INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Hydrogenase catalyzes the reversible reaction H₂2H⁺ + 2e (EC 1.18.99.1 and 1.12.2.1) (12). The enzyme occurs in a wide variety of eubacteria and archaea, and the catalytic direction depends on cellular location and metabolic function. Azotobacter vinelandii has a membrane-bound hydrogenase that primarily catalyzes the oxidative reaction (38). Membrane-bound [NiFe] hydrogenases consist of a membrane-bound catalytic heterodimer and an integral membrane cytochrome b (12). While the heterodimer alone can catalyze hydrogenase reactions with artificial electron acceptors and donors, the cytochrome b is essential for physiologically relevant H₂ oxidation (8-10, 20, 26, 40). Proposed functions for HoxZ, the cytochrome b in A. vinelandii, include electron transfer from the heterodimer to the respiratory chain, anchoring the heterodimer to the membrane, reduction of the catalytic site, and stabilization of the enzyme (40).

All three subunits of membrane-bound hydrogenases are metalloproteins, and the ligands in the subunits have unusual structural and functional features. The large subunit contains the site of catalysis, and the iron in the [NiFe] catalytic site has the nonprotein ligands CO and CN (18). The small subunit of the heterodimer has three aligned FeS clusters (45). Irons in FeS clusters are usually ligated by cysteines, but the distal 4Fe4S cluster includes a histidine ligand (45), which is essential for activity (16). Substitution of one of the cysteine ligands of the central 3Fe4S cluster in A. vinelandii resulted in the loss of O₂ sensitivity in hydrogenase, which is normally highly sensitive to O₂ inactivation (25). The hemes and ligands of the third subunit have been the subject of recent studies (5, 7, 9, 16).

Dross et al. (9) determined that the integral membrane component of hydrogenase in Wolinella succinogenes, HydC, had four hydrophobic segments, and they suggested that this protein was a cytochrome b. From their alignment of full and partial homologous sequences to HydC, they indicated proposed -helical transmembrane segments and two heme ligands. However, our alignment of these hydrogenase b-type cytochromes, which included subsequently released full sequences of Azotobacter chroococcum (10) and Bradyrhizobium japonicum (44), suggested that four conserved histidines may serve as ligands to two hemes, in agreement with Berks et al. (7) and Gross et al. (16). The homologous protein in Alcaligenes eutrophus, HoxZ, was shown to be a diheme cytochrome b that anchors the heterodimer to the periplasmic side of the cytoplasmic membrane and donates electrons to ubiquinone (6). When three of the conserved histidines in HydC of W. succinogenes were replaced with alanine or methionine, the mutant strains had negligible levels of H₂-dependent electron transfer to the anaerobic electron acceptor, menaquinone (16). The fourth conserved histidine was suggested as the fourth ligand (7, 16) but was not identified experimentally. The requirement of three ligands for physiologically relevant electron transfer from H₂ was shown (16), but further investigations of the roles of these ligands have not been carried out.

Unlike W. succinogenes, A. vinelandii is an aerobic, nitrogen-fixing bacterium that has ubiquinone as part of the respiratory chain. Along with the four conserved histidines, there are five histidines conserved to various degrees among 14 hydrogenase b-type cytochrome homologues. Four of these five residues occur in A. vinelandii HoxZ but not in W. succinogenes. We have investigated the effect of histidine substitutions in both O₂-dependent H₂ oxidation and H₂-dependent methylene blue (MB) reduction in vivo. Substitution of histidine residues with nonligands (alanine), alternative ligands (tyrosine), and exogenous ligands (Imd) provided a means of identifying the heme ligands and probing the heme site and the role of the ligands in HoxZ. Imd rescued mutants with substitutions in one of the putative heme ligands. This is the first report of imidazole (Imd) rescue of activity in a membrane protein and the first report of in vivo rescue in a nonheterologous system.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Schematic of A. vinelandii HoxZ. The sequence of A. vinelandii HoxZ (National Center for Biotechnology Information [NCBI] GenBank accession no. L23970, submitted by Menon et al. [27]) is shown in Fig. 1. The designated transmembrane segments were most similar in selected residues and length to those predicted by NCBI Entrez protein query for accession no. P23000 (A. vinelandii HoxZ).

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Schematic of A. vinelandii HoxZ.

Plasmids, bacterial strains, and mutagenesis. The plasmids and bacterial strains used in this study are listed in Table 1. Oligonucleotides for the mutagenesis were designed using the software Oligo (version 3; Molecular Biology Insights, Plymouth, Minn.). The oligonucleotide-directed PCR-based methods of Merino et al. (29) and Weiner et al. (46) were used for site-directed mutagenesis of the plasmid template pAVHoxZ⁺. The PCR product was then transformed into competent Escherichia coli DH5 (Gibco BRL, Grand Island, N.Y.), which were plated on Luria-Bertani agar medium (39) supplemented with ampicillin (60 mg/liter) (Sigma, St. Louis, Mo.) and grown at 37°C overnight. Single colonies were grown overnight in ampicillin-supplemented Luria-Bertani liquid medium, and plasmid preparations were made according to Lee and Rasheed (21). Sequencing by the Central Services Lab of the Center for Gene Research and Biotechnology, Oregon State University, was used to screen and verify the mutants.

Growth of A. vinelandii. A. vinelandii strains were grown at 30°C in Burk medium or on Burk plates (43). For rifampin-resistant (Rif^r) strains, the medium was supplemented with 20 mg of rifampin (Sigma) per liter (making Burk-Rif medium). Cultures were shaken continuously (150 cycles/min) during growth and grown to optical densities at 600 nm (OD₆₀₀) of 0.8 to 1.1 with a Beckman DU-70 spectrophotometer, except for the wild-type (Wt) activity assay, in which case cultures were grown to OD₆₀₀ of 0.5 to 1.2. For Imd (Sigma)-grown cultures, Imd (made to pH 7.5 with HCl) was added to a final concentration of 10 mM (4) at the time of inoculum addition. Cells were collected by centrifugation.

Transformation and screening. A. vinelandii strains were made competent by several transfers on iron- and molybdenum-deficient, 28 mM ammonium acetate-supplemented Burk medium (36) under O₂-limited conditions (33). To generate a Rif^s HoxZ A. vinelandii strain, pHoxZ (1 µg) was transformed into competent A. vinelandii DJ (Wt) without the aid of an antibiotic marker. Except for growth in Burk medium without rifampin, other aspects of transformation and screening were the same as for other mutants. Transformations with all other HoxZ constructs were done using plasmids linearized with the restriction enzyme SmaI (U.S. Biochemical, Cleveland, Ohio; Promega, Madison, Wis.). Linearized plasmid preparations (1 µg of DNA) of pHoxZ H33, H74, H194, H208, and H194,208A or -Y were individually cotransformed with the EcoRI (Promega) 1.7-kb fragment of pDB303 (10 ng), which confers Rif^r into Rif^s A. vinelandii DJ (Wt), using the method of Premakumar et al. (36). Linearized plasmids pHoxZ H97, H98, H136, and H191A or -Y were cotransformed similarly into Rif^s A. vinelandii HoxZ. To compare only hox mutants and not differences due to the lower growth rate in medium with rifampin, A. vinelandii DJ and A. vinelandii HoxKG were made Rif^r by transformation using the 1.7-kb EcoRI fragment of pDB303. The day following transformation, bacteria were transferred to Burk-Rif agar plates to select for Rif^r transformants. Single colonies were picked and transferred three times on Burk-Rif medium to allow for complete segregation (22, 36). A phenotypic assay of H₂ oxidation ability similar to that of Sayavedra-Soto and Arp (40) was used to screen for coincident transformation of the HoxZ mutants. Test tubes containing 2 ml of Burk-Rif were inoculated from single colonies; cultures were grown to an OD₆₀₀ of 0.4 and capped with butyl rubber stoppers, and H₂ (4.46 µmol) was added (high-purity commercial grade H₂ and N₂ were purchased locally). Cultures were then grown on a shaker table (150 rpm) for 16 h at 30°C. For screening, culture headspace was sampled and analyzed for the presence of H₂ using a Shimadzu GC-8A thermal conductivity detector gas chromatograph (Molecular Sieve 5A column; 5 ft long, 1/8-inch diameter, 200°C injection temperature, 120°C column temperature, 60-mA current). Strains with phenotypes differing from that of the background parental strain were confirmed as mutants by DNA sequence analysis of a PCR product of the genomic DNA preparation (3). Sequencing was performed at the Central Services Lab of the Center for Gene Research and Biotechnology, Oregon State University.

Whole cell O₂-dependent H₂ oxidation. To determine when to harvest A. vinelandii for assays, a Wt activity assay compared OD₆₀₀ (0.5 to 1.2) and protein contents (microbiuret assay [14] with bovine serum albumin as a standard) to O₂-dependent H₂ oxidation activity. A. vinelandii cultures for other assays were then harvested in the range of greatest activity, OD₆₀₀ of 0.8 through 1.0. The specific activities in nanomoles of H₂ oxidized per minute per milligram at these ODs were 0.8 (188), 0.9 (184), and 1.0 (173). Preliminary experiments showed that cells resuspended in growth medium had H₂ oxidation activity, but cells resuspended in phosphate buffer did not, probably due to oxidation of the hydrogenase enzyme. Therefore 3 mM sucrose, equivalent to a starvation level of carbon for E. coli (0.1% glucose) (2), was added to the buffer to maintain respiratory protection of O₂-sensitive enzymes. Suspension cultures (1.5 ml) were centrifuged for 1 min at 14,000 rpm, resuspended in 50 mM phosphate buffer (pH 7.0)-3 mM sucrose (1.5 ml), then placed in 10-ml serum vials, capped with a rubber sleeve stopper, and placed upside down in a 30°C water bath at a shaker rate of 200 cycles/min. To initiate the assay, H₂ (4.46 µmol) was added to the vial. Headspace samples (200 µl) were injected into the thermal conductivity detector gas chromatograph at 0, 15, and 30 min. For assays with Imd, either cultures were grown in 10 mM Imd, or Imd and chloramphenicol (Aldrich Chemical Company, Milwaukee, Wis.) were added at the initiation of the experiment to final concentrations of 10 and 3 mM, respectively. Four individual replications of this assay were recorded for each strain. In all assays, results for the control strains grown with Imd were not significantly different (least squares difference at P = 0.05 [LSD_0.05]) from the results for control strains grown without Imd.

Whole cell, H₂-dependent MB reduction. Suspension cultures (40 ml) were centrifuged 10 min at 7,500 rpm; the cells were resuspended in 1 ml of 50 mM Tris buffer-5 mM EDTA (pH 7.3) and placed in a 10-ml serum vial stoppered with a rubber sleeve cap. The vial was purged with N₂ to remove O₂. A sample cuvette (7.5 ml; Beckman 2097) containing a total of 1 ml of 50 mM Tris buffer-5 mM EDTA (pH 7.3) including 150 µM MB was similarly stoppered and flushed with N₂, and H₂ (1 ml) was added for assays with H₂. All reactions were started with the addition of N₂-purged culture (50 µl) to the sample cuvette. Under the conditions of these MB experiments, activation was not required for reduction of the catalytic site since the heterodimers were sufficiently reduced from the flushing of nitrogen gas and removal of O₂ to reduce MB without a lag period and in the absence of dithionite. The decrease in absorbance at 690 nm was monitored with a Beckman DU-70 spectrophotometer for 1 min, and the greatest difference in absorbance in a 10-s time period was recorded. Four individual replications of this assay were recorded for each strain. Calculations were based on a molar extinction coefficient of 11.4 mM¹ cm¹ at 690 nm. MSU Stat (Montana State University, Bozeman) software was used for analysis of variance of the means for the data from both O₂-dependent H₂ oxidation and H₂-dependent MB reduction assays.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Conservation of histidines in hydrogenase b-type cytochromes. An alignment and analysis of 14 eubacterial hydrogenase b-type cytochromes (alignment not shown) show that four conserved histidines are found in the predicted transmembrane -helical segments (Fig. 1). One histidine is near the center of helix A, one is at the beginning of helix B, and two are in helix D, in agreement with Gross et al. (16). Six histidines occur in A. vinelandii without corresponding histidines in W. succinogenes. Four of these in the loop regions were selected for mutagenesis. Three of the four are in loop BC, predicted to be on the cytoplasmic side, and the fourth is in loop CD, predicted to be on the periplasmic side of the cytoplasmic membrane (Fig. 1). A. vinelandii HoxZ is only distantly related (29% identical) to the hydrogenase cytochrome b of W. succinogenes (phylogeny not shown). However, the similarities seen in biochemical assays of these distantly related proteins imply similar structural and functional importance of all hydrogenase b-type cytochromes. To examine the roles of these histidines in HoxZ, mutants were constructed to produce altered proteins with specific histidines replaced with either alanine or tyrosine.

Effect of histidine replacements on whole cell O₂-dependent H₂ oxidation. To determine whether hydrogenase is functional in physiological electron transport in A. vinelandii hoxZ mutants, H₂ oxidation with O₂ as the terminal electron acceptor was assayed (Table 2). A. vinelandii DJ (Wt) and Rif^r A. vinelandii DJ (WtR) served as positive controls. HoxKG Rif^r (a partial deletion mutant of the catalytic heterodimer) served as a negative control. HoxZ Rif^r (a partial deletion mutant of hydrogenase cytochrome b) served as a control for the loss of a functional HoxZ. Wt and WtR were not significantly different. Physiologically relevant H₂ oxidation activity was undetectable for both Hox mutants.

Effect of Imd on whole cell O₂-dependent H₂ oxidation of mutants. In some proteins, Imd can serve as an exogenous ligand to heme in proteins where a natural ligand has been removed (4). To determine if Imd, a histidine analogue, could serve as a free ligand in physiologically relevant H₂ oxidation, 10 mM Imd was added to the growth medium. There was no significant difference in H₂ oxidation between Wt and WtR, with or without Imd (Table 2). Furthermore, Imd did not restore activity to HoxKG Rif^r or HoxZ Rif^r.

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Recovery of O2-dependent H2 oxidation by H194A with the addition of Imd. Symbols represent H194A alone (), H194A plus Imd added at initiation of assay (), H194A plus Imd and chloramphenicol added at initiation of assay (), H194A grown in Imd (), and no-cell control ().

Whole cell H₂-dependent MB reduction. MB reduction can be quantitatively measured spectroscopically and must be assayed anaerobically to prevent O₂ from reoxidizing the artificial electron acceptor. MB can accept electrons directly from the heterodimer (42) and may have other nonspecific binding locations along the electron transport chain and in the cell. Since MB can also be nonspecifically reduced, experiments were conducted in the absence and presence of H₂ (1) (Table 3) to account for the level of endogenous cellular reduction. In addition, assays were conducted on strains grown in the absence and presence of Imd.

View larger version (44K): [in this window] [in a new window]   FIG. 3.   Effect of Imd on H2-dependent MB reduction and O2-dependent H2 oxidation. Bars denote standard error; only top half of column error is shown. The x axis indicates A. vinelandii controls and HA mutants.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Influence of residue substitutions in HoxZ on O₂-dependent H₂ oxidation. In A. vinelandii HoxZ, the four histidines 33, 74, 194, and 208 are essential for physiologically relevant hydrogenase activity (Table 2) and are proposed as the ligands to two hemes in this cytochrome b. The first three residues are homologous to histidines in W. succinogenes HydC, which when substituted with other residues also resulted in cells with low or no physiologically relevant hydrogenase activity (16). In several quinone-interacting b-type cytochromes (7, 11, 16, 49), four conserved histidines occur in transmembrane segments. The invariant histidines in these proteins have been suggested to be ligands to two hemes. In succinate:ubiquinone oxidoreductase (11) and the bc₁ complex (49), residue replacements of conserved histidines resulted in loss of activity and changes in absorption spectra, which supported the conclusion that these histidines were ligands to two hemes. Crystal structures of various bc₁ complexes confirmed the invariant histidines as heme ligands (48). These biochemical approaches would have been valuable in demonstrating that the four conserved histidines in A. vinelandii HoxZ are indeed ligands to the hemes. Difference spectra (reduced minus oxidized) of A. vinelandii Wt and HoxZ membranes were examined, but given the large heme background found in this bacterium (e.g., cytochrome bc₁ complex and cytochrome bd oxidase), we were unable to identify the absorption associated with HoxZ. A. vinelandii hydrogenase, like most NiFe hydrogenases, was purified as a heterodimer without HoxZ (42). Our efforts to purify A. vinelandii HoxZ along with the heterodimer or HoxZ itself were not successful. However, the current site-specific mutagenesis approach shows that loss of any of these four ligands results in the complete loss of hydrogenase activity and supports the conclusion that the four fully conserved histidines in the hydrogenase b-type cytochrome are ligands to two hemes.

Influence of alterations in HoxZ on hydrogenase activity of the HoxKG heterodimer. To localize the loss of O₂-dependent H₂ oxidation activity to the substitutions in HoxZ, the hydrogenase activity in the mutants was determined with an assay (H₂-dependent MB reduction) requiring only functional heterodimers. Previous studies (5, 16, 32, 40) showed that the heterodimer is active in mutants in which the cytochrome b component of hydrogenase is disrupted. As expected, the hydrogenase heterodimers were active in all the HoxZ mutants (Table 3), which indicated that heterodimer activity could still occur without a functional HoxZ. However, quantitative assays revealed that the A. vinelandii HoxZ mutants did influence the heterodimer activity and that the effects depended on the residue substituted and whether or not Imd was included (Fig. 3). H33A/Y had full Wt activity, while H74A, H194A and H208A had less than Wt activity. The mechanism by which HoxZ influences the H₂-dependent MB reduction activity of the heterodimer is uncertain, particularly since MB can bind directly to the heterodimer (42). Perhaps these three HoxZ ligands, H74, 194, and 208, influence alternative and more efficient pathways of electron transfer to MB, or perhaps they fulfill a structural role in anchoring the heterodimer correctly. For H194 and H208, but not H74, Imd restored this function. Imd appears to influence a structural role in support of MB reduction by the heterodimer. The HA substitution in position 136, which is not a heme ligand, resulted in ~20% greater H₂-dependent MB reduction and 75% lower O₂-dependent H₂ oxidation rate than Wt. If H136 is part of a quinone binding site, perhaps the HA substitution decreases the efficiency of quinone binding while increasing the efficiency of MB binding. Since H136 is on the cytoplasmic side, this position probably is not involved in heterodimer binding.