Metabolic engineering to enhance bacterial hydrogen production

Hydrogen fuel is renewable, efficient and clean, and fermentative bacteria hold great promise for its generation. Here we use the isogenic Escherichia coli K‐12 KEIO library to rapidly construct multiple, precise deletions in the E. coli genome to direct the metabolic flux towards hydrogen production. Escherichia coli has three active hydrogenases, and the genes involved in the regulation of the formate hydrogen lyase (FHL) system for synthesizing hydrogen from formate via hydrogenase 3 were also manipulated to enhance hydrogen production. Specifically, we altered regulation of FHL by controlling the regulators HycA and FhlA, removed hydrogen consumption by hydrogenases 1 and 2 via the hyaB and hybC mutations, and re‐directed formate metabolism using the fdnG, fdoG, narG, focA, fnr and focB mutations. The result was a 141‐fold increase in hydrogen production from formate to create a bacterium (BW25113 hyaB hybC hycA fdoG/pCA24N‐FhlA) that produces the largest amount of hydrogen to date and one that achieves the theoretical yield for hydrogen from formate. In addition, the hydrogen yield from glucose was increased by 50%, and there was threefold higher hydrogen production from glucose with this strain.     

Enhanced hydrogen production from glucose by metabolically engineered <Emphasis Type="Italic">Escherichia coli</Emphasis>

To utilize fermentative bacteria for producing the alternative fuel hydrogen, we performed successive rounds of P1 transduction from the Keio Escherichia coli K-12 library to introduce multiple, stable mutations into a single bacterium to direct the metabolic flux toward hydrogen production. E. coli cells convert glucose to various organic acids (such as succinate, pyruvate, lactate, formate, and acetate) to synthesize energy and hydrogen from formate by the formate hydrogen-lyase (FHL) system that consists of hydrogenase 3 and formate dehydrogenase-H. We altered the regulation of FHL by inactivating the repressor encoded by hycA and by overexpressing the activator encoded by fhlA, removed hydrogen uptake activity by deleting hyaB (hydrogenase 1) and hybC (hydrogenase 2), redirected glucose metabolism to formate by using the fdnG, fdoG, narG, focA, focB, poxB, and aceE mutations, and inactivated the succinate and lactate synthesis pathways by deleting frdC and ldhA, respectively. The best of the metabolically engineered strains, BW25113 hyaB hybC hycA fdoG frdC ldhA aceE, increased hydrogen production 4.6-fold from glucose and increased the hydrogen yield twofold from 0.65 to 1.3 mol H₂/mol glucose (maximum, 2 mol H₂/mol glucose).     

Protein engineering of hydrogenase 3 to enhance hydrogen production

The large subunit (HycE, 569 amino acids) of Escherichia coli hydrogenase 3 produces hydrogen from formate via its Ni–Fe-binding site. In this paper, we engineered HycE for enhanced hydrogen production by an error-prone polymerase chain reaction (epPCR) using a host that lacked hydrogenase activity via the hyaB hybC hycE mutations. Seven enhanced HycE variants were obtained with a novel chemochromic membrane screen that directly detected hydrogen from individual colonies. The best epPCR variant contained eight mutations (S2T, Y50F, I171T, A291V, T366S, V433L, M444I, and L523Q) and had 17-fold higher hydrogen-producing activity than wild-type HycE. In addition, this variant had eightfold higher hydrogen yield from formate compared to wild-type HycE. Deoxyribonucleic acid shuffling using the three most-active HycE variants created a variant that has 23-fold higher hydrogen production and ninefold higher yield on formate due to a 74-amino acid carboxy-terminal truncation. Saturation mutagenesis at T366 of HycE also led to increased hydrogen production via a truncation at this position; hence, 204 amino acids at the carboxy terminus may be deleted to increase hydrogen production by 30-fold. This is the first random protein engineering of a hydrogenase.     

The Involvement of Hydrogenases 1 and 2 in the Hydrogen-Dependent Nitrate Respiration of Escherichia coli

The study of Escherichia coli mutants synthesizing either hydrogenase 1 (HDK203) or hydrogenase 2 (HDK103) showed that the nitrate-dependent uptake of hydrogen by E. coli cells can be accomplished through the action of either of these hydrogenases. The capability of the cells for hydrogen-dependent nitrate respiration was found to depend on the growth conditions. E. coli cells grown anaerobically without nitrate in the presence of glucose were potentially capable of nitrate-dependent hydrogen consumption. The cells grown anaerobically in the presence of nitrate exhibited a much lower capability for nitrate-dependent hydrogen consumption. The inhibitory effect of nitrate on this capability of bacterial cells was either weak (the mutant HDK203) or almost absent (the mutant HDK103) when the cells were grown in the presence of peptone and hydrogen. Hydrogen stimulated the growth of the wild-type strain and the mutant HDK103 (but not the mutant HDK203) cultivated in the medium with nitrate and peptone. These data suggest that hydrogenase 2 is much more active in catalyzing nitrate-dependent hydrogen consumption than hydrogenase 1.     

Hydrogen as an energy source for the human pathogen <Emphasis Type="Italic">Bilophila wadsworthia</Emphasis>

The gram-negative anaerobic gut bacterium Bilophila wadsworthia is the third most common isolate in perforated and gangrenous appendicitis, being also found in a variety of other infections. This organism performs a unique kind of anaerobic respiration in which taurine, a major organic solute in mammals, is used as a source of sulphite that serves as terminal acceptor for the electron transport chain. We show here that molecular hydrogen, one of the major products of fermentative bacteria in the colon, is an excellent growth substrate for B. wadsworthia. We have quantified the enzymatic activities associated with the oxidation of H₂, formate and pyruvate for cells obtained in different growth conditions. The cell extracts present high levels of hydrogenase activity, and up to five different hydrogenases can be expressed by this organism. One of the hydrogenases appears to be constitutive, whereas the others show differential expression in different growth conditions. Two of the hydrogenases are soluble and are recognised by antibodies against a [FeFe] hydrogenase of a sulphate reducing bacterium. One of these hydrogenases is specifically induced during fermentative growth on pyruvate. Another two hydrogenases are membrane-bound and show increased expression in cells grown with hydrogen. Further work should be carried out to reveal whether oxidation of hydrogen contributes to the virulence of B. wadsworthia.     

Differential expression of hydrogenase isoenzymes in Escherichia coli K-12: evidence for a third isoenzyme.

The cellular contents of the nickel-containing, membrane-bound hydrogenase isoenzymes 1 and 2 (hydrogenases 1 and 2) were analyzed by crossed immunoelectrophoresis. Their expression was differentially influenced by nutritional and genetic factors. Hydrogenase 2 content was enhanced after growth with either hydrogen and fumarate or glycerol and fumarate and correlated reasonably with cellular hydrogen uptake capacity. Hydrogenase 1 content was negligible under the above conditions but was enhanced by exogenous formate. Its expression was greatly reduced in a pfl mutant, which is unable to synthesise formate, but was restored to normal levels when the growth medium included formate. A mutation in the anaerobic regulatory gene, fnr, led to low overall hydrogenase activity and greatly reduced levels of both isoenzymes and abolished the formate enhancement of hydrogenase 1 content. Formate hydrogenlyase activity was similarly reduced in the fnr strain but, in contrast, was restored, as was overall hydrogenase activity, to normal levels by growth in the presence of formate. Low H2 uptake activity was found for the fnr strain under all growth conditions examined. Hydrogenase 1 content, therefore, does not correlate with formate hydrogenlyase activity and its role is unclear. A third hydrogenase isoenzyme, immunologically distinct from hydrogenases 1 and 2, whose expression is enhanced by formate, is present and forms part of the formate hydrogenlyase. We suggest that the effect of the fnr gene product on formate hydrogenlyase expression is mediated via internal formate.     

Induction and regulation of ribulose bisphosphate carboxylase activity in Rhizobium japonicum during formate-dependent growth

Rhizobium japonicum CJ1 was capable of growing using formate as the sole source of carbon and energy. During aerobic growth on formate a cytoplasmic NAD⁺-dependent formate dehydrogenase and ribulose bisphosphate carboxylase activity was demonstrated in cell-free extracts, but hydrogenase enzyme activity could not be detected. Under microaerobic growth conditions either formate or hydrogen metabolism could separately or together support ribulose bisphosphate carboxylase-dependent CO₂ fixation. A number of R. japonicum strains defective in hydrogen uptake activity were shown to metabolise formate and induce ribulose bisphosphate carboxylase activity. The induction and regulation of ribulose bisphosphate carboxylase is discussed.Abbreviations hup hydrogen uptake - MOPS 3-(N-morpholino)-propanesulphonate - TSA tryptone soya agar - RuBP ribulose 1,5-bisphosphate - FDH formate dehydrogenase     

Energy transformation coupled to formate oxidation during anaerobic fermentation

A correlation between the rate of ATP synthesis by F₀F₁ ATP synthase and formate oxidation by formate hydrogen lyase (FHL) has been found in inside-out membrane vesicles of the Escherichia coli mutant JW 136 (Δhya/Δhyb) with double deletions of hydrogenases 1 and 2, grown anaerobically on glucose in the absence of external electron acceptors at pH 6.5. ATP synthesis was suppressed by the H⁺-ATPase inhibitors N,N′-dicyclohexylcarbodiimide, sodium azide, and the uncoupler carbonyl cyanide m-chlorophenylhydrazone. Copper ions inhibited formate-dependent hydrogenase and ATP-synthase activities but did not affect the ATPase activity of the vesicles. The maximal rate of ATP synthesis (0.83 μmol/min per mg protein) was determined at simultaneous application of sodium formate, ADP, and inorganic phosphate, and was stimulated by K⁺ ions. The results confirm the assumption of a dual role of hydrogenase 3, the formate hydrogen lyase subunit that can couple the reduction of protons to H₂ and their translocation through membrane with chemiosmotic synthesis of ATP.     

Dissecting the roles of Escherichia coli hydrogenases in biohydrogen production.

Escherichia coli can perform at least two modes of anaerobic hydrogen metabolism and expresses at least two types of hydrogenase activity. Respiratory hydrogen oxidation is catalysed by two 'uptake' hydrogenase isoenzymes, hydrogenase -1 and -2 (Hyd-1 and -2), and fermentative hydrogen production is catalysed by Hyd-3. Harnessing and enhancing the metabolic capability of E. coli to perform anaerobic mixed-acid fermentation is therefore an attractive approach for bio-hydrogen production from sugars. In this work, the effects of genetic modification of the genes encoding the uptake hydrogenases, as well as the importance of preculture conditions, on hydrogen production and fermentation balance were examined. In suspensions of resting cells pregrown aerobically with formate, deletions in Hyd-3 abolished hydrogen production, whereas the deletion of both uptake hydrogenases improved hydrogen production by 37% over the parent strain. Under fermentative conditions, respiratory H2 uptake activity was absent in strains lacking Hyd-2. The effect of a deletion in hycA on H2 production was found to be dependent upon environmental conditions, but H2 uptake was not significantly affected by this mutation.     

Location, catalytic activity, and subunit composition of NAD-reducing hydrogenases of some Alcaligenes strains and Rhodococcus opacus MR22

Six new strains of Alcaligenes enriched for and isolated as nickel-resistant bacteria resemble Alcaligenes eutrophus H16 and contain both an NAD-reducing, tetrameric soluble hydrogenase and a membrane-bound hydrogenase. None of the soluble hydrogenases share with the Rhodococcus opacus MR11 enzyme tetramer the property of being cleaved easily into two dimeric moieties [a hydrogenase (βδ) and an NADH:acceptor oxidoreductase (αγ)], in the absence of nickel or at low ionic strength. The soluble hydrogenase of the newly isolated strain MR22 of R. opacus equalled that of strain MR11. The absence of a membrane-bound hydrogenase in Alcaligenes denitrificans strain 4a-2 and in Alcaligenes ruhlandii was confirmed. Received: 14 May 1996 / Accepted: 7 November 1996     

Transcriptional control of hydrogen production during mixed carbon fermentation by hydrogenases 4 (hyf) and 3 (hyc) in Escherichia coli

Escherichia coli molecular hydrogen (H(2)) production was studied during mixed carbon (glucose and glycerol) fermentation at pH 6.5. Wild type cells in the assays supplemented with glucose produced H(2) at ~2 fold lower level than cells grown on glucose only. When compared to the wild type, H(2) production in the assays added with glucose was decreased by ~2 fold in fhlA, hyfG and double fhlA hyfG mutants and by ~1.5 fold in hyaB, hybC, and double hyaB hybC mutants. However, in the assays with glycerol, no measurable H(2) production was detected. Taken together, these results suggest that during mixed carbon fermentation, H(2) could be produced with low efficiency via Hyd-3 and Hyd-4. This is a novel finding for Hyd-4 activity at pH 6.5. The insignificant decrease of H(2) production in the strains with defects in Hyd-1 and Hyd-2 was probably due to an interaction between the Hyd enzymes and their organization in the bacterial membrane. In the glucose assays, H(2) production in the wild type cells was inhibited ~2 fold by 0.3mM N,N'-dicyclohexylcarbodiimide (DCCD), an inhibitor of the F(0)F(1)-ATPase. This inhibition was the same for fhlA and hyfG fhlA mutants but not hyaB, hybC, hyfG or hyaB hybC mutants. The results indicate that the FhlA protein coded by the fhlA gene might interact with the F(0)F(1)-ATPase. We propose that this interaction is mediated by mixed carbon fermentation.     

Physiological characteristics and growth behavior of single and double hydrogenase mutants of Desulfovibrio fructosovorans

The presence of one periplasmic [NiFe] hydrogenase, one periplasmic [Fe] hydrogenase, and one cytoplasmic NADP-reducing hydrogenase has been previously established in Desulfovibrio fructosovorans. In the present work, marker-exchange mutagenesis was performed to determine the function of the tetrameric NADP-reducing hydrogenase encoded by the hndA, B, C, and D genes. The mutations performed were not lethal to the cells, although the H₂-dependent NADP reduction was completely abolished. The double-mutated DM4 (ΔhynABC, ΔhndD) strain was still able to grow on hydrogen plus sulfate as the sole energy source. The growth may have occurred under these culture conditions because of the presence of the remaining [Fe] hydrogenase. The cells grew differently on various substrates depending on whether fructose, lactate, or pyruvate was used in the presence of sulfate. The (hnd mutant growth rates were 25–70% lower than those of the wild-type strain, although the molar growth yield remained unchanged. By contrast, mutants devoid of both [NiFe] hydrogenase and NADP-reducing hydrogenase had 24-38% lower growth yields and showed a corresponding drop in the growth rates. We concluded that each of the three hydrogenases may contribute to the energy supply in D. fructosovorans and that the loss of one enzyme might be compensated for by another. However, the loss of two hydrogenases affected the phosphorylation accompanying the metabolism of fructose, lactate, and pyruvate. Received: 17 September 1996 / Accepted: 5 November 1996     

Purification and Characterization of Membrane-Associated Hydrogenase from the Deep-Sea Epsilonproteobacterium Hydrogenimonas thermophila

Membrane-associated hydrogenase was purified from the chemolithoautotrophic epsilonproteobacterium Hydrogenimonas thermophila at 152-fold purity. The hydrogenase was found to be localized in the periplasmic space, and was easily solubilized with 0.1% Triton X-100 treatment. Hydrogen oxidation activity was 1,365 μmol H₂/min/mg of protein at 80 °C at pH 9.0, with phenazine methosulphate as the electron acceptor. Hydrogen production activity was 900 μmol H₂/min/mg of protein at 80 °C and pH 6.0, with reduced methyl viologen as the electron donor. The hydrogenase from this organism showed higher oxygen tolerance than those from other microorganisms showing hydrogen oxidation activity. The structural genes of this hydrogenase, which contains N-terminal amino acid sequences from both small and large subunits of purified hydrogenase, were successfully elucidated. The hydrogenase from H. thermophila was found to be phylogenetically related with H₂ uptake hydrogenases from pathogenic Epsilonproteobacteria.     

Proteome Analysis of Enriched Heterocysts from Two Hydrogenase Mutants from Anabaena sp. PCC 7120

Cyanobacteria are oxygenic photosynthetic prokaryotes and play a crucial role in the Earth's carbon and nitrogen cycles. The photoautotrophic cyanobacterium Anabaena sp. PCC 7120 has the ability to fix atmospheric nitrogen in heterocysts and produce hydrogen as a byproduct through a nitrogenase. In order to improve hydrogen production, mutants from Anabaena sp. PCC 7120 are constructed by inactivation of the uptake hydrogenase (ΔhupL) and the bidirectional hydrogenase (ΔhoxH) in previous studies. Here the proteomic differences of enriched heterocysts between these mutants cultured in N₂‐fixing conditions are investigated. Using a label‐free quantitative proteomics approach, a total of 2728 proteins are identified and it is found that 79 proteins are differentially expressed in the ΔhupL and 117 proteins in the ΔhoxH variant. The results provide for the first time comprehensive information on proteome regulation of the uptake hydrogenase and the bidirectional hydrogenase, as well as systematic data on the hydrogen related metabolism in Anabaena sp. PCC 7120.     

Biodiversity of Hydrogenases in Frankia

Eighteen Frankia strains originally isolated from nine different host plants were used to study the biodiversity of hydrogenase in Frankia. In the physiological analysis, the activities of uptake hydrogenase and bidirectional hydrogenase were performed by monitoring the oxidation of hydrogen after supplying the cells with 1% hydrogen and the evolution of hydrogen using methyl viologen as an electron donor, respectively. These analyses were supported with a study of the immunological relationship between Frankia hydrogenase and other different known hydrogenases from other microorganisms. Uptake hydrogenase activity was recorded from all the Frankia strains investigated. A methyl-viologen-mediated hydrogen evolution was recorded from only four Frankia strains irrespective of the source of Frankia. From the immunological and physiological studies, we here report that there are at least three types of hydrogenases in Frankia: Ni-Fe uptake hydrogenase, hydrogen-evolving hydrogenase, and [Fe]-hydrogenase. An immunogold localization study, by cryosection technique, of the effect of nickel on the intercellular distribution of hydrogenase proteins in Frankia indicated that nickel affects the transfer of hydrogenase proteins into the membrane.     

Measuring the pH dependence of hydrogenase activities

The pH dependences of activities of homogenous hydrogenases of Thiocapsa roseopersicina and Desulfomicrobium baculatum in the reaction of hydrogen uptake in solution in the presence of benzyl viologen and the pH dependences of catalytic currents of hydrogen oxidation by electrodes on which these hydrogenases were immobilized were compared. Maximal activities of the hydrogenases from T. roseopersicina and D. baculatum in the reaction hydrogen uptake in solution were observed at pH 9.5 and 8.5, respectively. However, the steady-state current caused by catalytic uptake of hydrogen was maximal for the T. roseopersicina hydrogenase-containing electrode at pH 5.5-6.5 under overvoltage of 30-60 mV, whereas for electrodes with D. baculatum hydrogenase it was maximal at pH 6.0-6.5. Analysis of these data suggests that pH-dependent changes in the hydrogenase activities in solution during hydrogen uptake are due not only to the effect of proton concentration on the enzyme conformation or protonation of certain groups of the enzyme active center, but they are rather indicative of changes in free energy of the reaction accompanying changes in pH.     

Characterization of a HoxEFUYH type of [NiFe] hydrogenase from <Emphasis Type="Italic">Allochromatium vinosum</Emphasis> and some EPR and IR properties of the hydrogenase module

A soluble hydrogenase from Allochromatium vinosum was purified. It consisted of a large (M _r = 52 kDa) and a small (M _r = 23 kDa) subunit. The genes encoding for both subunits were identified. They belong to an open reading frame where they are preceded by three more genes. A DNA fragment containing all five genes was cloned and sequenced. The deduced amino acid sequences of the products characterized the complex as a member of the HoxEFUYH type of [NiFe] hydrogenases. Detailed sequence analyses revealed binding sites for eight Fe–S clusters, three [2Fe–2S] clusters and five [4Fe–4S] clusters, six of which are also present in homologous subunits of [FeFe] hydrogenases and NADH:ubiquione oxidoreductases (complex I). This makes the HoxEFUYH type of hydrogenases the one that is evolutionary closest to complex I. The relative positions of six of the potential Fe–S clusters are predicted on the basis of the X-ray structures of the Clostridium pasteurianum [FeFe] hydrogenase I and the hydrophilic domain of complex I from Thermus thermophilus. Although the HoxF subunit contains binding sites for flavin mononucleotide and NAD(H), cell-free extracts of A. vinosum did not catalyse a H₂-dependent reduction of NAD⁺. Only the hydrogenase module (HoxYH) could be purified. Its electron paramagnetic resonance (EPR) and IR spectral properties showed the presence of a Ni–Fe active site and a [4Fe–4S] cluster. Its activity was sensitive to carbon monoxide. No EPR signals from a light-sensitive Ni_a–C* state could be observed. This study presents the first IR spectroscopic data on the HoxYH module of a HoxEFUYH type of [NiFe] hydrogenase.     

Compartmentalisation of [FeFe]‐hydrogenase maturation in Chlamydomonas reinhardtii

Molecular hydrogen (H₂) can be produced in green microalgae by [FeFe]‐hydrogenases as a direct product of photosynthesis. The Chlamydomonas reinhardtii hydrogenase HYDA1 contains a catalytic site comprising a classic [4Fe4S] cluster linked to a unique 2Fe sub‐cluster. From in vitro studies it appears that the [4Fe4S] cluster is incorporated first by the housekeeping FeS cluster assembly machinery, followed by the 2Fe sub‐cluster, whose biosynthesis requires the specific maturases HYDEF and HYDG. To investigate the maturation process in vivo, we expressed HYDA1 from the C. reinhardtii chloroplast and nuclear genomes (with and without a chloroplast transit peptide) in a hydrogenase‐deficient mutant strain, and examined the cellular enzymatic hydrogenase activity, as well as in vivo H₂ production. The transformants expressing HYDA1 from the chloroplast genome displayed levels of H₂ production comparable to the wild type, as did the transformants expressing full‐length HYDA1 from the nuclear genome. In contrast, cells equipped with cytoplasm‐targeted HYDA1 produced inactive enzyme, which could only be activated in vitro after reconstitution of the [4Fe4S] cluster. This indicates that the HYDA1 FeS cluster can only be built by the chloroplastic FeS cluster assembly machinery. Further, the expression of a bacterial hydrogenase gene, CPI, from the C. reinhardtii chloroplast genome resulted in H₂‐producing strains, demonstrating that a hydrogenase with a very different structure can fulfil the role of HYDA1 in vivo and that overexpression of foreign hydrogenases in C. reinhardtii is possible. All chloroplast transformants were stable and no toxic effects were seen from HYDA1 or CPI expression.