Dissection of the Hydrogen Metabolism of the Enterobacterium Trabulsiella guamensis: Identification of a Formate-Dependent and Essential Formate Hydrogenlyase Complex Exhibiting Phylogenetic Similarity to Complex I

Biological H₂ production presents an attractive alternative for fossil fuels. However, in order to compete with conventional H₂ production methods, the process requires our understanding on a molecular level. FHL complexes are efficient H₂ producers, and the prototype FHL-1_Ec complex in E. coli is well studied. This paper presents the first biochemical characterization of an FHL-2-type complex. The data presented here will enable us to solve the long-standing mystery of the FHL-2_Ec complex, allow a first biochemical characterization of T. guamensis’s fermentative metabolism, and establish this enterobacterium as a model organism for FHL-dependent energy conservation.

ABSTRACT

Trabulsiella guamensis is a nonpathogenic enterobacterium that was isolated from a vacuum cleaner on the island of Guam. It has one H₂-oxidizing Hyd-2-type hydrogenase (Hyd) and encodes an H₂-evolving Hyd that is most similar to the uncharacterized Escherichia coli formate hydrogenlyase (FHL-2_Ec) complex. The T. guamensis FHL-2 (FHL-2_Tg) complex is predicted to have 5 membrane-integral and between 4 and 5 cytoplasmic subunits. We showed that the FHL-2_Tg complex catalyzes the disproportionation of formate to CO₂ and H₂. FHL-2_Tg has activity similar to that of the E. coli FHL-1_Ec complex in H₂ evolution from formate, but the complex appears to be more labile upon cell lysis. Cloning of the entire 13-kbp FHL-2_Tg operon in the heterologous E. coli host has now enabled us to unambiguously prove FHL-2_Tg activity, and it allowed us to characterize the FHL-2_Tg complex biochemically. Although the formate dehydrogenase (FdhH) gene fdhF is not contained in the operon, the FdhH is part of the complex, and FHL-2_Tg activity was dependent on the presence of E. coli FdhH. Also, in contrast to E. coli, T. guamensis can ferment the alternative carbon source cellobiose, and we further investigated the participation of both the H₂-oxidizing Hyd-2_Tg and the H₂-forming FHL-2_Tg under these conditions.

IMPORTANCE Biological H₂ production presents an attractive alternative for fossil fuels. However, in order to compete with conventional H₂ production methods, the process requires our understanding on a molecular level. FHL complexes are efficient H₂ producers, and the prototype FHL-1_Ec complex in E. coli is well studied. This paper presents the first biochemical characterization of an FHL-2-type complex. The data presented here will enable us to solve the long-standing mystery of the FHL-2_Ec complex, allow a first biochemical characterization of T. guamensis’s fermentative metabolism, and establish this enterobacterium as a model organism for FHL-dependent energy conservation.

INTRODUCTION

Future use of molecular hydrogen (H₂) as an energy carrier is a key step to replacing or reducing the anthropogenic use of fossil fuels. Sustainable H₂ production from biological sources is one approach to tackle this problem. Many microorganisms produce H₂ after the fermentation of glucose, but the use of glucose directly competes with human food production and requires the establishment of alternative carbon sources as the substrates.

The apathogenic enterobacterium Trabulsiella guamensis is able to grow on both cellobiose and glucose, fermenting these through the mixed acid fermentation (1). T. guamensis was isolated from the dust of a vacuum cleaner at a library on Guam (1). Phylogenetically related strains have also been isolated among cellulolytic microbes (2). It is phylogenetically most closely related to the pathogenic enterobacterium Salmonella enterica. Its genome sequence has been made available within the context of the project “Assembling the Tree of Life” to increase the breadth of genome sequence data available for the enterobacterium (https://atol.genetics.wisc.edu/index.php). With regard to its [NiFe]-hydrogenase (Hyd) complement, the genomic data predict that the organism encodes only one uptake Hyd and one enzyme complex similar to the formate hydrogenlyase 2 complex of Escherichia coli (FHL-2_Ec).

The FHL-2_Ec complex is related to the H₂-producing FHL-1_Ec complex, which catalyzes the oxidation of formate to CO₂ and H₂. The FHL-1_Ec complex is the main hydrogen-producing enzyme complex in E. coli under fermentative growth conditions, and it detoxifies formate. It consists of a formate dehydrogenase (FdhH) for formate oxidation and three iron-sulfur cluster-containing proteins, HycB, HycF, and HycG, that transfer electrons to the hydrogenase subunit HycE. It also has two membrane subunits, HycC and HycD, that are not involved in the redox reaction but are essential for full FHL-1_Ec activity (Fig. 1A) (3). The genes coding for the hydrogenase subunits and protein-specific chaperones of FHL-1_Ec are in the the hycA-hycI (hycA-I) operon. The FdhH protein is encoded by the fdhF gene, but its location on the chromosome is not in the vicinity of the hyc genes. The FHL-1_Ec architecture resembles the proton-translocating complex I of the respiratory chain (4, 5). Unfortunately, our previous attempts to prove the predicted proton translocation ability of the FHL-1_Ec complex have remained unfruitful (6).

FIG 1.

(A) Operon structure of formate hydrogenlyase complex-encoding genes from E. coli and T. guamensis. The numbers indicate the percentage of identical amino acid residues shared between the corresponding proteins. (B) Schematic representation of the FHL-2_Tg complex with high similarity to complex I. The soluble domain shows the formate dehydrogenase protein FdhH with selenocysteine (Se) and bis-molybdopterin adenine dinucleotide (Mo-bis-MGD) as cofactors, the HyfA and HyfH electron transfer subunits with iron-sulfur clusters (FeS cluster; orange and yellow spheres), the HyfI small subunit (FeS cluster), and the HyfG hydrogenase (Hyd) subunit with the [NiFe] cofactor. The membrane domain consists of 5 subunits, but the order is currently not known.

The second FHL complex (FHL-2_Ec) is predicted to have five instead of two membrane subunits. The FHL-2_Ec complex is encoded by the hyfA to -R (hyfA-R) genes, but neither gene expression nor protein activity for this FHL-2_Ec complex was yet directly demonstrated (7–11). It was further predicted that this complex might also rely on FdhH as an electron donor (8). The two FHL complexes can be unambiguously distinguished based on the presence of the FHL-2-specific protein HyfE (Fig. 1A).

The T. guamensis FHL-2_Tg complex, encoded by the hyf operon, is also predicted to have five membrane subunits and four subunits encoded in the operon forming the soluble domain (Fig. 1B) However, no formate dehydrogenase (FdhH) is encoded within the vicinity of the hyf operon but is encoded elsewhere on the chromosome. Notably, however, the FHL-2_Tg hydrogenase large subunit phylogenetically clusters together with the corresponding E. coli complex I protein NuoCD and its FHL-2_Ec catalytic subunit HyfG (see Fig. S1 in the supplemental material) (12). Other aspects of the T. guamensis hyf operon composition are highly similar to the hyf counterpart in E. coli but with significant variation in the organization of associated transcriptional regulators, protein maturation factors, and the surrounding genome context (Fig. 1A). For example, it contains an endoprotease gene, hyfK, whose gene product is predicted to be required for processing of the large subunit after completion of the metal cofactor insertion and is missing in the E. coli hyf operon (13). Instead, there is no formate transporter FocB encoded within the T. guamensis hyf operon. The E. coli FocB protein was presumed to participate in formate transport, but the main formate transporter in E. coli is FocA, located within the focA-pflB operon, encoding also the pyruvate formate lyase (14–16). The focA-pflB operon is also conserved in T. guamensis, and the FocA protein sequence is 92% identical to FocA in E. coli. Taken together, these features suggest an apparent functionality of the FHL-2_Tg complex.

The number and complement of the H₂-oxidizing Hyd vary significantly among enterobacteria. The E. coli Hyd-1 and Hyd-2 both catalyze H₂ oxidation and face the periplasmic side of the cytoplasmic membrane (5, 17). Although their respective catalytic subunits, HyaB and HybC, share 40% identity at the amino acid level, the architecture of the entire membrane complex differs significantly, conferring upon them unique enzymatic properties. The Hyd-1 complex has three different subunits and can function in the presence of oxygen (18). Hyd-2, on the other hand, combines four different subunits (HybOABC) and is highly catalytically active (18). Hyd-2 is also involved in the generation of a proton gradient during catalysis (19).

S. enterica has 3 H₂-oxidizing hydrogenases (catalytic subunits HyaB, HybC, and HydB in Fig. S1) and one FHL-1 complex (catalytic subunit HycE in Fig. S1), and the synthesis of these enzymes varies depending on its environmental situation (20). For example, in S. enterica and the pathogenic epsilonproteobacterium Helicobacter pylori, energy-conserving H₂ oxidation contributes to the pathogenicity of both bacteria, because anaerobic respiration enhances the survival of the pathogen in the host (21–23). Interestingly, T. guamensis encodes only one H₂-oxidizing hydrogenase, which is similar to Hyd-2 of E. coli (HybC in Fig. S1) (1).

The T. guamensis hydrogenase complement now poses the unique opportunity to study H₂ production by an organism that only has the FHL-2_Tg complex. Thus, we can gain considerable insight into the bioenergetics of this intriguing complex and we can determine what deficiencies in the FHL-2_Ec complex prevent it from being active under standard laboratory conditions.

RESULTS AND DISCUSSION

Growth conditions that favor Hyd-2_Tg activity.

T. guamensis can utilize cellobiose, the disaccharide derived from cellulose, for anaerobic growth, with a growth rate (μ) of 0.20 h⁻¹ (Fig. 2A). The growth rate with glucose was more than twice that of the cellobiose growth, with a μ of 0.57 h⁻¹. The growth rate of E. coli with glucose was 0.24 h⁻¹, which is in the same order of magnitude as the cellobiose growth rate of T. guamensis, while no growth of E. coli was observed with cellobiose. T. guamensis encodes a β-d-glucosidase similar to BglC from Bacillus subtilis, which is not present in E. coli, and a periplasmic β-d-glucosidase, BglX, whose function has not been determined in E. coli, possibly giving it the capacity to use cellobiose. Extracts from these cultures were applied to native PAGE and stained for Hyd activity, which revealed H₂ oxidation activity as staining of a red band. In this assay, E. coli showed a fast-migrating Hyd-1 activity band, two slower-migrating bands that can be assigned to Hyd-2, and one band belonging to the FHL-1_Ec complex (24) (Fig. 2B). Weaker bands can be observed near the top of the gel, which were previously identified as being due to weak H₂ oxidation activity catalyzed by formate dehydrogenases N and O (25).

FIG 2.

Growth and hydrogenase enzyme synthesis during cellobiose fermentation. (A) Cells of T. guamensis (black) and E. coli strain MG1655 (gray) were grown in triplicate in standing liquid cultures in M9 minimal medium with either 0.8% (wt/vol) glucose (circles) or cellobiose (squares) as the sole sources of carbon and energy. Growth was monitored in a Tecan plate reader. (B) Cells were harvested at the beginning of stationary phase, and 25 μg protein was separated on a 7.5% native-PAGE gel and subsequently stained for hydrogenase activity with benzyl viologen, 2,3,5-triphenyltetrazolium chloride, and H₂. The active hydrogenase complexes are labeled on the right of the gel based on the migration pattern of the E. coli enzymes. Tg, T. guamensis; Ec, E. coli.

The T. guamensis extracts show an intensively staining red band under both growth conditions, and this migrates at approximately the same position as the FHL-1_Ec complex. Furthermore, a faster-migrating band is weakly visible in glucose-grown cells but strongly increased in intensity in cellobiose-grown cells. Based on the homology to the E. coli Hyd proteins, the lower band likely represents Hyd-2. This apparent glucose repression of Hyd-2 is rather surprising, because synthesis of the E. coli Hyd-2 is not repressed during mixed-acid fermentation, as sufficient fumarate is produced internally that can be reduced to succinate with electrons derived from H₂ (26).

In order to unambiguously verify the identity of these bands, a Hyd-2-deficient strain was constructed that carried the insertion of a kanamycin resistance cassette within the hybC gene, encoding the catalytic subunit of Hyd-2. Strain TGH001 and its parental strain were grown in glucose, cellobiose, and glycerol-fumarate medium and extracts applied to a native-PAGE gel (Fig. 3A). This experiment verified that the faster-migrating band is indeed Hyd-2_Tg. It additionally confirmed that the activity of the enzyme is higher after growth with cellobiose and especially high in the presence of glycerol and fumarate. High activity for Hyd-2_Ec has been observed after growth in glycerol-fumarate medium, implicating a significant function in H₂ oxidation and energy conservation when formate is less abundant (26). Furthermore, the Hyd-2_Tg-coding genes were introduced into an E. coli strain (DLH04) that lacked the genes for the structural subunits of Hyd-2, hybOABC (Fig. 3B). In addition, this strain is devoid of Hyd-1 and FHL-1_Ec activity. When introduced, the plasmid pTg-hybOG resulted in an activity band that migrated slightly slower than that of the parental enzyme. This retardation is presumably caused by the addition of the histidine affinity tag on HybC and has been previously observed for other Hyd proteins in native PAGE (27). Initial attempts to isolate the Hyd-2 complex with this affinity tag from E. coli cells according to a procedure previously described (28) showed that only the tagged large-subunit HybC could be isolated but without any of the remaining structural subunits. The plasmid carries the enzyme-specific maturation enzymes but lacks the genes for the general maturation machinery of the [NiFe] cofactor, indicating that the E. coli maturation machinery recognizes the foreign HybC protein and introduces the [NiFe] cofactor. The induction of gene expression by isopropyl-β-d-thiogalactopyranoside (IPTG) resulted in the absence of the activity band, indicating that large amounts of protein have detrimental effects, possibly caused by an overburdening of the maturation machinery or the membrane-targeting apparatus.

FIG 3.

Identification of _TgHyd-2. (A to C) Cells of E. coli strain MC4100, T. guamensis, and TGH001 (T. guamensis hybC::kan) were grown anaerobically in M9 medium with either 0.8% (wt/vol) glucose, 0.4% (wt/vol) cellobiose, or 0.4% (vol/vol) glycerol and 15 mM fumarate (A) or TYEP with 0.4% (vol/vol) glycerol and 15 mM fumarate (B and C), as indicated. Panel B additionally shows E. coli strain DLH04 (MC4100 DE3 ΔhycA-I ΔhybO-C ΔhyaB) after complementation with the pTg-hybOG plasmid. + and – indicate addition and absence of 0.2 mM IPTG during growth, respectively. (C) Strain TGH001 (T. guamensis hybC::kan) complemented with plasmids pTg-hybOG and pEc-hybC. All bands represent Hyd-2 activities. Twenty-five-microgram portions of protein extracts were applied to a native PAGE gel and subsequently stained for Hyd activity. The migration pattern of the Hyd is indicated on the right.

Based on the findings described above, the TGH001 (T. guamensis hybC::kan) strain was used to test if the introduced mutation was the cause for the phenotype. Therefore, the growth conditions with glycerol and fumarate were chosen to favor optimal Hyd-2 synthesis, and the gel composition was slightly adjusted so that it revealed a second slower-migrating band for Hyd-2 in T. guamensis (Fig. 3C, lane 5), which was absent in TGH001 (Fig. 3C, lane 6), and even three bands belonging to Hyd-2 in E. coli (lane 1). These multiple bands represent various subcomplexes of Hyd-2 that differ in the presence or absence of the membrane anchor subunits HybA and HybB (29). A further unspecific faster-migrating band of Hyd-2 can be observed on the bottom of the gel after overexpression and might represent a form of degraded Hyd-2, similar to that observed for Hyd-1 when overexpressed (30). For the complementation, strain TGH001 was transformed either with plasmid pTg-hybOG or with pEc-hybC. Analysis on native PAGE showed that the main activity band of the pTg-hybOG construct returned in the complemented strain TGH001 (Fig. 3C, lane 4), but the migration pattern was slightly slower than that of the untagged complex in T. guamensis (Fig. 3C, lane 5) but identical to the activity band seen of the His-tagged version in E. coli strain DLH04 (Fig. 3C, lane 3). The E. coli hybC gene from plasmid pEc-hybC can also complement the lack of Hyd-2 activity in strain TGH001 (Fig. 3C, lane 7). Surprisingly, the activity band pattern of the chimeric complex in T. guamensis resembled that of E. coli Hyd-2 (Fig. 3C, lane 1), with the exception of the slower migration due to the fused Strep II-affinity tag on HybC. This finding would imply that the large subunit coordinates the assembly of the entire tetrameric complex, although in an alignment, the HybC from T. guamensis (TgHybC) and HybC from E. coli (EcHybC) have only 15 amino acid positions that are entirely different; those differences mostly locate on the solvent-exposed outer surface, except for positions 132 and 145, which face the respective other HybC protein. More research with purified chimeric complexes will be necessary to resolve this question, but currently, the affinity tags are not suitable for isolation of the entire Hyd-2 complex.

Hyd-2_Tg is an H₂-oxidizing Hyd.

The cultures from the different growth media were tested for their H₂ production, determined as H₂ accumulation in the headspace above the culture (Table 1). When T. guamensis cells were grown with glucose, they produced the greatest amount of H₂, with 48 μmol/optical density at 600 nm (OD₆₀₀)/ml. This is about 60% of that produced by an E. coli culture under the same conditions but in the same range as that of strain TGH001 (hybC::kan) (Table 1). The H₂ production from strain TGH001 indicates that Hyd-2 does not contribute to H₂ production under these conditions and is only responsible for a small fraction of reoxidation of the product. This is important because Hyd-2_Ec was previously shown to be able to produce H₂ from glycerol (31, 32), which represents a reaction driven by pmf in order to counteract the overreduction of the quinone pool (19). The amount of H₂ was reduced to 17% when T. guamensis was grown in the presence of cellobiose and was absent when the cells were grown with glycerol-fumarate. Here, the extent of the reoxidation of H₂ by Hyd-2 becomes evident, because strain TGH001 (hybC::kan) produced more than twice the amount of H₂ than the wild type grown with cellobiose and generated even greater amounts in glycerol-fumarate medium.

TABLE 1.

H₂ evolution from cultures

Culture addition	μmol H2 OD600−1 ml−1 for:
	T. guamensis	TGH001 (hybC::kan)	TGH001/pTg-hybOG	MC4100
Glucose	48.3 ± 9.1	51.5 ± 2.0	31.9 ± 2.3	82.8 ± 2.7
Cellobiose	8.2 ± 3.9	19.9 ± 5.4	<0.1	NAa
Glycerol-fumarate	<0.1	29.9 ± 2.3	<0.1	<0.1

^aNA, not available.

Formate is the substrate for the labile FHL-2_Tg complex.

Cell suspensions of T. guamensis and E. coli strain MC4100 were applied to a Clark-type electrode, and formate was tested to initiate H₂ production. The activity of the T. guamensis cells 3.21 ± 0.24 μmol min⁻¹ mg protein⁻¹) was in a range similar to that for E. coli (3.08 ± 0.79 μmol min⁻¹ mg protein⁻¹) (Table 2). The mutant TGH001 (hybC::kan) with the insertion that inactivated the Hyd-2 enzyme showed slightly reduced H₂ production, at 2.61 ± 0.24 μmol min⁻¹ mg protein⁻¹. This clearly demonstrated that the remaining activity was due to FHL-2_Tg.

TABLE 2.

Hydrogenase activities from FHL-2_Tg^a

Strain or bacterium	FHL activity (formate-dependent H2 production) (μmol min−1 mg−1)	Hydrogenase activity (H2-dependent BV reduction) (U mg−1) from:
		Whole cells	Crude extracts
MC4100	3.08 ± 0.79	0.12 ± 0.02	2.03 ± 0.54
T. guamensis	3.21 ± 0.24	0.06 ± 0.01	0.05 ± 0.01
TGH001 (hybC::kan)	2.61 ± 0.24	<0.01	<0.01
CPH022 (ΔhyfA-R)	<0.01	ND	<0.01
CPH022/pTg-hyf	3.30 ± 0.18	0.01 ± 0.01	0.04 ± 0.01
CPH023 (ΔhyfA-R ΔfdhF)	<0.01	ND	<0.01
CPH023/pTg-hyf	<0.01	ND	<0.01

^aCells were grown in the presence of 0.2% (wt/vol) l-arabinose. ND, not determined.

Experiments showing the optimum pH for activity revealed that when cells were grown at an initial pH of 5.5, the activity was up to 220% higher than when the medium pH was initially set to 6.5 or 7.5 (see Table S1 in the supplemental material). However, the assay conditions showed that of the three pHs tested, the highest activity was obtained when cells were assayed at pH 6.5. These data clearly show that the activity is most stable under slightly acidic pH, but overall enzyme activity is highest when the cells try to counteract acidification and formate accumulation during growth, a function that was originally proposed for FHL-1_Ec (33).

Using E. coli as a chassis, the cloned T. guamensis hyfA-K operon was introduced into E. coli strain CPH022, which is devoid of the entire operons encoding FHL-1_Ec and FHL-2_Ec. The T. guamensis hyfA-K operon showed H₂ production from the pBAD24 backbone, and this was entirely dependent on the presence of l-arabinose (Fig. S2). This plasmid restored FHL activity on the electrode to 3.30 ± 0.18 μmol min⁻¹mg protein⁻¹ to strain CPH022 (Table 2). Furthermore, deletion of the fdhF gene, coding for the FHL-1_Ec-associated formate dehydrogenase H, and the subsequent addition of pTg-hyf showed that the activity of the heterologous FHL-2_Tg complex was absent and thus identified FdhH as the electron input module for FHL-2_Tg in E. coli. The E. coli and T. guamensis formate dehydrogenases are both encoded at distinct locations from the operons encoding FHL, and recent work indicated that FdhH can form several protein complexes within the cell (34). The proteins share 95% amino acid identity between the two organisms, including the catalytic selenocysteine residue. This conservation is considerably higher than that of the remaining components of the complexes (Fig. 1). The E. coli FdhH can therefore substitute for the T. guamensis FdhH and provide electrons for H₂ production from formate.

The extracts from the heterologous expression of the FHL-2_Tg complex were further tested in a spectrophotometric assay, which measures the total hydrogenase activity of all Hyd proteins, including FHL-1_Ec, as H₂-dependent reduction of the artificial electron dye benzyl viologen (BV) (27). Performing the assay with whole cells allowed quantification of the periplasmically oriented Hyd enzymes, like Hyd-1 and Hyd-2 only, because oxidized BV does not diffuse across the intact membrane (29). This assay resulted in activities of 0.12 and 0.06 U mg protein⁻¹ for E. coli and T. guamensis, respectively (Table 2). Cell lysis increased the activity of the E. coli extract to 2.03 U mg protein⁻¹ due to the accessibility of the FHL-1_Ec complex. This increase was neither detectable in extracts of T. guamensis nor in CPH022 synthesizing FHL-2_Tg heterologously (Table 2). This result either implies that the FHL-2_Tg complex is unstable upon cell lysis or unable to perform the reverse H₂ oxidation reaction in the presence of large amounts of H₂.

The strain TGH002 carries a deletion of the entire hyfA-K operon and could only be stably obtained as a merodiploid in the presence of an in trans copy of the same operon. The regulation of the plasmid-carried operon was under the control of an arabinose-inducible promoter. Numerous attempts to generate the mutant without this extra copy failed. Similarly, anaerobic growth of TGH002 without arabinose was not possible. Small amounts of L-Ara (0.05% wt/vol) were already sufficient to induce H₂ production, but an increase of up to 2% (wt/vol) had no further significant effect on the H₂ production, showing that the amount of active FHL-2_Tg complex cannot be further increased (Fig. 4). This is similar to the observation of Hyd-2 activity, which cannot be increased when induced with IPTG. In contrast, the same plasmid showed that in E. coli strain CPH022, the absence of L-Ara still resulted in small amounts of H₂ (Fig. 4), rendering this construct unsuitable for further experiments in T. guamensis. In strain CPH022, the optimal l-Ara concentration for H₂ production was between 0.05 and 0.1% (wt/vol), with a further increase of l-Ara causing a decrease in H₂ accumulation. Similar to Hyd-2, overproduction left the protein mainly inactive.

FIG 4.

l-Arabinose-dependent H₂ production from FHL-2_Tg in TGH002 (circles) and CPH022 (squares). Cells were grown as liquid cultures in TGYEP medium (pH 6.5) in triplicate with the indicated amount of l-Ara in the growth medium. Headspace H₂ content was determined by gas chromatography, as described in Materials and Methods.

Conclusions.

We show in this study that the FHL-2_Tg of the enterobacterium T. guamensis functions as a highly efficient H₂-producing enzyme complex using formate as the substrate catalyzing the total reaction HCOO⁻ + H⁺ → H₂ + CO₂. The activity is similar to that of FHL-1_Ec, and it is dependent on the formate dehydrogenase H protein (FdhH) of E. coli. The numbers of subunits in the soluble domain between these two complexes are the same, but the difference is that five membrane subunits are present in the FHL-2_Tg complex, rather than the two in FHL-1_Ec. It appears from the lack of growth of the T. guamensis ΔhyfA-K mutant without concomitant coexpression of the operon in trans that the FHL-2_Tg complex is essential in T. guamensis. This could suggest that the extra membrane subunits HyfDEF are required for generation of a proton gradient in T. guamensis. Simultaneously, the pmf-building Hyd-2_Tg enzyme is less active with glucose because the FHL-2_Tg complex might provide enough energy under these conditions without the contribution of Hyd-2_Tg in energy conservation.

The Hyd proteins can be functionally interchanged between E. coli and T. guamensis, and this provides an excellent model system to investigate the proton-translocating potential of the FHL-2_Tg in E. coli, where it is not essential. Moreover, because our experiments show that T. guamensis is genetically tractable, this bacterium can be then used to reintroduce genetically modified FHL-2_Tg complexes to determine what makes FHL-2_Tg essential.

MATERIALS AND METHODS

Medium and growth conditions.

T. guamensis strain DSM16940 was obtained from the DSMZ in Germany. E. coli strain MC4100 was used for experiments throughout this study (35). The strain DLH04 (MC4100 DE3 ΔhycA-I ΔhybO-C ΔhyaB) has been described (D. Lubek, A. H. Simon, and C. Pinske, unpublished data) (Table 3). Strains of E. coli and T. guamensis were routinely grown in liquid LB medium (36) solidified with 1.5% (wt/vol) agar for plate growth. Antibiotics were added to a final concentration of 100 μg ml⁻¹ for ampicillin, 30 μg ml⁻¹ for chloramphenicol, and 50 μg ml⁻¹ for kanamycin. Unless otherwise stated, l-arabinose was added to a final concentration of 0.2% (wt/vol), which equates to 13.3 mM. For activity determination of Hyd, TGYEP medium (pH 6.5) was used, which includes 1% (wt/vol) tryptone, 0.5% (wt/vol) yeast extract, 0.1 M potassium buffer (pH 6.5), and 0.8% (wt/vol) glucose (37). For experiments with different carbon sources, M9 minimal medium was used. The concentrations of carbon sources were 0.4% (wt/vol) for cellobiose, 0.4% (wt/vol) for glycerol, and 15 mM for fumarate, or as otherwise indicated. IPTG was added to a final concentration of 0.2 mM.

TABLE 3.

Strains and plasmids used in the study

Strain or plasmid	Genotype or descriptiona	Reference or source
Strains
MC4100	E. coli parent	35
T. guamensis strain DSM16940	Parent	1
DLH04	MC4100 (DE3) ΔhycA-I ΔhybO-C ΔhyaB	D. Lubek, A. H. Simon, and C. Pinske, unpublished data
TGH001	T. guamensis hybC::kan	This study
TGH002	T. guamensis ΔhyfA-K/pBAD-hyf	This study
CP1170	MC4100 ΔhyaB ΔhybC ΔhycA-I	This study
CPH022	CP1170 ΔhyfA-R	This study
CPH023	CP1170 ΔhyfA-R ΔfdhF	This study
Plasmids
pMAK705	Temperature-sensitive replicon, Cmr	39
pBAD24	Cloning vector with araP, Ampr	38
pKD4	Contains kan cassette, Kanr	40
pKD46	Contains λ red genes γ, β, and exo, Ampr	40
pACYC-Duet1	Medium-copy-number vector, T7 promoter, Cmr	Novagen (Merck, Germany)
pTg-hyf	Derived from pBAD24, TghyfA-K Ampr	This study
pTg-hybOG	pACYC-Duet1; TghybO-G His-HybC Cmr	This study
pMAK-hybC::kan	Derived from pMAK705, TghybC::kan Cmr	This study
pMAK-Dhyf	Derived from pMAK705, Tghyf upstream and downstream regions, Cmr	This study
pEc-hybC	pASK-hybC; pASK-IBA5; hybC Ampr	27

^aCm^r, chloramphenicol resistance; Amp^r, ampicillin resistance; Kan^r, kanamycin resistance.

Strain and plasmid construction.

The T. guamensis hyfA-K operon comprising a total of 12 genes was cloned as three PCR fragments that were obtained from T. guamensis genomic DNA amplification and had sizes of 5,058 bp, 3,985 bp, and 4,048 bp. These were then assembled within PCR-amplified pBAD24 (38) with the NEBuilder kit, according to the manufacturer’s instructions (NEB), which generated plasmid pTg-hyf. The oligonucleotides used for this construction are listed as numbers 1 to 8 in Table 4.

TABLE 4.

Oligonucleotides used in this study

Oligonucleotide no.	Primer	Sequence (5′→3′)a	Purpose
1	pBAD24_fwd	GTACCCGGGGATCCTCTAGAG	pTg-hyf
2	pBAD24_rev	GGTGAATTCCTCCTGCTAGC
3	hyf1_5058_fwd	gtttttttgggctagcaggaggaattcaccATGAACCGCTTTGTGATTGC
4	hyf1_5058_rev	ttccggtcatcatcgACTCCCTTAGTTAAGCCAGAG
5	hyf2_3985_fwd	cttaactaagggagtCGATGATGACCGGAACCTTAATTG
6	hyf2_3985_rev	gtttcagcatcgttcAATCCTTACTTAAGCGGTGAC
7	hyf3_4048_fwd	gcttaagtaaggattGAACGATGCTGAAACTGTTTAAAAC
8	hyf3_4048_rev	cctgcaggtcgacagaggatccccgggtacTTAAAGCGGAAGAAATCCGC
9	pACYC184_fwd	AGGATCCGAATTCGAGCTCGG	pTg-hybOG
10	pACYC184_rev	GGTATATCTCCTTATTAAAGTTAAAC
11	HybOAB_fwd	taattttgtttaactttaataaggagatataccGTTCCTGGTTGCTCACCATC
12	HybOAB_rev	tgatggtgatggtgatgGCTCATGCTTTGCTCGCCTC
13	HybCDEFG_fwd	gagcaaagcatgagccatcaccatcaccatcaccatcacCAACGTATAACTATTGATCCGGTAACC
14	HybCDEFG_rev	gcaggcgcgccgagctcgaattcggatcctTTAGTTCGTTTGCACCGGCTC
15	DhybC up_BamHI	GCG GGATCC GTGGCAATTGAGGTATGTGC	pMAK-hybC::kan
16	DhybC down_BamHI	GCG GGATCC CCGGCAGCACGTACCGTTGC
17	pKD4_FW_SalI	GCG GTCGAG GCGATTGTGTAGGCTG
18	pKD4_RW_SalI	GCG GTCGAC CCATGGTCCATATGAATAT
19	hyf_KO_1_KpnI	GCG GGTACC CAGCACTCGGCCTGCTGCTC	pMAK-Dhyf
20	hyf_KO_2_XbaI	GCG TCTAGA GCAATCACAAAGCGGTTCAT
21	hyf_KO_3_PstI	GCG CTGCAG CGGATTTCTTCCGCTTTAAG
22	hyf_KO_4_HindIII	GCG AAGCTT GCGTCAGCAGAAATGAATAG
23	Ec_hyf_5′KO	ATGAACCGCTTTGTGGTGGCCGAACCACTGTGGTGTACAG ccatggtccatatgaatatc	ΔhyfA-R E.c.
24	Ec_hyf_3′KO	ACAACACCTCGCGAACCGAGATCCCCAGCCGCTGCATTCG gcgattgtgtaggctggagc
25	Ec-hyf_FW1_NcoI	GCG CCATGG TTACTTTGCTCCATTCCGTGC	PCR control of ΔhyfA-R
26	Ec-hyf_RW2_XhoI	GCG CTCGAG CGCCGGGCTGATTTTTCAGG

^aOverlapping regions of the oligonucleotides used in the NEBuilder assembly are indicated by lowercase letters.

Similarly, the T. guamensis hybO-G operon including a 650-bp upstream region was amplified as two fragments with oligonucleotides 11 and 12, and 13 and 14 (Table 4), which introduced a sequence coding for 8 His residues N terminally on HybC. The fragments were assembled on PCR-amplified (oligonucleotides 9 and 10) pACYC-Duet1 vector by the NEBuilder method (NEB), resulting in plasmid pTg-hybOG, which puts the operon under the control of its native promoter and additionally under the T7 promoter of the pACYC-Duet1 vector.

Construction of T. guamensis mutants was done according to the pMAK705 method, as described before (39). For the hybC::kan insertion mutant, the hybC gene was amplified from chromosomal DNA using oligonucleotides 15 and 16 (Table 4) and cloned into pJET1.2, and the intrinsic SalI site was then used to insert the kanamycin cassette from vector pKD4 (40), which had been previously amplified using oligonucleotides 17 and 18 (Table 4). The construct was then moved as a BamHI/BamHI fragment into pMAK705 and the resulting pMAK-hybC::kan used for recombination by selecting for the kanamycin resistance during the final step. The resulting strain was named TGH001 (T. guamensis hybC::kan).

To introduce the hyfA-K deletion into the T. guamensis genome, the upstream and downstream 500-bp regions of the operon were amplified and cloned successively into pMAK705 as KpnI/XbaI and PstI/HindIII fragments using oligonucleotides 19 and 20, and 21 and 22 (Table 4), resulting in pMAK-Dhyf. This plasmid was then used for genomic recombination according to the procedure of Hamilton et al. (39). However, mutants with the hyfA-K deletion on the chromosome were only obtained in the presence of the pTg-hyf plasmid, which was additionally selected for with ampicillin and 0.2% (wt/vol) l-arabinose during all steps of the recombination procedure. The resulting strain was named TGH002 (T. guamensis ΔhyfA-K/pBAD-hyf).

Genomic mutants were checked by colony PCR.

The deletion of hyfA-R operon in E. coli left focB intact. The strain CP1170 (ΔhyaB ΔhybC ΔhycA-I) was transformed with λ red carrying plasmid pKD46, grown according to Datsenko and Wanner (40), and a PCR fragment (oligonucleotides 23 and 24 in Table 4), including the chloramphenicol resistance (Cm^r) from pKD3 and 40 nucleotides flanking the deletion region. The authenticity of the resulting Cm^r clones was verified with oligonucleotides 25 and 26 (Table 4).

Enzyme activities.

Native PAGE was performed in 7.5% (wt/vol) polyacrylamide gels as described and stained for hydrogenase enzyme activity with 0.5 mM benzyl viologen and 1 mM 2,3,5-triphenyl tetrazolium chloride under a 2% (vol/vol) H₂ atmosphere (24).

Spectrophotometric hydrogenase enzyme assays were performed as described previously (41) with crude extracts that were obtained after sonication (20 W, 2 min, 0.5-s pulses) or with whole cells. Cuvettes were filled with 800 μl of 4 mM benzyl viologen in 50 mM morpholinopropanesulfonic acid (MOPS) buffer (pH 7.0), and the remaining gas phase was flushed with H₂. The reduction of benzyl viologen was monitored at 600 nm, and an ε_M of 7,400 M⁻¹cm⁻¹ was assumed. One unit of activity represents the conversion of 1 μmol substrate per min.

H₂ headspace content was determined with a Shimadzu-2010 gas chromatography system equipped with a ShinCarbon micropacked ST80/100 column, as described previously (34). The amount of H₂ was normalized to the OD₆₀₀ of the culture and the used culture volume.

The continuous H₂ production from whole cells was monitored using a Clark-type electrode (Hansatech Instruments) modified to detect H₂ and filled with 2 ml of degassed 50 mM MOPS buffer (pH 7.0). The reaction was started by the addition of formate to a final concentration of 50 mM. Calibration was done with H₂-saturated buffer according to reference 42.

Protein concentration was determined according to the method of Lowry (43).