Molecular characterization of H 2 O 2 -forming NADH oxidases from Archaeoglobus fulgidus Serve ´ W. M. Kengen, John van der Oost and Willem M. de Vos Laboratory of Microbiology, Department of Agrotechnology and Food Sciences, Wageningen University, the Netherlands Three NADH oxidase encoding genes noxA-1, noxB-1 and noxC were cloned from the genome of Archaeoglobus fulgidus, expressed in Escherichia coli, and the gene products were purified and characterized. Expression of noxA-1 and noxB-1 resulted in active gene products of the expected size. The noxC gene was expressed as well but the protein pro- duced showed no activity in the standard Nox assay. NoxA- 1 and NoxB-1 are both FAD-containing enzymes with subunit molecular masses of 48 and 69 kDa, respectively. NoxA-1 exists predominantly as homodimer, NoxB-1 as monomer. NoxA-1 and NoxB-1 showed pH optimum of 8.0 and 6.5, with specific NADH oxidase activities of 5.8 UÆmg )1 and 4.1 UÆmg )1 , respectively. Both enzymes were specific for NADH as electron donor, but with different apparent K m values (NoxA-1, 0.13 m M ; NoxB-1, 0.011 m M ). The apparent K m values for oxygen differed significantly (NoxA-1, 0.06 m M ; NoxB-1, 2.9 m M ). In contrast with all mesophilic homologues, both enzymes were found to pro- duce predominantly H 2 O 2 instead of H 2 O. Despite apparent similarities, NoxB-1 is essentially different from NoxA-1. Whereas NoxA-1 resembles typical H 2 O-producing Nox enzymes that are expected to have a role in oxidative stress defence, NoxB-1 belongs to a small group of enzymes that is involved in catalysing the reduction of unsaturated acids and aldehydes, suggesting a role in fatty acid oxidation. Moreover, NoxB-1 contains a ferredoxin-like motif, which is absent in NoxA-1. Keywords: Archaeoglobus; flavoprotein; NADH oxidase; oxygen stress. Archaeoglobus fulgidus is a strictly anaerobic hyperthermo- philic archaeon that has been isolated from marine hydro- thermal environments as well as subsurface oil fields. This sulfate reducer can grow organoheterotrophically with a variety of carbon sources, or lithoautotrophically on hydrogen, thiosulfate and CO 2 [1]. Besides its ability to grow at extremely high temperatures, this organism is unusual in that it is evolutionary unrelated to other sulfate reducers. Recently, the sequence of the entire genome of A. fulgidus was completed [2]. The sequencing revealed the presence of eight putative NADH oxidase genes, which were designated noxA-1 to noxA-5, noxB-1, noxB-2 and noxC, according to their homology to other NADH oxidase encoding genes. NADH oxidases (EC 1.6.99.3) catalyse the two-electron reduction of oxygen to peroxide or the four-electron reduction of oxygen to water. Although all so-called NADH oxidases share the ability to reduce oxygen, their physiological role may differ or is often not known. Moreover, some homologues have been shown not to reduce oxygen and to catalyse somewhat different reactions, such as NADH peroxidase (EC 1.11.1.1) and disulfide reductase (EC 1.8.1.14). The noxA homologues from A. fulgidus code for a group of typical H 2 O-forming NADH oxidases of  49 kDa, found in various prokaryotes, and including the well-studied NADH oxidase from Enterococ- cus faecalis [3] (Fig. 1). This group also contains NADH peroxidases and coenzyme A disulfide reductases [4,5]. The noxB homologues encode a small group of enzymes of  72 kDa, that are involved in the reduction of unsaturated acids and aldehydes, or whose function is not known. In the presence of oxygen, they also produce H 2 O, except for the enzyme from Thermoanaerobacter brockii that forms H 2 O 2 and some superoxide (O 2 ) )[6].NoxC codes for a small 20-kDa protein, and it was designated as NADH oxidase, due to its homology to the NADH oxidases from the thermophilic bacteria Thermus aquaticus and Thermus thermophilus [7,8]. Some putative NAD(P)H oxidoreduc- tases from thermophilic archaea also belong to this group (Fig. 1). The physiological role of the putative NADH oxidases in A. fulgidus is still enigmatic. Common NADH oxidases of mesophilic origin are assumed to protect the cells from oxidative stress by reducing oxygen to water, without the formation of harmful reactive oxygen species. Alternatively, NADH oxidases may recycle oxidized pyridine nucleotides during catabolism. Moreover, some homologues were shown to have functions other than NADH oxidase (see above). Concerning the homologues from (hyper)thermo- philic species, only a few have been studied in more detail. A recently purified NADH oxidase from A. fulgidus (NoxA-2) was proposed to be involved in electron transfer reactions Correspondence to S. W. M. Kengen, Laboratory of Microbiology, Hesselink van Suchtelenweg 4, 6703 CT Wageningen, the Netherlands. Fax: +31 317 483829, Tel.: +31 317 483748, E-mail: serve.kengen@wur.nl Abbreviations: NoxA-1, NADH oxidase A-1; NoxB-1, NADH oxidase B-1; NoxC, NADH oxidase C; DCPIP, 2,6 dichlorophenol- indophenol; DTNB, 5,5¢-dithiobis(2-nitrobenzoic acid). Enzymes: NADH oxidases (EC 1.6.99.3); NADH peroxidase (EC 1.11.1.1); disulfide reductase (1.8.1.14). (Received 28 February 2003, revised 25 April 2003, accepted 14 May 2003) Eur. J. Biochem. 270, 2885–2894 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03668.x during sulfate respiration [9]. The NADH oxidase of Pyrococcus furiosus (Nox1; PF1430634) purified from an overproducing Escherichia coli,wasanticipatedtoplaya role in protection against oxidative stress [10]. The function of the NADH oxidase of T. brockii is still unknown [6]. In several hyperthermophilic archaea and bacteria, other than A. fulgidus, various putative NADH oxidase genes have been identified, whose function remains to be established. Independent of their physiological role, H 2 O 2 -producing NADH oxidases may be applicable in biosensors, where they act as mediator between NADH-forming dehydrogen- ases and the electrode [11]. Concerning this, enzymes from (hyper)thermophiles may be superior to mesophilic coun- terparts, because they generally exhibit a higher stability not only with respect to temperature, but also towards chemical denaturants, like detergents or organic solvents [12,13]. In order to analyse the biochemical properties, to unravel their physiological role and to test the potential stability, the noxA-1 gene, the noxB-1 gene and the noxC gene were cloned and expressed in E. coli and the overproduced NADH oxidases were purified and characterized. Experimental procedures Materials 3,3¢-Dimethoxybenzidine, coenzyme A (oxidized), glutathi- one (oxidized), horseradish peroxidase and 2,3-dimethyl- 1,4-naphthoquinone were from Sigma Chemie. Q-Sepharose, Superdex 200 HR 10/30, and Mono-Q HR 5/5 were from Amersham Pharmacia Biotech. Hydroxy- apatite (Bio-Gel HT), SDS/PAGE calibration proteins (broad range), and the protein assay kit were from Bio-Rad. The pET9d expression vector was from Novagen Inc. E. coli BL21(DE3) and Pfu DNA polymerase was from Stratagene. A. fulgidus (DSM 4304) was from the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). Polysulfide was prepared by adding 12 g Na 2 S to 1.6 g elemental sulfur in 100 mL anoxic water [14]. Cytochrome c was purified from the mesophilic Syntrophobacter fumaroxidans and was a gift of Frank de Bok (Wageningen University, the Netherlands). Cloning of the NADH oxidase genes In the genome sequence of A. fulgidus [2] various putative NADH oxidase genes have been identified, of which three were selected for further research (noxA-1, AF0254; noxB-1, AF0455; noxC:, AF0226). The following primer sets were designed to amplify the selected Nox open reading frames: for noxA-1 primer BG852 (5¢-CGCGTCATGAAGGTT GCAATTATAGGCGGT-3¢, sense) and primer BG853 (5¢-CGCGGGATCCCTACGGCAATCCGAGCTTC-3¢, antisense), with BspHI and BamHI restriction sites in bold; for noxB-1 primer BG854 (5¢-CGCGCCATGGCCAAG CTTTTCGAGCCAATCGAG-3¢, sense) and BG855 (5¢-CGCGGGATCCCTAAACCTTCAAAGCCAGAT-3¢, antisense), with restriction sites NcoIandBamHI in bold; for noxC primer BG831 (5¢-GCGCGTCATGATGGAAT GCCTTGACTTGCTGTTC-3¢,sense)andBG832(5¢-CG CGCGGATCCTCACCATTTTTCGAAGTGCGTGAG-3¢, antisense), with BspHI and BamHI restriction sites in bold. The 50 lL PCR reaction mixture contained 200 ng A. ful- gidus SL-5 genomic DNA, isolated as described previously [15], 100 ng each primer, 0.3 m M dNTPs, Pfu polymerase buffer, and 2.5 U Pfu DNA polymerase and was subjected to 35 cycles of amplification (15 s at 94 °C, 30 s at 50 °C and 2 min at 68 °C) on a DNA Thermal Cycler (Perkin Elmer Cetus). The PCR product was digested with the appropriate enzymes and cloned into an NcoI/BamHI- digested pET9d vector, resulting in pWUR66, pWUR67, and pWUR68, for noxA-1, noxB-1 and noxC, respectively. The plasmids were transformed into E. coli TG1 and E. coli BL21(kDE3) by heat-shock. Sequence data were analysed using the computer program DNASTAR . Expression of the NADH oxidase genes in E. coli Ten millilitres of Luria–Bertani medium with 50 lgÆmL )1 kanamycin was inoculated (1%) from overnight cultures of E. coli BL21(DE3) containing either pWUR66, pWUR67 or pWUR68. After growth at 37 °Cto D 600 ¼ 0.8, 0.5 m M isopropyl thio-b- D -galactoside was added to induce expression. After overnight growth, 2-mL of each culture was centrifuged (10 min at 20 000 g)and cells were resuspended in 50 m M Tris/HCl buffer pH 7.8. Cells were sonicated and the supernatants were subse- quently heated for 20 min at 70 °CtodenatureE. coli proteins. The supernatants were analysed by SDS/PAGE and by activity measurements. For enzyme purification 2-L cultures were grown in essentially the same way as described above. Cells (5–6 g wet weight) were harvested by centrifugation (2200 g for 15 min at 10 °C) and resuspended in 28 mL 50 m M Tris/ HCl buffer pH 7.8. The suspension was passed twice Fig. 1. Phylogenetic tree of NADH oxidases and related enzymes. The tree was constructed from alignments using the CLUSTAL method [20] of the Megalign program ( DNASTAR , London, UK) and Nox sequences available at the NCBI data base. The units at the bottom indicate the number of substitution events. Genbank indentifiers are indicated in parentheses. 2886 S. W. M. Kengen et al.(Eur. J. Biochem. 270) Ó FEBS 2003 through a French press (110 MPa) and the resulting crude cell extract was used for purification of the recombinant NADH oxidases. Purification of recombinant NoxA-1 and NoxB-1 The E. coli cell extract was heated for 30 min at 70 °C (NoxA-1) or for 30 min at 50 °C (NoxB-1), and denatured proteins were pelleted by centrifugation (17 200 g for 15 min at 10 °C). This pellet fraction was washed with 10 mL Tris/HCl buffer pH 7.8 and the centrifugation step was repeated. The supernatants of both centrifugation steps were combined, filtered through a 0.45-lm filter and loaded onto a Q-Sepharose column (1.6 · 10 cm) equilibrated with 20 m M Tris/HCl buffer pH 7.8. Bound proteins were eluted by a 200-mL linear gradient of NaCl (0–1 M in Tris/HCl buffer pH 7.8). NoxA-1 eluted in a single peak at 0.38 M NaCl. In a similar purification NoxB-1 eluted at 0.15 M NaCl. Active fractions were pooled and applied to a hydroxyapatite column (Bio-Gel HT; 1.6 · 10 cm) equili- brated with 10 m M sodium phosphate buffer pH 7.0. Elution was performed with a 200-mL gradient from 10 to 500 m M sodium phosphate (pH 7.2). The NoxA-1 as well as the NoxB-1 eluted right after the flow-through fraction. Active fractions were pooled and concentrated by ultrafil- tration (Filtron; 10 kDa cut-off). A 200-lL aliquot of the concentrated samples was loaded onto a Superdex-200 HR column, equilibrated in 20 m M Tris/HCl pH 7.8 containing 150 m M NaCl.NoxA-1aswellasNoxB-1elutedastwo overlapping activity peaks. PAGE The purity of the various purification fractions was regularly checked by SDS/PAGE according to the procedure of Laemmli using 15% (w/v) gels [16]. Protein samples were denatured by heating in SDS-sample buffer for 5 min at 100 °C. SDS/PAGE was also used to determine the subunit molecular mass. Calibration was performed using a set of calibration proteins: myosin (200 kDa), b-galactosidase (116.25 kDa), phosphorylase b (97.4 kDa), serum albumin (66.2 kDa), ovalbumin (45 kDa) and carbonic anhydrase (31 kDa). Protein bands were stained with Coomassie brilliant blue R250. Enzyme assays NADH oxidase activity was measured spectrophoto- metrically in 1-mL quartz cuvettes on a Hitachi U-2010 spectrophotometer equipped with a thermostatted cuvette holder. Initially, one standard method was used for measuring NADH oxidase activity of the recombinant gene products. The standard assay mixture contained 100 m M potassium phosphate buffer (pH 7.0), 0.06 m M FAD, 0.29 m M NADH and an appropriate amount of enzyme. The activity was determined by monitoring the oxidation of NADH at 334 nm and at 70 °C(e 334 ¼ 6.18 m M )1 Æcm )1 ) [17]. In addition, separate assays were used for determining NoxA-1 and NoxB-1 activity, which were performed at 80 °Cand70°C, respectively. NoxB-1 was less stable than NoxA-1, and therefore the assays were performed at 70 °C instead of 80 °C. For NoxA-1 the assay mixture contained potassium phosphate/sodium citrate buffer (50 m M each; pH 8.0), 0.06 m M FAD, 0.29 m M NADH and an appropriate amount of enzyme. For NoxB-1 the same assay mixture was used, except that the pH was adjusted to 6.5. Specific activities were calculated from the initial linear change in absorbance. Absorbance changes were corrected for non- enzymatic NADH conversion. Where indicated FAD was omitted from the assay mixture. When electron acceptors other than oxygen were tested, all assay constituents were made anoxic by repeated evacuation and gassing with N 2 gas in stoppered serum bottles. The stoppered cuvettes were also evacuated and gassed with N 2 , and the different components were added by syringe. The following extinction coefficients were used to calculate the specific activities: ferricyanide, 1.00 m M )1 at 420 nm; 2,6 dichlorophenolindophenol (DCPIP), 20 m M )1 at 600 nm; 5,5¢-dithiobis(2-nitrobenzoic acid) (DTNB), 13.6 m M )1 at 412 nm, benzyl viologen, 8.6 m M )1 at 578 nm; cytochrome c, 21.1 m M )1 at 550 nm. For menadi- one and 2,3-dimethyl-1,4-naphthoquinone extinction coef- ficients were determined as 2.72 and 2.39 m M )1 at 334 nm, respectively, based on a 1 : 1 stoichiometry with NADH. To investigate whether the NADH oxidases produced either H 2 OorH 2 O 2 , an activity assay without FAD was run to completion at 50 °Canda50-lL sample from the reaction mixture was tested in a separate assay at 22 °Cfor the presence of H 2 O 2 . This second mixture (1 mL) con- tained potassium phosphate/sodium citrate buffer (50 m M each, pH 6.5), 0.5 m M 3,3¢-dimethoxybenzidine, and 7 U horseradish peroxidase. The increase in absorbance (460 nm) was compared to a standard curve, which was prepared separately using known amounts of H 2 O 2 .The assay was not disturbed by residual NADH, which might have remained in the Nox assay. The decrease in NADH in the first assay was related to the amount of H 2 O 2 found in the second assay. Protein was determined according to Bradford [18] using the Bio-Rad protein assay kit, with BSA as standard. Analysis of catalytic properties Kinetic parameters of NoxA-1 and NoxB-1 were deter- mined in the specific assay systems, by measuring the initial rate at different starting concentrations of NADH in the presence of ambient dissolved oxygen concentrations. K m (for NADH) and V max values were obtained by a computer-aided direct fit to the Michaelis–Menten curve ( TABLE CURVE 2D ). The K m values for O 2 were determined from one single assay, in which the stoppered cuvettes were completely filled with the specific assay buffers. The buffers, which were equilibrated at 60 °C were calculated to contain 0.135 m M of dissolved oxygen [19]. The reaction was started upon addition of anoxic NADH. The decrease in oxygen concentration was calculated from the decrease in NADH, assuming that for each mole of NADH one mole of O 2 was required (in case only H 2 O 2 was produced). The reaction rates at the various O 2 concentrations were subsequently fitted using the Michaelis–Menten equation. It was assumed that the NADH concentration was saturating, meaning that only apparent K m values were obtained. Ó FEBS 2003 H 2 O 2 -forming NADH oxidases from Archaeoglobus fulgidus (Eur. J. Biochem. 270) 2887 The temperature dependence of NoxA-1 and NoxB-1 was determined in the range 20–90 °C. The pH-dependence was determined in the standard potassium phosphate/ sodium citrate buffer at 80 °Cor70°C, for NoxA-1 or NoxB-1, respectively. Stability analysis The thermostability of the enzymes was tested by incubating the purified enzyme in potassium phosphate buffer (100 m M , pH 7.0) in a closed vial in a water bath at 80 °C. At regular time intervals a sample was taken and tested in the standard assay. For NoxB-1 the stability was determined also in the presence of 2 m M dithiothreitol. Half-life values were calculated from a fit of the data (exponential decay: y ¼ aÆe –bx ). Results Characterization based on amino acid sequence The sequences of the three NADH oxidases from A. fulgi- dus that were investigated here were aligned with various other NADH oxidase sequences, available at the NCBI database. The resulting phylogenetic tree (Fig. 1) clearly showed that the three NADH oxidases belong to different phylogenetic clusters. NoxA-1 falls within a group of typical NADH oxidases of  49 kDa, including the well-studied NADH oxidases from Enterococcus feacalis and Strepto- coccus mutans [3,21]. This group also contains a NADH peroxidase, which performs a NADH-dependent reduction of H 2 O 2 to H 2 O, and a coenzyme A disulfide reductase, which catalyses the NADPH-dependent reduction of CoA-S-S-CoA to CoA-SH [4,5]. The NADH oxidase and -peroxidase are H 2 O-forming enzymes and are believed to play a role in oxidative stress defence or in the recycling of oxidized pyridine nucleotides. The CoA disulfide reductase is involved in maintaining a sufficiently high intracellular thiol concentration [5]. The alignment (not shown) revealed conserved binding motives for FAD or NAD(P) and, moreover, most sequences in this group contain a redox active cysteine, which is regarded essential for the reduction of O 2 to H 2 O and for the disulfide reduction [22]. The NoxA-4 from A. fulgidus (GI:11498556), a NADH oxidase gene from P. horikoshii (GI:14590747), and three genes from Sulfolobus solfataricus (GI:15899854, GI:15897884, GI:15899669) do not contain this cysteine. NoxB-1 belongs to a small group of 72-kDa proteins, of which several are involved in the reduction of unsaturated acids or aldehydes. For instance, enoate reductase (enr) from Clostridium tyrobutyricum, 2,4 dienoyl-CoA reductase (fadH) from E. coli, and NADH:flavin oxidoreductase (baiH) from Eubacterium sp. strain VPI 12708, all perform a NAD(P)H-dependent reduction of a carbon–carbon double bond [23–25]. This group also contains the NADH oxidase of T. brockii, but no physiological role has been ascribed to it [7]. The alignment revealed potential ligands for an iron– sulfur cluster and FAD or NAD(P) binding domains (Fig. 2). NoxC belongs to a group of small 20-kDa proteins, that form H 2 O 2 instead of H 2 O. The NADH oxidases from Thermus aquaticus and Thermus thermophilus also belong to this group [7,8]. The physiological role of these enzymes is not known. Cloning and expression All three nox genes gave gene products when expressed in E. coli BL21 (DE3) as judged by SDS/PAGE (Fig. 3). NoxA-1 gave a clear band of the expected size (48 kDa). Expression of noxB-1 was less clear, but still visible on the gel. The expected molecular mass based on the sequence of NoxB-1 is 68 kDa. NoxC appeared as two proteins of 36 kDa and 20 kDa. The larger protein may represent undenatured NoxC, because upon prolonged boiling in SDS-sample buffer, it disappeared and the expected 20-kDa protein increased (data not shown). Cell-free extracts of E. coli producing NoxA-1 and NoxB-1 showed significant NADH oxidase activity (see below). For NoxC, however, no NADH oxidase activity was measured. Also, in the presence of FMN instead of FAD, or NADPH instead of NADH no NoxC activity was found For this reason, further purification and characterization of NoxC was abandoned. Purification of the recombinant enzymes NoxA-1 and NoxB-1 were purified essentially by the same procedure. The first step that capitalized on their thermo- stability was a heat treatment, resulting in the removal of E. coli proteins. Whereas this treatment worked fine with the NoxA-1, NoxB-1 remained contaminated with E. coli proteins. Four subsequent chromatographic steps were necessary to obtain a homogeneous preparation (Tables 1 and 2). Gelfiltration of NoxA-1 on Superdex 200 resulted in two activity peaks, corresponding to molecular masses of the native NoxA-1 of approximately 94 and 178 kDa, which suggests that the NoxA-1 exists as dimer and to some extent as tetramer. Most homologous NADH oxidases from mesophiles (Enterobacter feacalis, Streptococcus mutans and Serpulina hyodysenteria) are monomers or dimers [3,21,26]. A NADH oxidase from P. furiosus,which was recently described, also existed as dimer [10]. Gelfiltra- tion of NoxB-1 resulted in a major activity peak with a shoulder, corresponding to molecular masses of approxi- mately 70 and 152 kDa. These data suggested that the NoxB-1 exists as monomer and for a minor part as dimer. Other homologous enzymes have been shown to have different quaternary structures, like a trimer (Eubacterium sp.), hexamer (T. brockii) or dodecamer (Clostridium tyrobutyricum) [6,23,25]. Catalytic properties The NADH oxidase activity of NoxA-1 and NoxB-1 was stimulated upon addition of FAD (60 l M ) to the assay mixture. For NoxA-1 this stimulation was  2.5 fold, for NoxB-1 3.7-fold. Addition of FMN instead of FAD did not stimulate NADH oxidase activity. This result suggested that both Nox enzymes contain FAD as prosthetic group, and that part of the protein had apparently lost its cofactor. Indeed, during the purification of especially NoxA-1, it was observed that the yellow colour of the enzyme fractions gradually disappeared. From the UV/visible spectrum of 2888 S. W. M. Kengen et al.(Eur. J. Biochem. 270) Ó FEBS 2003 NoxA-1 (and NoxB-1), with an absorbance maximum at 450 nm, it could be concluded that a flavin is present (data not shown). For NoxA-1 and NoxB-1 different pH optima of 8.0 and 6.5 were found, respectively (Fig. 4). At their pH optimum, apparent V max values were found of 8.7 ± 0.5 UÆmg )1 for NoxA-1 and 1.5 ± 0.03 UÆmg )1 for NoxB-1. The latter activity was found to be strongly influenced by the presence of mercaptans. Dithiothreitol (DTT; 2 m M )or b-mercaptoethanol stimulated NoxB-1 activity up to two- fold. In contrast, NoxA-1 activity was inhibited by DTT (2 m M ), e.g. in its presence the activity rapidly decreases to <10% of the activity without DTT. The affinity for NADH was the highest for NoxB-1 (apparent K m ¼ 0.011 ± 0.001 m M ) (Fig. 5). NoxA-1 showed a rather low affinity for NADH (apparent K m ¼ 0.13 ± 0.014 m M )compared to NoxB-1 and to other thermoactive NoxA homologues from P. furiosus (K m <4l M )orA. fulgidus (NoxA-2: K m ¼ 3.1 l M ). NoxA-1 and NoxB-1 did not show activity with NADPH. For NoxA-1 an apparent K m for oxygen of 0.06 ± 0.03 was determined. The K m value for oxygen of NoxB-1 appeared to be much higher and for this reason difficult to assess. The fit-program resulted in an apparent K m of 2.9 m M , far above the maximum dissolved oxygen concen- tration (0.086 m M at a pO 2 of 0.2 10 5 Pa at 80 °C). Fig. 2. Multiple alignment of NoxB-1 homologues. Conserved and moderately conserved residues are shaded black or grey. Putative NAD- or FAD-binding motifs are boxed. Cysteine residues of a putative ferredoxin-like motif are indicated by arrowheads. The abbreviations used are as follows (Genebank identifier in parentheses): Nox Tbro, NADH oxidase of Thermoanaerobacter brockii (GI:48123); DienoylCoA, 2,4-dienoyl-CoA reductase of E. coli (GI:1176118); NADH:flav, NADH: flavin oxidoreductase of Eubacterium sp. (GI:416702); Enoate red, enoate reductase of Clostridium acetobutylicum (GI:15026455); Nox Sso, NADH oxidase (SSO2025) of Sulfolobus solfataricus (GI:15898816). Ó FEBS 2003 H 2 O 2 -forming NADH oxidases from Archaeoglobus fulgidus (Eur. J. Biochem. 270) 2889 Addition of EDTA to the assay mixture did not cause a decrease in activity of NoxA-1 or NoxB-1, suggesting that divalent cations are not required for activity. Effect of temperature on activity For NoxA-1and NoxB-1 an identical temperature optimum of 80 °C was found, corresponding to the physiological growth optimum of the organism (Fig. 6). Up to 70 °C, the increase in NADH oxidase activity followed a linear Arrhenius plot (ln k vs. 1/T; data not shown), from which activation energies of 76.6 kJÆmol )1 and 52.2 kJÆmol )1 could be calculated for NoxA-1 and NoxB-1, respectively. Despite the identical temperature optima, the thermostabi- lity of both Nox enzymes at this temperature was consid- erably different. Whereas NoxA-1 showed a half-life of  40 h at 80 °C (data not shown), NoxB-1 showed a half- life of only 40 min. For NoxB-1, the temperature stability was investigated in the presence and absence of DTT. In addition to the stimulating effect on the absolute activity, DTT also raised the stability about twofold (half-life 83 min) (Fig. 7). The product of the NADH oxidase reaction NADH oxidases can perform the bivalent reduction of oxygen to H 2 O 2 or the tetravalent reduction of oxygen to H 2 O. Production of H 2 O 2 was tested by analysing the assay mixture in a separate peroxidase assay, using horseradish peroxidase and 3,3¢-dimethoxybenzidine as electron donor. In the NoxA-1 assay between 71% and 95% of the amount Fig. 3. SDS/PAGE of extracts of recombinant E. coli containing NoxA-1, NoxB-1 or NoxC from A. fulgidus. M, Calibration proteins. The molecular mass of the calibration proteins is indicated (kDa). Table 1. Purification scheme of NoxA-1 from A. fulgidus. Activities were determined in phosphate/citrate buffer pH 8.0. Total volume (mL) Protein (mgÆmL )1 ) Total protein (mg) Specific activity (UÆmg )1 ) Total activity (U) Purification (fold) Recovery (%) Crude extract 28 21.5 602 0.59 355.2 1.0 100 Heat-treatment 30.4 3.55 108 3.29 355 5.57 100 Q-sepharose 47.54 1.32 62.75 4.69 294 7.94 82.7 Hydroxyapatite 44.9 1.06 47.61 4.69 223.3 7.94 62.8 Macrosep 6.92 6.62 45.79 5.07 232 8.59 65.3 Superdex 200 93.38 0.298 27.83 5.82 161.9 9.86 45.6 Table 2. Purification scheme of the NoxB-1 from A. fulgidus. Activities were determined in phosphate/citrate buffer (pH 6.5). Total volume (mL) Protein (mgÆmL )1 ) Total protein (mg) Specific activity (UÆmg )1 ) Total activity (U) Purification (fold) Recovery (%) Crude extract 28 20.84 583.5 1.139 664.6 1.0 100 Heat-treatment 28.5 6.16 175.56 2.35 412.56 2.06 62.1 Q-sepharose 60.16 2.09 125.7 2.52 316.8 2.21 47.7 Hydroxyapatite 83.8 0.67 56.13 3.43 192.5 3.01 29 Macrosep 7.74 8.2 63.5 2.27 144.16 1.99 21.7 Superdex 200 196.9 0.153 30.1 4.05 122 3.55 18.3 Fig. 4. pH dependence of purified NoxA-1 (s)andNoxB-1(d). 2890 S. W. M. Kengen et al.(Eur. J. Biochem. 270) Ó FEBS 2003 of NADH which was converted was recovered as H 2 O 2 .In the NoxB-1 assay, this value amounted to 97%. FAD was omitted from these assays, because unbound FAD may facilitate H 2 O 2 production via nonenzymatic oxidation of FADH 2 [27]. These results indicate that both enzymes probably produce exclusively H 2 O 2 . The fact that the recovery of H 2 O 2 was sometimes less than 100% (70–80%), could be explained by the observation that the amount of H 2 O 2 measured, was influenced by the time period between the NADH conversion and the actual measurement of H 2 O 2 . Apparently, the amount of H 2 O 2 in the assay mixture slowly decreased, despite the absence of NADH, which was already completely converted at that moment. Electron acceptors other than oxygen Because the physiological role of NoxA-1 and NoxB-1 is not known, it was of interest to test various electron acceptors other than oxygen. The results of these assays, which were performed in the absence of oxygen, are summarized in Table 3. Flavines, like FAD, are known to be able to react with various one- or two-electron acceptors. In accordance, NoxA-1 and NoxB-1 show activity with several of the e-acceptors tested. DCPIP, ferricyanide, menadione and 2,3-dimethyl-1,4-naphthoquinone are e-acceptors commonly used in the detection of membrane bound dehydrogenases. However, the activities were rather low (see Discussion). NoxA-1 also showed some activity with a cytochrome c, but again the activity is low. One of the homologues of NoxA-1 has been identified as NADH peroxidase [4]. For this reason H 2 O 2 was tested as electron acceptor under anoxic conditions. Neither NoxA-1 nor NoxB-1 showed convincing NADH peroxidase activity. Another NoxA-1 homologue has recently been recog- nized as CoA disulfide reductase, an enzyme that performs a Fig. 5. Rate dependence of NoxA-1 and NoxB-1 on the NADH con- centration. Data points were fitted according to the Michaelis–Menten equation. Fig. 6. Temperature dependence of purified NoxA-1 (s) and NoxB-1 (d). Fig. 7. Thermal stability of NoxB-1. The purified enzyme was incu- bated at 80 °Cin100 m M phosphate buffer pH 7.0 in the presence (d) and absence (s)of2m M DTT. Table 3. Specific activity of NoxA-1 and NoxB-1 with different electron acceptors. Presence of FAD a NoxA-1 (UÆmg )1 ) NoxB-1 (UÆmg )1 ) Oxygen + 5.8 4.1 H 2 O 2 +/– 0 0 DCPIP – 4.2 1.0 Ferricyanide – 5.8 1.8 Menadione – 1.29 1.93 2,3-Dimethyl-1,4-naphthoquinone – 1.56 1.29 Cytochrome c – 0.03 0 Benzyl viologen + 4.5 1.3 DNTB + 3.7 0.52 Coenzyme A (oxidized) +/– 0 0 Glutathione (oxidized) +/– 0 0 Cystine +/– 0 0 Polysulfide + 0 0.65 b Tiglic acid +/– NT 0 Cinnamic acid +/– NT 0 Crotonate +/– NT 0 Fumarate +/– NT 0 a Activity was determined in the presence (+) or absence (–) of FAD. +/–, Substrates were tested with and without FAD. b Only in 100 m M Tris/CL buffer pH 7.8 at 60 °C. NT, not tested; tiglic acid, trans-2-methyl-2-butenoic acid; cinnamic acid, 3-phenyl- 2-propenoic acid; crotonate, 2-butenoate; menadione, 2-methyl- 1,4-naphthoquinone. Ó FEBS 2003 H 2 O 2 -forming NADH oxidases from Archaeoglobus fulgidus (Eur. J. Biochem. 270) 2891 disulfide reductase activity via a single cysteine residue [5]. Forthisreason,NoxA-1aswellasNoxB-1weretested using DTNB as e-acceptor. An activity of 3.68 UÆmg )1 was determined for NoxA-1. NoxB-1 also showed a significant disulfide reductase activity of 0.52 UÆmg )1 . Both activities were determined in the presence of FAD. In the absence of FAD, the reduction of DTNB was substantially less. Other disulfides of more physiological nature like oxidized Coen- zyme A, glutathione or cystine did not cause a NADH oxidation in the absence of oxygen, nor did they stimulate the DTNB reduction. A special type of electron acceptor tested here was polysulfide. Polysulfide is a soluble form of elemental sulfur, which has been shown to act as electron acceptor by various hyperthermophiles. NoxB-1, but not NoxA-1, showed a significant NADH-dependent polysulfide reductase activity of 0.65 UÆmg )1 . However, this activity is again low when compared to other polysulfide reductases, like the NADPH- dependent sulfide dehydrogenase (7.0 UÆmg )1 )from P. furiosus [28]. Because NoxB-1 showed homology to enzymes involved in the reduction of unsaturated acids or aldehydes (Fig. 2), a few model substrates were tested as potential electron acceptors (Table 3). However, none of these caused an oxidation of NADH when added to the anoxic reaction mixture. Discussion The noxA-1 and the noxB-1 gene from the hyperthermo- philic archaeon A. fulgidus were successfully cloned and expressed in E. coli. The alignment to homologous genes in the database revealed that NoxA-1 and NoxB-1 belong to different phylogenetic groups (Fig. 1). Nevertheless, the similarity to other NADH oxidase gene sequences does not simply lead to their physiological function. For instance, NoxA-1 belongs to the family of pyridine nucleotide disulfide oxidoreductases (Pfam), which contains enzymes that may function as NADH oxidase, NADH peroxidase or as CoA disulfide reductase. The various electron acceptors tested here did not indicate an obvious function (Table 3). The reduction of DCPIP and ferricyanide suggests that NoxA-1 may have a role as NADH dehydrogenase as part of the electron transport chain for sulfate reduction. Moreover, it has recently been suggested that NoxA-2 of A. fulgidus may also function in electron transport for sulfate reduction, because the enzyme copu- rified with D -lactate dehydrogenase and both enzymes colocalized to the periplasmic side of the membrane [9,30]. However, the activities found here for NoxA-1 towards menadione, 2,3-dimethyl-1,4-naphthoquinone and cyto- chrome c are rather low, and thus do not support this hypothesis. For example, the F 420 H 2 : quinone oxido- reductase from A. fulgidus showed specific activities of 96 UÆmg )1 and 92 UÆmg )1 with 2,3-dimethyl-1,4-naphtho- quinone and menadione, respectively [29]. A novel type menaquinone, present in the membrane fraction of A. fulgidus, probably acts as the physiological electron acceptor [31]. NoxA-1 showed substantial disulfidereductase activity (3.7 UÆmg )1 ), but this DNTB-reducing activity was not stimulated by disulfides like oxidized coenzyme A, gluta- thione or cystine. The activity was, however, strongly stimulated upon addition of FAD, which indicated that free FAD was involved in the reduction of DNTB. This suggests that the observed disulfide reduction may not be the physiological role of NoxA-1. Moreover, when we compare the disulfide reductase activity of NoxA-1 with that of a true disulfide reductase (CoA disulfide reductase from Staphylo- coccus aureus; Spec. activity ¼ 4570 UÆmg )1 ), the latter is at least 1000-fold more active [5]. In this respect we also tested whether NoxA-1 exhibited polysulfide reductase activity. Polysulfide is a soluble form of sulfur consisting of predominantly tetrasulfide (S 4 2– ) and pentasulfide (S 5 2– ), and it is assumed that an S–S bond is cleaved similar to the disulfide reduction. This activity was of special interest because a similar NADH oxidase (NoxA-2) of P. furiosus was shown by DNA microarray analysis to be strongly up-regulated (7.4-fold) when cells were grown in the presence of sulfur [32]. The expression of two other ORFs in the P. furiosus genome increased more than 25-fold, and their products termed SipA and SipB are proposed to be part of an S-reducing protein complex. Although A. fulgidus is not able to grow by sulfur reduction, its genome contains homologues of the SipA and SipB encoding genes. Unfor- tunately, NoxA-1 did not show polysulfide reductase activity. Nevertheless, the similarity to the S-upregulated NoxA-2 of P. furiosus and to the membrane associated NoxA-2 of A. fulgidus, suggests some respiratory role. Alternatively, the function of NoxA-1 may actually be that of an NADH oxidase, using the reducing power of NADH to remove traces of oxygen that otherwise may lead to harmful oxygen species like O 2 – ,H 2 O 2 ,orOHÆ. The K m for oxygen of NoxA-1 is  60 l M , which is not very low compared to the amount of oxygen that can maximally dissolve at 80 °C(102l M at ambient oxygen concentrations and average marine salinity). However, because A. fulgidus is a strict anaerobe, it will most likely have to deal with much lower oxygen concentrations. A role as detoxicant has also been proposed for the NADH oxidase (NOX1) from the hyperthermophile P. furiosus [10]. The affinity for oxygen of the latter enzyme, however, is even lower (a K m of at least 110 l M has been reported), which makes this enzyme also not very efficient if it is assumed to remove small amounts of oxygen in an anaerobic environment. Unfortunately, oxygen affinity data of true NADH oxidases are not available in the literature, making a comparison impos- sible. The most plausible argument against a role as detoxicant, however, is the fact that the NoxA-1 and also theenzymefromP. furiosus produce predominantly H 2 O 2 . Thus, instead of preventing oxidative stress through O 2 removal, these Nox enzymes aggravate the problem by producing H 2 O 2 . On the other hand, H 2 O 2 which is produced by the Nox, may be converted further by a catalase-peroxidase, which has also been demonstrated in A. fulgidus [33]. But in this case H 2 O 2 is converted back to O 2 , which combined with the NADH oxidase lowers the amount of oxygen by only 50%. The K m for oxygen of NoxB-1 is even much higher ( 3m M ), making a role as oxygen detoxifying system very unlikely. Moreover, also NoxB-1 produces H 2 O 2 instead of H 2 O. In a recent paper Abreu et al. describe a superoxide scavenging system in A. fulgidus, and propose that 2892 S. W. M. Kengen et al.(Eur. J. Biochem. 270) Ó FEBS 2003 NAD(P)H oxidases may have a role in oxygen detoxifica- tion, not by directly reducing oxygen, but via intermediate redox enzymes like rubredoxin and neelaredoxin [34]. A similar system has been proposed for Desulfovibrio gigas [35]. This hypothesis certainly deserves further attention, but requires purification of the rubredoxin and neelaredoxin. The production of H 2 O 2 is in contrast with the data on mesophilic homologues of NoxA-1 or NoxB-1, which all produce H 2 O. Also the NoxA-2 from A. fulgidus,the NADH oxidase from P. furiosus,andtheNADHoxidase from T. brockii have been demonstrated to form H 2 O 2 instead of the usual H 2 O [9,10]. Thus, possibly the production of H 2 O 2 is a thermophilic feature. It has been put forward that reduction of oxygen to H 2 O 2 may be an artefact, because in anaerobes the flavin moiety of flavo- proteins is exposed to the solvent and can easily transfer electrons to oxygen to form H 2 O 2 . In aerobes the flavin is protected from this unwanted oxygen reduction, because the flavin is buried in the protein. Remarkably, for the NADH oxidase from P. furiosus only 61% of the NADH was recovered as H 2 O 2 (NADH/H 2 O 2 ratio of 0.61), suggesting that the enzyme produced both H 2 O 2 and H 2 O [10]. Occasionally, we also found <100% recovery of H 2 O 2 compared to NADH, but this could be diminished by shortening the time period between the Nox assay and the H 2 O 2 -assay. Possibly, this also applies for the enzyme from Pyrococcus. Concerning the physiological role of the NoxA-1, the direct neighbourhood of noxA-1 inthegenomewas investigated. A CTP synthase, a GMP synthase and several hypothetical proteins accompany noxA-1, which do not provide insight as to the physiological role of the NoxA-1. A STRING analysis [36] of the surrounding genes, however, reveals that all NoxA and NoxB homologues have a predicted redox protein, regulator of disulfide bond forma- tion (COG0425) in their neighbourhood. This COG belongs to a functional category involved in post-translational modification, protein turnover, and chaperones, which also does not reveal a physiological role of the Nox enzymes. Recent experiments by Pagala et al. [30] have shown that whole cell extracts of A. fulgidus exhibit multiple NADH oxidase activities, as judged by renatured SDS/PAGE gels. It was concluded that the majority of the Nox enzymes in A. fulgidus are expressed constitutively under strictly anaerobic conditions. The fact that the expression of the Nox enzymes is not regulated, also suggests that they have some fundamental metabolic role, and not an occasional role during oxygen stress. NoxB-1 shows homology to a small group of enzymes that is involved in the reduction of unsaturated acids or aldehydes. For instance, enoate reductase (enr) from Clostridium tyrobutyricum, 2,4 dienoyl-CoA reductase (fadH) from E. coli, and NADH:flavin oxidoreductase (baiH) from Eubacterium sp. strain VPI 12708, all perform a NAD(P)-dependent reduction of a carbon–carbon double bond [23–25]. However, several commonly used unsatur- ated compounds, like tiglic acid, cinnamic acid or crotonate did not show any activity when tested with NoxB-1. Possibly, the enzyme requires CoA-activated unsaturated compounds, which have not been tested here. As mentioned above, the adjacent genes of NoxB-1 do not reveal any information concerning its function. On the other hand, the gene encoding NoxB-2 of A. fulgidus, which is 98.9% identical to NoxB-1, lies upstream of a gene encoding a medium-chain acyl-CoA ligase, suggesting a role in fatty acid and phospholipid metabolism. Thus, despite an extensive analysis of the catalytic capabilities of NoxA-1 and NoxB-1, no obvious physio- logical role can be ascribed to them. Further studies, for instance using Northern blots or DNA microarrays may indicate conditions at which the enzymes are expressed and thereby unveil their cellular function. Because both NADH oxidases produce H 2 O 2 instead of H 2 O, they may find application in biosensors as mediator between a dehydrogenase and the electrode. For this purpose the enzymes should have sufficient stability and appropriate kinetics. Although, the stability of NoxB-1 is considerably lower than that of NoxA-1 (stable at 80 °Cfor 1 h and 40 h, respectively), the stability is likely to be sufficient at more moderate temperatures. Concerning the catalytic activity, k cat /K m values of 0.053 · 10 6 M )1 Æs )1 and 0.156 · 10 6 M )1 Æs )1 can be calculated for NoxA-1 and NoxB-1, respectively. These catalytic efficiencies are sub- stantially lower than the value found for the NADH oxidase of Thermus thermophilus, which was determined at room temperature (k cat /K m ¼ 1.250 · 10 6 M )1 Æs )1 )[37].Thus, compared to the latter enzyme, which also forms H 2 O 2 and which is also reasonably stable, NoxA-1 and NoxB-1 are less suited for biosensor application. Acknowledgements This work was partly funded by the European Community under the Industrial & Materials Technologies Programme (Brite-Euram III) (Contract BRPR-CT97-0484). References 1. Stetter, K.O. (1988) Archaeoglobus fulgidus gen. nov., sp. nov. a new taxon of extremely thermophilic archaebacteria. Syst. Appl. Microbiol. 10, 172–1720. 2. Klenk, H.P., Clayton, R.A., Tomb, J.F., White, O., Nelson, K.E., Ketchum, K.A., Dodson, R.J., Gwinn, M., Hickey, E.K., Peterson, J.D., Richardson, D.L., Kerlavage, A.R., Graham, D.E., Kyrpides, N.C., Fleischmann, R.D., Quackenbush, J., Lee, N.H., Sutton, G.G., Gill, S., Kirkness, E.F., Dougherty, B.A., McKenney, K., Adams, M.D., Loftus, B. & Venter, J.C. (1997) The complete genome sequence of the hyperthermophilic, sul- phate-reducing archaeon Archaeoglobus fulgidus. Nature 390, 364–370. 3. Ross, R.P. & Claiborne, A. (1992) Molecular cloning and analysis of the gene encoding the NADH oxidase from Streptococcus faecalis 10C1. Comparison with NADH peroxidase and the fla- voprotein disulfide reductases. J. Mol. Biol. 227, 658–671. 4. Ross, R.P. & Claiborne, A. (1991) Cloning, sequence and over- expression of NADH peroxidase from Streptococcus faecalis 10C1. Structural relationship with the flavoprotein disulfide reductases. J. Mol. Biol. 221, 857–887. 5. Del Cardayre, S.B., Stock, K.P., Newton, G.L., Fahey, R.C. & Davies, J.E. (1998) Coenzyme A disulfide reductase, the primary low molecular weight disulfide reductase from Staphylococcus aureus. Purification and characterization of the native enzyme. J. Biol. Chem. 273, 5744–5751. 6. Maeda,K.,Truscott,K.,Liu,X.L.&Scopes,R.K.(1992)A thermostable NADH oxidase from anaerobic extreme thermo- philes. Biochem. J. 284, 551–555. Ó FEBS 2003 H 2 O 2 -forming NADH oxidases from Archaeoglobus fulgidus (Eur. J. Biochem. 270) 2893 7. Liu, X.L. & Scopes, R.K. (1993) Cloning, sequencing and expression of the gene encoding NADH oxidase from the extreme anaerobic thermophile Thermoanaerobium brockii. Biochim. Biophys. Acta 1174, 187–190. 8. Park, H.J., Kreutzer, R., Reiser, C.O. & Sprinzl, M. (1992) Molecular cloning and nucleotide sequence of the gene encoding a H 2 O 2 -forming NADH oxidase from the extreme thermophilic Thermus thermophilus HB8 and its expression in Escherichia coli. Eur. J. Biochem. 205, 875–879. 9. Reed, D.W., Millstein, J. & Hartzell, P.L. (2001) H 2 O 2 -forming NADH oxidase with diaphorase (cytochrome) activity from Archaeoglobus fulgidus. J. Bacteriol. 183, 7007–7016. 10. Ward, D.E., Donnelly, C.J., Mullendore, M.E., van der Oost, J., de Vos, W.M. & Crane, E.J. III (2001) The NADH oxidase from Pyrococcus furiosus. Eur. J. Biochem. 268, 5816–5823. 11. Liu,Z.,Niwa,O.,Horiuchi,T.,Kurita,R.&Torimitsu,K.(1999) NADH and glutamate on-line sensors using Os-gel-HRP/GC electrodes modifed with NADH oxidase and glutamate dehydro- genase. Biosens. Bioelectron. 14, 631–638. 12. Jaenicke, R. (1991) Protein stability and molecular adaptation to extreme conditions. Eur. J. Biochem. 202, 715–728. 13. Leuschner, C. & Antranikian, G. (1995) Heat-stable enzymes from extremely thermophilic and hyperthermophilic microorganisms. World. J. Microbiol. Biotechnol. 11, 95–114. 14. Blumentals, I.I., Itoh, M., Olson, G.J. & Kelly, R.M. (1990) Role of polysulfides in reduction of elemental sulfur by the hyperthermophilic archaebacterium Pyrococcus furiosus. Appl. Environ. Microbiol. 56, 1255–1262. 15. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, 2nd edn. Cold Spring. Harbor Laboratory Press, Cold Spring Harbor, NY. 16. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680–685. 17. Bergmeyer, H.U. & Gawehn, K. (1978) Principles of Enzymatic Analysis. Verlag Chemie, New York, Weinheim. 18. Bradford, M.M. (1976) A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248–254. 19. Weiss, R. (1970) The solubility of nitrogen, oxygen and neon in seawater. Deep-Sea Res. 17, 721–735. 20. Higgins, D.G. & Sharp, P.M. (1988) CLUSTAL: a package for performing multiple sequence alignment on a microcomputer. Gene 73, 237–244. 21. Matsumoto, J., Higuchi, M., Shimada, M., Yamamoto, Y. & Kamio, Y. (1996) Molecular cloning and sequence analysis of the gene encoding the H 2 O-forming NADH oxidase from Strepto- coccus mutans. Biosci. Biotechnol. Biochem. 60, 39–43. 22. Mallett, T.C. & Claiborne, A. (1998) Oxygen reactivity of an NADH oxidase C42S mutant: evidence for a C (4a)-peroxyflavin intermediate and a rate-limiting conformational change. Biochemistry 37, 8790–8802. 23. Rohdich, F., Wiese, A., Feicht, R., Simon, H. & Bacher, A. (2001) Enoate reductases of Clostridia. Cloning, sequencing and expres- sion. J. Biol. Chem. 276, 5779–5787. 24. He, X.Y., Yang, S.Y. & Schultz, H. (1997) Cloning and expression of the fadH gene and characterization of the gene product 2,4-dienoyl coenzyme A reductase from Escherichia coli. Eur. J. Biochem. 248, 516–520. 25. Franklund, C.V., Baron, S.F. & Hylemon, P.B. (1993) Char- acterization of the baiH gene encoding a bile acid-inducible NADH: flavin oxidoreductase from Eubacterium sp. Strain VPI 12708. J. Bacteriol. 175, 3002–3012. 26. Stanton, T.B. & Jensen, N.S. (1993) Purification and character- ization of NADH oxidase from Serpulina (Treponema) hyody- senteria. J. Bacteriol. 175, 2980–2987. 27. Toomey, D. & Mayhew, S.G. (1998) Purification and character- isation of NADH oxidase from Thermus aquaticus YT-1 and evidence that it functions in a peroxide-reduction system. Eur. J. Biochem. 251, 935–945. 28. Ma, K. & Adams, M.W.W. (1994) Sulfide dehydrogenase from the hyperthermophilic archaeon Pyrococcus furiosus:anewmul- tifunctional enzyme involved in the reduction of elemental sulfur. J. Bacteriol. 176, 6509–6517. 29. Kunow, J., Linder, D., Stetter, K.O. & Thauer, R.K. (1994) F 420 H 2 : quinone oxidoreductase from Archaeoglobus fulgidus.Chara- cterization of a membrane-bound multisubunit complex contain- ing FAD and iron-sulfur clusters. Eur. J. Biochem. 223, 503–511. 30. Pagala, V.R., Park, J., Reed, D.W. & Hartzell, P.L. (2002) Cellular localization of D -lactate dehydrogenase and NADH oxidase from Archaeoglobus fulgidus. Archaea 1, 95–104. 31. Tindall, B.J., Stetter, K.O. & Collins, M.D. (1989) A novel, fully saturated menaquinone from the thermophilic, sulphate-reducing archaebacterium Archaeoglobus fulgidus. J. General Microbiol. 135, 693–696. 32. Schut, G.J., Zhou, J. & Adams, M.W.W. (2001) DNA microarray analysis of the hyperthermophilic archaeon Pyrococcus furiosus: evidence for an new type of sulfur-reducing enzyme complex. J. Bacteriol. 183, 7027–7036. 33. Kengen, S.W.M., Bikker, F.J., Hagen, W.R., de Vos, W.M. & van der Oost, J. (2001) Characterization of a catalase-peroxidase from the hyperthermophilic archaeon Archaeoglobus fulgidus. Extremophiles 5, 323–332. 34. Abreu, I.A., Saraiva, L.M., Carita, J., Huber, H., Stetter, K.O., Cabelli, D. & Teixeira, M. (2000) Oxygen detoxification in the strict anaerobic archaeon Archaeoglobus fulgidus:superoxide scavenging by Neelaredoxin. Mol. Microbiol. 38, 322–334. 35. Chen,L.,Liu,M.Y.,Legall,J.,Fareleira,P.,Santos,H.&Xavier, A.V. (1993) Purification and characterization of an NADH- rubredoxin oxidoreductase involved in the utilization of oxygen by Desulfovibrio gigas. Eur. J. Biochem. 216, 443–448. 36. Snel, B., Lehmann, G., Bork, P. & Huynen, M.A. (2000) STRING: a web-server to retrieve and display the repeatedly occurring neighbourhood of a gene. Nucleic Acids Res. 28, 3442–3444. 37. Park, H.J., Reiser, C.O.A., Kondruweit, S., Erdmann, H., Sch- mid, R.D. & Sprinzl, M. (1992) Purification and characterization of a NADH oxidase from the thermophile Thermus thermophilus. Eur. J. Biochem. 205, 881–885. 2894 S. W. M. Kengen et al.(Eur. J. Biochem. 270) Ó FEBS 2003 . Molecular characterization of H 2 O 2 -forming NADH oxidases from Archaeoglobus fulgidus Serve ´ W. M. Kengen, John. NADH oxidases belong to different phylogenetic clusters. NoxA-1 falls within a group of typical NADH oxidases of  49 kDa, including the well-studied NADH