Energy Conservation Model Based on Genomic and Experimental Analyses of a Carbon Monoxide-Utilizing, Butyrate-Forming Acetogen, Eubacterium limosum KIST612

Abstract

Eubacterium limosum KIST612 is one of the few acetogens that can produce butyrate from carbon monoxide. We have used a genome-guided analysis to delineate the path of butyrate formation, the enzymes involved, and the potential coupling to ATP synthesis. Oxidation of CO is catalyzed by the acetyl-coenzyme A (CoA) synthase/CO dehydrogenase and coupled to the reduction of ferredoxin. Oxidation of reduced ferredoxin is catalyzed by the Rnf complex and Na⁺ dependent. Consistent with the finding of a Na⁺-dependent Rnf complex is the presence of a conserved Na⁺-binding motif in the c subunit of the ATP synthase. Butyrate formation is from acetyl-CoA via acetoacetyl-CoA, hydroxybutyryl-CoA, crotonyl-CoA, and butyryl-CoA and is consistent with the finding of a gene cluster that encodes the enzymes for this pathway. The activity of the butyryl-CoA dehydrogenase was demonstrated. Reduction of crotonyl-CoA to butyryl-CoA with NADH as the reductant was coupled to reduction of ferredoxin. We postulate that the butyryl-CoA dehydrogenase uses flavin-based electron bifurcation to reduce ferredoxin, which is consistent with the finding of etfA and etfB genes next to it. The overall ATP yield was calculated and is significantly higher than the one obtained with H₂ + CO₂. The energetic benefit may be one reason that butyrate is formed only from CO but not from H₂ + CO₂.

INTRODUCTION

Acetogenic bacteria are a phylogenetically diverse group of strictly anaerobic bacteria able to reduce two molecules of CO₂ to acetate by the Wood-Ljungdahl pathway (WLP) (1–4). Electrons may derive from molecular hydrogen (autotrophic growth), from carbon monoxide, or from organic donors (heterotrophic growth) such as hexoses, pentoses, formate, lactate, alcohols, or methyl group donors (1). Not only does the WLP provide the cell with organic material for biomass formation, but it is also coupled to energy conservation for ATP supply by a chemiosmotic mechanism (2, 5). Every acetogen examined to date uses reduced ferredoxin (Fd) as the electron donor for an ion-translocating membrane protein complex, and acetogens can have either an Fd:NAD⁺ oxidoreductase (Rnf) or an Fd:H⁺ oxidoreductase (Ech) complex for generation of an ion motive force (5). In both cases, the ion gradient can be either an H⁺ or an Na⁺ gradient. The electrochemical ion gradient thus established is then used by a membrane bound, H⁺- or Na⁺-translocating F₁F_o ATP synthase (2).

Acetate production from CO₂ proceeds via formate that is converted to formyl-tetrahydrofolate (THF) in an ATP-consuming reaction (6). Water is split off from formyl-THF to yield methenyl-THF, which is reduced via methylene-THF to methyl-THF. The latter is condensed with CO (derived from another molecule of CO₂) and coenzyme A (CoA) to acetyl-CoA, which is the starting molecule for biosynthetic reactions (4, 7, 8). Acetyl-CoA is also the precursor of the end product, acetate, that is produced by the enzymes acetyltransferase and acetate kinase. ATP production in the acetate kinase reaction is of special importance for the WLP since it regains the ATP spent in the activation of formate.

Since the energy available for biosynthetic reactions ultimately stems from the oxidation of reduced ferredoxin, the number of redox reactions able to reduce ferredoxin and of the electron-accepting reactions that do not use ferredoxin as electron donors determines the ATP yield. In Acetobacterium woodii, which has only an electron-bifurcating hydrogenase to reduce ferredoxin (9) and uses 1 mol H₂, 2 mol NADH, and 1 mol of reduced ferredoxin for the four redox reactions of the WLP (10), only 0.5 mol reduced ferredoxin is available for the Rnf complex and only 0.3 mol of ATP/acetate are formed by this chemiosmotic mechanism (5).

A loss of the acetate kinase reaction would make the WLP highly energy consuming (by −0.7 ATP) if no other reactions improved the ATP yield either by substrate level phosphorylation or by reducing ferredoxin that then “fuels” the chemiosmotic mechanism. Therefore, it is the more surprising that acetogens can also make a living from producing ethanol that is produced by way of acetyl-CoA (1). Apparently, ethanol production must be coupled to energy conservation, but the mechanisms are obscure. Some acetogens may also produce other reduced end products from acetyl-CoA and face the same energetic problem (11).

The conversion of gases to acetate is of special interest since it is a promising way of producing biofuels and biocommodities independent of carbohydrates. The conversion of syngas, a mixture of H₂, CO, and CO₂, to ethanol by microbial fermentation is already demonstrated in large industrial-scale pilot plants by companies like Coskata, INEOS Bio, or LanzaTech. LanzaTech is using the acetogen Clostridium autoethanogenum as a catalyst (12, 13). Molecular techniques have been used to engineer acetogens to produce longer carbon compounds such as butanol or 2,3-butanediol from acetyl-CoA (14, 15). So far, production is low and the process in its infancy.

Eubacterium limosum KIST612 is a Gram-positive acetogen that grows on syngas, H₂ + CO₂ or CO (16, 17). Interestingly, during growth on CO it produces butyrate alongside acetate (18). E. limosum KIST612 is one of the few acetogens known to produce butyrate (11) and may become an interesting alternative in the production of higher carbon chain compounds from gases. A prerequisite for improving product yields is the knowledge of the biosynthetic pathway and its energetics. To address this question, we have used a genome-guided experimental approach to identify the biosynthetic route for acetyl-CoA and butyrate and its coupling to energy conservation in E. limosum KIST612.

MATERIALS AND METHODS

Gene analysis, comparison, and alignment.

The website of Kyoto Encyclopedia of Genes and Genomes (KEGG) was used for gene classification and clustering analysis (19). Gene comparison based on protein function was carried out by using the protein blast program of the National Center for Biotechnology Information (NCBI) (20). Multiple-sequence alignment was performed using Clustal Omega (21, 22).

Bacterial strain and cultivation.

E. limosum KIST612 was cultivated in anaerobic phosphate-buffered basal medium (PBBM) at 37°C (23). Glucose (20 mM) or 101 kPa of H₂-CO₂ (80:20, vol/vol) was used as the carbon and energy source for cultivation. When H₂-CO₂ was used, KHCO₃ (6.0 g/liter) was added to PBBM.

Ferredoxin:NAD⁺ oxidoreductase assay.

The ferredoxin:NAD⁺ oxidoreductase activities were measured according to the method detailed in reference 24.

Acetate formation in cell suspensions.

All the following steps were performed under strictly anoxic conditions at room temperature in an anaerobic chamber (Coy Laboratory Products, USA) filled with 96% N₂ and 4% H₂. The CO-grown cells were harvested by centrifugation for 10 min at 8,000 rpm and 4°C and washed with imidazole buffer (50 mM imidazole, 20 mM MgSO₄, 20 mM KCl, 4.4 μM resazurin, 4 mM DTE [dithioerythritol], pH 7.0). The cells were resuspended to a protein concentration of 20 mg/ml and transferred to Hungate tubes, and the gas phase of the cell suspensions was changed to N₂ to remove the remaining H₂ from the anaerobic chamber. Protein concentration was determined as described in reference 25. For the cell suspension experiments, the resting cells were resuspended in the same buffer to a protein concentration of 1 mg/ml in glass bottles. NaCl (20 mM) and H₂-CO₂ (80:20, vol/vol; 101 kPa) were added when indicated. The cell suspensions were incubated at 37°C in a shaking water bath (BS-11 Lab Companion; Jeio Tech, Kimpo, South Korea) at 180 rpm. Samples were taken every 20 min, cells were removed via centrifugation (13,000 rpm, 5 min), and the acetate concentration was determined in the supernatant using an enzymatic assay commercially available (acetic acid-UV method; Boehringer Mannheim GmbH, Mannheim, Germany).

Cell harvest and cell lysis.

Glucose-grown cells were harvested by centrifugation for 10 min at 8,000 rpm and 4°C. The cells were suspended in buffer A (25 mM Tris, 420 mM saccharose, 4.4 μM resazurin, 2 mM DTE, pH 7.5) and were centrifuged again (10 min, 8,000 rpm, 4°C). The cell pellet was suspended in buffer A, and lysozyme (0.02 mg/ml) was added. After an incubation for 45 min at 37°C, cells were pelleted again (15 min, 10,000 rpm, 4°C). The cell pellet was suspended in 15 ml of buffer B (50 mM Tris, 20 mM MgSO₄, 20% glycerol, 2 mM DTE, pH 7.5) with phenylmethylsulfonyl fluoride (PMSF; 0.5 mM) and DNase I (0.1 mg/ml). Cells were disrupted by passing three times through a French pressure cell press (SLM Aminco; SLM Instruments, Rochester, NY, USA) at 1,000 lb/in². The cell debris was removed by centrifugation for 30 min at 14,000 rpm and 4°C, and the cell-free crude extract was collected. The membrane fraction was removed from the crude extract by ultracentrifugation at 42,000 rpm for 60 min. The cytoplasmic fraction was stored under anoxic conditions at 4°C.

Activities of the butyryl-CoA dehydrogenase.

The cytoplasmic fraction containing 352 mg of protein was applied to a Q Sepharose high-performance column (2.6 by 5 cm) equilibrated with buffer B. Proteins were eluted with an NaCl gradient (0 to 1 M) at a flow rate of 2.0 ml/min. Enzymatic measurements were performed at 30°C in 1.8-ml anoxic cuvettes (Glasgerätebau Ochs, Germany) sealed by rubber stoppers under an N₂ atmosphere. The oxidation of NADH was monitored at 340 nm (ε = 6.3 mM⁻¹ cm⁻¹), and the reduction of ferredoxin was monitored at 430 nm (ε = 13.1 mM⁻¹ cm⁻¹). The activities were measured in 50 mM Tris buffer (pH 7.5) containing 2 mM DTE. Flavin adenine dinucleotide (FAD; 5 μM), 0.25 mM NADH, 20 μM Fd, and 1 mM crotonyl-CoA were added, and the change of absorbance was monitored. Ferredoxin was purified from Clostridium pasteurianum as described before (26).

RESULTS AND DISCUSSION

Genes involved in the Wood-Ljungdahl pathway.

The genes for the Wood-Ljungdahl pathway are localized in three distinct clusters (Fig. 1), similar to what is seen in the closely related A. woodii (27). The first cluster contains a gene encoding a formate dehydrogenase (fdhA-II, ELI_0994). It is clustered with genes involved in the incorporation of the cofactor molybdenum and for the activity of the protein (ELI_0996–0997). The formate dehydrogenase (FDH) of A. woodii contains an additional hydrogenase subunit, which enables the direct use of H₂ for CO₂ reduction (9), but a hydrogenase-encoding gene is not found next to fdhA-II in E. limosum KIST612. An N-terminal extension of 218 amino acids is present in the putative FDH of the strain, in contrast to the FDH of A. woodii. This extension contains conserved domains for a [2Fe2S] cluster and two [4Fe4S] clusters, which are also present in NAD(P)-dependent FDHs (28), but in these FDH complexes the NAD(P)-binding site is present in another subunit. No gene encoding this second subunit was found in the genome, and the FDH of E. limosum KIST612 does not contain a predicted NAD(P)-binding site.

FIG 1.

Wood-Ljungdahl pathway in E. limosum KIST612 and genetic organization of the encoding genes. The formation of acetate as well as that of butyrate via the Wood-Ljungdahl pathway are indicated. THF, tetrahydrofolate. Gene cluster I contains genes for a formate dehydrogenase (fdhA-II) and an accessory protein (fdhD). Gene cluster II contains genes for the formyl-THS synthetase (fhs1), methenyl-THF cyclohydrolase (fchA), methylene-THF dehydrogenase (folD) and the methylene-THF reductase (metF, metV). Gene cluster III contains the genes for the CO dehydrogenase (acsA) and the acetyl-CoA synthase (acsB), the corrinoid/iron sulfur protein (acsCD), a methyltransferase (acsE), CODH nickel-insertion accessory proteins (cooC1 and cooC2), a corrinoid activation and regeneration protein (acsV), and a hypothetical protein (orf).

The genes encoding the proteins required for the activation of formate to formyl-THF and the further reduction to methyl-THF are localized in cluster II (ELI_0372–0376). The methylene-THF reductase (MTHFR) contains two subunits, MetF and MetV, which are also present in other acetogens (10, 14, 27, 29). No other genes that encode additional subunits of the MTHFR are present in this cluster, in contrast to A. woodii, which contains rnfC2 in addition (10), and Moorella thermoacetica, which has the genes for a putative electron-bifurcating heterodisulfide reductase (Hdr) module next to metFV (29). Therefore, the MTHFR of E. limosum KIST612 seems to be of the MetFV type, as in Clostridium ljungdahlii (14). Noteworthy is the presence of a putative second MTHFR-encoding gene in the genome (ELI_2371). This gene encodes a putative MTHFR of the MetF type, but the product shares only very little similarity (21% identity) to the MetF encoded by cluster II. The highest identities are with MetF of Clostridium carboxidovorans (75%) and Clostridium drakei (74%). It is also 51% identical to the MTHFR of Blautia producta, which was characterized to be a FAD-containing, O₂-insensitive enzyme that uses NADH as the electron donor for the reduction of methylene-THF (30). Neither metV nor other genes encoding a distinct function in relation to the reduction of methylene-THF were found upstream or downstream of ELI_2371.

The third cluster contains all the genes required for the key enzyme of the WLP, the carbon monoxide dehydrogenase/acetyl-CoA synthase (CODH/ACS) (ELI_3601–3609). The genes for the CODH/ACS (acsAB) are clustered together with the genes for the corrinoid-iron-sulfur protein (acsCD) and the methyltransferase that transfers the methyl group from methyl-THF to the corrinoid-iron-sulfur protein (acsE). Also present are the genes coding for a corrinoid activation/regeneration protein (acsV) and two copies of a CODH nickel insertion accessory protein (cooC). The proteins encoded by this cluster are similar to the corresponding proteins from A. woodii with identities ranging from 67% for AcsD to 87% for AcsE.

When growing on syngas, E. limosum KIST612 has to oxidize the gaseous substrate CO and/or H₂. Genes encoding a monomeric CODH are apparently not present in the genome; therefore, the CODH that is part of the CODH/ACS is the key enzyme when growing on CO. Two genes that encode putative hydrogenases can be found in the genome. The first encodes a putative electron-bifurcating hydrogenase, HydABCD (ELI_0843–0847) similar to the one present in A. woodii (9). The second putative hydrogenase (ELI_2538) has all the characteristic sequence signatures for a [FeFe] hydrogenase, and the cysteine ligands of the H cluster are also conserved (31).

Energy conservation during autotrophic growth of E. limosum KIST612.

So far, acetogens are only known to have either an Rnf complex or an Ech complex to couple oxidation of reduced ferredoxin to the generation of a membrane potential (5). Inspection of the genome of E. limosum KIST612 revealed genes encoding a Rnf complex but did not reveal Ech genes. The Rnf-encoding genes are ELI_2638 to ELI_2643 (encoding RnfABCDEG). Each Rnf subunit is most similar to the corresponding one from A. woodii (Awo_2201 to Awo_2206) with identities of 86% (RnfA), 63% (RnfB), 68% (RnfC), 75% (RnfD), 79% (RnfE), and 53% (RnfG). RnfB is suggested to have iron sulfur clusters and four [4Fe-4S] ferredoxin-like binding domains. RnfC most likely has two [4Fe-4S] binding domains. RnfA, RnfD, and RnfE are expected to be transmembrane proteins. RnfG has a conserved region for a covalently attached flavin mononucleotide (FMN).

To confirm the presence of the Rnf complex and to address the nature of the coupling ion used, we attempted to measure its enzymatic activity in membranes of E. limosum KIST612. Indeed, the activity could be measured, and without Na⁺ added, the ferredoxin-dependent NAD⁺ reduction as carried out by these membranes amounted to 110 mU/mg. Interestingly, this activity was stimulated 3-fold by addition of 5 mM NaCl (Fig. 2). The Na⁺ stimulation is in accordance with the hypothesis that the Rnf complex uses Na⁺ as coupling ion. The activity of a Na⁺-dependent Rnf complex should result in a sodium ion dependence of acetate formation from H₂ + CO₂. And indeed, resting cells produced almost no acetate from H₂ + CO₂ in the absence of NaCl, but in the presence of 20 mM NaCl, acetate was produced with a rate of 4 nmol/min · mg (Fig. 3).

FIG 2.

Ferredoxin:NAD⁺-oxidoreductase activity in E. limosum KIST612. The reduction of 1 mM NAD⁺ with 30 μM reduced ferredoxin (reduced with CO dehydrogenase of A. woodii under an atmosphere of 100% CO) as the electron donor was measured with 139 μg membrane protein in 50 mM Tris, 2 mM DTE, pH 7.5. NaCl (5 mM) was added to one assay (solid line), and the control did not receive NaCl (dashed line). The data shown are representative of 3 independent measurements.

FIG 3.

Effect of sodium ions on acetate production from H₂ + CO₂ in resting cells. Resting cells (1 mg/ml) were incubated in buffer (50 mM imidazole–HCl, 20 mM MgSO₄, 20 mM KCl, 4.4 μM resazurin, 4 mM DTE, pH 7.0) containing no additional Na⁺ (○) or 20 mM Na⁺ (●) under an atmosphere of H₂-CO₂ (80:20, vol/vol). The acetate concentration was measured enzymatically. The data shown are representative of 3 independent measurements.

If Na⁺ is the coupling ion for the Rnf complex, the electrochemical ion gradient established should be used to drive ATP synthesis. Inspection of the genome sequence led to an unprecedented finding: the presence of gene clusters encoding two different multisubunit ATPases/synthases with no one enzyme being similar to bacterial F₁F_o ATP synthases. ELI_2184–2192 encodes an enzyme with highest similarity to an archaeal A₁A_o ATP synthase. Most interestingly, the gene encoding the rotor subunit c (ELI_2186) is 474 bp long and encodes a 15.72-kDa protein with four transmembrane helices (TMH). Like the c subunits from the F₁F_o ATP synthase from A. woodii (32) or the A₁A_o ATP synthase from Pyrococcus furiosus (33), it contains a conserved Na⁺ binding motif (Q…ET) but only one ion-binding site. As in P. furiosus but in contrast to A. woodii c₁, the Na⁺-binding site is in TMH3 and TMH4 (Fig. 4). ELI_2424–2416 codes for an enzyme with similarity to V₁V_o ATPases; however, the order of the genes is unusual. The gene encoding the subunit c (ELI_2421) is 435 bp long and encodes a 14.48-kDa protein with four TMHs. The c subunit does not contain the conserved Na⁺ binding motif but only the glutamate residue essential for proton binding, indicating that the enzyme is a proton pump.

FIG 4.

The c subunit of the putative A₁A_o ATP synthase has the conserved Na⁺-binding motif. The amino acid sequence of the c subunit of the putative A₁A_o ATP synthase of E. limosum KIST612 (ELI_2186) was aligned with the sequences of the c₁ subunit of the F₁F_o ATP synthase of Acetobacterium woodii (Awo_c02160) and the c subunit of the A₁A_o ATP synthase of Pyrococcus furiosus (PF0178). The Na⁺ binding motif (Q…ET) is indicated in bold.

Taken together, the data provide evidence that E. limosum KIST612 conserves energy by a chemiosmotic mechanism with Na⁺ as coupling ion and an Na⁺-translocating Rnf complex and an Na⁺-translocating ATP synthase as key players. Therefore, the more reduced ferredoxin is available for the Rnf complex, the more ATP can be synthesized. To be able to calculate the ATP yield, we analyzed other reactions of the WLP for the reductant used. The reduction of CO₂ to CO (E₀′ = −520 mV) requires reduced ferredoxin as reductant. The redox potential of methenyl-THF/methylene-THF is −295 mV, and NADH is therefore sufficient as the electron donor. Thus, the two bioenergetically interesting redox reactions are the reduction of CO₂ to formate and the reduction of methylene-THF (Fig. 5). The predicted FDH does not contain an NAD(P)-binding site and is related to Fd-dependent enzymes (34); therefore, ferredoxin is a plausible electron donor. But since there is a monomeric hydrogenase encoded in the genome, it is also possible that this hydrogenase forms a complex with the FDH, similar to the hydrogen-dependent CO₂ reductase from A. woodii (35) that can use hydrogen directly. Reduction of methylene-THF is exergonic and thought to be electron bifurcating in Moorella thermoacetica with the subunits Moth_1191–1196 assumed to convey electron bifurcation to the base module (MetFV). MetFV proteins of E. limosum KIST612 are highly similar (67% and 63%) to the corresponding proteins from C. ljungdahlii (14). As in C. ljungdahlii, there are a couple of etfAB genes in the genome of E. limosum KIST612, encoding potential electron-transferring flavoproteins (Etf). This could be interpreted to result in an MetFV-EtfAB complex that may reduce ferredoxin with NADH as reductant by electron bifurcation. A final conclusion regarding the nature of electron donors/acceptors for FDH and MTHFR requires purification and characterization of these enzymes. However, the theoretical ATP gain can be calculated, dependent on the different electron donors and the assumption that the ATP synthase uses 4 Na⁺/ATP. Considering an Fd-dependent FDH and a nonbifurcating MTHFR, the ATP gain would be zero under autotrophic conditions. Since the organism does grow on H₂ + CO₂ and produces only acetate, this combination can be excluded. If the FDH used H₂ instead of Fd, this would allow the synthesis of 0.25 ATP/acetate. If the MTHFR reduced Fd via electron bifurcation, this would allow the synthesis of 0.5 ATP/acetate. Thus, having an H₂-dependent FDH and a bifurcating MTHFR in combination, the ATP gain would be 0.75 ATP/acetate. That the formate dehydrogenase forms a complex with the monomeric hydrogenase that is encoded somewhere else in the genome is very unlikely. It is much more plausible that the MTHFR of E. limosum KIST612 reduces ferredoxin via electron bifurcation. For the sake of simplicity, we will assume a bifurcating MTHFR and an Fd-dependent FDH, yielding 0.5 mol ATP/mol acetate.

FIG 5.

Biochemistry of and energy conservation during acetogenesis by E. limosum KIST612. It is not known whether the methylene-THF reductase reduces ferredoxin, and the electron donor for the formate-dehydrogenase is not known. THF, tetrahydrofolate.

Butyrate is produced with CO as the substrate.

When E. limosum KIST612 grows on CO as a carbon and electron donor, it starts to produce butyrate at the mid-exponential growth phase (23). Inspection of the genome sequence revealed genes encoding thiolase (thlA, ELI_0537), 3-hydroxybutyryl-CoA dehydrogenase (hbd, ELI_0538), crotonase (crt, ELI_0539), and butyryl-CoA dehydrogenase (bcd, ELI_0540) (Fig. 6A). Next to the bcd gene, there are two genes (etfAB; ELI_0541 and ELI_0542) encoding two subunits of a potential Etf protein. This gene cluster is well conserved in butyrate- or butanol-forming clostridia such as Clostridium acetobutylicum (36), Clostridium beijerinckii (37), Clostridium saccharobutylicum (38), and C. carboxidivorans (39). The encoded proteins are most similar to the corresponding proteins from Butyrivibrio proteoclasticus (Clostridium proteoclasticus), with identities ranging from 62% (EtfA) to 70% (Bcd).

FIG 6.

Genetic organization (A) and enzymatic activity (B) of the butyryl-CoA dehydrogenase. (A) thlA, thiolase gene; hbd, 3-hydroxybutyryl-CoA dehydrogenase gene; crt, crotonase gene; bcd, butyryl-CoA dehydrogenase gene; etfB, gene encoding the small subunit of electron transfer flavoprotein; etfA, gene encoding the large subunit of electron transfer flavoprotein. (B) The enzyme assay was performed in anoxic cuvettes under an atmosphere of 100% N₂ in 50 mM Tris-HCl, 2 mM DTE, pH 7.5. Protein (1.7 mg), 5 μM FAD, 0.25 NADH, 20 μM ferredoxin, and 1 mM crotonyl-CoA were used. NADH oxidation was measured at 340 nm (ε = 6.2 mM⁻¹ cm⁻¹), and ferredoxin reduction was measured at 430 nm (ε = 13.1 mM⁻¹ cm⁻¹). The data shown are representative of 2 independent measurements.

The presence of the genes etfA and etfB next to bcd is reminiscent of the genetic organization of the prototype of flavin-dependent, electron-bifurcating enzymes, the butyryl-CoA dehydrogenase complex from Clostridium kluyveri (40). This prompted us to determine whether or not the butyryl-CoA dehydrogenase from E. limosum KIST612 reduces ferredoxin by electron bifurcation. Since we were not able to detect Bcd activity in the cytoplasm, we separated cytoplasmic proteins via anion-exchange chromatography. One of the fractions did oxidize NADH and reduce Fd, but only after crotonyl-CoA was added (Fig. 6B). Oxidation of NADH and reduction of Fd were observed simultaneously at 340 and 430 nm, respectively. The oxidation of NADH and reduction of Fd were seen only after addition of 1 mM crotonyl-CoA. The specific activities of NADH oxidation and Fd reduction were calculated as 17.5 mU/mg (1 U = 1 μmol of NADH/min) and 6.5 mU/mg (1 U = 1 μmol of Fd/min), respectively. This activity was strictly dependent on the presence of NADH, ferredoxin, and crotonyl-CoA. Addition of FAD stimulated the activity (Fig. 7A). In comparison to the values obtained without crotonyl-CoA, NADH oxidation was increased 12-fold and Fd reduction was increased 9-fold. Both activities were stimulated 3-fold by the addition of FAD as a cofactor. The stoichiometry of NADH oxidized and ferredoxin reduced was calculated to be 2.6:1 (Fig. 7B). Taken together, these data are consistent with a Bcd/Etf complex in E. limosum KIST612 that couples NADH-dependent reduction of crotonyl-CoA to the reduction of ferredoxin via flavin-dependent electron bifurcation, similar to the Bcd/Etf complex from C. kluyveri (40).

FIG 7.

NADH, ferredoxin, and FAD requirements of crotonyl-CoA reduction and stoichiometry of NADH oxidation and ferredoxin reduction. (A) Enzymatic assays were done as described in the legend to Fig. 6. (B) The oxidation of NADH and the reduction of ferredoxin were monitored simultaneously at 340 and 430 nm, respectively. The data shown are representative of 2 independent measurements.

Conclusion.

E. limosum KIST612 does produce significant amounts of butyrate only when fermenting CO, not when growing on H₂ + CO₂. The formation of butyrate starts with two moles of acetyl-CoA, which are condensed to acetoacetyl-CoA. Acetoacetyl-CoA is reduced to hydroxybutyryl-CoA, and water is split off, giving crotonyl-CoA. Crotonyl-CoA is reduced to butyryl-CoA by action of the ferredoxin-reducing, electron-bifurcating butyryl-CoA dehydrogenase. A phosphotransacetylase reaction gives butyryl phosphate, which is the phosphoryl group donor for ADP, giving butyrate and ATP. When E. limosum KIST612 is growing on H₂ + CO₂, butyrate will be formed as follows: 10H₂ + 4CO₂ → 1 butyrate + H⁺ + 6H₂O.

Having the electron-bifurcating hydrogenase for H₂ oxidation, this reaction theoretically yields only 1 mol ATP/mol butyrate (Fig. 8). If E. limosum KIST612 is growing with CO as energy source, the reducing equivalents are provided by the CODH/ACS as reduced ferredoxin. NADH is generated from reduced ferredoxin by action of the Rnf complex, which translocates Na⁺. Altogether, butyrate formation from CO occurs according to the following reaction: 10CO + 4H₂O → 1 butyrate + H⁺ + 6CO₂. This yields 3.5 mol ATP/mol butyrate and therefore considerably more ATP than does butyrate formation from H₂ + CO₂ (Fig. 9). In addition, regulatory effects may also be the basis for this phenomenon.

FIG 8.

Energy conservation during H₂ + CO₂-dependent butyrate formation by E. limosum KIST612. For the sake of simplicity, a ferredoxin-dependent formate dehydrogenase and an electron-bifurcating methylene-THF reductase are assumed.

FIG 9.

Energy conservation during CO-dependent butyrate formation by E. limosum KIST612. For the sake of simplicity, a ferredoxin-dependent formate dehydrogenase and an electron-bifurcating methylene-THF reductase are assumed.