Cloning, Overexpression, and Characterization of Peroxiredoxin and NADH Peroxiredoxin Reductase from Thermus aquaticus *

The genes for peroxiredoxin (Prx) and NADH:peroxiredoxin oxidoreductase (PrxR) have been cloned from the thermophilic bacterium Thermus aquaticus. prxis located upstream from prxR, the two genes being separated by 13 bases. The amino acid sequences show that Prx is related to two-cysteine peroxiredoxins from a range of organisms and that PrxR resembles NADH-dependent flavoenzymes that catalyze the reduction of peroxiredoxins in mesophilic bacteria. The sequence of PrxR also resembles those of thioredoxin reductases (TrxR) from thermophiles but with an N-terminal extension of about 200 residues. PrxR has motifs for two redox-active disulfides, one in the FAD-binding site, as occurs in TrxR, and the other in the N-terminal extension. The molecular masses of the monomers of Prx and PrxR are 21.0 and 54.9 kDa, respectively; both enzymes exist as multimers. The recombinant flavoenzyme requires 3 mol equivalents of dithionite for full reduction, as is consistent with 1 FAD and 2 disulfides per monomer. PrxR and Prx together catalyze the anaerobic reduction of hydrogen peroxide. The activity of Prx is much less than has been observed with homologous proteins. Prx appears to be inactivated by cumene hydroperoxide. PrxR itself has low peroxidase activity.

Peroxiredoxins (Prx) 1 are a group of small proteins that catalyze the reduction of hydrogen peroxide to water and of alkyl hydroperoxides to the corresponding alcohols (1)(2)(3). They have been identified in many organisms from bacteria to mammals and implicated in a wide variety of cellular processes including proliferation (4), differentiation (5), and the immune response (6) as well as in the detoxification of peroxides. They have been given various names, including thiol-specific antioxidant (7), heme-binding protein 23 (HBP 23) (8), and alkyl hydroperoxide reductase cysteine protein (9). They occur as homo-dimers and higher multimers of monomers of molecular mass 19 to 29 kDa. The active site at which peroxide is reduced includes a cysteine residue toward the N terminus of the protein whose side chain is thought to be oxidized to a sulfenic acid during the reaction (10,11). It appears that in many peroxiredoxins the sulfenic acid reacts with the thiol of a second conserved cysteine residue toward the C terminus of the other monomer to form a cystine disulfide, which is subsequently reduced back to the dithiol. The second cysteine is not present in some peroxiredoxins from higher organisms, and in these cases, the sulfenic acid is reduced directly to the thiol (11). The external reductant for peroxiredoxins varies with the source of the enzyme, but it is invariably a thiol-containing molecule such as cysteine or trypanothione (3) or, more commonly, the protein thioredoxin (Trx) in combination with NADPH thioredoxin reductase (TrxR) (12). In bacteria the reducing equivalents are provided by an NAD(P)H-dependent flavoenzyme in which a reduced cystine disulfide is the direct electron donor to peroxiredoxin (10,13,14). The first enzyme to be recognized in this group was isolated from Salmonella typhimurium and termed AhpF (9). Similar enzymes from other sources were sometimes first isolated as NADH dehydrogenases or NADH oxidases, and their roles in peroxiredoxin reduction only became apparent later. Poole et al. (15) recently proposed the name NADH:peroxiredoxin oxidoreductase or peroxiredoxin reductase (PrxR) for enzymes in this group, a name that is adopted in this paper. Electron transfer in the bacterial systems occurs as follows: NAD(P)H 3 PrxR 3 Prx 3 H 2 O 2 .
A flavoenzyme in the thermophilic bacterium Thermus aquaticus was first isolated as an NADH oxidase (16). More recently, several properties of the enzyme, including its chemical composition and its ability to catalyze cumene hydroperoxide reduction in the presence of Prx from S. typhimurium, suggested that it is related to NADH peroxiredoxin reductases from other organisms (17,18). These observations implied that a corresponding peroxiredoxin is synthesized in T. aquaticus. The N-terminal amino acid sequence of a small protein that was found to copurify with the flavoenzyme was shown to resemble that of peroxiredoxins from other bacteria (18). However, it proved difficult to obtain the putative peroxiredoxin free from the flavoenzyme. Since large amounts of both proteins are required for studies on their structures and catalytic mechanisms, we have cloned their genes, overexpressed them separately in Escherichia coli, purified the recombinant proteins, and characterized them biochemically. This paper describes the first peroxiredoxin system from a thermophile and shows that the two enzymes comprise a system similar to the peroxiredoxin-dependent systems of other bacteria. A preliminary report has been published on the cloning of the two genes (19).

MATERIALS AND METHODS
Growth of Bacteria-T. aquaticus YT-1 (NCIMB 11243) was cultured as described previously (18). E. coli strains TG1, DH5␣, and BL21(DE3) * This work was supported by Forbairt/Enterprise Ireland through Basic Research Grant SC/96/213 and a Postgraduate Research Scholarship (to C. L.). 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) AF276071.
were maintained and propagated in Luria-Bertani medium (20) in cultures of up to 4 liters. The growth medium was supplemented with ampicillin (100 g/ml) or kanamycin (30 g/ml) according to the vector used and with IPTG (1 mM added in log phase, A 600 ϭ 0.5) when induction of the synthesis of PrxR was required.
Protein Purification-Native PrxR and Prx were purified from T. aquaticus with the following modifications to the published method for the flavoprotein (18). DNase (30 g ml Ϫ1 ) was included in the buffer used to suspend the cell paste. Elution of the flavoprotein from AMP-Sepharose with a pH gradient gave yellow fractions that also contained Prx as the only contaminant, as judged by SDS-PAGE analysis. The fractions were combined, concentrated by ultra-filtration (Amicon Corp.; PM10 membrane), dialyzed versus 0.1 M potassium phosphate buffer, pH 6.0, containing 0.5 M NaCl, and applied to a column of Sephacryl 200 (82 cm x 1.6 cm diameter) equilibrated with the same buffer. The first fractions from this gel filtration column contained both proteins, whereas a second band of protein contained only the flavoprotein. Material from the second band was used as the source of native PrxR; the mixture of PrxR and Prx was used to compare the catalytic activities of native and recombinant Prx.
Recombinant T. aquaticus PrxR was purified as follows from E. coli BL21(DE3) that had been transformed with pDK6/PrxR. All buffers contained 0.3 mM EDTA. A crude extract of E. coli was made by suspending 9.3 g of cell paste in 30 ml of 20 mM Tris-HCl, pH 7.4 (buffer A), lysing the cells by sonication (Branson sonicator type 7532B, operated at 80% full power), and centrifuging (23,500 ϫ g for 20 min). The extract was mixed with DEAE cellulose (300 ml of a slurry of Whatman DE32 that had been allowed to settle from buffer A), and the mixture stirred for 1 h. The DEAE-cellulose was separated by filtration on a filter funnel fitted with a sintered disc and washed in the funnel with 1.5 liters of buffer A plus 0.15 M NaCl. The PrxR was then eluted with 0.45 M NaCl in buffer A, and the yellow solution was dialyzed for 12 h against 0.1 M potassium phosphate, pH 7.0 (buffer B), using three changes of buffer. The solution was applied to a column of AMP-Sepharose (12 ϫ 1.5-cm diameter) equilibrated with buffer B. The column was washed with 120 ml buffer B, leaving the enzyme in a yellow band at the top of the column. A pH gradient was applied to the column by continuous dilution of 150-ml buffer B with 150 ml of 0.1 M Tris-HCl, pH 9.0 (buffer C). This eluted part of the enzyme. The remainder was eluted with more buffer C. Fractions (2 ml) were collected and analyzed by SDS-PAGE. Those that contained only PrxR were pooled, concentrated by ultrafiltration as described above, and stored at Ϫ20°C.
Recombinant Prx was purified as follows from E. coli BL21(DE3) that had been transformed with pRSETB/Prx. All buffers contained 0.3 mM EDTA. Cell paste (20.5 g) was suspended in 400 ml of cell lysis solution (50 mM Tris-HCl, 40 mM EDTA, pH 8.0, and 1 mg/ml lysozyme). After incubating the mixture on ice for 30 min, the suspension was sonicated and centrifuged as described for PrxR. The supernatant was incubated at 70°C in a water bath for 30 min, then cooled to 4°C and centrifuged (20,000 ϫ g, 20 min). The precipitate was discarded, and the supernatant was applied to a DEAE-cellulose column (Whatman DE32, 15.5 ϫ 4.8-cm diameter) equilibrated with buffer A. The column was washed with buffer A until the A 280 of the eluate was zero. The Prx was eluted with a linear salt gradient in which 1 liter of buffer A was continuously diluted with 1 liter of 0.4 M NaCl in buffer A. Fractions of 10 ml were collected and analyzed by SDS-PAGE, and those giving a single protein band corresponding to Prx were combined. The protein was precipitated and separated from much contaminating DNA by adding ammonium sulfate to 90% saturation and centrifuging (20,000 ϫ g, 20 min). The precipitate was dissolved in 0.1 M potassium phosphate, pH 6.0, containing 0.5 M NaCl, and the colorless solution was applied to a Sephacryl 200 column (82 cm x 1.6-cm diameter) to separate the remaining DNA. Fractions (2.5 ml) were collected and analyzed by SDS-PAGE. Those that gave only the band corresponding to Prx were pooled, concentrated by ultrafiltration as described above, and stored at Ϫ20°C. Prx (AhpC) from S. typhimurium was provided by Leslie Poole, Wake Forest University, North Carolina.
Enzyme Assays-The assay for NADH oxidase activity used during the purification of PrxR contained, in 1-ml final volume, 20 mM Tris-HCl, pH 7.4, 0.3 mM EDTA, 0.126 mM FAD, 0.176 mM NADH, and enzyme to start the reaction (18). The oxidation of NADH was followed at 340 nm and 25°C. A small blank reaction due to the chemical oxidation of NADH by FAD and oxygen was subtracted from the rate measured in the presence of enzyme.
The enzymic reduction of 5,5Ј-dithiobis(2,2Ј-nitrobenzoate) (DTNB) was measured in a final volume of 1 ml. The assay mixture contained 50 mM potassium phosphate, pH 7.0, 0.3 mM EDTA, 0.02% (mass/volume) bovine serum albumin, 0.5 mM NADH, 0.4 mM DTNB, and PrxR to start the reaction. The production of TNB Ϫ was followed at 412 nm and 25°C. The small blank rate of TNB Ϫ production that occurred in the absence of enzyme was subtracted from the initial rate given in the presence of enzyme.
Assays for peroxidase activity were carried out in anaerobic solution at 25-70°C in a total volume of 1 ml. The assay mixture contained 50 mM potassium phosphate, pH 7.0, 0.3 mM EDTA, 150 mM ammonium sulfate, 0.02% (mass/volume) bovine serum albumin, 0.15 mM NADH, 1 mM hydrogen peroxide, and enzyme(s). PrxR-dependent assays contained 5 M Prx and up to 151 nM PrxR. Reactions were carried out in a cuvette fitted with a 14/23 socket and a rubber cap (Subaseal 25). The cuvette, buffer components, and PrxR, as appropriate, was made anaerobic by several cycles of evacuation via a syringe needle inserted through the rubber cap and filling with N 2 that had been purified by passing it over a column of BASF catalyst (R3-11; BASF Corp., Mount Olive, NJ) at 120°C and bubbling through a solution of photo-reduced methyl viologen. The NADH and peroxide was then added to the cuvette from a syringe, and the low rate of NADH oxidation in the absence of Prx was measured at 340 nm. The Prx was then added, and the new rate of NADH oxidation was determined. The peroxidase activity of PrxR in the absence of Prx was examined using up to 4 M PrxR. In these assays the blank oxidation of NADH by PrxR was determined before adding hydrogen peroxide to start the peroxidase reaction.
Chemical Reduction-The stepwise anaerobic reduction of PrxR by sodium dithionite was carried out in a quartz cuvette (21) that contained, in 3 ml, 50 mM potassium phosphate buffer, pH 7.0, 0.3 mM EDTA, 0.45 M methyl viologen to promote electron transfer, and flavoenzyme. A side arm on the cuvette contained 0.4 ml of 0.1 M Tris-HCl, pH 8.0, 10 mM EDTA, 0.3 mM methyl viologen, and 3 M 3-methyl-5deaza lumiflavin. The cuvette was made anaerobic by repeated cycles of evacuation and filling with purified nitrogen. The methyl viologen in the side arm was then photo-reduced to act as a sink for residual oxygen and as a visible check that the cell remained anaerobic. A solution of sodium dithionite (in 0.1 M sodium pyrophosphate-HCl buffer, pH 9.0) that had been standardized with FMN was added stepwise to the cuvette (21). An absorption spectrum was recorded after each addition when all changes in the visible spectrum were complete.
Analytical Methods-Protein was determined by the dye binding method using bovine serum albumin as the standard (22). The quantitative extraction of FAD from recombinant PrxR was achieved by incubating the enzyme in a sealed tube at 100°C for 12 min (23). The protein precipitated during the heat treatment, and it was subsequently removed by centrifugation (20,500 ϫ g, 10 min) before measuring the absorption spectrum of the extracted flavin in the supernatant.
Protein size determination was performed using a calibrated FPLC column of Superdex-200 HR (Amersham Pharmacia Biotech) to analyze the apparent molecular mass of the purified native and recombinant proteins under non-denaturing conditions. The column (30 cm x 1-cm diameter) was equilibrated with 50 mM potassium phosphate, pH 6.0, containing 0.15 M NaCl and 0.3 mM EDTA at a flow rate of 0.5 ml/min. Samples of 100 -200 l were applied to the column, and the elution was monitored at 280 nm. SDS-PAGE was performed (24) using a vertical slab gel (ATTO AE-6450). The separated proteins were stained with Coomassie Blue.
DNA Manipulations-Cloning and transformation techniques were carried out as described (20). Plasmid DNA was isolated by the alkaline lysis method (20). T. aquaticus genomic DNA was isolated and purified from cultures that were grown for no more than 16 h (25).
The polymerase chain reaction (PCR) was used together with a Techne thermocycler to amplify genomic DNA. Reactions were carried out in a reaction volume of 100 l that contained 0.1 g of genomic DNA, either primers A and B or primers C and D as given below (1 M each), and 2 mM MgCl 2 , 10 mM Tris-HCl, pH 8.3, 50 mM KCl, the four dNTPs (0.25 mM each), and 5 units of Taq polymerase (Sigma). Glycerol (30 l 50% (v/v)) was included in amplification reactions in which primers A and B were used. The DNA was denatured by incubating at 95°C for 5 min before adding it to the reaction. The conditions used were as follows: 40 cycles of denaturation at 95°C for 1.5 min, annealing at 40°C for 1.5 min, and extension at 72°C for 3 min. Primer A was ATAGAYGCNGAYATHAARGC. Primer B was ATGTTRAAYTCYTC-YTCNCC. Primer C was non-degenerate GCAAATACTGGGCTAGTT-GCG. Primer D was ATGGGNAARAARGTNCARCC. The regions that are underlined at the 5Ј ends of the primers were included to increase the stability of the priming duplex. The reaction products were analyzed by agarose gel electrophoresis. They were isolated from the gel using the GeneClean kit (Bio 101) and subcloned into the pCR® 2.1 plasmid (Invitrogen) to generate the plasmids pCR2.1PRXR990 and pCR2.1PRX600. The nucleotide sequences of the fragments were determined (Cambridge Bioscience), and the correctly amplified sequence was identified by comparison with the N-terminal amino acid sequences of PrxR and Prx.
Identification and Cloning of PrxR and Prx Genes-The DNA fragments amplified by PCR were used as probes in Southern blot analysis. The probes were labeled with digoxigenin-11-dUTP in a random-primed labeling reaction (Roche Molecular Biochemicals kit, 1175033). Purified genomic DNA was digested with a range of restriction enzymes, both singly and in pairs, to generate fragments that could be screened for both PrxR and Prx. Restriction mapping of PRX600 PRXR990 identified several restriction sites that were used in the mapping of the complete Pxr and PrxR (Fig. 1). The restriction digests were separated by electrophoresis in 0.8% agarose gel and transferred to positively charged nylon membranes (26). DNA fragments that contained the genes encoding both PrxR and Prx were identified by direct hybridization of the blots with the digoxigenin-11-dUTP-labeled probes and subsequent detection in the specific antibody-mediated chemiluminescence reaction. Larger amounts of DNA were digested with the enzyme PstI, which produced single-sized fragments of DNA that hybridized to both probes and were therefore likely to contain the full-length genes. Sizeselected fragments were isolated from agarose gel and ligated into pBluescript SK ϩ (Stratagene) that had been digested with the same enzyme and treated with calf-intestinal alkaline phosphatase (New England Biolabs) to prevent self-ligation. The ligation products were transformed into E. coli TG1, and the plasmid DNA was isolated from colonies that contained recombinant plasmids. These plasmid preparations were screened by restriction mapping, Southern blot analysis, and DNA sequencing. A plasmid that contained the full-length PrxR and Prx genes in a 3.8-kb fragment was identified and termed pBluescript/PrxPrxR.
Expression of Recombinant Genes-prxR was subcloned into pDK6 (27) to give the expression plasmid pDK6/PrxR. The prx gene was similarly subcloned into pRSETB (28), giving the expression plasmid pRSETB/Prx. The full-length genes were amplified (from start codon to stop codon) by PCR with the high fidelity polymerase Pfu (Stratagene) and using non-degenerate primers, which incorporated appropriate restriction sites to facilitate subcloning. Reactions were carried out in a reaction volume of 100 l that contained 50 ng of pBluescript/PrxPrxR plasmid DNA, either primers PrxRNterm and PrxRCterm (for prxR amplification) or primers PrxNterm and PrxCterm (for prx amplification) as given below (0.5 M of each), 20 mM Tris-HCl, pH 8.8, 10 mM KCl, 10 mM ammonium sulfate, 20 mM magnesium sulfate, 0.01% Triton® X-100, 0.01 mg/ml bovine serum albumin, the four dNTPs (0.25 mM each), and 5 units of cloned Pfu DNA polymerase. The conditions used for primers PrxRNterm and PrxRCterm were as follows: 30 cycles of 94°C for 0.75 min, annealing at 50°C for 0.75 min, and extension at 72°C for 4 min. Reaction conditions for amplification with primers PrxNterm and PrxCterm were 30 cycles of 94°C for 0.75 min, annealing at 55°C for 0.75 min, and extension at 72°C for 2 min. All reactions were terminated with a 10-min cycle at 72°C. Primer PrxRNterm incorporating an EcoRI restriction site was GG2AATTCATGCTGTTAG-ATGCCGACATT. Primer PrxRCterm incorporating a PstI restriction site was CTGCA2GTTAATGGCGAATCAAATAATC. Primer PrxNterm incorporating an NdeI restriction site was GGGAATTCCAT2AT-GTCTTTGGTTGGGAAAAAA. Primer PrxCterm incorporating a PstI restriction site was CTGCA2GTTAAATTTTGCCGACAAGGTC. The restriction sites (underlined, with an 2 indicating the points of cleavage) allowed the PCR products to be cloned into the required expression vector. DNA isolation and manipulation procedures involved the use of the GeneClean kit for isolation of amplified products from agarose gels, restriction digestion of the purified PCR products and expression vectors, and other standard procedures for subsequent cloning of DNA fragments (20). Following transformation of E. coli BL21(DE3) with the appropriate expression construct (pDK6/PrxR or pRSETB/Prx), cultures of the bacterium were grown at 37°C in the presence or absence of IPTG as required. Crude lysates of E. coli were made to determine the levels of expression under a variety of growth conditions. The lysates were made by resuspending the cell pellets in cell lysis solution and incubating as outlined above. After centrifugation (18,500 ϫ g, 20 min) the supernatant was incubated at 70°C for 30 min. After a further centrifugation (18, 500 ϫ g, 20 min) the supernatant was analyzed by SDS-PAGE using 12.5% gels and Coomassie Blue staining.
Computer-based Methods-DNA sequence analysis was carried out using the Macmolly package (Softgene, Berlin). The amino acid sequences were aligned using ClustalW (29). The secondary structure of the polypeptide sequences was determined using Predict Protein (30 -32).

Cloning and Sequences of PrxR and Prx Genes-
The relative proximity and arrangement of the PrxR and Prx genes in the T. aquaticus genome was not known at the beginning of the investigation. However, the N-terminal amino acid sequences of PrxR (40 residues) and Prx (19 residues) had been determined, as had the sequences of four internal peptides of PrxR generated by trypsin digestion (now known to be residues 185-195, 275-285, 321-330, and 458 -486) (18). These sequences were used to generate oligonucleotide primers for PCR amplification of genomic DNA. A sequence similarity search of the protein/ DNA data bases using the BLAST search algorithm (33) revealed that in a number of bacterial homologs the PrxR gene is located downstream from the Prx gene. It was therefore assumed that the two genes in T. aquaticus are arranged similarly. Two oligonucleotide primers were designed and synthesized for use in a PCR amplification of part of the PrxR gene. Primer A was degenerate (96-fold), and it corresponded to the N-terminal amino acid sequence of PrxR from residues 3 to 8. Primer B was also degenerate (64-fold), and it corresponded to a amino acid sequence determined from a tryptic fragment of PrxR from residues 330 to 325. Amplification only occurred with primers A and B when glycerol was present in the reaction mixture. The amplified product was cloned into pCR® 2.1 (Invitrogen). Nucleotide sequencing showed that the 990-base pair product contained base sequence that corresponded to the amino acid sequence of part of PrxR; this fragment was designated PRXR990. Two further oligonucleotides were designed and synthesized for use in a second PCR amplification. Primer C was based on the PrxR sequence obtained above, and it corresponded to the amino acid residues 15 to 10. Primer D was 128-fold degenerate, and it corresponded to the N-terminal amino acid sequence of Prx from residues 5 to 9. The amplified product was similarly cloned, and base sequencing showed that it contained most of Prx; this product was designated PRX600. After labeling with digoxigenin-11-dUTP, the PRXR990 and PRX600 PCR products were used for Southern blot analysis and sub-genomic library screening.
The sizes of T. aquaticus PrxR and Prx suggested that if the two genes occur sequentially in the genome then they would both occur on a 2.1-kb fragment of DNA. Southern blot analysis was carried out with T. aquaticus genomic DNA that had been digested with restriction enzymes that were selected from information obtained from the restriction mapping of the base sequences for PRXR990 and PRX600 (Fig. 1). Genomic DNA was digested singly with an enzyme known not to cut within the PRXR990 and PRX600 fragments and, doubly, with this enzyme and one of the enzymes known to cut within the PRXR990 or PRX600 sequences. Single sized (3.8 -4.0-kb) fragments produced from the digestion of genomic DNA with PstI were identified as potential candidates to contain the genes for both Prx and PrxR because both the PRXR990 and the PRX600 probes were observed to hybridize. A digest of genomic DNA with PstI and HindIII produced three fragments that hybridized to the PRXR990 probe, the sizes of which were determined to be approximately 0.6, 1.4, and 1.9 kb. The HindIII restriction sites within the PRXR990 sequence were determined to be 604 bases apart. The 1.4-and 1.9-kb fragments produced from the double digest were therefore known to be from either side of the HindIII fragment, either orientation of which would produce a fragment that contained the full-length prxR and prx (Fig. 1). Size selected fragments produced from the digestion of genomic DNA with PstI were ligated into pBluescript SK ϩ to produce a subgenomic library. Of 300 colonies screened, 3 were found to contain DNA that hybridized to both probes. Identical restriction patterns were given by the three plasmids. One was se-quenced and found to contain the complete genes for PrxR and Prx. This plasmid was designated pBluescript/PrxPrxR.
The 3833-base pair insert from pBluescript/PrxPrxR was sequenced on both strands. ORF Finder (NCBI) identified one partial and three complete open reading frames that are homologous to entries in the GenBank, Swiss-Prot, or PIR data bases (BLAST search algorithm (33)) ( Fig. 1). The amino acid sequence of the Prx is derived from bases 1-561 (Fig. 2). A putative ribosome-binding site similar to those found in E. coli is positioned 10 bases upstream from the initiation codon. The gene is initiated by an ATG start codon and is terminated by a single TAA stop codon. It encodes a polypeptide with 187 amino acids and a molecular mass of 20,982 Da. The gene that encodes PrxR is separated from prx by 13 bases. The amino acid sequence of PrxR is derived from bases 578 -2104 (Fig. 2). This gene is also preceded by a putative ribosome-binding site (bases 562-569) 8 bases upstream from the initiation codon, and it is terminated by a single TAA stop codon. It encodes 509 amino acids, and the polypeptide has the molecular mass 54,863 Da. The two genes are translated in different reading frames, indicating that although they might be transcribed as a single polycistronic mRNA molecule, they are translated as individual polypeptides.
A third open reading frame was identified from bases Ϫ955 to Ϫ491, encoding a polypeptide with 154 amino acids (Fig. 1). The gene is initiated by a single ATG start codon and is terminated by a single TGA stop codon. The sequence of the proposed gene product is most similar to conserved hypothetical proteins from Bacillus subtilis (A69981) and Thermotoga maritima (E72338) (72 and 58% identity, respectively). A partial open reading frame was identified from bases 2587-2679 encoding a polypeptide with 31 amino acids (Fig. 1). A single ATG codon initiates the gene that is present in the DNA sequence in the same reading frame as the prxR. A search of the protein data bases showed sequence identity of the partial gene product with the N-terminal sequences of five bacterial histidyl tRNA synthetases. The sequence is most similar to the proteins from B. subtilis and Staphylococcus aureus (68 and 57% identity, respectively) and to which it is related by many conservative amino acid substitutions (80 and 95%, respectively).
The G/C contents of T. aquaticus prxR and prx are 53.1 and 51.1%, respectively. There is a slight bias in favor of G/C in the third codon position (62.9 and 60.1% for prxR and prx, respectively). The G/C content in the first and second codon positions is 59.4 and 38.6%, respectively, for prxR and 56.9 and 36.1%, respectively, for prx. The G/C content of the two genes is less than the overall G/C content of other coding sequences in T. aquaticus DNA. The overall content is 66.6% as determined for 65 coding sequences that contain 21,160 codons, with a bias for G/C in the first, second, and third codon positions of 67.8, 43.0, and 88.9% respectively (GenBank, codon usage data base).
Amino Acid Sequence Comparisons-A search of the sequence data bases using BLAST (33) showed that the amino acid sequences of PrxR and Prx from T. aquaticus are similar to those of corresponding proteins from other organisms ( Table I).
The sequence of T. aquaticus Prx is most similar to the sequences of Prxs (AhpCs) from Amphibacillus. xylanus, B. subtilis, and Streptococcus mutans (77, 75, and 73% identity, respectively). The T. aquaticus protein contains three cysteine residues. Two of them (positions 47 and 166) are conserved in the proteins from other organisms, where they are known to be redox-active and to function in the peroxidase activity. The third cysteine (position 37 in T. aquaticus Prx) is present in the B. subtilis and S. mutans proteins but not in the A. xylanus Prx (AhpC) or in the well characterized S. typhimurium protein, which shares 67% identity with T. aquaticus Prx. The sequence of the T. aquaticus protein is also similar to the sequences of peroxiredoxins in eukaryotes. For example the identity with the 2-cys Prx from Arabidopsis thaliana is 43%.
The amino acid sequence of the flavoprotein from T. aquaticus is similar to the sequences of flavoproteins from a variety of bacteria that have been variously identified as NADH dehydrogenases, NADH oxidases, and/or components of alkyl hydroperoxidases termed AhpF. Table I shows that the greatest similarity is with the enzymes in B. subtilis, Bacillus alcalophilus, and A. xylanus. As noted earlier, the name NADH:peroxiredoxin oxidoreductase (PrxR) has recently been proposed for enzymes in this group (15). A lower level of sequence identity is observed with bacterial thioredoxin reductases, most notably with the proteins identified from the thermophilic bacteria Pyrococcus abyssi and Thermotoga maritima (Table I). Sequence alignment shows that the PrxRs contain an N-terminal extension of approximately 200 amino acids that is not present in the NADPH thioredoxin reductases. The motifs for two pairs of redox-active cysteines, CXXC, occur in T. aquaticus PrxR (residues 128 -131 and 337-340). The NADPH thioredoxin reductase proteins contain only one of these motifs, and it corresponds with the more C-terminal of the two motifs in PrxR.
One class of flavoproteins that contain FAD and a single pair of redox-active cysteines involved in the transfer of reducing equivalents from the FAD cofactor to the substrate (class II pyridine nucleotide disulfide oxidoreductases, PNDRII) is characterized by the active site fingerprint sequence; CX 2 CD(G/ A)X 2-4 (F/Y)X 4 (L/I/V/M)X(L/I/V/M) 2 G 3 (D/N) (as defined by PROSITE, PDOC00496) (34). The class includes prokaryotic and eukaryotic thioredoxin reductases (35,36) and bacterial PrxRs (37). The sequence is present in T. aquaticus PrxR between residues 337 and 357 (Fig. 3). The T. aquaticus PrxR also contains two motifs that correspond to the consensus sequence for the binding site of ADP (38,39) and that are possible FIG. 1. Organization of prx and prxR and partial restriction map of 3.8-kb fragment of DNA from T. aquaticus. The location of the ORF prx and prxR are indicated. The ORF for a putative protein is indicated with slanted lines. A partial ORF that is homologous to histidyl tRNA synthetases is indicated in black. The numbering of each ORF indicates the position along the 3.8-kb fragment at which the sequence is located (ϩ1 corresponding to the first base of prx ORF). PRX600 and PRX990 were the PCR products that were labeled and used to screen a sub-genomic library. The positions are indicated for the restriction sites EcoRI, HindIII, and SalI that were used to map the 3.8 fragment. Intergenic regions are shown as single horizontal lines. The coding and noncoding regions are drawn to the same scale.
binding sites for the ADP moieties of FAD and NAD(P)H. They contain three conserved glycine residues flanked by small hydrophobic residues, with an acidic residue occurring at the C terminus of the sequence (V/I/L/A) 3

G(A/G/S)G(A/G/I/L/S) 2 G(A/ S)(A/G/I/S/V)X 12 (F/I/L/M/V)(D/E).
Furthermore, the predicted secondary structures surrounding these two motifs are in the form of ␤␣␤ folds, as is diagnostic of nucleotide binding sites (Fig. 3). The amino acid sequence of T. aquaticus PrxR (residues 469 -479) is in complete agreement with the proposed consensus sequence motif for the binding site of the FAD flavin moiety (38).
A comparison of the amino acid compositions of proteins from thermophiles with those from their mesophilic homologs has shown that thermostable proteins tend to have a greater content of charged and hydrophobic amino acids and a smaller content of uncharged polar amino acids (40). Grouping of the amino acids of T. aquaticus Prx and PrxR into these three classes shows that they do not differ significantly from the corresponding groups for the nine most homologous proteins from mesophiles that are listed in Table II.
Overexpression of Recombinant PrxR and Prx-A major aim of the present investigation was the expression in E. coli of the recombinant genes for the T. aquaticus proteins and purification of the two proteins in amounts sufficient for biochemical characterization. Neither gene is expressed from the plasmid pBluescript/PrxPrxR in which the 3.8-kb insert is not correctly oriented for transcriptional control to be exerted by the lac promoter. It was shown that this plasmid does not survive in E. coli during culture of the organism at 37°C in the presence or absence of IPTG (0.1-1.0 mM). The genes for the two proteins were therefore separately amplified using the high fidelity polymerase Pfu, with simultaneous incorporation of appropri- ate restriction sites at each end. The gene prxR was cloned into the kanamycin-resistant vector pDK6 (27) (yielding the expression plasmid pDK6/PrxR), the gene being inserted into the EcoRI (5Ј-end) and PstI (3Ј-end) sites with transcription controlled by the trp-lac (tac) promoter. The greatest expression of the recombinant protein was observed with strain BL21(DE3) of E. coli and after induction by 1 mM IPTG. A lower level of protein expression occurred in E. coli TG1. Expression of T. aquaticus PrxR was also investigated in the ampicillin-resistant expression vector pKK 223-3 (41) with strains BL21(DE3) and TG1 of E. coli. The gene was cloned into the EcoRI (5Ј-end) and PstI (3Ј-end) sites of the vector with control of expression by the tac promoter. High levels of protein expression were observed using this vector and E. coli BL21(DE3). The pDK6/ PrxR expression construct was used for the large scale production of PrxR during the present investigation.
The gene prx was cloned into the NdeI (5Ј-end) and PstI (3Ј-end) sites of pRSET B (yielding the expression plasmid pRSETB/Prx), an ampicillin-resistant vector with transcription regulated by the T7 promoter. The greatest expression of the soluble form of this protein was also observed in the BL21(DE3) strain of E. coli, but in this case, in the absence of IPTG. It was observed that in the presence of IPTG, much of the recombinant protein was insoluble. Overexpression of Prx from pR-SETB/Prx was not observed in E. coli TG1. Expression of Prx was also investigated in the expression vectors pKK223-3 (41) and pDK6 (27). Prx was inserted into the XmaI (5Ј-end) and PstI (3Ј-end) sites of both vectors, and protein expression was examined in strains BL21(DE3) and TG1 of E. coli. Analysis of cell lysates by SDS-PAGE showed that overexpression of Prx did not occur.
Purification of PrxR and Prx-Recombinant T. aquaticus Prx was purified from E. coli grown in a 4-liter culture. The cell extract was heat-treated to denature and precipitate most of the E. coli protein. The recombinant protein from the thermophile remained in solution, where it composed about 95% of the protein. The absorbance maximum of the sample after heat treatment was at lower wavelength (265 nm) than is usual for protein. The main residual contamination was DNA, and this was not completely separated by chromatography on DEAEcellulose or by a subsequent precipitation of the protein with ammonium sulfate. The remaining DNA was therefore removed from the protein by gel filtration. After this step the protein was judged by SDS-PAGE to be at least 99% pure (Fig.  4). Approximately 144 mg was obtained from 20.5 g of cell paste. A specific assay for Prx was not used during the three steps of the fractionation, the purification being monitored by measurement of the UV absorption spectrum and SDS-PAGE analysis. Therefore, the overall recovery of this recombinant enzyme is not known; it is estimated to be about 50%.
Recombinant PrxR was purified from cell paste from a 2-liter culture of E. coli using two chromatographic steps similar to those used for the native enzyme. It was not possible to initiate the purification of this enzyme with heat denaturation because the treatment led to incomplete binding of the enzyme to the AMP-Sepharose used in the following step. The flavoenzyme was therefore separated from the extract using DEAE-cellulose as used previously with the native enzyme (18). Approximately  128 mg of enzyme, estimated to be more than 99% pure, was obtained from 9.3 g of cells. The overall yield of NADH oxidase activity was 85%. Based on these figures the amount of recombinant enzyme obtained from E. coli is much greater than the amount of native enzyme available from T. aquaticus. Isolation of the two T. aquaticus enzymes as recombinant proteins has the further advantage that the two proteins can be purified separately.
It was observed during the present investigation that although fractionation of extracts of T. aquaticus leads to highly purified native PrxR, about 20% of the enzyme isolated from the final gel filtration column is contaminated with Prx. When native PrxR that is contaminated with Prx is applied to a column of Superdex 200, two protein peaks are separated, the first of which (apparent molecular mass ϭ 400 kDa) contains PrxR and Prx, and the second (apparent molecular mass ϭ 150 kDa) contains only PrxR (19). Attempts to separate the proteins under reducing conditions in the presence of 10 mM dithiothreitol failed, and it was not possible to separate the two proteins in the mixture unless they were first denatured, as occurs in SDS-PAGE analysis.
Physico-chemical Properties-Both recombinant proteins gave one band of protein after SDS-PAGE under reducing conditions. The electrophoretic mobilities of the two recombinant proteins were identical with those of the native proteins, and the molecular masses were found to be 21.1 and 53.5 kDa for Prx and PrxR, respectively (Fig. 4). These values are in close agreement with the values calculated from the derived amino acid compositions of 21.0 and 54.9 kDa for the Prx and PrxR, respectively. Analysis of the pure recombinant proteins by FPLC in a column of Superdex 200 produced a single peak in each case with an apparent molecular mass of 235 kDa for Prx and 130 kDa for PrxR (Fig. 4), implying that both proteins occur as multimers. Solutions of Prx were colorless, the absorption spectrum showing a single band with a maximum at 280 nm. A value for the molar absorption coefficient at the maximum was calculated from the amino acid composition and the molecular mass of the protein using the program ProtParam (⑀ 280 ϭ 26,200 M Ϫ1 cm -1 ). The yellow solution of recombinant PrxR has absorbance maxima at 270, 384, and 450 nm. The ratio A 270 /A 450 is 6.1, similar to that reported previously for the native enzyme. The flavin in the supernatant after heat denaturation of the apoenzyme was shown to be FAD. This quantitative extraction of the flavin also confirmed an earlier report with the native enzyme (18) that the visible absorbance of the enzyme-bound FAD is greater than that of FAD in free solution (⑀ 450 ϭ 11.3 (42) and 13.0 mM Ϫ1 cm Ϫ1 for free and bound FAD, respectively). In further contrast to FAD, the absorbance at 384 nm for the enzyme-bound flavin is greater than the absorbance at 450 nm (A 450 /A 384 ϭ 0.92). The FAD content of the recombinant protein was determined using the absorption coefficient at 450 nm, protein analysis using a modification of the Bradford assay, and the value 55,692 Da for the molecular mass of the subunit of PrxR calculated from the derived amino acid composition plus a molecule of FAD. The flavin content of the isolated recombinant PrxR (0.98 mol of FAD/subunit) is in close agreement with the flavin content of the native enzyme (18).
The anaerobic spectrophotometric titration of recombinant PrxR with sodium dithionite showed that the enzyme requires approximately 6 electrons to fully reduce the FAD to the hydroquinone and confirmed earlier observations on the native enzyme that the reduction in such static titrations occurs in several phases (Fig. 5). In the first phase, during the stepwise addition of approximately 0.8 mol of dithionite/mol of enzyme FAD, the flavin undergoes a rapid reduction after each addition of reductant. The flavin spectrum then slowly changes to give a spectrum that in the long wavelength region is characteristic of the blue neutral form of the semiquinone. At the end of this phase, approximately 11% of the flavin is in form of semiquinone, as judged by the absorbance at 580 nm (⑀ 580 ϭ 4.3 mM Ϫ1 cm Ϫ1 ). In the second phase of the titration, during the further addition of almost 2 mol of dithionite/mol of FAD, the semiquinone increases to a maximum. During this phase the initial rapid reduction of the flavin after each addition of dithionite is followed by a slow further reduction of the flavin. At the end of this second phase 94% of the flavin was found to be in the form of the semiquinone. The third phase involves the stoichiometric conversion of the semiquinone to the hydroquinone, as is evident from the decrease in absorbance at 580 nm (0.5 mol of dithionite/mol of enzyme FAD). Exposure of the fully reduced enzyme to air causes a complete and rapid reoxidation of the flavin. A comparison of these observations on the reduction of T. aquaticus PrxR by dithionite with similar titrations of PrxRs from other organisms is given in the discussion.
Catalytic Properties-PrxR from T. aquaticus couples the oxidation of NADH to the reduction of several electron acceptors, including molecular oxygen (which it reduces to hydrogen peroxide), H 2 O 2 , DTNB, and Prx. The NADH oxidase activity is quite low unless exogenous FAD is added to mediate electron transfer from the enzyme-bound flavin (Table II ( saturating FAD are similar to the corresponding turnover numbers of the native enzyme. However the specific activities determined for the native (37.0 units/mg) and recombinant (38.5 units/mg) forms of the enzyme in the standard assay used to monitor the purification of the enzyme (0.126 mM FAD) are about 50% greater than reported for an earlier preparation of native enzyme (24.9 units/mg (18)). The explanation for this difference is not known. The possibility that the discrepancy is due to the use of a different protein assay in the earlier work is ruled out by the observation in the present study that the two protein assays give identical results.
T. aquaticus PrxR also catalyzes the reduction of DTNB, a thiol-disulfide interchange activity that it shares with PrxR isolated from A. xylanus (43) and S. typhimurium (10). The kinetic constants determined for the recombinant enzyme are similar to those determined for native enzyme. However, the turnover number for the native enzyme determined under standard conditions ( at 0.4 mM DTNB) is 26 times greater than was determined with an earlier preparation of enzyme (18), and again, the explanation for this difference is not known.
Earlier work showed that native PrxR (NADH oxidase) from T. aquaticus couples NADH oxidation to the reduction of cumene hydroperoxide in the presence of Prx (AhpC) from S. typhimurium (18). Recombinant T. aquaticus PrxR similarly couples NADH oxidation to the reduction of hydrogen peroxide in assays with S. typhimurium Prx ( Fig. 6; Table II). The corresponding activity with peroxiredoxin from T. aquaticus is much lower (Fig. 6). Thus, the turnover number for recombinant T. aquaticus PrxR in the presence of 5 M S. typhimurium Prx is 1107 min Ϫ1 at 25°C, whereas it is greater than 80 times smaller when Prx from T. aquaticus replaces the S. typhimurium enzyme. Efforts to enhance the activity by pretreating Prx with dithiothreitol have failed. The turnover number for the mixture of the two T. aquaticus enzymes increases to 171 min Ϫ1 when the assay temperature is 70°C, the temperature at which the organism is grown. The overall activity at both temperatures depends on the concentrations of the two coupled enzymes. The rate at 25°C increases linearly with concentration up to 144 nM PrxR when the assay is done with 5 M Prx. It was not possible to study the effect of varying the concentration of Prx in the presence of a similarly high concentration of PrxR because the flavoprotein itself appears to have peroxidase activity, and this relatively high blank oxidation of NADH in the presence of H 2 O 2 obscures the activity enhancement given by Prx. The turnover numbers for the peroxidase activity of PrxR alone are 0.034 s Ϫ1 and 0.47 s Ϫ1 at 25°C and 70°C, respectively. The peroxidase activity of PrxR in the presence of Prx is enhanced 1.3-fold by ammonium sulfate (apparent K act ϭ 11.4 mM at 25°C), similar to the effect of ammonium sulfate on the activities of the homologous enzyme system from S. typhimurium (14). It is not known if this activation by ammonium sulfate is a specific one or a general ionic strength effect as observed in the A. xylanus system (44). It appears that in the case of the T. aquaticus enzyme the effect is on PrxR because the blank peroxidase activity of this enzyme is similarly enhanced by ammonium sulfate and with a similar activation constant (apparent K act ϭ 14.6 mM at 25°C).
As noted above, T. aquaticus PrxR couples the oxidation of NADH to the reduction of cumene hydroperoxide in the presence of S. typhimurium Prx. This organic hydroperoxide appears to lead to inactivation of T. aquaticus Prx, because when cumene hydroperoxide replaces hydrogen peroxide in the com- plete anaerobic assay with the two recombinant enzymes from T. aquaticus, the rate of NADH oxidation declines rapidly with time. A similar inactivation by peroxides has been reported for peroxiredoxin from yeast (45).
The recombinant form of the flavoenzyme PrxR shows high thermal stability, similar to that previously reported for the native enzyme (18). The stability at high temperature is enhanced by including FAD in the incubation mixture (Fig. 7). This suggests that the main cause of loss in activity at high temperature is dissociation of flavin from the enzyme and subsequent denaturation of the apoenzyme. The temperature stability of Prx has not been examined. DISCUSSION We have shown that T. aquaticus contains a peroxiredoxin (Prx) and a corresponding reductase (PrxR) that allows the oxidation of NADH to be coupled to the reduction of hydrogen peroxide. Peroxiredoxins have been identified recently in a wide variety of organisms, and flavoenzymes that catalyze the reduction of Prxs are known to occur in certain other bacteria. This is the first time that the two genes for the enzyme system have been cloned and overexpressed from a thermophile and the enzymes characterized biochemically.
The function of the system in T. aquaticus is not known. It is surmised that the potential for cell damage by reactive oxygen species, such as peroxides, is high in an aerobic organism that grows at 70°C and that the Prx/PrxR system serves to help to remove such species. However the catalytic activity of the recombinant form of T. aquaticus Prx with hydrogen peroxide is surprisingly low by comparison with the activity of S. typhimurium Prx. In addition, cumene hydroperoxide, an alkyl hydroperoxide that is rapidly reduced by other bacterial Prxs, seems to inactivate the T. aquaticus enzyme. The native form of T. aquaticus Prx has not been obtained in a homogeneous form, and therefore, it has not been possible to directly compare its catalytic properties with those of the recombinant enzyme. However, the overall peroxidase activity of the mixture of the two native enzymes that is obtained by fractionation of cell-free extracts of T. aquaticus and that contains Prx and PrxR in a molar ratio of approximately 4.5:1 is very similar to the activity of an identical mixture of the recombinant Prx and native PrxR enzymes. In addition, the mixture of the two native proteins is also inactivated by cumene hydroperoxide. Therefore we conclude that the activity of recombinant Prx is similar to that of native protein. The possibility that Prx is isolated in a partly inactive form is being investigated. The overexpression of the two thermophilic proteins in the E. coli, a mesophile, allows the relatively rapid isolation of recombinant Prx, thus minimizing the possibility for inactivation during the purification procedure. Nevertheless, the procedure that is described above might be shortened still further if a method can be found to rapidly remove DNA, the major contaminant after the initial heat treatment of the E. coli extract. This might be achieved by introducing a gel filtration chromatography step immediately after the heat treatment. As noted earlier, heat treatment could not be used as the first step in the purification of recombinant PrxR. It is possible that the poor resolution of this enzyme on the subsequent AMP-Sepharose chromatography step when heat treatment was used is also due to interference by fragments of DNA.
Recombinant PrxR was found to differ from earlier data on the native form of the enzyme not only in two catalytic properties, as discussed above, but also in the number of electrons required for its full reduction by dithionite ion. The overall stoichiometry for reduction of the recombinant enzyme is smaller than that reported previously for native PrxR from T. aquaticus. The earlier measurement, which gave a value of 5 mol of dithionite/mol of enzyme FAD (18), also reported an additional phase at the beginning of the titration during which no changes occurred in the absorbance at 450 and 580 nm when the first mol equivalent of dithionite was added. The explanation for the differences between the results of the two measurements is not known. The measurement with the recombinant protein was made in the presence of methyl viologen to facilitate electron transfer between the different redox components of the enzyme (46), and for this reason it was possible to complete the titration in a much shorter time than the earlier titration with native enzyme. In addition, a greater concentration of enzyme was used in the titration with recombinant enzyme. Both conditions would decrease the likelihood that the titration would be distorted by oxygen contamination. Even more likely, however, is the possibility that the native enzyme that was used earlier contained either apo-PrxR and/or Prx. It is now known that it is difficult to completely separate Prx from PrxR when the two proteins are extracted from T. aquaticus. The presence of Prx and/or apo-PrxR in the titration of PrxR would increase the stoichiometry of reduction by contributing additional cystine disulfides to the overall reaction.
The stoichiometry of reduction of T. aquaticus PrxR by dithionite ion is similar to the stoichiometries reported for similar titrations of peroxiredoxin reductases (AhpF) from A. xylanus (43) and S. typhimurium (10), as are the changes in the optical spectra in the various phases of reduction. In all cases, a high proportion of the flavin is converted to the semiquinone. However, the semiquinone is formed slowly in the titrations and evidently is stabilized thermodynamically. It is probably not important as an intermediate in catalysis (47).
In conclusion, the successful cloning, overexpression of the genes for T. aquaticus Prx and PrxR and the isolation of the two proteins in large amounts opens the way to detailed studies on their structures and catalytic mechanisms, including the use of site-directed mutagenesis to study the effects of replacing amino acids whose side chains may play important roles in the overall functions of the enzyme system.