INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Pyrococcus furiosus is an archaeon that grows optimally near 100°C by the fermentation of carbohydrates and peptides to produce acetate, CO₂, and H₂ (9). If elemental sulfur (S⁰) is present in the medium, it is reduced to H₂S. It has previously been shown that the organism contains two enzymes that are able to catalyze the reduction of S⁰ to H₂S, hydrogenase I (H-I) and sulfide dehydrogenase. The hydrogenase, also known as sulfhydrogenase because of its S⁰-reducing activity, is a Ni-containing, iron-sulfur flavoprotein that also serves to reduce H⁺ to H₂ (16). Sulfide dehydrogenase is an iron-sulfur flavoprotein which also functions as a ferredoxin:NADP oxidoreductase (13). The two enzymes are thought to transfer electrons from reduced ferredoxin and NADPH, generated from the fermentation pathways, to the terminal electron acceptors, H⁺ and S⁰.

It was assumed that S⁰ reduction by P. furiosus was merely a means of removing the H₂ produced during fermentation, as H₂ inhibits growth (9). However, the presence of S⁰ in the growth medium both stimulates the growth rate and increases the cell yield by about 50%, suggesting that H₂S production is an energy conservation process (19). Yet, H-I and sulfide dehydrogenase are cytoplasmic enzymes and a conventional respiratory mechanism with S⁰ as a terminal electron acceptor is not present in this organism. It is thought that S⁰ reduction may increase the ratio of oxidized to reduced electron carriers, thereby favoring the oxidative rather than the nonoxidative decarboxylation of the 2-keto acids produced by ferredoxin-linked oxidoreductases during fermentation (14). How this process is regulated, however, is not known.

In this paper, we show that the energy metabolism of S⁰ by P. furiosus may be even more complicated than was originally thought. A second enzyme, termed hydrogenase II (H-II), has been identified and purified. It is demonstrated that both it and H-I catalyze the reduction of S⁰ to H₂S using NADPH as the electron donor. In fact, both enzymes preferentially reduce S⁰ rather than protons. The biochemical and molecular characterization of this new H₂-evolving, S⁰-reducing enzyme is presented.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Growth of P. furiosus. P. furiosus (DSM 3638) was routinely grown at 90°C in a 600-liter fermentor with maltose as the carbon source as described previously (6).

Enzyme assays. The activity of H-II was routinely measured during purification by the H₂-dependent reduction of benzyl viologen at 80°C. The assay mixture (2 ml) contained 100 mM EPPS [N-(2-hydroxyethyl)piperazine-N'-3-propanesulfonic acid] (pH 8.0) and 1 mM benzyl viologen with H₂ in the gas phase. The reaction was monitored at 580 nm, and a molar absorptivity of 7,800 M¹ cm¹ was used for reduced benzyl viologen. The S⁰ reduction activity of hydrogenase was determined at 80°C by measuring H₂S production by methylene blue formation (16) or by NADPH oxidation at 365 nm using a molar absorbance of 3,200 M¹ cm¹. Reaction mixtures contained either S⁰ (5% [wt/vol]) under H₂ (1 atm), S⁰ (5% [wt/vol]) and NADPH (0.4 mM) under Ar, or polysulfide (up to 3 mM) and NADPH (0.4 or 1 mM) under Ar. Rates of sulfide production measured by methylene blue formation were calculated over a time of 15 min. The H₂ oxidation and H₂ evolution activities of the enzyme were also measured at 80°C using methyl viologen, NAD(H), NADP(H), and P. furiosus ferredoxin as the electron carriers as previously described (16, 17). When ferredoxin was reduced using pyruvate as the electron donor via pyruvate ferredoxin oxidoreductase (POR) from P. furiosus, the 2-ml assay mixture contained 100 mM EPPS buffer (pH 8.0), pyruvate (10 mM), coenzyme A (4 mM), thiamine PP, (0.05 mM), ferredoxin (44 µM), POR (324 µg), and H-II (250 µg) or H-I (234 µg). Where indicated, the reaction mixtures also contained ferredoxin:NADP oxidoreductase (FNOR) from P. furiosus (100 µg) and NADP (1.6 mM). For the H₂-dependent reduction of ferredoxin, the 2-ml assay mixture contained, under an atmosphere of H₂, 75 mM EPPS buffer (pH 8.0), ferredoxin (0.11 mM), and H-I (234 µg) or H-II (250 µg). H-I (16), POR (3), FNOR (13), and ferredoxin (1) were purified from P. furiosus as described in the references. All activities for hydrogenase are expressed in units per minute, where 1 U catalyzes the oxidation of production of 1 µmol of H₂ per min.

Enzyme purification. P. furiosus H-II was purified at 23°C under anaerobic conditions. Frozen cells (100 g [wet weight]) were thawed in 400 ml of buffer A (50 mM Tris-HCl [pH 7.8] containing 2 mM dithiothreitol and 2 mM sodium dithionite) containing DNase I (10 µg/ml) and 0.2 mg of lysozyme/ml. The cells were lysed by incubation at 37°C for 2 h. A cell extract was obtained by centrifugation at 50,000 × g for 2 h. The supernatant (340 ml) was loaded onto a column (5 by 12 cm) of DEAE-Sepharose Fast Flow equilibrated with buffer A. The column was eluted with a linear gradient (1.2 liter) of 0 to 0.6 M NaCl in buffer A, and 50-ml fractions were collected. H-II started to elute as 0.4 M NaCl was applied to the column. The active fractions were combined (250 ml) and loaded onto a column (5 by 10 cm) of hydroxyapatite (Bio-Rad) equilibrated with buffer A. The flow rate was 4 ml/min, and 50-ml fractions were collected. The column was eluted with a 1.0-liter linear gradient (0 to 0.5 M potassium phosphate) in buffer A. The H-II activity started to elute as 0.25 M potassium phosphate was applied to the column. Fractions containing H-II activity were combined (230 ml), and 2 M (NH₄)₂SO₄ was added to give a final concentration of 0.8 M. This was loaded onto a column of phenyl-Sepharose (3.5 by 10 cm) equilibrated with buffer A containing 0.8 M (NH₄)₂SO₄. The column was eluted with a 320-ml linear gradient of 0.8 to 0 M (NH₄)₂SO₄. The flow rate was 4 ml/min, and 30-ml fractions were collected. H-II activity started to elute from the column as 0.01 M (NH₄)₂SO₄ was applied. Fractions containing H-II activity were combined and concentrated by ultrafiltration (Amicon ultrafilter; PM-30 membrane). The concentrated fractions (10 ml) were applied to a column of Superdex 200 (6 by 60 cm) equilibrated with buffer A containing 100 mM NaCl. The flow rate was 6 ml/min, and 30-ml fractions were collected. Those fractions containing pure H-II as judged by electrophoretic analysis were combined (120 ml), concentrated by ultrafiltration to 2 ml, and stored in liquid N₂.

Other methods. Protein concentrations were routinely estimated by the method of Bradford (4) with bovine serum albumin as the standard. The iron (12) and acid-labile sulfide (7) contents were measured as described elsewhere. A complete metal analysis (32 elements including nickel and iron) was performed by plasma emission spectroscopy using a Jarrel Ash Plasma Comp 750 instrument at the Riverbend Research Laboratories, University of Georgia. The flavin adenine dinucleotide (FAD) content was estimated by the change of absorption at 450 nm upon addition of sodium dithionite using a molecular absorptivity of 11,500 M¹ cm¹ (2). Flavin was analyzed using thin-layer chromatography and UV-visual light spectroscopy (22). The molecular weight of H-II was estimated by gel filtration (15), and sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis was carried out by the method of Laemmli (11). The N-terminal amino acid sequences were determined by using an Applied Biosystems model 477 sequencer (8). Electron paramagnetic resonance (EPR) spectra were recorded on an IBM-Bruker ER 300D spectrometer interfaced to an ESP 3220 data system and equipped with an Oxford Instrument ITC-4 flow cryostat. H-II in 50 mM Tris-HCl (pH 7.8) was oxidized by thionine (0.5 mM) under anaerobic conditions in an Amicon ultrafilter (PM-30 membrane), and the dye was removed by repeated washings with buffer. The oxidized enzyme (3.6 mg/ml) was reduced by adding 5 mM sodium dithionite, and after about 30 s at 80°C, the sample was frozen immediately in liquid nitrogen for analysis.

Nucleotide sequence accession number. The DNA sequences of the genes encoding the four subunits (---) of H-II are available from GenBank under accession no. AF176650.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Purification of H-II. More than 90% of H₂-dependent S⁰-reducing activity of P. furiosus was in the supernatant after centrifugation (100,000 × g for 2 h) of a cell extract, indicating that the enzyme(s) responsible is located in the cytoplasmic fraction. During the routine purification of H-I (6, 16), the activity of which is measured by the H₂-dependent reduction of benzyl viologen, a second distinct peak of hydrogenase activity eluted from the first ion-exchange (DEAE-Sepharose) column. The first peak eluted near 0.26 M NaCl (pH 8.0), contained approximately 90% of the total activity, and represented H-I. The second peak, containing the remaining 10%, eluted as 0.40 M NaCl was applied to the column. The hydrogenase responsible for the latter activity was purified further and, as discussed below, was shown to also catalyze the H₂-dependent reduction of S⁰ to H₂S. This second hydrogenase of P. furiosus is therefore referred to as H-II (or sulfhydrogenase II). The results of a typical purification are shown in Table 1. Compared to the cell extract, H-II was purified 24-fold with a yield of 4%, but since H-I is responsible for most of the activity in the extract, the true yield of H-II is closer to 40%.

Physical properties. The purified enzyme gave rise to four protein bands after SDS-gel electrophoresis (Fig. 1) which corresponded to M_rs of 52,000, 39,000, 30,000, and 24,000 Da. H-I also contains four subunits with M_rs of 50,000, 43,000, 33,000, and 29,000 Da, which are of comparable size although clearly distinct from those of H-II (see below). The apparent M_r of the H-II holoenzyme was 320,000 ± 2,000, which suggests that the enzyme is a dimer of heterotetramers, ()₂. Plasma emission spectroscopy and chemical analyses showed that purified H-II contained (mole(s) per mole of ) Ni (0.9 ± 0.1), Fe (21 ± 1.6), and acid-labile sulfide (19 ± 2.1). Flavin was also present, and this was identified using thin-layer chromatography as FAD. Visible spectroscopy showed that the enzyme contained 0.83 ± 0.1 mol of FAD per mol of . As shown in Fig. 2, the dithionite-reduced H-II exhibited at 50K an axial EPR spectrum characteristic of a single reduced [2Fe-2S] center. The spectrum became more complex and was considerably broadened as the temperature was decreased, consistent with the presence of multiple reduced [4Fe-4S] centers. The EPR absorption at 4K represented 2.7 ± 0.3 spin/mol of heterotetramer [], indicating that the enzyme contained at least two 4Fe and one 2Fe cluster. The oxidized enzyme showed an isotropic-type EPR signal near g = 2.03 which is typical of an oxidized [3Fe-4S] cluster (data not shown). This corresponded to 0.2 ± 0.03 spins per heterotetramer [] and presumably arose from oxidative damage of a [4Fe-4S] center. EPR resonances that might originate from Ni were not apparent from the reduced or oxidized enzyme.

View larger version (74K): [in this window] [in a new window]   FIG. 1.   SDS-12.5% polyacrylamide gel electrophoresis of H-II purified from P. furiosus. Lanes 1 and 4, molecular mass markers; lanes 2 and 3, 1 and 2 µg of H-II, respectively.

View larger version (22K): [in this window] [in a new window]   FIG. 2.   EPR spectra of reduced H-II. The spectra were recorded using the dithionite-reduced enzyme (3.6 mg/ml in 50 mM Tris-HCl [pH 7.8]) at the indicated temperature. The spectrometer settings were as follows: microwave power, 20 mW; microwave frequency, 9.596 GHz; modulation frequency, 100 kHz; modulation amptitude, 6.366 G; time constant, 81.92 ms; gain, 104.

Catalytic properties. The activities of H-II under various assay conditions are shown in Table 2. Values for H-I from P. furiosus are given for comparison. H-II was about five times less active than H-I in the standard H₂-dependent, benzyl viologen reduction assay. Hence, from the total activity in the cytoplasmic extract (Table 1), the cellular concentrations of the two enzymes are roughly comparable. Similarly, while H-II catalyzed the reduction of S⁰ to H₂S using H₂ as electron donor, the specific activity was about 30-fold lower than that of H-I. H-II was also about an order of magnitude less active than H-I in catalyzing both H₂ oxidation and H₂ evolution using methyl viologen as the electron carrier (Table 2).

View larger version (17K): [in this window] [in a new window]   FIG. 3.   H2 production from pyruvate by H-II (A) and H-I (B). H2 production was measured using ferredoxin (Fd) reduced by POR as the electron donor to H-I and H-II as described in Materials and Methods. Abbreviations: + Fd, with ferredoxin;  Fd, without ferredoxin;  H-I, without H-I;  H-II, without H-II.

View larger version (16K): [in this window] [in a new window]   FIG. 4.   Reduction of P. furiosus ferredoxin by H-I and H-II. The reactions were carried out as described in Materials and Methods.

View larger version (17K): [in this window] [in a new window]   FIG. 5.   H2 production from pyruvate using a coupled system of P. furiosus enzymes. The reactions were carried out as described in Materials and Methods and the legend to Fig. 3. (A) H-II; (B) H-I. Where indicated, the reaction mixtures also contained P. furiosus FNOR and NADP.

View larger version (17K): [in this window] [in a new window]   FIG. 6.   Stimulation of NADPH oxidation and inhibition of H2 production from NADPH by polysulfide. (A) H-II. (B) H-I. The 2-ml reaction mixture contained 100 mM EPPS (pH 8.0), 0.4 mM NADPH, polysulfide as indicated, and 250 µg of H-II (A) or 78 µg of H-I (B). The rate of NADPH oxidation was measured spectrophotometrically over 2 min. The amount of H2 produced (expressed as a percentage) was determined after 10 min.

Sequence analyses. The N-terminal sequences of four subunits of H-II were determined after they were separated from each other by SDS-polyacrylamide gel electrophoresis. They are as follows (where X is an unknown residue): , MIXELDEFTRVEGNXKAEIV-; , MRYVKLHSEYFPEFFNRLKE-; , MNPYRSYDARII-; and , MKLGVFELTDXGG-. They do not correspond to the N-terminal sequences of the four subunits of H-I, the complete sequences of which are known (18). The sequences of the H-II subunits were used to search the genome sequence of P. furiosus (http://combdna.umbi.umd.edu.genemate.html). Each one corresponded exactly to the translated N-terminal sequences of four distinct open reading frames, and as shown in Fig. 7, these appear to constitute a single operon. The calculated molecular masses of the encoded proteins (; 39,178 Da; , 32,943 Da; , 26,282 Da; and , 46,177 Da) are in agreement with the subunit sizes of H-II determined by SDS-polyacrylamide gel electrophoresis (Fig. 1) and, as shown in Table 4, are comparable to those reported for H-I.

View larger version (68K): [in this window] [in a new window]   FIG. 7.   DNA and amino acid sequences of the genes proposed to encode H-II from P. furiosus. The amino acid sequence of each open reading frame is given in one-letter code above the DNA sequence. Amino acids given in boldface represent putative motifs that bind the Ni-Fe binding site in the  subunit (400 to 408), the [4Fe-4S] clusters in the  subunit (229 to 240 and 306 to 317) and the  subunit (11 to 133, 160 to 178, and 188 to 202), and the [2Fe-2S] cluster in the  subunit (250 to 274). The motifs are extended to include the identity in sequence between H-II and H-I (20). The dashed lines above the amino acid sequence of the  subunit indicate binding regions for two nucleotide cofactors. Nucleotide sequences in boldface indicate possible RNA polymerase binding sites; underlined sequences indicate putative ribosome binding sites. =>, start of open reading frame; ===, transcription termination signal.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The molecular properties of H-II of P. furiosus are remarkably similar to those of the enzyme of this type previously characterized from this organism, H-I (6, 16) (Table 5). Although the size of the holoenzyme of H-II is twice that of H-I, both consist of four subunits and contain Ni, multiple Fe/S centers, and FAD (6, 20). The genes encoding H-I had been previously cloned and sequenced (18). The four subunits of the two enzymes are not only of comparable size but also show significant sequence similarity (55 to 63%). In fact, the gene order () for H-II is the same as that for H-I, and in both cases, the four genes are tightly linked. For H-II, those encoding  and , and  and , have 4 and 1 overlapping nucleotide, respectively, while  and  are separated by only 2 nucleotides (Fig. 7). Each translational start codon is preceded by a typical archaeal ribosome binding site (5), while the start codon of the gene encoding the  subunit is TTG, which is rarely used in members of the domain Archaea. In addition, there is a transcription termination signal (T-rich region) beginning 9 nucleotides following the stop codon of the last gene which encodes the  subunit. As shown in Fig. 7, the gene encoding the  subunit is preceded by two possible RNA polymerase binding sites, suggesting that a regulatory mechanism may be involved in H-II expression. What such effectors might be remains unknown.

The EPR properties of reduced H-II indicate that it contains at least two 4Fe and one 2Fe center (per heterotetramer). These would account for only about half of the Fe present in the enzyme, suggesting that additional FeS centers are present but that they are not EPR active under the experimental conditions, presumably due to incomplete reduction. This is substantiated by the presence of numerous cysteines and of several cysteine-containing motifs in the sequences of the subunits. Specifically, the results of our analysis of the genes of H-II are in agreement with those recently proposed by Silva et al. (20) for H-I, and these are summarized in Table 5. Thus, the  and  subunits of H-II correspond to the so-called large and small subunits of conventional NiFe-hydrogenases, the prototypical example of which is the crystallographically characterized enzyme from Desulfovibrio gigas (21). In H-II, by analogy, the  subunit contains the binuclear NiFe catalytic site while the  subunit contains three [4Fe-4S] clusters with all cysteinyl ligation. In the D. gigas enzyme, one of the 4Fe clusters is of the [3Fe-4S]-type while another (the distal cluster) has a His ligand. This is not the case in H-II or in H-I. In fact, as indicated in Fig. 7, the cluster-binding motifs show high sequence identity between these two enzymes for all four subunits. These also include those for two [4Fe-4S] clusters in the  subunit and one [2Fe-2S] cluster in the  subunit.

The  and  subunits of H-II also contain a significant number (10 and 4, respectively) of cysteinyl residues, most of which are conserved in H-I (7 and 4, respectively). It seems likely that these coordinate additional FeS clusters, since the proposed cluster content (Table 5) accounts for 22 Fe/heterotetramer, which is in good agreement with what is found by chemical analysis of purified H-II. Whether these additional residues are involved in S⁰ reduction remains to be seen. The  subunit of H-II also contains two putative nucleotide-binding sites analogous to those previously identified in H-I (18). These are consistent with the presence of FAD in H-II and with its ability to use NAD(P)(H) as electron carriers. The kinetic data presented herein demonstrate that ferredoxin cannot be considered as a physiological electron carrier for either H-II or H-I.

Hence, P. furiosus contains two hydrogenase-type enzymes that are extremely similar in their molecular properties, including their gene organization. Both are capable of reducing S⁰ and polysulfide, as well as protons, using electrons supplied by NADPH. Indeed, both have a preference for sulfide production (from polysulfide) rather than H₂ production. In fact, polysulfide both stimulates NADPH oxidation and leads to a decrease in H₂ production. Whether the latter reaction is specifically inhibited by polysulfide, or whether the lower rate results from the diversion of electron flow to polysulfide rather than protons, remains to be elucidated. In any event, it is clear that both hydrogenases have the potential to catalyze sulfide production in vivo.

The two hydrogenases of P. furiosus appear to differ primarily in their relative catalytic activities, with H-II being approximately an order of magnitude less active than H-I in most of the assays investigated (Table 3). The two enzymes are present in the cytoplasm at roughly similar concentrations, and it seems reasonable to conclude that they serve different functions. H-II has a higher affinity for both S⁰ and polysulfide in the standard assays, and perhaps this enzyme becomes physiologically relevant at low S⁰ concentrations. In the absence of S⁰, the role of H-I is proposed to be H₂ production using NADPH as the electron donor (16, 17), a reaction in which it is about 60 times more efficient than H-II (Table 3). The only reaction catalyzed by H-II that is very poorly carried out by H-I is the H₂-dependent reduction of NAD (Table 3). However, the apparent K_m value for NAD (125 µM) seems high for this to be physiologically relevant, and H-II appears to have a much higher affinity for NADP (K_m, 17 µM). Hence, in spite of the fact that H-II contains FAD and has a putative nucleotide-binding site, it is hard to imagine that NAD(P) reduction has any physiological significance. Clearly, insight into the function of this enzyme will require additional analyses such as proteomics and mRNA studies, and these are in progress.