The tungsten-containing formate dehydrogenase from Methylobacterium extorquens AM1: Purification and properties

Authors


J. A. Vorholt, Laboratoire de Biologie Moléculaire des Relations Plantes – Microorganismes, INRA/CNRS, BP27, 31326 Castanet-Tolosan, France. Fax: +33 5 61 28 50 61, Tel.: +33 5 61 28 54 58, E-mail: vorholt@toulouse.inra.fr

Abstract

NAD-dependent formate dehydrogenase (FDH1) was isolated from the α-proteobacterium Methylobacterium extorquens AM1 under oxic conditions. The enzyme was found to be a heterodimer of two subunits (α1β1) of 107 and 61 kDa, respectively. The purified enzyme contained per mol enzyme ≈ 5 mol nonheme iron and acid-labile sulfur, 0.6 mol noncovalently bound FMN, and ≈ 1.8 mol tungsten. The genes encoding the two subunits of FDH1 were identified on the M. extorquens AM1 chromosome next to each other in the order fdh1B, fdh1A. Sequence comparisons revealed that the α-subunit harbours putative binding motifs for the molybdopterin cofactor and at least one iron–sulfur cluster. Sequence identity was highest to the catalytic subunits of the tungsten- and selenocysteine-containing formate dehydrogenases characterized from Eubacterium acidaminophilum and Moorella thermoacetica (Clostridium thermoaceticum). The β-subunit of FDH1 contains putative motifs for binding FMN and NAD, as well as an iron–sulfur cluster binding motif. The β-subunit appears to be a fusion protein with its N-terminal domain related to NuoE-like subunits and its C-terminal domain related to NuoF-like subunits of known NADH-ubiquinone oxidoreductases.

Abbreviations
FDH

formate dehydrogenase

FDH1

NAD-dependent formate dehydrogenase

H4F

tetrahydrofolate

H4MPT

tetrahydromethanopterin

DCPIP

2,6-dichlorophenolindophenol

The conversion of formate to CO2 is the terminal enzymatic step in C1 unit oxidation to CO2 in the α-proteobacterium Methylobacterium extorquens AM1 and other aerobic methylotrophic bacteria [1,2]. M. extorquens AM1 possesses two separate pathways for conversion of C1-units between the oxidation levels of formaldehyde and formate that are essential for growth on methylotrophic substrates, methanol and methylamine [1,3,4]. One of these pathways involves tetrahydrofolate (H4F)-dependent enzymes [4,5]. Its main function seems to be the provision of methylene–H4F for the assimilatory serine cycle and the H4F-bound C1-intermediates at different oxidation levels for various biosynthetic reactions. Formate is an intermediate in this pathway, a result of the formyl–H4F ligase reaction [6]. The second C1-converting pathway involves tetrahydromethanopterin (H4MPT)-dependent enzymes, and its main function seems to be in energy metabolism [4,5,7,8]. Some of the enzymes in this latter pathway exhibit sequence identity to enzymes that are an integral part of the energy metabolism in methanogenic archaea (methanogenesis) [9]. However, in contrast with the methanogenesis pathway, the H4MPT-dependent pathway in M. extorquens AM1 involves formate as an intermediate which is formed by the formyltransferase/hydrolase complex [2,10].

NAD-dependent formate dehydrogenase (FDH) activity in cell extracts of M. extorquens AM1 has been previously reported, but the enzyme has not been purified to homogeneity [11]. We were interested to learn more about the formate oxidation step in this bacterium, i.e. determine whether different FDH enzymes are present, and to which class they belong based on cofactor content and electron acceptor specificity (for a review see [12]).

FDHs from a number of methylotrophic bacteria have already been analysed, and it became evident that their occurrence is not uniform. Pseudomonas sp. 101 and Mycobacterium vaccae 10 were shown to express different NAD-linked FDH enzymes dependent on the presence of molybdenum: one devoid of a prosthetic group and one containing molybdenum. The latter was suggested to be partially active with tungsten as well [13,14]. The cofactor-free homodimeric NAD-dependent FDH from Pseudomonas sp. 101 has been studied in detail [15,16] and a similar enzyme was also purified from Moraxella sp. C-1 [17]. The molybdenum-containing FDH from the methane-oxidizing bacterium Methylosinus trichosporium was studied in detail as well. This FDH was shown to contain iron–sulfur clusters, a flavin and molybdenum. It is reported to be composed of either two [18] or four different subunits [19]. Methylobacterium sp. RXM was reported to exhibit high specific activity of NAD-dependent FDH when either molybdate or tungstate were present in the growth medium. In the absence of molybdate or tungstate, NAD-dependent FDH activity was detected only at low levels [20]. It was suggested that this bacterium contains only one FDH that is active with both tungsten and molybdenum [20]. The enzyme from Methylobacterium sp. RXM grown in molybdate-containing medium supplemented by methanol was partially purified [21], and its molecular mass reported to be 75 kDa.

While tungsten has been recognized as an important component in some formate dehydrogenases from anaerobic bacteria [22–24], the evidence for its presence in aerobic methylotrophic bacteria has been only indirect. Here we describe, for the first time, purification and properties of a tungsten-containing NAD-dependent FDH from the aerobic bacterium, M. extorquens AM1.

Materials and methods

Organism and growth conditions

M. extorquens AM1 was grown in the presence of methanol (120 mm) at 30 °C in a minimal medium as described previously [25]. Na2WO4 and (NH4)Mo7O21 were added separately to the culture medium, where specified, to the final concentration of 0.3 µm. The cells were cultivated in 10-L glass fermenters containing 8 L medium. The fermenters were stirred at 500 r.p.m and gassed with air (2 L·min−1). The cultures were harvested in the late exponential phase at a cell density of D578 = 3.5. Cells were pelleted by centrifugation at 5000 g and stored at −20 °C. Where indicated, cells were grown in 2-L Erlenmeyer flasks filled with 800 mL medium, shaken at 150 r.p.m.

FDH assays

A standard optical assay for FDH activity was performed at 30 °C, by following the reduction of NAD+ at 340 nm (ε340 = 6.2·mm−1·cm−1). The reaction mixture contained in a final volume of 0.72 mL 50 mm Tricine/KOH pH 7.0, 30 mm sodium formate, 0.5 mm NAD+ and an appropriate amount of protein. One unit of activity was defined as the amount of enzyme catalysing the reduction of 1 µmol NAD+ or an alternative electron acceptor. The following alternative electron acceptors were used: ferricyanide (FeCN ε420 = 1.02·mm−1·cm−1); 2,6-dichlorophenolindophenol (DCPIP ε600 = 16.3·mm−1·cm−1); FMN (ε445 = 12.5·mm−1·cm−1); FAD (ε450 = 11.3·mm−1·cm−1); NADP+340 = 6.2·mm−1·cm−1); benzyl viologen (ε578 = 6.25·mm−1·cm−1). Benzyl viologen was tested under anoxic conditions. To test the inhibitory effect of sodium azide, the enzyme was pre-equilibrated with 0.9 mm sodium azide for 2 min at 30 °C, then the reaction was started with the addition of sodium formate.

Protein purification

Frozen cells of M. extorquens AM1 were resuspended in 50 mm Mops/KOH pH 7,0 at 4 °C and passed three times through a French pressure cell at 120 MPa. Centrifugation was performed at 150 000 g for 45 min to remove cell debris, whole cells and the membrane fraction, which was shown to contain only ≈ 3% of the NAD-dependent FDH activity and benzyl viologen-dependent FDH activity in comparison with the cytosolic fraction (both under oxic and anoxic conditions, see below). Protein was determined by the Bradford assay using Bio-Rad reagent with BSA as a standard [26].

NAD-dependent FDH (FDH1) was purified from M. extorquens AM1 via four chromatographic steps at 4 °C under oxic conditions. All chromatographic materials were from Amersham Pharmacia Biotech. The soluble fraction of the cell extract (41 mL) was loaded on to a DEAE–Sephacel column (2.6 cm × 10 cm) equilibrated with 50 mm Mops/KOH pH 7.0. Protein was eluted with the following gradients of NaCl in this buffer: 50 mL 0 m NaCl, 5 mL 0–0.16 m NaCl, 50 mL 0.16 m NaCl, 325 mL 0.16–0.6 m NaCl, 5 mL 0.6–1 m NaCl, 65 mL 1 m NaCl, 5 mL 1–2 m NaCl, 60 mL 2 m NaCl. NAD-dependent FDH was eluted at ≈ 0.4 mm NaCl. Combined active fractions (76 mL) were diluted 1 : 2 in 50 mm Mops/KOH pH 7.0, and loaded on to a Source 15Q column (1.6 cm × 10 cm) equilibrated with the same buffer. Protein was eluted with the following gradients of NaCl: 250 mL 0–0.6 m NaCl, 20 mL 0.6–1 m NaCl, 50 mL 1–2 m NaCl. Active fractions (9 mL) were recovered at ≈ 0.6 m NaCl. These fractions were concentrated using the 30 kDa cut-off Centricon centrifugal filter units (Millipore) to a final volume of 0.15 mL. The protein was loaded on a Superdex 200 column, equilibrated with 50 mm Mops/KOH 0.1 m NaCl pH 7.0. Active fractions were loaded on a Resource Q column, equilibrated with 50 mm Mops/KOH pH 7.0. Purified protein was eluted with an increasing NaCl gradient (0–1 m NaCl). The purified enzyme was eluted at ≈ 0.4 m NaCl in 3 mL.

For anoxic purification of FDH1, the gas phase of serum bottles containing frozen cells of M. extorquens AM1 was replaced by 100% N2. Passing through the French Pressure cell and the centrifugation were performed under N2 atmosphere as well. All of the buffers used during the purification were depleted of oxygen by boiling for 5 min followed by cooling under vacuum with stirring, and the addition of 2 mm dithiothreitol. All of the chromatographic purification steps were performed in an anaerobic chamber (Coy) under gas atmosphere of 95% N2/5% H2 at 15 °C. Elution methods and profiles were similar to those described above.

Gel electrophoresis and molecular weight determination

Purified protein was subjected to electrophoresis in a 10% polyacrylamide gel and stained with Coomassie brilliant blue R250. The molecular masses of the subunits of purified FDH1 were also determined by MALDI-TOF analysis using Voyager-DE-RP (Applied Biosystems). The molecular mass of the native enzyme was estimated by gel filtration on a Superdex 200 column using the following standards: ferritin (440 kDa), catalase (232 kDa), aldolase (158 kDa), ovalbumin (43 kDa), and chymotrypsinogen (25 kDa). Peptide mass finger-printing was performed after trypsin digestion using Voyager-DE-STR (Applied Biosystems).

Determination of the N-terminal amino-acid sequence

Purified enzyme was separated by electrophoresis in the presence of SDS and electroblotted on to a poly(vinyl trifluoride) membrane (Applied Biosystems). The amino acid sequence was determinated on a 477-protein/peptide sequencer (Applied Biosystems).

Analytical methods

For the determination of tungsten, preparations of purified FDH1 were washed three times with 50 mm Mops/KOH pH 7.0 using Centricon centrifugal filter units (30-kDa cut-off, Millipore). Samples were analysed by neutron activation analysis. Molybdenum was determined with an atomic adsorption spectrometer Zeeman 3030 (Perkin Elmer). Nonhaem iron was quantified colorimetrically with neocuproin (2,9-dimethyl-1,10-phenanthroline) and ferrozine [3-(2-pyridyl)-5,6-bis-(4-phenylsulfonate)-1,2,4-triazine by the method of Fish [27] with Titrisol iron solution (Merck) as a standard. Acid-labile sulfur was determined as methylene blue [28] using Na2S as standard. Covalently and noncovalently bound flavins were determined as described [29]. For pterin cofactor determination, the pH of purified formate dehydrogenase (0.2 mg in 50 mm Mops/KOH pH 7.0) was adjusted to pH 2.5 with 2 mHCL, then 1% I2/2% KI was added to the acidified protein at a ratio of 1 : 20 (v/v), and the sample was heated for 30 min in boiling water bath, followed by cooling and centrifugation at 35 000 g for 10 min. The supernatant was filtered through the 30-kDa cut-off Centricon centrifugal filter units (Millipore) to remove any precipitate. As a positive control, milk xanthine oxidase (Sigma-Aldrich) was treated in the same way. As negative controls, FMN and NADH were used. Fluorescence spectra were recorded in a Carian Eclipse spectrofluorometer (Varian) at a fixed excitation wavelength of 380 nm and emission wavelengths of 380–700 nm.

Sequence analysis

The genes encoding the subunits of FDH1 purified in this study were identified via blast search against the genomic database of M. extorquens AM1 (http://vixen.microbiol.washington.edu/), using the N-terminal amino acid sequence of the β-subunit as a query. The sequence of 4741 bp containing fdh1A and fdh1B has been deposited with GenBank under the accession number AF489516. The amino acid sequences translated from fdh1A and fdh1B were used as queries to search the nonredundant database (http://www.ncbi.nlm.nih.gov). The sequences for the putative formate dehydrogenase subunits homologous to the subunits of FDH1 were also identified in the Methylococcus capsulatus genome whose sequence has been released before publishing by the Institute for Genomic Research (http://tigrblast.tigr.org/ufmg/).

Results and discussion

Purification of NAD-dependent formate dehydrogenase and identification of encoding genes

Cell extracts of M. extorquens AM1 grown in the presence of methanol in a fermenter under standard conditions (0.3 µm molybdenum, no addition of tungsten) contained NAD-dependent FDH activity of about 0.1–0.2 U·mg−1 and benzyl viologen-dependent FDH activity in the same range. Purification of FDH was attempted under both oxic and anoxic conditions. Both conditions resulted in similar protein yields, specific activity and stability of the enzyme, therefore only results of purification under oxic conditions are shown in Table 1. One major FDH activity peak was detected via NAD- or benzyl viologen-dependent activity determination upon each purification step. After four chromatographic steps, FDH was enriched 520-fold with a yield of 15% and a specific activity of 73 U·mg−1.

Table 1. Purification of NAD-dependent formate dehydrogenase (FDH1) from M. extorquens AM 1. Enzyme activity was determined at 30 °C under standard assay conditions. Cells were cultivated in 8 L-fermenters in the presence of methanol in minimal medium supplemented with 0.3 µm molybdenum.
 Total protein
(mg)
Total activity
(U)
Specific activity
(U·mg−1)
Purification
(fold)
Recovery
(%)
Soluble fraction13431850.141100
DEAE–Sephacel741522.11582
Source 15 Q8.89410.77651
Superdex 2001.2504230027
Resource Q0.37277352115

SDS/PAGE analysis revealed the presence of two polypeptides, of molecular masses of ≈ 105 and ≈ 60 kDa, respectively (Fig. 1). MALDI-TOF analysis also showed the presence of two molecules, of 107 and 61 kDa. The N-terminal amino acid sequence of the smaller subunit was determined to be: SEASGTV?SFAHPG?G?NVA?AVPKG?QVDP. It was, however, not possible to determine the N-terminal amino acid sequence of the larger subunit using a number of different preparations of the protein. The gene encoding the smaller subunit (fdh1B) was identified in the unfinished genome database of M. extorquens AM1 (L. Chistoserdova and M. E. Lidstrom, unpublished data), via blast search with the N-terminal amino acid sequence shown above. The 26 amino acid residues identified by Edman degradation were identical to the respective N-terminal 26 amino acid residues in the polypeptide translated from fdh1B (see Fig. 3). Fdh1B has a predicted molecular mass of 62 kDa, which is in agreement with the experimentally determined mass for the β-subunit (61 kDa). The gene located 56 nucleotides downstream of fdh1B potentially encodes a polypeptide with a predicted molecular mass of 107 kDa, which is in perfect agreement with the determined molecular mass of the larger subunit. The identity of this ORF as the larger subunit of FDH1 was confirmed by peptide mass finger-printing analysis. All of the seven most intense mass peaks obtained after trypsin digestion of the larger subunit of FDH1 fitted within a 20-p.p.m. range of the predicted masses calculated using peptidemass (http://www.expasy.org). The ORF dowstream of fdh1B was therefore designated fdh1A.

Figure 1.

SDS/PAGE analysis of purified FDH1 of M. extorquens AM1. Proteins were separated in a 10% polyacrylamide gel and stained with Coomassie brilliant blue R250. Lane A, molecular mass standards: phosphorylase B (97 kDa), albumin (66 kDa), ovalbumin (45 kDa), carbonic anhydrase (30 kDa); lane B, 5 µg purified FDH1. The use of a higher percentage polyacrylamide gel (14%) did not indicate the presence of smaller polypeptides (data not shown).

Figure 3.

Alignment of amino acid sequences for Fdh1B from M. extorquens AM1 with subunits of NAD(P)-dependent nickel-hydrogenases, subunits of the soluble FDH from Ralstonia eutropha and subunits of NADH-ubiquinone oxidoreductases. The N-terminal part of Fdh1B from M. extorquens AM1 (amino acid 1–185) is aligned with HoxE from Synechococcus sp. [38], the γ-subunit of FDH from Ralstonia eutropha[30], and NuoE from Aquifex aeolicus[37]. The C-terminal part of Fdh1B from M. extorquens AM1 (amino acid 186–578) is aligned with HoxF from Anabaena variabilis[36], the β-subunit of FDH from Ralstonia eutropha[30], and NuoF from A. aeolicus[37]. Identical residues are shown by dark-grey boxes. Conserved regions for binding FMN (according to expasy, http://www.expasy.ch), 4Fe−4S iron–sulfur cluster, and NAD (NCBI, http://www.ncbi.nlm.nih.gov/BLAST/) are underlined.

Apart from the genes for FDH1, the genome of M. extorquens AM1 contains two additional gene clusters potentially encoding formate dehydrogenases (L. Chistoserdova, M. Laukel, J. A. Vorholt & M. E. Lidstrom, unpublished data), one similar to the soluble NAD-dependent FDH characterized from R. eutropha[30] and one similar to the membrane-bound FDH from Wolinella succinogenes[31]. The presence of multiple FDH enzymes in M. extorquens AM1 lead us to use a nonstandard gene nomenclature (fdh1AB), which will aid in the future in discriminating between the three different enzymes (FDH1, FDH2 and FDH3).

Sequence analysis

Analysis of the amino acid sequence translated from fdh1A revealed similarity to the molybdopterin binding family of FDHs. Fdh1A shares ≈ 40% of identical amino acid residues with the catalytic α subunits of the two tungsten and selenocysteine-containing FDHs from Eubacterium acidaminophilum[32] and with the α-subunit of FDH from the thermophilic acetogenic bacterium Moorella thermoacetica (Clostridium thermoaceticum) (Acc. No. U73807), and it shows about 35% identity with FDHH from Escherichia coli[33](Fig. 2). The highest sequence identity, however, was found with the polypeptides translated from, respectively, the genomic sequence of Methylococcus capsulatus (63%), another aerobic methylotrophic bacterium (http://tigrblast.tigr.org/ufmg/) and a DNA region sequenced in the course of Leishmania major genome sequencing project that is believed to belong to an unknown bacterium (64%, Acc. No. AC091510; data not shown).

Figure 2.

Alignment of amino acid sequences for Fdh1A from M. extorquens AM1, FdhAI and FdhAII from Eubacterium acidaminophilum[32], FdhA from Moorella thermoacetica(U73807), and FdhH from E. coli[33]. The alignment was performed using the clustal w method (dnastar). Identical residues present in at least four of the sequences are marked by grey boxes. Conserved regions for the 4Fe−4S iron–sulfur cluster binding are shown in blue. Amino acids coordinating the two molybdopterin guanine dinucletoide cofactors of FdhH of E. coli (MGD801 and MGD802) that were found to be invariant among molybdopterin-containing FDH [34] are shown in red. Amino acids coordinating MGD801 and MGD802 that are less well conserved among the family of molybdopterin-containing FDH [34] but are conserved in the sequences shown here are marked in pink. The selenocysteine at position 140 of FdhH that is a direct ligand of the Mo [34] is shown in dark-red.

FDHH from E. coli was studied in detail biochemically and crystal structures of the enzyme are known [34]. The enzyme contains selenocysteine, molybdenum and two molybdopterin guanine dinucleotide cofactors, and a [4Fe−4S] cluster. Twelve conserved amino acid residues have been identified that coordinate directly the two molybdopterin cofactors and are conserved among the known sequences of molybdopterin-containing FDH enzymes [34]. The molybdopterin cofactors each provide two sulfur atoms for the ligation of the central Mo/W atom. Besides, the Mo/W is coordinated by the selenium atom of selenocysteine and cysteine, respectively, that corresponds to position 140 of FDHH of E. coli. The alignment shown in Fig. 2 indicates that the sequence of FDH1 from M. extorquens AM1 fits well in the family of molybdopterin- containing FDH. All of the amino acid residues that have been shown to participate in the coordination of the central Mo/W atom via two pterin cofactors are also identified in the sequence of FDH1.

The primary sequence of the β-subunit of FDH1 from M. extorquens AM1 indicates that the protein is composed of two regions: an N-terminal region (amino acid residues 1–185) and a C-terminal region (amino acid residues 186–572) (Fig. 3). The latter exhibits sequence identities of ≈ 30% to the subunit HoxF of pyridine-nucleotide-reducing nickel hydrogenases [35,36], the subunit NuoF of NADH-ubiquinone oxidoreductases from various organisms, e.g. Aquifex aeolicus[37], and to the β-subunit of formate dehydrogenase of Ralstonia eutropha[30] (Fig. 3). All of these sequences contain putative motifs for flavin, NAD, and iron–sulfur cluster binding sites. The N-terminal part exhibits sequence identities to the subunit HoxE of nickel hydrogenases, e.g. from Synechococcus sp. [38], to the subunit NuoE of NADH-ubiquinone oxidoreductases and the γ-subunit of formate dehydrogenase from R. eutropha[30]. All of these sequences contain four conserved cysteines, which might be involved in iron–sulfur cluster binding. Polypeptides showing the highest identities with Fdh1B (both at 58%) are translated from the chromosomes of M. capsulatus and of the unknown bacterial contaminant of L. major DNA (see above). In these two latter cases, the polypeptides also reveal the two-domain nature described above for Fdh1B.

Properties

The optimum pH for formate oxidation with NAD+ was determined to be between pH 8.0 and 8.5 in 120 mm potassium phosphate buffer.

Purified FDH1 could reduce the artificial electron acceptors DCPIP and benzyl viologen. However, none of the natural electron acceptors, i.e. FAD, FMN, or NADP+, could replace NAD+. The apparent Km values for FDH1 were determined to be 1.6 mm for sodium formate and 0.07 mm for NAD+. Sodium azide is known as a transition state analogue of formate and therefore a general inhibitor of FDHs [39]. A 50% inhibitory effect was observed in the presence of 0.9 mm sodium azide in standard assay conditions. A stabilizing effect of sodium azide and potassium nitrate on some FDHs upon enzyme purification was reported [21,40,41]. This effect was not observed for FDH1 from M. extorquens AM1.

The UV/visible spectrum of the purified enzyme (Fig. 4), as well as its brownish/yellowish colour were indicative of the presence of cofactors, and this finding is in agreement with the primary sequence analysis data (see above). The spectrum shows shoulders at about 362, 375, 415, 455, and 570 nm. The absorption peaks at 375 and 455 might originate form a flavin (see below), other peaks might be due to FeS centres in the enzyme. The enzyme could be partially reduced with either NADH or dithionite leading to a decreased absorption in the spectral range of interest.

Figure 4.

UV/visible absorption spectra of purified FDH1. Spectra shown are as isolated in an air-oxidized state (black line), reduced by the addition of crystalline NADH (dark grey line) or dithionite (pale grey line). Spectra were recorded with the enzyme (0.14 mg·mL−1) in 50 mm Mops/KOH buffer pH 7.0, against a buffer blank.

The iron content was determined to be at 5.4 mol·mol enzyme−1 and the acid-labile sulfur content was determined to be at ≈ 4.7 mol·mol enzyme−1, which might indicate that iron–sulfur clusters have been partially lost upon purification of the protein. In addition, the enzyme was found to contain 0.6 mol·mol−1 FMN while FAD could not be detected. Upon iodine oxidation, a fluorescent compound was liberated from the enzyme that exhibited an emission maximum at 470 nm and is supposed to be the pterin cofactor (Fig. 5). For comparison, a spectrum of identically treated milk xanthine oxidase is given. The amounts of purified FDH1 used for isolation of the pterin cofactor were not sufficient to allow quantification or the identification of the exact nature of the molybdopterin cofactor. However, the primary sequence analysis of FDH1 clearly indicates that all of the amino acid residues required for coordination of the two molecules of molybdopterin cofactor are present (Fig. 2). It therefore seems very likely that FDH1 contains two molybdopterin cofactors per active site as is generally found for molybdopterin-containing prokaryotic oxotransferases [42].

Figure 5.

Emission fluorescence spectra of purified FDH1 and milk xanthine oxidase after iodine treatment. (I) Fluorescence spectrum of FDH1 after oxidation with KI/I2; (II) the same spectrum corrected for FMN and Raman; (III) fluorescence spectrum of milk xanthine oxidase. The emission fluorescence spectra were measured at pH < 2.5. The emission spectra were taken at an excitation wavelength of 380 nm. The peak at 470 nm indicates the presence of a pterin cofactor in FDH1.

No molybdenum was detected in the enzyme by neutron activation analysis, or by atomic adsorption spectrometry. Instead, tungsten could clearly be detected by neutron activation analysis in the purified FDH1. The stoichiometric calculation indicated a ratio of about 1.8 mol tungsten·mol enzyme−1. This value is probably somewhat overestimated as the coordination of more than one tungsten in the active site could not be expected. Thus, FDH1 from M. extorquens AM1 is a tungsten-containing enzyme. Even though the sequence of FDH1 indicates a closer relatedness to known tungsten-containing FDHs than to known molybdenum-containing FDHs, this finding is still very surprising. Untill now, FDHs of aerobic bacteria were generally believed to be molybdenum-dependent enzymes or enzymes devoid of prosthetic groups [12,43]. The presence of a tungsten-containing formate dehydrogenase in a strictly aerobic bacterium may indicate that tungsten-containing enzymes are not restricted to anaerobic organisms and are probably more widespread than previously believed. For example, M. capsulatus, another aerobic methylotroph, also contains genes potentially encoding an enzyme very similar to FDH1 (see above), and a membrane-bound tungsten FDH has been detected in R. eutropha[44]. Another surprising property of FDH1 is the lack of oxygen sensitivity, while all of the previously characterized tungsten-containing FDHs were reported to be extremely oxygen-sensitive [22].

Tungsten was determined in FDH1 preparations from M. extorquens AM1 even if no tungstate was added to the growth medium, when cultivated in fermenters. We assume that in these cases tungsten must have been leached from the steel of the fermenters as the fermenters used in this study have been routinely used for cultivating methanogenic archaea, and respective media have been supplemented with tungstate. Since tungsten was clearly preferred over molybdenum for incorporation into FDH1, even in the presence of excess of molybdate in the growth medium, M. extorquens AM1 must possess a specific high-efficiency tungstate transporter. Such a transporter belonging to the ABC transporter group was recently identified in Eubacterium acidaminophilum[45]. Its existence was initially predicted based on the fact that tungstate was present in FDH preparations isolated from cells grown in the absense of tungstate [46]. TupA, the substrate binding subunit of this transporter was shown to have a Kd value of 0.5 µm for tungstate, whereas molybdate and sulfate were bound weakly when added at a more than 1000-fold molar excess [45]. The analysis of the genomic database of M. extorquens AM1 (L. Chistoserdova and M. E. Lidstrom et al. unpublished data) indicates the presence of a putative ABC transporter whose putative substrate-binding protein shows 46% sequence identity to TupA from Eubacterium acidaminophilum. Gene clusters similar to the one in Eubacterium acidaminophilum containing tupA are also found in Vibrio cholerae, Campylobacter jejuni, Haloferax volcanii and Methanothermobacter thermautotrophicus, and a function was suggested for these genes in the specific uptake of tungstate [45]. It is very likely that the TupA orthologue in M. extorquens AM1 serves such a function.

Effect of molybdate and tungstate on methylotrophic growth of M. extorquens AM1 and NAD-dependent FDH activity

To test the effect of the addition of molybdate and tungstate on methylotrophic growth of M. extorquens AM1 and FDH activity, we performed batch culture experiments using Erlenmeyer flasks. No significant effect on growth of the wild type M. extorquens AM1 was observed when molybdate or tungstate or none of these trace elements were added to the methanol-containing growth medium. However, the activity of NAD-dependent FDH varied depending on the presence of these trace elements. Extracts of cells grown in the medium to which tungstate and no molybdate were added exhibited a specific activity of approximately 0.2 U·mg−1. About half of this specific activity was found in extracts of cells grown in the medium containing molybdate and no tungstate. In the absence of either of the trace elements, FDH activity was only ≈ 0.003 U·mg−1. This low activity is probably an indication that no tungsten contamination was present in the flasks. The activity of NAD-dependent FDH found in these conditions in the presence of molybdate might be due to an alternative, molybdenum-containing enzyme that we have so far been unable to detect biochemically in fermenter grown cultures, possibly due to tungstate inhibition. This alternative enzyme might be encoded by a cluster of four genes homologous to the genes encoding the soluble FDH of R. eutropha[31] (see above). A third gene cluster is present in the M. extorquens AM1 genome, potentially encoding a membrane-bound FDH similar to the one characterized from W. succinogenes[32]. Work is in progress focusing on the roles of the three different FDHs in M. extorquens AM1 and their expression pattern under different growth conditions.

Acknowledgements

This work was supported by the Max-Planck-Gesellschaft, the Centre National de la Recherche Scientifique, the Deutsche Forschungsgemeinschaft, and the Public Health Service National Institutes of Health (GM58933). We thank D. Alber (Hahn-Meitner-Institut, Berlin, Germany) for the determination of tungsten, A. Pierik (University of Marburg, Germany) for helpful discussions, M. Rossignol (UMR 5546 CNRS/Université P. Sabatier, Castanet-Tolosan, France) for the peptide mass finger-printing analysis and D. Linder (University of Giessen, Germany), for determination of the N-terminal amino-acid sequence of the purified FDH1.

Ancillary