Characterization of Genes Responsible for the CO-Linked Hydrogen Production Pathway in Rubrivivax gelatinosus

Abstract

Upon exposure to carbon monoxide, the purple nonsulfur photosynthetic bacterium Rubrivivax gelatinosus produces hydrogen concomitantly with the oxidation of CO according to the equation CO + H₂O ↔ CO₂ + H₂. Yet little is known about the genetic elements encoding this reaction in this organism. In the present study, we use transposon mutagenesis and functional complementation to uncover three clustered genes, cooL, cooX, and cooH, in Rubrivivax gelatinosus putatively encoding part of a membrane-bound, multisubunit NiFe-hydrogenase. We present the complete amino acid sequences for the large catalytic subunit and its electron-relaying small subunit, encoded by cooH and cooL, respectively. Sequence alignment reveals a conserved region in the large subunit coordinating a binuclear [NiFe] center and a conserved region in the small subunit coordinating a [4Fe-4S] cluster. Protein purification experiments show that a protein fraction of 58 kDa molecular mass could function in H₂ evolution mediated by reduced methyl viologen. Western blotting experiments show that the two hydrogenase subunits are detectable and accumulate only when cells are exposed to CO. The cooX gene encodes a putative Fe-S protein mediating electron transfer to the hydrogenase small subunit. We conclude that these three Rubrivivax proteins encompass part of a membrane-bound, multisubunit NiFe-hydrogenase belonging to the energy-converting hydrogenase (Ech) type, which has been found among diverse microbes with a common feature in coupling H₂ production with proton pumping for energy generation.

Hydrogen is a clean fuel, and if produced from renewable resources, it has the potential to address both energy security and environmental concerns (14). One such renewable feedstock is waste biomass, which can be gasified to generate CO (along with H₂) as an energy-enriched substrate. Numerous anaerobic microbes have been reported to carry out a CO oxidation reaction leading to H₂ production according to the equation CO + H₂O ↔ CO₂ + H₂ (5, 6, 17, 26, 37, 38, 39). More in-depth understanding of the underlying metabolic pathways along with the genes and enzymes involved thereby has immense potential in developing a feasible approach for renewable H₂ production from biomass-derived CO. Carbon monoxide oxidation is catalyzed initially by the enzyme CO dehydrogenase (CODH), and the resulting reducing equivalents are shuttled via a series of iron-sulfur proteins to a terminal hydrogenase, yielding H₂ as the end product. The CO oxidation reaction also yields energy, as evidenced by the CO-supported ATP generation and cell growth in the purple nonsulfur photosynthetic bacteria Rubrivivax gelatinosus (Rx. gelatinosus) and Rhodospirillum rubrum and in Methanosarcina barkeri, with the former two microbes capable of using CO as the sole carbon substrate in darkness (7, 8, 20, 25, 38). Biological CO oxidation thereby plays a significant role not only in the generation of H₂ from CO (and H₂O) but also in yielding ATP, although the mechanism of the latter is unknown thus far.

Genes and enzymes involved in the overall CO oxidation-H₂ production pathway have been elucidated in R. rubrum. There are two CO-inducible transcripts, with the cooFSCTJ operon encoding CODH and related proteins (19, 21) and the cooMKLXUH operon encoding a NiFe-hydrogenase and related proteins (11, 12). Both transcripts are under the exclusive control of CooA, a heme-containing CO-sensing transcriptional factor (15, 29, 33). Sequence comparison indicates that the hexameric CooMKLXUH hydrogenase in R. rubrum belongs to a group of multisubunit, membrane-bound NiFe-hydrogenases with homologs identified in several organisms, including Escherichia coli (4, 32), M. barkeri (23), Carboxydothermus hydrogenoformans (35), Thermoanaerobacter tengcongensis (36), and Desulfovibrio gigas (30). A feature common to this type of hexameric hydrogenase is that it is composed of two integral membrane proteins and four hydrophilic subunits; the latter display sequence similarity to the energy-conserving NADH:quinone oxidoreductase (complex I) (2, 13). For these hydrogenases, their characteristic sequence and energy conservation features coupling the oxidation of a carbonyl group (originating from CO, formate, or acetate) to proton reduction thus classify them as the group 4 energy-converting (or energy-conserving) hydrogenase (Ech) (40). Of the four peripheral subunits, one is putatively the small subunit (CooL in R. rubrum, HycG in E. coli hydrogenase-3, and EchC in the other aforementioned strains) and another is the large NiFe-containing catalytic subunit (CooH in R. rubrum, HycE in E. coli hydrogenase-3, and EchE in the other aforementioned strains), based on the crystal structure of the “model” heterodimeric NiFe-hydrogenase from D. gigas (41). The Ech hydrogenases from M. barkeri, T. tengcongensis, and D. gigas have each been purified as a hexameric complex. Yet biochemical characterization is thus far limited to M. barkeri, where the EchC small subunit harbors the proximal [4Fe-4S] cluster mediating electron and plausible proton transfer to the catalytic subunit (EchE) (10, 24). Very limited success has been reported with respect to the purification and characterization of the hydrogenase complex in R. rubrum (11).

The purple nonsulfur photosynthetic bacterium Rubrivivax gelatinosus CBS contains a CO-inducible, CO oxidation-H₂ production system similar to those in R. rubrum and C. hydrogenoformans. The partially purified hydrogenase is sensitive to CO, with a K_i near 3.9 μM, provided that CODH is removed during hydrogenase purification (26). Biochemical characterization of this partially purified hydrogenase also reveals tolerance to O₂ and linkage to the photoreduced ferredoxin for in vitro H₂ production (26). It displayed a half-life of 6 h when partially purified hydrogenase was stirred in ambient air (21% O₂). This level of oxygen tolerance has not been reported for other Ech-type hydrogenases. Despite the fact that genes encoding the CO oxidation pathway have been elucidated in R. rubrum, nothing is known as to the genetic elements encoding the CO-inducible hydrogenase in Rx. gelatinosus, which therefore warrants further study. In this report, we document the isolation of a transposon-mutagenized Rx. gelatinosus cell line deficient in the CO-inducible hydrogenase activity and the subsequent cloning of several genes involved in the CO-linked H₂ production pathway. Most of these genes show high degrees of sequence similarity as well as arrangement with those Ech genes from M. barkeri and the coo genes of C. hydrogenoformans and R. rubrum, with the latter serving as a model for this study.

MATERIALS AND METHODS

Organism, medium, and growth conditions.

Rubrivivax gelatinosus CBS was cultivated in RCVBN medium (26) supplemented with yeast extract (0.05%, wt/vol) and sodium malate (37 mM) (hence black-green [BG] medium), with or without CO (15%, vol/vol). The latter was accomplished by injecting 3 ml pure CO into a stoppered 25-ml Bellco tube containing 10 ml BG medium, resulting in a dissolved CO concentration of 108 nm/ml liquid. Medium preparation, growth conditions, and cell dry weight determinations were described previously (26). Illumination was provided by a band of incandescent lamps with a light intensity reaching the culture bottle of approximately 45 μmol m⁻² s⁻¹.

Transposon mutagenesis and mutant selection.

Rx. gelatinosus CBS was mutagenized using the EZ::TN (R6Kγori/Kan-2) Tnp transposome kit (Epicentre) by following the manufacturer's protocol. Cells were grown in 15-ml culture tubes under photoheterotrophic conditions in BG medium to an optical density at 660 nm (OD₆₆₀) of 0.5 and washed four times with sterile, ice-cold glycerol (10%, vol/vol). After each wash, the cells were suspended in one-half the starting volume. After the final wash, cells were suspended in 0.5 ml of medium and either electroporated immediately or frozen at −80°C. Approximately 50 μl of cells was mixed with 1 μl of the transposon prior to electroporation inside an 0.2-cm-gap cuvette using a Gene Pulser II (Bio-Rad, Hercules, CA) set at 2.0 kV, 200 Ω, and 25 μF. After electroporation, cells were immediately added to 15 ml of sterile BG medium and incubated in the dark for 6 h. Transformants were selected by plating on BG agar plates containing 15 μg/ml kanamycin. Transposon insertion was confirmed via Southern blot analysis targeting the kanamycin resistance gene (data not shown).

To assay CO-to-H₂ activity and to identify mutants, individual transformed colonies were inoculated into 10-ml anaerobic vials containing 5 ml BG medium supplemented with 15 μg/ml kanamycin. These cultures were grown photosynthetically for 16 h to an OD of ∼1.0, followed by induction of the CO oxidation pathway by addition of 1.0 ml of pure CO and shaking of the mixture at 250 rpm for 2 h. The culture vials were then placed in an autosampler of a gas chromatograph (Agilent 5890) for analysis of H₂ and CO contents, using a molecular sieve 5A column and a thermal conductivity detector. Data were analyzed with the HP3365 ChemStation software.

Rescue cloning of mutated genes.

DNA was prepared from the above-described mutants using the Qiagen genomic DNA extraction kit (Qiagen). Approximately 1 μg of DNA was digested to completion with SacI (New England Biolabs), self ligated, and transformed into E. coli cells expressing the pir gene. Transformants were selected on Luria-Bertani agar plates containing 50 μg/ml kanamycin. DNA was isolated from two colonies per mutant. DNA flanking the transposon was sequenced at Davis Sequencing (Davis, CA).

Creation, screening, and sequencing of an Rx. gelatinosus genomic library.

To isolate the DNA flanking the transposon insertion, a genomic library was constructed. Genomic DNA was partially digested with HaeIII (GG/CC) and separated on a 1% (wt/vol) agarose gel. Fragments ranging from 8 to 10 kb were gel purified and ligated into the blunt, dephosphorylated pSMART LC-Kan vector by following the manufacturer's protocol (Lucigen). Pooled transformants were screened by high-fidelity PCR for the 317-bp cooH fragment with primers 1762f (5′-CCCTCCACGACGAACGCCA-3′) and 1762r (5′-CCGATCAGCAGCAGAAAGCCG-3′). PCR products were analyzed on a 2.5% Tris-acetate-EDTA (TAE)-buffered agarose gel. DNA pools positive for PCR products were deconvoluted by repeating the PCR on smaller pools of plasmid clones and then on single colonies plated from the frozen stock cultures. Genomic library clones were sequenced at Davis Sequencing (Davis, CA).

In addition to sequencing of the aforementioned genomic clones, rescue clones from the GV1762 mutant (described in Results) were sequenced by the Joint Genome Institute through their Sequencing for Others project (http://spider.jgi-psf.org/programs/SFO/index.html). DNA was prepared from the GV1762 mutant using the Qiagen genomic DNA extraction kit (Qiagen). Approximately 1 μg of DNA was sheared by repeatedly passing it though a 25-gauge needle, self ligated, transformed, and selected for as noted above.

Sequence analysis.

Open reading frames (ORFs) were predicted with FGENESB (Softberry, Inc.) and SYCO (synonymous codon usage Gribskov statistic plot) trained on the Rx. gelatinosus codon usage table (GenBank) and manually using blastX and blastP (from conceptual translations) alignments.

Complementation of cooH.

To restore functionality in the GV1762 (CooH⁻ Kan^r) mutant, a 1.7-kb PCR product containing a majority part of the cooH ORF (codons at the 5′ end for the first three amino acids [MSA] were excluded) and flanking downstream sequence was amplified with primers cooH 1762 comp frd1 (5′-ATATAAGCTTGCACCACGACACGTCGAAGTC-3′; primer a in Fig. S1 at http://www.nrel.gov/basic_sciences/technology_staff.cfm/tech=16/ID=7 and cooH 1762 comp rev2 (5′-ATATGAATTCACCATGCCGAGCACGAACAG-3′; primer b in Fig. S1) using the KOD Hot Start DNA polymerase kit (EMD Biosciences) per the manufacturer's protocol, but with an annealing temperature of 60°C and a 1-min extension time. This fragment was cloned into the broad-host-range pBBR1MCS-4 vector (22) at the unique HindIII and EcoRI sites, sequenced, and electroporated into GV1762. The resultant transformants from single colonies were selected based on resistance to ampicillin (10 μg/ml) on successive platings. The CO-to-H₂ assay of the transformant was described above.

Enzyme assay.

Both the CO dehydrogenase assay (based on methyl viologen reduction by CO) and the hydrogenase assay (based on H₂ evolution from reduced methyl viologen) were described previously (26). The hydrogenase assay based on H₂ oxidation coupling to the reduction of methyl viologen, and the reverse direction from above, was described previously (27).

Membrane preparation, antibody generation, and Western blotting.

Rubrivivax gelatinosus chromatophore membranes were prepared as described previously (26) from photosynthetic cultures grown in the absence and presence of 15% (vol/vol) CO for 2 days. Antibodies against CooH were generated in rabbits (YenZym Antibodies LLC, San Francisco, CA) and affinity purified by YenZym using synthesized peptide. The peptide antigen was CooH#317-330 (CRTDGSDKPERLKWR). Antibody against CooL was generated in rabbit by a “genetic immunization” technique at Strategic Diagnostics Inc. (Newark, DE) and affinity purified by Strategic Diagnostics Inc. using E. coli expressed peptide. The peptide antigen for both genetic immunization and affinity purification was CooL#1-56 (MFDIFKKAAGRSPWLYRINAGACNGCDVELVTTALIPRYDIERLGCKYCGSPKHAD). For SDS-PAGE, total protein extracts from the same amount of cells (based on OD₆₆₀) were loaded into a 4 to 20% gradient gel with UV-activated protein imaging (NuSep, Lane Cove, Australia). Western blotting was performed using a semidry blotter (Bio-Rad, Hercules, CA), a 1:1,000 dilution of primary antibodies, a 1:1,000 dilution of goat anti-rabbit secondary antibody conjugated with alkaline phosphatase (Calbiochem, San Diego, CA), and an alkaline phosphatase substrate kit (also from Bio-Rad).

RESULTS

Mutant isolation and rescue cloning of the cooH gene encoding the hydrogenase catalytic subunit.

The genome of the Rx. gelatinosus PM1 strain (reclassified as Methylibium petroleiphilum, a nonpigmented methylotroph) has been sequenced, and yet this nonphotosynthetic strain does not contain the necessary genes to confer any CO oxidation activity linking to H₂ production (18). It is therefore necessary to clone these genes from Rx. gelatinosus CBS. In the initial strategy, primers and DNA probes were designed based on the conserved cooH gene sequence of R. rubrum. Our effort failed to yield any positive products via either PCR or Southern hybridization. An EZ::TN transposon mutagenesis approach (in Materials and Methods) was therefore undertaken. The transposon mutagenesis occurred with high frequency, with 1 in 100,000 cells becoming kanamycin resistant. After screening approximately 2,000 mutants, we identified a cell line unable to consume CO or to produce H₂ from CO. This cell line also lost its ability to grow photoautotrophically using CO as the sole carbon substrate. Enzymatic assays of partial reactions indicated that this mutant, designated GV1762, has no H₂ production activity, as assayed either with CO as the reductant for the linked CO-to-H₂ pathway or with chemically reduced methyl viologen as the electron source for hydrogenase determination (Table 1). Nevertheless, this mutant retained the wild-type (WT) level of its CODH activity, measured by the reduction of methyl viologen from CO. The genomic region encompassing the GV1762 mutation was rescue cloned, sequenced, and compared to known entries in GenBank. Mutant GV1762 carries an interrupted gene with a sequence similar to cooH, which encodes the CO-induced NiFe-hydrogenase catalytic subunit CooH in R. rubrum (11) and in C. hydrogenoformans (35). The wild-type cooH fragment was amplified and sequenced, based on the flanking sequence obtained in GV1762.

TABLE 1.

CO oxidation and H₂ production activities in whole cells of Rubrivivax gelatinosus strains

Strain	Activity (μmol min−1 mg [dry weight] cell−1)
	CO dehydrogenasea	Hydrogenaseb	CO to H2c
CBS parent	14.8	0.36	0.87
GV1762	14.5	0	0

^aCO dehydrogenase was assayed by measuring the reduction of methyl viologen from CO oxidation at 578 nm.

^bHydrogenase was determined by the evolution of H₂ from methyl viologen (5 mM), reduced by sodium dithionite (5 mM).

^cCO-to-H₂ rate was determined by measuring H₂ production from CO (15%, vol/vol) oxidation without exogenous electron mediator.

CooH.

Rx. gelatinosus CooH is predicted to be a soluble protein (using the program http://bp.nuap.nagoya-u.ac.jp/sosui/) with 368 amino acids (41 kDa) and an isoelectric point of 5.97. It shares 56% identity and 76% similarity with CooH from both R. rubrum and C. hydrogenoformans (Table 2). Except for the extra eight amino acids in the N terminus, it can be almost perfectly aligned with these two CooH sequences (Fig. 1A). It displays less similarity with its counterpart from M. barkeri (23), T. tengcongensis (36), D. gigas (30), and E. coli (4, 32) (Table 2). The Rx. gelatinosus CooH contains strictly conserved sequence motifs near its N terminus (RxC[S/G]xCx₃H) and at its C terminus (DPCx[A/S]Cxx[H/R]) (both marked with asterisks in Fig. 1A), which are known to jointly coordinate (via the four conserved cysteine residues) a binuclear [NiFe] active site in the large catalytic subunit of all NiFe-hydrogenases reported thus far (42). CooH therefore may be the catalytic subunit of a NiFe-hydrogenase in Rx. gelatinosus involved in the CO oxidation pathway. Moreover, the presence of two highly conserved motifs, L1 (RGxEx₁₄LxERxCx₂CSx₂H) and L2 (Dx₈DPCx₂CTER), on CooH (underlined in Fig. 1A) classifies this hydrogenase as the group 4 membrane-associated, H₂-evolving, Ech-type energy-converting hydrogenase (40). Rubrivivax CooH ends with an arginine residue at its C terminus, a feature common to all Ech hydrogenases reported thus far, yet without the extra stretch of 32 amino acids present only in HycE of E. coli (data not shown).

TABLE 2.

Identity and similarity percentages between the predicted Coo polypeptides and their homologs, generated by NCBI P-BLAST search

Gene and organism	Protein	% Identity	% Similarity
cooH
R. rubrum	CooH	56	76
C. hydrogenoformans	CooH	56	76
M. barkeri	EchE	35	58
T. tengcongensis	EchE	40	61
E. coli	HycE	37	58
D. gigas	EchE	41	60
cooL
R. rubrum	CooL	66	76
C. hydrogenoformans	CooL	69	83
M. barkeri	EchC	43	65
T. tengcongensis	EchC	42	59
E. coli	HycG	39	59
D. gigas	EchC	40	59
cooX
R. rubrum	CooX	42	61
C. hydrogenoformans	CooX	35	53
M. barkeri	EchF	31	48
T. tengcongensis	EchF	35	51
E. coli	HycF	34	54

FIG. 1.

Hydrogenase subunit-encoding genes in the CO-to-H₂ pathway in Rubrivivax gelatinosus CBS. (A) Alignment of Rubrivivax gelatinosus hydrogenase subunit CooH sequence with biochemically characterized Coo and Ech catalytic subunit sequences from other bacteria. Rg, Rubrivivax gelatinosus; Mb, Methanosarcina barkeri; Ec, Escherichia coli; Ru, Rhodospirillum rubrum; Tt, Thermoanaerobacter tengcongensis; Ch, Carboxydothermus hydrogenoformans. The conserved L1 and L2 signatures are underlined, and the NiFe cluster binding motifs are marked with asterisks. N and C termini of E. coli HycE are not shown (see text). (B) Organization of the hydrogenase subunit genes. The fragment is 2,609 bp in length and is putatively part of an operon. Open reading frames (ORFs) occur in the direction of the arrow. The ORF cooU (sequence data not shown) overlaps with cooX and cooH. (C) CooL alignment. The conserved cysteine residues coordinating a [4Fe-4S] cluster are underlined. Abbreviations are as defined for panel A. (D) CooX alignment. The 12 conserved cysteine residues are marked as ligands coordinating potentially up to three [4Fe-4S] clusters. The assignment of the first 8 cysteines is typical of a 2x [4Fe-4S] ferredoxin (9), with numerals 1 and 2 designating the two respective 4x Cys-binding motifs. A conserved acidic residue (see text) is underlined.

Genetic complementation of the hydrogenase mutant.

Mutant GV1762 lacking an evolving hydrogenase was complemented by transformation with a plasmid harboring a majority part of the cooH ORF and its flanking downstream sequence (see Fig. S1 at http://www.nrel.gov/basic_sciences/technology_staff.cfm/tech=16/ID=7). The broad-host-range plasmid pBBR1MCS-4 has not been used in Rx. gelatinosus CBS but is expected to replicate in it. The complemented cell line remained resistant to kanamycin and showed partially restored in vitro hydrogenase activity, roughly 16% of that of the wild type (WT) (236 and 1,520 nmol H₂ h⁻¹ ml⁻¹ OD₆₆₀⁻¹, respectively) when assayed using sodium dithionite as the electron donor mediated by methyl viologen. When supplemented with CO in anaerobic conditions, the complemented strain, like the WT, converted CO to H₂ quantitatively in the dark, albeit at an initial rate that is roughly also 16% of the WT rate (Fig. 2). The cooH mutant GV1762 cell line displayed no hydrogenase or CO-to-H₂ activity (Table 1).

FIG. 2.

Anaerobic CO-to-H₂ conversion in WT Rubrivivax gelatinosus and in the complemented strain GV1762c. Both strains were grown photosynthetically in CO (15%, vol/vol) to mid-log phase to fully induce the CO-to-H₂ activity. The two cultures were then sparged with argon gas, followed by injection of CO (15%, vol/vol) and incubation in the dark with shaking (50 rpm). Headspace CO and H₂ were measured at indicated time intervals using gas chromatography. Data are averages of three repeats. The mutant strain GV1762 showed no CO-to-H₂ conversion (data not shown).

Identification of other hydrogenase subunit genes.

The cooH sequence was used to design specific primers and to probe for additional genes that might be clustered with cooH. Sequencing of the flanking region immediately upstream of cooH reveals three additional coo genes displaying a high degree of sequence similarity (Table 2) and conserved gene arrangement (Fig. 1B) with their respective counterparts in R. rubrum. Consequently, they are designated cooL, cooX, and cooU (the features of which will not be discussed in this report). These genes were proposed to encode proteins with functions similar to the CO-inducible hydrogenase in R. rubrum (12).

CooL.

CooL is predicted to be a soluble protein of 143 amino acids (16 kDa) with an isoelectric point of 8.21. Its sequence shows a high level of similarity to CooL from C. hydrogenoformans (69% identity) (35) and R. rubrum (66% identity) (12) and lower similarity to its counterparts EchC in M. barkeri (23) and D. gigas (30), CooL in T. tengcongensis (36), and HycG in E. coli (32) (Table 2). It shares 19% identity with the small subunit of the “standard” heterodimeric uptake hydrogenase in D. gigas (41). Among the sequences, only HycG of E. coli contains the extra 25 amino acids in its N terminus (Fig. 1C) and an extra 88 amino acids in its C terminus (part of those are not shown). Sequence alignment also identifies four conserved cysteine residues (underlined in Fig. 1C) in Rx. gelatinosus, with the conserved sequence pattern CxxCx_nGxCxxxGx_mGCPP (n = 61 to 106 with n = 62 in Rx. gelatinosus CBS and m = 24 to 61 with m = 24 in Rx. gelatinosus CBS). This motif is reported to bind a [4Fe-4S] cluster in the small subunit of all NiFe-hydrogenases reported thus far (1). This Fe-S cluster also corresponds to the proximal Fe-S cluster of the hydrogenase small subunit based on the crystal structure of the “standard” uptake hydrogenase in D. gigas (41), with a proposed role in mediating electron flow directly from/to the [NiFe] active site located within the catalytic subunit.

CooX.

Immediately downstream of cooL is the cooX gene, putatively encoding a soluble protein of 176 amino acids (20 kDa) with an isoelectric point of 8.03. It shows higher levels of sequence similarity to CooX from R. rubrum (12) than to its counterparts from C. hydrogenoformans (35), E. coli (32), M. barkeri (23), and T. tengcongensis (36) (Table 2). Sequence alignment predicts the following two conserved cysteine-containing motifs: CxxCxxCx_nCP (n = 36 to 38) and Cx_mCxxCxxC (m = 20 to 24) (Fig. 1D) that are typical in ferredoxin containing two [4Fe-4S] clusters (9). The second four-cysteine motif contains a strictly conserved acidic residue, glutamate (underlined in Fig. 1D), displaying the pattern CxxCxxCx3CPx_8-10-E. The location of this acidic residue (glutamate or aspartate) is conserved among all EchF homologs and the corresponding subunit of complex I from various sources (2). Figure 1D also shows that the C-terminal regions of CooX in Rx. gelatinosus, C. hydrogenoformans, and R. rubrum all contain four additional cysteines with a CxxC motif; therefore, they could potentially coordinate an additional Fe-S cluster.

Hydrogenase molecular mass estimation.

Hydrogenase was solubilized from Rx. gelatinosus chromatophore membranes and partially purified via anion-exchange and gel filtration chromatography according to the method of Maness et al. (26). Figure S2 available at http://www.nrel.gov/basic_sciences/technology_staff.cfm/tech=16/ID=7 shows that the solubilized and partially purified Rx. gelatinosus hydrogenase has a molecular mass of approximately 58 kDa based on a gel chromatography elution profile (V_e/V₀ = 1.36). Previous work showed that this partially purified 58-kDa protein fraction is active in the forward H₂ production direction mediated by reduced methyl viologen (26). We demonstrated in this work that it is also active in the reverse direction, coupling H₂ oxidation to the reduction of methyl viologen. It displays a ratio of nearly 61 when activities in the forward direction are compared with those in the reverse direction using the same methyl viologen dye for mediation, suggesting a preference for the H₂ production reaction (versus H₂ oxidation).

Expression of the hydrogenase by CO treatment.

The appearance of the hydrogenase proteins upon CO addition was confirmed by Western blotting using antibodies against CooH and CooL subunits in whole-cell extracts prepared from wild-type, GV1762, and GV1762c strains. As shown in Fig. 3, CooH (41 kDa) and CooL (16 kDa) were clearly detectable in the wild type only after CO treatment. The accumulation of CooH was observed within 30 min of CO injection (data not shown) into a wild-type culture and became more pronounced over time. A similar expression profile was also observed for CooL. In addition, both CooH and CooL subunits were found in Rx. gelatinosus chromatophore membranes following CO treatment (data not shown). The Western blotting data on the wild-type strain are therefore in agreement with and complement the studies of R. rubrum showing that the hydrogenase genes are transcriptionally induced by CO (12), with a concomitant accumulation of the CooH protein (34).

FIG. 3.

Expression of Rubrivivax gelatinosus hydrogenase subunits upon CO addition. Wild-type, GV1762, and GV1762c cultures were treated with CO (15%, vol/vol) for up to 4 h. Cell optical density (at 660 nm) of the wild-type culture was 0.273 at time zero and 0.394 at 4 h. Cells from 0.5-ml cultures were spun down, and the pellet was resuspended in 80 μl SDS loading buffer. A 10-μl sample was loaded into each well of a 4 to 20% polyacrylamide gradient gel with NUView, a UV-activated protein-binding fluorescence dye (NuSep, Lane Cove, Australia) for electrophoresis. After Western transfer onto nitrocellulose membranes, the proteins were probed with antibodies produced in rabbits against various hydrogenase subunit peptides. A UV-activated fluorescence image of the gel was used as a loading control. Lane M, prestained molecular mass markers.

The Western blots confirmed the absence of CooH in GV1762 following a 4-h CO treatment and its restoration in GV1762c. The latter finding correlates well with the recovery of the hydrogenase activity and CO-to-H₂ production in this complemented line (Fig. 2). The CooH band in GV1762c appeared to be at 44 kDa, 3 kDa higher than the wild-type band, and has been repeatedly observed without CO treatment, albeit at low levels. The intensity of this protein band increased with CO treatment, possibly due to stabilization by binding to other subunits such as CooL that were expressed only following CO treatment. Surprisingly, CooL was observed in both GV1762 and GV1762c at levels comparable to that in the wild type. The accumulation of CooL therefore seems to be independent of the large subunit CooH, at least within the time frame of this test.

DISCUSSION

Rubrivivax gelatinosus catalyzes the conversion of CO (and H₂O) to CO₂ and H₂ in light and in darkness, with the latter mode yielding energy during CO oxidation in support of cell growth (25, 38). In this study, we report the sequence of three genes, cooL, cooX, and cooH, along with biochemical and immunological evidence corroborating their roles as part of a NiFe-hydrogenase involved in the H₂ production reaction coupled to CO oxidation, a first such report for Rx. gelatinosus. As in R. rubrum, the above three genes are arranged as cooLXUH in a putative operon (cooU data not shown). No hydrogenase or its related genes were found downstream of cooH. Upstream of cooLXUH, we identified two genes, cooM and cooK (data not shown; cooA was not found upstream of cooM), putatively encoding two additional hydrogenase subunits, similar to the cooMKLXUH gene cluster of R. rubrum (11, 12). Sequence similarity and conserved gene arrangement strongly suggest that the hydrogenase in Rx. gelatinosus is a multisubunit Ech type. One feature shared by all Ech hydrogenases is that their H₂ production reactions are linked to the oxidation of a carbonyl group. With the exception of E. coli, another common feature shared by Ech hydrogenases is that the C terminus of their catalytic subunit lacks the extra stretch of amino acids known to be required for Ni insertion based on the model developed for the three NiFe-hydrogenases in E. coli (3). Maturation of the catalytic subunit CooH or EchE therefore must proceed without C terminus processing, with its Ni insertion mechanism remaining a puzzle.

Alignment of Rx. gelatinosus CooH amino acid sequence with the corresponding putative homologs found in other organisms suggests that CooH is the large subunit of an Ech-type NiFe-hydrogenase (Fig. 1). Moreover, we presented several direct lines of evidence supporting the role of the cooH gene in encoding H₂ productivity in Rx. gelatinosus. The disruption of cooH via transposon insertion abolishes both in vivo and in vitro H₂ production (Table 1). Hydrogen production is partially restored (Fig. 2) after complementation of the GV1762 mutant with a replicating plasmid carrying the 5′-truncated version of the cooH gene. Restoration of the H₂ production activity also coincides with the presence of the CooH protein in Western blotting in the GV1762c strain (Fig. 3). The low rate of H₂ production exhibited by the complement (roughly 16% of that of the wild type) could be due to the structural differences between the fusion protein and the wild-type CooH. With the omission of the first three codons from the 5′ end of cooH, the complementation experiment was originally designed to test if homologous recombination occurs in this strain. However, it was found later that the LacZ-alpha-CooH fusion protein could be expressed off the plasmid, since cloning of the cooH fragment created an in-frame fusion with the lacZ-alpha gene on the plasmid (see Fig. S1 at http://www.nrel.gov/basic_sciences/technology_staff.cfm/tech=16/ID=7). The fusion protein contains the N-terminal 33 residues of LacZ-alpha (MTMITPSAGLTLTKGNKSWVPGPPSRSTVSISL) that replace the first three residues (MSA) of CooH. Plasmid expression of the fusion protein has two characteristics: the expression is driven by the constitutive lacZ promoter independently of CO induction and the fusion protein is 30 residues longer and 3 kDa larger than the wild-type CooH. Indeed, both characteristics are found in the Western blot of CooH in GV1762c (Fig. 3).

The structure and function of CooL can be inferred from electron paramagnetic resonance (EPR) study of its homolog EchC in M. barkeri, where the lone [4Fe-4S] cluster interacts magnetically with the unpaired electrons localized in the [NiFe] catalytic center (10, 24). The conserved cysteine motif and overall sequence similarity between CooL and EchC lend support to the idea that CooL is the putative hydrogenase small subunit in Rx. gelatinosus. In addition to the lone Fe-S cluster identified in the small subunits of all the Ech hydrogenases reported above, the small subunit (27 kDa) of the “standard” heterodimeric uptake hydrogenase in D. gigas also coordinates a medial and a distal Fe-S cluster serving as a plausible electron transfer channel going from the active site of the large subunit to the molecular surface of the small subunit (41). CooL and its homolog may work in concert with other Fe-S proteins within the complex, with CooX being a likely candidate. Similarly, biochemical data on CooX homologs are limited to the EchF subunit from M. barkeri (10, 24). EPR, site-directed mutagenesis, and redox titration data all indicate that the two [4Fe-4S] clusters in EchF are involved in simultaneous electron transport and proton pumping, with the conserved glutamate or the aspartate residue serving as part of the proton translocation machinery (Fig. 1D).

Albeit the two organisms are similar in gene arrangement, attempts to purify the NiFe-hydrogenase failed in R. rubrum (11) and yet had partial success in Rx. gelatinosus (this work). Size exclusion chromatography revealed a smaller size (∼58 kDa) of the solubilized hydrogenase fraction than of the Ech hydrogenases from M. barkeri, T. tengcongensis, and D. gigas; the purification from all of the latter yielded an intact hexameric complex with molecular masses ranging from 172 to 184 kDa (28, 30, 36). A hexameric protein complex of this size would have been eluted in the void volume of the Sephacryl S-200 column used in this study, and yet no hydrogenase activity (assayed with reduced methyl viologen) was detected in the column void volume. It is likely that the 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS) detergent used for the extraction is inefficient in solubilizing all the subunits from Rx. gelatinosus, with the bulk remaining on the membranes. Nevertheless, this 58-kDa protein fraction confers active hydrogenase activity mediated by the artificial mediator methyl viologen, implying the presence of, minimally, the CooH catalytic subunit based on its homology alignment (Fig. 1A). This 58-kDa protein fraction was also active in a hydrogen-deuterium exchange reaction (26). Its preference for H₂ production over H₂ oxidation is another feature shared by the Ech hydrogenases in R. rubrum (11, 27) and Pyrococcus furiosus (31), in contrast to most other NiFe-hydrogenases, which prefer the H₂ oxidation over the H₂ evolution direction (16).

In summary, we have taken the first step toward uncovering the genetic elements controlling the H₂ production reaction coupled to CO oxidation in the photosynthetic bacterium Rx. gelatinosus. The hydrogenase that we identified belongs to a growing family of the energy-converting, proton-pumping NiFe-hydrogenases. The energy-yielding reaction during H₂ production may confer a unique advantage for these microbes to thrive in diverse ecosystems with important ramifications in the global carbon and energy cycles.