Impact of alterations near the [NiFe] active site on the function of the H2 sensor from Ralstonia eutropha


B. Friedrich, Institut für Biologie, Humboldt Universität zu Berlin, Chausseestr. 117, 10115 Berlin, Germany
Fax: +49 30 209 38102
Tel: +49 30 209 38101


In proteobacteria capable of H2 oxidation under (micro)aerobic conditions, hydrogenase gene expression is often controlled in response to the availability of H2. The H2-sensing signal transduction pathway consists of a heterodimeric regulatory [NiFe]-hydrogenase (RH), a histidine protein kinase and a response regulator. To gain insights into the signal transmission from the Ni–Fe active site in the RH to the histidine protein kinase, conserved amino acid residues in the L0 motif near the active site of the RH large subunit of Ralstonia eutropha H16 were exchanged. Replacement of the strictly conserved Glu13 (E13N, E13L) resulted in loss of the regulatory, H2-oxidizing and D2/H+ exchange activities of the RH. According to EPR and FTIR analysis, these RH derivatives contained fully assembled [NiFe] active sites, and para-/ortho-H2 conversion activity showed that these centres were still able to bind H2. This indicates that H2 binding at the active site is not sufficient for the regulatory function of H2 sensors. Replacement of His15, a residue unique in RHs, by Asp restored the consensus of energy-linked [NiFe]-hydrogenases. The respective RH mutant protein showed only traces of H2-oxidizing activity, whereas its D2/H+-exchange activity and H2-sensing function were almost unaffected. H2-dependent signal transduction in this mutant was less sensitive to oxygen than in the wild-type strain. These results suggest that H2 turnover is not crucial for H2 sensing. It may even be detrimental for the function of the H2 sensor under high O2 concentrations.


membrane-bound hydrogenase




regulatory hydrogenase


soluble hydrogenase

Hydrogenases are redox enzymes that catalyse the reversible oxidation of hydrogen into protons and electrons. Based on their metal content, three classes of hydrogenases are distinguished: [FeFe]-hydrogenases, [NiFe]-hydrogenases and the small group of Fe–S cluster-free hydrogenases [1–5]. [NiFe]-hydrogenases are most abundant in nature and display different physiological functions, e.g. energy conservation via H2 oxidation, disposal of reducing equivalents via H2 production and transcriptional regulation by sensing the presence of H2 in the environment. The H2-sensing regulatory hydrogenases (RHs) represent a subgroup of [NiFe]-hydrogenases found in a few proteobacteria that are capable of using H2 as a facultative energy source, e.g. Bradyrhizobium japonicum[6], Rhodobacter capsulatus [7] and Ralstonia eutropha H16 [8]. These microorganisms recognize the availability of H2 by means of an H2 sensor and a histidine protein kinase. For R. eutropha and Rh. capsulatus it has been demonstrated that the RH forms a complex with the corresponding histidine kinase. This sensor complex controls the phosphorylation state of its cognate response regulator, which directs hydrogenase expression at the transcriptional level. Unlike in orthodox two-component systems, this response regulator is active in the nonphosphorylated form. It is suggested that H2 generates a redox signal at the Ni–Fe active site that is further transmitted to the kinase module, which in turn triggers the activity of the regulator [9–13]. The molecular basis of this sensing mechanism is scarcely understood. Recently, the three components of the H2-dependent signal transduction chain have also been found in Thiocapsa roseopersicina and Rhodopseudomonas palustris. Interestingly, in both investigated strains, the H2-signalling pathway is interrupted by gene mutations leaving hydrogenase gene expression H2 independent, a similar phenotype as observed for H2-insensitive strains of R. eutropha[11,14,15].

What are the typical features of an H2-sensing vs. an energy-conserving hydrogenase? A prototypic [NiFe]-hydrogenase consists of a large active-site-containing subunit and a small electron-transferring subunit accommodating several Fe–S clusters. The Ni–Fe active site, buried deep inside the large subunit, is coordinated by four thiol groups of cysteine residues. Three additional diatomic ligands, one CO and two CN groups, are bound to the iron [1–5,16–21]. The crystal structure of an H2-sensing hydrogenase is not yet available. However, comprehensive biochemical and spectroscopic data provide insights into the properties of the well-studied RH from R. eutropha H16 [9,10]. Like the standard hydrogenase module, the RH is composed of two heterologous polypeptides, a large subunit (HoxC) of 52 kDa and a small subunit (HoxB) of 36 kDa. IR spectroscopy revealed that the RH harbours a common Ni–Fe active site with one CO and two CN ligands [22,23]. Assembly of the Ni–Fe cofactor relies on the assistance of the Hyp proteins [24–26]. Proteolytic cleavage of a C-terminal peptide from the large subunit of most [NiFe]-hydrogenases, however, does not occur during the maturation of H2-sensing proteins [11,23]. Moreover, the small subunit of H2-sensing hydrogenases lacks both a signal peptide for membrane translocation and a C-terminal hydrophobic domain that normally anchors the protein to the membrane [27–29]. These features correlate with a cytoplasmic location for H2 sensors [6–8]. Instead of a hydrophobic extension, the small subunit of H2-sensing proteins carries a conserved peptide that is crucial for the formation of a double dimer (HoxBC)2 and its interaction with the kinase HoxJ [9,23,30].

In addition to these structural characteristics, the H2-sensing hydrogenase of R. eutropha H16 displays unique catalytic properties. Its activity, albeit two orders of magnitude lower than that of standard [NiFe]-hydrogenases, is completely insensitive to oxygen, carbon monoxide and acetylene [10]. Hydrogenases are usually sensitive to inhibition by these gaseous compounds [1,2]. Normally oxygen does not affect the structural integrity of the Ni–Fe active site, but reversibly inactivates its catalytic function by occupying the position between nickel and iron that is required for binding of a formal hydride under turnover conditions. The oxygen species has to be reductively removed prior to hydrogen binding resulting in a typically delayed reaction initiation [31–35]. Aerobically isolated RH protein, however, does not require activation prior to H2 binding. Thus, H2 sensors are permanently on alert and ready to react with H2, an ability that is crucial for the sensing function [10,22,36]. At the molecular level, resistance to O2 was proposed to be attributed to a narrowed gas channel that prevents access of O2 to the RH active site. On the basis of a structural model, two more bulky hydrophobic residues seem to shield the Ni from O2 inactivation, whereas the small molecule H2 is still able to enter the Ni–Fe centre. Restoring the standard consensus in the RHs of R. eutropha and Rh. capsulatus by replacing Phe and Ile with Leu and Val yielded O2-sensitive mutant proteins thus supporting the structural predictions [37,38].

To obtain deeper insights into the catalytic and regulatory mechanisms of H2-sensing proteins, a systematic mutagenesis has been initiated by introducing mutations into highly conserved regions of the large subunit [37,39,40]. Of the six highly conserved motifs surrounding the active site in [Ni–Fe] hydrogenases [31,41,42], H2-sensing proteins exhibit alterations in four of these signatures [23,31,39]. In this study, site-directed alterations were introduced into the L0 motif (R[V/I]EG[H/D]) and the resulting mutant proteins were characterized with respect to their regulatory, catalytic and spectroscopic properties.


Amino acid replacements in motif L0 of HoxC

Alignment of more than 100 amino acid sequences of [NiFe]-hydrogenases revealed five consensus motifs in the active-site-containing large subunit of [NiFe]-hydrogenases [31,41]. A substantial number of hydrogenases harbour an additional conserved motif close to the N-terminus of the active-site-containing subunit, referred to as motif L0 [42]. The consensus sequence of the L0 motif in standard hydrogenases consists of mostly charged residues (RIEGH). This consensus is altered at two positions (RVEGD) in H2-sensing proteins (Fig. 1).

Figure 1.

 Sequence and localization of motif L0, the target of mutational alterations in the RH large subunit HoxC. (A) Alignment of the N-terminal region of the catalytic subunits of [NiFe]-hydrogenases harbouring the L0 motif. R.e., Ralstonia eutropha; A.h., Alcaligenes hydrogenophilus; R.c., Rhodobacter capsulatus; B.j., Bradyrhizobium japonicum; D.g., Desulfovibrio gigas. The first four hydrogenases belong to the group of H2 sensors, the remaining four represent energy-linked [NiFe]-hydrogenases. (B) Close-up view of the D. gigas crystal structure [16]. Coordinates were taken from the Brookhaven Protein Database (2frv). The Cα backbone of the large subunit is drawn in black and that of the small subunit in light grey. The amino acids of motif L0 are shown in a stick representation alternately coloured in black and yellow. The Ni–Fe centre with the diatomic ligands and the proximal and medial [Fe−S] clusters are depicted as spheres.

Site-directed mutagenesis, performed by using a two-step PCR procedure, focused on replacing Val12, Glu13 and Asp15 in HoxC. The modified hoxC alleles were introduced into plasmid pCH861 (Table 1) containing the RH genes under the control of the strong R. eutropha soluble hydrogenase (SH) promoter. Subsequently, the mutagenized DNA fragments were subcloned in the broad-host-range vector pEDY309 [23]. Six HoxC mutant proteins were constructed bearing the following amino acid replacements: the two residues exclusively conserved in H2-sensing proteins, Val12 and Asp15, were replaced by Ala/Ile and His/Leu, respectively. The commonly conserved Glu13 was replaced by Gln/Leu (Table 2).

Table 1.  Strains and plasmids.
Strains or plasmidsRelevant characteristicsSource or reference
Ralstonia eutropha
HF574SH (hoxHΔ), MBH (hoxGΔ), RH(hoxCΔ)[39]
SH (hoxHΔ), MBH (hoxGΔ), RH(hoxBCΔ), HoxJ*(hoxJa1264g),
HF435RH(hoxCΔ), HoxJ*(hoxJa1264g)[11]
Escherichia coli
F′traD36 lacIq, Δ(lacZ)M15 proA+B+/e14 (McrA) Δ(lac-proAB)
thi gyrA96 (Nalr) endA1 hsdR17(rkmk+) relA1 supE44 recA1
S17-1Tra+, recA, pro, thi, hsdR chr:RP4-2[60]
pBluescriptSK(+)Ampr, lacZStartagene Cloning Systems
pEDY309RK2 ori, Tcr, Mob+, promoterless lacZ gene[23]
pCH3973.3 kb PstI fragment in pACYC177 containing hoxBC[23]
pCH5942.8 kb HindIII–NcoI fragment in pCH591 containing[23]
pCH861pCH594 without BamHI and BglII sites[39]
pCH9710.64 kb NcoI–BglII cut PCR product in pQE60 containing hoxJInput–His6[9]
pCH9890.36 kb SacII-cut PCR product in pBluescript SK(+)This study
(HoxC D15H) 
pCH9900.36 kb SacII fragment of pCH989 in pCH861This study
pCH9910.36 kb SacII-cut PCR product in pBluescript SK(+)This study
(HoxC V12A) 
pCH9920.36 kb SacII-cut PCR product in pBluescript SK(+)This study
(HoxC V12I) 
pCH9930.36 kb SacII-cut PCR product in pBluescript SK(+)This study
(HoxC E13Q) 
pCH10290.36 kb SacII-cut PCR product in pBluescript SK(+)This study
(HoxC D15L) 
pCH10300.36 kb SacII fragment of pCH991 in pCH861This study
pCH10320.36 kb SacII cut PCR product in pBluescript SK(+)This study
(HoxC E13L) 
pCH10340.36 kb SacII fragment of pCH992 in pCH861This study
pCH10350.36 kb SacII fragment of pCH993 in pCH861This study
pCH10360.36 kb SacII fragment of pCH1032 in pCH861This study
pCH10370.36 kb SacII fragment of pCH1029 in pCH861This study
2.8 kb HindIII–XbaI fragment of pCH594 in pEDY309 containing
This study
pGE5382.8 kb SpeI–HindIII fragment of pCH990 in pEDY309 (HoxC D15H)This study
pGE5392.8 kb SpeI–HindIII fragment of pCH1035 in pEDY309 (HoxC E13Q)This study
pGE5402.8 kb SpeI–HindIII fragment of pCH1034 in pEDY309 (HoxC V12I)This study
pGE5412.8 kb SpeI–HindIII fragment of pCH1030 in pEDY309 (HoxC V12A)This study
pGE5422.8 kb SpeI–HindIII fragment of pCH1036 in pEDY309 (HoxC E13L)This study
pGE5432.8-kb SpeI-HindIII fragment of pCH1037 in pEDY309 (HoxC D15L)This study
Table 2.  Mutant hoxC alleles.
PlasmidpBluescript cloning intermediatepCH861
cloning intermediate
Codon changeAmino acid replacement

H2-sensing function

To investigate the regulatory function of the RH mutant proteins, recombinant plasmids were introduced into R. eutropha HF570 by conjugation. This recipient carries deletions in the SH, membrane-bound hydrogenase (MBH) and RH structural genes (Table 1) to exclude interferences between wild-type and mutant hydrogenases. A chromosomal Φ(hoxK′-lacZ) fusion in strain HF570 allows monitoring of β-galactosidase activity as a reporter for the MBH promoter whose function in this strain depends strictly on the availability of H2[40].

The results are illustrated in Fig. 2. Mutants HoxC[E13Q] and HoxC[E13L] completely lost the ability to sense H2, whereas the Val12 substitutions HoxC[V12A] and HoxC[V12I] downregulated MBH promoter activity to 81 and 25%, respectively. A similar regulatory pattern was obtained when Asp15 was replaced by His and Leu, respectively.

Figure 2.

 H2-sensing function of RH mutant proteins. R. eutropha strains harbouring the plasmid-based Φ(hoxK′-lacZ) fusion were grown in FGN medium with (black bars) and without (grey bars) hydrogen. Cells were harvested at D436 = of 8.0 ± 0.3 and β-galactosidase activity was measured as previously described [57]. Lanes: 1, HF570(pGE377); 2, HF570(pEDY309); 3, HF570(pGE541); 4, HF570(pGE540); 5, HF570(pGE539); 6, HF570(pGE542); 7, HF570(pGE543); 8, HF570(pGE538).

Lithoautotrophic growth

Lithoautotrophic growth of R. eutropha H16, the wild-type strain, occurs in the presence of high concentrations of O2. Cells can be cultivated under an atmosphere of 30% O2, 10% CO2 and 60% H2[43]. H2-responsive strains, such as HF433 [11], do not tolerate O2 at comparably high levels. Both the H2-responsive and H2-independent strains differ only in the activity of the histidine kinase HoxJ [11]. Thus, O2 sensitivity must be related to H2 signal transduction. Interestingly, the HoxC[D15H] derivative isolated in this study proved to be fairly tolerant to O2(Fig. 3), despite maintaining its H2-sensing function (Fig. 2). This O2 tolerance is therefore not conferred by a single amino acid exchange in HoxJ, as in wild-type H16, but relies on an alteration near the H2-binding site of the RH. Unlike HoxC[D15H], the HoxC[D15L] isolate was as O2-sensitive as the control strain containing wild-type RH (Fig. 3).

Figure 3.

 Lithoautotrophic growth at different O2 concentrations. The strains were cultivated for 8 days at 30 °C under a gas atmosphere of 70% H2, 10% CO2, 5% O2 and 15% N2 or 70% H2, 10% CO2, 10% O2 and 10% N2, respectively. Lanes: 1, HF435; 2, HF435(pGE378); 3, HF371(pGE378); 4, HF435(pGE543); 5, HF435(pGE538).

Catalytic activity

To test whether changes in the regulatory capacity in RH mutant proteins correlate with changes in their catalytic activity, recombinant plasmids (Table 2) were transferred to R. eutropha HF574 (Table 1). This recipient is SH-, MBH- and RH-negative but, unlike strain HF570, does not require H2 for hydrogenase gene expression due to inactivation of the kinase HoxJ [11]. To examine the ability of the mutant proteins to react with H2, cells were cultivated under hydrogenase-depressing conditions in fructose–glycerol mineral medium. Two different assays were performed using soluble extracts: (a) amperometric determination of the H2-oxidation rate with methylene blue as the artificial electron acceptor [22], and (b) determination of the D2/H+ isotopic exchange activity that does not rely on electron transfer [10]. The data are presented in Table 3. Replacement of the highly conserved Glu13 led to a complete loss of H2-oxidizing and D2/H+-exchange activity which correlates well with the inability of the respective mutants to sense H2. Furthermore, the HoxC[V12A] mutant exhibited wild-type-like catalytic activity in line with its H2-responsive regulatory phenotype. The remaining mutants, however, displayed catalytic properties (Table 3) that did not correlate directly with the respective regulatory patterns (Fig. 2). Despite its low H2-sensing capacity, the HoxC[V12I] mutant RH showed only a minor decrease in the H2-oxidation rate but a twofold higher D2/H+-exchange activity compared with wild-type. Replacement of Asp15 by Leu almost abolished both catalytic functions, whereas 20% of the regulatory activity was maintained. Replacement of Asp15 by His led to a severe decrease in the H2-oxidizing activity, but the D2/H+-exchange rate and the H2-sensing function were scarcely affected. The low H2-oxidizing activity of HoxC[D15L] and HoxC[D15H] was corroborated using alternative artificial electron acceptors like benzyl viologen or ferricyanide (data not shown).

Table 3.  Catalytic activities of RH wild-type and mutant proteins. Values are the mean of two independent experiments. Activities are given in percent of wild-type activity; 100% corresponds to 0.036 U·mg−1 protein H2-oxidizing activity, 0.110 U·mg−1 protein D2/H+ exchange activity and 7.2 µmol ortho-H2 produced per mg of protein after 100 min of reaction, respectively. ND, not determined.
StrainRelevant characteristicsH2-oxidizing
activity (%)
D2/H+ exchange
activity (%)
Para-/ ortho-
  1. a H 2-oxidizing activity could not be increased using the alternative electron acceptors methyl viologen, benzyl viologen, ferricyanide or phenazine methosulfate.

HF574(pEDY309)RH  0  0ND
HF574(pGE541)HoxC(V12A) 83 92ND
HF574(pGE540)HoxC(V12I) 69207 54
HF574(pGE533)HoxC(E13Q)  0  0 72
HF574(pGE542)HoxC(E13L)  0  0 50
HF574(pGE538)HoxC(D15L)  1a  0ND
HF574(pGE543)HoxC(D15H)  4a 99115

To test whether the low catalytic activities result from protein instability, immunochemical analyses were conducted using antibodies raised against the two subunits of the R. eutropha RH (Fig. 4). The immunoblot illustrates that nearly all mutants contain wild-type levels of HoxB and HoxC. Only the HoxC[D15L] mutant displayed a weak HoxB signal in the soluble extract indicating a decreased stability of the small hydrogenase subunit.

Figure 4.

 Stability of RH mutant proteins. Immunoblot analysis of soluble extracts (20 µg protein per lane) separated on a 12% SDS/PAGE gel. The blot was developed with a mixture of antibodies raised against the RH subunits HoxB and HoxC. Lanes: 1, HF574(pGE377); 2, HF574(pEDY309); 3, HF574(pGE541); 4, HF574(pGE540); 5, HF574(pGE542); 6, HF574(pGE539); 7, HF574(pGE543); 8, HF574(pGE538).

Nickel incorporation

To obtain further information about the incorporation of nickel in the RH mutant proteins, 63Ni autoradiography was performed. Soluble extracts prepared from cells grown heterotrophically in the presence of 63NiCl2 were separated by native PAGE [23]. Characteristic signals corresponding to HoxBC and the double dimer (HoxBC)2 can be distinguished [9,10](Fig. 5, lane 1). Only the double dimer, which is the predominant form in the wild-type, is able to form a complex with HoxJ [9]. A wild-type-like migration pattern was observed for HoxC[E13Q] and HoxC[E13L] mutant proteins (Fig. 5, lanes 3,4) indicating that a Ni-containing double dimer is formed that is, however, devoid of regulatory and catalytic activities. For the HoxC[D15L] and HoxC[D15H] exchanges, by contrast, a dominance of the HoxBC dimer was observed over low amounts of (HoxBC)2 (Fig. 5, lanes 5,6). Thus, all mutants tested remained capable of incorporating nickel into the RH. The Asp15 replacements, however, seem to decrease the stability of the (HoxBC)2 complex in the soluble extract.

Figure 5.

 Nickel content and oligomerization state of selected RH mutant proteins. Cells were grown in FGN medium in the presence of 120 nm63NiCl2. Soluble extracts (200 µg protein per lane) were separated on a 4–15% native PAGE. Lanes: 1, HF574(pGE377); 2, HF574(pEDY309); 3, HF574(pGE533); 4, HF574(pGE542); 5, HF574(pGE543); 6, HF574(pGE538).

Spectroscopic properties

RH mutant proteins were purified taking advantage of the fact that the RH interacts with the input domain of the histidine protein kinase (HoxJInput). Previous experiments have shown that a hexahistidine fusion protein of HoxJInput bound to a Ni–NTA column forms a complex with the RH from soluble extracts of R. eutropha. These RH– HoxJInput complexes can be isolated in a one-step purification procedure [9]. Because of the failure to form a double dimer, the HoxC[D15L] mutant protein was not accessible to this purification technique and was excluded from these experiments. For the remaining mutants, however, sufficient amounts of the RH–HoxJ complex were obtained for spectroscopic analysis.

The RH–HoxJ complexes were examined by EPR and FTIR spectroscopy. Contrary to standard hydrogenases, the RH does not show an EPR signal in the as-isolated, oxidized state. Upon reduction with H2, a rhombic signal characteristic of the Nia-C* state with principal g-values at 2.19, 2.14 and 2.01 has been observed [10,33,39]. A similar signal was also obtained for the HoxC[D15H] mutant. A more heterogeneous pattern was observed for the HoxC[V12I] mutant because additional signals occurred at g = 2.22, 2.17 and 2.02. Apart from the fact that Ni had been incorporated, no Nia-C* signal could be detected for the HoxC[E13Q] and HoxC[E13L] proteins after H2 treatment indicating that no hydride is bound to the bridging position between Ni and Fe (data not shown).

FTIR spectroscopy of the as-isolated wild-type protein revealed two faint bands (2080 and 2071 cm−1) and one intense band (1943 cm−1) corresponding to two cyanides and one CO bound to the iron (Fig. 6A, trace a). Incubation with H2 leads to a characteristic shift in the CO stretch vibration in the FTIR spectrum to 1961 cm−1 (Fig. 6B, trace a), as observed previously [22]. The position of the CN bands, however, remained unchanged under these conditions. Similar spectra were obtained for the HoxC[D15H] mutant protein, although the intensity of the signals was notably lower (Fig. 6, traces b). The HoxC[V12I] mutant protein showed several shoulders of the CO band (Fig. 6, trace e) pointing to different conformations of the Ni–Fe active site. Both HoxC[E13] mutant proteins showed uniform spectral properties (Fig. 6, traces c,d). Identical FTIR spectra were obtained in the presence as well as in the absence of H2 which underlines that these proteins are not reduced by H2. Compared with the wild-type protein, the CO and CN bands are shifted 10–20 cm−1 to higher wavenumbers. Broadening of the signals indicates an increased inhomogeneity of the active site.

Figure 6.

 FTIR spectra of R. eutropha RH wild-type and mutant proteins. RH–HoxJInput–His6 fusion proteins were used for FTIR measurements at room temperature in the ‘as-isolated’ (A) and H2-reduced state (B). (a) RH wild-type (36 mg·mL−1), (b) HoxC[D15H] (62 mg·mL−1), (c) HoxC[E13Q] (100 mg·mL−1), (d) HoxC[E13L] (89 mg·mL−1), (e) HoxC[V12I] (28 mg·mL−1).

The availability of purified mutant proteins permitted the application of a third catalytic assay, namely para-/ortho-H2 conversion which is not accessible to the RH of soluble extracts. This method is based on binding of H2 to the active site and subsequent weakening or cleavage of the H–H bond leading to para-/ortho-H2 conversion upon reversal of the process. In contrast to the previously applied assays, this activity is independent of H2 cleavage and does not imply hydride formation necessarily. Interestingly, para-/ortho-H2 conversion was detected in the HoxC[E13Q] and HoxC[E13L] mutant proteins indicating that H2 can still bind to the active site although H2 cleavage is not detected (Table 3).


Characteristic sequence pattern determine the unique catalytic properties of H2 sensors

RH from R. eutropha is part of a multicomponent signal transduction chain that triggers hydrogenase gene expression depending on the availability of H2 in the environment. Compared with standard [NiFe]-hydrogenases, H2-sensing hydrogenases show variations in the amino acid composition of the conserved L0 motif that is located at the interface of the large and small subunit (Fig. 1). Three of the residues in the L0 motif, two of them unique to RHs, were replaced in order to solve the question whether there is a functional correlation between H2 turnover and H2 sensing.

H2 binding is not sufficient for H2 sensing

The glutamate residue at position 13 in the RH large subunit HoxC is highly conserved in all [NiFe]-hydrogenases. Exchange of HoxC[E13] by either glutamine or leucine led to mutant RH proteins that were incapable of H2 sensing, H2 oxidation and D2/H+ isotopic exchange. Only para-/ortho-H2 conversion was detected, which is indicative of H2 binding but not hydride transfer. For the standard hydrogenase from Desulfovibrio fructosovorans it was shown that the corresponding Glu25 residue of motif L0 has a role in proton transfer [44]. The X-ray structure and spectroscopic data in combination with model compounds and theoretical calculations suggest that the proton resulting from H2 cleavage is transferred to one of the terminal Ni-ligating cysteine residues. The protonated cysteine, by contrast, should donate the proton to the conserved glutamate of motif L0. Replacing Glu25 with Gln abolished both the D2/H+-exchange and the H2-oxidizing activity but H2 binding was still possible as indicated by wild-type-like para-/ortho-H2 conversion. A Nia-C signal was obtained for the mutant protein after redox titration indicating that electron flow to the active site and hydride formation was still possible, whereas fast proton transfer was impaired [44]. Similar results were obtained for RH variants carrying the HoxC[E13Q] and HoxC[E13Q] exchanges with the exception that a Nia-C signal was absent after reduction of the enzyme. This implicates that H2-cleavage and proton transfer are necessary for H2 sensing, and that H2 binding is not sufficient for signal transduction.

O2 tolerance of the H2-sensing complex

Asp15 in motif L0 is strictly conserved within the large subunit of H2 sensors. The energy-conserving hydrogenases harbour a histidine instead. The HoxC[D15H] mutant displayed a remarkable phenotype. Only 4% of the H2-oxidizing activity was maintained in this mutant, although the regulatory and D2/H+-exchange activities reached wild-type levels. The low RH turnover activity of mutant HoxC[D15H] correlated with a higher O2 tolerance during growth on H2 and CO2. While the H2-responding wild-type strain does not tolerate O2 concentrations above 5%, the HoxC[D15H] mutant cells were capable of growing lithoautotrophically at 10% O2.

Sensing of H2 demands tight complex formation between the RH and the kinase HoxJ. According to the current model, H2-dependent signal transduction controls the phosphorylation state of the response regulator HoxA, which activates transcription of the SH and MBH genes in the nonphosphorylated form [11,45]. In strains lacking an active HoxJ protein, hydrogenase gene transcription escapes the H2 control and concomitantly becomes O2 insensitive. Similar effects of mutations in homologous histidine protein kinases involved in H2 sensing have been described for Rh. capsulatus and B. japonicum[46,47]. In this study, we demonstrated experimentally that an alteration in the catalytic subunit HoxC of the RH results in H2-dependent, but O2-tolerant hydrogenase gene transcription. Because O2 does not affect the autophosphorylation status of HoxJ (O. Lenz & B. Friedrich, unpublished results) or H2 cleavage at the Ni–Fe site [9,10,37], we postulate an effect of O2 on intramolecular electron transfer and/or subsequent signal transmission to the input domain of the kinase HoxJ.

H2 sensing is independent of H2 turnover

As described above, the D2/H+ activity of the HoxC[D15H] protein was wild-type-like, whereas its H2-oxidizing activity was very low (Table 3). These results indicate that the mutant protein still catalyses H2 cleavage, resulting in a bridging hydride that could be detected by EPR as the Nia-C* state. Moreover, proton transfer was not affected in this mutant protein, as deduced from the high D2/H+-exchange activity. The low H2-turnover rate indicates that electron transfer via the Fe–S clusters in HoxB is almost completely abolished in the HoxC[D15H] protein, whereas its H2-sensing activity is almost unaffected. This observation strongly suggests that H2 turnover is not required for signal transmission.

Interaction between the RH and the kinase HoxJ depends on the C-terminal extension of HoxB and the N-terminal input module of HoxJ, which accommodates a Per-Arnt-Sim (PAS) domain [9,48]. In many cases, PAS domains carry redox cofactors such as flavins or Fe–S clusters [49]. No evidence for a bound cofactor has been obtained in the case of HoxJ, leaving open the question as to whether there is a direct electron transfer between the RH and the kinase HoxJ. H2 oxidation normally leads to reduction of the Fe–S clusters in the small subunit of [NiFe]-hydrogenases which can be monitored by EPR spectroscopy [2]. Surprisingly, no EPR signals were detected upon reduction of the RH by H2[10]. Recent UV-Vis and analysis by X-ray absorption spectroscopy point to an unusual pattern of Fe–S clusters in HoxB, two [2Fe−2S] clusters and a [4Fe−3Fe−3O] cluster being reduced by H2[50]. Based on this observation, it is proposed that the reduction of the Fe–S clusters is accompanied by a conformational change that is then transmitted to HoxJ and in turn controls the kinase activity of the transmitter domain.

Experimental procedures

Bacterial strains and growth conditions

Bacterial strains and plasmids are listed in Table 1. Strains with the initials HF were derived from R. eutropha H16 harbouring the endogenous megaplasmid pHG1. R. eutropha strains were cultivated in mineral salt medium containing 0.2% (w/v) fructose and 0.2% (v/v) glycerol (FGN medium) [51]. For protein purification 30 g of cells (wet weight) obtained from 4-L FGN cultures were used.

Escherichia coli strains were grown in Luria–Bertani medium [52]. Antibiotics were used at the following concentrations: 15 µg·mL−1 tetracycline for R. eutropha, and 15 µg·mL−1 tetracycline and 100 µg·mL−1 ampicillin for E. coli.

Recombinant DNA techniques and plasmid construction

Standard DNA techniques were used [53]. Site-directed mutagenesis in hoxC was performed according to the method of Chen and Przybyla [54]. Plasmid pCH397 containing hoxC served as a template, and Vent DNA polymerase (New England Biolabs, Frankfurt, Germany) was used for the amplification of 0.65-kb fragments with the synthetic oligonucleotide 5′-CTGCCGGTGATCAATGTCGCCGGTTGCCCG-3′ and the appropriate mutagenic oligonucleotide. The resulting PCR product and the synthetic oligonucleotide 5′-TCACGCCGGCAAGATCCGCCAATGCGCGGC-3′ were used as primers in a second amplification step yielding a 0.79-kb PCR product. The 0.36-kb SacII-cut PCR products were cloned into SacII-digested pBluescriptSK(+) and checked by DNA sequencing. Subsequently, mutagenized SacII fragments were cloned into pCH681. Of these derivatives, the entire RH gene region under the control of the SH promoter [23] was excised as 2.8-kb SpeI–HindIII fragment and subsequently ligated to the broad-host-range vector pEDY309. The resulting plasmids were transferred from E. coli S17-1 to R. eutropha HF435, HF570 and HF574 (Table 1) using a spot-mating technique. The codon changes in hoxC and the resulting plasmids are shown in Table 2.

Ni-NTA affinity chromatography procedure

RH-containing cells (30 g) were resuspended in 30 mL 50 mm potassium phosphate buffer pH 7. Cells were disrupted by two passages through a chilled French pressure cell. Cell debris and membrane pellets were removed by ultracentrifugation (30 min, 100 000 g, 4 °C). Concomitantly the soluble extract (2 mL) of a 400 mL LB culture of E. coli JM109(pCH971) was prepared. One hundred milligrams of protein containing HoxJInput were applied to a Ni-NTA superflow (Qiagen, Hilden, Germany) column. The column was washed with 20 column vol. of washing buffer (50 mm Tris/HCl, pH 8.0, 300 mm NaCl, 20 mm imidazole). Subsequently, the RH-containing soluble extract (15 mL) was applied to the column which was again washed with 20 column vol. of washing buffer. Finally, the RH–HoxJInput complex bound to the Ni-NTA matrix was eluted with 5 vol. of elution buffer (50 mm Tris/HCl, pH 8.0, 300 mm NaCl, 250 mm imidazole). Protein-containing fractions were analysed by SDS/PAGE. Fractions containing RH were combined, dialysed against 20 mm Tris/HCl, pH 8.0 and finally concentrated (Amicon Ultra 100 000 MWCO, Millipore Corp., Bedford, MA). Typically, ≈ 8 mg enriched protein were obtained. Protein concentrations were determined using the method described by Lowry [55].

Immunoblot analysis

Proteins were separated by electrophoresis in 12% polyacrylamide SDS gels and transferred to nitrocellulose membranes (Pall) according to a standard protocol [56]. The R. eutropha RH was identified with a mixture of antibodies raised against the individual subunits HoxB and HoxC.

63NiCl2 autoradiography

Cells were grown in FGN medium in the presence of 63NiCl2 (6.38 mCi·mL−1). Soluble extracts were prepared and subjected to native gradient PAGE (4–15%). Gels were run in a continuous buffer system (90 mm Tris, 80 mm borate, and 2.5 mm EDTA, pH 8.3) at 200 V and 4 °C for 3000 V*h. After electrophoresis, gels were dried under vacuum and subjected to autoradiography using an SI 550 storage PhosphorImager (Molecular Dynamics, Sunnyvale, CA).

Activity assays

H2-oxidizing activity was measured by an amperometric H2-uptake assay, as described earlier, using a Clarke-type H2-electrode with methylene blue as the artificial electron acceptor [22]. One unit of H2-methylene blue oxidoreductase activity is defined as the amount of enzyme which catalyses the consumption of 1 µmol H2 per min. D2/H+-exchange activity was quantified in a stirred membrane leak chamber fitted to a mass spectrometer (Masstorr 200 DX quadrupole, VG Quadrupoles Ltd, Middlewich, UK). For the assays, 10 mL of 50 mm Mes buffer pH 5.5 were saturated with 20% D2 in Ar and 1 µL of 1 m sodium dithionite was added in order to scavenge residual oxygen. The reaction was started by addition of RH sample to a final concentration of 0.3 mg·mL−1 of protein. Evolution of D2, HD and D2 were followed by scanning masses 1–6 at 1 atomic mass·s−1 and the temperature of the chamber was controlled at 30 °C. β-Galactosidase activity was determined as described previously [57]. Para-/ortho-H2 conversion was measured in 50 mm Tris buffer, pH 8.0 in the presence of 10 mm sodium dithionite under an atmosphere of para-enriched hydrogen (approximately 50%para, 50%ortho). RH samples were added to 1.5 mL of buffer to a final protein concentration of 1.9 mg·mL−1. Sample vials were placed in a shaker at 35 °C and at different times the atmosphere was analysed in a gas chromatograph with a thermal conductivity detector. Para-enriched H2 was prepared as described previously [58].

FTIR and EPR spectroscopy

IR spectra were recorded in a Nicolet 860 Fourier-transform spectrometer, equipped with a MCT detector and a purge gas system for removal of CO2 and H2O (Whatman Inc., Brentford, UK). The samples were concentrated by ultrafiltration with Microcon-30 (Millipore, Schwalbach, Germany) to 15–30 mg·mL−1 and introduced into a CaF2 cell with a 82 µm path length. Reduced samples were flushed previously under 1 atm H2 for 10 min and transferred anaerobically to the cell with a gas-tight syringe. The temperature of the cell was controlled at 25 °C with a Huber CC 230 thermostat. IR spectra were averaged from 1024 scans and the spectral resolution was 2 cm−1. Spectra were blank-subtracted and baseline corrected using omnic software (Nicolet, Thermo Scientific, Waltham, MA).

EPR measurements were carried out with a Bruker ER200D instrument (Bruker Optics Inc., Billerica, MA) operating in the X-band and using a DPPH standard (in a T-type double cavity) for frequency calibration. Portions of sample were introduced into a spectroscopic quartz probe cell closed with a rubber septum and reduced under 1 atm H2 for 10 min. Spectra were recorded at 77 K and were typically performed by accumulating 10 scans (in continuous mode, ≈ 20 min of total recording time) at 9.55 GHz microwave frequency, 19.5 mW microwave power, 100 kHz modulation frequency, 2 G modulation amplitude, 40 ms time constant and 1 × 105 receiver gain. No significant signal saturation was observed under those conditions.


We thank Angelika Strack for excellent technical assistance and Thorsten Buhrke for his help in the initial phase of this work. This project was supported by the Deutsche Forschungsgemeinschaft (Project C1 within the Sfb498), Fonds der Chemischen Industrie and Spanish Ministry of Science and Technology (Project BQU2003-04221).