Staphylococcus aureus Coenzyme A Disulfide Reductase, a New Subfamily of Pyridine Nucleotide-Disulfide Oxidoreductase

The cdr gene encoding coenzyme A disulfide reductase (CoADR) from Staphylococcus aureus8325-4 was cloned, sequenced, and overexpressed. The gene encodes a 438-amino acid polypeptide that has a calculated molecular weight of 49,200 and sequence similarity to the pyridine nucleotide-disulfide oxidoreductase family of flavoenzymes. The deduced primary structure contains consensus sequences for flavin adenine dinucleotide and NADPH-binding regions but lacks the catalytic disulfide signature sequence typical of the glutathione reductase family of disulfide reductases. The active site region of CoADR has only a single cysteine residue that is similar to that in the conserved SFXXC active site motif of NADH oxidase and NADH peroxidase fromEnterococcus faecalis. CoADR is the first disulfide reductase reported having this active site region, and sequence comparisons of CoADR to representative members of the pyridine nucleotide-disulfide reductase superfamily placed CoADR in a distinct subfamily. CoADR was overexpressed in Escherichia coliusing the pET expression system, and 5–10 mg of fully active recombinant enzyme were recovered per liter of E. colicells.

The cdr gene encoding coenzyme A disulfide reductase (CoADR) from Staphylococcus aureus 8325-4 was cloned, sequenced, and overexpressed. The gene encodes a 438-amino acid polypeptide that has a calculated molecular weight of 49,200 and sequence similarity to the pyridine nucleotide-disulfide oxidoreductase family of flavoenzymes. The deduced primary structure contains consensus sequences for flavin adenine dinucleotide and NADPH-binding regions but lacks the catalytic disulfide signature sequence typical of the glutathione reductase family of disulfide reductases. The active site region of CoADR has only a single cysteine residue that is similar to that in the conserved SFXXC active site motif of NADH oxidase and NADH peroxidase from Enterococcus faecalis. CoADR is the first disulfide reductase reported having this active site region, and sequence comparisons of CoADR to representative members of the pyridine nucleotide-disulfide reductase superfamily placed CoADR in a distinct subfamily. CoADR was overexpressed in Escherichia coli using the pET expression system, and 5-10 mg of fully active recombinant enzyme were recovered per liter of E. coli cells.
We are investigating thiol chemistry and biology in Grampositive pathogenic bacteria and are interested in how the thiol metabolisms of these organisms compare with that of their human hosts. The human pathogen Staphylococcus aureus does not utilize the thiol/disulfide redox system based on glutathione and glutathione reductase (GSR) 1 found in eukaryotes and Gram-negative bacteria (1,2). Instead, S. aureus appears to use a redox system based on CoA and coenzyme A disulfide reductase (CoADR) (22). CoADR is a dimeric flavoprotein that specifically catalyzes the NADPH dependent reduction of oxidized CoA (Equation 1), thereby contributing to the high ratio of CoA/oxidized CoA (Ͼ450) and the intracellular reducing environment. An initial physical and chemical characterization of CoADR indicated that this enzyme was related to the family of pyridine nucleotide-disulfide oxidoreductases, which includes glutathione reductase (GSR), mercuric reductase, and dihydrolipoamide dehydrogenase. In addition, the disparate disulfide specificities of CoADR and its presumed human counterpart, GSR, identified this enzyme as a possible target for the design of selective inhibitors that would interrupt the thiol metabolism of S. aureus and function as anti-staphylococcal agents.
To better compare CoADR to human GSR and to facilitate the recovery of large amounts of recombinant enzyme for structural and functional studies, we have cloned and expressed the cdr gene in Escherichia coli. Our analysis of the CoADR primary structure reveals striking features that distinguish this enzyme not only from GSR but from other members of the pyridine nucleotide-disulfide oxidoreductase family. This analysis also highlights regions of the enzyme that likely mediate disulfide specificity and provokes questions as to the catalytic mechanism of this enzyme and to the function of homologous enzymes.

EXPERIMENTAL PROCEDURES
Strains and Media-S. aureus 8325-4 was the source of CoADR and genomic DNA (from John Iandolo, Kansas State University, Manhattan, KS). E. coli DH5␣ was from Life Technologies, Inc. (Burlington, ON, Canada), strain BL21 (DE3) and plasmid pET22B(ϩ) were from Novagen (Madison, WI). S. aureus was grown in tryptic soy broth (Difco Laboratories, Detroit, MI) at 37°C. Isolation of S. aureus genomic DNA was carried out as described (3). E. coli was grown at 37°C in LB and TB medium (4) and when required in the presence of ampicillin (100 g/ml).
Reagents and General Methods-Tetramethyl ammonium chloride, and CNBr were from Sigma (Mississauga, ON). Restriction enzymes and Taq DNA polymerase were from Life Technologies, Inc., and T4 DNA ligase and Calf Intestine Alkaline Phosphatase were from New England Biolabs. DNA fragments and PCR products were purified using Qiaquick spin columns (Qiagen, San Diego, CA). DNA fragments were labeled with digoxigenin by random primed PCR using the digoxygenin DNA Labeling and Detection Kit (Boehringer Mannheim, Laval, PQ, Canada). PCR products were cloned using the TA Cloning Kit (Invitrogen, San Diego, CA). Plasmids were purified on Qiawell cartridges (Qiagen) and sequenced using the Dye Termination Cycle Sequencing Kit and AmpliTaq DNA polymerase FS (Perkin-Elmer) and analyzed on an Applied Biosystems (Applied Biosystems, Inc., Foster City, CA) 373 automated DNA sequence analyzer. Sequences were further analyzed and assembled using the program SEQED (Applied Biosystems, Inc.). Oligonucleotides were prepared on a Beckman (Fullerton, CA) oligonucleotide synthesizer using standard phosphoramidite chemistry. SDS-PAGE and blotting of proteins were carried out using a Mini Protean Electrophoresis System (Bio-Rad, Richmond, CA), using a Tris-glycine buffer system as described (5). Immobilon polyvinylidene difluoride membranes were from Millipore (Mississauga, ON, Canada). Protein sequence determinations were carried out by the University of British Columbia Nucleic Acid and Protein Sequencing * This work was supported by funding from Abbott Laboratories. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF041467.
Laboratory on an ABI 476A Protein Sequencer. CoADR was purified from S. aureus as described previously (22).
Identification of a DNA Fragment Encoding the N terminus of CoADR-The gene encoding CoADR was identified and cloned using a reverse genetic approach. Purified CoADR (5 g) was cleaved with cyanogen bromide as described (6), and the resulting fragments were separated by SDS-PAGE, blotted onto a polyvinylidene difluoride membrane, and stained with Coomassie Blue. Protein bands corresponding to native CoADR and a 35-kDa CNBr cleavage product were excised and sequenced. Degenerate oligonucleotide primers (SD111 and 113; see Table I), designed from the N-terminal and internal sequences (see Table I), were used to amplify the N-terminal of the CoADR gene using the PCR. The PCR was performed in PCR buffer (Life Technologies, Inc.) containing S. aureus genomic DNA (10 ng), primers SD111 and SD113 (100 pmol each), MgCl 2 (2.4 mM), tetramethylammonium chloride (60 mM), and dNTPs (0.25 mM each). The reaction was incubated at 95 (30 s), 47 (30 s), and 72°C (30 s) for 30 cycles. The 600-bp product was ligated directly with the pCRII cloning vector (Invitrogen) and electroporated into E. coli DH5-␣. The fragment was entirely sequenced from the resulting plasmid, pCR:CDR.
Cloning and Sequencing of the Gene Encoding CoADR-The cloned PCR fragment encoding the N-terminal of CoADR was excised from pCR:CDR by digestion with EcoRI and purified by agarose gel electrophoresis. The excised fragment was labeled with digoxigenin and used to probe Southern blots of S. aureus genomic DNA digested with various restriction enzymes (4). A single 3.8-kilobase pair HindIII fragment that hybridized to the probe under stringent conditions (68°C, 0.1 ϫ SSC buffer containing 0.1% SDS) was subcloned from a size selected library (3-5 kilobase pairs) into plasmid pUC18. The resulting plasmid, pUCDR4, was identified by colony hybridization to the PCR fragment and purified. Both strands of the gene encoding CoADR were then sequenced by primer walking. The open reading frame encoding CoADR was compared with a nonredundant protein data base using BLASTP 2 and was aligned to individual members of the pyridine nucleotidedisulfide oxidoreductase family using the GAP algorithm of the Genetics Computer Group software package (Madison, WI). CoADR was then compared by multiple sequence alignment to representative members of this superfamily using the CLUSTAL algorithm of the Genetic Data Environment software package, version 2.2, using a PAM 100 weighting matrix, a k topple of 2, and a fixed and floating gap penalty of 10. TERMINATOR (Genetics Computer Group) was used to identify possible stem loop structures functioning as transcriptional stop signals.
Heterologous Overexpression of cdr in E. coli-The structural gene for CoADR was overexpressed in E. coli using the pET expression system (7,8). The open reading frame encoding CoADR was amplified by the PCR using primers SD301 and SD302 (see Table I), which were designed to create NdeI and HindIII sites at the 5Ј and 3Ј termini of the CoADR gene. The amplified fragment was digested with NdeI and HindIII, purified by agarose gel electrophoresis, and ligated with pET22B(ϩ) (Novagen, Madison WI) digested with the same two enzymes and purified. The resulting plasmid, pXCDR (expression of Co-ADR; see Fig. 3A), was sequenced to confirm the presence of cdr and transformed into the expression strain BL21 (DE3). For overexpression, TB (10 ml) containing ampicillin (400 g/ml) was inoculated with a fresh colony of BL21 (DE3)/pXCDR and shaken overnight. Cells were harvested by centrifugation, washed twice in TB (10 ml), used as an inoculum for TB (1 liter) containing ampicillin (400 g/ml), and incu-bated with shaking at 37°C until it reached an A 600 of 1.2. Isopropyl-␤-D-thiogalactoside (to 1 mM) was added, and incubation was continued for 3 h. The cells were harvested, and the recombinant enzyme was purified as described for native CoADR (22) except that lysozyme (0.2 mg/ml) was used in place of lysostaphin in cell disruption. The purity of the enzyme was assessed by SDS-PAGE and staining with Coomassie Blue. The specific activity was measured spectrophotometrically by following the oxidation of NADPH (22). GSR activity of the final sample was measured to confirm that no E. coli GSR had been copurified.

Isolation and Nucleotide Sequence Analysis of cdr-The cdr
gene encoding CoADR was identified and cloned using protein sequencing and a reverse genetic strategy. The N-terminal amino acid sequences of CoADR and a 35-kDa CNBr fragment of CoADR are shown in Table I. Degenerate oligonucleotides (Table I) were designed to encode these peptides and used in the PCR to amplify the N-terminal portion of the CoADR gene. This fragment was used as a probe in Southern blots to identify a single HindIII fragment carrying the entire CoADR gene. The structure of the 3.8-kilobase pair HindIII fragment containing cdr and carried in the plasmid pUCDR4 is depicted in Fig. 1A. The complete nucleotide sequence of cdr, the deduced amino acid sequence of CoADR, and the 5Ј and 3Ј flanking sequences are shown in Fig. 1B. The 1317-bp gene encodes a 438-amino acid polypeptide having a calculated molecular weight of 49,200. The gene contains 35% guanosine and cytosine (GϩC) nucleotides consistent with that of S. aureus (3). The N-terminal and internal peptides (the residues with asterisks in Fig. 1B) sequenced from the native and CNBr digested protein are both encoded within this polypeptide, and the sequence of the PCR fragment used as probe (the underlined region in Fig. 1B) is present within the nucleotide sequence. A putative ribosome-binding site (3) was tentatively identified 9 bp upstream from the start codon (ATG). A putative transcriptional start site recently mapped by primer extension 3 is located 23 bases upstream from the ATG start codon. The structural gene for CoADR is preceded by another open reading frame (ORF1 in Fig. 1A) that terminates 50 bp upstream from the CoADR start codon. There was no evidence for stem loop structures that might serve as factor independent transcriptional terminators.
Sequence Comparisons-The sequence of CoADR shared greatest similarity to members of the pyridine nucleotidedisulfide oxidoreductase superfamily (P(N) ϭ 10 Ϫ6 -10 Ϫ30 ). Members of this superfamily are characterized by a conserved arrangement of functional domains: an N-terminal ADP-binding motif, GXGXXG, for binding the nucleotide moiety of FAD, followed by a catalytic region having a pair of redox active cysteines, CXXXXC, another ADP-binding motif for the binding 2 http://www.ncbi.nlm.nih.gov. 3 M. K. Pope, personal communication.  of the pyridine nucleotide, and a flavin-binding region. A pairwise comparison revealed that CoADR was 20 -30% identical and 40 -50% similar to representative members of this superfamily (Table II) and that sites of similarity were concentrated in areas corresponding to the FAD and pyridine nucleotidebinding motifs. However, an active site disulfide motif was missing from the sequence. Instead, CoADR has a stretch of amino acids containing a single cysteine residue that is similar to that of the conserved active site signature sequence SFXXC of NADH oxidase (NOX) and NADH peroxidase (NPX) from Enterococcus faecalis ( Fig. 2A), enzymes considered to be members of the pyridine nucleotide-disulfide reductase superfamily. Furthermore, a multiple sequence alignment of CoADR to members of the superfamily places CoADR in a new subfamily of disulfide reductase that is separate from GSR, dihydrolipoamide dehydrogenase, thioredoxin reductase, and mercuric reductase and that more closely resembles NPX and NOX (Fig.  3). Sequence alignments of CoADR with NOX and NPX are shown in Fig. 2A and with GSR from E. coli and human erythrocyte in Fig. 2B. It is clear from these comparisons that the CoADR sequence is divergent from both classes of enzyme but shares the catalytic region of the NPX and NOX subfamily and sequences typical of NADPH as opposed to NADH specific enzymes with the GSR subfamily. The upstream open reading frame, ORF1, which was also compared with the protein data base using BLASTP, encodes a homolog to a family of "hypothetical proteins" identified in the DNA sequences of a variety of microorganisms. The closest member of that family was the protein from Bacillus subtilis, P(N) ϭ 10 Ϫ30 (GenBank accession number for the Bacillus homolog is Z79580).
Overproduction of CoADR in E. coli-To overcome the difficulty of isolating CoADR from S. aureus, the cdr gene was cloned into the vector pET22B(ϩ) and overexpressed in E. coli. Extracts of BL21 (DE3) cells harboring pXCDR (Fig. 4A) and induced with isopropyl-␤-D-thiogalactoside contained significant CoADR activity. CoADR was purified from the soluble fraction at 5-10 mg/liter of culture to greater than 99% purity according to SDS-PAGE and Coomassie Blue staining (Fig. 4B). The specific activity (4,600 units/mg) and kinetic parameters (K m (NADPH) ϭ 1.9 Ϯ 0.5 M, K m (CoA disulfide) ϭ 14 Ϯ 3 M, and k cat ϭ 1100 Ϯ 200 s Ϫ1 ) of the recombinant enzyme were indistinguishable from that of CoADR isolated from S. aureus (Ref. 22 and Tables I and II). The purified CoADR sample had  E. faecalis (B). The primary structures were aligned using the CLUSTAL algorithm of the Genetic Data Environment software package, version 2.2, using a PAM 100 weighting matrix, a k topple of 2, and a fixed and floating gap penalty of 10. no detectable GSR activity, and the N-terminal sequence of the recombinant enzyme was identical to that of native CoADR purified from S. aureus (Table I). DISCUSSION The cdr gene of S. aureus encodes a polypeptide having a calculated molecular weight of 49,200 that is consistent with a subunit M r of 49,000 and native M r of 90,000 Ϯ 10,000 estimated previously (22). The gene contains a putative ribosomebinding site (Fig. 1B) and putative transcriptional start site. 3 The deduced primary structure of CoADR is similar to that of the pyridine nucleotide disulfide oxidoreductase superfamily and contains a catalytic region, two FAD-binding regions, and a single NADPH-binding region in appropriate arrangement. The most striking feature of the CoADR primary structure is its lack of a conserved pair of catalytic cysteine residues. Co-ADR instead has a region containing a single cysteine residue (Cys 43 ) that is similar to the active site cysteine (Cys 42 ) of the NPX and NOX from E. faecalis. These enzymes are members of the pyridine nucleotide-disulfide reductase superfamily, but their unique active sites have thus far distinguished them from enzymes that actually reduce disulfide bonds (9). Claiborne and co-workers have extensively characterized NPX and demonstrated that Cys 42 forms a stabilized sulfenic acid intermediate during the catalytic reduction of H 2 O 2 by this enzyme (10,11). Their analysis of the x-ray crystal structure of NPX further showed that Cys 42 is structurally and functionally similar to Cys 63 , the charge transfer cysteine, of human erythrocyte GSR (12,13). Cys 43 of CoADR likely participates in substrate reduction as well but in a manner that more closely resembles the reduction of O 2 and H 2 O 2 by NOX and NPX than the reduction of disulfides by GSR. Indeed, we find that during catalysis CoADR forms a stable mixed disulfide intermediate with CoA. These findings are discussed in detail in the accompanying paper (22).
A phylogenetic tree produced from a multiple sequence alignment of representative members of the pyridine nucleotidedisulfide reductase superfamily suggests that CoADR represents a unique subfamily that lies between the GSR and NPX branches (Fig. 3). A comparison of the pyridine nucleotide and FAD-binding regions of CoADR, NPX, NOX, and GSR provides further support for this divergence (Fig. 2). The N-terminal of CoADR (residues 3-32) contains a putative ␤␣␤ ADP-binding fold ((14) and references therein) that aligns well with the N-terminal (residues 2-31) FAD-binding region of NPX and NOX ( Fig. 2A). Like NPX and NOX, the central glycine of this GXGXXG pyrophosphate-binding fingerprint is not conserved (valine in CoADR). However, unlike NPX and NOX, the residue following the central glycine in CoADR is alanine, a residue that could not functionally replace the conserved histidine (His 10 ) that is believed to stabilize the sulfenic acid that forms at Cys 42 during catalysis by these enzymes (11,13,15). CoADR has two cysteine residues, one in the proposed active site region (Cys 43 ) and the other in the center of the N-terminal ␤␣␤ motif (Cys 16 ). Thiol titrations of CoADR suggest that one of these cysteines remains in the reduced form in both the reduced and the oxidized enzyme (22). Residue 16 of NPX is buried in the core of the FAD-binding region. Cys 16 of CoADR would be buried in the analogous region of CoADR and is thus likely the cysteine that is recalcitrant to oxidation. In catalysis by GSR a C-terminal histidine residue, His 467 of the human erythrocyte enzyme, acts as general base in the activation of the interchange cysteine; however, no analogous residue in CoADR is obvious. Another ␤␣␤ ADP-binding fold starts at Lys 150 in CoADR and presumably binds the ADP moiety of NADPH. The replacement of the terminal glycine in the GXGXXG motif with serine is common among enzymes that bind NADPH rather than NADH (14). The bulkier serine residue is proposed to expand the fold to sterically permit the 2Ј phosphate into the binding cleft. An arginine residue (Arg 179 ) just following the terminal ␤ sheet likely binds the 2Ј phosphate of NADPH. Both NPX and NOX, which use NADH as cofactor, maintain the terminal glycine residue and have a conserved aspartate following the terminal ␤ sheet that is believed to form a hydrogen bond with the 2Ј-hydroxyl of the ribityl moiety (9,14,16). CoADR has an additional fingerprint region (residues 267-277) that likely participates in the binding of the flavin moiety of FAD. The terminal aspartate of this region (Asp 277 ) is well conserved and in other enzymes is known to interact with the 3Ј hydroxyl of the FAD ribityl moiety (17)(18)(19).
The structural differences between CoADR and human GSR that result in the distinct disulfide specificities of these enzymes are of particular interest. In the absence of tertiary structural information for CoADR, a comparison of the x-ray structures for NPX and human GSR was made instead. Although the overall fold of these two enzymes are quite similar, the region following the cysteine redox centers are of dramatically different structure. In human GSR this region (residues 63-122) comprises two long ␣-helices connected by a loop region (19), whereas in NPX this region (residues 43-68) comprises several short stretches of helix and coil (12,13,20). Because these regions in both enzymes span the substrate-binding pocket, it is possible that the analogous region in CoADR (residues 44 -70) may participate in the specificity for CoA disulfide. Future structure-function studies of the recombinant enzyme will be necessary to address these speculations.
CoADR is the first disulfide reductase reported to have a SFXXC active site motif and represents a new subfamily of disulfide reductase (Fig. 3). This finding implies that proteins that are identified as similar to NPX and NOX due to the presence of the SFXXC motif may not reduce H 2 O 2 or O 2 but instead reduce disulfide bonds. For example, in the sequence of Methanococcus jannaschii (21) a gene was identified as encoding an NADH oxidase. This gene has the SFXXC motif but shares greater sequence similarity to CoADR than to NOX (Fig.  3) and lacks a residue similar to the conserved His 10 of NOX. What is the true function of the protein encoded by this gene? The similarity of CoADR to NPX and NOX also provokes the question whether NPX, NOX, and CoADR share catalytic activities. We have shown that CoADR has only minimal O 2 or H 2 O 2 reductase activity, Ͻ0.1% that of its CoA disulfide reductase activity (see Table II of Ref. 22), but can NPX or NOX reduce CoA disulfide? E. faecalis NPX was shown to have no disulfide reductase activity using the substrates lipoamide, glutathione disulfide, and DTNB, however, CoADR activity was never measured. NOX was recently shown to have no detectable CoADR activity. 4 Previous purifications of CoADR yielded approximately 200 g of enzyme from 10L of S. aureus cells. This level was sufficient for an initial characterization of the enzyme but has hampered more detailed structure/function investigations. The ability to rapidly purify milligram quantities of enzyme will now permit a structural analysis of CoADR. Protein engineering studies aimed at the identification of those residues within CoADR that mediate substrate recognition and turnover are now underway. Low molecular weight thiols play diverse and important roles in aerobic organisms. CoADR was originally identified as an enzyme central to the thiol metabolism of the human pathogen S. aureus and as a potential point of intervention in the treatment of staphylococcal infections (22). The cdr sequence has further confirmed that CoADR is structurally divergent from other disulfide reductases, and recent investigations show that these enzymes reduce disulfide bonds by different chemical mechanisms. A key question now is whether CoADR is essential for S. aureus pathogenicity. The molecular genetics of S. aureus are well established and provide a means of testing this possibility directly. The pathogenicity of a mutant strain of S. aureus in which cdr has been disrupted can be compared with that of an otherwise isogenic organism. Such a study would provide insight as to the role thiols play during infection and would provide the first direct test of whether interrupting the thiol metabolism of a pathogen could curtail virulence.