N2 Fixation by Streptomyces thermoautotrophicus Involves a Molybdenum-Dinitrogenase and a Manganese-Superoxide Oxidoreductase That Couple N2Reduction to the Oxidation of Superoxide Produced from O2by a Molybdenum-CO Dehydrogenase*

N2 fixation byStreptomyces thermoautotrophicus follows the equation N2 + 4–12MgATP + 8H+ + 8e − → 2NH3 + H2 + 4–12MgADP + 4–12Pi and exhibits features which are not obvious in the diazotrophic bacteria studied so far. The reaction is coupled to the oxidation of carbon monoxide (CO) by a molybdenum-containing CO dehydrogenase that transfers the electrons derived from CO oxidation to O2, thereby producing superoxide anion radicals (O·̄2). A manganese-containing superoxide oxidoreductase reoxidizes the O·̄2 anions to O2 and transfers the electrons to a MoFeS-dinitrogenase for the reduction of N2 to ammonium. Among the most striking properties of the S. thermoautotrophicus nitrogenase system are the dependence on O2 and O·̄2, the complete insensitivity of all components involved toward O2 and H2O2, the inability to reduce ethine or ethene, and a low MgATP requirement. In addition, the subunit structure of theS. thermoautotrophicus nitrogenase components and the polypeptides involved seem to be dissimilar from the known nitrogenases.

Streptomyces thermoautotrophicus UBT1 occurs in natural enrichments in the covering soil of burning charcoal piles (1,2). The bacterium is characterized by the utilization of gases as sources of energy, carbon and nitrogen (gasotrophy). S. thermoautotrophicus grows with CO or H 2 plus CO 2 under aerobic, chemolithoautotrophic, and thermophilic conditions (3). It is a free living dinitrogen fixer (4). Although CO and H 2 are known as inhibitors of nitrogenase activity, S. thermoautotrophicus is able to fix N 2 with CO or H 2 plus CO 2 as growth substrates as well as in the presence of CO plus ethine (4). The N 2 -fixing system of S. thermoautotrophicus is expressed constitutively (5). Intact bacteria reduced ethine to ethene (6) at a negligible level of less than 0.001% of the activity of Azotobacter vinelandii (7). We were, therefore, interested in the analysis of N 2 fixation by S. thermoautotrophicus with respect to the unusual components, structure, and reactivity of the nitrogenase system.

EXPERIMENTAL PROCEDURES
Bacterial Strain and Growth Conditions-S. thermoautotrophicus UBT1 (DSM 41605, ATCC 49746) was employed throughout this study (3). Bacteria were grown chemolithoautotrophically with CO as a sole source of carbon and energy in mineral medium supplied with trace elements (8) and 1.5 g of NH 4 Cl/liter (8) at 65°C. Fermentors of 70 liters total volume were flushed with a gas mixture composed of (v/v) 78% air, 13% CO, and 9% CO 2 at a flow rate of 3 liters/min. Bacteria were harvested by centrifugation and stored at Ϫ20°C until use.
Preparation of Subcellular Fractions-Crude extracts were prepared by passing bacterial suspensions (200 g of wet weight in 200 ml of 50 mM potassium phosphate buffer, pH 7.5, containing 0.5 mM phenylmethylsulfonyl fluoride and few crystals of DNase I) about 20 times through a French pressure cell under anoxic or oxic conditions. Solutions were made anoxic by evacuation, sparging with N 2 , and the addition of sodium dithionite (100 mg/liter). Cytoplasmic fractions were obtained from crude extracts by ultracentrifugation.
Purification of Nitrogenase Proteins-Cytoplasmic fractions were loaded onto a DEAE-52 cellulose column (height, 10.6 cm; diameter, 6.2 cm; bed volume, 250 ml) and eluted with 250 ml of phosphate buffer (50 mM, pH 7.5), followed by a linear gradient of 0 to 1 M NaCl in phosphate buffer. Fractions were analyzed for nitrogenase activity. The St2 1 protein appeared after elution in 100 -125 ml of phosphate buffer; the St1 protein eluted at 0.3 mM NaCl.
St1 protein was then subjected to centrifugation in a linear gradient of 8 to 35% (w/v) sucrose in phosphate buffer for 18 h at 100,000 ϫ g and 4°C. The St1 protein appeared at 19% sucrose. Fractions with nitrogenase activity were pooled, applied to a Sephadex G-150 gel filtration column (height, 44 cm; diameter, 2.5 cm; bed volume, 200 ml) and chromatographed with 200 ml of phosphate buffer. Unless otherwise stated, a 50 mM phosphate buffer, pH 7.5, was employed throughout this study. The St2 protein from DEAE cellulose chromatography was subjected to gel filtration on Sephadex G-100 with phosphate buffer as the eluent.
The purity of native proteins was examined by PAGE (9) at different pH values. Isoelectric focusing was performed with commercially available Isogel agarose isoelectric focusing plates (FMC Bio-Products).
Enzyme Assays-Nitrogenase was assayed in serum-stoppered 37-ml Wheaton vials containing 2 ml of phosphate buffer supplied with 2.5 mM ATP, 5 mM MgCl 2 , 25 mM Na 2 S 2 O 4 , 0.1 mg of St2 and 0.8 mg of St1. The gas atmosphere was pure N 2 or pure helium. Assays were kept unshaken at 65°C. Ammonium formation (10) was followed with time (for 2 h if not otherwise indicated) in samples withdrawn from the vials. Calculations of specific activities of nitrogenase were based on the total amount of protein (St1 plus St2) present in the assay.
Nitrogenase activity in the experiments of Table IV was assayed in Wheaton vials of 100 ml of total volume, containing 15 ml of phosphate buffer, 2.5 mM ATP, 5 mM MgCl 2 , and 30 mg of the cytoplasmic fraction. To this basic assay the additions specified in Table IV were made. The * This work was financially supported by a grant from the Deutsche Forschungsgemeinschaft (Bonn) and the Fonds der chemischen Industrie (Frankfurt/Main). 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.
‡ Recipient of a stipend from the Freistaat Bayern (Mü nchen). § To whom correspondence should be addressed. Tel.: 49-921-552729; Fax: 49-921-552727; E-mail: ortwin.meyer@uni-bayreuth.de. anoxic assays were amended with cysteine and dithiothreitol (4 mg each) and O 2 -free gases, and the vials were equipped with a central cylinder containing 50 mg of sodium dithionite in 1 ml of phosphate buffer to trap traces of potentially contaminating O 2 .
CO dehydrogenase was assayed spectrophotometrically using 1-methoxyphenazine methosulfate and INT as the electron acceptor (11). Formation of superoxide was assayed by the reduction of p-nitro blue tetrazolium as described previously (12). Electron transfer from St2 to physiological or artificial electron acceptors (50 M acceptor in phosphate buffer) was assayed spectrophotometrically. H 2 O 2 was analyzed colorimetrically with the iron-sensitive dye xylenol orange (13).
Measurements of ATP Hydrolysis-Hydrolysis of MgATP during the reduction of N 2 to ammonium as shown in Fig. 5 was examined in Wheaton vials containing 0.625 mM ATP, 1.25 mM MgCl 2 , 6.25 mM Na 2 S 2 O 4 , 6 mg of St1, and 0.75 mg of St2 in 8 ml of phosphate buffer under a gas atmosphere of pure N 2 at 65°C. Analysis of ammonium was as described above for the conventional nitrogenase assay. ATP hydrolysis was analyzed by isocratic high pressure liquid chromatography (HPLC) on an anion exchange column (ET 250/4 NUCLEOSIL 100 -10 SB, Macherey-Nagel, Dü ren, Germany) using 500 mM potassium phosphate buffer (pH 3.5) as a mobile phase at a flow rate of 1 ml/min. The column outflow was continuously analyzed in a diode array detector (Waters 991, Waters, Eschborn, Germany).
Determination of Metals, Acid-labile Sulfide, and Sulfhydryls-Protein samples were wet-ashed with concentrated sulfuric acid and H 2 O 2 and analyzed for molybdenum with the dithiol method (14) and for iron with the bathophenanthrolinedisulfonic acid method (15). Manganese, as well as molybdenum and iron, were also analyzed by inductively coupled plasma mass spectroscopy. The methylene blue method (16) was employed for the determination of acid-labile sulfide. Sulfhydryl groups of native or SDS-treated proteins were estimated with 5,5Јdithio-bis-2-nitrobenzoic acid (17).

Purification and Subunit Composition of Nitrogenase Components-Centrifugation of cytoplasmic fractions of CO grown
S. thermoautotrophicus in linear sucrose density gradients revealed three distinct protein bands of brown color with approximate molecular masses of 115, 49, and 147 kDa, designated protein St1, protein St2, and protein St3, respectively. The St3 protein was identified as CO dehydrogenase on the basis of the ability to catalyze the reduction of INT with CO, a molecular mass ϳ123 kDa ( Fig. 1) and the presence of the polypeptides CoxL (molecular mass ϳ87 kDa), CoxM (molecular mass ϳ32 kDa) and CoxS (molecular mass ϳ17 kDa) (Fig. 1). Previous studies had identified the CO dehydrogenase of S. thermoautotrophicus as a molybdo-iron-sulfur-flavoprotein containing the MCD type of molybdenum cofactor (21). Mixtures of the proteins St1 and St2 catalyzed the reduction of N 2 to NH 4 ϩ with dithionite as electron donor, whereas the individual components did not, suggesting a two-component type of nitrogenase.
In four independent preparations under anoxic conditions, the St1 protein was purified 8-fold with a yield of 11% and a mean specific activity for the reduction of N 2 of 109 nmol of NH 4 ϩ /mg of protein/h ( Table I). The St2 protein was purified about 20-fold, with a yield of 15% and a mean specific activity of 278 nmol of NH 4 ϩ /mg of protein/h ( Table I) Upon nondenaturating PAGE, the St2 protein revealed a single 40-kDa band at pH 6.5 or pH 7.5 ( Fig. 1), or a single 98-kDa band at pH 8.5. Other methods revealed masses of 49 kDa (sucrose density gradient centrifugation) or 41.5 kDa (gel filtration). Denaturating PAGE showed a single noncovalently bound 24-kDa subunit, designated D (Fig. 1), suggesting a homodimeric subunit structure. The N terminus 2 MFELPPLPYPYDALEPYFDAKKMEIHYYGGHGA of the St2-D polypeptide revealed no significant sequence similarity to sequenced nitrogenase polypeptides. Instead, St2-D showed 96.0% sequence similarity and 72.0% identity to the N terminus of the manganese-containing SODs of Bacillus stearothermophilus (22) or Bacillus caldotenax (23). In assays containing xanthine oxidase, xanthine, and O 2 for the production of superoxide anion radicals ( Ultraviolet and Visible Spectra, Contents of Metals, and Acid-labile Sulfide-The UV-visible spectrum of air-oxidized St1 protein revealed a protein peak at 274 nm and a broad nondistinct band in the region from 400 to 800 nm (Fig. 2). The dithionite-reduced spectrum was relatively featureless (Fig. 2). The visible spectra of oxidized and reduced St1 protein were similar to those of Kp1 (25) and different from those of Cp1 (26) or Av1 (27). St1 is a molybdo-iron-sulfur protein which contains molybdenum, iron, and acid-labile sulfide in a 1:17:11 molar ratio (Table II).
The St2 protein exhibited a pale yellow-greenish color and contained substoichiometric amounts of manganese and zinc (Table II). Other components could not be demonstrated. In its overall shape the UV-visible spectrum of St2 was similar to that reported for the "SOD active" protein P1 from Euglena gracilis (28). The spectrum revealed a protein absorption at 276 nm and a broad absorption extending from 310 to 700 nm with a small maximum at 411 nm and a broad shoulder around 474 nm.
Reactions Catalyzed by the Proteins St1 and St2- Maximum activity for the formation of NH 4 ϩ from N 2 was obtained at a molar ratio of St1 to St2 of 2.2:1 (Fig. 3). An ATP-regenerating system (e.g. creatine phosphate and creatine kinase) was not required. A 10-fold excess of MgADP over MgATP had no effect on nitrogenase activity. MgADP is apparently not inhibitory for S. thermoautotrophicus nitrogenase. Formation of NH 4 ϩ from N 2 as well as formation of H 2 from H ϩ was linear with time for at least 2.5 h (Fig. 4, 5). In the absence of N 2 , only H 2 but no ammonium was formed (Fig. 4). Reduction of ethine or ethene by mixtures of St1 and St2 was below the detection limit of 0.2 pmol of carbon of the gas chromatographic assay applied. There was a linear correlation between MgATP hydrolysis to MgADP and NH 4 ϩ formation from N 2 (Fig. 5). At the assay temperature of 65°C, ATP was almost chemically stable, whereas significant hydrolysis of ADP to AMP and inorganic phosphate occurred (Fig. 5). There was a linear correlation between the molar formation of ammonium from N 2 and the molar amount of MgATP under different assay conditions (Fig. 6). The regression line followed the equation [NH 4 ϩ ] ϭ 0.35833 ϫ [ATP] ϩ 0.0409. Formation of 2 mol of NH 4 ϩ , which is equivalent to the reduction of 1 mol of N 2 , had a mean requirement of 5.5 Ϯ 1.0 mol of MgATP (n ϭ 22). The minimum and maximum MgATP/N 2 ratios were 3.85 or 11.34, respectively (Fig. 6). This is significantly less than the minimum of 16 ATP reported for nitrogenases from other sources (29 -31).
In assays with limiting amounts of N 2 in helium, 1 mol of N 2 was reduced yielding 2.2 Ϯ 0.2 mol of NH 4 ϩ and 0.9 Ϯ 0.2 mol  (Table IV). MnSODs from Escherichia coli and B. stearothermophilus could not substitute for the St2 protein (Table IV). Oxidation of H 2 by CO dehydrogenase in the presence of O 2 also leads to O 2 . formation (Table IV).     anions are free intermediates and can be trapped by SOD (Table IV) with similar efficiency as ferredoxins, flavodoxins, or hydroquinones in the known nitrogenase systems (34,35). In addition, the use of O 2 . in N 2 fixation by S. thermoautotrophicus seems to be a powerful mechanism to scavange O 2 . radicals.
The dinitrogenase reductases known so far are ␥ 2 dimeric iron proteins with molecular mass ϳ63 kDa and contain 4Fe and 4S 2Ϫ atoms per dimer (36). In contrast, the St2 protein of S. thermoautotrophicus has been identified as a manganesesuperoxide oxidoreductase with molecular mass ϳ48 kDa and does not contain Fe or S 2Ϫ (Table II (Fig. 7).
Like the known nitrogenases, the S. thermoautotrophicus system also has a requirement for MgATP. With S. thermoautotrophicus nitrogenase the most efficient MgATP/N 2 ratio is 4, instead of 16 reported for nitrogenases from other sources, indicating its superb efficiency. Future work must unravel how MgATP interacts with the nitrogenase proteins. It is likely that the reduction of N 2 takes place at the St1 protein of S. thermoautotrophicus. The known MoFeS-dinitrogenases are complex ␣ 2 ␤ 2 tetrameric proteins with M r ϳ230 kDa containing 2 molybdenum atoms and 30 -34 iron atoms and an approximately equivalent number of acid-labile sulfide atoms (36). The St1 protein was also identified as a MoFeSprotein but with molecular mass ϳ144 kDa and a differing LMS heterotrimeric subunit structure ( Figs. 1 and 7). It revealed per molybdenum atom, 13.8 -21.7 iron atoms and 8.8 -15 acid-labile sulfide atoms (Table II). A substoichiometric amount of acid-labile sulfide compared with iron is frequently found with nitrogenases (37). There was a moderate sequence similarity between the N-terminal sequences of the subunits St1-L and Kp1-␤ or St1-M and Kp1-␣, although we are aware of the limitations of a comparison of subunit N termini. The overall reaction catalyzed by S. thermoautotrophicus nitrogenase compares to that of known nitrogenases, including the concomitant formation of H 2 (Fig. 4).
In consequence of the inability of the S. thermoautotrophicus nitrogenase to reduce ethine or ethene, the widely used ethine reduction assay is not applicable for measuring nitrogenase activity in environmental samples inhabited by bacteria performing this new type of N 2 fixation. It can be expected that the insensitivity of the S. thermoautotrophicus nitrogenase to O 2 will open new avenues for practical applications. Studies on the electron paramagnetic resonance properties of the metal centers in the S. thermoautotrophicus nitrogenase system as well as on the genetics and molecular biology of CO dehydrogenase (St3 protein) and St1 and St2 proteins, which are currently underway, will add further clues to the understanding of the novel nitrogenase system.