R. Hedderich, Max-Planck-Institut für terrestrische Mikrobiologie, Karl-von-Frisch-Straße, D-35043 Marburg/Germany. Fax: + 49 6421178299, Tel.: + 49 6421178230, E-mail: email@example.com
From the membrane fraction of the Gram-positive bacterium Carboxydothermus hydrogenoformans, an enzyme complex catalyzing the conversion of CO to CO2 and H2 was purified. The enzyme complex showed maximal CO-oxidizing:H2-evolving enzyme activity with 5% CO in the headspace (450 U per mg protein). Higher CO concentrations inhibited the hydrogenase present in the enzyme complex. For maximal activity, the enzyme complex had to be activated by either CO or strong reductants. The enzyme complex also catalyzed the CO- or H2-dependent reduction of methylviologen at 5900 and 180 U per mg protein, respectively. The complex was found to be composed of six hydrophilic and two hydrophobic polypeptides. The amino-terminal sequences of the six hydrophilic subunits were determined allowing the identification of the encoding genes in the preliminary genome sequence of C. hydrogenoformans. From the sequence analysis it was deduced that the enzyme complex is formed by a Ni-containing carbon monoxide dehydrogenase (CooS), an electron transfer protein containing four [4Fe−4S] clusters (CooF) and a membrane bound [NiFe] hydrogenase composed of four hydrophilic subunits and two membrane integral subunits. The hydrogenase part of the complex shows high sequence similarity to members of a small group of [NiFe] hydrogenases with sequence similarity to energy conserving NADH:quinone oxidoreductases. The data support a model in which the enzyme complex is composed of two catalytic sites, a CO-oxidizing site and a H2-forming site, which are connected via a different iron–sulfur cluster containing electron transfer subunits. The exergonic redox reaction catalyzed by the enzyme complex in vivo has to be coupled to energy conservation, most likely via the generation of a proton motive force.
Microorganisms can utilize a variety of exergonic redox-reactions to gain energy for growth. NADH, H2, and formate play an important role as electron donors, while O2, nitrate, Fe3+, fumarate, and sulfate are widespread electron acceptors . There are also several organisms known that can oxidize CO using O2 as electron acceptor. Among these carboxydotrophic bacteria, Oligotropha carboxidovorans has been intensively studied [2,3]. Only a few microorganisms have been identified that can grow anaerobically with CO under chemolithoautotrophic conditions and couple the oxidation of CO to CO2 with the reduction of protons to H2.
Organisms known to grow at the expense of this reaction are the gram-negative bacteria Rhodospirillum rubrum and Rubrivax gelatinosus[4–6] and the Gram-positive bacterium Carboxydothermus hydrogenoformans[7,8].
The biochemical process underlying the conversion of CO to CO2 and H2 has been most intensively studied in R. rubrum. The carbon monoxide oxidation system (Coo) is encoded by the coo regulon, which consists of two gene clusters regulated by the cooA gene . The cooFSCTJ gene cluster encodes the catalytic subunit (CooS) of the CO dehydrogenase, an electron transfer protein (CooF) of the CO dehydrogenase, and proteins required for the insertion of Ni into the enzyme (CooC, T, and J) [10,11]. The CO dehydrogenase is a nickel iron–sulfur protein and has been purified as a single subunit protein (CooS) . The crystal structure of this enzyme has been determined . Under certain purification conditions, CooS copurifies with the iron–sulfur protein CooF, which mediates the electron transfer from CooS to a membrane-bound hydrogenase .
The second gene cluster cooMKLXUH encodes the hydrogenase [15,16], which belongs to a small group of membrane-bound [NiFe] hydrogenases. Members of this family include Ech hydrogenase from Methanosarcina barkeri[17,18] and Escherichia coli hydrogenase 3 [19,20]. These membrane-bound hydrogenases characteristically contain six conserved subunits – four hydrophilic proteins and two integral membrane proteins. Two of the hydrophilic subunits (HycE and HycG in E. coli hydrogenase 3) are related to the hydrogenase large and small subunit conserved in all [NiFe] hydrogenases. The overall sequence similarity is, however, very low. Furthermore, the hydrogenase small subunit is considerably smaller than that of ‘standard’[NiFe] hydrogenases and contains only the cysteine ligands for the proximal [4Fe−4S] cluster. The large and small subunits of the membrane-bound hydrogenases are more closely related to subunits of the energy-conserving NADH:quinone oxidoreductase (complex I) . The membrane-bound hydrogenases contain at least four additional subunits not found in standard [NiFe] hydrogenases, but which have homologues in complex I.
Other members of this class of multisubunit membrane-bound [NiFe] hydrogenases are Eha and Ehb hydrogenases from the methanogenic archaeon Methanothermobacter marburgensis and Mbh hydrogenase from the hyperthermophilic archaeon Pyrococcus furiosus. The structure of these enzymes appears to be more complicated than that of the related [NiFe] hydrogenases. Ech hydrogenase from M. barkeri is the only member of this hydrogenase family that has been purified as an intact membrane complex; the other members have been found to be too labile to be purified intact . The characterization of a mutant strain lacking Ech hydrogenase has led to the elucidation of the physiological function of this hydrogenase in M. barkeri.
The thermophilic Gram-positive bacterium C. hydrogenoformans is also able to utilize CO as sole energy source, and it was therefore predicted to have a CO oxidation system similar to that of R. rubrum. The purification of two closely related CO dehydrogenases from C. hydrogenoformans, designated as CO dehydrogenase I and CO dehydrogenase II, has recently been reported . The two purified enzymes are homodimers of the catalytic subunit (CooS). The crystal structure of CO dehydrogenase II has been solved .
Here we report on the purification and catalytic properties of a membrane-bound enzyme complex from C. hydrogenoformans composed of a hydrogenase and a CO dehydrogenases. The complex catalyzes the conversion of CO to CO2 and H2.
Materials and methods
Dodecyl-β-d-maltoside was from Glycon Biochemicals. Carbon monoxide (99.997%) was from Messer Griesheim. All chromatographic materials were from Amersham Pharmacia Biotech or Bio-Rad. All other chemicals were from Merck or Sigma.
Growth of the organism
C. hydrogenoformans Z-2901 (DSM 6008) was from the Deutsche Sammlung für Mikroorganismen und Zellkulturen (Braunschweig). C. hydrogenoformans was grown strictly anaerobically in 10-L fermentors at 70 °C and pH 7.0 in the medium described in , with slight modifications. To avoid precipitation, the CaCl2 and MgCl2 concentrations were lowered to 1.2 and 1.1 mm, respectively; in addition pyruvate was added at a concentration of 10 mm. The fermentors were continuously supplied with 350 mL of CO/H2S (99.9 : 0.1, v/v) per min and stirred at 1000 r.p.m. When the culture reached an D578 of ≈ 2, the cells were harvested under N2 and stored at −20 °C.
Purification of a CO-oxidizing:H2-evolving enzyme complex
All purification steps were carried out under strictly anoxic conditions under an atmosphere of N2/H2 (95 : 5, v/v). Cell extracts were routinely prepared from 40 g cells (wet mass) suspended in 50 mL 50 mm Mops/KOH (pH 7.0) containing 2 mm dithiothreitol (buffer A). Lysozyme (5 mg) was added, and the suspension was incubated for 15 min at 20 °C. Cells were disrupted by sonication at 4 °C in intervals of 3 × 6 min using an energy output of 200 W (Bandelin sonicator) at 18 °C. Undisrupted cells and cell debris were removed by centrifugation at 10 000 g for 20 min.
Crude membranes were isolated from cell extracts by ultracentrifugation at 160 000 g for 2 h. Membranes were resuspended in 60 mL buffer A containing 0.5 mm detergent (dodecyl-β-d-maltoside) using a Teflon Potter homogenizer. After a second ultracentrifugation at 160 000 g for 2 h, washed membranes were homogenized in ≈ 70 mL buffer A (1.8 mg protein·mL−1) using a Teflon Potter homogenizer. Dodecyl-β-d-maltoside was added to a concentration of 16 mm[4.6 mg·(mg protein)−1]. The suspension was incubated for 12 h at 4 °C with slight swirling. After centrifugation at 160 000 g for 40 min, the solubilized membrane proteins present in the supernatant were loaded onto a Q-Sepharose HiLoad column (2.6 × 15 cm) equilibrated with buffer A containing 2 mm dodecyl-β-d-maltoside (buffer A + detergent). The column was washed with 60 mL buffer A + detergent. Protein was eluted in a stepwise NaCl gradient at a flow rate of 5 mL·min−1 (60 mL each in buffer A + detergent): 0.1, 0.2, 0.3, 0.4, and 0.5 m. The CO-oxidizing:H2-evolving enzyme complex was recovered in the fractions eluting with 0.3 m NaCl. Protein was concentrated and desalted by ultrafiltration (Molecular/Por cellulose ester ultrafiltration membranes, 100-kDa cut-off, Spectrum) and further purified by chromatography on ceramic hydroxyapatite (Bio-Rad). The column (1.6 × 20 cm) was equilibrated with 0.03 m potassium phosphate buffer pH 7.0 containing 2 mm dithiothreitol and 2 mm detergent. Protein was loaded and the column was washed with 50 mL of 0.03 m potassium phosphate buffer + detergent. Protein was eluted using a linear gradient from 0.03 m to 1 m potassium phosphate (400 mL). The CO-oxidizing:H2-evolving enzyme complex was recovered in the fractions eluting with 1 m potassium phosphate while part of the CO dehydrogenase activity was found in the 0.03 m potassium phosphate washing fraction.
Fractions containing CO-oxidizing:H2-evolving enzyme activity were concentrated by ultrafiltration as described above and applied to a Superdex 200 gel filtration column (2.6 × 60 cm) equilibrated with buffer A + detergent + 0.1 m NaCl. Protein was eluted using the same buffer. The enzyme complex eluted after 187 mL (peak maximum) corresponding to an apparent molecular mass of 450 kDa. Thyroglobulin (670 kDa), apoferritin (443 kDa), β-amylase (200 kDa) and alcohol dehydrogenase (150 kDa) were used to calibrate the column. Protein was concentrated by ultrafiltration and stored in buffer A + detergent at a protein concentration of 3 mg·mL−1 at 4 °C.
Purification of CO dehydrogenase
The hydroxyapatite fraction containing CO dehydrogenase activity but no hydrogenase activity (see above) was further fractionated by gelfiltration on a Superdex 200 column (2.6 × 60 cm) equilibrated with buffer A + detergent + 0.1 m NaCl. CO dehydrogenase eluted after 215 mL corresponding to an apparent molecular mass of 120 kDa. Protein was concentrated by ultrafiltration and analyzed by SDS/PAGE.
Determination of enzyme activities
The assays were routinely carried out under anoxic conditions at 70 °C either in 8-mL serum bottles or in 1.5-mL cuvettes. All assays contained 50 mm Mops/KOH (pH 7), 2 mm dodecyl-β-d-maltoside, and 2 mm dithiothreitol. One unit of enzyme activity corresponds to 1 µmol H2 or CO formed or consumed per min.
Hydrogen-uptake activity with methylviologen as electron acceptor was determined by following the reduction of methylviologen at 578 nm. The 0.8-mL assays contained 20 mm methylviologen and 0.1 mm sodium dithionite. In standard assays, cuvettes were allowed to equilibrate with 100% H2 in the headspace (1.2 × 105 Pa). CO-oxidizing activity was assayed under the same conditions, except that H2 was replaced by 100% CO in the headspace. One unit of H2- or CO-oxidation activity is defined as the reduction of 2 µmol of methylviologen per min, which is equivalent to 1 µmol of CO or H2 oxidized per min. When methylene blue was used as electron acceptor in the hydrogen-uptake assay or the CO dehydrogenase assay, the assay mixture contained 1.3 mm methylene blue instead of methylviologen.
The H2 formation activity with reduced methylviologen as electron donor was measured by following the oxidation of reduced methylviologen at 578 nm. The standard assay contained 2 mm methylviologen, which was reduced with sodium dithionite to a ΔE578 of 2, and N2 (1.2 × 105 Pa) as the gas phase. The reaction was started by the addition of protein.
The CO-dependent formation of H2 was followed by determining the H2 concentration in the gas phase in 1-mL assays in 8-mL serum bottles. Where indicated, the enzyme was activated with 1 mm Ti(III)citrate prior to the assay and 1 mm Ti(III)citrate was added to the assay mixture. The gas phase was 100% CO (1 × 105 Pa) or as indicated. The reaction was started by the addition of enzyme. The solution was stirred vigorously with a magnetic bar. At 1.5-min intervals, samples from the gas phase were withdrawn, and H2 was quantified after separation by gas chromatography. One unit of H2 formation activity is defined as 1 µmol of H2 produced min.
The gas chromatograph (model Carlo Erba GC Series 6000) was equipped with a thermal conductivity detector. Gases were separated by a molecular sieve (5 Å). The oven and injection port were at 110 °C; the detector was at 150 °C. The carrier gas was N2 at a flow rate of 30 mL·min−1.
Nonheme iron was quantified colorimetrically with neocuproin (2,9-dimethyl-1,10-phenanthroline) and ferrozine[3-(2-pyridyl)-5,6-bis(4-phenyl sulfonate)-1,2,4-triazine] as described by Fish . Acid-labile sulfur was analyzed as methylene blue .
Nickel was determined by atomic absorption spectroscopy on a 3030 Perkin Elmer atomic absorption spectrometer fitted with an HGA-600 graphite furnace assembly and an AS-60 autosampler.
Protein concentration was routinely determined by the method of Bradford using serum albumin as the standard .
Extraction of hydrophobic proteins from the purified enzyme complex
For the extraction of hydrophobic proteins a modification of the recently published procedure was used . 350 µg purified enzyme complex in 100 µL buffer A + detergent was added to 900 µL chloroform/methanol (67 : 33, v/v). The mixture was kept on ice for 15 min before centrifugation (4 °C) at 10 000 g for 20 min. The organic phase, which contained proteins soluble in chloroform/methanol, was dried in a SpeedVac. The dried pellet was solved in SDS sample buffer and was analysed by SDS/PAGE.
Determination of the stoichiometry of the different subunits present in the CO-oxidizing:H2-evolving enzyme complex
Two different methods were applied. Both methods rely on the binding of Coomassie brilliant blue to proteins. First, Coomassie-stained SDS gels were scanned (Scantouch 210, Nikon) and protein bands were quantified using the imagequant software (Molecular Dynamics).
Alternatively, Coomassie-stained protein bands were excised from SDS gels. In general protein bands from four lanes were combined. Protein-bound Coomassie was extracted with 200 mm NH4CO3 in 50% acetonitrile for 14 h at room temperature under vigorous shaking which resulted in a complete extraction of protein-bound dye. Gel-pieces were sedimented by centrifugation at 5000 g. The amount of dye present in the extract was determined spectrophotometrically at 590 nm.
In-gel tryptic digestion or CNBr cleavage of proteins and identification of peptides by MALDI-TOF mass spectrometry
Protein samples were separated by SDS/PAGE and the gel bands containing protein were excised after staining with Coomassie brilliant blue. Gel pieces were completely destained with 200 mm NH4CO3 in 50% acetonitrile (three times) and dried with a SpeedVac.
Tryptic digestion was started with the addition of a solution containing 0.2 µg trypsin per µL in 40 mm ammonium hydrogencarbonate buffer pH 8.1 containing 10% acetonitrile. Gel pieces were covered with this solution. The protein was digested for 14 h at 35 °C. The CNBr cleavages were performed for at least 14 h in the dark under N2 at 37 °C by adding a solution containing 1 m CNBr in 70% trifluoroacetic acid to the gel pieces. Peptides were extracted three times from gel pieces by sonication for 20 min in 60% acetonitrile, 1% trifluoroacetic acid. Extracts were combined and concentrated in a SpeedVac. Concentrated solutions were desalted with a ZipTipC18 (Millipore). Peptides were eluted from the ZipTipC18 using a solution of 10 mg·mL−1α-cyano-4-hydroxysuccinnamic acid in acetonitrile/H2O/trifluoroacetic acid (70 : 30 : 0.1, v/v/v). Subsequently, aliquots of 1 µL of the eluate were spotted onto a target disk and allowed to air-dry. Spectra were obtained using a Voyager DE RP MALDI time-of-flight mass spectrometer (Applied Biosystems). The accuracy of external calibration was, in general, better than 0.02%.
Purification of a CO-oxidizing:H2-evolving enzyme complex from Carboxydothermus hydrogenoformans
After cell breakage and separation of the membrane fraction from the soluble fraction, 80–90% of the hydrogenase activity and 60–70% of the CO dehydrogenase activity, both tested with methylviologen as electron acceptor, were found in the membrane fraction. The distribution of the enzyme activities strongly depended on the sonication conditions; prolonged sonication resulted in higher portions of the two enzyme activities in the soluble fraction. The membrane fraction also catalyzed the conversion of CO to CO2 and H2 at high rates (Table 1). Three enzyme assays were used to follow the purification of hydrogenase, CO dehydrogenase, and a possible complex of both enzymes (CO-oxidizing:H2-evolving enzyme activity) from the membrane fraction of C. hydrogenoformans (Table 1). The crude membrane fraction was washed with buffer containing 0.5 mm dodecyl-β-d-maltoside in order to remove included soluble proteins and proteins loosely associated with the membrane. In this washing step normally less than 1% of the hydrogenase but about 40% of CO dehydrogenase activity were released into the supernatant. Washed membranes containing tightly bound membrane proteins were solubilized using dodecyl-β-d-maltoside at a concentration of 16 mm. All three activities were almost completely recovered in the solubilized protein fraction. Proteins were further fractionated by anion-exchange chromatography on Q-Sepharose HiLoad, chromatography on hydroxyapatite and by gel filtration chromatography on Superdex 200. In these chromatographic steps, the three enzyme activities coeluted (Table 1). Only after chromatography on hydroxyapatite a partial separation was observed. A fraction eluting with 30 mm potassium phosphate from this column only contained CO dehydrogenase activity while a fraction eluting at 1 m potassium phosphate contained CO dehydrogenase-, hydrogenase- and CO-oxidizing:H2-evolving activity (Table 1). Proteins in this latter fraction were further purified by gelfiltration on Superdex 200. Hydrogenase-, CO dehydrogenase and CO-oxidizing:H2-evolving activity was found in a single peak eluting with an apparent molecular mass of 450 kDa. A fraction containing only hydrogenase and no CO dehydrogenase activity has never been obtained.
Table 1. Purification of a CO-oxidizing:H2-evolving enzyme complex from C. hydrogenoformans. The enzyme was purified from 40 g cells (wet mass). One unit of enzyme activity is defined as the amount of enzyme that catalyzes the reduction of 2 µmol methylviologen by 1 µmol CO (CO dehydrogenase activity) or H2 (hydrogenase activity) or the production of 1 µmol H2 from 1 µmol CO (complex activity). Activities were determined with either 100% H2 or 100% CO in the gas phase. Enzymes were not activated with Ti(III)citrate.
CO dehydrogenase activity
Washed membrane fraction
Solubilized membrane proteins
Hydroxyapatite (1 M potassium phosphate eluate)
The enzyme preparation thus obtained was subjected to SDS/PAGE. Prior to electrophoresis, the samples were either boiled in SDS sample buffer or incubated in SDS sample buffer at room temperature for 1 h. In the nonboiled samples, nine major polypeptides were observed with apparent molecular masses of 115, 89, 62, 41, 29, 23, 21, 18, and 16 kDa (Fig. 1A, lane 1). In boiled samples of the same fraction, the 89-kDa polypeptide was no longer detectable (Fig. 1A, lane 2). The intensity of the 115-kDa polypeptide decreased in boiled samples and a smear at the interface to the stacking gel was observed indicating protein aggregation. The apparent molecular mass of the 115-kDa polypeptide varied with the acrylamide concentration in the gel. This protein band was shifted to higher apparent molecular masses with increasing acrylamide concentrations. The value given was obtained with a 14% acrylamide gel. In 12% gels this polypeptide migrated with an apparent molecular mass of about 90 kDa. To identify hydrophobic polypeptides the purified protein fraction was extracted with chloroform/methanol (67 : 33, v/v). The 115-kDa and the 29-kDa proteins were selectively extracted using this solvent mixture, indicating a hydrophobic nature of these polypeptides (Fig. 1A, lane 3).
The hydroxyapatite fraction, which only contained CO dehydrogenase activity but no hydrogenase activity (see above), was further purified by gel filtration chromatography on Superdex 200. CO dehydrogenase activity eluted from this column with an apparent molecular mass of 120 kDa. An SDS/PAGE analysis of this fraction showed one protein band with an apparent molecular mass of 62 kDa in samples boiled prior to SDS/PAGE (Fig. 1B). This polypeptide was identified as CooSI by peptide mass finger printing. Trypsin digestion of this protein yielded peptides with the following masses (in Da), which matched a theoretical digest of CooSI (mass deviations from theoretical values are given in brackets): 1069.58 (0.05), 1100.52 (0.02), 1144.57 (0.00), 1144.57 (0.03), 1947.19 (−0.15), 2021.14 (−0.08), 2021.14 (−0.08), 2162.06 (−0.17), 2259.33 (−0.13), 2278.31 (−0.17), 2279.35 (−0.17), 2727.59 (−0.17).
Protein analysis by amino-terminal sequencing and mass fingerprinting, and identification of the encoding genes
The amino-terminal sequences of seven of the nine polypeptides detected in nonboiled protein samples were determined. The results are summarized in Table 2. The amino-terminal sequences of the 89- and 62-kDa protein bands were found to be identical. Since the 89-kDa band was not observed in boiled samples, this may indicate that this protein band is the dimer of the 62-kDa protein (see below). This is also supported by the observation that the intensity of the 62-kDa band increased in boiled samples. The amino-terminal sequences of the 115-kDa and of the 29-kDa polypeptides could not be determined (see below).
Table 2. Determination of the amino-terminal sequences of the subunits of CO-oxidizing:H2-evolving enzyme complex by Edman degradation. Polypeptides were separated by SDS/PAGE, blotted onto poly(vinylidene difluoride) membranes (Applied Biosystems) as described previously . Sequences were determined using an Applied Biosystems 4774 protein/peptide sequencer and the protocol given by the manufacturer. The names of the identified gene products are given in parentheses. Protein sequences derived from gene sequences are shown. Amino acids identified by Edman degradation are highlighted in bold. The 89-kDa protein band was found to be a mixture of two polypeptides. The major sequence corresponded to CooSI, the minor sequence corresponded to CooSII. For CooSI ≈ 20–40 pmol and for CooSII ≈ 5–10 pmol of each amino acid were found after each reaction cycle.
(CooSI) (main sequence)
(CooSII) (minor sequence)
62-kDa polypeptide (CooSI)
41-kDa polypeptide (CooH)
23-kDa polypeptide (CooU)
21-kDa polypeptide (CooFI)
18-kDa polypeptide (CooX)
16-kDa polypeptide (CooL)
Using the sequence information obtained, the encoding genes were identified in the genome of C. hydrogenoformans (Table 2). Preliminary sequence data were obtained from The Institute for Genomic Research website at http://www.tigr.org. The genes are organized in one hydrogenase gene cluster and a CO dehydrogenase gene cluster (Fig. 2). The amino acid sequences deduced from the nucleotide sequences of these genes show high similarity to the amino acid sequence of the subunits of CO-induced hydrogenase and of subunits of the carbon monoxide dehydrogenase from R. rubrum[10,16]. Therefore, the same nomenclature was chosen for the respective genes and proteins from C. hydrogenoformans (Fig. 2). Four of the polypeptides in the purified enzyme could be clearly assigned as subunits of the hydrogenase: CooH is the hydrogenase large subunit that contains the [NiFe] active site, CooL is the hydrogenase small subunit with one binding motif for a [4Fe−4S] cluster, CooX is an iron–sulfur protein predicted to ligate two [4Fe−4S] clusters, and CooU is a small protein with no known cofactor binding motifs. The hydrogenase gene cluster contains three additional genes, cooM, cooK, and hypA. cooM encodes a large integral membrane protein (136 kDa) predicted to form 34 transmembrane helices. The cooK gene encodes a second integral membrane protein with a molecular mass of 34.5 kDa and eight predicted transmembrane helices. Possible candidates for these integral membrane proteins are the 115- and 29-kDa polypeptides of the enzyme preparation, for which the amino-terminal sequence could not be determined. Both polypeptides are hydrophobic as shown by the chloroform/methanol extraction experiment. These polypeptides were in-gel trypsin or CNBr digested and the peptides obtained were analysed by MALDI-TOF mass spectrometry. The peptide masses were compared with the masses obtained from theoretical digests of the cooM and cooK derived polypeptides. Trypsin digestion of the 29-kDa polypeptide resulted in four major polypeptides, three of which matched the theoretical digest of CooK. These peptides with masses of 1557.87 Da (−0.01), 1564.86 Da (−0.01) and 1713.97 Da (−0.005) were found to be located in a large hydrophilic loop of CooK between amino acids 28 and 73. Most probably the hydrophobic regions of the protein were not accessible by the protease. CNBr cleavage, which is a more appropriate method to obtain peptides from integral membrane proteins , could not be used for the 29-kDa protein because of the low methionine content of this protein. CNBr cleavage of the 115-kDa polypeptide resulted in 12 peptides with molecular masses between 800 and 3000 Da, which could be assigned to CooM. Mass deviations of the measured masses from expected masses are given in brackets (in Da): 1004.3 (−0.07), 1134.54 (0.04), 1440.64 (0.19), 1501.81 (0.09), 1519.77 (0.05), 1568.77 (0.06), 2034.89 (0.08), 2147.07 (0.12), 2307.07 (0.23), 2458.20 (0.18), 2467.25 (0.13), 2584.17 (0.20). A theoretical CNBr cleavage of CooM resulted in 31 peptides in the mass range between 800 and 3000 Da, which is generally used for MALDI-TOF analysis. Based on these results the 115- and 29-kDa subunits can be tentatively assigned as CooM and CooK, respectively. The apparent molecular masses of these polypeptides, as determined from SDS gels, differ from the calculated molecular masses. This is typical for integral membrane proteins, which often show an anomalous migration in SDS gels . Both polypeptides shifted to higher apparent molecular masses at higher acrylamide concentrations as has been described for other membrane proteins .
A large intergenic region is found upstream of cooM. Directly downstream of cooH, a gene designated as hypA is located (Fig. 2). The gene product is closely related to the E. coli HypA and HybF proteins, which are essential for the maturation of E. coli hydrogenases .
The hypA gene is followed by an intergenic region of 116 base pairs. Downstream of this intergenic region are two genes designated cooF1 and cooSI. cooFI encodes an electron transfer protein predicted to ligate four [4Fe−4S] clusters. cooSI encodes the Ni- and Fe-containing catalytic subunit of CO dehydrogenase. A catalytically active CooSI dimer has been previously purified and characterized from C. hydrogenoformans. The 89- and 62-kDa polypeptides were assigned as CooSI, and the 21-kDa polypeptide was assigned as CooF1 via amino-terminal sequencing.
C. hydrogenoformans was previously shown to form a second CO dehydrogenase, designated CO dehydrogenase II. Upon inspection of the amino-terminal sequence of the 89-kDa polypeptide a background sequence was identified, which corresponds to the amino-terminal sequence of CooSII (Table 2). The data described in Materials and methods indicate a CooSI to CooSII ratio of 4 : 1 in the 89-kDa protein band. In contrast, the amino-terminal sequence of the 62-kDa protein band does not contain a background sequence corresponding to CooSII. From a densitometric analysis of Coomassie stained SDS-gels it can be estimated that ≈ 60–70% of CooS migrate as the monomer (62 kDa protein band) (data not shown). Hence, the overall content of CooSII on the total CooS content of the complex can be estimated to be less than 10%.
The two CooS proteins of C. hydrogenoformans have been shown to form homodimers [25,26]. This explains the occurrence of two forms of CooS in the SDS-polyacrylamide gels of nonboiled protein samples. The deviation of the apparent molecular mass of the dimer from the calculated value of 124 kDa may be related to the fact that the protein dimer is not completely unfolded and thus shows an anomalous migration behavior. The CooSII dimer might be more stable in SDS-containing buffer at room temperature explaining the observation that the CooSII monomer was not found under these conditions.
Subunit stoichiometry and cofactor content of the enzyme complex
In Coomassie-stained SDS gels the CO dehydrogenase subunits, CooS and CooF, showed a higher intensity compared to the hydrogenase subunits. A densitometric analysis revealed a molar ratio of CooS to CooH, the catalytic subunit of the hydrogenase, of 2.1 ± 0.1 : 1 and a molar ratio of CooF to CooH of 1.9 ± 0.1 : 1. The same results were obtained when the Coomassie dye, bound to the different subunits, was quantified after extraction with 200 mm ammonium carbonate in 50% acetonitrile. These data suggest a CO dehydrogenase to hydrogenase stoichiometry of 2 : 1 in the purified complex.
Gel filtration chromatography of the complex on a calibrated Superdex 200 column revealed an apparent molecular mass of the complex of 450 kDa. This value is consistent with the molecular mass of a complex containing a single copy of each hydrogenase subunit (CooM, K, L, X, U and H) and two copies of each CO dehydrogenase subunit (CooF and CooS) with a calculated molecular mass of 441 kDa.
The purified enzyme had a deep brown color. The air-oxidized minus sodium-dithionite-reduced difference spectrum showed a broad absorbance peak between 350 and 500 nm, indicative for the presence of iron–sulfur centers. The enzyme complex was found to contain 6.9 nmol Ni per mg protein, 139.5 nmol nonheme iron and 133 nmol acid-labile sulfur per mg protein. These values correspond to 3 mol Ni per mol enzyme, 61.5 mol nonheme iron and 59 mol acid labile sulfur per mol enzyme using the calculated molecular mass of the complex of 441 kDa. They are consistent with calculated values of 3 mol Ni, 65 mol nonheme iron and 66 mol acid labile sulfur per mol enzyme complex.
CO dehydrogenase and hydrogenase form a tight complex
Further experiments were performed to elucidate if CO dehydrogenase and hydrogenase form a tight complex or whether the two enzymes coelute on the different chromatography columns by coincidence. The chromatographic properties of purified CO dehydrogenase I were compared with the properties of the proposed complex of CO dehydrogenase I and hydrogenase. CO dehydrogenase I eluted from the hydroxyapatite column with 0.03 m potassium phosphate while CO dehydrogenase I associated with the hydrogenase was strongly bound to this column and could only be eluted from this column with 1 m potassium phosphate. CO dehydrogenase I eluted from a Superdex 200 gel filtration column with an apparent molecular mass of 120 kDa while the proposed complex eluted with an apparent molecular mass of 450 kDa. These data together with kinetic data described below strongly suggest that CO dehydrogenase is tightly associated with the hydrogenase.
Catalytic properties of the enzyme complex
The purified CO-oxidizing:H2-evolving enzyme complex is composed of a hydrogenase and a CO dehydrogenase. The activities of the two enzymes were determined individually with methylviologen as artificial electron donors or acceptors. The purified enzyme complex catalyzed the formation of H2 with reduced methylviologen as electron donor at a rate of 170 U per mg protein, the reduction of methylviologen by H2 at a rate of 180 U per mg protein, and the reduction of methylviologen by CO at a rate of 5900 U per mg protein. To determine whether hydrogenase and CO dehydrogenase are electrically connected, the enzyme preparation was tested for its ability to catalyze the conversion of CO to CO2 and H2. With 100% CO in the headspace, the enzyme complex catalyzed this reaction at a rate of 120 U per mg protein. This value increased to 450 U per mg protein when the assay was carried out with 5% CO in the headspace and the enzyme complex was activated with Ti(III)citrate (see below). A linear increase of activity with increasing protein concentrations was obtained, even at protein concentrations as low as 45 µg per mL assay mixture, corresponding to concentrations of the complex of 0.1 µm. This behavior suggests that all components involved in this reaction form a tight complex with an active site for CO oxidation, an active site for H2 formation, and the required electron transfer components. The presence of two distinct active sites was also supported by inhibition studies; CO dehydrogenase activity was specifically blocked by cyanide, and hydrogenase activity was blocked by CO or acetylene.
The rate of H2 formation with CO as electron donor and the rate with reduced methylviologen as electron donor showed the same pH dependence, with an optimum at pH 6.5.
Redox-dependent activation of the CO-oxidizing:H2-evolving enzyme complex
CO dehydrogenase from R. rubrum has recently been shown to be mostly inactive at redox potentials higher than −300 mV. The enzyme can be converted to an active form by the addition of strong reductants or by incubation with CO. This activation process is dependent on the concentration of CO and the incubation time . The CO-oxidizing:H2-evolving enzyme activity of the C. hydrogenoformans enzyme complex was also found to increase with increasing CO concentrations. Maximal activities were obtained with approximately 40% CO in the headspace of the assay mixture, corresponding to 250 µm CO in solution at 70 °C. Since the apparent Km value of the purified C. hydrogenoformans CO dehydrogenase I is 18 µm, a maximal H2-formation rate from CO was therefore expected at very low CO concentrations (< 100 µm). At higher CO concentrations, the inhibition of the hydrogenase in the enzyme complex by CO was expected to predominate (see below). The finding that maximal activity could only be observed at rather high CO concentrations indicates that CO dehydrogenase of the enzyme complex is activated in the presence of CO. To activate the enzyme independently of CO, the strong reductant Ti(III)citrate, which has been successfully used to activate other redox enzymes that require low redox potentials , was used. Incubation of the enzyme complex in the presence of 1 mm Ti(III)citrate for 1 min at 65 °C resulted in an increase of activity that was maximally pronounced at low CO concentrations. With 5% CO in the headspace, corresponding to 31 µm CO in solution, the initial activity of the CO-oxidizing:H2-evolving enzyme complex increased about 45-fold after activation with Ti(III)citrate. In general, the extent of activation was dependent on how long the enzyme preparation had been stored before used. Freshly prepared enzyme showed a higher specific activity and could only be activated to a lower extent as compared to older preparations. Figure 3 shows the dependence of CO-oxidizing:H2-evolving enzyme activity on the CO concentration with enzyme not activated by Ti(III)citrate (Fig. 3A) and enzyme activated with Ti(III)citrate (Fig. 3B).
To elucidate if Ti(III)citrate activates CO dehydrogenase, hydrogenase or both enzymes the influence of Ti(III)citrate on CO dehydrogenase- and hydrogenase activity was tested. In these assays methylene blue was used instead of methylviologen as electron acceptor. Reduced methylviologen is a strong reductant, which could also activate the enzyme(s). It was found that preincubation of the complex with Ti(III)citrate had no influence on the rate of methylene blue reduction by H2 which proceeded at a rate of 30 U per mg protein. However, the initial rate of methylene blue reduction by CO increased by a factor of 21 to a final activity of 1500 U per mg protein. These findings indicate that the CO dehydrogenase in the complex is activated by Ti(III)citrate.
Inhibition of the hydrogenase present in the enzyme complex by CO and acetylene
The rate of hydrogen production with reduced methylviologen as electron donor was determined at different CO concentrations. To follow this reaction, it was necessary to inhibit the CO dehydrogenase present in the enzyme complex. This was achieved with 3 mm potassium cyanide, which is a potent inhibitor of CO dehydrogenase, but does not inhibit the hydrogenase . The inhibition curve shows 50% inhibition with 45% CO in the headspace, corresponding to ≈ 300 µm CO in solution at 70 °C (Fig. 4). A similar value (50% inhibition at 300 µm CO in solution) has been determined for CO-induced hydrogenase from R. rubrum. Hence, both hydrogenases are quite insensitive to inhibition by CO compared to most other [NiFe] hydrogenases: typically, a 50% inhibition of [NiFe] hydrogenases at CO concentrations of ≈ 40 µm in solution has been observed .
H2 formation from reduced methylviologen or from CO was inhibited by acetylene when added to the gas phase. A 50% inhibition was observed with 30% acetylene in the gas phase. The reduction of methylviologen by CO was not inhibited by acetylene.
C. hydrogenoformans catalyzes the conversion of CO to CO2 and H2 to gain energy for growth. We have obtained an enzyme preparation composed of eight polypeptides catalyzing this reaction at high specific rates. The enzyme preparation was found to be composed of a CO dehydrogenase and a hydrogenase. Amino-terminal sequence analysis allowed the assignment of four of the polypeptides present in the enzyme preparation as the hydrophilic subunits of the hydrogenase. Two hydrophobic proteins were identified which most likely constitute the membrane spanning part of the hydrogenase. The genes encoding these six polypeptides could be identified in the genome of C. hydrogenoformans. The deduced proteins show high sequence similarity to proteins encoded by the cooMKLXUH gene cluster of R. rubrum and to the six subunits of Ech hydrogenase from M. barkeri[16,18]. Purified Ech hydrogenase is composed of six subunits: four hydrophilic proteins and two integral membrane proteins.
The two additional polypeptides present in the enzyme preparation from C. hydrogenoformans could be identified as subunits CooSI and CooFI of the CO dehydrogenase. The catalytic subunit CooSI of CO dehydrogenase has been previously purified from the soluble fraction of C. hydrogenoformans as a homodimer. In this enzyme preparation the CooFI protein was lacking . CooFI can be regarded as a four [4Fe−4S] cluster containing polyferredoxin, which is proposed to mediate the electron transfer between the catalytic subunit CooS and the hydrogenase. A homologue of CooF, the HycB protein, is also encoded by the hyc operon of E. coli, which encodes E. coli hydrogenase 3. HycB is thought to mediate the electron transfer between formate dehydrogenase and hydrogenase 3 in E. coli.
The data obtained for C. hydrogenoformans strongly suggest that the catalytic subunit of CO dehydrogenase (CooSI), the electron transfer protein CooF and the hydrogenase form a tight complex with a fixed stoichiometry. First, a coelution of these proteins on three different chromatography columns was observed. In particular a very strong binding of these proteins to hydroxyapatite was found. Elution of the complex from this column was only possible at a potassium phosphate concentration of 1 m. In contrast, the purified CooSI dimer did not bind to hydroxyapatite at a concentration of 30 mm potassium phosphate. Second, purified CooSI dimer eluted from a calibrated gel filtration column with a molecular mass of 120 kDa while the complex eluted with an apparent molecular mass of 450 kDa under the same conditions. Third, the rate of H2 formation from CO increased linearly with protein concentration. For a multicomponent system, containing CO dehydrogenase, the electron transfer protein (CooF) and the hydrogenase as individual components, a nonlinear protein dependence would have been expected. Fourth, in different enzyme preparations a constant subunit stoichiometry was always observed, the molar concentration of CooS and CooF being ≈ twofold higher than the concentration of CooH, the catalytic subunit of the hydrogenase. A hydrogenase preparation containing a lower or zero content of CO dehydrogenase was not obtained using the purification procedure described. In contrast, CO dehydrogenase devoid of hydrogenase can be easily obtained. Svetlitchnyi et al.  have recently purified the CooSI homodimer from the soluble fraction of C. hydrogenoformans. Also the work presented here indicates that only a part of CooSI present in the cell is tightly associated with the hydrogenase. A possible explanation could be that the synthesis of CO dehydrogenase and hydrogenase are not completely coregulated resulting in an excess of CooSI in the cell. Svetlitchnyi et al.  have recently purified the CooS dimer of a second CO dehydrogenase (CO dehydrogenase II) from C. hydrogenoformans. Electron microscopy studies had shown that in vivo this enzyme is associated with the cytoplasmic membrane. Purified CO-oxidizing:H2-evolving enzyme complex was found to contain low amounts of CooSII. CooSII was estimated to account for approximately 10% of the total CooS content. Meyer and coworkers  have proposed that CooSI interacts with a membrane-bound hydrogenase and thus has a function in energy conservation. This proposal is a based on the in vitro reconstitution of a CO-oxidizing:H2 forming system composed of cytoplasmic membranes, CooSI and a greenish-brown-coloured protein fraction called factor B. Addition of CooSI resulted in a fourfold increase of activity while addition of CooSII had no influence on the activity. The isolation of an enzyme complex, which in addition to a membrane-bound hydrogenase and the polyferredoxin CooFI mainly contains CooSI, supports the data by Meyer and coworkers. It is, however, not yet clear why small amounts of CooSII are present in the enzyme complex. The function of CO dehydrogenase II, which has thus far only been purified as CooSII dimer, is not yet known. Meyer and coworkers  have suggested that the enzyme provides the cell with reducing equivalents for anabolic reactions. It is, however, also possible that CooSI and CooSII are isoenzymes that both are able to interact with the hydrogenase via their specific CooF proteins. Directly upstream of the cooSII gene, a gene encoding a second CooF protein, CooFII, is located. The CooFII protein, which has a calculated molecular mass of 20 kDa, has not been detected in the enzyme complex from C. hydrogenoformans.
Thus far Ech hydrogenase from Methanosarcina barkeri[18,24] and the enzyme complex described in this work are the only members of a growing family of energy converting [NiFe] hydrogenases which can be purified and thus are accessible to further biochemical studies.
Preliminary sequence data was obtained from The Institute for Genomic Research website at http://www.tigr.org. Sequencing of C. hydrogenoformans is accomplished with support from the United States Department of Energy. This work was supported by the Max-Planck-Gesellschaft, by the Deutsche Forschungsgemeinschaft, and by the Fonds der Chemischen Industrie. We thank Karen Brune for editing the manuscript.