢/reverse 5¢-GGTGGTA GATGATCGGCAAGGTTTGC-3¢), C130S (forward 5¢- CCAATTGGTGTGCTAAAGACTTTG-3¢/reverse 5¢-AA GGATAAATATGTGAGAAGATATTC-3¢), C133 (for- ward 5¢-CTTTGAAAAATATATCAGAGAAAATG-3¢/ reverse 5¢-TCTTTAGCAGACCAGTTGTAAGG-3¢), C159S (forward 5¢-GCCCACAATAAAGTCAATAAG AAAT-3¢/reverse 5¢-CTCAGACATCCACCTCCCAAG TTCTT-3¢),C176S(forward5¢-CTCCAATTTCTGG GAAAAAAGATGGAAG- 3¢/reverse 5¢-TCAAATTTGG GCTTCCTCAATTTC-3¢). Sequences of the primers con- tained changes of restriction sites that allowed identification of the successfully mutated plasmids. Purification of Erv1p-6His proteins from E. coli All pET24a(+) constructs were expressed in E. coli strain BL21(+). The presence of the hexahistidyl-tag at the C-terminus of the proteins allowed rapid purification with nickel nitrilotriacetic acid-agarose according to the standard protocols for the isolation under native conditions in phosphate buffer (50 m M NaCl, 50 m M KH 2 PO 4 ,10m M imidazole, pH 7.5) (Qiagen). Proteins were bound to nickel agarose and eluted with 50 m M NaCl, 50 m M KH 2 PO 4 , 200 m M imidazole, pH 7.5. Purification to homogeneity was verified by SDS/PAGE and by Western blot analysis. Spectroscopy of yeast Erv1p-6His The visible spectra of the purified proteins in elution buffer were recorded with an S-10 diode-array photometer (Zeiss). Under identical conditions a reference of 15 m M purified FAD (Sigma) was measured. To test the reversible reduc- tion of protein-bound FAD samples of Erv1p and mutant proteins in elution buffer (300 m M imidazole, 50 m M phosphate pH 7.5, 200 m M NaCl) were completely reduced with a few grains of Na-dithionite. Release of protein-bound FAD by heat and acid treatment Protein aliquots were adjusted to an A 460 of 0.200 before releasing the FAD by heat and treatment with 5% trichloric acetic acid. Denatured proteins were pelleted by centrifu- gation. Free FAD in the supernatant was determined by measuring the A 450 . The measurements of the released FAD were very similar to the values obtained from the A 460 in the purified proteins. The only exception was the mutant protein C133S that gave lower values in these experiments because a substantial amount of the FAD was not liberated and therefore found in the pellet together with the denatured protein. EPR spectroscopy EPR measurements were conducted with a Brucker EMX 1/6 spectrometer operating at 9.2 GHz. Sample temperature was maintained at 4 K by an Oxford instruments EPR-9 helium flow cryostat. The magnetic field was calibrated using strong or weak pitch standard. Spectra of the oxidized and reduced (dithionite) enzymes were observed. Enzyme assay for sulfhydryl oxidase activity Erv1 protein of the full-length and short form and from the respective cysteine mutants were adjusted to 10 pmol protein-bound FAD per 100 lL reaction mixture. The enzyme reaction was started in NaCl/P i buffer (68 m M NaCl, 75 m M potassium phosphate buffer pH 7.5 contain- ing 3 m M EDTA) together with dithiothreitol substrate that corresponded to 50 nmol reduced thiol groups. Aliquots of 100 lL of this reaction mixture were used for each time point. For determination of the thiol content the 0.1 mL samples were diluted with 800 lLofNaCl/P i buffer and then 100 lL5,5¢dithio-bis(2-nitrobenzoic acid) were added to a final concentration of 10 m M . After 2 min the extinction at 412 nm was measured and the thiol content was calculated using an extinction coefficient of 14 m M )1 Æcm )1 [16]. The initial content of thiol groups in the reaction mixture was determined from a sample without enzyme. Western analysis For immunological studies, aliquots with about 200 ng of purified Erv1p and mutant proteins were applied to 4%-12% nonreducing SDS/polyacrylamide gels (Novex/ Invitrogen). Protein samples were set up with or without 20 m M dithiothreitol as indicated. The primary anti-His 5 Ig (Qiagen) was detected by alkaline phosphatase-conjugated secondary antibodies and chemiluminescence. Miscellaneous methods Plasmid DNA was isolated from E. coli by alkaline lysis and using the Qiagen kit. Purification, restriction enzyme digestion, ligation and analysis of DNA or PCR products on agarose gels were performed as described previously [17]. DNA sequences of all gene constructs were controlled by completely sequencing the reading frames (MWG-Biotech company). The modelling and the energy minimizing of the Fig. 1. Comparison of Erv1p, DN-Erv1p and Erv2p. The figure displays the corresponding cysteine residues and identity values for the N- and the C-terminal parts of the proteins [8,10]. The hatched area marks the part of Erv2p whose structure has been solved [11]. Ó FEBS 2003 Cysteine mutagenesis of yeast ERV1 (Eur. J. Biochem. 270) 1529 Erv1p model were carried out using the WHAT IF program [18,19]. Results The C-terminal FAD binding domain of Erv1p and Erv2p are conserved Partial structural information for the homologous Erv2p is available in the databases [11]. The crystallized protein is a 119 residue proteolytic fragment of Erv2p. This fragment represents the highly conserved C-terminal region with an identity of 30% between Erv1p and Erv2p (see Fig. 1). To investigate whether Erv1p can fold into the same overall structural assembly, we carried out a modelling study using the molecular modelling software package WHAT IF [18–20]. The high resolution of the 3D structure with 1.5 A ˚ and the high sequence identity make this study reliable. Briefly, the sequence of Erv1p was aligned to the Erv2p sequence. The corresponding Erv1p residues were modelled and the resulting structure was energy minimized using the software (Fig. 2) WHAT IF [18]. The modelling suggests that Erv1p and Erv2p can obtain the same structural fold. There are no clashes of amino acid side chains in the core of the structure, and most mutations occur on the protein surface. The FAD binding groove is conserved with most ligand binding residues identical. However, mutations result in minor differences in the homodimer interface. A fundamental difference between the proteins concerns the CGC motif near the C-terminus of Erv2p. The structure of Erv2p revealed two possible conformations of the C-terminal arm with one conformation bringing the cysteine close to the redox centre of the partner monomer. This finding sugges- ted that the CGC motif mediates redox shuffling between the redox centre and possible substrates [11]. This shuffling mechanism is not possible for Erv1p due to the lack of cysteines near the C-terminus. However, Erv1p contains an additional cysteine pair near the N-terminus that could functionally substitute the CGC motif. Therefore the systematic functional analysis of the six cysteines in yeast Erv1p was the major goal of this paper. All three pairs of cysteines in Erv1p are indispensable for in vivo activity The six cysteines of yeast Erv1p can be grouped into three pairs (Fig. 1): one pair near the N-terminus of the protein, the central pair close to the FAD binding motif and two cysteines in the C-terminal region that are 17 residues apart. For our experiments the wild-type ERV1 gene was extended such that the resultant Erv1p carried six histidine residues at the C-terminus, facilitating the purification of the proteins by chromatography on Ni-nitrilotriacetic acid agarose. Cysteine codons were then successively replaced by serine- triplets and exchanges were verified by restriction analysis and sequencing (see Material and methods). To check the in vivo activity of the mutated proteins we used a haploid yeast strain with a temperature-sensitive Erv1 protein [12]. Yeast cells were transformed with single copy pRS416 constructs containing the genes for the respective mutated proteins. At elevated temperatures of 36 and 38 °C the temperature-sensitive Erv1p is not functional and growth therefore depends on the expression of the genes from the plasmids. The data in Table 1 demonstrate that under these conditions only the wild-type protein supports growth at 36 °C, the restrictive tempera- ture for the ts-protein. The only cysteine mutant that is capable of supporting some growth at elevated temperatures is C30S. Because Erv1p exists as a dimeric protein in vivo it is possible that heterodimers between the temperature-sensi- tive protein and the C30S mutant protein lead to intra- molecular complementation (see Discussion). To exclude heterodimer formation, we repeated the experiment using tetrad analysis of an insertion mutant of the ERV1 gene (Table 2). The analysis is complicated by recombination events due to the position of ERV1 close to the centromere, but the results at higher temperature support the findings of the first experiments: only the C30S mutation allows some growth at higher temperatures. Tetrad analysis showed that Table 1. Complementation studies with erv1-ts. The erv1-ts mutant (pet492) was transformed with the single copy plasmid pRS416 bearing the respective Cys mutants (exchange of Cys against Ser: C-S) of ERV1. Complementation activity (+ growth/– no growth) was tested on glucose complete medium at 28, 36 or 38 °C after 4 days. Yeast strain Temperature (°C) 28 36 38 Wild-type + + + erv1-ts + – – erv1-ts + Erv1p C30S + + – erv1-ts + Erv1p C33S + – – erv1-ts + Erv1p C130S + – – erv1-ts + Erv1p C133S + – – erv1-ts + Erv1p C159S + – – erv1-ts + Erv1p C176S + – – Table 2. Complementation studies with Derv1. The diploid strain Derv1/ERV1 was transformed with the single copy plasmid pRS416 bearing the respective Cys mutants (exchange of Cys against Ser: C-S) of ERV1. After tetrad dissection complementation activity (+ growth/– no growth) for the haploid Derv1wastestedonglucose complete medium at 28, 36 or 38 °C after 4 days. Due to the close association of the ERV1 gene with the centromere frequent abnormal recombination between the genomic copy and the plasmid encoded ERV1 gene were observed. Therefore only complete tetrads with cor- rect genetic markers were used for evaluation of the phenotype listed in Table 2. Yeast strain Temperature (°C) 28 36 38 Wild-type + + + Derv1 – – – Derv1 + Erv1p C30S + – – Derv1 + Erv1p C33S – – – Derv1 + Erv1p C130S – – – Derv1 + Erv1p C133S – – – Derv1 + Erv1p C159S + – – Derv1 + Erv1p C176S + – – 1530 G. Hofhaus et al. (Eur. J. Biochem. 270) Ó FEBS 2003 at 28 °C the mutants C30S, C159S and C176S displayed a residual activity that allows survival, whereas the mutants C33S, C130S and C133S had no complementation activity. To investigate the in vitro activity of the enzyme, the mutagenized genes were overexpressed in E. coli and the proteins were purified to homogeneity taking advantage of the His-tag. In contrast to earlier activity measurements, sulfhydryl oxidase activity was measured with dithiothreitol, which was a good substrate in NaCl/P i ([21] and Material and methods). Due to the sulfhydryl oxidase activity the dithiothreitol is oxidized. At different time points, the remaining free thiol groups of the substrate were quantified with Ellman’s reagent [16]. The initial slope of the time course was used to calculate the turnover numbers for mutated and wild-type proteins. As shown in Fig. 3, most of the mutated proteins show no or strongly diminished in vitro sulfhydryl oxidase activities. Exchange of the two cysteine residues in the N-terminal part of the proteins seems to interfere with but not abolish the activity. It has been reported in previous studies that the N-terminally truncated form of Erv1p has the same in vitro activity as the wild-type enzyme with artificial substrates such as dithiothreitol or reduced lysozyme [15]; nevertheless, truncated Erv1p is not able to replace wild-type Erv1p in vivo [22]. C30 and C33 are involved in dimer formation Erv1p is isolated as a homodimer from yeast cells [15]. The dimeric form is probably important for function. SDS/ PAGE has shown that the dimer is stabilized by a disulfide bond that keeps the two monomers covalently linked under nonreducing conditions [15]. While the wild-type proteins exist as a mixture of dimers and monomers under these conditions, the DN-Erv1p is found exclusively in the monomeric form (Fig. 4). This already demonstrates the importance of the N-terminal domain for dimer-formation. Under nonreducing conditions all of the cysteine mutant proteins display only a small amount of monomeric form Fig. 3. Sulfhydryl oxidase in vitro activity of Erv1p and mutated enzymes. Activities of the purified enzymes were measured with dithiothreitol as an artificial substrate (for details see Materials and methods). Turnover numbers were calculated per protein-bound FAD molecule as determined spectroscopically. Three independ- ent measurements were performed for each enzyme preparation and the respective mean values were listed. SD are given by black bars. Fig. 2. The functional dimer of the proteolytic Erv2p fragment as determined by X-ray crystallography (pdb accession no. 1JR8 [11]) is presented. One monomer of the dimeric protein is shown in blue, the second monomer is presented in a simplified sketch (grey). Cysteine residues and the FAD are displayed in different shades of yellow. The C-terminal CGC motif (C176–C178) is in proximity to the postulated active site of the partner monomer and hence could participate in thiol-exchange. The modelled structure of the corresponding C-terminal part of Erv1p (red) is displayed on the right side of the figure. A C-terminal CGC motif is missing in Erv1p. Ó FEBS 2003 Cysteine mutagenesis of yeast ERV1 (Eur. J. Biochem. 270) 1531 with the exception of C133S. In comparison to the wild- type protein the C30S, C33S and to a lesser extend the C130S mutant proteins are predominantly found in the dimeric form. The absence of a high molecular smear points to proper protein folding of these mutants in contrast to mutants C133S, C159S and C176S that exhibit high molecular mass aggregates indicating nonspecific aggregation. Unique colour changes of the protein are associated with the exchange of certain cysteine residues for serine So far, the analysis has revealed the importance of all cysteine residues in the protein, but has not given any clues to their molecular function. Luckily, some mutations resulted in interesting colour changes. The wild-type protein and the short form of the protein exhibit an intensive yellow colour, due to the bound FAD (Fig. 5). The two most striking changes upon cysteine exchange are the black appearance of the C30S protein and the orange colour of the C130S mutant. Changing the other cysteine within these pairs did not result in matching colour changes. While the mutation C30S resulted in a black protein, C33S produced a wild-type colour; the orange colour of C130S is contrasted by the pale C133S and exchange of C159 against serine produced a colourless protein, while the exchange C176S resulted in a wild-type appearance. These diverse colour changes point to different biophysical properties of the two cysteines within all three pairs. The colour changes were analysed in more detail by spectroscopy. The C30S mutant revealed an additional absorbance around 580 nm, while for the C130S mutant the FAD peak around 460 nm seems to be broadened (Fig. 6). The spectra of the remaining coloured mutant proteins (not shown) were indistinguishable from the spectra of the wild-type protein whereas the colourless proteins did not exhibit maxima typical for FAD. EPR spectroscopy gave no indications of radicals or metals bound to the proteins (data not shown). All colours disappeared after reduction with dithionite. Upon reoxida- tion of the solution in air, the yellow wild-type colour appeared. Likewise, upon storage for several days the appearance of the mutated proteins turned towards the Fig. 4. Dimer formation of Erv1p and mutant proteins. Purified proteins were separated on a nonreducing SDS/PAGE with (+) or without (–) 20 m M dithiothreitol in the sample buffer. Detection of protein bands was carried out with a His 6 -antibody and chemiluminescence. Fig. 5. Colours of purified Erv1p and mutant proteins. Protein concentrations vary due to different expression levels of the mutant proteins in E. coli. Erv1p (5.9 mgÆmL )1 ), C30S (36 mgÆmL )1 ), C33S (10.8 mgÆmL )1 ) C130S (14.1 mgÆmL )1 ), C133S (4.1 mgÆmL )1 ), C159S (9.9 mgÆmL )1 ), C176S (6.0 mgÆmL )1 ), DN-Erv1p (4.1 mgÆmL )1 ). Fig. 6. Spectra for wild-type Erv1p, C30S and C130S. Purified proteins were analysed in a Zeiss S10 diode-array spectrophotometer. The absorption between 320 and 700 nm is shown. Spectra are interpreted qualitatively only because the FAD content is different and the molar absoprtion coefficient might be different for mutant proteins. Protein concentrations are: 2.2 mgÆmL )1 (Erv1p), 5.1 mgÆmL )1 (C30S) and 4.2 mgÆmL )1 (C130S). 1532 G. Hofhaus et al. (Eur. J. Biochem. 270) Ó FEBS 2003 wild-type colour, possibly due to oxidation by air. So far, we have not been able to restore the initial colour of the mutant protein once it is lost. Calculation of the FAD content of the purified mutant proteins turned out to be difficult due to variations in FAD/ protein ratios of the recombinant proteins. Control experi- ments with liberated FAD from heated and acid-treated samples gave similar values. It appears that growth conditions and gene expression of the recombinant proteins cause the variations. A reproducible qualitative finding was that C133S and C159S always contained a substantially lower amount of protein-bound FAD than all other mutant proteins. The FAD content of the mutants C30S and C33S were similar to that of the wild-type proteins arguing against a general misfold of the proteins caused by the mutations. Discussion The CXXC motif at the reaction centre is indispensable for the interaction with FAD The interpretation of the observed colour changes is facilitated by the available structural data for Erv2p [11]. As can be deduced from Fig. 1 the crystallized part of Erv2p contains the conserved FAD-binding domain and four of the six cysteines present in Erv1p. The structure (see Fig. 2) reveals that the redox-active residues C130, C133 are in close proximity to the FAD, with the C-terminal cysteine (C133) being closer to the flavin structure. For the reduced central CXXC motif an intermediate has been suggested [23], where a thiol anion is stabilized by a charge-transfer- complex with the bound FAD. Subsequently, the FAD is reduced and a disulfide bridge is formed between the cysteines. Replacement of C130 by serine interferes with the second step and thereby may stabilize the charge transfer complex of C133 with the FAD resulting in the orange colour of the mutant protein. However, a similar colour change was not reported for the replacement of the corresponding cysteine by alanine in Erv2p [11]; thus it is possible that the serine residue also contributes to colour formation. In any case, the overall reaction is impaired and the mutated enzyme does not show any activity in vitro or in vivo. This phenotype is in agreement with the finding that the C130–C133 pair is part of the primary redox-active centre. Cysteine 159 and 176 stabilize the FAD-binding domain The second pair of cysteines in the crystallized part of Erv2p is separated by 17 residues and the exact distance is found between the corresponding cysteines of Erv1p. The reason for this exact match is clear from the structure of Erv2p, because the disulfide bridge between these residues has to tether the short helix 5 to the four-helix-bundle [11]. Replacing C159 by a serine destroys the disulfide bridge and the introduction of a polar group probably interferes with binding of the adenine portion of the FAD that is normally attached to this region of the protein. Conse- quently, the capability of the mutant protein for binding FAD is affected. Although the corresponding exchange of C176S destroys the same disulfide bridge, it has a less drastic effect. Due to its more peripheral position in a part of the protein without a pronounced secondary structure, this introduced serine might be accommodated without disturbing the FAD binding too much. Although FAD is bound to the protein, structural adjustments are not perfect, as shown by the low in vitro enzymatic activity. In agreement with this, one also observes a temperature- sensitive in vivo activity (Table 2). The N-terminal CXXC motif is essential for the function of Erv1p in vivo Like DN-Erv1p, the C33S mutant protein cannot function- ally replace the wild-type enzyme in vivo. At least for the point mutation, localization in an incorrect subcellular compartment is unlikely to explain this finding. The similar FAD binding and the absence of unspecific, large aggregates on the SDS/PAGE argue against a general misfold of the protein. Thus, the fact that the mutant cannot complement the wild-type enzyme in vivo (Tables 1 and 2) provides genetic evidence for an essential involvement of the N-terminal cysteine pair in the in vivo function. A possible function of this N-terminal CXXC motif in Erv1p is suggested by the discovery of a CGC motif as part of the flexible C-terminal arm in the related Erv2p [11]. In one conformation, this arm brings the CGC close to the redox-active centre of the other monomer, while in the second conformation the arm reaches out to the open space. Thus, the CGC motif on the flexible arm might exchange the de novo synthesized disulfides from the CXXC motif of the primary redox-active centre and passes them on to specific substrates. Erv1p lacks the C-terminal CGC motif, but possesses the additional CXXC motif near the N-terminus. Based on the structure of Erv2p the authors have already speculated that in Erv1p the N-terminal domain could fulfil the task of a flexible arm [11]. Our data present several new lines of evidence supporting this idea. As shown for the Erv2 protein, the dimer formation depends on an interaction of the CGC motif with the CXXC motif of the opposite monomer. The short form of the Erv1 protein, which lacks the N-terminal pair of cysteines, does not form any dimers. This is not due to a general misfolding of the short Erv1p because FAD binding and in vitro enzyme activity are unchanged. Changing C30 or C33 to serine drastically increases the amount of observed dimers. Probably, the remaining single cysteine residue interacts with the CXXC motif at the reaction centre of the opposite monomer. Due to the mutation, the reaction of the mutant protein gets trapped halfway and the two monomers are permanently crosslinked, explaining the increased amount of dimers. The C30 mutation drastically changes the colour of the protein. EPR spectroscopy gave no indication for radicals or metals bound to the protein. On the other hand, the colour of the protein is completely bleached upon reduc- tion with dithionite. Thus, we conclude, that the serine introduced at position 30 is close enough to the reaction centre to change the spectroscopic characteristics of the FAD or the charge-transfer complex. One possible explan- ation would be a charge-transfer complex between a persulfide anion and the FAD, that causes a similar colour Ó FEBS 2003 Cysteine mutagenesis of yeast ERV1 (Eur. J. Biochem. 270) 1533 change in the FAD containing butyryl-CoA dehydrogenase [24]. Again, the change to serine might contribute to colour production as no colour changes were reported for the corresponding alanine mutations of Erv2p. While a detailed band analysis and FT-IR might reveal more about the exact interactions at the reaction centre, we take the colour change of C30S as strong evidence for physical interactions of the N-terminal cysteine pair with the primary reaction centre. The situation in the C33S mutation appears to be different with respect to the intermediate formed. Biophys- ical differences between the cysteine residues of redox active CXXC motives are well known and it might be speculated, that C33 is the stronger nucleophile that therefore initiates the transfer reaction with other cysteine residues [4,5]. Deletion of the N-terminal Erv1p domain does not disturb the reaction with artificial substrates like dithio- threitol. In contrast, the two point mutations C30S/C33S reduce the in vitro activity. The most plausible explanation for this finding is that the changes of cysteine 30 or 33 to serine may result in unproductive contacts with the primary redox centre of the partner monomer as demonstrated by the increased dimer formation (Fig. 4). The analogous bacterial compartment of the mitochond- rial intermembrane space has developed similar strategies for the transfer of disulfide bridges. It has recently been reported that the bacterial dsbB protein in the periplasm also contains two cysteine pairs acting in concert for the transfer of disulfides to dsbA [25]. Interestingly, the redox function of dsbB is linked to the bacterial respiratory chain [26,27], providing another clue for the localization of Erv1p in the mitochondrial intermembrane space. The presented genetic, biochemical and spectroscopic data indicate a possible involvement of the flexible N-terminal arm for intersubunit disulfide transfer. While final proof could possibly only be achieved by structural information, it will be interesting to identify the natural substrates, which may be proteins or iron–sulfur clusters within proteins that due to structural restrictions depend solely on a flexible arm for the transfer of disulfide bridges. Acknowledgements The help of Dr Thorsten Friedrich with the EPR experiments and for reading the manuscript are gratefully acknowledged. For critical reading and useful suggestions we thank Dr Robert N. Lightowlers. We thank Martin Ingenhoven for help with the preparation of the mutant proteins. The authors acknowledge support of their work by the Deutsche Forschungsgemeinschaft (DFG) (Grant SFB 575). References 1. Kadokura, H. & Beckwith, J. (2001) The expanding world of oxidative protein folding. Nat. Cell Biol. 3, E247–E249. 2. Ritz, D. & Beckwith, J. (2001) Roles of thiol-redox pathways in bacteria. Annu. Rev. Microbiol. 55, 2001. 3. Senkevich, T.G., White, C.L., Koonin, E.V. & Moss, B. (2000) A viral member of the ERV1/ALR protein family participates in a cytoplasmic pathway of disulfide bond formation. Proc. Natl Acad. Sci. USA 97, 12068–12073. 4.Hoober,K.L.,Glynn,N.M.,Burnside,J.,Coppock,D.L.& Thorpe, C. (1999) Homology between egg white sulfhydryl oxi- dase and quiescin Q6 defines a new class of flavin-linked sulfhydryl oxidases. J. Biol. Chem. 274, 31759–31762. 5. Thorpe, C., Hoober, K.L., Raje, S., Glynn, N.M., Burnside, J., Turi, G.K. & Coppock, D.L. (2002) Sulfhydryl oxidases: emerging catalysts of protein disulfide bond formation in eukaryotes. Arch. Biochem. Biophys. 405, 1–12. 6. Lange,H.,Lisowsky,T.,Gerber,J.,Mu ¨ hlenhoff, U., Kispal, G. & Lill, R. (2001) An essential function of the mitochondrial sulf- hydryl oxidase Erv1p/ALR in the maturation of cytosolic Fe/S proteins. EMBO Report 2, 715–720. 7. Klissenbauer, M., Winters, S., Heinlein, U.A.O. & Lisowsky, T. (2002) Accumulation of the mitochondrial form of the sulfhydryl oxidase Erv1p/Alrp during early stages of spermatogenesis. J. Exp. Biol. 205, 1979–1986. 8. Sevier, C.S., Cuozzo, J.W., Vala, A., Aslund, F. & Kaiser, C.A. (2001) A flavoprotein oxidase defines a new endoplasmic reticulum pathway for biosynthetic disulphide bond formation. Nat. Cell Biol. 3, 874–882. 9. Gerber,J.,Mu ¨ hlenhoff, U., Hofhaus, G., Lill, R. & Lisowsky, T. (2001) Yeast ERV2p is the first microsomal FAD-linked sulf- hydryl oxidase of the Erv1p/Alrp protein family. J. Biol. Chem. 276, 23486–23491. 10. Stein, G. & Lisowsky, T. (1998) Functional comparison of the yeast scERV1 and scERV2 genes. Yeast 14, 171–180. 11. Gross, E., Sevier, C.S., Vala, A., Kaiser, C.A. & Fass, D. (2002) A new FAD-binding fold and intersubunit disulfide shuttle in the thiol oxidase Erv2p. Nat. Struct. Biol. 9, 61–67. 12. Lisowsky, T. (1992) Dual function of a new nuclear gene for oxidative phosphorylation and vegetative growth in yeast. Mol. Gen. Genet. 232, 58–64. 13. Mumberg, D., Mu ¨ ller, R. & Funk, M. (1994) Regulatable pro- moters of Saccharomyces cerevisiae: comparison of transcriptional activity and their use for heterologous expression. Nucleic Acids Res. 22, 5767–5768. 14. Hanahan, D. (1983) Studies on transformation of Escherichia coli with plasmids. J. Mol. Biol. 166, 557–580. 15. Lee, J., Hofhaus, G. & Lisowsky, T. (2000) Erv1p from Sac- charomyces cerevisiae is a FAD-linked sulfhydryl oxidase. FEBS Lett. 477, 62–66. 16. Riddles, P.W., Blakely, R.L. & Zerner, B. (1979) Ellman’s reagent: 5,5¢-dithiobis(2-nitrobenzoic acid) – a re-examination. Anal. Bio- chem. 94, 75–81. 17. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Press, Cold Spring Harbor, New York. 18. Vriend, G. (1990) WHAT IF: a molecular modeling and drug design program. J. Mol. Graph. 8, 52–56. 19. Sato, S., Tews, I. & Wilson, R.J. (2000) Impact of a plastid-bearing endocytobiont on apicomplexan genomes. Int. J. Parasitol. 30, 427–439. 20. Tews,I.,Perrakis,A.,Oppenheim,A.,Dauter,Z.,Wilson,K.S.& Vorgias, C.E. (1996) Bacterial chitobiase structure provides insight into catalytic mechanism and the basis of Tay-Sachs disease. Nat. Struct. Biol. 3, 638–648. 21. Hoober, K.L., Sheasley, S.L., Gilbert, H.F. & Thorpe, C. (1999) Sulfhydryl oxidase from egg white. A facile catalyst for disulfide bond formation in proteins and peptides. J. Biol. Chem. 274, 22147–22150. 22. Lisowsky, T. (1996) Removal of an intron with unique 3¢ branch site creates an amino-terminal protein sequence direct- ing the scERV1 gene product to mitochondria. Yeast 12, 1501– 1510. 23. Hoober, K.L. & Thorpe, C. (1999) Egg white sulfhydryl oxidase: kinetic mechanism of the catalysis of disulfide bond formation. Biochemistry 38, 3211–3217. 24. Williamson, G., Engel, P.C., Mizzer, J.P., Thorpe, C. & Massey, V. (1982) Evidence that the greening ligand in native butyryl-CoA dehydrogenase is a CoA persulfide. J. Biol. Chem. 257, 4314–4320. 1534 G. Hofhaus et al. (Eur. J. Biochem. 270) Ó FEBS 2003 25. Kadokura, H. & Beckwith, J. (2002) Four cysteines of the mem- brane protein DsbB act in concert to oxidize its substrate. EMBO J. 15, 2354–2363. 26. Bader, M.W. & Xie, T., Yu, C.A. & Bardwell, J.C. (2000) Disulfide bonds are generated by quinone reduction. J. Biol. Chem. 275, 26082–26088. 27. Kadokura, H., Bader, M., Tian, H., Bardwell, J.C. & Beckwith, J. (2000) Roles of a conserved arginine residue of DsbB in linking protein disulfide-bond-formation pathway to the respira- tory chain of Escherichia coli. Proc.NatlAcad.Sci.USA97, 10884–10889. Ó FEBS 2003 Cysteine mutagenesis of yeast ERV1 (Eur. J. Biochem. 270) 1535 . The N-terminal cysteine pair of yeast sulfhydryl oxidase Erv1p is essential for in vivo activity and interacts with the primary redox centre Go¨ tz Hofhaus 1 ,. six cysteines in yeast Erv1p was the major goal of this paper. All three pairs of cysteines in Erv1p are indispensable for in vivo activity The six cysteines