Bacillus subtilis ResA Is a Thiol-Disulfide Oxidoreductase involved in Cytochrome c Synthesis*

Covalent attachment of heme to apocytochromesc in bacteria occurs on the outside of the cytoplasmic membrane and requires two reduced cysteinyls at the heme binding site. A constructed ResA-deficient Bacillus subtilis strain was found to lack c-type cytochromes. Cytochrome csynthesis was restored in the mutant by: (i) in transexpression of resA; (ii) deficiency in BdbD, a thiol-disulfide oxidoreductase that catalyzes formation of an intramolecular disulfide bond in apocytochrome c after transfer of the polypeptide across the cytoplasmic membrane; or (iii) by addition of the reductant dithiothreitol to the growth medium. In vivo studies of ResA showed that it is membrane-associated with its thioredoxin-like domain on the outside of the cytoplasmic membrane. Analysis of a soluble form of the protein revealed two redox reactive cysteine residues with a midpoint potential of about −340 mV at pH 7. We conclude that ResA, probably together with another thiol-disulfide oxidoreductase, CcdA, is required for the reduction of the cysteinyls in the heme binding site of apocytochromec.

Thiol-disulfide oxidoreductases contain the active site motif Cys-Xxx-Yyy-Cys in which the two cysteine residues reversibly cycle between oxidized disulfide and reduced dithiol forms, and thus participate in redox reactions. Cytoplasmically located thiol-disulfide oxidoreductases, such as thioredoxin, have low redox potentials and are involved in maintaining the reducing environment of the cytoplasm. The more recently discovered thiol-disulfide oxidoreductases located on the outside of the cytoplasmic membrane in bacteria are found to be involved in: oxidation reactions, e.g. Escherichia coli DsbA, which catalyzes the formation of disulfide bonds in proteins transported across the membrane (1); disulfide bond isomerization reactions, e.g. E. coli DsbC, which functions in the redistribution of disulfide bonds among the cysteine residues of target proteins (2); and reduction reactions, e.g. Bradyrhizobium japonicum TlpA, which functions in the assembly of cytochrome aa 3 (3).
Thiol-disulfide oxidoreductases play a central role in the assembly of c-type cytochromes, in which the heme group is covalently attached via (usually) two thioether bonds between the thiol side chains of cysteine residues, occurring in the conserved motif Cys-Xxx-Yyy-Cys-His, and the vinyl side chains of the heme (4). Bacterial c-type cytochromes are localized to the outside of the cytoplasmic membrane where they function in electron transport processes. The mechanisms by which cytochromes c assemble are currently of great interest. Assembly takes place on the outside of the cytoplasmic membrane, involving separate transport of heme and the apocytochrome across the membrane (5); the latter occurs via the general protein secretory (Sec) pathway of the cell (6). Following translocation, assembly proceeds via a pathway involving a number of specific proteins. To date, two distinct systems have been identified in bacteria. System I is found in ␣and ␥-proteobacteria, including E. coli, Rhodobacter capsulatus, B. japonicum, and Paraccocus denitrificans (7)(8)(9). System II seems more widespread than system I, occurring in a range of bacteria including cyanobacteria, Gram-positive bacteria such as Mycobacterium species, ␤-, ␥-, and ␦-proteobacteria such as Bordetella pertussis, Thiobacillus ferrooxidans, and Helicobacter pylori, respectively, and some extremophiles such as Aquifex aeoliticus (10,11). Common to both systems is that the apocytochrome cysteine thiols and the heme iron must be in their reduced states for thioether bond formation to occur. Thiol-disulfide oxidoreductases are required for this, and specific proteins have been identified in a number of system I organisms, including E. coli (CcmG), R. capsulatus (HelX), B. japonicum (CycY), and P. denitrificans (CcmG) (12)(13)(14)(15)(16), and one system II organism, B. pertussis (CcsX) (11). We are studying cytochrome c assembly in the system II Gram-positive bacterium Bacillus subtilis. To date three genes required for this assembly have been identified: ccdA, resB, and resC (10,17,18). CcdA, ResB, and ResC are all polytopic integral membrane proteins. CcdA shows sequence similarity to a domain of E. coli DipZ (DsbD) that functions in the transfer of reducing equivalents from the cytoplasm to the outer side of the cytoplasmic membrane (19). In the B. subtilis chromosome the ccdA gene is not located close to other genes that are important for cytochrome c synthesis (20). The resBC genes reside in the resABCDE cluster in the chromosome. resD and resE encode a two-component signal transduction system (21,22), whereas resA encodes a putative thiol-disulfide oxidoreductase, which has been proposed to be involved in cytochrome c assembly (9,10). Previous attempts to disrupt resA were unsuccessful, leading to the conclusion that it (along with resB and resC) is an essential gene (21). Here, by generating a resA knockout mutant, we demonstrate that ResA is not essential for cell viability and that it is required for cytochrome c assembly. We show that ResA is located on the outside of the cytoplasmic membrane, attached to it via a single transmembrane segment and is shown to be a low potential thiol-disulfide oxidoreductase. The specific role of ResA in cytochrome c assembly-associated redox processes has been investigated and is discussed.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Media-Bacterial strains and plasmids used in this work are listed in Table I. E. coli cells were grown at 37°C in LB broth or on LA plates (23). B. subtilis strains were cultivated at 37°C in LB or NSMP (24), or on tryptose blood agar base (Difco) plates. Strains with an insertion mutation in resA were grown in the presence of 1 mM IPTG. 1 Antibiotics were used at the following concentrations when appropriate: spectinomycin, 150 mg/liter; erythromycin, 1 mg/liter; chloramphenicol, 5 mg/liter; phleomycin, 1.5 mg/liter (for B. subtilis); and ampicillin, 50 -100 mg/liter; chloramphenicol, 20 -25 mg/liter (for E. coli).
DNA Techniques-Standard molecular genetics techniques were employed (23). Plasmid DNA was isolated using Quantum miniprep (Bio-Rad) or Qiagen midiprep (Qiagen) kits, or by CsCl density gradient centrifugation. Chromosomal DNA from B. subtilis was isolated according to Marmur (25). E. coli was transformed by electroporation and B. subtilis was grown to natural competence as described by Hoch (26). PCR was carried out using Pfu (Promega) or Taq (Roche Diagnostics GmbH) DNA polymerases, and B. subtilis 1A1 chromosomal DNA used as template. Cloning was carried out using E. coli JM109 as host. All constructs obtained by PCR were verified by DNA sequence analysis.
Construction of Plasmids-Primers used for DNA amplification by PCR are listed in Table II. pLLE36 was constructed by amplifying an internal fragment of resA using PCR with primers LE032 and LE033. The 0.35-kb PCR product was cut with HindIII and EcoRI and cloned into pMutin2 cut with the same enzymes. pRAN1 was constructed by amplifying a DNA fragment containing the resA gene and its natural promoter using PCR with primers resA4 and resA5. The 0.73-kb product was cut with PstI and SalI and ligated into pHPSK cut with the same enzymes. pRAN1Es was obtained by cloning the resA fragment of pRAN1 into pHP13Es.
pLLE56 and pLLE57 were constructed by amplifying two DNA fragments, containing the natural resA promoter and a truncated resA reading frame, using PCR with primer pairs LE040/LE041 and LE040/ LE042, respectively. The PCR products were cut with BamHI and KpnI and cloned into pPHO1 digested with the same enzymes. The resulting plasmids were cut with BamHI and HindIII and the appropriate fragments cloned into pHPKS cut with the same enzymes. pRAN11 and pRAN8 were constructed by amplifying DNA fragments using PCR with primer pairs resA8/resA3 (soluble ResA) and resA8/ resA2 (His-tagged soluble ResA). The products were ligated into SmaIcut pUC18, generating pRAN10 and pRAN6, respectively. These were digested with NdeI and XhoI (pRAN6) or NdeI and EcoRI (pRAN10) and the resulting resA fragments ligated into pET21a(ϩ) cut with the same enzymes.
Construction of B. subtilis LUL9 -Strain LUL9 was isolated by transforming 1A1 with pLLE36, and selection on plates containing erythromycin and 1 mM IPTG. The pMutin2 insertion in resA was confirmed by PCR amplification of a DNA fragment using primers LE040 and MUT01 (which hybridize upstream of resA and within pMutin2, respectively); and by gene linkage analysis using strain 1A1C, which carries a chloramphenicol resistance marker within the spmAB locus located just upstream of resA in the B. subtilis chromosome. Ͼ90% of the erythromycin-resistant transformants obtained when strain 1A1C was transformed with limiting amounts of LUL9 chromosomal DNA were found to be chloramphenicol-sensitive.
Cytochrome c Oxidase Activity Assay-Isolated membranes from B. subtilis strains were added at a final protein concentration of 40 g/ml to a 40 M solution of reduced Saccharomyces cerevisiae cytochrome c in 20 mM MOPS, pH 7.4, in a stirred 3-ml cuvette at 30°C. Oxidation of the c-type cytochrome was followed at the wavelength pair 540 and 550 nm (27), using ⌬⑀ 550 -540 ϭ 19.5 mM Ϫ1 cm Ϫ1 to calculate activities. TMPD oxidation by colonies on NSMP plates was assayed as described by Le Brun et al. (10) or by Erlendsson and Hederstedt (28) for cells grown on NSMP plates containing DTT. 1 The abbreviations used are: IPTG, isopropyl-␤-D-thiogalactopyranoside; DTT, dithiothreitol; EI-MS, electrospray ionization-mass spectrometry; MOPS, 3-morpholinopropanesulfonic acid; NSMP, nutrient sporulation medium with phosphate; TMPD, N,N,NЈ,NЈ-tetramethyl-p-phenylenediamine. Immunoblot Analysis-SDS-PAGE was carried out using the Schä gger and von Jagow system (29). Proteins were blotted onto a polyvinylidene difluoride membrane (Millipore) by wet electroblotting using 20 mM Tris, 150 mM glycine, and 20% methanol. Rabbit antiserum raised against His-tagged soluble ResA or E. coli alkaline phosphatase was used as the primary antibody. Peroxidase-labeled anti-rabbit antibodies were used as secondary antibodies and SuperSignal West pico chemiluminescent substrate (Pierce) was used for visualization of bound primary antibodies.
Purification of His-tagged and Soluble ResA-His-tagged soluble ResA (htsResA) and soluble ResA (sResA) were purified from BL21(DE3)/pRAN8 and BL21(DE3)/pRAN11, respectively, after induction of resA expression with 1 mM IPTG. Harvested cells were washed and resuspended in 20 mM sodium phosphate, 0.5 M NaCl, 10 mM imidazole, pH 7.4, and sonicated while on ice. For htsResA, the soluble fraction obtained after centrifugation was applied to, and eluted from, a HiTrap metal chelation affinity column (Amersham Biosciences) according to the manufacturer's instructions. Purified htsResA was exchanged into 100 mM sodium phosphate, 100 mM NaCl, 1 mM EDTA, pH 7.0, using an ultrafiltration unit (Amicon). For sResA, the soluble fraction obtained after centrifugation was subjected to ammonium sulfate fractionation. The pellet from the 50 -80% (w/v) (NH 4 ) 2 SO 4 fraction was resuspended in 100 mM sodium phosphate, 100 mM NaCl, 1 mM EDTA, pH 7.0, desalted using a HiTrap desalting Sephadex G-25 Superfine column (Amersham Biosciences), and applied to a HiTrap Q-Sepharose HP anion exchange column (Amersham Biosciences) equilibrated in the same buffer. Proteins were eluted using a linear gradient of 0.1-1 M NaCl in 10 column volumes of the same buffer. Fractions containing sResA were pooled, exchanged into 100 mM sodium phosphate, 100 mM NaCl, 1 mM EDTA, pH 7.0, and applied to a XK26/60 Sephacryl S-100 gel filtration column equilibrated in the same buffer, giving pure sResA. Concentrations of htsResA and sResA were calculated using extinction coefficients, ⑀, at 278 nm of 18.25 and 18.15 mM Ϫ1 cm Ϫ1 , respectively, determined as previously described (30).
Reactivities of Thiols and Determination of Redox Potentials-The number of reactive thiols per protein was determined using iodoacetamide and iodoacetate, as previously described (31), except that samples were separated using PAGE according to Laemmli (32) with 8 M urea in place of SDS. The reactivities of cysteine residues of ResA as isolated and following reduction were determined using 5,5Ј-dithiobis(2-nitrobenzoic acid) (Ellman's reagent) as previously described (33).
For redox potential determinations, 20 M ResA (sResA or htsResA) in deoxygenated 100 mM sodium phosphate, 1 mM EDTA, 100 mM NaCl, pH 7.0, was incubated with 5 mM oxidized or reduced DTT for 3 h at 4°C to generate the fully oxidized or reduced species, respectively. Treatment with diamide was also used to generate the fully oxidized form (14). DTT or diamide was removed using a HiTrap desalting Sephadex G-25 Superfine column, and ResA was diluted to a final concentration of 1 M. Varying ratios of oxidized and reduced DTT, at a total concentration of 2 or 5 mM, were added to generate different potentials and the reaction was allowed to equilibrate for 3 h at 25°C. The redox state of ResA was followed by measuring fluorescence emission intensity at 350 nm, following excitation at 280 nm, using a PerkinElmer LS55 fluorimeter with excitation and emission slit widths set to 10 nm. Measurements indicated that equilibrium was achieved within 1 h. Similar results were obtained at different total DTT concentrations; when starting with fully oxidized or fully reduced ResA protein; and, when using de-oxygenated buffers in an anaerobic glove box in which oxygen was Ͻ2 ppm (Faircrest Engineering). Fluorescence intensities at ratios of 399:1 and 1:399 oxidized to reduced DTT were taken to represent 100% oxidized and reduced ResA, respectively. The data were fitted using the following equation, which is derived from the Nernst expressions for the two redox couples at equilibrium, where r is the fraction of protein in the reduced form, E h is the potential of the DTT couple, E m is the midpoint redox potential of ResA. n, the number of electrons involved in the reaction, was set to 2, the expected value for a thiol-disulfide oxidoreductase. Setting n ϭ 1 gave very poor fits. All E h calculations were based on the value of Ϫ330 mV for the standard redox potential (E m ) of DTT at pH 7.0 (34). E m values for DTT at other pH values were calculated using a correction of Ϫ59 mV per pH unit increase. Analytical Gel Filtration Chromatography-Molecular weights/association states of sResA and htsResA were determined using a Superdex 75 HR 10/30 column (Amersham Biosciences). sResA or htsResA (3 mg/ml) were applied to the column equilibrated with 100 mM sodium phosphate, 100 mM NaCl, 1 mM EDTA, pH 7.0, and eluted in the same buffer at a flow rate of 1.0 ml min Ϫ1 . Molecular weights were determined by reference to a calibration curve prepared using bovine lung aprotinin (6,500), horse heart cytochrome c (12,400), bovine erythrocyte carbonic anhydrase (29,000), bovine serum albumin (66,000), and blue dextran (ϳ2,000,000).
Other Methods-Sporulation efficiency assays were performed as described previously (28). Radiolabeling of cytochromes using 5-[4-14 C]aminolevulinic acid (51 mCi mmol Ϫ1 ) was performed as described by Schiött et al. (18). Membranes were isolated from bacteria grown in NSMP at 37°C as described previously (35). Membrane protein concentrations were determined using the bicinchoninic acid protein assay (Pierce Chemical Co.), with bovine serum albumin as standard. Nterminal sequencing was carried out by Edman degradation (Biomolecular Analysis Facility, University of Leeds, UK) on proteins separated by SDS-PAGE, blotted onto polyvinylidene difluoride membranes, and stained with 0.005% (w/v) sulforhodamine B in 30% (w/v) methanol, 0.2% (v/v) acetic acid. Blots were dried and stored at Ϫ20°C. EI-MS was performed using a VG platform electrospray mass spectrometer. For this, ϳ20 pmol of protein was prepared in 20 l of 1:1:0.005, acetonitrile:water:formic acid. Horse heart myoglobin (16951.48 Da) was used as a calibrant. Alkaline phosphatase activities of cell lysates of E. coli were measured using p-nitrophenyl phosphate as substrate (36). UVvisible absorbance spectra were recorded on a PerkinElmer 800 spetrophotometer.

RESULTS
Inactivation of resA-To investigate the function of ResA in B. subtilis, a derivative of pMutin2 was constructed (pLLE36), containing bp 55-409 of resA (537 bp). A Campbell-type integration of the entire plasmid into the chromosome resulted in strain LUL9 (Table I), in which resA was disrupted. pMutin2 contains an IPTG-inducible spac promoter that, following integration, can drive the expression of genes located downstream of the insertion (37). Growth of the resA knockout mutant LUL9 was similar to the parental strain 1A1 in the presence of 1 mM IPTG. However, in the absence of IPTG it grew poorly. This indicates that the pMutin2 insertion into resA has a polar effect on resBC and/or resDE expression. In subsequent experiments involving strain LUL9 the cells were grown in the 5Ј-AAAGAGCATATGTCCGAAGGGTCCGACG-3Ј NdeI presence of 1 mM IPTG. The properties of LUL9 show that the resA gene is not essential for viability of the cell. A ResA-deficient Strain Is Defective in Cytochrome c Synthesis-Colonies of LUL9 on NSMP plates did not oxidize TMPD, indicating a deficiency in cytochrome caa 3 activity. This was confirmed by analysis of membranes isolated from LUL9 cells grown in NSMP medium, which showed Ͻ0.1% activity of wild-type membranes. Plasmid pRAN1Es, containing the resA gene with its natural promoter, in LUL9 complemented the defect in TMPD-oxidation activity.
Under denaturing conditions cytochromes lose their heme unless it is covalently bound to the polypeptide. In a wild-type B. subtilis strain, grown in the presence of [ 14 C]aminolevulinic acid (a precursor to heme), the four membrane-bound cytochromes c and QcrB are observed as radioactive bands by SDS-PAGE. QcrB is the b subunit of the cytochrome bc complex and one of its two hemes is probably covalently bound to a cysteine residue in the polypeptide (38). Fig. 1 shows that strain LUL9 lacks all four c-type cytochromes but contains QcrB. The cytochrome c content in LUL9 was restored when resA was expressed in trans from pRAN1Es. These results combined show that ResA is required for cytochrome c synthesis.
The tested properties of LUL9 were found to be very similar to those of mutants defective in CcdA, e.g. those of strain LU60A1 (Fig. 1). Strain LUL14, defective in both ccdA and resA, showed the same TMPD-oxidation phenotype as LUL9.
The ResA-deficient Phenotype Can Be Suppressed by Inactivation of BdbD or Addition of DTT to the Growth Medium-B. subtilis BdbD and BdbC are thiol-disulfide oxidoreductases that catalyze formation of disulfide bonds in proteins on the outer side of the cytoplasmic membrane (28,39). Strain LUL15, deficient in both ResA and BdbD, was found to be TMPDoxidation positive. The same result was obtained when strain LUL9 was grown on plates in the presence of the reducing thiol reagent DTT (15 mM). Addition of DTT to the medium did not restore TMPD oxidation in strains deficient in ResB and ResC (28). Our in vivo findings indicate that ResA is a thiol-disulfide oxidoreductase involved in disulfide reduction during biosynthesis of c-type cytochromes.

A ResA-deficient Strain Is Not Defective in Sporulation-
Some proteins in the B. subtilis spore coat are rich in cysteine residues and are heavily cross-linked by disulfide bonds in the final spore (40). B. subtilis strains lacking CcdA are defective, but not completely blocked, in the synthesis of endospores (20). The exact role of CcdA in the sporulation process is not understood but sporulation in a ccdA mutant can be restored by inactivation of bdbD or bdbC (28). CcdA is therefore thought to function in breaking or isomerization of disulfide bonds during spore maturation and perhaps also in germination. The ResAdeficient strain LUL9, however, showed normal sporulation efficiency when compared with the parental strain (data not shown). Thus, ResA, in contrast to CcdA, does not play an important role in spore synthesis.
ResA Is a Protein of 179 Amino Acid Residues-The B. subtilis resA gene has been reported to encode a putative 181amino acid residue protein (21,41). Sequencing of the resA insert of pRAN1, and two other independent resA constructs, revealed a guanine residue that is not present in the data base resA sequence (Fig. 2A). Thus, the reading frame either upstream or downstream of the extra residue is different from that indicated in the data base. Alignment of ResA with thioldisulfide oxidoreductases that are known to be involved in cytochrome c assembly showed significant sequence similarity, particularly around the dicysteine motif, but not in the Nterminal region (not shown). Therefore, we concluded that the reading frame upstream of the point of insertion is affected by the correction. Fig. 2A shows an ATG start codon 7 bp downstream of the originally proposed GTG translation start codon (21) and gives the sequence of ResA shown in Fig. 2B. Alignment of the revised ResA sequence with other system II thioldisulfide oxidoreductases (data not shown) showed that similarity does extend into the N-terminal region, thus supporting our conclusion.
ResA Is a Membrane Protein-Analysis of the original erroneous ResA sequence led to the prediction that it is a soluble protein (10). However, the revised sequence of the resA gene encodes a protein with one predicted transmembrane segment (42, 43) and a putative type I signal peptidase cleavable N- FIG. 1. Cytochrome c contents of B. subtilis membranes. Autoradiograph of a SDS-polyacrylamide gel of membranes of B. subtilis strains labeled with 14 C-heme. The strains and plasmids are presented in Table I 2. Analysis of B. subtilis resA and its encoded protein  product. A, nucleotide sequence flanking the translational start point of resA. The original designated start codon is underlined (21). A G residue that was missing in the original published sequence and revealed in this work is indicated by the asterisk and bold lettering. The ATG translational start codon predicted from the corrected sequence is indicated in bold. B, corrected amino acid sequence of ResA. The asterisks indicate the positions at which truncated ResA was fused to PhoA for topology studies (see text). The cysteine residue at a position characteristic of lipoproteins is underlined. C, schematic representation of the deduced topology of ResA (corrected sequence) based on the transmembrane segment prediction program TMHMM (43) and data from the topology and localization studies described in the text. The model features one transmembrane ␣-helical segment, and a major domain with the thioredoxin-like motif on the outside of the cytoplasmic membrane. The numbers indicate amino acid residues.
terminal signal peptide (44) (Fig. 2B). Using immunoblot analysis of B. subtilis subcellular fractions we found that ResA is a membrane-bound protein (Fig. 3) and is probably anchored to the membrane by the N-terminal hydrophobic segment (Fig.  2C). Strain 1A1 contained membrane-bound ResA but LUL9 containing the vector pHP13Es did not. LUL9/pRAN1Es overproduced membrane-bound ResA, as expected. Thus, the disruption of resA with pMutin2 in LUL9 abolishes formation of stable ResA antigen in the cells.
The transmembrane topology of ResA was analyzed by using N-terminal segments of ResA fused to E. coli alkaline phosphatase (PhoA) lacking its native signal sequence. Active alkaline phosphatase is only formed if it is transported to the outer side of the cytoplasmic membrane in E. coli (45). Lysates of E. coli cells harboring plasmid pLLE56 or pLLE57 (which encode residues 1-35 and 1-140 of ResA, respectively, fused to PhoA) contained PhoA antigen of the expected sizes (immunoblot not shown) and contained alkaline phosphatase activity (0.22 and 0.17 mol ϫ min Ϫ1 ϫ (mg protein) Ϫ1 , respectively), whereas lysates of E. coli cells containing the vector pHPKS showed no activity (Ͻ0.01 mol ϫ min Ϫ1 ϫ (mg protein) Ϫ1 ). The alkaline phosphatase activity was found to be associated with the particulate subfraction of the lysates. Therefore, the N-terminal part of ResA probably functions as a membrane anchor and the C-terminal thioredoxin-like domain is exposed on the outer side of the cytoplasmic membrane, see Fig. 2C. ResA contains one cysteine residue in the N-terminal region of the polypeptide, in a position characteristic of lipoproteins (Fig. 2B). However, the protein does not satisfy other prediction criteria for lipoproteins (46,47), and analysis by SDS-PAGE and Western botting of ResA antigens in strains LUH102 and LUH104, which are deficient in Lgt (diacylglyceride transferase) and Lsp (type II signal peptidase), respectively (48), showed the same apparent size of ResA as found in 1A1 (data not shown). Therefore, ResA is most likely not a lipoprotein.
Production and Isolation of the Thioredoxin-like Domain of ResA-A version of resA encoding a shortened protein missing the first 35 amino acid residues that anchor ResA to the membrane and with the 36 th , Ile, changed to Met, was cloned, E. coli expression vectors encoding His-tagged and non-tagged forms were constructed, and the proteins were purified. N-terminal sequence analysis indicated that the N terminus is processed, resulting in the removal of the initial Met residue. EI-MS gave masses of 15921.9 and 16988.8 Da for sResA and htsResA, respectively, which are 134.4 and 132.6 Da lower than the predicted masses, consistent with the processing of the Nterminal Met residue (131.2 Da).
Molecular weights of 15,900 Ϯ 120 Da and 16,500 Ϯ 130 Da were determined for sResA and htResA, respectively, using analytical gel filtration. These are in good agreement with predicted values and clearly indicate that both proteins are monomers in solution.
Reactivity of ResA Cysteine Thiols-The number of thiolcontaining residues in sResA and htResA was investigated using the cysteine modification reagents iodoacetamide and iodoacetate, as described under "Experimental Procedures." Urea-polyacrylamide gels of sResA (not shown) and htsResA (Fig. 4) contained three bands corresponding to the protein modified with no, one, and two carboxymethyl groups, consistent with the presence of 2 cysteine thiol residues.
Investigation of cysteine thiol reactivity with Ellman's reagent showed that, as isolated, both sResA and htsResA contained ϳ0.5 reactive cysteine residues per protein. Reduction with DTT followed by removal of the excess reductant immediately prior to the addition of Ellman's reagent showed that, as expected, sResA contained 2 reactive cysteine residues per protein, although only 1.4 reactive cysteine residues could be detected per htsResA. The reason for this difference is unknown, but may indicate a difference in the cysteine reactivities of the two proteins.
Thiol-disulfide oxidoreductases such as E. coli thioredoxin, calf liver protein-disulfide isomerase, and B. japonicum TlpA (49 -51) exhibit an enhanced fluorescence signal at ϳ350 nm when in their reduced state. This is because of the significant quenching, in the oxidized state, of the fluorescence arising from tryptophan residues close to the dicysteine motif. This can be used to monitor the redox state of the protein and, when allowed to reach equilibrium with varying ratios of oxidized and reduced thiol reagents, can be used to determine the midpoint redox potential (E m ) of the thiol-disulfide oxidoreductase (15,51). Like thioredoxin, ResA has two tryptophan residues close to the dicysteine motif (i.e. WXXWCXXC) and, as isolated, both sResA and htsResA gave fluorescence intensity at 350 nm. Excess reduced glutathione did not affect the fluorescence signal, whereas reduced DTT caused a significant increase in intensity. Comparison of the fluorescence intensities of the isolated and reduced proteins indicated that ϳ29% of the proteins were in the reduced state following purification, consistent with the Ellman's reagent assay. DTT has a significantly lower standard redox potential than glutathione (Ϫ330 mV compared with Ϫ245 mV at pH 7 (34,52)) and was subsequently used for the determination of the midpoint potential of ResA (53)(54)(55). Fig. 5, A and B, show plots of the fraction of reduced protein versus the potential generated by DTT for sResA and htsResA, respectively. Fitting the data from multiple titrations gave an E m of Ϫ340 Ϯ 10 mV at pH 7 for both ResA proteins. The pH dependence of the midpoint potential of sResA was investigated between pH 6.0 and 8.0, see Fig. 5C. The data were fitted to a straight line with a gradient of Ϫ59 Ϯ 1 mV per pH unit. This is characteristic of the involvement of two protons and two electrons in the redox process, as expected for a thiol-disulfide oxidoreductase in this pH range. DISCUSSION In this paper we have shown that ResA is a low potential thiol-disulfide oxidoreductase that is involved in cytochrome c assembly on the outer side of the B. subtilis cytoplasmic membrane. The revised sequence of resA presented here leads to the prediction that ResA has one transmembrane segment and a thioredoxin-like motif. Our results confirmed that the protein is membrane-associated and showed that the C-terminal domain, containing the thioredoxin-like motif, is exposed on the outside of the membrane.
We have previously proposed that ResA accepts reducing equivalents from CcdA (28). The ResA-deficient strain LUL9 described here showed the same pleiotropic cytochrome c defect as a CcdA-deficient strain. That the deficiency can be complemented by inactivation of BdbD, or by supplementing the medium with DTT is consistent with the proposed roles of ResA and CcdA in breaking disulfide bonds during cytochrome c synthesis. In Fig. 6 we present a scheme for the proposed functions of ResA and CcdA. Apocytochrome c is transported across the membrane in an unfolded state by the Sec-protein transport machinery. BdbD and BdbC catalyze the formation of an intramolecular disulfide bond in the heme binding site. This disulfide bond is specifically broken by the action of reduced ResA; once broken, heme can be covalently attached to the cysteine residues of the apocytochrome. ResA is re-reduced by the activity of CcdA. The ResA/CcdA pathway seems to be dispensable in the absence of BdbD/BdbC; that is, if the disulfide bond-forming pathway is disrupted, the disulfide bondbreaking pathway is not required. This is consistent with the findings of Daltrop et al. (56) who showed that heme can be covalently attached to apocytochrome c in vitro without biosynthetic enzymes if the two cysteine residues in the heme binding site are reduced.
In contrast to ResA, CcdA is also involved in spore synthesis. This indicates that CcdA, in addition to transferring reducing equivalents for cytochrome c assembly, also transfers reducing equivalents to components required for sporulation (28).
The redox potentials of numerous thiol-disulfide oxidoreductases have been determined. In general terms, they fall into one of three approximate types: high potential (about Ϫ100 mV and above), e.g. E. coli DsbA (Ϫ89 mV) (57,58); mid potential (about Ϫ100 to Ϫ200 mV), e.g. eukaryotic protein-disulfide isomerase (Ϫ110 to Ϫ190 mV) (59,60); and low potential (about Ϫ200 and below), e.g. thioredoxins (about Ϫ230 to Ϫ270 mV) (61)(62)(63). Each of these types is characteristic of a different cellular function. The low potential, cytoplasmic thioredoxins are involved in maintaining protein cysteine residues in their re- duced form. The mid potential protein-disulfide isomerases are involved in thiol-disulfide interchanges and an intermediate redox potential apparently facilitates a role in breaking (reducing) and making (oxidizing) disulfide bonds. High potential DsbA is located on the outside of the cytoplasmic membrane and functions in the oxidation of protein thiols to form disulfide bonds in periplasmic proteins. A number of characterized thiol-disulfide oxidoreductases in Gram-negative bacteria are involved specifically in cytochrome c assembly, including E. coli CcmG, R. capsulatus HelX, B. japonicum CycY, and P. denitrificans CcmG (12)(13)(14)(15)(16)55). Where measured, these have generally been found to be of the low potential, reductant-type, but they are distinct from thioredoxins because they are located on the outside of the cytoplasmic membrane, where they are believed to be involved in the reduction of specific cellular components, for example, apocytochrome c prior to heme attachment. Assuming that the redox properties of sResA and full-length B. subtilis ResA are similar, the E m ϭ Ϫ340 mV obtained here for sResA indicates that it is a reductant-type thiol-disulfide oxidoreductase. This is consistent with a role for ResA in reducing the cysteine residues of apocytochrome c immediately prior to covalent heme attachment.